Icom Metal2022

METAL2022
SEPTEMBER 5–9, 2022

HELSINKI, FINLAND
PROCEEDINGS OF THE INTERIM MEETING OF THE ICOM-CC METALS WORKING GROUP
EDITED BY PAUL MARDIKIAN, LIISA NÄSÄNEN, AND AKI ARPONEN

METAL2022
PROCEEDINGS OF THE INTERIM MEETING OF THE ICOM-CC METALS WORKING GROUP
SEPTEMBER 5–9, 2022

HELSINKI, FINLAND
EDITED BY PAUL MARDIKIAN, LIISA NÄSÄNEN, AND AKI ARPONEN

© 2022 International Council of Museums – Committee for Conservation (ICOM–CC) and The National Museum
of Finland Collections and Conservation Centre (NMF)
The Metals Working Group is part of the Committee for Conservation (ICOM–CC), a committee of the International
Council of Museums (ICOM) network.
Publishers: International Council of Museums – Committee for Conservation (ICOM–CC) and The National Museum
of Finland
All rights reserved. No part of this book may be reproduced in any form or by any means, whether electronic or
mechanical, including photocopying, recording, or otherwise, without the prior permission in writing from the
copyright holder.
ISBN: 978-2-491997-61-8
Proceedings and user interface design and production: Eduardo Pulido ([email protected])
Copy editing: Carla Nunes ([email protected]), Per Christopher Foster ([email protected]), and Wendy
Ran ([email protected])
Paper-flow contribution platform production and management: use.it® – Virott e Associados, Lda. (www.useit.pt)
Cover image: Viking Age swords found in Finland, now in the Archaeological Collections of the Finnish Heritage
Agency. The hilts in Viking Age swords are true masterworks of metal technology. The mythical animals and palmettes
in the hilt of KM10833:1 (front cover, left) were created with gilt silver and niello; in the pommel of KM9243:1 (back
cover, right), copper and silver wires weave a herringbone motif. © Finnish Heritage Agency
Table of contents
Foreword. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ix
Preface. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . x
Acknowledgments. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xi
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xiii
In memoriam David Hallam (1956–2020). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xiv
Sustainability in Practice
Natural Conservation of Archaeological Iron
H. Matthiesen, J. Frydendahl, L.B. Andersen, C. Carré, P. Dillmann, D. Neff, T. Wiinblad .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
Exploiting Biologically Synthetized Chelators in Conservation: Gel-based Bio-cleaning of Corroded
Iron Heritage Objects
L. Cuvillier, A. Passaretti, A. Raimon, V. Dupuy, E. Guilminot, E. Joseph .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
Microbial Stabilization of Archaeological Iron Objects (YCP)
S. James, M. Monachon, S. Ganesan, E. Joseph . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
Deterioration Tests Using a Simulated Tumulus to Evaluate In Situ Preservation of Metal Artifacts in a
Stone Chamber
A. Yanagida, S. Wakiya, H. Xie, D. Ogura, N. Takatori, H. Yasui, S. Hokoi, Y. Kohdzuma . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
Development of a Sunflower-Oil-based Bio-lubricant for Use in Gear Models at the ISEP Museum (YCP)
M. Gonçalves, B. Campos, E. Vieira, P.R. Moreira . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
Modern Metals
A Review of Treatments of Magnesium Alloy Objects at the National Air and Space Museum
K. Wilcox, M. Sweeney, L. Horelick . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
Mitigating Magnesium Corrosion: Testing the Efficacy of Reapplying Conversion Coatings in
Conservation Treatment (YCP)
A.E. Rodriguez. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
Corrosion and Conservation of Nickel Silver Alloys Recovered from Historic Shipwrecks
I.D. MacLeod. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
First Step in the PROCRAFT Project on WWII Aircraft Heritage: Investigation and Conservation of the
Aluminium Alloys
M. Brunet, L. Robbiola, C. Deshayes, E. Bernardi, C. Martini, C. Chiavari, A. Balbo, C. Monticelli, J. Fišer,
T. Vyhlídal, J. Echinard, E. Guilminot . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78
DiscoveryMat: A Freeware Electrochemical Tool for the In Situ Analysis of Aluminum Alloys in the Cooling Systems
of Historical Vehicles (YCP)
E. Granget, B. Chalançon, C. Degrigny . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
Coatings, Consolidants & Corrosion Inhibitors

A Comparison of the Long-Term Outdoor Performance of Two Modern Paint Coating Systems and a
Traditional Lead-based Paint Applied to Historic Wrought Iron
P. Meehan, N. Emmerson, D. Watkinson . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94
iv ICOM-CC | METAL 2022 | HELSINKI, FINLAND
Keris Blade Protection – A Comparison of Traditional and Contemporary Coatings (YCP)

Janine Meier . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103
Preliminary Investigation of the Use of Chitosan-based Coatings for Outdoor Cultural Heritage Objects (YCP)
C. Carvalho, N. Silva, E. Vieira, P.R. Moreira . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109
Testing the Compressive Strength and Reversibility of Consolidants Applied to Marine Archaeological Gray
Cast Iron
E. Farrell, M.K. McGath, J. Echerd . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114
Calcium Sulfonate: Adsorptive Inhibitors for Metal Conservation?
E. Wentland .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123
Investigating a Sustainable Alternative: L-Cysteine as a Non-Toxic Corrosion Inhibitor for Copper Alloy
Conservation (YCP)
E. Tréhu, D. Sully .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 132
Technical Studies
Sustainable Growth of Mewari Mail Makers: Traditional to Contemporary Practices
V. Singh . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 140
Development of Mexican Bell Documentation Tools and Methodologies
J. Contreras Vargas, D. Lira Pacheco, Á.E. García Abajo, F. Llop Álvaro, F. Llop I Bayo, F. Magaña Nieto . . . . . . . . . . . 146
Corrosion Detection by Color Change Using Crowdsourced Photographs. Preliminary Results of the MIPAC Project
B. Ramírez Barat, M.T. Molina, E. Cano . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 153
Reflectance Transformation Imaging Feature Maps for Visual Documentation of Metal Objects
A. Siatou, M. Nurit, G. Le Goïc, A. Mansouri, L. Brambilla, C. Degrigny . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 160
Revealing Degradation Patterns: Imaging Techniques for the Study of Metal Soap Formation on Painted
Metal Objects
S. Russo, L. Brambilla, J.-B. Thomas, E. Joseph . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 168
Monitoring of Metal Sculptures and Their Environment in Art Museums in Sweden
S. Golfomitsou, E. Canosa, M. Pullano, M. Borin, E. Nygårds, M. Florescu, K. Hermerén . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 176
Discovery of a 12th-Century Enamelled Reliquary Pendant: Elemental Analysis and Content Visualisation
Using Prompt Gamma Neutron Activation Analysis and Neutron Tomography
M. Heinzel, E. Kluge, D. Kemper, B. Schillinger, C. Stieghorst .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 184
Analytical Investigations
Sulfur, the ‘Enemy of Copper’: Replication of Sulfurous Efflorescence on Copper-based Heritage Materials with
Elemental Sulfur
C.F. Kuhn-Wawrzinek, G. Eggert, T. Schleid . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 194
The Role of Patina on Archaeological Copper Alloy Coins in the Outbreak and Progression of Bronze Disease
J.C. Thunberg, D.E. Watkinson, N.J. Emmerson, Z. Kis, I. Harsányi, Z. Kasztovszky, M. Lewis . . . . . . . . . . . . . . . . . . . . . . . . 203
Non-Invasive Analysis: The Mirage and the Reality
P. Northover. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 212
ICOM-CC | METAL 2022 | HELSINKI, FINLAND v
An Analytical Study of the Corrosion Behavior and Microstructural Properties of a Group of Copper Alloy
Artifacts from the Khirbet Yajuz Archaeological Site, Jordan
A.N. Abu-Baker, L.A. Khalil . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 221
Corrosion Analysis and Assessment of a Collection of Archaeological Copper Alloy Objects from Sagzabad,
Northern Iran
O. Oudbashi, S.Bahadori, A. Aliyari . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 231
The Effect of Synthesis Routes on the Ability of Akaganeite to Corrode Iron
N.J. Emmerson, D.E. Watkinson, K. Roche, J.H. Seifert, J.C. Thunberg . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 240
Analysis of Heterogeneous Tarnish on Silver-based Alloys Using the Pleco for Local, Controlled
Electrolytic Cleaning (YCP)
N. Ricotta, A. Cagnini, C. Degrigny .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 248
Using Epidemiology to Validate Scientific Results for Complex Situations
D. Thickett . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 253
Long-Term Provision of Stable Environments for Metals Conservation
P. Lankester, D. Thickett, S. Johnson . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 261
Conservation Practices
The Reanimation of a Renaissance Automaton: Diana and Stag at the Museum of Fine Arts, Boston
S. Gänsicke, R. Lang .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 272
Building an Ironclad System: A Quarter-Century of Innovation in the Conservation of USS Monitor
E. Farrell, E. Sangouard, L. Haines, L. King, M.K. McGath, W. Hoffman . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 281
Conservation and Technical Study of Ten Objects in the Thiriyaya Collection at the Department of
Archaeology in Sri Lanka
K.A. Anusha Kasthuri, D. Ruckmal Athukorala, A. Fernando . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 290
Conservation of Copper and Copper Alloy Objects: A Training Programme Conducted at the Mehrangarh Museum
Trust, Jodhpur (YCP)
S. Raja, S. Rathore .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 300
Keeping it Kosher: The Care and Treatment of the Judaica Collection at the Israel Museum, Jerusalem
J. Lewinsky, H. Seri, M. Delano, I.L. Beyth . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 308
Commercially Available Vacuum Chambers as an Alternative in the Deoxygenated Desalination Treatment of
Archaeological Iron (YCP)
A. Pienimäki. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 317
POSTER SESSION
Subcritical Fluids in the Conservation of Archaeological Iron Objects: From Experiments to Permanent Practice
A. Arponen, A. Hyppönen, P. Kilpeläinen, K. Kaipanen .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 324
When Corrosion is Useful: Absolute Dating of Metals by the Radiocarbon Method
L. Beck, C. Messager. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 325
Corrosion, Communication, and Comparison: Collaborative Conservation Approaches Towards Two Medieval
Axes from Norfolk, England
K. Berlewen, R. Hudson . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 326
vi ICOM-CC | METAL 2022 | HELSINKI, FINLAND
Atmospheric Particulate Matter: How to Include it in Artificial Ageing?

E. Bernardi, C. Martini, C. Velino, C. Chiavari, I. Vassura . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 327
Study and Restoration of the Trajan’s Column Electrotypes at the National Museum of Archaeology of
Saint-Germain-en-Laye
I. Bonora Andujar, J. Jouet, A. Amarger, B. Branche, A. Molineri . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 328
Acoustic Emission Techniques for the Detection and Monitoring of Corrosion Phenomena
L. Brambilla, B. Chalançon, A. Roda-Buch, S. Mischler . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 329
The Bronze Hand of Prêles: Protecting a Valuable Archaeological Object Using 3D Techniques
S. Brechbühl. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 330
Corrosion of Strings on Musical Instruments
V. de Bruyn-Ouboter, A. Erbe, E.F. Gustad . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 331
Do the Lubricating Oils Used in Scientific-Technological Objects Protect Metals Against Corrosion?
E. Cano, B. Ramírez Barat, J. Leal, M.T. Molina . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 332
The Bluish Green and the Greenish Blue
M. Cardoso . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 333
Heritage Biota and Bronze Patina Composition: A Correlation?
C. Chiavari, A. Timoncini, C. Martini, E. Bernardi, F. Costantini . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 334
Testing Silver Lacquers: What about Agateen #27?
G. Eggert, G.D. Smith, M.J. Samide . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 335
Possibilities for Implementing Aesthetic Unity, Reversibility and Distinctiveness: Two Conservation Case Studies
E. Tóth . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 336
Supercritical Carbon Dioxide: Cleaning and Stabilization Tests for Copper Alloys and Iron Keys from a Scientific
Collection of the Centro Hospitalar Conde de Ferreira (Porto, Portugal)
C. Figueiredo, N. Camarneiro, C. Bottaini, R. Bordalo, I. Silva, M. Duarte, E. Vieira . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 337
A Simple Treatment System for the Deaerated Desalination of Archaeological Iron
J. Frydendahl, L.B. Andersen, T. Wiinblad, C. Carré, D. Neff, P. Dillmann, H. Matthiesen .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 338
Metals Conservation Teaching at University Level in Greece and the Cultural Vision of EYDAP S.A.
M. Giannoulaki, C. Panagiotopoulou, S. Tzimopoulou, V. Argyropoulos .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 339
A Lacquer to Dye For! Exploring a UV Fluorescent Additive for Coating Applications in Silver Conservation
R. Grayburn, L. Fair . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 340
Port-Jeanne-d’Arc: Is Conservation of an Isolated Site Possible?
M. Grima . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 341
Evaluation of the Impact of Dechlorination Treatments on the Organic Parts of Composite Objects
E. Guilminot, C. Pelé-Meziani, S. Labroche . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 342
Historical Repairs on Ancient Chinese Bronzes: Identification and Conservation Issues
K.-S. (Tracy) Han . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 343
Microclimates for Marine Archaeological Iron Artillery: Three Case Studies in Treatment, Storage, and Display
L. King, E. Farrell . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 344
ICOM-CC | METAL 2022 | HELSINKI, FINLAND vii
Results of Research on 17th–19th-Century Coffin Portraits in the Collection of the National Museum in Krakow
M. Labut, A. Stępień, J. del Hoyo-Meléndez, M. Obarzanowski, P. Krupska, K. Stefańczyk, M. Goryl .. . . . . . . . . . . . . . . 345
Examination of an Unusual Apulian-Corinthian Helmet Using X-Ray and Computed Tomography
M. Leroux, E. Lambert . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 346
Diagnostic, Restoration and Maintenance of Outdoor Bronze Artworks: Investigating Patinas in Marine
Environments and Their Stabilisation with Low Environmental Impact Treatments
P. Letardi, G. Monari . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 347
The Curious Case of Storage at the Petrie Museum of Egyptian Archaeology
G. McArthur, M. Chow . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 348
Comparing Simple Portable XRF with PIXE Analysis for Archaeological Metal Objects: Suitability, Advantages
and Drawbacks
E. Menart, Ž. Šmit . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 349
Methods of Marking Small Museum Objects: A Joint Project of the National Museum in Warsaw and the National
Institute of Cultural Heritage in Poland
A. Mistewicz, J. Kwiatkowska . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 350
Influence of the Application Methodology on the Performance of Coatings for the Protection of Metallic Heritage
Objects
M.T. Molina, B. Ramírez Barat, E. Cano . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 351
Archaeological Tinned-Copper Objects: A Study on Corrosion and Cleaning Treatments
M. Mortazavi, H.R. Bakhshandehfard, A. Abed-Esfahani .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 352
Archaeometallurgical Investigations of a Late Iron Age Helmet: An Interdisciplinary Approach
N. Nemeček, T. Lazar, L. Grahek, M. Nečemer, P. Fajfar, B. Žužek . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 353
The Care and Maintenance of Heritage Sample Collections
P. Northover, V. Cheel . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 354
Monitoring the Effect of Humidity on the Tarnishing of Silver 0.800 by Means of Electrochemical Noise
J. Ortíz-Corona, E.M. García-Ochoa, F.J. Rodríguez-Gómez . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 355
Investigating Precious Artefacts Using Non-destructive FIB on FEG-SEM: A Case Study of 10th-Century Hollow
Gilded Copper Alloy Pendants (Gombiky) from Prague Castle
E. Ottenwelter, C. Josse, L. Robbiola . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 356
The Enemy Within: Dezincification and Cyclic Corrosion on a Brass Patinated Equestrian Sculpture
M.J. Pereira, S. Fragoso, R. Silva, A.P. Rodrigues . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 357
Standardizing Acid Digestion Methodology to Determine the Chloride Content of Ferrous Corrosion Samples
K. Roche, J. Rivera, L. Arslaner, S. Crette . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 358
Treatment of Two Fanghu Vessels (1st Century BC): Conservation Issues of Archaeological Painted Bronze
C. Scarrone, D.M. Scalarone, F. Varallo, M. Demmelbauer, T. Poli . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 359
Investigating Three Types of Dry Cell Battery Deterioration
M. Sweeney . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 360
The Complexity of Developing a Restoration Plan for an 18th-Century Cast Iron Fountain
J. Tauber .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 361
viii ICOM-CC | METAL 2022 | HELSINKI, FINLAND
Testing Gap Fillers for Archaeological Metal

S. Tsvetkov . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 362
The ‘Hallmarks on (Dutch) Silver’ Project: Where Are We Now?
E. van Bork, R. Erdmann, T. Davidowitz . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 363
Metallotheek: A Collaborative Exploration of Bronze Patination
H. van Santen, L. van Santen . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 364
Author index .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 365
Keyword index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 367
ICOM-CC | METAL 2022 | HELSINKI, FINLAND ix
Foreword
The National Museum of Finland and Metropolia this field. Another reason for pride is this publication,
University of Applied Sciences were proud to jointly in which all the papers and poster abstracts presented at
host Metal 2022, the 10th Interim Meeting of the Metals the conference are featured. The publication adds to the
Working Group, part of the International Council of already vast body of scientific articles disseminated in the
Museums – Committee for Conservation. This 30-year Metals Working Group’s conference publications. These
milestone event took place for the very first time in publications are of huge importance and a remaining
Finland and the Northern European region. point of reference to all those working in the field of
metals conservation worldwide.
After nearly three years of global uncertainty and
isolation caused by the COVID-19 pandemic, and Conservation is at the core of preserving cultural heritage,
considering the Russian attack on Ukraine, we were transmitting its value, as well as documenting and enabling
very pleased that, despite all the challenges, Metal 2022 its interpretation in various ways. Without conservation,
went full steam ahead as a hybrid meeting at the there would be no tangible heritage displayed in museums
spectacular Suomenlinna Fortress, a UNESCO World and its preservation, presentation and use for research,
Heritage Site in Helsinki’s harbour. The decision to interpretation, learning or enjoyment would not be
offer a hybrid conference format was another first that possible. Conservation is a key element in the cultural
greatly expands professional inclusion and outreach of heritage sector and thanks to it, museums can better
the Metals Working Group and strongly reflects upon fulfil their function of serving society. Conservation is
and promotes the overarching theme of the conference: also an ever-growing and developing science in its own
‘Sustainability in the Conservation of Cultural Heritage’. right. Because of our role, knowledge, interdisciplinary
This is a vibrant topic reflecting the professional and expertise and responsibility, the National Museum of
personal priorities of today: eco-friendly treatment Finland aims to be on the frontline in developing the
approaches, cross-disciplinary collaborations, social field. We have ambitious goals for introducing new
justice, conservation education, as well as data science, conservation methods and promoting research efforts
computational techniques and visualisation. and perspectives within the field, as well as embracing
more sustainable practices.
The Metals Working Group is concerned with all issues
related to preserving and presenting cultural historical All of this is very much in line with the papers presented
objects made of metals or metallic alloys. These consti- in this publication, which is the result of a tremendous
tute a vast percentage of almost any museum collection. amount of work from contributors, authors and reviewers
This is also the case at the National Museum of Finland, alike from all over the world, the Metals Working Group
which is why we were delighted to see professionals coordinating team and the conference organisers in
in metals conservation and related sciences gather in Helsinki, to whom we are deeply thankful. It was indeed
Helsinki to bring together the latest developments in our great pleasure to welcome you to Helsinki!
Elina Anttila Eero Ehanti

Director General Keeper, Head of the Conservation Department
The National Museum of Finland The National Museum of Finland
x ICOM-CC | METAL 2022 | HELSINKI, FINLAND
Preface
This publication contains the proceedings of Metal 2022, virtual attendance was offered for those who were not
the Interim Meeting of the ICOM-CC Metals Working able to travel to Helsinki. The live-streamed and recorded
Group that was held in Helsinki, Finland, on September virtual component expanded the outreach and access of
5–9, 2022, hosted by The National Museum of Finland the Metals Working Group community worldwide. We
and Metropolia University of Applied Sciences. hope that a hybrid conference model can remain an
option for future Interim Meetings.
The conference program for Metal 2022 formed a
dynamic ensemble of paper and poster presentations, Metal 2022 also marked the 10th international meeting
as well as invited lectures by leading voices in our field. of the Metals Working Group and offered numerous
Authors from 26 countries, including Young Conserva- opportunities to celebrate the resilience and many
tion Professionals, presented their work in Helsinki. In achievements of our members and friends. Gathering
addition, two pre-conference workshops on Cleaning together in person after a worldwide pandemic felt like
Silver Objects and Oxygen Consumption were offered, as a remarkable achievement and clearly demonstrated
well as two wonderful post-conference day trips. the most important pandemic takeaway—our desire for
community and face-to-face dialogue.
We were thrilled to welcome Dr. Caitlin Southwick, who
Congratulations to all the authors, who have put so much
delivered a passionate keynote address on how to become
of themselves into this conference and worked tirelessly
better advocates for sustainability in our field. Caitlin
to have their contributions included in this volume. Their
also served as a moderator during an inspirational panel
collective work not only reflects the constant dialogue
discussion on sustainability—the theme of the confer-
between conservation practice and conservation research,
ence—exploring with our panelists and the audience the
but it also provides an important snapshot of the vitality
future of cultural heritage conservation.
and creativity of the metals conservation community
For the first time in the history of our Working Group, worldwide.
Paul Mardikian
Terra Mare Conservation, LLC
Coordinator, ICOM-CC Metals Working Group

Liisa Näsänen, Aki Arponen, and Eero Ehanti

The National Museum of Finland
Heikki Häyhä
Metropolia University of Applied Sciences
ICOM-CC | METAL 2022 | HELSINKI, FINLAND xi
Acknowledgments
The proceedings were edited by Paul Mardikian, Liisa We thank the Program Committee for volunteering their
Näsänen, and Aki Arponen. time to peer review and discuss abstracts and papers:
The ICOM-CC Coordination Team for 2020–2023 Ahmad N. Abu-Baker

comprised Coordinator Paul Mardikian and Assistant Yarmouk University, Jordan
Coordinators Jerrad Alexander, Valentin Boissonnas,
Jerrad Alexander
Nicola Emmerson, Elodie Guilminot, Vandana Singh,
Smithsonian National Air and Space Museum, USA
and Ellen van Bork.
Vasiliki Argyropoulos
Liisa Näsänen from the National Museum of Finland
Technological Educational Institute of Athens, Greece
served as Program Chair.
Tonny Beentjes
The Steering Committee comprised Liisa Näsänen, Eero The University of Amsterdam, The Netherlands
Ehanti, Heikki Häyhä, and Paul Mardikian.
Valentin Boissonnas
The local organizing committee comprised Liisa Näsänen, Haute Ecole Arc Conservation-restauration,
Eero Ehanti, Heikki Häyhä, Elisa Ahverdov, Aki Arponen, Switzerland
Stina Björklund, Satu Haapakoski, Anna Hyppönen,
Roni Iilomo, Pia Klaavu, Rachel Fay-Leino, Jenniina Emilio Cano Díaz
Laine, Kari Nordfors Ane Orue-Etxebarria, Päivi Paaja- National Center for Metallurgical Research (CENIM),
nen-Salmi, Sanna Paakkanen, Viktor Sohlström, Toni Spain
Spännäri, Tuomas Aatola, and Jussi Linkola. Claudia Chemello
We thank the students and lecturers from the bachelor’s Terra Mare Conservation LLC, USA
degree program in Conservation at the Metropolia Cristina Chiavari
University of Applied Sciences, students of Information University of Bologna, Italy
and Communication Technology at the Metropolia
University of Applied Sciences, the Design and Imple- Jannen Contreras Vargas
mentation services Valovirta at the Helsinki XR Center, National Institute of Anthropology and History,
Avanio Oy, and the Collections and Conservation Centre Mexico
of the Finnish Heritage Agency. Nicola Emmerson
Cardiff University, United Kingdom
Susanne Gänsicke
J. Paul Getty Museum, USA
Stavroula Golfomitsou
University of Gothenburg, Sweden
Elodie Guilminot
Arc’Antique, France
Heikki Häyhä
Metropolia University of Applied Sciences, Finland
Paola Letardi
Institute of Marine Sciences (ISMAR), Italy
Ian D. MacLeod
Heritage Conservation Solutions, Australia
xii ICOM-CC | METAL 2022 | HELSINKI, FINLAND
Paul Mardikian We are grateful to Nicola Emmerson, João Cura D’Ars de

Terra Mare Conservation LLC, USA Figueiredo Junior, Henning Matthiesen, David Thickett,
and David Watkinson for presenting pre-conference
Henning Matthiesen
workshops, and to Vasilike Argyropoulos, Valentin
National Museum of Denmark, Denmark
Boissonnas, Lauren Fair, Heikki Häyhä, Vandana Singh,
Liisa Näsänen and David Watkinson for their participation in the panel
The National Museum of Finland, Finland discussion on the future of our profession, moderated by
Caitlin Southwick.
Delphine Neff
The Archaeomaterials and Alteration Prediction Thank you to invited speakers Laura Brambilla, Philippe
Laboratory (LAPA), France de Viviés, Arie Pappot, and Lisa Young for accepting
our invitation to be part of Metal 2022 and contribute
Omid Oudbashi
to the diverse program; to Ahmad N. Abu-Baker, Jerrad
University of Isfahan, Iran
Alexander, Aki Arponen, Emilio Cano Díaz, Nicola
Lyndsie Selwyn Emmerson, Elodie Guilminot, Ian D. MacLeod, Ellen
Canadian Conservation Institute (retired), Canada van Bork, and Cátia Viegas Wesołowska for chairing
Vandana Singh the paper sessions; and to Anna Hyppönen and Ane
Centre for Art Conservation and Research Experts, Orue-Etxebarria for chairing the poster sessions.
India We thank Miguel Mertens, Guy Silva, and the team
Stina Björklund from use.it, Portugal, for their hard work and unwa-
The National Museum of Finland, Finland vering support of Paper-flow, the paper management
platform used for our conferences, and Joan Reifsnyder
David Thickett for assisting with collecting the copyright transfer agree-
English Heritage, United Kingdom ments.
Johanna Thunberg Last but not least, we are thankful for the work of Carla
Cardiff University, United Kingdom Nunes and Eduardo Pulido, Portugal, who were respon-
David Thurrowgood sible for the layout and design of another beautiful Metals
Applied Conservation Science, Australia Working Group volume. Carla Nunes, Per Christopher
Foster, and Wendy Ran also provided copy editing.
Ellen van Bork
Rijksmuseum, The Netherlands
David Watkinson
Cardiff University, United Kingdom
Cátia Viegas Wesołowska

National Museum of Gdansk, Poland
ICOM-CC | METAL 2022 | HELSINKI, FINLAND xiii
Introduction
For nearly three decades, the Interim Meeting of the the conference on Suomenlinna, a UNESCO World
Metals Working Group has served as a network to keep Heritage Site located at the entrance to Helsinki’s harbor,
members of our growing community connected, forming added a unique dimension to the whole experience. The
one of the largest Working Groups within ICOM-CC and exceptional social program and the post-conference day
possibly the largest group of metal conservation experts trips made the conference a truly memorable event. I
in the world. This success has only been possible through would also like to thank our hosts for their determined
the determination, hard work, and stamina of generations fund-raising efforts, which resulted in 22 travel grants
of colleagues and amazing hosting institutions. for presenting authors.
I extend my heartfelt thanks to our conference hosts Like previous Metals Working Group conferences,
for taking on the task of organizing an in-person and Metal 2022 is the result of a vast amount of unseen work
virtual meeting, and staying the course, through such a and preparation. I am particularly grateful for a great
difficult time. Bringing Metal 2022 to the Nordic Coun- team of Assistant Coordinators and thank Jerrad Alex-
tries has been a wonderful experience. The National ander, Valentin Boissonnas, Elodie Guilminot, Nicola
Museum of Finland and Metropolia University of Applied
Emmerson, and Vandana Singh for co-leadership and
Sciences surmounted enormous challenges during an
wisdom at much-needed times.
unprecedented world pandemic and economic downturn.
Navigating through a world of unknowns prompted the Finally, I would like to thank the former Coordinator of
organizing committee to offer a hybrid meeting with both the Metals Working Group, Claudia Chemello, for her
safe in-person and virtual attendance options—a first in guidance during this Triennium. None of this work would
the history of the Metals Working Group. Organizing have been possible without her unconditional support.
Paul Mardikian
Coordinator, ICOM-CC Metals Working Group
xiv ICOM-CC | METAL 2022 | HELSINKI, FINLAND
In memoriam
David Hallam (1956–2020)
This volume is dedicated to our dear friend, colleague, mentor, and teacher David Leslie Hallam,
who left us for another great journey on February 1, 2020. Probably in one of his old Volvos.
We remember and celebrate David’s curiosity, originality, great sense of humor, cross-disciplinary
thinking, and his diverse work in our field, particularly at the Metals Working Group Interim
Meetings, when we come together to celebrate the field’s many achievements and connect with
our friends and colleagues.
Cheers, mate. We miss you.
“Tell me, I forget. Show me, I remember. Involve me, I understand.”

SUSTAINABILITY IN PRACTICE
Natural Conservation of Archaeological Iron
Henning Matthiesen* Lars Brock Andersen Philippe Dillmann Trine Wiinblad

National Museum of Denmark Centre for Preservation of Cultural LAPA NIMBE – IRAMAT, CEA/ National Museum of Denmark
Lyngby, Denmark Heritage CNRS/U P Saclay, UMR 3685 Lyngby, Denmark
[email protected] Vejle, Denmark Gif-sur-Yvette, France [email protected]
[email protected] [email protected]
Jacob Frydendahl *Author for correspondence
Royal Danish Academy – Charlotte Carré Delphine Neff
Conservation LAPA NIMBE – IRAMAT, CEA/ LAPA NIMBE – IRAMAT, CEA/
Copenhagen, Denmark CNRS/U P Saclay, UMR 3685 CNRS/U P Saclay, UMR 3685
[email protected] Gif-sur-Yvette, France Gif-sur-Yvette, France
Abstract
The study aimed to test whether reburial of chloride- over 1.5 years. Their corrosion products were characterised
containing iron artefacts in a natural protective environment and compared to untreated reference samples. Desalination
can be used for the desalination of these objects. This could efficiency was compared to burial in a peat microcosm in
constitute an alternative procedure for bulk treatment of the conservation laboratory, as well as to a more traditional
the large amounts of archaeological iron that is currently desalination treatment (deaerated NaOH). Analysis by Raman
discarded or stored without treatment and/or environmental spectroscopy showed corrosion products were beginning to
control. An anoxic, iron-reducing, pH-neutral waterlogged peat transform into carbonated phases, especially along cracks,
environment was used that was rich in organic matter and had where they also seemed to give the objects increased physical
low chloride and high carbonate content. Thermodynamics stability. Scanning electron microscopy revealed that the
predict that Cl−-containing Fe(II) compounds (such as ferrous spatial distribution of Cl− was heterogeneous in both treated
hydroxychloride) will, over time, transform into carbonates and untreated samples, while digestion experiments suggested
in this environment, while iron-reducing bacteria in the peat a lower total Cl− content for the treated samples. The results
are expected to reduce reactive Fe(III) compounds (including indicate that reburial has potential for natural conservation,
akaganeite) in the corrosion layers. In addition, anoxic and but that the process can take a long time.
pH-neutral conditions are expected to stop active corrosion
during the reburial period. Newly excavated iron artefacts Keywords
from a terrestrial chloride-rich environment were reburied in archaeological iron, desalination, reburial, iron-reducing
waterlogged peat and samples were recovered at intervals bacteria, natural conservation, in situ treatment
Introduction
Chloride-containing iron artefacts cause severe problems extraction in alkaline solutions. None of these methods
in museums. The chloride (Cl−) can be tightly embedded have been quantitatively proven to be 100% efficient at
in the thick corrosion layers of iron objects and cause removing Cl− and are often labour intensive, ill-adapted
the objects to crack and disintegrate upon excavation to mass treatment and may cause irreversible changes
(Selwyn et al. 1999). If Cl− is not removed, it is necessary to the objects.
to store artefacts in either very dry conditions (< 12% Cl− is mainly found either as a counter ion or within
RH; Watkinson and Lewis 2005) or oxygen-free envi- Cl−-bearing compounds at the interface between the
ronments (Paterakis and Hickey-Friedman 2011), both metal core and the corrosion products. The release of
of which are expensive and may harm their ability to be Cl− is determined by how it is bound. For example, the
exhibited. Alternatively, attempts may be made to remove mineral akaganeite (β-FeOOH) that is formed under
the Cl− using electrochemical techniques, annealing in an aerated conditions can contain both lattice-bound and
inert atmosphere, hydrogen plasma, subcritical fluid or surface-bound Cl− with different properties (Watkinson
NATURAL CONSERVATION OF ARCHAEOLOGICAL IRON 17
and Emmerson 2017), while ferrous hydroxychloride compared with state-of-the-art Cl − removal under
(β-Fe2(OH)3Cl) that is formed under anoxic conditions oxygen-free, alkaline conditions in the conservation
can release Cl− upon oxidation or dissolution during workshop (Kergourlay et al. 2018).
storage and conservation of objects (Reguer et al. 2007,
Kergourlay et al. 2018). Potentially, both phases are able Samples and methods
to release Cl− depending on the conditions and their
The study was carried out on archaeological material from
stability, and ideally both phases should be transformed
Stavsager in Western Jutland, Denmark (N 55°21.179’ E
during conservation into other more stable phases that
008°56.387’) where the soil is sandy and normally rich in
do not contain Cl−.
Cl− due to proximity to the North Sea. A total of 13 lances
It is well established that soil contains a large group of measuring 14–20 cm in length and dating from approx-
iron-reducing bacteria that under anoxic conditions imately AD 200–500 (Grundvad 2020) were used. Four
reduce iron(III) minerals to iron(II) in order to oxidise of these were divided into ca. 1.5 cm long subsamples
organic material and obtain energy for their metabolism using a diamond saw in order to have sufficient replicate
(Kirk 2004). During preliminary experiments, archaeo- samples for the study. Both the subsamples and whole
logical iron artefacts covered by thick layers of iron(III) lances were treated for either 0.5, 1 or 1.5 years, and the
oxide corrosion products were reburied in a waterlogged results were compared to the untreated reference samples.
peat, and after three years the main phase in the corrosion The treatments included reburial in a wetland, burial
layer was iron(II) carbonate (Saheb et al. 2011). Similar in waterlogged peat in the laboratory and treatment in
effects using pure strains of iron-reducing bacteria in the anoxic NaOH (Table 1). Due to space restrictions, only
laboratory have been recorded (Comensoli et al. 2017, selected results are presented here that focus on the
Kooli et al. 2018). samples treated for 1.5 years. Some of the samples (four
of the lances and several subsamples) are still buried for
This study investigated a new low-tech and low-cost
longer-term examination.
approach to remove Cl− from corrosion layers in archaeo-
logical iron within a natural protective environment. The Reburial took place in waterlogged peat at Nydam
low-tech approach is based on reburying the artefacts Mose in Southern Jutland, Denmark (N 54°57.637’
in a waterlogged freshwater peat that is organic-rich, E 009°43.657’). Nydam is an archaeological site where
oxygen-free, pH-neutral, carbonate-rich, low in Cl− and large amounts of weapons and personal equipment were
with a natural content of iron-reducing bacteria. Our deposited between AD 250–500. Parts of the site were
hypothesis was that the artefacts would be preserved excavated from 1989 to 1997, and the environmental
in this environment, that the new thermodynamic conditions were investigated and monitored from 1997
conditions and natural iron-reducing bacteria would, to 2021 to preserve the remaining unexcavated artefacts
over time, transform the corrosion products into more in situ (Matthiesen et al. 2004 and 2020). Reburial of
stable compounds and that Cl− would slowly be removed three lances and nine subsamples was carried out in
through diffusion. The results of the experiment were a backfilled pit from the former excavations. Samples
Table 1. Overview of the number of samples and subsamples used in each treatment and for the different analyses. ‘Total number’ shows how many
samples were treated, ‘Retrieved after 0.5/1/1.5 year(s)’ shows how many were retrieved at the different time intervals and ‘Analyses’ shows how many
samples were analysed by the different methods. The remainder of the samples are either under continued treatment or stored for future analysis
Total Retrieved Retrieved Retrieved

Treatment Type Analyses
number 0.5 year 1 year 1.5 years
NaOH Subsample 8 2 2 4 SEM & Raman (2), or total-Cl (1)
NaOH Whole lance 1 0 0 1 Cleaning (1)
Reburied Subsample 9 2 2 2 SEM & Raman (2), or total-Cl (1)
Reburied Whole lance 3 0 0 1 Cleaning (1)
Peat, lab Subsample 11 2 2 2 SEM & Raman (2), or total-Cl (1)
Peat, lab Whole lance 3 0 0 1 Cleaning (1)
Reference Subsample 9 2 0 3 SEM & Raman (3), or total-Cl (2)
Reference Whole lance 2 0 0 2 Cleaning (2)
18 ICOM-CC | METAL 2022 | HELSINKI, FINLAND SUSTAINABILITY IN PRACTICE
were placed in large-mesh polyester nets that could be Treatment of one lance and eight subsamples in indi-
pressed down into pre-drilled holes (3 cm in diameter) vidual baths of 0.5 M NaOH (analytical grade) was
to a depth of 0.8 m below the surface of the soil. The nets carried out in a nitrogen atmosphere in a custom-made
were attached to a polyester string (1.5 mm in diameter) chamber (Frydendahl et al. 2022). Anoxic conditions
to the soil surface to allow easy retrieval of the samples were verified by oxygen sensors (PSt3 from PreSense).
(Figure 1). The pre-drilled holes were backfilled with Treatment solutions were changed after 1, 2, 3, 4, 6 and
peat whose spongy nature created a good seal. On site, 10 months and analysed for extracted Cl− using ion
the water level and temperature within the peat were chromatography (IC) (Metrohm 881 Compact IC pro
monitored continuously using a Van Essen TD-Diver and ion chromatograph equipped with a Metrosep A supp
during site visits oxygen concentration, redox potential 16 250/4.0 column with ion suppression).
and pH were measured directly in the soil at the depth of Two untreated reference lances and nine untreated
the samples using the methods described in Matthiesen et reference subsamples were stored under anoxic and dry
al. (2004 and 2013). Samples of soil water were retrieved conditions until analysis as described above.
and analysed for alkalinity, dissolved iron, sulfate and
Subsamples from each treatment as well as untreated
Cl− as described in Matthiesen et al. (2004). Buried iron
reference samples were embedded in epoxy resin, cut
samples were retrieved with two subsamples after 0.5, 1
using a diamond saw and ground using 320, 800, 1200
and 1.5 years and a whole lance after 1.5 years. They were
and 4000 grit SiC grinding paper with absolute ethanol
pulled up using the attached string so no digging was
as a coolant and lubricant. They were then polished on
required. After retrieval, the samples were transported to
velvet pads using a diamond spray (3 μm and 1 μm).
the laboratory in absolute ethanol where they were dried
The cross-sections were analysed by scanning electron
for a few minutes in warm air and kept under anoxic microscopy (SEM) coupled with energy-dispersive x-ray
and dry conditions (in Escal oxygen barrier film with spectroscopy (EDS) (JEOL 7001F, acceleration voltage of
nitrogen atmosphere and dried silica gel) until analysed. 15 kV, Cl detection limit of ca. 0.5 mass%) and Raman
spectroscopy (inVia Renishaw, 532 nm, 1 µm spot size,
100 µW). Distinction was made between the dense
product layer (DPL, the inner corrosion products that
do not contain markers of the soil such as quartz grains)
and the transformed medium (TM, the outer part of
the corrosion system that is a mix between corrosion
products and markers from the soil), as described in Neff
et al. (2005). Cl content in the DPL was determined by
EDS (Aztec software) for selected zones excluding the
signal from metal, resin and TM. Raman spectroscopy
was used on both DPL and TM. Other subsamples
(both treated and untreated) were dissolved in 4.5 M
sulfuric acid (analytical grade). During dissolution, the
containers were covered with PE film (parafilm) and
lidded, but loosely. After dissolution, the solutions were
Figure 1. Image of net with iron subsample for reburial. The net was neutralised using NaOH (analytical grade), filtered and
attached by the white string to the surface of the soil to allow easy the Cl− content was quantified using IC as described
retrieval of the sample above (modified from Rimmer et al. 2012).
Three lances and 11 subsamples were placed in a micro- After 1.5 years of treatment, two whole lances (one from
cosm constituting a closed system of 100 L of waterlogged Nydam and one from the peat in the laboratory) were
Nydam peat in the laboratory, and the same analyses as cleaned to the corrosion layer representing the original
described for Nydam were carried out. The microcosm surface by scalpel and air abrasion before being coated
was briefly opened for recovery of samples at the same with microcrystalline wax (Mobil Kote 420). The NaOH-
time intervals as those at Nydam. treated lance was impregnated with wax before being
cleaned due to its fragility. Cleaning was carried out

by two conservators working in parallel on each lance,
discussing and noting hardness, tendency to flaking/
spalling and ease of locating the original surface of the
artefacts to compile subjective qualitative assessments
of object condition. Two untreated lances were cleaned
for comparison. The subsamples were too small and
their corrosion products too fragile to test mechanical
cleaning.
Thermodynamic modelling of the stability of different

minerals under the environmental conditions prevailing
at Nydam was carried out using The Geochemist’s Figure 2. Water level and soil temperature variations at Nydam. The
Workbench 2021 software version 15.0.0 (available from samples were buried during spring 2019 at a depth of 80 cm, where the
soil temperature was also monitored
www.gwb.com/, accessed on 3 March 2022). The built-in
thermodynamic database was used and supplemented
with data for ferrous hydroxychloride (Nemer et al.
2011), chukanovite (Kim et al. 2017) and akaganeite
(Rémazeilles et al. 2009).
Results
Environmental conditions varied between the different
treatments. For the reburial site at Nydam, Figure 2
shows limited seasonal variation in temperature and
constant waterlogged conditions at the reburial depth.
Other environmental parameters measured during the
four site visits and the results from the laboratory peat Figure 3. Photomicrograph of a cross-section of an untreated reference
sample with four colour EDS maps, where iron is purple and Cl− is blueish.
microcosm and NaOH treatments are reported in Table 2.
Five areas marked by red boxes (no. 1–5) indicate zones where the Cl−
All parameters except Cl− showed a modest variation content of DPL was quantified and Raman measurements were performed
over time. In the peat microcosm, the Cl− content in the (Figure 4). The green line shows the interface between DPL and TM
water samples increased after burial of the archaeological
iron samples, which was not the case at Nydam. In the
objects (Figure 3). The distribution of Cl− within the
NaOH treatment, the solutions were changed at inter-
corrosion layers was extremely heterogeneous and mostly
vals, and only the first bath showed markedly increased
found in small pockets (blueish colour) near the metal
Cl− concentrations.
surface. Some of the zones (marked by red rectangles)
SEM-EDS analysis of a cross-section from an untreated have been further investigated by Raman point measure-
subsample is shown as an illustrative example of the iron ments to identify corrosion products (Figure 4), and the
Table 2. Comparison between the three treatment environments. Values given as mean ± standard deviation for four site visits (except temperature
at Nydam, which is measured daily). SHE = standard hydrogen electrode; nm = not measured
Reburied Peat, lab NaOH

Temperature (°C) 9.8 ± 1.4 19 ± 1.5 21 ± 1
Oxygen Anoxic Anoxic Anoxic
Redox potential (mV vs SHE) −208 ± 5 −224 ± 7 nm
pH 6.2 ± 0.1 6.5 ± 0.1 13.7
Alkalinity (meq/L) 8.8 ± 0.2 16.3 ± 0.4 nm
Cl− (mg/L) 36 ± 4 62 before sample burial 10–120 in first baths,
89 ± 6 after < 2 in next baths
Fe2+ (mg/L) 6.7 ± 1.0 16 ± 4 nm
Figure 4. Backscatter SEM images of zones 1–5 from an untreated

reference sample (red rectangles in Figure 3) with superimposed
identification of corrosion products by Raman spectroscopy. The Figure 5. Backscatter SEM images of a subsample treated for 1.5 years in
different corrosion products are identified in the key, with blue symbols a peat microcosm in the laboratory with superimposed identification of
for iron oxides (magnetite, maghemite and hematite), yellow for iron corrosion products. No. 1–3 show different areas on the sample. Symbols
hydroxy oxides (goethite and ferrihydrite), red for chlorinated phases
and colours as described in Figure 4
(akaganeite and ferrous hydroxychloride) and green for carbonated
phases (chukanovite and siderite). A few analyses showed fayalite (light
blue circle with a cross), which is not a corrosion product but part of the
slag from iron production. The metallic iron is white, while the black part
of the images consists of sand grains and embedding media
average Cl− content was quantified in the DPL within

Figure 6. Photomicrographs of three different areas with carbonated
the same zones by SEM-EDS mapping.
phases from a subsample treated for 1.5 years at Nydam: (1) is from
Raman analysis revealed heterogeneous and highly the inner corrosion products (DPL), (2) shows a crack in the DPL that
is covered with siderite and (3) shows siderite precipitated next to a
complex corrosion layers, with up to nine different corro- sand grain in the outer corrosion products (TM). Symbols and colours
sion products being identified in the untreated sample as described in Figure 4
(Figure 4). At the interface between metal and corrosion
products, large areas of ferrous hydroxychloride and to
The results from the Raman analyses are summarised in
a lesser extent akaganeite were identified in the chlorin-
Figure 7, which shows a relative abundance of different
ated areas (most distinct in Figure 4.2 and 4.4), while
types of corrosion products. Although over 100 meas-
chukanovite and other compounds were identified in the
urements were carried out for each sample, abundance
non-chlorinated areas. The DPL primarily consisted of
measurements cannot offer exact areal distribution of
magnetite, with goethite and maghemite marbling, while
different corrosion products, as this will depend upon
the outer transformed layer (TM) primarily consisted of
where the operator decides to conduct analyses.
goethite (Figure 4, 1–2). Similar analyses were carried out
on samples from all treatments. The results for samples Measurements of total Cl− (as a percentage of the total
treated in peat are similar to the layout of the untreated weight of the sample) for five digested samples (two
sample but with some distinctively different features:
Figure 5 demonstrates how a sample treated for 1.5 years
in the peat microcosm has a distinct presence of siderite
in the outer corrosion layers (TM) as well as siderite layers
in cracks and exposed parts of the DPL. Figure 6 shows
some more detailed investigations of a sample buried
for 1.5 years at Nydam, which confirms the tendencies Figure 7. Relative abundance of phases identified in the corrosion layers
observed from the peat microcosm, and also shows how after 1.5 years of treatment. The total number of analyses for each sample
the surface of cracks is covered by siderite (Figure 6.2). is shown in parenthesis. The phases are grouped as: blue for iron oxides
(magnetite, maghemite and hematite), yellow for iron hydroxy oxides
The results for samples treated in NaOH are markedly
(goethite and ferrihydrite), red for chlorinated phases (akaganeite and
different, showing the presence of magnetite, goethite ferrous hydroxychloride) and green for carbonated phases (chukanovite
and ferrihydrite only (not shown). and siderite)
untreated reference samples and three treated samples)

are shown in Figure 8a. The results are very uncertain
and recovery tests showed that there may be a loss of
Cl− of up to 56% due to matrix effects and evaporation
during dissolution. Results from mapping Cl− content
by SEM-EDS (as a percentage of the weight of the DPL)
in five zones on each sample (three untreated reference
samples and three treated samples) are compiled in Figure
8b. SEM-EDS focused on the DPL alone, where most Cl
is found; this results in higher Cl content compared to
digestion, as the latter also includes metallic iron and Figure 9. Lance shown before reburial, after reburial for 1.5 years at
Nydam and after final cleaning (with air abrasion)
TM with lower Cl content.
that the original surface was easily recognised. No loss
of corrosion products had occurred during burial in
peat (Figure 9). In NaOH, the corrosion layer became
extremely porous and fragile, requiring impregnation of
the lance with wax before cleaning, which proved difficult
to control and made the original surface hard to identify.
Discussion
Evidence of change and desalination
The study material was very heterogeneous both in terms
of corrosion products (Figure 4) and Cl distribution
(Figure 3), making it challenging to document changes
unless they were very comprehensive.
NaOH treatment caused significant changes to the

samples: the Cl content was lowered (Figure 8), the corro-
sion products consisted almost exclusively of magnetite
and goethite (Figure 7) and the corrosion layers became
porous and fragile (as reported by Selwyn and Argy-
ropoulos (2005) for heavily degraded objects). Increased
porosity can to some extent be caused by transformation
of chlorinated phases to magnetite, which has a low
Figure 8. Measurement of Cl− content after 1.5 years of treatment: volume per iron atom (28–33 cm3/mol Fe for the chlo-
(a) total Cl− (as a percentage of total sample weight) measured in five rinated phases and only 15 cm3/mol Fe for magnetite).
subsamples after dissolution in acid; (b) mapping of Cl− by SEM-EDS
While increased porosity may speed up the removal of
on six other subsamples. The reference boxplot represents 15 areas on
three untreated reference samples, while the other boxplots represent Cl− , it may also produce more fragile corrosion layers.
five areas on one sample for each treatment. Circles show individual
measurements, crosses show mean value, boxes show 1st, 2nd (line)
The changes in the samples treated in peat in situ
and 3rd quartile, and whiskers show minimum and maximum values and in the laboratory were more subtle: Raman anal-
yses identified more carbonated phases and less iron
Treated lances were cleaned to reveal a corrosion layer oxides (Figure 7) and a change in phase distribution
representing the original surface. During this procedure, (Figures 5, 6) when compared to the untreated samples,
it was subjectively noted that the corrosion layers of even if the transformation was by no means complete
objects from reburial and laboratory peat were coherent, after 1.5 years. The transformation from chlorinated
seemed slightly softer than the untreated material and phases (28–33 cm 3/mol Fe) to carbonated phases
(28–29 cm3/mol Fe) is a modest change in volume, so

porosity will not increase as was the case with NaOH
treatment. Local precipitation of carbonates was observed
especially along cracks (Figure 6), which may explain why
the lances treated in peat showed less tendency to flake
than the untreated lances. Total Cl− in samples treated in
peat was lower than in the untreated samples (Figure 8a)
and the Cl− content in the peat microcosm increased after
burial of the samples (Table 2), both indicating some Cl−
removal from the samples. However, desalination was not
complete and there were still zones with high amounts
of Cl− in the DPL in samples both from the microcosm
and from Nydam (Figure 8b). Thus, overall, there were
some indications of changes to the objects and a tendency
towards both desalination and physical stabilisation, but
the changes were slow and it is too early to reach any Figure 10. Pourbaix diagram calculated by The Geochemist’s Workbench
for an environment with 0.12 mM of dissolved iron, alkalinity of
firm conclusions about the removal of Cl− due to the 10 meq/L, 1 mM of Cl− and suppressing hematite. The red dot shows
heterogeneity of the material. the average pH and redox potential measured at Nydam during four
site visits
Burial environment and thermodynamics
ychloride, on the other hand, requires high concentra-
The burial environment at Nydam is very stable with tions of Cl− and Fe2+ to precipitate (Remazeilles et al.
waterlogged, carbonate-rich, Cl −-poor, anoxic and 2009), and thermodynamic modelling indicates that
pH-neutral conditions that are favourable for the pres- it is not stable at the concentrations found in Nydam.
ervation of iron. High carbonate contents in pH-neutral However, the kinetics are slow, which may be explained
conditions indicate high CO2 pressure, and previous
by slow ion transport through compact corrosion layers.
measurements of CO2 pressure have shown up to 0.5 bar
Furthermore, the samples in this study have a relatively
CO2 in the peat (Matthiesen et al. 2004). During this
high content of Fe(II) corrosion products that cannot be
project, the environmental conditions were monitored
utilised by iron-reducing bacteria but have to be trans-
for two years, and the results were similar to values from
formed through abiotic processes such as dissolution and
the last 20 years of monitoring (Matthiesen et al. 2020).
re-precipitation, which can be relatively slow processes.
In the longer term, the presence of very well-preserved
iron objects from AD 200–500 in Nydam show that pres- Safety and accessibility of the objects should also be
ervation conditions for iron have been favourable during considered as far as reburying archaeological iron is
this period. The corrosion products on archaeological concerned, with risk of changes to the environment
objects from Nydam were predominantly siderite, which (such as physical disturbance or changes in water level)
seems to be a long-term stable compound (Matthiesen and risk of theft. At Nydam, these risks are very low, as
et al. 2003). the site is listed. There are strong restrictions with regard
Thermodynamics indicate that corrosion products on the to the water level and use of the site, but this will not be
reburied artefacts will change over time. Figure 10 shows the case everywhere and some monitoring will normally
a Pourbaix diagram for iron adjusted to the environ- be needed. As for accessibility, it is always possible to
mental conditions prevailing at Nydam, and where the pH retrieve the objects, but this is obviously more tedious
and redox potential measured at the site are marked. The than retrieving them from a storage room and therefore
diagram indicates that the system is saturated regarding reburial is not suited for objects that need to be studied
siderite, which means that if Fe2+ is produced during very frequently. Some of these limitations may be solved
transformation or dissolution of corrosion products, it by reburial in peat at the laboratory, where the transfor-
will re-precipitate as siderite, as was indeed observed in mations seem to be slightly faster than in situ (Figure 7),
cracks in the buried artefacts (Figure 6). Ferrous hydrox- but there are also disadvantages to using a closed system,
as Cl− from artefacts will accumulate in the peat, and it is References

not practical for long-term treatment or storage.
Comensoli, L., J. Maillard, M. Albini, F. Sandoz, P. Junier,
and E. Joseph. 2017. Use of bacteria to stabilize archae-
Conclusion
ological iron. Applied and Environmental Microbiology
Reburial as reported here may offer temporary storage 83(9): art. e03478.
and/or possible desalination treatment options. As Frydendahl, J., L.B. Andersen, T. Wiinblad, C. Carré, P.
storage, it can provide a stable and completely anoxic Dillmann, D. Neff, and H. Matthiesen. 2022. A simple
environment that may be compared to current expensive treatment system for the deaerated desalination of
museum storage under either very dry or anoxic condi- archaeological iron. Poster presented at Metal 2022: 10th
tions. In terms of desalination, it is too early to evaluate
Interim Meeting of the ICOM-CC Metals Working Group,
the efficiency of reburial – it is certainly significantly
Helsinki, 5–9 September 2022.
slower than alkaline extraction but may offer other
advantages such as physical stabilisation of the objects, González, K.S. and M. Salvato. 2019. Pictures speak for
sustainability (no use of chemicals) and easy upscaling. themselves: Case studies proving the significance and
affordability of X-ray for archaeological collections. In
Thus, overall, reburial should be considered a supple- New life for archaeological collections, eds. R. Allen and
mentary procedure rather than a replacement for existing B. Ford, 226–44. Lincoln: University of Nebraska Press.
methods. It could, for instance, be used for the large
Grundvad, L. 2020. Jernalderofringer fra Stavsager Høj
amounts of archaeological iron that is currently discarded
after being recorded due to the high costs of conservation ved Fæsted – En foreløbig præsentation af deponeringer
and storage, or even as a storage solution for some of the og kontekster. Arkæologi i Slesvig 18: 119–38.
accessioned but unstable iron artefacts that continue to Kergourlay, F., S. Réguer, D. Neff, E. Foy, F.E. Picca, M.
degrade in museum collections (González and Salvato Saheb, S. Hustache, F. Mirambet, and P. Dillmann. 2018.
2019). Stabilization treatment of cultural heritage artefacts: In
situ monitoring of marine iron objects dechlorinated in
This initial study has only scraped the surface, exam-
alkali solution. Corrosion Science 132: 21–34.
ining principles and tendencies rather than quantitative
effects. The results so far have indicated that reburial and Kim, S., M. Cassandra, M. Nemer, and J.-H. Jang. 2017.
storage in waterlogged peat could be feasible but that Solubility model for ferrous iron hydroxide, hibbin-
transformation of corrosion products and desalination gite, siderite, and chykanovite in high saline solutions
is very slow for the studied artefacts. More studies are of sodium chloride, sodium sulphate, and sodium
needed to reach more firm conclusions. These include carbonate. ACS Earth Space Chemistry 1(10): 647–63.
study of iron artefacts from other sites. For instance, it Kirk, G. 2004. The biogeochemistry of submerged soils.
is likely that artefacts dominated by Fe(III) minerals will Chichester: Wiley.
show faster transformation due to iron-reducing bacteria.
Consequently, removal of Cl− from ferric akaganeite could Kooli, W.M., L. Comensoli, J. Maillard, M. Albini, A. Gelb,
be faster than from ferrous hydroxychloride. Studies on P. Junier, and E. Joseph. 2018. Bacterial iron reduction
the transformation of synthetic corrosion products and and biogenic mineral formation for the stabilisation of
on microbial activity are also needed. corroded iron objects. Scientific Reports 8(764): 1–11.
Matthiesen, H. 2013. Oxygen monitoring in the corrosion
Acknowledgements and preservation of metallic heritage artefacts. In Corro-
sion and conservation of cultural heritage artefacts, eds. P.
We would like to thank Lars Grundvad from Sønder-
Dillmann, D. Watkinson, E. Angelini, and A. Adriaens,
skov Museum and Lasse Sørensen from the National
368–91. Cambridge: Woodhead Publishing Limited.
Museum of Denmark for providing the sample material,
the Danish Cultural Ministry for financing the study Matthiesen, H., L.R. Hilbert, and D. Gregory. 2003.
and the Ingwersenske Fond, William Demant Fonden, Siderite as a corrosion product on archaeological iron
Augustinus Fonden and Elisabeth Munksgaard Fonden from a waterlogged environment. Studies in Conservation
for supporting Jacob Frydendahl’s research visit to LAPA. 48(3): 183–94.
Matthiesen, H., D. Gregory, P. Jensen, and B. Sørensen. Selwyn, L.S. and V. Argyropoulos. 2005. Removal of
2004. Environmental monitoring at Nydam, a waterlogged chloride and iron ions from archaeological wrought iron
site with weapon sacrifices from the Danish Iron age. I: A with sodium hydroxide and ethylenediamine solutions.
comparison of methods used and results from undisturbed Studies in Conservation 50(2): 81–100.
conditions. Journal of Wetland Archaeology 4: 55–74. Watkinson, D. and M.T. Lewis. 2005. Desiccated storage
Matthiesen, H., K. Strætkvern, D. Gregory, and P. Jensen. of chloride-contaminated archaeological iron objects.
2020. Preservation of the Nydam site and finds – In situ Studies in Conservation 50(4): 241–52.
and at the museum. In Excavating Nydam. Archaeology, Watkinson, D. and N.J. Emmerson. 2017. The impact of
palaeoecology and preservation. The National Museum’s aqueous washing on the ability of βFeOOH to corrode
research project 1989–99, eds. S. Holst and P.O. Nielsen, iron. Environmental Science and Pollution Research 24(3):
345–71. Copenhagen: The Royal Society of Northern 2138–49.
Antiquaries.
Neff, D., P. Dillmann, L. Bellot-Gurlet, and G. Beranger. Authors
2005. Corrosion of iron archaeological artefacts in soil: Henning Matthiesen is senior researcher at the National
Characterisation of the corrosion system. Corrosion Museum of Denmark and an honorary professor at the
Science 47: 515–35. Royal Danish Academy – Conservation. He undertakes
Nemer, M.B., Y. Xiong, A.E. Ismail, and J.-H. Jang. 2011. research on the preservation of archaeological remains
Solubility of Fe2(OH)3Cl (pure-iron end-member of both in situ and in museum storage.
hibbingite) in NaCl and Na2SO4 brines. Chemical Geology Jacob Frydendahl has an MSc in conservation from
280: 26–32. the Royal Danish Academy – Conservation where he
Paterakis, A.B. and L. Hickey-Friedman. 2011. Stabi- developed his thesis on the present project. He works
lization of iron artifacts from Kaman-Kalehoyuk: A in the field of dechlorination of archaeological iron and
comparison of chemical and environmental methods. the transformation of corrosion products.
Studies in Conservation 56(3): 179–90. Lars Brock Andersen has a BSc in conservation. He
Reguer, S., P. Dillmann, and F. Mirambet. 2007. Buried has been working for many years in the field of method
iron archaeological artefacts: Corrosion mechanisms development and hands-on conservation of chloride-rich
related to the presence of Cl− containing phases. Corrosion archaeological iron.
Science 49(6): 2726–44. Charlotte Carré obtained her master’s from La Rochelle
Remazeilles, C., D. Neff, F. Kergourlay, E. Foy, E. Conforto, University (France) and her PhD from New-Caledonia
E. Guilminot, S. Reguer, P. Refait, and P. Dillmann. 2009. University. Her current research interests include the
Mechanisms of long-term anaerobic corrosion of iron study of the behaviour of metallic materials depending
archaeological artefacts in seawater. Corrosion Science on their environment.
51: 2932–41. Philippe Dillmann is director of the Laboratoire
Rimmer, M., D. Watkinson, and Q. Wang. 2012. The archéomatériaux et prévision de l’altération (LAPA). He
effiency of chloride extraction from archaeological iron has a PhD in material science and undertakes research
objects using deoxygenated alkaline solutions. Studies in on the long-term alteration of metals and glasses.
Conservation 57(1): 29–41. Delphine Neff is a research scientist at LAPA. She
Saheb, M., F. Marsal, H. Matthiesen, D. Neff, P. Dillmann, undertakes research on understanding the transfor-
and D. Pellegrini. 2011. Fluctuation of redox conditions in mation of ancient corrosion layers during desalination
radioactive waste disposal cell: Characterisation of corro- treatments as well as their interaction with organic
sion layers formed on archaeological analogues. Corrosion compounds applied as protective coatings.
Engineering, Science and Technology 46: 199–204. Trine Wiinblad has an MSc in conservation from the
Selwyn, L.S., P.J. Sirois, and V. Argyropoulos. 1999. The Royal Danish Academy – Conservation. She has been
corrosion of excavated archaeological iron with details working for many years in the field of method devel-
on weeping and akaganéite. Studies in Conservation 44: opment and hands-on conservation of composites and
217–32. chloride-rich archaeological iron.
Exploiting Biologically Synthetized Chelators in
Conservation: Gel-based Bio-cleaning of Corroded Iron
Heritage Objects
Luana Cuvillier Virginie Dupuy
Haute Ecole Arc Conservation-restauration (HE-Arc CR), HES-SO University of Laboratoire Arc’Antique, Grand patrimoine de Loire-Atlantique (GPLA)
Applied Sciences and Arts Western Switzerland Nantes, France
Laboratory of Technologies for Heritage Materials (LATHEMA), Institute of [email protected]
Chemistry, University of Neuchâtel
Neuchâtel, Switzerland Elodie Guilminot
[email protected] Laboratoire Arc’Antique, Grand patrimoine de Loire-Atlantique (GPLA)
Nantes, France
Arianna Passaretti [email protected]
Haute Ecole Arc Conservation-restauration (HE-Arc CR), HES-SO University of
Applied Sciences and Arts Western Switzerland Edith Joseph*
Laboratory of Technologies for Heritage Materials (LATHEMA), Institute of Haute Ecole Arc Conservation-restauration (HE-Arc CR), HES-SO University of
Chemistry, University of Neuchâtel Applied Sciences and Arts Western Switzerland
Neuchâtel, Switzerland Laboratory of Technologies for Heritage Materials (LATHEMA), Institute of
[email protected] Chemistry, University of Neuchâtel
Neuchâtel, Switzerland
Aymeric Raimon [email protected]
Laboratoire Arc’Antique, Grand patrimoine de Loire-Atlantique (GPLA) *Author for correspondence
Nantes, France
[email protected]
Abstract
The complexing agents widely used for the removal of undesired results showed the comparable cleaning efficiency of DFO and
corrosion products from iron artworks can be difficult to control Na2EDTA at neutral pH. However, the DFO-hydrogel performed
and their environmental impact is often overlooked. This paper better on iron rust. A frequent renewal of the treatment
compares the efficiency of deferoxamine B (DFO), a siderophore formulation ensured more effective cleaning. Among the different
synthesised by the bacterium Streptomyces pilosus, with that carriers tested, agar gels applied when hot and gellan gum gels
of disodium ethylenediaminetetraacetic acid (Na2EDTA) in the prepared at room temperature were the most effective, and left
removal of iron corrosion phases. Several bio-derived hydrogels few residues on the treated surfaces. The methodology was then
were exploited as carriers and preliminary tests were carried out tested on altered steel artefacts belonging to heritage institutions
on naturally aged samples. The investigated parameters included in Nantes (France). The encouraging outcomes point the way to
siderophore concentration, treatment duration and potential a more sustainable approach to treating historical iron artworks.
reapplication, as well as a reliable application protocol. Cleaning
efficacy was evaluated using optical microscopy, colourimetry and Keywords
Raman spectroscopy. Iron uptake was monitored in gel matrixes siderophores, iron objects, corrosion, biocleaning, green
through colourimetry and atomic absorption spectroscopy. The chemistry, hydrogels, deferoxamine B
Introduction
Like most metals, iron is vulnerable to corrosion. For either case, the removal of active or unaesthetic corro-
iron artworks, this problem requires unique solutions to sion stains is needed to arrest ongoing damage processes
ensure their long-term preservation (Kooli et al. 2018). and to facilitate the appreciation, readability and use of
For artworks stored in museums without appropriate the object. The cleaning agent most commonly used
environmental control (Monnier et al. 2013), the corro- by conservators is the sodium salt of ethylenediamine-
sion layer that forms is often a combination of several tetraacetic acid (Na2EDTA), which complexes iron at a
iron oxyhydroxides, including goethite (α-FeOOH) and maximum molar ratio of 1:1, according to the pH, and
lepidocrocite (γ-FeOOH) (Monnier et al. 2010 and 2013). turns light yellow after iron complexation (Filipsky et al.
In the case of archaeological iron objects, however, the 2013). However, despite its easy availability and low cost,
corrosion layer is more complex and heterogeneous. In EDTA is classified as an irritant and it must be used with
appropriate caution (Kaurin et al. 2021). In addition, action of the latter is unaltered even at the relatively high
although the quantities used in cultural heritage are temperatures (up to 110°C) needed to prepare the gels
much smaller than in other industrial applications, EDTA (Pawlaczyk and Schroeder 2021). Only naturally derived
is not biodegradable and its environmental impact is a gelling agents were considered here, namely gellan and
worldwide concern (Gluhar et al. 2020). Thus, in this xanthan gums, which are obtained from bacteria, and
study we tested a bio-derived complexing agent in an agar, which is extracted from red seaweed.
effort to develop more sustainable cleaning practices.
The application protocol and case studies described
Specifically, we examined the efficacy of the siderophore below demonstrate the potential of this ground-breaking
deferoxamine B (DFO), produced by the bacterium methodology in terms of its greater efficiency and
Streptomyces pilosus. DFO is a natural iron chelator sustainability compared to Na2EDTA.
whose synthesis is stimulated in iron-deficient environ-
ments (Albelda-Berenguer et al. 2019). DFO and other
siderophores are biodegradable and have a pH close to
neutral, which makes them less harmful to the envi-
ronment and in their applications (Fazary et al. 2016).
Figure 1. Molecular structure or of deferoxamine B (DFO)
DFO has been tested on iron-contaminated wood, paper
and textile artworks, due to its high affinity for iron,
Materials and methods
as evidenced by its stability constant (logβ), a measure
of the interaction strength between two components Corroded iron substrates
forming a complex: 30.4 for DFO (Farkas et al. 1999)
versus 25 for EDTA (Rapti et al. 2021) (Figure 1). DFO Naturally corroded mild steel samples (20 × 30 × 2 mm)
chelates iron at a 1:1 molar ratio and iron binding causes were obtained from a private conservation workshop,
it to turn bright orange-red, allowing its activity to be where over a period of years they underwent atmos-
easily ascertained. pheric corrosion due to uncontrolled indoor conditions.
Prior to their use in the experiments, the steel samples
A further consideration in the development of were degreased with a cotton swab soaked in 70% v/v
DFO-based cleaning methods is the delivery system. ethanol and the corrosion layer was characterised by
Although the use of gels as delivery systems is not as Raman spectroscopy, which revealed both goethite and
widespread in metal conservation as in other sectors of lepidocrocite, common compounds that develop during
art conservation, their acceptance is rapidly increasing indoor iron corrosion.
(Passaretti et al. 2021). The advantages of gels in conser-
vation include: controlled delivery of the embedded
Gelling agent selection
active substances, the need for smaller quantities of
reagents and easy clean-up after treatment, all of which The suitability of several naturally derived hydrogels
contribute to the protection of the treated artefact, the amended with a solution of DFO mesylate salt (Desferal,
operator and the environment. Several polymers, espe- Novartis) was assessed using triplicate steel samples
cially polysaccharides, can be employed in the design of (Table 1). Each gel formulation was applied to the
DFO-based hydrogels as they are compatible with the corroded surface and removed after 10 min using a cotton
neutral pH (6.8) of the DFO solution and the chelating swab soaked in 70% v/v ethanol.
Table 1. Overview of the different gel formulations tested
Xanthan gum Agar Gellan gum

Brand, supplier Vanzan©, CTS AgarArt©, CTS Phytagel™, Sigma Aldrich
Chelator DFO DFO and Na2EDTA DFO
Preparation 5% w/v in H2O and 3% w/v in H2O, heated to 90°C twice 3% w/v in H2O, manual stirring 3% w/v in H2O, heated to 90°C
manual stirring
Application With a spatula at Dripped when still Cooled preformed With a spatula at room Dripped when still hot
room temperature hot rigid foil temperature
Label in the text A B C D E
EXPLOITING BIOLOGICALLY SYNTHETIZED CHELATORS IN CONSERVATION: GEL-BASED BIO-CLEANING OF CORRODED IRON HERITAGE OBJECTS 27
Evaluation of the application parameters historical objects; (ii) a gorget and a helmet belonging to the
personal collection of Raoul de Rochebrune (1849–1924)
Iron chelation was assessed after the immersion of the
and bequeathed to the Dobrée Museum of Nantes (France)
corroded steel samples for 10 min, 30 min, 1 h, 5 h and 24
in 1930, as archaeological objects (Table 2).
h in a solution of DFO at concentrations of 0, 3·10−4, 3·10−3,
3·10−2 and 6·10−2 M. A 3% w/v agar gel amended with the The objects were marked by heterogeneous corrosion
same concentrations of DFO was applied to the naturally phases and conditions, as described in Table 2.
corroded iron samples for 10 min. The application was
Consequently, the treatment aims for the two sets of
repeated five more times, for a total application time of 60
objects differed and had to be specifically defined for
min. In addition, a 3% w/v agar DFO-gel at a concentration
each type (religious artefacts or archaeological objects).
of 6·10−2 M was applied to five iron coupons for 10 min,
For the diocesan calix and candle holder, a continuous
30 min, 1 h, 5 h and 24 h, respectively. For comparison, a
and homogeneous thin red layer, typical of surface oxida-
3% w/v agar gel amended with Na2EDTA (PanReac Appli-
tion and probably consisting of goethite, was seen. The
chem) was tested using the same reiterative application
metallic surface was well preserved, and a black patina was
protocol. All tests were carried out on triplicate samples.
sporadically present. The goal was to reduce the reddish
corrosion layer until the metallic surface by means of gentle
Case studies
chemical cleaning, in order to limit metal brightness and
Four iron-based artefacts were selected: (i) a calyx and preserve the black patina. Based on mock-up results, the
a candle holder from the Diocese of Nantes (France), as religious artworks received two 15-min applications of 3%
Table 2. Overview of the artefacts treated
Material Description Picture

Calyx Steel
1 – atmospheric corrosion layer

2 – black patina
3 – metallic surface
Candle Holder Cast iron
Gorget Steel
(internal side)
1 – dirt and sediments

2 – corrosion products
3 – original surface
4 – metallic surface
Helmet Steel
(internal side)
1 – external elements (pebbles)

2 – red corrosion products
3 – dense corrosion crust
4 – original surface
5 – metallic surface, visible in the crevices
w/v agar gels amended either with DFO or EDTA (both where L*1, a*1 and b*1 are the coordinates in the colouri-
at a concentration of 6·10-2 M); the application of agar gel metric space of the first measurement, and L*2, a*2, b*2
to three distinct areas served as a control. those of the second measurement. For each analysed
area, measurements were performed in triplicate.
For the gorget and helmet, the lack of conservation of
the original surface can be explained by the fact that, Optical microscopy images were acquired using a Leica
like most of the collection, these archaeological objects Digital microscope system (DMS1000) with Leica Appli-
were probably discovered in a fluvial environment, which cation Suite software.
would account for the stripped aspect of both the gorget
A double-beam atomic absorption spectrometer
and helmet. Heterogeneous crevices or pit corrosion were
(iCE 3300, Thermo Fisher) equipped with an acety-
seen on the helmet and the strong areas of abrasion on
lene-air flame was used to determine the quantity of iron
its outer parts indicated that it had been cleaned in the
ions present after treatment. The gels were dissolved in 10
past using chemical and/or mechanical means, typical
mL of 70% v/v HNO3 and diluted with deionised water to
of the late 19th century. The black patina on the helmet
100 mL. The solutions were not filtered, in order to also
and gorget was present only sporadically, with its loss allow a determination of the amount of corrosion that
probably the result of uncontrolled cleaning. Further could be mechanically scraped from the surface during
loss would lead to the undesired exposure of the metallic gel removal. The analyses were performed in triplicate.
surface. As the dense red corrosion layer visible on the
crevices served as historical evidence, its retention was Results and discussion
desirable. Three different areas on the gorget and inte-
rior aspect of the helmet were treated. Due to the thick Gelling agent selection
corrosion present on both objects, the overall cleaning Based on visual observations, DFO-agar hot gel (B)
consisted of the application of 3% w/v agar gels amended and DFO-gellan gel prepared at room temperature (D)
either with DFO or EDTA (both at a concentration of showed the best performances in terms of corrosion
6·10−2 M) for 3× 20 min. The control consisted of plain removal (Figure 2a, 2b). The DFO-gellan hot gel (E)
agar gel applied following the same protocol. In the achieved mild cleaning following the same protocol
case of the gorget, due to the variable thickness of the (Figure 2c). The performances of the DFO-xanthan gel
corrosion, thinner corrosion areas were cleaned first, prepared and applied at room temperature (A) and the
which softened the thicker areas and allowed their later rigid DFO-agar gel (C) were poor (Figure 2d, 2e).
removal using a scalpel.
The L*a*b* coordinates of the iron substrates before and
Equipment after cleaning confirmed the visual observations. After
cleaning with DFO-loaded hot agar (B), room-temper-
A Jobin Yvon Horiba T640000 Raman spectrometer ature gellan (D) and to a lesser extent hot gellan (E), the
equipped with a microscope (50× magnification) was surface colour became brighter (higher values of L*) and
used to examine the helmet and gorget. Spectra were the hue became blue-green (higher values of a* and b*).
recorded in the spectral range of 100–1500 cm−1 at This can be attributed to the removal of the red-orangish
0.1 mW using a 633 nm laser, with four accumulations rust compounds. By contrast, few visual modifications
of 30 s each. The data were collected using Spectragryph were seen on the samples treated with DFO-amended gels
software. based on xanthan gum (A) or rigid agar (C). After xanthan
A Minolta CM-508D spectrophotometer was used on the gum treatment, a thin layer of gel remained that could not
iron coupons. The setup was as follows: specular compo- be removed by swab rinsing. For the rigid DFO-agar gel
(C), the lack of visible results was due to the failure of the
nent excluded (SCE), illuminant D65, d/8° geometry,
agar to adhere to or form a close contact with the metal
10° observer, window size 8 mm, CIELab colour space.
surface (Guilminot et al. 2019). Due to its good perfor-
ΔE was calculated using the standard colour variation
mance further tests were carried out using 3% w/v agar gel
formula in CIELab:
amended with DFO and applied when still hot, at ~90°C
(B). Despite its good cleaning potential, DFO-gellan gum
at room temperature was not further tested because its
viscous texture implied the need for a thorough rinsing molecules had complexed with a ferric ion molecule on
after removal, especially when applied to carvings. the basis of a 1:1 molar reaction. At a concentration of
6·10−2 M, the best-performing DFO solution was the 3%
Evaluation of the application parameters w/v hot agar gel (B). The decrease in the reaction rate
Tests with the corroded iron samples immersed in after 1 h of treatment (Figure 3b) can be explained by
DFO at concentrations of 0, 3·10−4, 3·10−3, 3·10−2 and the drying of the gel matrix, which made it more rigid,
6·10 −2 M were performed to better understand the thus hindering interaction of the iron ions with the
chelating kinetics of DFO on iron corroded substrates. chelator (Zaitoun and Lin 1997) as well as contact with
The pH of the different DFO solutions was 7.51, 7.68, the object’s surface. AAS measurements showed that the
7.16, 6.79 and 6.65, respectively. The atomic absorption total iron concentration was higher when the formulation
spectrometry (AAS) analysis showed that DFO was was applied 6× 10 min rather than as a single 60 min
effective in the complexation of iron at concentrations application. According to this result, frequent treatment
starting from 3·10−2 M (Figure 3a). Some iron ions were reiteration and gel replacement provide a more efficient
detected in the solution with no siderophore, attributable cleaning action (Table 3). This finding was consistent
to a partial dissolution of the powdery corrosion layer. with the decrease in the complexation reaction over
After 5 h of immersion, the quantity of iron detected time, as the first 10 min of treatment yielded the highest
in solution reached a plateau, indicating that all DFO complexation rates (Figure 3a).
Figure 2. Iron samples treated for 10 min with

6·10−2 M DFO-amended gels: (a) hot (~90°C)
3% w/v agar gel (B); (b) gellan gum at room
temperature (D); (c) gellan gum at 90°C (E);
(d) rigid agar gel (C); (e) xanthan gum (A).
(f) L* (bright/dark) and colour coordinates a*
(red/green) and b* (yellow/blue) coordinates
of the iron coupons before (circle) and after
(triangles) gel application; the standard errors
are also shown (A-dark blue, B-yellow, C-grey,
D-light blue, E-orange). Scale bar: 5 mm

Figure 3. (a) Quantity of iron ions (mg/L) measured in solution at different times after the immersion of the corroded iron samples in DFO at
concentrations of 0 (light blue), 3·10−4 (red), 3·10−3 (grey) 3·10−2 (yellow), 6·10−2 (dark blue) M. (b) Quantity of iron ions (mg/L) measured in the gel
after the application of 6·10−2 M DFO-amended gel to the corroded iron coupons for 0 min, 10 min, 30 min, 1 h, 5 h and 24 h
After a single 60 min application of 3% w/v agar gel action. However, no direct correlation could be drawn
amended with 6·10−2 M DFO, weathering pits were between the quantity of extracted iron and the visual
still visible whereas a more homogeneous surface was appearance of the gel during treatment. Indeed, the
obtained when the same gel was applied for 6× 10 min. hues of the agar gel amended with 3·10−2 M and 6·10−2
This was confirmed by colourimetric measurements, in M DFO were similar despite the greater complexation
which multiple treatment reiteration resulted in a higher action of the higher concentration, as determined by
colour variation than achieved with the same formulation AAS (Figure 4b).
applied as continuous treatment (Table 3).
The results obtained with 3% w/v agar gel amended with
The complexation reaction caused the uncoloured 6∙10−2 M Na2EDTA, also applied to the iron corroded
gel to turn vivid red-orange (Figure 4a). This feature samples for 6× 10 min, were similar to those obtained
allows the operator to monitor and assess the cleaning with DFO-amended 3% w/v agar gel in terms of the
Table 3. Summary of the colour variation of the treated samples before and after cleaning (ΔE), the amount of iron extracted using 3% w/v agar
gels amended with DFO or EDTA at 6·10−2 M, and micrographs of the samples’ surfaces after the intervention
Extracted [Fe] mg/L

Mean ΔE of triplicates Optical microscope picture (0.75×)
per g of gel
Untreated sample n/a n/a
6·10−2 M - DFO-amended gel; 17.76 ± 3.43 19.26 ± 0.78

1× 60 min
6·10−2 M -DFO-amended gel; 21.22 ± 0.68 37.86 ± 6.68

6× 10 min
6·10−2 M EDTA-amended gel; 16.15 ± 2.97 41.41 ± 2.57

6× 10 min
Figure 4. (a) Different concentrations of DFO-amended

3% w/v agar gels applied to the corroded iron
coupons for 10 min. (b) Graphical representation of
ΔE and the extracted iron concentration before and
after treatment with DFO-amended 3% w/v agar gels
visual appearance of the coupons after cleaning and iron gorget and helmet had been buried. Typical iron indoor
uptake. However, in the EDTA-amended gel the colour corrosion compounds, such as lepidocrocite and goethite,
variation before and after treatment was weaker (Table 3). were still found in areas after treatment with either DFO-
or EDTA-amended 3% w/v agar gels, but the proportion
Treatment outcome in the case studies in the former was smaller (Criado et al. 2015).
Visual observations of the religious artefacts (calyx AAS of the gels was performed to evaluate the amount
and candle holder) indicated efficient cleaning with of iron extracted using the different formulations.
DFO-amended 3% w/v agar gel (Figure 5a, 5b). The
removal of the corrosion products revealed the under-
lying metal and the objects’ details. On the gorget and
helmet, the gel was less effective in the removal of the
sediment layer but a controlled cleaning was nonetheless
achieved (Figure 5c, 5d, green rectangles). EDTA treat-
ment, however, was aggressive and both the cleaning
process and the achievement of the desired degree of
cleaning were difficult (Figure 5c, 5d, red rectangles). In
both armour pieces, the presence of sediments implied
the presence of calcium-based compounds. Indeed,
unlike EDTA, which readily binds iron and calcium
(logβFe3+-EDTA = 25 versus logβCa2+-EDTA = 10.9) (Rapti et al.
2021), DFO is specific for iron and its affinity for calcium
ions is poor (logβFe3+-DFO = 30.4 versus logβCa2+-DFO ≤ 3.03)
(Farkas et al. 1999). Hence, the use of a DFO-amended
gel resulted in an appropriate surface appearance from
a conservator’s perspective, as the heterogeneous corro-
sion layers were preserved, whereas the use of Na2EDTA
resulted in the removal of the sediment layer, thus
revealing the metallic surface.
Post-treatment Raman analyses confirmed this finding

and detected calcium-based compounds (e.g. CaSO4, Figure 5. Photographs of the treated zones of (a) calyx, (b) candle
holder, (c) gorget and (d) helmet before (left) and after (right) cleaning.
CaCO3) on the zones treated with DFO but not with Red-delimited zones were treated with EDTA-amended agar gel, green-
EDTA (Figure 6) (Antunes et al. 2014). The calci- delimited zones with DFO-amended agar gel, and blue-delimited zones
um-based compounds derived from the soil in which the with pure water agar gel
Figure 6. Raman spectra (633 nm, 0.1 mW,

4 × 30 s, 50×) of the treated objects. The main
Raman signals are reported for each spectrum
displayed: (a) goethite, (b) lepidocrocite,
(c) anhydrite and (d) calcite
The spectra showed that iron extraction using the formulation, made from renewable sources, is versatile
DFO-amended gel was hindered by the presence of an in its applications and does not require specific storage
upper sediment layer of calcium, especially in the case conditions. These features should facilitate its wide
of the gorget (Table 4). This specific affinity of DFO adoption by conservators. Optimal cleaning with the
for iron supports the use our newly developed DFO DFO-amended agar gel was achieved when it was applied
gel formulation as a suitable and reliable agent for the hot, at a concentration not lower than 3∙10−2 M and with
cleaning of historical iron-based indoor objects. When frequent reiteration of the treatment, which also allows
the corrosion layer is more complex, such as when it better control of the treatment by the operator.
includes sediments, a preliminary mechanical cleaning
It is worth mentioning that the application parameters
or renewal of the application of DFO-amended gel may
and protocol will depend on the nature of the objects to
be effective. Although less sustainable, EDTA offers
be cleaned, therefore our results should be interpreted as
the benefit of tackling sediments as well as corrosion
general guidelines. DFO-amended agar gel may be more
products in one step. The choice of DFO versus EDTA
effective when applied to objects with slight corrosion,
thus depends on the balance between the ease of use
such as indoor historical artefacts, rather than archae-
and sustainability.
ological items heavily corroded and with the potential
Table 4. Quantity (mg/L) of iron ions detected in the gels after 20 min presence of residual sediments. Nonetheless, as an
of treatment of the Dobrée Museum objects eco-friendly, water-soluble, agar-based cleaning agent,
Gorget Helmet DFO is likely to be suitable for a broad range of objects,
DFO 0.16 5.45 with the potential for widespread adoption by conserva-
EDTA 8.73 9.95 tors. Further research will be carried out to optimise the
Water-based/plain gels Below limit of detection 0.19 application protocol for different degradation features.
Environmental and economic aspects should also be
Conclusion considered, from the production of the DFO-agar agent
to its final use (time of application, reiteration, etc.), with
This study demonstrated the potential of a sustainable
comparisons to Na2EDTA.
alternative to traditional complexing agents for use in
metal conservation, to avoid the health and environ-
Acknowledgements
mental concerns associated with, for example, the use
of EDTA. Cleaning gels amended with the siderophore This research was funded by the Swiss National Science
DFO allowed the removal of iron corrosion from natu- Foundation (SNSF), grant number 205121_188755,
rally altered samples with a performance similar to that HELIX (Investigating metal bioremediation for the pres-
of Na2EDTA. Moreover, the use of DFO at neutral pH ervation of historical metal artworks) project 2020–2024.
avoids the need for post-treatment rinsing, in contrast to The authors would like to acknowledge Jean-Yves
EDTA solutions, which have an acidic pH. Our innovative Mevellec from Institut des Matériaux de Nantes and Nath-
alie Guichard from the laboratory Chimie et Interdisci- Committee for Conservation and Haute Ecole Arc
plinarité, Synthèse, Analyse, Modélisation (CEISAM) Conservation-restauration.
for their contribution to the Raman spectroscopy and Kaurin, A., S. Gluhar, I. Maček, D. Kastelec, and D. Lestan.
AAS analyses. We are grateful to Michel Chaillou from 2021. Demonstrational gardens with EDTA-washed soil.
Nantes Diocese for the provision of objects. Part II: Soil quality assessment using biological indicators.
Science of the Total Environment 792: art. 148522.
References
Kooli, W.M., L. Comensoli, J. Maillard, M. Albini., A.
Albelda-Berenguer, M., M. Monachon, and E. Joseph. Gelb, P. Junier, and E. Joseph. 2018. Bacterial iron reduc-
2019. Siderophores: From natural roles to potential tion and biogenic mineral formation for the stabilisation
applications. In Advances in applied microbiology, eds. G. of corroded iron objects. Scientific Reports 8(1): 1–11.
Gadd and S. Sariaslani, 1st ed., vol. 106, ch. 5. Cambridge:
Monnier, J., L. Bellot-Gurlet, D. Baron, D. Neff, I. Guillot,
Academic Press.
and P. Dillmann. 2010. A methodology for Raman
Antunes, V., A. Candeias, M.J. Oliveira, S. Longelin, V. structural quantification imaging and its application to
Serrão, A.I. Seruya, J. Coroado, L. Dias, J. Mirão, and M.L. iron indoor atmospheric corrosion products. Journal of
Carvalho. 2014. Characterization of gypsum and anhy- Raman Spectroscopy 42: 773–81.
drite ground layers in 15th and 16th centuries Portuguese
Monnier, J., I. Guillot, L. Legrand, and P. Dillmann.
paintings by Raman spectroscopy and other techniques.
2013. Reactivity studies of atmospheric corrosion of
Journal of Raman Spectroscopy 45(11–12): 1026–33.
heritage iron artefacts. In Corrosion and conservation of
Criado, M., S. Martínez-Ramirez, and J.M. Bastidas. cultural heritage metallic artefacts, eds. P. Dillmann, D.
2015. A Raman spectroscopy study of steel corrosion Watkinson, E. Angelini, and A. Adriaens, ch. 14. 285–310.
products in activated fly ash mortar containing chlorides. Cambridge: Woodhead Publishing.
Construction and Building Materials 96: 383–90. Passaretti, A., L. Cuvillier, G. Sciutto, E. Guilminot, and
Farkas, E., É.A. Enyedy, and H. Csóka. 1999. A compar- E. Joseph. 2021. Biologically derived gels for the cleaning
ison between the chelating properties of some dihy- of historical and artistic metal heritage. Applied Sciences
droxamic acids, desferrioxamine B and acetohydroxamic 11(8): art. 3405.
acid. Polyhedron 18(18): 2391–98. Pawlaczyk, M. and G. Schroeder. 2021. Deferoxam-
Fazary, A.E., A.S. Al-Shihri, M.Y. Alfaifi, K.A. Saleh, M.A. ine-modified hybrid materials for direct chelation of
Alshehri, S.E.I. Elbehairi, and Y.-H. Ju. 2016. Microbial Fe(III) ions from aqueous solutions and indication of
production of four biodegradable siderophores under the competitiveness of in vitro complexing toward a
submerged fermentation. International Journal of Biolog- biological system. ACS Omega 6(23): 15168–81.
ical Macromolecules 88: 527–41. Rapti, S., S.C. Boyatzis, S. Rivers, and A. Pournou. 2021.
Filipsky, T., M. Ríha, R. Hrdina, K. Vávrová, and P. Siderophores and their applications in wood, textile, and
Mladěnka. 2013. Mathematical calculations of iron paper conservation. In Microorganisms in the deteriora-
complex stoichiometry by direct UV-Vis spectropho- tion and preservation of cultural heritage, ed. E. Joseph,
tometry. Bioorganic Chemistry 49: 1–8. 301–39. Cham: Springer International Publishing.
Gluhar, S., A. Kaurin, and D. Lestan. 2020. Soil washing Zaitoun, M.A. and C.T. Lin. 1997. Chelating behavior
with biodegradable chelating agents and EDTA: Tech- between metal ions and EDTA in sol–gel matrix. Journal
nological feasibility, remediation efficiency and envi- of Physical Chemistry B 101(10): 1857–60.
ronmental sustainability. Chemosphere 257: art. 127226.
Authors
Guilminot, E., A. Gomez, A. Raimon, and M. Leroux.
2019. Use of gels for the treatment of metals. In Metal Edith Joseph is an assistant professor at both the
2019: Proceedings of the Interim Meeting of the ICOM-CC University of Neuchâtel and the Haute Ecole Arc Conser-
Metals Working Group, Neuchâtel, 2–6 September 2019, vation-restauration. She holds a PhD in environmental
eds. C. Chemello, L. Brambilla, and E. Joseph, poster and heritage chemistry from the University of Bologna,
session, 473. Paris: International Councils of Museums– Italy. Her main research activities are the application
of green chemistry for the preservation of artistic and

archaeological objects. The characterisation of hetero-
geneous matrixes and the interaction between organic
substances and inorganic compounds, in particular
microorganisms and metals, are some of her research
interests. She has published around 60 papers on analyt-
ical chemistry and conservation in international journals
and books.
Luana Cuvillier and Arianna Passaretti are PhD

students at Haute Ecole Arc Conservation-restauration,
University of Neuchâtel, Switzerland, working on the
HELIX project.
Elodie Guilminot is a senior conservation scientist at

Arc’Antique, Nantes, France.
Aymeric Raimon and Virginie Dupuy are respectively

conservator and curator at Grand Patrimoine Loire
Atlantique (GPLA) in Nantes, France.
Young Conservation Professional
Microbial Stabilization of Archaeological Iron Objects
Sarah James Edith Joseph*

Laboratory of Technologies for Heritage Materials, University of Neuchâtel Laboratory of Technologies for Heritage Materials, University of Neuchâtel
Neuchâtel, Switzerland Haute École Arc Conservation-restauration, HES-SO University of Applied
[email protected] Sciences and Arts Western Switzerland
Mathilde Monachon [email protected]
Laboratory of Technologies for Heritage Materials, University of Neuchâtel
Neuchâtel, Switzerland *Author for correspondence
[email protected]
Sathiyanarayanan Ganesan
Laboratory of Technologies for Heritage Materials, University of Neuchâtel
[email protected]
Abstract
Chlorides and sulfides in the archaeological environment react biomineralization and iron uptake by microorganisms to
with iron ions to produce corrosion products such as akaganeite, remove or stabilize iron corrosion products. In this study, the
goethite, and iron sulfates. Traditional conservation efforts iron reduction and dechlorination properties of Pseudomonas
include chemical and mechanical corrosion removal together putida mt-2 (KT2440) were examined.
with climate control. The primary chemical treatments for
archaeological iron generally consist of either alkaline baths Keywords
for desalination or the application of complexing agents, such archaeology, iron corrosion, green conservation,
as tannic acid, to transform the corrosion products. However, biomineralization, iron uptake
green treatments are emerging in the heritage field, including
Introduction
The stability of an archaeological iron object depends oxides (Cornell and Schwertmann 2003, Selwyn 2004).
on the specific environment in which it was found and For objects recovered from freshwater or seawater, the
subsequently stored post-excavation (Groysman 2010). presence of iron sulfides reflects exposure to sulfur-re-
Both pre- and post-excavation, the object may be exposed ducing bacteria (Rémazeilles et al. 2010).
to environmental stresses, such as freeze-thaw cycles
and chemical contamination. The resulting corrosion The active corrosion and deterioration of archaeological
crust will differ accordingly with respect to its compo- iron are primarily caused by contamination with anions
sition, thickness, and reactivity (Figure 1). For instance, (Cl−, S2−) and the presence of oxygen and moisture, which
depending on pH, burial environments can produce together support electrochemical reactions (Thickett
chloride-containing compounds such as akaganeite 2012; Table 1). As the corrosion products accumulate,
β-ferric oxyhydroxide or combinations of different iron they may cause delamination of the corrosion layers
Figure 1. Formation of iron

corrosion crust
Table 1. Common chemical equations for iron according to the surrounding environment: aerated soil, acidic soil (pH < 4), with the presence
of chlorides or sulfur-reducing bacteria (SRB). (a) Watkinson 1983; (b) Watkinson 1983, Scott and Eggert 2016; (c) Scott and Eggert 2016; (d)
Groysman 2010
Aerated a Acidic b Chlorides c SRB d
Table 2. Microbial mechanisms and their effect on iron-containing heritage materials
Microbial mechanisms Microorganisms Substrate References

Iron phosphating Rhodococcus sp. C125; Mild steel (industrial process) Volkland et al. 2000a; Volkland et al. 2000b;
Biomineralization
Geobacter sulfurreducens; Volkland et al. 2001; Cote et al. 2015

Pseudomonas putida mt-2
Iron reduction Desulfitobacterium hafniense; Corroded iron (akaganeite, iron Comensoli et al. 2017; Kooli et al. 2018;
Shewanella loihica oxides) Kooli et al. 2019
Iron complexation Siderophore-producing Wood; paper (iron oxides) Rapti et al. 2021
Uptake
bacteria: Pseudomonas sp.

Sulfur oxidation Thiobacillus denitrificans Wood (iron sulfides) Monachon et al. 2020
(Thickett 2012). The increased acidification results in was synthesized according to Cornell and Schwertmann
a loss of mechanical strength, especially in iron-wood (2003). The ground archaeological mixture consisted of
composite objects (Fors et al. 2012). archaeological iron pieces ground with a mortar into a
fine powder. Pyrite was obtained from a geological spec-
The treatment of archaeological iron mostly relies on
imen (Pierre qui Roule, Switzerland) and was ground and
desalination methods developed in the 1970s (Watkinson
then washed in HCl− acetone before use.
2010). These methods include alkaline baths and
complexing agents (Scott and Eggert 2016). Chemical
Bacterial treatment
treatments have proven to be relatively effective, but
determining the treatment endpoint can be difficult, Pseudomonas putida mt-2 (PPmt2, DSM 6125) was
thus risking damage to the objects. Accordingly, there is selected for the biological treatment as it was previously
growing interest in less toxic methods, particularly green shown to produce a cohesive iron-phosphate layer on
treatments, for long-term stabilization. For example, in a mild steel (Volkland et al. 2000a; Table 2)underwent
biologically based approach natural microbial processes a surface reaction leading to the formation of a corro-
are exploited to convert the corrosion layer into an inert sion-inhibiting vivianite layer [Fe3(PO4.
biomineralized layer or to remove undesired compounds A modified nutrient broth (NBS; peptone 15 g/L, yeast
(iron uptake), as summarized in Table 2. In this study, 5 g/L, D±glucose1 g/L) was used to prepare thio1 medium
the iron reduction and dechlorination properties of (thio1; NBS with sodium thioglycolate 250 mg/L)
Pseudomonas putida mt-2 (KT2440) were examined.
Iron sources
Four iron sources were tested in this study: iron powder,
akaganeite, ground archaeological mixture (composed
of iron, iron oxyhydroxides, and traces of quartz from
the soil), and pyrite. The iron powder was sourced from Figure 2. Summary of the design of the biomineralization experiment,
Carl Roth Chemicals (≥ 99.5% p.a., powder). Akaganeite created using biorender.com
MICROBIAL STABILIZATION OF ARCHAEOLOGICAL IRON OBJECTS 37
according to James and Joseph (2021sulphur species). fact that some corrosion products are less amenable to
MM9 medium was modified from Volkland et al. (2000a) conversion than others due to their particular arrangement
by the addition of 200 mL of salt solution (NH4NO35 g/L, of atoms and their electron charges. However, a conversion
NaCl 2.5 g/L, Na2HPO464 g/L, KH2PO415 g/L) to 20 mL of iron powder and archaeological mixture was observed
of glucose (20% w/v), 2 mL MgSO4 (1 M), 0.1 mL (Figure 3). The effective conversion of iron powder over
Ca(NO3)2 (1 M). other oxidized iron species suggests its use as an additive
PPmt2 pre-cultures were grown in NBS medium under to boost less favorable reactions. For example, iron powder
180 rpm agitation at 28°C–30°C for 24 h with a bottle could be added to initiate a more uniform phosphating
volume ratio of 1:2. The bacterial cells were harvested and layer. The conversion of the archaeological mixture,
washed via centrifugation (5000 rpm, 15 min), including while significant, cannot be fully attributed to PPmt2 as
two washes with NBS followed by one with either thio1 the mixture was not sterilized before use and perhaps
or MM9 medium. because the burial biome was still present. However, a
reactivation of the microorganisms present in the burial
For each iron source, PPmt2 was inoculated in triplicate
environment during stabilization treatments can also result
in thio1 or MM9 media. Two abiotic controls were also
in the formation of more stable compounds, resulting in
included. Growth was monitored from photographs taken
at 1-week intervals for a duration of 6 weeks. The bacteria an object’s preservation (Orr et al. 2021).
were cultured without agitation, to allow for different An advantage of PPmt2 is that it produces siderophores,
application modes (solutions, gels) in future practice. which enhance iron uptake. Thus, PPmt2 treatment
could allow for the removal of thick iron-containing
Results and discussion concretions, while potentially increasing the availability
In thio1 and MM9 media, there was no bacterial conver- of iron for conversion (biomineralization). Iron uptake
sion of pyrite and only a partial conversion of akaganeite has been successfully employed in the removal of harmful
(data not shown). These results might be explained by the iron species from waterlogged wood (Monachon et al.
Figure 3. Archaeological iron powder treated with Pseudomonas putida mt-2 in the presence of sodium thioglycolate (250 mg/L) as an oxygen
scavenger: (A) day 0 and (B) after a 6-week incubation at 30°C under stationary conditions. The precipitated black particulate material was identified
as a combination of vivianite, quartz, and magnetite
2020). Promising iron extraction rates of up to 75% were Groysman, A. 2010. Corrosion for everybody. Dordrecht:
obtained with a commercial siderophore, deferoxamine. Springer.
Isolation and identification of the natural biome have thus James, S. and E. Joseph. 2021. Microbial-driven stabi-
been initiated by our group to determine the impact of lisation of archaeological iron artefacts. Corrosion and
these microorganisms on the conversion of the treated Materials Degradation 2(2): 274–92.
iron compounds and their potential relationship with Kooli, W.M., L. Comensoli, J. Maillard, M. Albini, A. Gelb,
exogenously added microbes (whole biomes or isolated P. Junier, and E. Joseph. 2018. Bacterial iron reduction
strains). In addition, the use of Methylophilus methy- and biogenic mineral formation for the stabilisation of
lotrophus is currently being assessed as an alternative corroded iron objects. Scientific Reports 8(1): 1–11.
treatment option, as suggested by Orr et al. (2021).
Kooli, W.M., T. Junier, M. Shakya, M. Monachon, K.W.
Conclusion/future prospects Davenport, K. Vaideeswaran, A. Vernudachi, I. Marozau,
T. Monrouzeau, C.D. Gleasner, K. McMurry, R. Lienhard,
The use of microbial processes to treat corrosion on L. Rufener, J.-L. Perret, O. Sereda, P.S. Chain, E. Joseph,
archaeological iron requires further studies before this and P. Junier. 2019. Remedial treatment of corroded iron
approach can be applied to recovered or stored objects. objects by environmental aeromonas isolates. Applied and
The results will allow versatile methods to be proposed to Environmental Microbiology 85(3): e02042–18.
end users. For composite objects (i.e., iron parts of water-
logged wood objects), a sequential protocol that includes Monachon, M., M. Albelda-Berenguer, T. Lombardo, E.
iron extraction from the wood substrate and stabilization Cornet, F. Moll-Dau, J. Schramm, K. Schmidt-Ott, and E.
of the corroded iron parts may be effective. This would Joseph. 2020. Evaluation of bio-based extraction methods
avoid a lack of compatibility between the microorganisms by spectroscopic methods. In Minerals 10(2): 203–20.
used in the treatment of different materials or between Orr, C.H., R. Williams, H.H. Halldórsdóttir, A. Birley,
the microorganisms and the object itself. E. Greene, A. Nelson, T.K. Ralebitso-Senior, and G.
Taylor. 2021. Unique chemical parameters and microbial
Acknowledgments activity lead to increased archaeological preservation at
the Roman frontier site of Vindolanda, UK. Scientific
This work was supported by the Swiss National Science
Foundation (grants no. PP00P2_163653/1, 2016–2020, Reports 11(1): 1–10.
and PP00P2_190081, 2020–2023). Rapti, S., S.C. Boyatzis, S. Rivers, and A. Pournou. 2021.
Siderophores and their applications in wood, textile, and
References paper conservation. In Microorganisms in the deteriora-
Comensoli, L., J. Maillard, M. Albini, F. Sandoz, P. Junier, tion and preservation of cultural heritage, ed. E. Joseph,
and E. Joseph. 2017. Use of bacteria to stabilize archae- 301–39. Cham: Springer Nature Switzerland AG.
ological iron. Applied and Environmental Microbiology Rémazeilles, C., M. Saheb, D. Neff, E. Guilminot, K. Tran,
83(9): e03478–16. J.A. Bourdoiseau, R. Sabot, M. Jeannin, H. Matthiesen,
Cornell, R.M. and U. Schwertmann. 2003. The iron oxides: P. Dillmann, and P. Refait. 2010. Microbiologically
Structure, properties, reactions, occurences and uses, 2nd influenced corrosion of archaeological artefacts: Char-
ed. Weinheim: Wiley-VCH. acterisation of iron(II) sulfides by Raman spectroscopy.
Journal of Raman Spectroscopy 41(11): 1425–33.
Cote, C., O. Rosas, and R. Basseguy. 2015. Geobacter
sulfurreducens: An iron reducing bacterium that can Scott, D. and G. Eggert. 2016. Iron and steel in art.
protect carbon steel against corrosion? Corrosion Science London: Archetype Publications.
94: 104–13. Selwyn, L. 2004. Overview of archaeological iron: the
Fors, Y., F. Jalilehvand, E. Damian Risberg, C. Björdal, corrosion problem, key factors affecting treatment, and
E. Phillips, and M. Sandström. 2012. Sulfur and iron gaps in current knowledge. In Metal 04: Proceedings of the
analyses of marine archaeological wood in shipwrecks Interim Meeting of the ICOM-CC Metals Working Group,
from the Baltic Sea and Scandinavian waters. Journal of Canberra, 4–8 October 2004, eds. J. Ashton D. Hallam,
Archaeological Science 39(7): 2521–32. 294–306. Canberra: National Museum of Australia.
MICROBIAL STABILIZATION OF ARCHAEOLOGICAL IRON OBJECTS 39
Thickett, D. 2012. Post excavation changes and preventive

conservation of archaeological iron. PhD dissertation,
University of London, UK.
Volkland, H.P., H. Harms, K. Knopf, O. Wanner, and
A.J.B. Zehnder. 2000a. Corrosion inhibition of mild steel
by bacteria. Biofouling 15(4): 287–97.
Volkland, H.P., H. Harms, B. Müller, G. Repphun, O.
Wanner, and A.J.B. Zehnder. 2000b. Bacterial phos-
phating of mild (unalloyed) steel. Applied and Environ-
mental Microbiology 66(10): 4389–95.
Volkland, H.P., H. Harms, O. Wanner, and A.J.B. Zehnder.
2001. Corrosion protection by anaerobiosis. Water
Science and Technology 44(8): 103–6.
Watkinson, D. 1983. Degree of mineralization: Its signif-
icance for the stability and treatment of excavated iron-
work. Studies in Conservation 28(2): 85–90.
Watkinson, D. 2010. Measuring effectiveness of washing
methods for corrosion control of archaeological iron:
Problems and challenges. Corrosion Engineering, Science
and Technology 45(5): 400–406.
Authors
Sarah James has a master’s degree in conservation
science from the University of Cardiff, UK. She is
currently a PhD student at the University of Neuchâtel,
where she investigates biomineralization processes to
stabilize archaeological iron artifacts.
Mathilde Monachon is currently a postdoctoral

researcher at the University of Neuchâtel, where she
studies green extraction methods for iron and sulfur
species to preserve waterlogged archaeological wood.
Her main research area is analytical chemistry with a
focus on conservation science.
Sathiyanarayanan Ganesan is currently a senior

postdoctoral researcher at the University of Neuchâtel
working on green extraction methods for iron and sulfur
species to preserve waterlogged archaeological wood.
His main research area is microbiology with a focus on
conservation science.
Edith Joseph is an assistant professor at the University

of Neuchâtel and the Haute École ARC Conservation-res-
tauration. She has a PhD in chemistry from the University
of Bologna, Italy. Her main research activities are the
application of green chemistry for the preservation of
artistic and archaeological objects.
Deterioration Tests Using a Simulated Tumulus to
Evaluate In Situ Preservation of Metal Artifacts in a Stone
Chamber
Akinobu Yanagida* Huarong Xie Nobumitsu Takatori Shuichi Hokoi
Nara National Research Institute for School of Architecture, Southeast Department of Architecture School of Architecture, Southeast
Cultural Properties University and Architectural Engineering, University
Nara City, Japan Nanjing, China Graduate School of Engineering, Nanjing, China
[email protected] [email protected] Kyoto University [email protected]
Kyoto, Japan
Soichiro Wakiya Daisuke Ogura [email protected] Yohsei Kohdzuma
Nara National Research Institute for Department of Architecture Nara National Research Institute for
Cultural Properties and Architectural Engineering, Hiroyuki Yasui Cultural Properties
Nara City, Japan Graduate School of Engineering, Department of Architecture Nara City, Japan
[email protected] Kyoto University and Architectural Engineering, [email protected]
Kyoto, Japan Graduate School of Engineering, *Author for correspondence
[email protected] Kyoto University
Kyoto, Japan
[email protected]
Abstract
To verify the possibility of in situ preservation of metal artifacts emergence of a tin compound corrosion layer. These results were
inside the stone chamber of a tumulus, the corrosion behavior congruent with genuine artifact corrosion behavior and revealed
of iron and bronze artifacts was studied by conducting a the relationship between the corrosion process and the buried
deterioration test in a simulated tumulus. The corrosion rate environment within the stone chamber.
of the suspended carbon steel samples in the stone chamber
was extremely low, at one-tenth that of samples buried in Keywords
soil. Malachite formed on the bronze samples because of an iron artifact, bronze artifact, tumulus, in situ preservation,
increasing CO2 concentration in the stone chamber after the atmospheric corrosion, soil corrosion, buried environment
Introduction
There are approximately 200,000 tumuli1 in Japan with Because of the difficulty of investigating the physical
soil mounds and stone chambers (Figure 1). In most environment and conducting corrosion monitoring of
of the tumuli, significant amounts of iron and bronze metal artifacts over time in a genuine stone chamber,
artifacts are buried in the stone chamber (Otsuka 2012). experimental methods need to be adopted.
In a tumulus, the position of artifacts within the stone
This study aimed to investigate the initial corrosion of
chamber relates to their academic value (Tateishi 2020).
Removing significant amounts of artifacts for archeo- iron and bronze artifacts in the stone chambers of tumuli
logical investigation would make their conservation in typically found in Japan. A simulated tumulus was used
a storage facility expensive and time-consuming. Conse- and environmental investigations (Xie et al. 2020) and
quently, there is a need to preserve the grave contents in corrosion tests conducted to discover the effect of buried
situ. For this purpose, a quantitative investigation of the environments on the corrosion of these metal artifacts.
relationship between the physical environment of the
artifacts and their corrosion is essential. Furthermore, Experimental methods
by understanding this relationship in the stone chamber,
Structure of the simulated tumulus
the artifacts can be actively preserved by adjusting the
environment, such as placing soil on the surface of the The simulated tumulus was located in a hilly area at
tumulus. This method has benefits in terms of lower the Katsura Campus of Kyoto University (Figure 1) and
energy consumption and sustainability compared to consisted of an antechamber and a stone chamber. The
controlling the environment using an air conditioner. chamber was 50 × 50 × 100 cm3 in size and its floor was
DETERIORATION TESTS USING A SIMULATED TUMULUS TO EVALUATE IN SITU PRESERVATION OF METAL ARTIFACTS IN A STONE CHAMBER 41
Table 1. Environmental monitoring of simulated tumulus
Monitoring variables Position of monitoring sensors Measuring instruments

Stone Temperature Ceiling, floor, and air T-type thermocouple
chamber Temperature and relative-humidity sensor
Relative humidity Air
(Onset HOBO U23 Pro v2)
Water potential Both sides of the floor soil Dielectric water-potential sensor (METER MPS-2)
Oxygen concentration Air and soil Soil O2 sensor (EMJ MIJ-03)
Carbon dioxide concentration Air Carbon dioxide sensor (BARON BEC-CO2SA-A40)
Outdoor Air temperature, humidity, air pressure, precipitation, Weather station
weather solar radiation, wind direction, and speed (DAVIS)
1.5 m above the ground. The sides and ceiling of the spectroscopy (EDX) coupled with SEM, micro-X-ray
chamber were lined with 2 cm-thick granite slabs, while diffraction (µ-XRD), and micro-Raman (µ-Raman)
the floor consisted of earth. The environmental investi- spectroscopy. The samples installed for approximately
gations and corrosion tests were conducted in the stone three years were mounted in epoxy resin, cut, polished—
chamber, whereas the dataloggers for the environmental the final polishing being with a 1 µm polycrystalline
measurements and corrosion sensors were placed in the diamond paste—under ethanol, and subjected to the
antechamber (Figure 1a). The inside of the stone chamber same analysis.
was observed through an acrylic window that separated
it from the antechamber. In the environmental investiga- Calculation of corrosion rate
tions, in addition to meteorological observations of the Table 4 shows the samples used to calculate the corrosion
model tumulus, the temperature, water potential,2 and rate. To remove the corrosion products, carbon steel
O2 and CO2 concentrations in the mound were meas- samples were soaked in 6 mol/L of hydrochloric acid
ured. Inside the stone chamber, the temperature, relative containing hexamethylenetetramine (0.35 wt%) for 10 min
humidity (RH), and O2 and CO2 concentrations in the (JIS Z 2371; Japanese Standards Association 2015). The
atmosphere were measured according to a previously bronze samples were soaked in 6 mol/L of hydrochloric
described method (Table 1; Xie et al. 2020). acid for 3 min. The weight loss in the samples was meas-
ured after washing in distilled water and degreasing in
Deterioration test acetone using an ultrasonic cleaning machine. The weight
Samples loss was determined by subtracting the control weight
loss from the sample weight loss. The corrosion rate was
Carbon steel (SS400) and two types of bronze—namely,
calculated using the following equation:
low-tin (C5191, Sn: 6.69 wt%) and high-tin (CAC502A,
Sn: 9.91 wt%) bronze—were cut to size (50 × 50 × 5 mm3),
polished (final polishing: #400 emery paper), washed in (1)
distilled water, and degreased in acetone using an ultra-
sonic cleaning machine. These samples were installed In Equation (1), Rcorr, ΔW, S, Δt, and ρ indicate corrosion
on February 14, 2014. One set of samples was suspended rate (mm/y), weight loss (g), sample area (mm2), and
in the stone chamber, and the other was buried in the experimental period (year).
ground (Figure 1b-2, 1b-3). Table 3 describes the appa-
ratus. Table 2. Date of stone chamber opening and sample extraction
Analytical methods Opening the stone

Date Sample extraction
chamber
The samples were observed from the antechamber 1/9/2014  Sample installation
approximately every three months and removed at 9/30/2014  
regular intervals for analysis. They were observed using 11/24/2015  , Sample installation
optical microscopy (OM) and scanning electron micros- 11/27/2016  , Sample installation
copy (SEM). The corrosion products that formed on the 11/27/2017  
samples were analyzed using energy-dispersive X-ray 10/10/2018  

Table 3. The samples for corrosion rate measurement in the stone chamber and underground increased in
Apparatus Model Manufacturing company winter and decreased in summer (Figure 2b), in contrast
Ultrasonic cleaning US CLEANER AS ONE Co., to the CO2 concentration, which reached approximately
machine 10% in summer. This behavior was attributed to the accel-
Optical microscope Z16 APO A Leica Microsystems Co. Ltd. eration of aerobic microbial respiration in summer owing
Scanning electron JSM-IT200 JEOL Co. Ltd. to the increased temperature. The water potential of the
microscope
soil underground remained high throughout the year.
Micro-X-ray Smart lab Rigaku Co. Ltd.
diffractometer
Micro-Raman LabRAM HR HORIBA Ltd.
spectroscope Evolution
Table 4. The samples for corrosion rate measurement
Installation Removal Experimental Sample

date date period (day) number
1/9/2014 9/30/2014 264 1
1/9/2014 11/24/2015 684 2
1/9/2014 11/27/2017 1418 1
1/9/2014 10/27/2016 1022 2
1/9/2014 10/10/2018 1735 2
11/24/2015 10/27/2016 338 1 Figure 2. Change in the environmental sensors: (a) surface temperature
11/24/2015 10/10/2018 1051 1 of the ceiling and floor; (b) O2 and CO2 concentrations; (c) water potential
10/27/2016 11/27/2017 396 1 in the ground and due to precipitation
Results and discussion Corrosion of carbon steel

Figure 3 shows photographs of each sample. On the
Burial environment of the simulated tumulus
carbon steel, a dense rust film formed at the lower edge
The RH in the stone chamber was maintained at 100% of the suspended samples (Figure 4a), causing extensive
throughout the experiments. The surface temperature at corrosion underneath. In contrast, porous rust formation
the ceiling was higher than that at the floor in summer, occurred over the entire surface of the buried samples.
whereas in winter it was the opposite (Figure 2a), indi- SEM-EDX showed that the samples thinned locally due to
cating that condensation occurred near the floor in corrosion (indicated by the red ↑ in Figure 4b). Because
summer and the ceiling in winter. The O2 concentration Si—attributed to soil particles—was detected in the upper
Figure 1. Illustration of the different

types of ancient tumuli and
schematic diagram of the simulated
tumulus: (a-1) tumulus shape; (a-2)
type of stone chamber; (b-1) the
geometry of the simulated tumulus;
(b-2) location of the metal samples;
(b-3) setup of the corrosion sensors
section of the rust layer, the carbon steel presumably lower edge is possibly higher than that in Figure 5. In
corroded heavily when in contact with the soil particles. contrast, the corrosion rate of the buried samples grad-
Goethite (α-FeOOH), lepidocrocite (γ-FeOOH), and ually decreased: 0.025–0.038 mm/y after 1 year falling
magnetite (Fe3O4) were detected by XRD on the rust on to 0.015 mm/y after 5 years. The corrosion rate of the
both suspended and buried samples. buried samples was approximately 10 times higher than
that of the 4–5-year-old suspended samples.
The corrosion rate in the 2–5-year-old suspended samples
was 0.0003–0.0024 mm/y but was slightly higher in the
Corrosion mechanism of iron artifacts
1-year-old samples (Figure 5a). Because the corrosion rate
was calculated as an average for the entire sample, the The corrosion rate in the atmosphere depends on the
localized corrosion behavior is not presented. Therefore, wetting time and water film thickness on the surface
in the suspended samples, the actual corrosion rate at the of the metal (Hœrlé et al. 2004). In the soil, corrosion
Figure 3. Photographs of the

removed samples from the
simulated tumulus
Figure 4. OM and SEM-EDX

images of the carbon steel
samples: (a) suspended
samples; (b) buried samples.
OM images indicate the surface
corrosion conditions and SEM
images show a cross-section of
the samples
Figure 5. Change in corrosion

rate: (a) carbon steel samples;
(b) bronze samples
proceeds due to Fe2+ removal from the metal surface was formed due to the oxidation of Fe2+ to Fe3+ at the
and O2 supply from the soil to the metal (Neff et al. gas-liquid interface (Figure 6a-IV).
2005). Based on previous studies, three factors cause
For the buried samples (Figure 6b), thick water films
intense corrosion in the soil. First, the buried samples
formed when in contact with large soil particles (He et
remained wet because of the high water content of the
al. 2017) (Figure 6b-I). Consequently, localized thinning
soil, whereas the wetting area and time on the suspended
presumably occurred in the thick water film in contact
samples changed depending on the season. Second, the
with large soil particles via the fixing of an anode (Figure
conductivity of the pore water in the soil was higher
6b-II, 6b-IV). Previous studies showed that the supply
than that of the water film on the suspended samples, of dissolved oxygen to the metal or removal of Fe2+ from
because the latter was due to condensation. Third, the the metal through a rust layer controls the corrosion rate
pore water on the surface of the samples resulted from in soil (Neff et al. 2005, He et al. 2017, Shibata 2019).
rainfall, stimulating the removal of Fe2+, whereas Fe2+ The corrosion rate on the buried samples presumably
remained on the surfaces in the stone chamber. decreased over time because the growth of the rust
Figure 6a shows a schematic of the corrosion mechanism layer suppressed the supply of dissolved oxygen and the
on the suspended samples. The water film was thin on removal of Fe2+.
the sides of the samples and thick at the lower edge;
that is, water flowing along the sides accumulated at the Corrosion of bronze samples
lower edge. In the initial stage of corrosion, dissolved In the low-tin bronze sample, spots of green corrosion
oxygen in the water film was consumed on both the products, identified using µ-XRD and µ-Raman spec-
sides and lower edges of the samples. The supply of troscopy as malachite (Cu2(CO3)(OH)2), were formed on
dissolved oxygen in the water film at the lower edges of the dense black corrosion layer (DBCL) (Figure 7a-1).
the carbon steel gradually decreased compared to that The distribution of the DBCL—where O and Sn were
at the sides because of its slow diffusion (Figure 6a-I, detected by EDX—was based on where the water
6a-II). Consequently, macrocell corrosion proceeded droplets flowed. Moreover, malachite was generated
because of the formation of an oxygen concentration cell at the DBCL cracks (Figure 7a-1). The DBCL was an
via the fixing of a cathode on the side of the sample and aggregate of minute particles, with a corrosion layer of
an anode on the lower edge (Figure 6a-III). Moreover, 10–20 µm (Figure 7a-2). Because obvious spectra were
by increasing the concentration of Fe2+ via the corrosion not detected in the DBCL using µ-XRD and µ-Raman
reaction in the water film at the lower edge, a rust film spectroscopy, this corrosion layer was presumably
Figure 6. Schematics of the corrosion mechanism of carbon steel: (a) suspended samples; (b) buried samples. DO = dissolved oxygen
Figure 7. OM and SEM images

of the suspended bronze
samples: (a-1) surface of the
low-tin bronze samples; (a-2)
cross-sections of low-tin bronze
samples; (b-1) surface of the
high-tin bronze samples; (a-2)
cross-sections of high-low
bronze samples
amorphous. Previous research reported that a DBCL Corrosion mechanism of bronze samples
can occur on the surface of bronze artifacts (Wang
Figure 8a shows a schematic of the corrosion on bronze
et al. 1995) and that this corrosion layer consists of
samples suspended in the stone chamber. In the low-tin
amorphous tin oxide or tin hydroxide (Trentelman et
bronze sample, the corrosion of the alloy began with
al. 1999, Chiavari et al. 2007). Therefore, the DBCL was
the reduction reaction of dissolved oxygen due to the
presumably tin oxide or tin hydroxide. The cross-sec-
tional images also indicated that there was substantial formation of the water film (Figure 8a-I). Because of
thinning of the alloy under malachite (indicated by the the low solubility of the tin compound, it was deposited
red △ in Figure 7a-2). on the surface of the alloy and became a DBCL. For
copper, cuprite occurs during atmospheric corrosion
In Figure 7b-1, corrosion of the high-tin bronze sample (Graedel et al. 1987). In contrast, amorphous tin oxide
was more evident in the upper-left area and less evident or tin hydroxide forms on bronze (Cilvia et al. 2007,
in the lower-right area of the image. The dendrite-shaped Souissi et al. 2007) and acicular malachite occur on the
alloy changed to black, and on it formed an acicular of tin compound layer (Figure 8a-II, 8a-III). The formation
malachite. In the dendrite-shaped α phase, O and Sn were of malachite is attributed to the reaction between cupric
detected, while the intensity of Cu decreased. Therefore, and carbonate ions by the dissolving of atmospheric CO2
the α phase corroded to leave oxide or tin hydroxide. In in the water film (Graedel et al. 1987). The concentration
contrast, even in the heavily corroded area, the eutectoid of CO2 decreased to 3% in winter and increased to 10%
of the (α+δ) phase showed high luminance in the SEM in summer in the stone chamber (Figure 3b). When
image. The eutectoid of the (α+δ) phase does not corrode dependent only on the equilibrium of carbonate (Stumm
easily compared to the α phase. In the cross-section of and Morgan 1996), the pH of the water film varied from
the high-tin bronze sample, acicular malachite had also
4.4–4.7, with a total carbon dioxide (Ct) content of
formed on the DBCL, similar to the low-tin bronze
10−3 mol/L in winter and 10−2 mol/L in summer. Figure
sample (Figure 7b-2). Furthermore, malachite growth
9 shows the stable phase change from tenorite in winter
occurred during summer in both the low-tin and high-tin
(Ct: 10−2 mol/L) to malachite in summer (Ct: 10−3 mol/L)
bronze samples. Figure 5b shows that the corrosion rate
depending on the Ct concentration. That is, malachite
of the low-tin bronze sample in the stone chamber was
formed because of the increasing CO2 concentration in
slightly higher than that of the high-tin bronze sample.
the stone chamber. In addition, the water film on the
In the buried samples, the low-tin and high-tin bronze bronze sample decreased because the temperature of the
samples showed corrosion behaviors similar to that in ceiling in the stone chamber changed, becoming higher
the stone chamber. However, acicular malachite was not than that of the ground in summer. Acicular malachite
observed in the buried samples. presumably formed because of an increase in the concen-
Figure 8. Schematic of bronze corrosion mechanism in the stone chamber: (a) low-tin bronze sample; (b) high-tin bronze sample
tration of cupric ions (Figure 8a-III). This anode reaction attributed to an increase in the CO2 concentration in
possibly occurred in a relatively large gap between the the stone chamber, similar to the low-tin bronze sample
tin compounds (Figure 8a-IV). (Figure 8b-III).
In the high-tin bronze sample, Cu was leached from the

Comparison with the condition of genuine artifacts
center of the dendritic structure where Cu was highly
concentrated, thus increasing the Sn concentration on the Figure 10a shows that the corrosion rate on genuine
surface (Figure 8b-I). In this process, selective leaching iron artifacts buried in the soil was lower than that
occurs because the α phase is an anode and the eutectoid of the buried samples. In a buried iron artifact, the
is a cathode. As corrosion proceeds, the α-phase grain corrosion rate decreases due to the growing corrosion
boundary is observed, and the DBCL gradually spreads layer suppressing the diffusion of dissolved oxygen and
Fe2+. In contrast, the corrosion rate of the suspended
(Figure 8b-II). Furthermore, malachite growth was
samples in the simulated tumulus was identical to that
of the genuine artifacts in the stone chamber of the real
tumulus. This indicates that the corrosion rate in the
stone chamber was suppressed after the initial stages,
because an extremely dense corrosion film was formed.
Figure 10b shows a cuirass excavated from a tumulus,
which was heavily corroded by the soil. In contrast, the
part where the armor was exposed to the atmosphere
presumably corroded very slowly because its original
shape remained intact.
Figure 9. Effects of Ct concentration on the formation of malachite and
tenorite. This figure shows the copper compound formed by the pH at Figure 10c shows bronze artifacts with different Sn
each Ct in the water film on the bronze samples concentrations. The high-tin bronze retained its surface
Figure 10. Comparison of the experimental results and genuine artifacts: (a) relationship between the corrosion depth and time in carbon steel samples
and iron artifacts (Yoshikawa et al. 2008, Yoshikawa et al. 2009, Mitsui et al. 2012, Mitsui 2015a, Mitsui 2015b). The corrosion rate of genuine artifacts
was calculated by dividing the corrosion thickness measured using X-ray computed tomography by the duration of burial; (b) the cuirass excavated
from tumulus; (c) the different Sn content of bronze artifacts
decoration due to protection by the DBCL, whereas the Scientific Research C (Number 21K00998). We would
low-tin bronze corroded heavily. Because the concen- like to thank Katsuhiko Aketa (Revive HIKO), Shinji
tration of Sn was 1.71 and 18.77 wt% in the low-tin and Yamamoto (Okumura Corporation), Nobuhiro Ichikawa
high-tin bronze (Saito et al. 2016), respectively, the metal- (Live Architects Office), Nippon Steel Technology, and
lographic structure of these was presumably the α phase Madoka Keshi (Kyoto University).
and eutectoid of the (α+δ) phase. This behavior shows
that the corrosion condition of the bronze samples in the Notes
simulated tumulus corresponded to the bronze artifacts. 1
One tumulus in Japan was constructed in the mid-3rd
century and remained in use until the end of the 7th
Conclusion
century. The size of the soil mound ranged from a few
In this study, a deterioration test using carbon steel and meters up to 525 m.
two types of bronze samples in a simulated tumulus 2
The decrease and increase in water potential under
was combined with an environmental investigation to
the same soil corresponded to a decrease and increase
evaluate the corrosion behavior in a stone chamber. The
in moisture in the soil.
corrosion rate of the iron artifacts in the stone chamber
was extremely low compared to that of buried iron arti-
References
facts. In bronze artifacts, the seasonal change in the CO2
concentration affected malachite formation. In addition, Chiavari, C., K. Rahmouni, H. Takenouti, S. Joiret,
the metallographic structure of bronze strongly affected P. Vermaut, and L. Robbiola. 2007. Composition and
the corrosion behavior, bearing in mind that the high-tin electrochemical properties of natural patinas of outdoor
bronze sample was less corrosive. The simulated tumulus bronze monuments. Electrochimica Acta 52(27): 7760–9.
results corresponded to the corrosion conditions of Gill, W.R. and G.E. Vanden Berg. 1968. Soil dynamics
genuine artifacts. in tillage and traction – Agricultural handbook no. 316,
The study confirmed the initial corrosion behavior 42–52. Washington, DC: Agricultural Research Service,
of metal artifacts in a tumulus. In future work, in situ United States Department of Agriculture.
preservation in a tumulus will be assessed by analyzing Graedel, T.E, K. Nassau, and J.P. Franey. 1987. Copper
the data from corrosion sensors placed in the simulated patinas formed in the atmosphere—I. Introduction.
tumulus. Corrosion Science 27(7): 639–54.
Hœrlé, S., F. Mazaudier, P. Dillmann, and G. Santarini.
Acknowledgments
2004. Advances in understanding atmospheric corrosion
This work was supported by a JSPS KAKENHI Grant- of iron. II. Mechanistic modelling of wet-dry cycles.
in-Aid for Scientific Research B (Number 26289203) and Corrosion Science 46(6): 1431–65.
Japanese Standards Association (JSA). 2015. JIS G 3101: Tateishi, T. 2020. Mogi kofun: An experiment in site and
Rolled steels for general structure. Tokyo: JSA. artifact preservation and utilization: Part 1. The in-situ
Japanese Standards Association (JSA). 2015. JIS Z 2371: preservation of historic sites and open-air display of
Methods of salt spray testing. Tokyo: JSA. archaeological sites: Principles and history. Quarterly of
Archaeological Studies 67(1): 12–16. (in Japanese)
Japanese Standards Association (JSA). 2016. JIS H 5120:
Copper and copper alloy castings. Tokyo: JSA. Trentelman, K., L. Stodulski, R. Lints, and C. Kim. 1999.
A comparative study of the composition and corrosion of
Japanese Standards Association (JSA). 2018. JIS H 3110:
branches from Eastern Han dynasty money trees. Studies
Phosphor bronze and nickel silver sheets, plates, and strips.
in Conservation 44(3): 170–83.
Tokyo: JSA.
Wang, C., B. Lu, J. Zuo, S. Zhang, S. Tan, M. Suzuki,
Mitsui, S. 2015a. Burial environment and corrosion of
and W.T. Chase. 1995. Structure and elemental analysis
iron artifacts excavated from Mukibanda archaeolog-
on the nanocrystalline SnO2 in the surface of ancient
ical remains. Mukibanda Iseki Hakkutsu Chosa Kenkyu
Chinese black mirrors. Nanostructured Materials 5(4):
Nempo 2014: 27–44. (in Japanese)
489–96.
Mitsui, S. 2015b. Burial environment and corrosion of
Xie, H., D. Ogura, H. Yasui, N. Takatori, S. Hokoi, S.
socketed iron axe excavated from Yoshida Nakamichi site
Wakinya, A. Yanagida, and Y. Kohdzuma. 2020. Envi-
in Tottori City, Tottori Prefecture. Koeki Zaidan Hojin
ronment in stone chamber of an unexcavated tumulus
Tottoriken Kyoiku Bunka Zaidanhen, Yoshida Nakam-
and preservation of buried relics: Part 1. Environmental
ichi Iseki, Tottoriken Kyoiku Iinkai 2015/03: 221–30. (in
monitoring for simulated tumulus. Journal of Building
Japanese)
Physics 44(4): 287–308.
Mitsui, S., A. Fujii, M. Higuchi, and K. Nishimura. 2012.
Yoshikawa, H., E. Gunji, and M. Tokuda. 2008. Long term
Long-term corrosion of 2,000-year-old ancient iron
stability of iron for more than 1500 years indicated by
sword. MRS Online Proceedings Library 1475: 545–50.
archaeological samples from the Yamato 6th tumulus.
https://doi.org/10.1557/opl.2012.630
Journal of Nuclear Materials 379(1–3): 112–17.
Neff, D., P. Dillmann, L. Bellot-Gurlet, and G. Beranger.
Yoshikawa, H., S. Lee, and T. Matsui. 2009. A sampling
2005. Corrosion of iron archaeological artefacts in soil:
method and data evaluation of archaeological samples
Characterisation of the corrosion system. Corrosion
to support long-term corrosion prediction. Corrosion
Science 47(2): 515–35.
65(4): 227–32.
Otsuka, H. 2012. Kofun to hisosya no nazo ni semaru
[Exploring the mysteries of burial mounds]. Tokyo: Authors
Syoudensya Publishing. (in Japanese)
Akinobu Yanagida is a researcher at the Nara National
Revie, R.W. and H.H. Uhlig. 2008. Corrosion and corrosion
Research Institute for Cultural Properties. He specializes
control—An introduction to corrosion science and engi-
in conservation science.
neering, 4th ed., 116–31. New Jersey: John Wiley & Sons.
Soichiro Wakiya is the head of the Conservation Science
Saito, T., T. Koezuka, Y. Kohdzuma, M. Sawada, Y.
Laboratory at the Nara National Research Institute for
Nakagawa, and K. Masuda. 2016. About a chemical
Cultural Properties. He specializes in conservation
analysis of the bronze bell found at Kamo-Iwakura site.
science.
Studies of the Ancient Culture 24: 1–17. (in Japanese)
Souissi, N., E. Sidot, L. Bousselmi, E. Triki, and L. Huarong Xie is a postdoctoral researcher at the School
Robbiola. 2007. Corrosion behaviour of Cu–10Sn bronze of Architecture, Southeast University. He specializes in
in aerated NaCl aqueous media – Electrochemical inves- engineering building environments.
tigation. Corrosion Science 49(8): 3333–47. Daisuke Ogura is a professor at the Department of
Stumm, W. and J.J. Morgan. 1996. Aquatic chemistry: Architecture and Architectural Engineering, Graduate
Chemical equilibria and rate in natural waters, 3rd ed, School of Engineering, Kyoto University. He specializes
396–8. New Jersey: Wiley–Interscience. in engineering building environments.
Nobumitsu Takatori is an assistant professor at the

Department of Architecture and Architectural Engi-
neering, Graduate School of Engineering, Kyoto Univer-
sity. He specializes in engineering building environments.
Hiroyuki Yasui is a graduate at the Department of

Architecture and Architectural Engineering, Graduate
School of Engineering, Kyoto University. He specializes
in engineering building environments.
Shuichi Hokoi is a professor at the School of Architec-

ture, Southeast University. He specializes in engineering
building environments.
Yohsei Kohdzuma is the deputy director at Nara

National Research Institute for Cultural Properties. He
specializes in conservation science.
Materials list
SS400 is ordinary structural carbon steel made by
hot-rolling sheet steel to JIS G 3101 (JAS 2015).
C5191 is a JIS H 3110-standard phosphor bronze (JAS

2018). The samples were α monophase and contained
Cu: 93.32 wt% and Sn: 6.69 wt% according to SEM-EDX.
C502A is a JIS H 5120-standard cast phosphor bronze

(JAS 2015). The samples consisted of an α phase and a
eutectoid in the (α+δ) phase. The chemical composi-
tion of the sample was Cu: 93.32 wt% and Sn: 9.91 wt%
according to SEM-EDX.
Development of a Sunflower-Oil-based Bio-lubricant for

Use in Gear Models at the ISEP Museum
Maria Gonçalves Bruno Campos Eduarda Vieira Patrícia R. Moreira*

School of Arts, Universidade Research Centre for Science and School of Arts and Research Centre School of Arts, Research Centre
Católica Portuguesa Technology of the Arts (CITAR) and for Science and Technology of the for Science and Technology of
Porto, Portugal Centre for Biotechnology and Fine Arts (CITAR), Universidade Católica the Arts (CITAR), and Centre for
[email protected] Chemistry – Associate Laboratory Portuguesa Biotechnology and Fine Chemistry
(CBQF), Universidade Católica Porto, Portugal – Associate Laboratory (CBQF),
Portuguesa [email protected] Universidade Católica Portuguesa
Porto, Portugal Porto, Portugal
*Author for correspondence
Abstract
In this case study, we describe the development of an eco- the FTIR analyses, the bio-lubricant showed good resistance to
friendly, sunflower-oil-based bio-lubricant for use in the oxidation and good UV resistance.
conservation of exhibited metal gears. The bio-lubricant was
produced from two different sunflower oil bases. In addition to Keywords
the methodology used to develop the vegetable-oil lubricant, antioxidant, Fourier transform infrared (FTIR) spectroscopy,
we present the results of oxidation and UV resistance tests. In lubricant, oxidation, sunflower oil, thickener, ultraviolet (UV)
Introduction
Although the importance of industrial and scientific methods. Consequently, new approaches adapted to the
collections has long been recognized, the conservation requirements of this type of heritage are needed. In this
and restoration of scientific and technological objects are case, the objects of study were a set of ten scientific-di-
still very much guided by traditional, not entirely effective dactic pieces belonging to the Schröder collection of the
Figure 1. The Schröder collection

DEVELOPMENT OF A SUNFLOWER-OIL-BASED BIO-LUBRICANT FOR USE IN GEAR MODELS AT THE ISEP MUSEUM 51
Museum of Instituto Superior de Engenharia do Porto vegetable oils, its biodegradability, and the fact that its
(ISEP) (Figure 1). The lubricating oil present in the gears concentration can be easily increased or decreased as
of the pieces had degraded and was no longer protective, needed (Kreivaitis et al. 2015). For resource management
evidenced by its visible dripping onto undesired areas reasons, the compatibility tests were performed using the
of the pieces due to its fluidity. The lubricant had to be commercial sunflower oil base. After the antioxidant had
replaced rather than simply removed, to prevent wear been selected, it was added to the high-grade sunflower
between the gears and to allow their smooth functioning oil together with the thickener, and the two bio-lubricants
during use. As this provided the opportunity to take a were then compared.
“green” approach to conservation, a lubricant of vegeta-
The miscibility of the selected antioxidants was tested by
ble-oil origin was developed in our laboratory.
adding each one to the commercial sunflower oil base at
a final concentration of 5% (w/w), followed by thorough
Objective
mixing using a magnetic stirring and induction plate
Commercial lubricants are not biodegradable, since they first at room temperature and then gradually increasing
are made from mineral oils and other toxic components both the stirring rate and the temperature to 1000 rpm
that, when discarded, are environmentally damaging and 100°C, respectively. The antioxidant with the best
(Woma et al. 2019). Since there are no publications compatibility with the sunflower oil base was selected.
regarding the adaptation of bio-lubricants for use in
conservation and restoration, including for scientific Next, beeswax at final concentrations of 5%, 7.5%, 10%,
and industrial collections, we developed a bio-lubri- 20%, and 30% (w/w) were tested in the commercial
cant from vegetable oil based on previous reports in sunflower oil containing the antioxidant.
other research areas. Sunflower oil was selected due to The resistance to oxidation of the formulation was tested
its desirable physical characteristics with respect to its using a continuous air injection system consisting of an
kinematic viscosity, density, and pour and cloud points at EHEIM air pump (model Air 400) run at maximum flow
temperatures below 0°C. It also has a high percentage of (200 L/h for one air outlet) for 90 consecutive hours. The
unsaturated fatty acids, thus readily allowing the addition test was conducted in the dark to avoid the influence
of antioxidants to improve stability and of thickeners of light on the results. One hundred mL of sunflower
to obtain the appropriate consistency (Karmakar et al. oil containing the selected antioxidant was added to a
2017). However, as the gears are exhibited and therefore 250 mL borosilicate Duran glass bottle. A stainless steel
mostly static, the performance features of the lubricant multiport connector cap with two tubes allowed the
are not further considered herein. entrance and exit of air from the system. All equipment
was sterilized before use and two Whatman HEPA-VENT
Method filters were placed in the air inlet and outlet.
Two formulation bases were selected for the bio-lubri-
To evaluate the commercial-oil-based lubricant’s resist-
cant: a commercial virgin biological sunflower oil from
ance to photooxidation, it was exposed to direct UV
Naturefoods (purchased in the Portuguese supermarket
light at a power of 15 W for 720 consecutive hours. The
chain Celeiro) and a sunflower seed oil from Belianthus
lubricant was placed on a ~3 mm thick glass slide, and
annuus available from Sigma-Aldrich. The bio-lubricants
light was delivered using an HQ Power device.
developed from these two sources were compared to
determine whether a low-cost bio-lubricant could be The chemical changes to the bio-lubricant that occurred
developed from an easily available product. As a disad- during the oxidation tests were analyzed by Fourier
vantage of sunflower oil as a base for a lubricant is its poor transform infrared (FTIR) spectroscopy (Hamed and
resistance to oxidation, three antioxidants were tested: Allam 2006), using a PerkinElmer Spectrum 100 FT-IT
propyl gallate (PG), 4,4’-methylenebis(2,6-di-tert-butyl- instrument. Spectra were obtained between 4000 and
phenol) (MBP), and tert-butylhydroquinone (TBHQ) 600 cm−1 from formulations with and without air expo-
(Merrill et al. 2008), all purchased from Sigma-Aldrich. sure and with and without UV exposure. FTIR was also
Purified beeswax (KahlWax 8105) was selected as the used to compare the qualities of the lubricants developed
thickener due to its good compatibility when added to from the commercial and high-grade oils.
Results and discussion

Tests of the three antioxidants showed that optimal
results were obtained with TBHQ, as the solute dissolved
easily in the solvent, forming a homogeneous solution
that remained stable after heating to 100°C followed by
cooling to room temperature and then further storage.
By contrast, in the formulation containing PG, while
at ~100°C the solute was dissolved in the oil base,
after the mixture had cooled to room temperature
the solution became turbid, and 1 h later two distinct
phases, including a precipitate, had formed. Similarly,
while MBP completely dissolved in the oil base and the
mixture remained homogeneous after cooling to room
temperature, after 48 h a precipitate had formed. Thus,
the antioxidant TBHQ was selected for use in the bio-lu-
bricant (Table 1).
Table 1. Antioxidant test results of three antioxidants added to com-

mercial sunflower oil at a final concentration of 5%
Antioxidant During test During cooling After cooling

down to room down to room
temperature temperature
Figure 2. Air injection system used in the oxidation test
PG The solute The solution Two distinct
dissolved in the oil became cloudy phases, with
base, becoming a the solute
crystalline solution deposited at the
bottom
MBP Completely No changes After 48 h, it
dissolved; occurred was possible
crystalline to observe the
appearance formation of
crystals
TBHQ Homogeneous Stable Stable
crystalline solution
Beeswax was initially added to the formulation at a Figure 3. FTIR spectra of the formulation prepared using commercial
sunflower oil and TBHQ, before and after the oxidation test
concentration of 30% (w/w). After heating at 60°C–70°C
with stirring (1000 rpm), the beeswax had completely
melted and a homogeneous mixture was obtained. of TBHQ under the conditions tested. Similarly, in the
However, after cooling, the mixture had a very high hard- UV resistance tests, no changes in color or consistency
ness, with a texture that was too waxy for the consistency were visible to the naked eye and the FTIR spectra of
required for lubricating the gears. Further tests using the samples with and without light exposure did not
the same procedure showed that the best results were significantly differ (Figure 4). Therefore, the developed
achieved with concentrations of 5%, 7.5%, and 10%, all of lubricant showed good UV resistance under the condi-
which yielded a homogeneous lubricant with a viscosity tions tested.
and fluidity appropriate for use with the gears.
The bio-lubricants fabricated from the two oils showed
During the oxidation tests (Figure 2), no visible changes no significant differences in either peak intensity or peak
were recorded. In the FTIR analysis of the formulations location in the FTIR spectra (Figure 5). A comparison
with and without air exposure (Figure 3), there were no of the two bio-lubricants at a thickener concentration of
significant differences in the pattern or intensity of the 10% (Figure 6) showed only a negligible difference in the
peaks, thus demonstrating the antioxidative efficacy intensity of the asymmetric and symmetric stretching of
DEVELOPMENT OF A SUNFLOWER-OIL-BASED BIO-LUBRICANT FOR USE IN GEAR MODELS AT THE ISEP MUSEUM 53
References
Hamed, S.F. and M.A. Allam. 2006. Application of
FTIR spectroscopy in the determination of antioxidant
efficiency in sunflower oil. Journal of Applied Sciences
Research 2(1): 27–33.
Karmakar, G., P. Gosh, and B.K. Sharma. 2017. Chemi-
cally modifying vegetable oils to prepare green lubricants.
Lubricants 5(4): 44.
Figure 4. FTIR spectra of the formulation prepared using commercial
Kreivaitis, R., J. Padgurskas, M. Gumbyte, and A. Kupcin-
sunflower oil and TBHQ, before and after the UV test
skas. 2015. An assessment of beeswax as a thickener for
environmentally friendly lubricating grease production.
Lubrication Science (27): 347–58.
Merrill, L.I, O.A. Pike, L.V. Ogden, and M.L. Dunn.
2008. Oxidative stability of conventional and high-oleic
vegetable oils with added antioxidants. Journal of the
American Oil Chemists’ Society 85(8): 771–76.
Woma, T.Y., S.A. Lawal, A.S. Abdulrahman, M.A. Olutoye,
Figure 5. FTIR spectra of the commercial and high-grade sunflower oils and M.M. Ojapah. 2019. Vegetable oil based lubricants:
Challenges and prospects. Tribology Online 14(2): 60–70.
Authors
Maria Gonçalves has a bachelor’s degree in art conser-
vation and restoration and is currently a master’s student
in conservation and restoration of cultural heritage at
Universidade Católica Portuguesa (UCP), Porto. She
has a great interest in green conservation focused on the
Figure 6. FTIR spectra of the bio-lubricants developed with commercial research of new intervention methodologies, mainly for
and high-grade sunflower oils inorganic materials.
Bruno Campos holds a PhD in chemistry. He has a vast

-C-H (CH2) peaks at 2925 and 2855 cm−1, respectively.
experience in analytical methods, research, and devel-
These results demonstrated that both oils can be used
opment of organic molecules and inorganic materials.
to formulate the lubricant.
Eduarda Vieira holds a PhD in conservation and resto-
Conclusion ration of historic and artistic heritage. She is currently
an assistant professor at UCP and has been the director
The bio-lubricant developed with commercial or high-
of the UCP Research Centre for Science and Technology
grade sunflower oil, TBHQ as antioxidant, and beeswax
of the Arts (CITAR) since July 2019.
as thickener proved to be a sustainable alternative to
commercial lubricants. The physical and the chemical Patrícia R. Moreira holds a PhD in biotechnology, is an
stability of the bio-lubricant were demonstrated in air assistant professor at UCP, and coordinates the Heritage,
and UV oxidation tests. As the formulation will be used Conservation and Restoration focus area at CITAR. She
to maintain machinery for exhibition purposes, a high is interested in innovation in the field of cultural heritage
performance is not needed and was therefore not tested. with an emphasis on biodeterioration, sustainability,
However, to ensure that our newly developed product circular economy, citizen science, and green conservation.
has broader application, further tests in other settings
are needed.
MODERN METALS
A Review of Treatments of Magnesium Alloy Objects at
the National Air and Space Museum
Karen Wilcox Meredith Sweeney Lauren Horelick*

National Air and Space Museum, National Air and Space Museum, National Air and Space Museum,
Smithsonian Institution Smithsonian Institution Smithsonian Institution
Chantilly VA, USA Chantilly VA, USA Chantilly VA, USA
[email protected] [email protected] [email protected]
Abstract
This paper assesses the efficacy of treatments used to stabilize protective coatings. The objects under study were treated on a
magnesium-containing alloys at the National Air and Space case-by-case basis, and the methods and levels of intervention
Museum (NASM) over the course of a grant-funded project from varied, particularly as understanding developed during the
2012 to the present. Throughout the 20th and 21st centuries, eight years of the project. Following treatment, the objects
the desirable physical qualities of magnesium alloys have led to were kept in controlled museum storage conditions. Past
their extensive use in automotive and aerospace applications, documentation was compared to their current condition to
of which there are numerous examples in NASM’s collection. determine the efficacy of the treatments after real-time aging.
The reactivity of these alloys, which is widely acknowledged The criteria focused on the development of new corrosion or
as a key limitation for industry, is especially problematic for other signs of deterioration. Most of the objects were found to
historic collections. Yet, there is a paucity of conservation- have a relatively unchanged appearance, irrespective of the
specific examples dealing with these alloys, which this survey specific methods used to conserve them. These results factor
sought to remedy. In this study, past treatments performed into a more nuanced discussion about levels of intervention for
at NASM were re-examined, drawing out examples of various NASM’s complex composite magnesium alloy objects.
situations a conservator may encounter. It details the methods
used for surface preparation and protection, including Keywords
corrosion removal, chemical conversion, and the selection of magnesium, aerospace, corrosion, past treatments
Introduction
The desirable physical qualities of magnesium alloys have Given the widespread use of magnesium alloys over
led to their extensive use in automotive and aerospace the last century, one would expect them to be present
applications throughout the 20th and 21st centuries. in museum collections. However, as NASM sought to
However, the reactivity of these alloys is widely acknowl- develop enduring treatment approaches during the Triage
edged as a limitation for industry and presents a challenge Project, only a short list of conservation-focused literature
for conservators tasked with their stewardship. was encountered. This is curious given the relative insta-
This research offers a review of past treatment approaches bility of magnesium alloys. Treatment options and useful
by conservators at the National Air and Space Museum insights into corrosion mitigation methods are discussed
(NASM) to stabilize magnesium alloys. The examples by Dornheim (1999), Worle et al. (2010), Norquest (2013),
discussed were compiled from treatment reports prepared Taggart (2016), and Rodriguez and Ravizza (2021). These
during a Smithsonian Institution Collections Care and publications discuss the balancing act required of conser-
Preservation Fund (CCPF) grant (2012–present), known vators, who must walk a fine line between the aggressive
in-house as the Triage Project. This initiative supports corrosion mitigation measures required for this alloy and
documentation and stabilization of artifacts across the necessity to preserve integral markings and surface
NASM’s diverse collection and has included objects with finishes. Museum professionals are therefore confronted
active magnesium corrosion, a condition for which there with the challenge of finding nuanced preservation strat-
are no established conservation methodologies. egies to balance these two extremes.
A REVIEW OF TREATMENTS OF MAGNESIUM ALLOY OBJECTS AT THE NATIONAL AIR AND SPACE MUSEUM 57
The impetus for this research stems from Riddle et al.

(2019), who highlighted the necessity of reviewing past
treatments on magnesium alloys after observing the
failure of a treatment on NASM’s Surveyor spacecraft.
Corrosion re-formed within only a few years, indicating
that the methods originally used were insufficient. The
current project, therefore, sought to identify what other
treatments might also have been unsuccessful. The Triage
Project is fast paced, with an average of 300 objects
completed annually. A diversity of tools and techniques
have been employed throughout the years, and it was
anticipated that revisiting these treatments would reveal a
variety of results. This paper shares NASM’s approaches,
reflections, and treatment outcomes to discuss the unique Figure 1. (a) Light surface corrosion on a 1960s Satellite Momentum
preservation challenges of magnesium alloys. Wheel; (b) compact crust on an Engine Starter; (c) spiral threads on an
Antenna Drive; (d) voluminous corrosion on the Surveyor 8 battery cell
Magnesium alloy corrosion

Objects in the Triage Project were previously stored in
less-than-ideal conditions, which contributed to the range
and extent of the magnesium corrosion observed. The
high reactivity of magnesium alloys is due to their low
ionization energy, which means they lose electrons easily
in ambient conditions through normal aerobic corrosion
processes (Keronite 2021). A study by Green and Thickett
(1993) characterized corrosion found on museum objects
that included magnesium oxide, magnesium carbonate
hydroxide hydrate, and magnesium acetate. Magnesium
alloys are sensitive to surface contamination, such as Figure 2. Boost Control module: (a) corrosion covering the surface before
metallic particulates, chlorides, sulfates, and any foreign treatment and (b) after treatment showing retention of the surface
material trapping moisture on the surface. As these alloys paint. Apollo-era Keyboard Display: (c) dense, spherical corrosion before
treatment and (d) preserved inscriptions after treatment
do not form a passivating film, and given their sensitivity
to impurities and contamination, manufacturers aggres- recovered with thoughtful selection of surface prepara-
sively clean surfaces before any protective layers can be tion and protection methods (Figure 2). In other cases,
applied (Guo 2010). Types of historic corrosion protec- this crust was associated with severe pitting and surface
tion include anodizing, chemical conversion, plating, or loss. Öteyaka et al. (2012) used microscopy to show
organic coatings (Chen et al. 2011). that corrosion creates a complex morphology, with pits
Conservators at NASM have observed such protective ranging from shallow and wide to narrow and deep, and
surfaces disrupted by magnesium corrosion, which with subsurface cavities. This creates a complex structure
for treatment.
appears as bright-white-colored, loosely adherent, spher-
ical, acicular, or spiral products that can range from a light At NASM, it has also been observed that components
powder on relatively intact original surfaces through to within a single object can be differentially affected.
dramatic, thick crusts (Figure 1). Corrosion has been Figure 3 shows areas of spalling on a machined bezel next
found migrating through painted surfaces, conversion to light superficial corrosion on the same cast exterior.
coatings, and, in severe cases, causing blistering, flaking, Portable X-ray fluorescence spectroscopy (pXRF) of
and loss to paint layers. Original surface details such as both regions revealed similar elemental compositions,
inscriptions and evidence of casting processes might perhaps reflecting that fabrication methods influence
be unaffected by the voluminous crust and could be corrosion (Hu et al. 2014).
58 ICOM-CC | METAL 2022 | HELSINKI, FINLAND MODERN METALS
products were a multi-elemental reflection of the associ-

ated bulk alloy. However, specific corrosion mechanisms
were not investigated due to this project’s tight scope of
providing stabilizing treatments. From reviewing the 66
reports, two major stages in treatment became evident:
surface preparation and surface protection.
Figure 3. Satellite camera lens: (a) before treatment overall; (b) detail
of the machined bezel
Surface preparation
All treatments began with cleaning, reportedly aiming for
The variety of corrosion formations, their underlying
an even surface with no evidence of white-colored corro-
surfaces, the composite nature of the objects, and their
sion products and escalating the intervention until this
overall construction greatly influenced conservation
was achieved. However, in some cases, the intervention
decisions, as evidenced by the variety of past treatments
was limited by surface layers (paint, coatings) or phys-
performed during the Triage Project.
ical construction (access to the interior) and proceeded
directly to consolidation or coating. Key materials and
Methods used in the Triage Project
techniques are summarized in Table 1.
In 2020, a search of 1,782 Triage treatment documents
Typically, the removal of corrosion products, dirt, and
commenced, and 66 reports were identified with the term
contamination prepared the surface for a protective
“magnesium” mentioned in the description, condition, or
coating or, less frequently, concluded the treatment.
treatment text. Within these 66 reports, a wide variety of
treatment approaches and methodologies were recorded,
Surface protection
reflecting the experiences of the many contributors to the
project since 2012. All objects examined during the Triage Protective conversion layers are added during the fabrica-
Project underwent before-treatment written and photo- tion of magnesium alloys, but they can become disrupted
graphic documentation with a view to their stabilization by corrosion. The creation of chromate conversion layers
under curatorial guidance. Occasionally objects were was historically performed in industry and at NASM,
characterized with pXRF, revealing that loose corrosion but treatments of this type were discontinued due to
Table 1. Summary of surface preparation techniques used by Triage conservators from 2012–2019
Tool type Specifics Notes

Hand tools Stainless steel scalpels (sharp or dulled), stainless steel micro-tools, wooden Residues cleared by swabbing with solvents
skewers, or fiberglass bristle brushes
Abrasive media Walnut shell Aqueous surface cleaning after application
to remove residues
Not suitable for objects that cannot be
disassembled or have vulnerable areas that
cannot be masked off
Polishes 3M™ Mag and Aluminum Polish Residues cleared by swabbing with mineral
Nuvite™ NuShine II (Grades A, S, or C; fine to finest) spirits
Chemical: Acids o-Phosphoric acid (20% in deionized water) Applied by brushing, swabbing, or
Chemfil™ DX520 Metal Conditioner (phosphoric acid with a detergent) ultrasonic bath if components can be
Cortec™ VpCI-426 disassembled or isolated
(phosphoric acid with corrosion inhibitors) diluted with deionized water Cleared by swabbing with deionized water,
(from 1:1 up to 1:4) followed by ethanol and/or mineral spirits
Sanchem™ 560RTU Mild Acid Cleaner (glycolic acid)
Chemical: Bases Cortec™ VpCI-427 sodium hydroxide with corrosion inhibitors diluted with Same as above
deionized water (1:3 or 1:4)
OTHER METHODS TESTED
Chemical swabbing Cotton swabbing with acetone, ethanol, sodium hydroxide solution (2% w/v Minimally effective
in deionized water)
Electrolytic reduction Electrolytic reduction or softening using a sodium hydroxide electrolyte (10 V) Minimally effective
health and safety concerns. Safer formulations have specifics of the object, personal preference, and a devel-
been developed industrially, and a commercial two-step oping understanding of magnesium alloys over time.
system by Sanchem was trialed at NASM using a glycol- In 2020, the authors had the opportunity to review this
ic-based cleaner (560RTU Mild Acid Cleaner) for surface topic further by revisiting a selection of case study objects
preparation followed by a chromate-free conversion previously treated and stored at 69°F (±2) (20°C) and
coating (SafeGard CC 13062Mg RTU Rust Preventive 40% RH (±10) in the intervening years.
& Conversion Coating). Whilst the cleaner is suitable
for a range of light-metal alloys, the coating was devel- Evaluation of past treatments
oped specifically for magnesium alloys. Rodriguez and
Eight case study objects were purposefully selected
Ravizza (2021) reported the use of the Sanchem system
for re-examination to exemplify varied fabrication
on the Tiros meteorological satellite. They demonstrated methods and types of treatment. Analysis considered:
it was possible to retain the original surface layers even (a) the efficacy of the treatment methods for the desired
when walnut shell blasting was followed by the Sanchem outcome, i.e., achieving a stable surface; (b) the extent
products. This treatment was completed with a final of the treatment intervention; and (c) the longevity of
surface coating of microcrystalline wax (Figure 4). The corrosion protection.
Sanchem system has been the preferred choice of Triage
conservators since this treatment in 2018. Methodology
The presence of magnesium in each object was confirmed
with pXRF (40 kV, 10 µA, 120 s, vacuum, no filter), and
its appearance was compared with past written and
photographic documentation. One challenge was how
best to parse this documentation to draw meaningful
comparisons.
The evaluation criteria focused on:
a) alterations to the object during treatment, such as to

original surface finishes; and,
b) the post-treatment development of corrosion or other

Figure 4. Tiros meteorological satellite: (a) before treatment and (b) after;
signs of deterioration after real-time aging, including
photomicrographs (c) before (40×) and (d) after (100×)
changes in NASM-applied coatings.
Before this product was identified, the Triage treatments
This was a purely empirical evaluation of treatment
progressed directly from surface cleaning to a protective
success and “stability” based on appearance before
coating. Protective coatings used during the project are
treatment, combined with the experience that objects
summarized in Table 2.
could catastrophically re-corrode with voluminous,
Collating the treatment reports revealed an array of white-colored corrosion product within a few years after
materials and methods, as well as the general protocol for treatment—as in the case of the Surveyor spacecraft—or
surface preparation and protection. Variations reflected even in a matter of hours during a treatment.
Table 2. Protective coatings used
Material Specifics Notes

Waxes Renaissance™ microcrystalline wax Applied hot or cold, with or without buffing
Commercial oils and CRC SP-350™ (corrosion inhibitor oil) CRC SP-350™ brushed onto surface, allowing to penetrate and dry for 24
waxes containing CRC SP-400™ (a corrosion inhibitor wax) hours up to several days, followed by buffing with CRC SP-400™.
corrosion inhibitors Cortec™ VpCI®-368 (a petroleum-based film) Cortec™ VpCI®-368, applied and allowed to dry followed by buffing
Acrylics Paraloid B-72 and B48N Applied by brush in various concentrations and solvent vehicles
Cellulose nitrate Agateen™ lacquer Applied by brush on a gold-plated magnesium alloy
Results purposefully selected to include “worst-case” scenarios

where previous corrosion had been very severe or where
Table 3 organizes the eight selected case study objects.
treatment was limited by factors of the object or the
The table breaks down the relevant factors used to
conservator’s judgement.
evaluate the objects. The specific connections between
the objects’ original surface coatings, condition issues, Surprisingly, empirical observations showed that no
treatment methods, and the re-emergence of corrosion new corrosion had formed in the years after treatment,
are paramount to the discussion of intervention levels, although a longer time-depth study is warranted. The
the success of corrosion removal, and the efficacy of review of the objects resulted in three salient trends:
coatings. (a) no method completely removed all evidence of corro-
sion; (b) there are multiple suitable treatment pathways
Discussion
for magnesium alloys kept in controlled storage condi-
Based on the Surveyor experience, it was expected that tions; and (c) the coatings, in combination with stable
some treatments would have also failed. The objects were storage conditions, proved to be very effective in the
Table 3. Summary of the case study objects
Year 2020 Corrosion

Object name Original surface Condition Surface preparation Protective coating
treated re-formation
2012 Apollo Chromate Round spheres of Dry brush, scalpel; 3M™ 3 coats of Paraloid B-72 (8% No
Keyboard conversion white corrosion, polish, cleared with MS acetone/ethanol)
Display coating thick
2013 Gemini Gimbal Chromate White Cleaning: vacuum; wooden 3 coats of Paraloid B-72 (5% No
Control conversion layer, efflorescence and skewer; with brush, wooden acetone:ethanol)
white paint pitting skewer; swabbed with Nuvite
polish (grades F9/F10, C)
cleared with MS
2014 German Earth Non extant due White, powdery Scalpel, pointed tools, CRC SP-350, followed by No
Induction to corrosion corrosion cleared with MS; repeated CRC SP-400, dried over 48
Compass immediately prior to coating h, excess removed with soft
Corrosion re-appeared during cloth
treatment
2014 Momentum Conversion White, powdery Dull scalpel, natural rubber No coatings Minor
Wheel coating bloom; conversion eraser, swabbed with MS superficial
layer intact corrosion
2014 ATS 6 Satellite Chromate White corrosion Scalpel, wooden skewer, Several coats of Paraloid No
conversion layer, erupting through swabbed with ethanol and MS B-72 (5% acetone/ethanol),
red primer and surface layers, also used to consolidate
black paint on top minimal white paint
surface, yellow bloom over paint
anti-static film on
bottom surfaces
2014 Northrop P-61C Conversion Extensive Scalpel, wooden skewer, 2 coats of Paraloid B-72 (5% No
Turret Motor coating corrosion fiberglass brush, micro-tools in acetone/ethanol)
cleared with MS; swabbed
with PA (20%) cleared with DI
then MS; swabbed with 3M™
polish cleared with MS
2017 German Chromate Extensive Areas of intact conversion 2 coats of Paraloid B48N (6% No
Aircraft De-icer conversion corrosion, coating masked; walnut shell in xylene); microcrystalline
coating especially at abrasive blasting; swabbed wax around iron-alloy
contact with with acetone fasteners
dissimilar fittings
2018 Intervalometer Chromate Extensive friable Mechanical Sanchem™ products; hot No
Kit Camera conversion with corrosion microcrystalline wax
tope colored
paint
Key: DI = deionized water; MS = mineral spirits; PA = phosphoric acid
prevention of corrosion re-emergence, which is critical This object was originally treated in 2014 and did not
given then reactivity of magnesium alloys. have a coating applied. The original, brown-colored
chromate conversion layer appeared relatively intact,
Incomplete corrosion removal
and the conservator recorded their concern that further
Triage treatments reportedly aimed for surfaces cleared intervention might disrupt this thin layer. Of the case
of corrosion products prior to the surface protection. study objects, this was the only one to noticeably re-de-
Microscopic examination of the case study objects velop new, bright-white corrosion, albeit minimally over
revealed fine, white-colored material embedded in a seven-year period.
the pits. This suggests that no preparation technique
Experimentally, as NASM conservators began to expand
completely removed all corrosion products. Despite the
their techniques, they found that some materials could
apparent incomplete removal, the objects did not appear
be used with care over thin-surface conversions layers
to worsen while in controlled storage. Even with a tech-
with minimal disruption. These included Nuvite metal
nique as effective overall as walnut shell blasting, used
polish used on the Gemini-era Gimbal Control (Figure 7)
to treat the German De-icer, some minimal corrosion
and the glycolic-based Sanchem system used on the
product was still observed microscopically in the pits
Intervalometer Kit Camera.
after treatment (Figure 5). In addition, whilst all metallic
contamination must be cleaned from magnesium alloys The treatment of the ATS-6 satellite was also limited due
due to their sensitivity to galvanic corrosion, the use of to prioritizing retention of the original black paint film
metallic tools also did not seem to create loci for corro- over stabilizing the corrosion. In 2020, corrosion product
sion—likely due to solvent clearance. These observations was seen microscopically permeating through the film.
have major practical significance for all collections. However, there has been no more overt flaking or loss
of the paint due to this underlying instability, suggesting
Multiple suitable pathways
the value of a good environment, limited handling, and
In the case study objects, both minimal and interventive the protection afforded by the original paint surface.
treatments were characterized by a trend of increasing
The most compelling example for an interventive re-treat-
intervention levels as the project progressed. This reflects
ment is the P-61C Turret Motor, which was minimally
a growing vocabulary of tools, techniques, and materials.
treated in 2014. Surface cleaning ceased when the
Some of the earliest treatments were the most cautious,
conservator noted spalled metal and opted to move to
as in the example of the Momentum Wheel (Figure 6).
surface coating with an acrylic resin to limit further
material loss. A visible layer of corrosion remained extant.
Interestingly, when re-examined in 2020, this had not
worsened (Figure 8). At this time a decision was made to
remove more of this corrosion using chemical means. The
bottom portion of the artifact was submerged in Cortec
VpCI-426 (1:4 in deionized water) in an ultrasonic bath
Figure 5. German De-icer: (a) before and (b) after treatment in 2017;
(c) photomicrograph (180×) showing corrosion in the pits with the for five minutes, rinsing with deionized water, and then
wax coating
Figure 6. Momentum Wheel: (a) before and (b) after treatment in 2014; Figure 7. Gemini Gimbal Control: (a) before and (b) after treatment; (c)
(c) 2020 condition with location of photomicrographs identified: (d) 2020 condition with location of photomicrographs identified: (d) 65×
65× and (e)178× and (e) 178×
Figure 8. Northrop P-61C Black Widow Turret Motor: (a) before and (b) Figure 9. German Earth Induction Compass coated with CRC™ products:
after treatment in 2014; (c) after re-treatment in 2020 (a) before and (b) after treatment in 2012; (c) condition in 2020
with mineral spirits. Whilst more corrosion was removed

and a more even, gray surface appearance was achieved,
etching was noted around the edges. Submersion while in
the ultrasonic bath also resulted in some metal accumu-
lating in the bottom of the bath. Despite this more inter-
ventive re-treatment, minor corrosion product persisted,
and once again the treatment was discontinued out of
concern for further loss. Upon reflection, the immersive
treatment might have been more successful if the glycol- Figure 10. Intervalometer Kit Camera coated with acrylic: (a) before
and (b) after treatment in 2019; (c) condition in 2020 with location of
ic-based Sanchem cleaner had been used rather than
photomicrographs identified: (d) 65× and (e) 180×
Cortec. Both products are identified by manufacturers
as suitable for use on magnesium alloys. This inference is a less-than-ideal micro-climate than the Intervalometer
based on experience from the Intervalometer Kit Camera Kit Camera, both examples highlight the functionality
and TIROS satellite, where the Sanchem product was of commercial wax and corrosion-inhibitor coatings as
observed to be a “gentler” acid cleaner.
distinct from the acrylic coatings used in most of the
This re-treatment spurred discussion about the neces- case study objects.
sity of further intervention. Some conservators felt the
As with other facets of these Triage treatments, there
re-treatment was appropriate due to the amount of extant
corrosion. Others felt it was too interventive given the are multiple suitable pathways for successful coatings to
lack of new corrosion formation from 2012−2020, and adequately prevent corrosion re-formation.
that re-treatment also did not completely remove all Objects must be treated on a case-by-case basis. This
visible corrosion products.
discussion summarized the nuances, variety, and chal-
Utility of coatings lenges encountered with some of the complex, composite
magnesium alloy artifacts found in NASM’s collections.
Where coatings (acrylics, waxes, and commercial formu-
lations with corrosion inhibitors) were applied to objects,
no corrosion re-formed. This was true even for objects Conclusion
stored in less-than-ideal microclimates, specifically This paper details materials and methods used during
wooden enclosures. Two of the objects, the German Earth
NASM’s Triage Project from 2012 to 2019 and reviews
Induction Compass (Figure 9) and the Intervalometer
past treatments. Variations in techniques reflected
Kit Camera (Figure 10), were returned to their original
specifics of the objects, personal preferences, and devel-
wooden enclosures after treatment, and no new corrosion
or alteration of the coating was observed. This is compel- oping approaches. Earlier in the project, approaches were
ling as studies have tied magnesium alloy corrosion to more cautious, drawing from tried-and-true methods
incompatible wooden storage (Green and Thickett 1993, used for other metals and finding their suitability for
Worle et al. 2010). Although the German Earth Induction magnesium alloys, with the added caveat that the relative
Compass, treated in 2014, represents a longer duration in reactivity of these alloys requires a higher level of inter-
vention. Since 2018, the preference has been to use the paper will spur discussion about refining stewardship
Sanchem system in concert with wax coatings. practices for these alloys.
The Triage Project provided a compelling breadth of

Acknowledgments
examples for evaluation and added surety to previous
conservation methodologies. It was expected that the The authors wish to thank the National Collections Pres-
review would identify examples of treatment failures ervation grant committee, NASM staff, the numerous
due to the reactivity of magnesium alloys. Surprisingly, CCPF Triage contractors, interns, and fellows.
the case study objects mostly remained unchanged up
to seven years after treatment in a controlled environ- References
ment. While this is a short timescale for evaluation, it is Chen, X., N. Birbilis, and T. Abbott. 2011. Review of
warranted due to the experience with Surveyor’s failed corrosion-resistant conversion coatings for magnesium
treatment, where corrosion re-formed within five years. and its alloys. Corrosion 67(3): 035005-1–035005-16.
This paper demonstrated the unambiguous need Dornheim, J. 1999. Behandlung korrodierter Oberflächen
for protective coatings with examples of corrosion von Objekten aus Aluminium- und Magnesiumlegierungen,
re-forming when none was applied. The utility of coat- 257 pp. Beijing: Kexue Chubanshe. https://aata.getty.edu/
permalink/f/1kjitv/GETTY_AATA9934093547401551
ings is also evident in the case studies where objects
(accessed 6 March 2022).
were kept in their original (acidic) wooden housings
after treatment. The study concluded that no new Green, L.R. and D. Thickett. 1993. Modern metals in
corrosion developed in instances where there was still museum collections. In Saving the Twentieth Century:
extant corrosion post-treatment. This suggests that The Conservation of Modern Materials. Proceedings
minor corrosion, if both sealed and maintained in a of a Conference Symposium, Ottawa, Canada, 15–20
controlled environment, can remain stable. September 1991, ed. D.W. Grattan, 261–89. Ottawa:
Canadian Conservation Institute.
Further work Guo, K.W. 2010. A review of magnesium/magnesium
alloys corrosion and its protection. Recent Patents on
While this paper examines a small sample size over a
Corrosion Science 2: 12–21.
relatively short period of time, the findings have gener-
ated the following areas for future research: Hu, H., X. Nie, and Y. Ma. 2014. Corrosion and surface
• How much do protective coatings—particularly treatment of magnesium alloys. In Magnesium alloys:
waxes—impede future re-treatment if needed? Properties in solid and liquid sates, ed. F. Czerwinski,
• Are there other simple methods beyond visual exam- chap. 3. London: IntechOpen.
ination that could be used to qualitatively determine Keronite. 2021. How does magnesium corrode? Posted
treatment success and stability? on Keronite website, April 19, 2021. https://blog.keronite.
• To what extent can corrosion protection be proactively com/how-does-magnesium-corrode (accessed 6 March
measured in situ, using nondestructive methods? 2022).
• Can magnesium and magnesium-containing alloys Norquest, S. 2013. Spiral threads of corrosion overtake
(such as in the aluminum series) be differentiated using an antenna drive. Posted on the Smithsonian Air and
simple instrumental analysis? And how does this factor Space Museum website, July 24, 2013. https://airandspace.
into necessary levels of intervention, e.g., for purposely si.edu/stories/editorial/spiral-threads-corrosion-over-
engineered corrosion-resistant formulations? take-antenna-drive (accessed 6 March 2022).
Through the process of writing this paper, NASM now Öteyaka, M.Ö., E. Ghali, and R. Tremblay. 2012. Corro-
has an established baseline survey process based on sion behaviour of AZ and ZA magnesium alloys in alka-
visual examination coupled with pXRF and photomi- line chlorine media. International Journal of Corrosion
croscopy. This will be used for future monitoring of 2012: art. 452631.
magnesium alloys, and it is a useful starting point for a Riddle, J., E. Beesley, L. Young, and M. Collum. 2019.
much needed, longer-term study. The authors hope this Decision-making in context: Conservation of gold and
magnesium alloy components on a surveyor spacecraft.

In Objects Specialty Group Postprints, 47th Annual
Meeting in Uncasville, Connecticut, May 12–June 17,
2019, eds. K. Dodson, M. Wilcop, and J. Arista, vol. 26.
Washington DC: AIC.
Rodriguez, A. and D. Ravizza. 2021. Tough love for
magnesium. Poster presented at Transform 2021: AIC/
SPNHC 49th Joint Virtual Annual Meeting, May 10–June
24, 2021.
Taggart, C. 2016. Conserving a WWII Black Widow
compressor. Posted on the Smithsonian Air and Space
Museum, February 04, 2016. https://airandspace.si.edu/
stories/editorial/conserving-wwii-black-widow-com-
pressor (accessed 6 March 2022).
Worle, M., O. Berger, E. Hildbrand, V. Hubert, and K.
Hunger. 2010. Conservation of pigeon cameras: A collab-
orative approach between conservators and scientists.
In Metal 2010: Proceedings of the Interim Meeting of the
ICOM-CC Metals Working Group, Charleston, 11–15
October 2010, eds. P. Mardikian, C. Chemello, C. Watters,
and P. Hull, 100–106. Charleston: Clemson University.
Authors
This work was completed whilst Meredith Sweeney
and Karen Wilcox were Engen Conservation Fellows at
NASM, under the supervision of Conservator Lauren
Horelick.
Mitigating Magnesium Corrosion: Testing the Efficacy

of Reapplying Conversion Coatings in Conservation
Treatment
Abigail Elizabeth Rodriguez
Winterthur/University of Delaware Program in Art Conservation
Winterthur DE, USA
[email protected]
Abstract
Magnesium and its alloys are highly desirable structural materials methods: localized and immersion. The treatment employs two
with a propensity to corrode. Used in a variety of industrial proprietary materials: (1) Sanchem 560 Cleaner, a glycolic acid-
applications since the early 20th century, magnesium alloys based solution used to remove surface oxides, followed by (2)
can be found in objects ranging from automobiles to satellites. Sanchem SafeGard CC 13062Mg conversion coating, an adherent
Many of these objects have found their way into both museums protective layer. Analysis was conducted using scanning electron
and private collections, providing a unique opportunity for microscopy and energy-dispersive x-ray spectroscopy, x-ray
conservators to consider industry standards and proprietary diffraction, and digital microscopy. The results of the study
materials in the mitigation of magnesium corrosion. Placing supported the use of immersion treatment as it produces a more
the preservation of original material at the forefront, this study effective coating.
presents a possible treatment methodology for applying a new
conversion coating to replace worn areas of an aged coating on Keywords
a magnesium alloy (AZ31). The study examines the effects of the conversion coating, magnesium corrosion, modern metal
new coating on the original coating, comparing two application conservation
Introduction
Magnesium and its alloys have a variety of desirable advances in corrosion mitigation have prompted the
properties including a high strength to weight ratio and development of several methods to protect magnesium
a low density. For these reasons, they have been used alloys (Umehara et al. 2003, Shashikala et al. 2008).
in many structural applications across a diverse array Among these different technologies, chemical conver-
of industries. The preferable properties of magnesium sion has become the least expensive and simplest to
alloys are often offset by the metal’s high reactivity and execute (Shashikala et al. 2008, Makhlouf 2011, Frayret
susceptibility to corrosion—particularly in the presence et al. 2017). The conversion coating process changes
of humidity and contact with dissimilar metals (Figure 1). the surface of the metal into a more stable non-metallic
The galvanic corrosion of magnesium alloys is a result of compound, providing a barrier between the reactive
the standard electrode potential of the metal. surface and the environment.
Given magnesium’s predisposition to corrode, it can be When considering the efficacy of a coating, it is important
difficult to work with as a structural material. However, to note that film formation and uniformity play a crucial
role in decreasing corrosion susceptibility. The protective
layer must be uniform, well-adhered, and robust for
applications where physical wear may occur (Gray and
Luan 2002). While many factors contribute to effective
film formation and successful coatings, the number one
cause of coating failure on magnesium is poor adhesion
due to the improper removal of oxide and hydroxide
Figure 1. Photomicrographs of magnesium corrosion taken at (a) 35× layers on the metal surface (Mordike and Ebert 2001,
and (b) 100× Frayret et al. 2017). Thus, processes including cleaning,
degreasing, and oxide/hydroxide removal play a critical Table 2. Sample identification numbers
role in the application of successful coatings (Gray and Sample ID Details
Luan 2002, Umehara et al. 2003). Industry standards 1C control of bare, uncoated magnesium
in the application of conversion coatings include the 2C control of conversion coated magnesium,
CONTROLS
complete removal of previous coatings. This study aims pre-corrosion
to test two application methods for the replacement of 3C control of corroded magnesium with worn
conversion coating
worn areas of aged conversion coatings.
4C control of immersion-treated samples
The conversion coating system selected for this study 5C control of locally treated samples
is Sanchem’s 560 Cleaner (glycolic acid-based) and A Immersion-treated sample
TEST SAMPLES
SafeGard CC-13062Mg, an acid bath and a conversion B Immersion-treated sample
solution, respectively (Sanchem, Inc. 2017, Sanchem, C Immersion-treated sample
Inc. 2018). This system was chosen based on its ease of D locally treated sample
application and disposal. Sanchem’s SafeGard magnesium E locally treated sample

F locally treated sample
conversion coating and cleaner contain “no toxic mate-
rials;” the spent solution can be diluted and disposed of
Table 3. Sample preparation
in the sewer system. The final coating is a transparent
surface that can readily accept primer or topcoats. 1 A 3% v/v (recommended by Sanchem) aqueous solution of the
560 Cleaner was prepared in a small plastic bin
The alloy AZ31, a wrought magnesium alloy with 2 The samples were immersed for three minutes each at a pH
of 2.2–2.4 (recommended by Sanchem) to remove oxides and
aluminum and zinc (Table 1), was selected for this study
hydroxides on the metal surface
given its diverse applications (Dziubinska et al. 2015). 3 The samples were rinsed with deionized water, dried with
compressed air, and weighed
Experimental 4 The samples were then immediately immersed in the conversion
coating solution at pH 4.0–4.2 (recommended by Sanchem) for
Materials and sample preparation 2 min and set up to air dry
5 Once dry, the samples were weighed and photographed
Eleven samples were cut into 5E-5 m3 coupons and
die-stamped with sample identification numbers
(Table 2). Five were used as controls and six as test
samples (three for localized treatment and three for
immersion treatment). samples in a commercial humidity chamber for 48 h.
To ensure corrosion growth, an aerosolized 1% aqueous
Table 1. Chemical composition of magnesium alloy AZ31 solution of NaCl was introduced into the chamber at
Magnesium 94.2%–96.6% the 12 h, 24 h, and 36 h marks using a spray bottle. This
Aluminum 2.5%–3.5% corrosion environment was designed based on the cyclic
Zinc 0.7%–1.3% corrosion standards for automotive and aircraft mate-
Manganese 0.2%–1.0% rials in VDA 233-102 (Verband der Automobilindustrie
2013). The chamber was kept at 80%–100% RH. After
All 11 samples were polished with 600- to 1200-grit
the samples were removed, they were rinsed, air-dried,
Micro-Mesh, rinsed with acetone, and dried with
compressed air. The starting mass of each sample was weighed, and photographed.
recorded. These corroded samples were then treated using two
Six conversion-coated samples and their associated application methods of the conversion coating system:
controls were prepared to represent an original coating immersion (the industry standard) vs. localized (a
(Table 3). conservative approach) (Table 4).
The six samples and associated controls were then

Analytical techniques
subjected to a controlled corrosion environment. This
corrosion was grown by mechanically abrading areas of The analytical techniques employed in this study are
the coating on one side of each sample and placing the detailed in Table 5.
MITIGATING MAGNESIUM CORROSION: TESTING THE EFFICACY OF REAPPLYING CONVERSION COATINGS IN CONSERVATION TREATMENT 67
Table 4. Immersion vs. localized treatment procedure
Immersion Localized
• Loose corrosion products • Loose corrosion products
removed with Scotch-Brite removed with Scotch-Brite
abrasive pads, glass bristle abrasive pads, glass bristle
brushes, and stiff nylon brushes brushes, and stiff nylon
• Samples submerged for 3 min in brushes
a 3% v/v aqueous solution of the • Samples locally treated with a
acid-based 560 Cleaner brush application of a 3% v/v Figure 2. SEM images (25–26×) of (a) a bare polished sample, (b) a
• Samples rinsed, dried, and aqueous solution of the acid- converted sample, and (c) a corroded sample
weighed; weight compared based 560 Cleaner; attempted
to the weight of the samples targeting of areas of extant
before corrosion, applying a oxides and hydroxides highlight the morphological changes that occurred
“weight-loss” method to estimate without damaging the first during the coating and corrosion of the samples. The
an average corrosion rate (see coating
converted surface, as shown in Figure 2b, is smooth
Appendix A) • Samples coated by brush
• Samples coated by immersion in application and glass-like, indicating a cohesive and successful
the conversion solution for 2 min film formation. This successful coating application was
likely influenced by the polishing (the results are shown
Results and discussion in Figure 2a) and acid bath––both of which aided in
removing oxides and hydroxides from the surface.
The sample preparation phase of the study provided the
opportunity to document the conversion coating process X-ray diffraction (XRD) was used to characterize the
using a known alloy (AZ31) and a known set of prod- corrosion products formed during the sample preparation
ucts. The scanning electron microscopy (SEM) images phase. As shown in Figure 3, magnesium oxide (MgO)
in Figure 2 show the contrast between the polished, was the only detected component of the white corrosion
converted, and corroded surfaces of the magnesium collected from sample A. This data is corroborated by
samples. The images, taken at 25–26× magnification, magnesium’s propensity to form oxide and hydroxide
Table 5. Analytical techniques
Instrument Application Sample prep

Hirox digital microscopy Used to capture morphological N/A
changes during the sample preparation
and treatment phase; images were
taken of the etched/polished surfaces,
converted surfaces, corroded surfaces,
areas of coating loss, and morphology
of the new coating
Scanning electron microscopy Used to capture morphological The samples were cut into 1 × 1 cm squares and mounted on SPI Supplies
(SEM) (Zeiss EVO MA15) changes during the sample preparation Zeiss aluminum slot head stubs (12.7 × 3.1 mm) with SPI Supplies
and treatment phase; images taken double-sided carbon tabs (12 mm diameter). SPI Supplies conductive
of the cohesiveness of the coating, carbon paint (colloidal graphite in isopropanol 20% solids) was applied
corrosion boundaries, and treatment on the side and top surfaces of each sample. The samples were examined
effects using a Zeiss EVO MA15 scanning electron microscope with LaB6 source
Energy-dispersive Used to characterize the changes in the at an accelerating voltage of 20 kV for the electron beam, working
spectroscopy (EDS) chemical composition of the samples distance of approximately 10 mm, and sample tilt of 0°. The EDS data was
as they were treated with the Sanchem collected with a Bruker Nano X-flash detector 6 | 30 and analyzed with
SafeGard method Quantax 200/Esprit 1.9 software.
X-ray diffraction (XRD) Used to investigate the phase Powdered samples were collected by scraping the surfaces of the
(Rigaku D/max Rapid II with structures of the polished samples, coupons with a diamond blade scalpel. The powdered materials were
a copper anode x-ray tube converted samples, and corrosion adhered to the tip of a silicon loop sample holder and secured to the
(40 kV, 30 mA) and 0.3 mm sample stage. Each sample was analyzed in spin mode (0°–360° rotation)
collimator) at a speed of 10°/sec, and total collection time of 3 h. Rigaku RAPID/XRD
software (v.2.4.2) was used for instrument operation and data collection
and Rigaku 2DP software (v.2.0.1.1) was used to select the portion of
diffraction rings for interpretation. Rigaku PDXL 2 software (v.2.3.1.0) was
used to interpret the diffraction pattern, and the Powder Diffraction File
from the International Center for Diffraction Data (ICDD) was used as a
reference database.
layers. Subsidiary corrosion products likely included treatment was conservative, due to the decision not to use
chlorides and hydroxides, given the conditions of the strong abrasives. Industry standards recommend the use
corrosion growth. of air abrasives to dislodge corrosion from pits prior to an
acid bath. Walnut shells and plastic beads are preferred
media for reducing corrosion on magnesium-aluminum
alloys (Ravizza and Rodriguez 2019).
Depending on an object’s needs or the treatment inten-

tions, these methods can be adapted to enhance the
cleaning power of the acid bath and the application
technique for the conversion coating.
Figure 3. XRD pattern of magnesium oxide (MgO) compared to corrosion

from sample A
Figure 4. Photomicrographs (35×) of (a) a polished sample, (b) a corroded

Immersion treatment sample, and (c) an immersion-cleaned sample
Figure 4 shows photomicrographs of the magnesium
sample (B) at various stages of treatment. The immersion
treatment for the corroded samples shows significant
visual contrast before (4b) and after (4c) the acid-bath
treatment step. The sequence of the three photomicro-
graphs illustrates that the corrosion in Figure 4b was
successfully cleared from the surface to reveal areas of the
original coating and extensive pitting. There is damage Figure 5. (a) SEM image and (b) EDS elemental map (25×) of an
to the extant coating where the corrosion pustules were immersion-cleaned sample
washed away in the acid bath. While the losses to the

original coating are not ideal for a cultural heritage Localized treatment
setting, the removal of the corrosion from the surface In contrast to the immersion treatment discussed above,
and pits was necessitated for proper coating formation. the conservative approach of locally treating samples with
The SEM image and energy-dispersive spectroscopy the acid-bath solution was unsuccessful in reducing the
(EDS) elemental map in Figure 5 show that the pits on white corrosion residues on sample D (Figure 6c). This
the magnesium surface (sample 4C) have oxygen and lack of clearance was confirmed with EDS (Figure 7b),
chloride-containing elements present. Surrounded by a showing concentrated areas of oxygen and chloride-con-
smooth coated surface, the pits likely have traces of oxides taining elements on sample 5C. As stated in the section
and chlorides that were not removed during the acid above, these are likely oxides and chlorides from the
bath. This finding is important as it indicates a lowered corrosion environment. The SEM image shown in
corrosion resistance for the applied conversion coating. Figure 7a depicts layers of corrosion products protruding
These areas of pitting are now points of weakness below from the surface of the sample. Given the extent of
the coating that are susceptible to future corrosion. corrosion residues beneath the conversion coating, the
adhesion and effective lifetime of the coating are signif-
This can be corrected in future applications of this
icantly compromised.
methodology. While the samples during this study
were immersed in the acid bath without any agitation Within the context of this study, the localized treatment
or scrubbing on the surface, abrasive pads can be used approach for acid baths was developed as an experimental
to enhance the cleaning power of the bath (Ravizza and alternative to standard industry practice, or immersion
Rodriguez 2019). Additionally, the mechanical reduction treatment. This method of selectively cleaning a corroded
of the corrosion that was conducted during this mock sample to preserve extant coating elucidates several
MITIGATING MAGNESIUM CORROSION: TESTING THE EFFICACY OF REAPPLYING CONVERSION COATINGS IN CONSERVATION TREATMENT 69
key issues. The first is the inefficacy of locally treating References

corroded magnesium alloys. This method, together Dziubinska, A., A. Gontarz, K. Horzelska, and P. Piesko.
with magnesium’s propensity to corrode, creates areas of 2015. The microstructure and mechanical properties of
concentrated corrosion development along the bound- AZ31 magnesium alloy aircraft brackets produced by
aries of cleaned areas and extant coating. Additionally, a new forcing technology. Procedia Manufacturing 2:
while the localized treatment in this study did not actively 337–41.
remove the original coating, it also did not produce a
Frayret, J., J.C. Dupin, and S. Pommiers. 2017. Corrosion
viable new coating to protect the surface.
protection of magnesium alloys: From chromium VI
process to alternative coatings technologies. In Magne-
sium alloys, ed. M. Aliofkhazraei, 161–87. London:
IntechOpen.
Gray, J.E. and B. Luan. 2002. Protective coatings on
Figure 6. Photomicrographs (35×) of (a) a polished sample, (b) a corroded
magnesium and its alloys – A critical review. Journal of
sample, and (c) a locally cleaned sample Alloys and Compounds 336: 88–113.
Makhlouf, A.S.H. 2011. Current and advanced coating
technologies for industrial applications. In Nanocoat-
ings and ultra-thin films: Technologies and applications,
eds. A.S.H. Makhlouf and I. Tiginyanu, Woodhead
Publishing Series in Metals and Surface Engineering,
3–23. Cambridge: Woodhead Publishing.
Mordike, B.L. and T. Ebert. 2001. Magnesium: Properties,
Figure 7. (a) SEM image and (b) EDS elemental map (25×) of a locally
applications, potential. Materials Science and Engineering
cleaned sample
A302: 37–45.
Conclusion Ravizza, D. and A. Rodriguez. 2019. Tough love for

magnesium. Poster presented at Challenges of the Modern
While a conservative approach is traditionally taken Object, Science Museum London, 17 October 2019.
within the context of heritage preservation, this study
Sanchem, Inc. 2017. Sanchem SafeGard CC-13062Mg.
explored an interventive option––balancing the preser-
SDS, issued 5 April 2017. Chicago, IL: Sanchem.
vation of authenticity with the execution of an effective
remedy. For the passivation of modern metal surfaces, Sanchem, Inc. 2018. Sanchem 560 Cleaner. SDS, issued
traditional conservation values are often challenged to 5 December 2018. Chicago, IL: Sanchem.
provide a successful solution. The importance of acid Shashikala, A.R., R. Umarani, S.M. Mayanna, and A.K.
baths was explored through two application methods: Sharma. 2008. Chemical conversion coatings on magne-
localized and immersive. The results of the study support sium alloys – A comparative study. International Journal
the use of immersion treatment where possible and the of Electrochemical Science 3: 993–1004.
limited application of localized treatment approaches Umehara, H., M. Takaya, and S. Terauchi. 2003. Chrome-
to situations where objects cannot be disassembled for free surface treatments for magnesium alloy. Surface and
immersion. Future research could include exploration Coatings Technology 169: 666–69.
of additional conversion coating systems as well as the Verband der Automobilindustrie. 2013. VDA 233-102
incorporation of mechanical corrosion reduction prior (06/2013): Cyclic corrosion testing of materials and compo-
to treatment. nents in automotive construction. Berlin: VDA.
While these methods may not be ideal for every treat-
ment, they should be considered an example of the
Author
practical application of “tough love” to magnesium, a Abigail Elizabeth Rodriguez received her BA in art
very temperamental metal. conservation from Scripps College and is currently in her
third year of the Winterthur/University of Delaware MS

Program in Art Conservation, where she is majoring in
objects and minoring in textiles with a focus on modern
and industrial materials. Her research interests include
the intersection of industrial treatment methods and
conservation practice, alternative approaches to site-
based heritage preservation, and adhesives for plastics.
She is dedicated to promoting community-based pres-
ervation efforts, diversity, equity, and inclusion in the
conservation field.
Materials list
AZ31 – 0.0010-meter sheet from Buy Metal
https://store.buymetal.com/
Sanchem’s SafeGard CC-13062Mg and 560 Cleaner

www.sanchem.com/
Appendix A. Weights of the samples
Polished Coated Corroded Cleaned Final coated Δ after corrosion Average rate over
Sample ID
sample (g) sample (g) sample (g) sample (g) sample (g) (g) 48 h (g/min)
AZ31-A 4.472 4.472 4.475 4.463 4.463 0.009 3.12 × 10−6
AZ31-B 4.429 4.431 4.433 4.421 4.421 0.010 3.47 × 10−6
AZ31-C 4.703 4.704 4.705 4.696 4.697 0.008 2.78 × 10−6
Corrosion and Conservation of Nickel Silver Alloys
Recovered from Historic Shipwrecks
Ian D. MacLeod
Shipwrecks Museum
Fremantle WA, Australia
[email protected]
Abstract
The impact of the corrosion environment on the decay of nickel shallow and well-oxygenated wreck site off Rottnest Island. The
silver is illustrated by the contrasts between shallow waters off corrosion products were identified by X-ray diffraction (XRD)
the Western Australian coast and those found in the extremely and revealed significant dezincification and mobilisation of
deep and icy waters at the RMS Titanic (1912) wreck site. A nickel. Desalination data indicates different chloride release
19th-century nickel silver fob watch and spoon were recovered mechanisms to normal brasses. The heavy silver plate covering
from anaerobic sediments on the site of the former Fremantle the nickel silver cutlery from the Titanic was re-examined and
Long Jetty at a depth of 2.5 m. The scanning electron microscopy a new interpretation of the delamination mechanism has been
(SEM) examination of the internal parts of the fob watch revealed developed.
a corrosion matrix with high concentrations of nickel and
lesser amounts of copper and zinc. The anaerobically corroded Keywords
watch is contrasted with a set of concreted nickel silver spurs shipwrecks, corrosion, conservation, nickel silver alloys,
and chains from the wreck of the Macedon (1883) from the aerobic, anaerobic
Shallow aerobic corrosion at the SS Macedon (1883) wreck

A set of nickel silver spurs and chains (stamped with the any standard alloy but are similar to a 16% nickel silver
sovereign’s initials ‘VR’) from an ill-fated expeditionary (leaded nickel brass) used in small castings but with more
force, which left Fremantle for the Kimberley region of zinc and less nickel, which would have made for an easier
Western Australia, were recovered from the wreck of a casting (Simons 1969, 119). Wrought alloys, such as those
ship located on a reef off Rottnest Island, only 22 km used in the chains, are often known as German silver.
from its departure point. The spurs were covered with They have a white tint, are tough, malleable, ductile and
an adherent layer of concretion, while parts of the chains corrosion-resistant, and their high electrical resistance
were relatively free of marine deposits (Figure 1). There is a major factor in affecting corrosion performance
was selective corrosion on the chain links corresponding (Simons 1969). A detailed study of the oxide film on
to corrosion relief from the stress of the drawn metal, cupronickel alloys in seawater shows substitution of Ni2+
which is seen in the fine striations around the metallic into vacant positions of Cu+ in the Cu2O lattice, which
rings (Figure 2). Chemical analysis of the nickel silver increases the electrical resistance of the oxide film. The
spurs and chain, reported in Table 1, shows a composition film changes from straight Cu2O to a mixture of CuxO
of 60.2 ± 1.1% Cu, 24.2 ± 2.7% Zn and 13.0 ± 0.1% Ni,
and a copper hydroxide/oxide layer, which protects the
which is typical of nickel silver alloys of that period
underlying alloy (Jin et al. 2019).
(Simons 1969). The concretion was also analysed, and
the normalised results are also reported in Table 1. The Both nickel silver objects from the Macedon wreck come
corrosion products trapped in the concretion showed from the same section of the ternary alloy phase diagram,
that there had been selective mobilisation of zinc, along which shows that there is only a single α-phase present.
with impurities such as lead and tin from the single- The ratio of copper and nickel in the concretion is the
phase alloy. The spurs and chains do not conform to same at 0.8, which is consistent with the less reactive
found in the concretion are similar to the values for brass

objects found in an aerobic environment with low oxygen
levels (MacLeod 1985). The impact of the concretion on
the corrosion mechanism is to provide an oxygen diffu-
sion barrier to the corroding surfaces. The large amount
of iron in the concretion is due to the incorporation of
iron corrosion products from the adjacent structural steel
of the vessel. The properties of the nickel silver alloys are
sensitive to the relative proportions of zinc and nickel,
since at equal amounts of nickel and zinc the disorder
of the alloys at low concentrations of the third element
Figure 1. Concreted and corroded nickel silver spurs and chains from changes to strong order (Chadwick and Argent 1961).
the Macedon (1883) wreck
The corrosion potentials of the spurs and chains in
seawater were +0.093 and +0.055 V, respectively,
compared to the normal hydrogen electrode. The more
positive Ecorr value of the spurs may be due to a higher zinc
content (Table 1). These voltages are within ± 0.025 V of
the in situ corrosion potentials recorded on the wreck of
the Day Dawn (1886) for dezincified brass bolts which
had been submerged in shallow waters for the same time
as the other wreck. The nature of the zinc-rich corroded
surfaces on the nickel silver objects from the Macedon is
essentially the same as the zinc-rich corroded surfaces
on the brass drifts in the Day Dawn, both vessels being
shipwrecked within three years of each other. Areas of the
objects on the Macedon that were not covered in concre-
tion showed extensive dezincification of the surface. The
corrosion potentials at a pH of 8.2 are equivalent to a
Figure 2. Backscattered secondary electron SEM image of the Macedon surface concentration of 1.1 × 10−9 M Cu+ in equilibrium
chain links showing corrosion along stress lines induced during their
with a cuprite layer, i.e. very low. Infrared analysis of the
fabrication (90×)
corrosion products showed primary Cu2O with CaCO3.
Table 1. Analyses of standard nickel silver alloys and artefacts from the
The X-ray diffraction (XRD) analyses showed that the
Macedon (1883) and Long Jetty sites
principal oxidation product on the spurs was cuprite
Cu Ni Zn Pb Sn Fe (Cu2O), along with major amounts of zincian parata-
Macedon spurs 59.4 13.0 26.1 0.52 0.26 0.31
camite, previously called anarakite (Cu,Zn)2(OH)3Cl)
Macedon chains 60.9 12.9 22.3 0.31 0.10 0.55
(25-325). Areas of the spurs not protected by the concre-
Macedon concretion 17.5 3.5 9.8 0.6 0.09 4.1
concretion to 100% 49.1 9.8 27.5 1.7 0.3 11.6
tion showed extensive mineralisation and scanning
Long Jetty spoon 60.7 21.2 17.1 0.006 n.d. n.d. electron microscopy (SEM) studies indicated that there
Nickel silver 16 60 16 16 5 3 was little mobilisation of nickel but significant loss of zinc
Nickel silver 20 64 20 8 4 4 from the parent metal. Among the corrosion products, the
lemon-green nickel hydroxy chloride (NiClOH) (2-935)
was characterised with the main copper-chloride phase
nature of the parent metals, while the zinc is present in (Cu7Cl4(OH)10.H2O) (23-948). Owing to the ion exchange
the concretion at 1.1 times the alloy composition, which between amorphous FeO.OH corrosion products on the
indicates that there is preferential corrosion of the zinc 532-ton iron-screw steamer, the local surface concentra-
from the alloy, corresponding to its position in the elec- tion of sulfate was increased to the point of precipitating
trochemical series. The relative proportions of the metals mineral species such as Cu19Cl4SO4(OH)32.2H2O (8-135)
CORROSION AND CONSERVATION OF NICKEL SILVER ALLOYS RECOVERED FROM HISTORIC SHIPWRECKS 73
and the grey, white ZnSO4.3Zn(OH)2.4H2O (9-204), as the corrosion products expected at the deepest sites at
evidenced by 0.53% sulfate and 4.4 wt% chloride in the 6,000 ft. (1,829 m), with minerals such as CuCl2.2 H2O
concretion analyses. and Cu2(OH)3Cl being formed. The ratio of sulfate to
chloride in the corrosion products found on nickel silver
Desalination of the nickel silver spurs was completed after
corrosion coupons at depth was 11.7, which is much
three months of washing in 2 wt% sodium sesquicarbo-
higher than the 0.12 found in the Macedon corrosion
nate. The desalination of nickel silver objects is different
matrixes. The sulfate levels were 11% by mass and the
to binary alloys, owing to the nature of the mixed oxide
chlorides were only 0.9 wt%. Previous reports on the
coating on the surface of the ternary alloy. The chloride
decay matrixes on the deeper Titanic (1912) site also
release rate remains the same after the first plateau has
showed up a dominance of sulfate-containing corrosion
been reached, as shown in Figure 3, which contrasts
products due to electrostriction of the sulfate ions in cold
with a marked diminution for second and successive
and very deep water, which effectively increases their
washes for brass and bronze objects (MacLeod 1987).
surface concentration to the point of bringing about
The normalised chloride release rate for the Macedon
precipitation of sulfate-containing corrosion products
spurs and chains was 3.5 ± 0.2 μg.cm−2hr−½, which is
(MacLeod and Pennec 2004).
significantly lower than the mean rate for brasses of
14.2 ± 5.7 μg.cm−2hr−½. However, the total treatment The corrosion products on nickel silver alloys reported
times for the Macedon spurs and chains was the same by the US Navy at 6,000 ft. are the same as observed on
as most aerobically corroded marine brasses (MacLeod recovered nickel silver cutlery from the deeper Titanic
1987). The similarity of the chloride release slopes in both wreck site. Coupons of copper-nickel alloys produced
the washes and the lower amounts of chlorides being nickel(II) hydroxide {Ni(OH)2} as a significant corrosion
released into the sesquicarbonate solution are consistent product. It is likely that the hydroxides are precipitated
with decreased permeability of the oxide film covering due to localised increased pH associated with the cathodic
the original surface, in which a combination of mixed reduction of dissolved oxygen on the non-ferrous metal
oxide and hydroxides of nickel and copper have created alloy surface. This decay product is not unexpected as
a different chloride diffusion barrier (Jin et al. 2019). A surface corrosion of the Macedon nickel silver spurs
consequence of the different barrier is that longer wash produced the nickel hydroxy chloride {NiOHCl} from
times are needed for nickel silver alloy objects compared corrosion at less than 10 m of turbulent water, which
to straight brass artefacts. It was noted that under the would decrease the local pH to the point where copre-
standard pH 10 conditions of the sesquicarbonate cipitation of the mixed hydroxy chloride would be
wash, the molar ratio of released chloride to copper kinetically favoured. Data on copper coupons showed
was 1.5 ± 0.5; however, the lack of data on the amount a steady decrease in corrosion rate over time, and the
of nickel and zinc ions released into the wash solutions sediment decay rate was 27% lower than the open ocean
prevented detailed modelling of the processes. rate at 6,000 ft. After three years in the deep dark ocean
waters, the corrosion rate had fallen by 65%, and this is
Aerobic corrosion at depth on the RMS Titanic a common passivation process as the oxide film grows
(1912) and slows down the long-term corrosion rate, as shown
in Table 2 (Reinhart 1976).
One of the most comprehensive reports on the corro-
sion of ternary copper-nickel-zinc in deep ocean waters Table 2. Corrosion rates at 4,000 ft. in the Atlantic Ocean, US Navy
was written in 1976 by the US Navy when hundreds of (table 23, pp. 85, 86)
alloys were assessed. The normally good shallow water Corrosion rate: Corrosion rate:
Alloy
performance of copper alloys can fall away as parting sea (mm/year) sediment (mm/year)
corrosion (dezincification and denickelification) results Copper 0.041 0.030
in an uneven attack on the solid solution alloys (Reinhart CA 752 American

0.015 0.010
65% Cu, 18% Ni, 17% Zn
1976). At depths of 2,500 ft. (762 m), copper corrodes
at the same rate as at the surface, and the addition of The corrosion products on the tested cupronickel
only 8% nickel normally confers protection against alloys included tenorite, nickel oxide (NiO) and the
dezincification at 40 wt% of zinc. Ternary alloys created corresponding hydroxide (Ni(OH)2), cupric chloride,
copper(II) hydroxy chlorides and a trace of nickel sulfide

(NiS). The corrosion on the electroplated nickel silver
cutlery on the wreck of the RMS Titanic (1912) was
much more extensive than on the base metal alloys,
which were exposed for only a few years compared with
65 years at the Titanic wreck site. The nature of the passive
film in seawater changes with pressure, as seen when
the corrosion rate goes through a minimum at around
2,500 ft. (762 m), where the corrosion rate increases from
0.016 mm/year, to reach 0.051 mm/year at the depth of Figure 4. Delamination of the thick electrochemically deposited silver
4,000 ft. (1,828 m) due to changes in the nature of the layer on cutlery from the Titanic. Photo: Chemistry Centre of Western
passive surface film. If we assume a similar corrosion Australia
rate on the Titanic wreck site, it is not unreasonable to

electrostriction brought about by the near-freezing water
expect to find significant decay problems with the nickel
and high pressure which reduces their size. They also
silver objects. Recovered artefact treatments were based
on traditional electrolysis in sesquicarbonate solutions bring about accelerated pitting corrosion in the under-
(Montluçon and Lacoudre 1989). Although the results lying nickel silver alloy (MacLeod and Pennec 2004).
were initially promising, there were post-treatment issues Even though most of the tableware was recovered from
with continued internal corrosion which led to surface the debris field between the main structural elements of
exfoliation despite good storage conditions. The products the ill-fated wreck, where anoxic sediment layers abound
also indicated continued high concentrations of nantokite and create massive differential aeration problems for
within the objects (Pennec, pers. comm., 1998).
brass artefacts, these anaerobic sediments do not appear
Most of the cutlery on the Titanic was made by Elkington to have played a significant role in the decay of the nickel
in the United Kingdom and they typically coated their silver cutlery. The voltage difference between the pure
nickel silver base metal castings and mouldings with a silver and the underlying nickel silver alloy would have
35 μm layer of silver, which is readily seen as the surface been sufficient to drive significant internal galvanic
material peeling off the dinner plates shown in Figure 4. corrosion. The difference in the electrochemical reac-
Elkington silver plate was the thickest that could be tivity of the interior of the plated cutlery is like that of
bought at the time, with most other suppliers happy to a forged coin recovered from the Batavia (1629) wreck,
sell their EPNS with coatings of 15–25 μm (MacLeod where the deeply etched silver-enriched outer layer was
and Pennec 2004). For these objects, the deep water has connected by chloride ions to a tin-silver-copper alloy
exacerbated pinhole corrosion in microscopic defects in core (MacLeod 1984). During 350 years in the surf, all the
the silver plate, which has deformed due to the pressure in
debased silver-tin-copper core had corroded, with most
the hollowed-out pits beneath the silver layer (Figure 4).
decay products being lost through the etching pinholes. It
Chloride ions readily penetrate the plating owing to the
is not unexpected to have significant amounts of residual
metal left on the Titanic cutlery after only 80 years of
corrosion. It is possible that the microporous surfaces of
the Elkington plate became occluded during their initial
treatment, which left a reactive matrix of copper and
nickel chlorides trapped within the corroded base metals.
Owing to greater solubility, the zinc corrosion products
were expected to have been lost to the Atlantic Ocean.
During the normal cycling of temperature and relative
humidity in non-ideal storage conditions, moisture
would have penetrated the pores and reactivated the pits,
Figure 3. Desalination plot of chloride vs square root of time for Macedon allowing the oxidation products to mobilise and undergo
nickel silver spurs surface oxidation with concomitant precipitation.
The impact of the changing nature of the decay products

on the original surface is significant, as indicated by the
molar volumes of corrosion products. The ternary nickel
silver alloys have a density of 7.7 ± 0.5 g.cm−3, so even
the formation of cuprite, with a density of 6 g.cm−3, gives
double the molar volume. Such increases are typical for
decay matrixes that include the corrosion products of
nickel oxide (NiO), which is 50% larger than the parent
alloy. The formation of nickel(II) chloride is associated
with an increase in molar volume by a factor of 3¾, and
the heavy silver plating was readily disbonded under
such pressure. Treatment with a 50% volume aqueous
solution of acetonitrile and water for 48 h in the LP3
laboratories at Semur-en-Auxois stabilised the objects,
since they did not corrode after six months in open
storage with no relative humidity control. During this
process, the acetonitrile complexed and extracted the
copper(I) chlorides, which breaks the cyclic corrosion
process (MacLeod et al. 1981). Apart from the specific
bonding of acetonitrile with Cu+ ions, the formation of
hexacoordinated Ni2+ ions with acetonitrile will also be
effective in removing nickel corrosion products from the Figure 5. Reverse of a nickel silver fob watch from the Long Jetty site
showing the escapement adjustment
degraded matrix (Inada and Funahashi 1997).
Shallow anaerobic corrosion of a nickel silver

pocket watch
The Long Jetty extended 1 km from the shore into
the Indian Ocean from 1873 until 1897 when the new
Fremantle Harbour was opened. Owing to the exposed
nature of the Long Jetty site, localised wave action would
cause the sailing ships to move, and a vast range of
goods, including a pocket watch, were lost by the unco-
ordinated, and often drunk, sailors (Figure 5). During
examination in the conservation laboratory, the inner
workings were assessed and the whole assembly was
examined using backscattered secondary electron SEM
images (Figures 6, 7). Figure 6. Iron screw with (1) lepidocrocite, (2) copper sulfide patina,
(3) redeposited copper, (4) original surface and (5) CaCO3 deposit (90×)
The grey-black corrosion products on the surface of
the fob watch case showed that the parent nickel silver surmounted by a grey-black sulfide layer. This indicates
alloy had been in an anaerobic microenvironment for that the burial position is 180° to the image (Figure 5).
significant periods. This is consistent with the artefacts The details of the image of the cog screw blocked with
being recovered from sediment layers up to 1.8 m deep corrosion products are shown in Figure 6. A detailed
that were rich in decayed sea grass. The patina on the examination of the corrosion products adjacent to the
watch case is a thin mottled layer of chalcocite (Cu2S), outer ring rear locking plate showed that colonies of
and the main flange plate of the watch has a mixed cuprite anaerobic bacteria had penetrated the area and replicated
(Cu2O), in the bottom section, and chalcocite patina, in the as-cast dendritic structure of the watch as an isomor-
the upper section, as indicated by the red-brown patina phous Cu2S layer. A series of corrosion dimples filled with
the microenvironment, the main body of the alloys is

subject to selective corrosion of the zinc- or nickel-rich
sections of the alloy. In aerobic corrosion, the greater
solubility of zinc chloride corrosion products means that
zinc corrosion products are often concentrated in the
concretion phases attached to the objects. Underneath
the protective microenvironment of the marine growth,
zinc can accumulate to produce zincian paratacamite
((Cu, Zn)2(OH)3Cl), formerly called anarakite), which
is a strong indicator that the underlying alloy is nickel
silver. The low corrosion rate associated with nickel silver
alloys is due to the changes in the nature of the oxidation
Figure 7. SEM backscattered second electron image of the copper sulfide film on top of the original surface, since the copper(I)
layer at 900× oxide incorporates nickel ions in its structure and this
creates a barrier of higher electrical resistance, which
decay matrixes are indicators that colonies of anaerobic slows down the reduction of dissolved oxygen.
bacteria had been present inside the back of the watch
The problems of electroplated nickel silver cutlery
(Figure 7). The object was returned to the finder after
recovered from the wreck of the RMS Titanic have been
six weeks of washing in a 2 wt% sodium sesquicarbonate
compared to the way in which the internal debased
solution, which prevented a more detailed examination.
coinage on the Batavia (1629) shipwreck was eaten away
A nickel silver spoon recovered from the same location during 350 years of corrosion in the surf. Silver surfaces
as the watch showed a mixed copper sulfide patina function as excellent cathodic sites for the reduction of
and patches of metal covered with a thin oxide film. dissolved oxygen, and chloride ions readily penetrate
The higher nickel content of the spoon had minimised the microscopic pinholes in both the electroplate and
segregation attack in the anaerobic microenvironment in the depletion-etched surfaces on the forged coins.
(Figure 8; Table 1). This leads to localised pitting corrosion in the vicinity of
the defects. This mechanism became apparent with the
decay of heavily electroplated nickel silver on the wreck
of the RMS Titanic (1912), as sections of the electroplated
layer had depressions around pinholes as the pressure
squashed the noble metal to conform to the shape of the
underlying substrate. During their original electrolysis
treatment, it is possible that the plating pores above the
pits became occluded and sealed the internal structure
from conservation treatment. However, after years of
storage in ambient conditions, moisture was able to
Figure 8. Image of the sulfide-patinated Long Jetty nickel silver spoon penetrate the pores and reactivate the corrosion cycle.
(Table 1) Owing to the much lower density of the corrosion prod-
ucts, the volume expansion of the decaying metal caused
Conclusion disbondment of the silver coating. Treatment of these
objects with 50% volume aqueous acetonitrile removed
The deceptively simple microstructure of nickel silver
large volumes of copper(I) and nickel(II) chlorides and
alloys, in which both the nickel and zinc are completely
stopped corrosion in its tracks.
soluble in the copper-rich α-phase, can lead to the
false assumption that the problems of internal galvanic
Acknowledgements
corrosion associated with brass and bronze artefacts
on historic shipwrecks are not going to be present in The authors would like to pay special thanks to Stephane
such inherently homogeneous materials. Depending on Pennec, Jon Carpenter and Kalle Kasi for their assistance.
References Simons, E.N. 1969. A dictionary of alloys, 118–19.

London: Frederick Muller.
Chadwick, G.A. and B.B. Argent. 1961. Thermodynamic
properties of solid solutions. Part 4. – Copper-rich solid
Author
solutions of copper+zinc+nickel. Transactions of the
Faraday Society 57: 2138–42. Ian D. MacLeod has studied the corrosion and conser-
Inada, Y. and S. Funahashi. 1997. Solvation structure vation of metals recovered from historic shipwrecks
determination of nickel(II) ion in six nitriles using since 1978. He developed in situ conservation of iron
extended X-ray absorption fine structure spectroscopy. artefacts using sacrificial anodes and achieved significant
Analytical Sciences 13(3): 373–7. success in reducing the post-recovery treatment times.
His studies on the WWII shipwrecks of Chuuk Lagoon
Jin, T., W. Zhang, N. Li, X. Liu, L. Han, and W. Dai.
in the Federated States of Micronesia have enabled
2019. Surface characterization and corrosion behaviour
quantification of the decay mechanisms and prediction
of 90/10 copper-nickel alloy in marine environment.
of the longevity of iron shipwrecks. He previously served
Materials (Basel) 12(11): 1869. https://doi.org/10.3390/
ICOM-CC as coordinator of the Metals Working Group
ma12111869
and sat on the Directory Board for two triennial terms.
MacLeod, I.D. 1984. A genuine sixteenth century forged
coin. The Bulletin of the Australian Institute for Maritime
Archaeology 8(2): 1–9.
MacLeod, I.D. 1985. The effects of concretion on the
corrosion of non-ferrous metals. Corrosion Australasia
10(4): 10–13.
MacLeod, I.D. 1987. Stabilization of corroded copper
alloys: A study of corrosion and desalination mecha-
nisms. In ICOM-CC 8th Triennial Meeting Preprints,
Sydney, 6–11 September 1987, ed. K. Grimstad, vol. III,
1079–85. Los Angeles: The Getty Conservation Institute
for the ICOM Committee for Conservation.
MacLeod, I.D., A.J. Parker, and P. Singh. 1981. Electro-
chemistry of copper in aqueous acetonitrile. Journal of
Solution Chemistry 10: 757–74.
MacLeod, I.D. and S. Pennec. 2004. Characterisation of
corrosion products on artifacts recovered from the RMS
Titanic (1912). In Metal 2001: Proceedings of the Inter-
national Conference on Metals Conservation, Santiago,
Chile, 2–6 April 2001, eds. I.D. MacLeod, J.M. Theile,
and C. Degrigny, 270–78. Welshpool: Western Australian
Museum.
Montluçon, J. and N. Lacoudre. 1989. Les objets du
Titanic: La mémoire des abîmes. Paris: Hermé JFG.
Pennec, S., LP3 Conservation Laboratories, Semur-en-
Auxois, France, personal communication, 30 May 1998.
Reinhart, F.M. 1976. Corrosion of metals and alloys in
the deep ocean. Technical Report 834, Civil Engineering
Laboratory, Naval Construction Battalion Center, US
Department of Commerce, National Technical Infor-
mation Service, AD-AO21 279, 264 pp.
First Step in the PROCRAFT Project on WWII Aircraft
Heritage: Investigation and Conservation of the
Aluminium Alloys
Magali Brunet* Carla Martini Cecilia Monticelli Jane Echinard
Centre for Materials Elaboration Department of Industrial Corrosion and Metallurgy Study Arc’Antique Laboratory, Grand
and Structural Studies (CEMES- Engineering, University of Bologna Centre ‘Aldo Daccò’, Department patrimoine de Loire-Atlantique
CNRS), University of Toulouse Bologna, Italy of Engineering, Technopole of the (GPLA)
Toulouse, France [email protected] University of Ferrara Nantes, France
[email protected] Ferrara, Italy [email protected]
Cristina Chiavari [email protected]
Luc Robbiola Department of Cultural Heritage, Elodie Guilminot
TRACES Laboratory (UT2J-CNRS), University of Bologna Jaromír Fišer Arc’Antique Laboratory, Grand
University of Toulouse Bologna, Italia Department of Instrumentation patrimoine de Loire-Atlantique
Toulouse, France [email protected] and Control Engineering, Faculty (GPLA)
[email protected] of Mechanical Engineering, Czech Nantes, France
Andrea Balbo Technical University in Prague Elodie.GUILMINOT@loire-
Christophe Deshayes Corrosion and Metallurgy Study Prague, Czech Republic atlantique.fr
Centre for Materials Elaboration Centre ‘Aldo Daccò’, Department [email protected]
and Structural Studies (CEMES- of Engineering, Technopole of the *Author for correspondence
CNRS), University of Toulouse University of Ferrara Tomáš Vyhlídal
Toulouse, France Ferrara, Italy Department of Instrumentation
[email protected] [email protected] and Control Engineering, Faculty
of Mechanical Engineering, Czech
Elena Bernardi Technical University in Prague
Department of Industrial Prague, Czech Republic
Chemistry, University of Bologna [email protected]
Bologna, Italy
[email protected]
Abstract
The conservation of excavated wrecks or abandoned aircraft is and forging. The corrosion protection treatments applied during
of cultural and historical interest. A first step in understanding manufacturing of the aircraft parts were also examined, which
the challenges posed by this type of cultural heritage and revealed cladding with a thin aluminium foil, anodisation or
in finding solutions for its conservation is to document the the application of primer/paint. Based on typical examples
constituent materials and their subsequent alterations. of the alterations that occurred in the metal components, we
In the project PROCRAFT, aircraft wrecks from six nations demonstrate the importance of this original protection against
involved in WWII were collected and then studied. Analysis the aggressiveness of the environment.
of the elemental composition together with archival research
allowed identification of the wrought aluminium alloys used Keywords
in the non-structural and structural parts of the aircraft. Three aircraft heritage, aluminium alloys, Al cladding, protective
manufacturing processes were recognised: rolling, extrusion coatings
Introduction
Although WWII is often considered the golden age of The challenges in the conservation and protection of
military aviation, battles in the air resulted in enormous aircraft heritage include the documentation of constit-
human and material losses, with the latter including a uent historical materials and their alterations as well as
large number of wrecks on European soil and in the sea. the need for innovative approaches to the development
Volunteers and associations have long been engaged in of protective coatings and to the monitoring and control
recovering the wrecks, but despite the undeniable histor- of atmospheric conditions in storage and display areas.
ical and emotional value of WWII aircraft for Europeans, These challenges are addressed in project PROCRAFT
only recently have these heritage objects come to the (PROtection and conservation of heritage airCRAFT),1
attention of archaeologists and conservators. Nonetheless, a joint programming initiative on cultural heritage. The
they are still rarely exhibited in museums. project also aims at connecting the multiple actors in
FIRST STEP IN THE PROCRAFT PROJECT ON WWII AIRCRAFT HERITAGE: INVESTIGATION AND CONSERVATION OF THE ALUMINIUM ALLOYS 79
the operational chain, and specifically those in France, the aircraft, details about its excavation or recovery, the
Italy and the Czech Republic, from the recovery to the global condition of the aircraft (sometimes less than 1%
exhibition stage. of the whole aircraft is recovered) and the availability of
any archives linked to the aircraft. All original documents
The present study was based on the work carried out
are clearly very important to correctly classifying the
by several associations (Aérocherche, Aérothèque,
aircraft, whether historically or technically.
Ailes Anciennes Toulouse, Aéroscope and ABSA 39-45
in France; Volandia and Air Club & Fun in Italy) that In a second step, the constituent materials (aluminium
excavated or collected wrecks from various environments alloys and protective coatings) are identified and their
(terrestrial ground, sea, lakes), mainly in France and conservation state is assessed. Macro- to micro-scale
Italy, and gathered information/testimonies related to observations of the samples coupled with elemental and
the crashes. The first task of the project was to identify structural analyses are performed to extract precise infor-
and record a significant group of WWII aircraft/wrecks, mation on the elemental composition and microstructure
most of which are bombers and fighters from Germany, of the alloys as well as on the nature and composition of
France, Italy, the USA, the UK and the USSR that crashed the original protective coatings. Corrosion features are
during military operations between 1939 and 1945. Other also examined.
aircraft dating from the same historical period, including
For this study, representative samples were prepared by
civil planes abandoned after the end of their lifetime and
cutting the aircraft parts along three spatial directions:
recovered years later (such as the Dewoitine D.338), were
rolling or longitudinal (L), short transverse (i.e. thick-
considered as well.
ness, ST) and long transverse (LT), defining the main
Documentation of the artefacts, parts of wrecks or full metallurgical planes: rolling (L, LT), transverse (LT, ST)
aircraft, and their constituent materials is an important and normal (L, ST), as shown in the insert of Figure 1.
part of understanding aeronautical heritage, especially The sections were then prepared, embedded in epoxy
since archival data are difficult to recover. Moreover, resin, mechanically polished down to 2400 SiC paper
recording the condition of the artefacts (their alteration) grade and then polished on diamond paste cloths down
will provide insights into the main parameters driving to 1 µm particle size to obtain a mirror-like finish. The
corrosion. This knowledge will be used by PROCRAFT elemental compositions of the alloys were determined
to select restoration treatments, determine the appro- using a scanning electron microscope (SEM) (LaB6 Zeiss
priate protective coating for these materials and, even- EVO 25) operating at 20 keV (1 nA) equipped with a
tually, develop the appropriate preventive conservation Bruker XFLAsh energy dispersive X-ray spectroscopy
procedures and facilities (large storage hangar, outdoor (EDS) detector. Three measurements of 5 million counts
storage). on ~0.2 mm2 areas were averaged. Quantitative determi-
nation was carried out with a Quantax system (Bruker
This article focusses on the first task of the PROCRAFT
Esprit 2.1 software), applying the phi-rho-z method.
project, which was to obtain detailed information on a
The results were normalised to 100 wt% after carbon
wide set of WWII aircraft, including the identification of
deconvolution. Images in backscattering electron (BE)
wrought aluminium alloys, the principal materials used
mode completed with EDS maps were acquired for the
in the construction of aircraft, and their anti-corrosion
detection of protective coatings and the observation of
protection. A summary of the conservation state of the
corrosion features.
aircraft is provided and the relations between the alloy
manufacturing, the environment and the extent and
morphology of the corrosion are discussed.
Identification of alloys and their anti-corrosion
Methodology and experimental setup protection
A methodology was developed to uniformly record The 34 aircraft obtained thus far are from Germany (6),
information on each aircraft/wreck. Briefly, in a first France (7), Italy (5), USA (7), UK (2) and USSR (1). Five
step information is collected on the historical context, artefacts have yet to be attributed to a specific aircraft
the circumstances of the crash or the abandonment of and thus to a country. As invasive analyses could not
be performed on some aircraft, 16 artefacts for which the stringer of DO 217 and plate 2 of the Junkers 88, the
sampling could be performed are presented below. measured Mg content was at the border of Duralumin and
Super Duralumin. In these cases, the alloy was classified
Wrought aluminium alloys are used in three manu-
according to the content of other elements, Mn and Si
facturing processes (Davis and Semiatin 1989): rolled
in particular.
products (plates or sheets, mainly fuselage skin), extruded
products (stiffeners, stringers, both with a structural Duralumin (low Mg rate < 1 wt%) was used in most
role; Alderliesten 2018) and forged products (parts with French aircraft for structural and non-structural parts.
complex shapes). In addition, some stiffeners riveted to However, contemporary studies have shown that Super
plates are fabricated by rolling and folding. Duralumin was used in France only after WWII (Guil-
lemin 1946). For example, the Espadon, an aircraft
Identification of alloys
from 1948, was fabricated with Duralumin, which was
Since corrosion is directly linked to microalloying selected for both the structural and non-structural parts
elements and microstructure, identification of the alloy (stiffeners).
type in relation to the manufacturing processes is critical
In American aircraft, Super Duralumin (24S alloy) was
for establishing a diagnosis and, subsequently, conser- used for non-structural (skin) and structural parts.
vation protocols. The alloy was employed in aeronautical construction
In most of the collected and analysed samples (Table 1), already in 1937, first in the DC-3 and then more widely
the wrought alloys were exclusively Al-Cu-Mg (Si), corre- (ALCOA 1935).
sponding to the Duralumin family, nowadays referred Among British and Italian aircraft, only Duralumin has
to as Al alloys of the 2000 series. The nominal composi- been identified so far in the collected parts.
tion of these alloys is Cu (3–5 wt%), Mn (0.4–1.5 wt%),
Mg (0.5–1.7 wt%) and Si (0.5 wt%). Si is either not The two forged products in the collection are propeller
controlled or it is added in small quantities whereas blades, one from a Lockheed P38 and the other from
Fe, also present in the alloy composition, is always an an unidentified aircraft. The P38 alloy, which does not
impurity. Trace amounts of other elements (< 0.1 wt%) contain any Mg, is a 25S alloy. As noted in ALCOA’s
may occur as well: Zn, Ti, Ni, Pb. archives (ALCOA 1935), the 25S was specifically devel-
oped for forged products, with its advantage consisting
As reported in Table 1, two specific sub-families of alloys of a higher resistance to high temperatures. The alloy of
can be classified. Their denomination differs according the blade from the non-identified aircraft is Duralumin.
to country (Ouissi et al. 2019, Ouissi 2020):
• Duralumin, with a Mg content < 1 wt%: the forerunner Identification of the protection type
of the modern 2017A alloy, called Duralumin or The corrosion of aluminium alloys was an early concern
A-U4G in France (Chevigny 1954), AlCuMg1 or 3115 for aluminium manufacturers. Although paint was
in Germany (Böhle 1956), Hiduminium DU Brand in initially applied mainly as colouring/camouflage, it also
the UK (High Duty Alloys 1948) and 17S in the USA ensured corrosion protection. Primers with anti-corro-
(ALCOA aluminum and its alloys 1935); sion pigments (chromates) were applied underneath.
• Super Duralumin, with a Mg content of 1–1.5 wt%: Other types of protection in the 1920s included anodi-
the forerunner of 2024 alloy, called Duralumin FR or sation. However, due to cost issues and difficulties in
A-U4G1 in France, AlCuMg2 or 3125 in Germany, treating large areas with this process, it was replaced in
Hiduminium 72 in the UK and 24S in the USA. This the late 1920s by a method in which the alloy was cladded
alloy has a high yield, a high ultimate strength and a between two thin sheets of pure Al, called Alclad in the
high degree of hardness. USA (Dix 1927) and Vedal in France (Pubellier 1929).
For German aircraft, both Duralumin (AlCuMg1) and Most of the recovered WWII aircraft in the collection
Super Duralumin (AlCuMg2) were used in the manu- (Table 1) were originally painted but the paint has been
facture of rolled and extruded products. The choice of retained on only a few of them. In cases of severe corro-
alloy for the different parts of the aircraft was left to the sion, the original protection is missing and cannot be
manufacturers. It should be noted that on some parts, e.g. inferred (ND).
Table 1. Elemental composition (wt%) of wrought aluminium alloys and the protective coatings for aircraft from five countries
Elemental composition (wt%) Protection*/

Manufacturer / Aircraft Product / Manufacturing Identified
Thickness
(year of manufacture) / Thickness (mm) Al Cu Mg Mn Fe Si alloys
(µm)
Sheet / Rolled / 1 93.65 4.04 0.83 0.50 0.40 0.35 AlCuMg1 C / 50
Dornier / Do 217 (1943)
Stringer / Extruded / 3.5–5 92.71 4.20 1.14 0.81 0.33 0.51 AlCuMg2 ND
Heinkel / He 111 (1937) Stringer / Extruded / 2 93.21 4.29 0.61 0.75 0.34 0.46 AlCuMg1 O
Germany
Heinkel / He 177 (1943) Plate / Rolled / 1–1.4 93.39 4.43 0.71 0.55 0.29 0.41 AlCuMg1 C / 50
Messerschmitt / Bf 109 (1944) Plate / Rolled / 1.1–1.2 93.59 4.13 1.52 0.49 0.15 0.02 AlCuMg2 C / 50
Engine plate 1 / Rolled / 1.1 93.21 4.64 1.34 0.45 0.15 0.07 AlCuMg2 C / 75
Engine plate 2 / Rolled / 1.5 93.67 4.15 0.97 0.39 0.24 0.40 AlCuMg1 C / 75
Junkers / Ju 88 (1943)
Stiffener / Rolled and A
92.84 4.75 0.80 0.51 0.32 0.59 AlCuMg1
folded / 0.8
Dewoitine / D.520 O
Plate / Rolled / 0.7–2 93.20 4.35 0.50 0.56 0.31 0.46 Duralumin
(1940)
Dewoitine / D.338 Stiffener / Extruded / 2.5 94.05 3.58 0.59 0.52 0.34 0.64 Duralumin A
France
(1937) Skin / Rolled / 0.5 93.13 4.45 0.79 0.62 0.31 0.50 Duralumin C
SNCASO / C
Skin / Rolled / 2.0 93.68 4.32 0.65 0.47 0.36 0.34 Duralumin
Espadon (1948)
Latécoère / Laté 298 (1940) Plate / Rolled / 1.5 93.69 4.16 0.69 0.58 0.24 0.56 Duralumin C / 95
Plate / Rolled / 1–1.2 93.26 4.19 0.94 0.65 0.43 0.49 Hidu DU Brand ND
UK
Supermarine / Spitfire (1942) Stiffener / Rolled and ND

94.06 3.68 0.91 0.63 0.38 0.28 Hidu DU Brand
folded / 1.8
Republic Aviation / P47D Plate / Rolled / 1 92.72 4.78 1.45 0.54 0.27 0.15 24S/ C / 50
Thunder (1943) Stiffener / Extruded / 1 93.15 4.33 1.32 0.55 0.31 0.18 24S ND
USA
Lockheed / P-38 Lightning Skin / Rolled / 1 91.63 5.44 1.55 0.57 0.39 0.42 24S C / 30
(1943) Stiffener / Extruded / 1 91.63 5.64 1.51 0.62 0.31 0.30 24S A
Lockheed / P38 (< 1944) Blade / Forged / 2–112 93.58 4.57 0.05 0.70 0.44 0.53 25S A
IMAM / RO37 (1936) Plate / Rolled / 0.7 94.35 3.86 0.53 0.55 0.42 0.21 Duralumin ND
Italy
Macchi / MC202 (1942) Sheet / Rolled / 0.9 91.87 4.64 0.94 1.43 0.39 0.66 Duralumin ND
Unknown Blade / Forged / 3–66 93.70 4.05 0.59 0.64 0.40 0.24 Duralumin A
* C = Al cladding; A = anodisation; O = organic (primer/paint); ND = not determined
Nonetheless, Al cladding was largely found on rolled Several plates of the Dewoitine D.520 (1940) had not been
products, especially fuselage skin and plates. Figure 1 cladded whereas plates of the D.338, an earlier aircraft
shows an example of an Al-cladded plate in cross-sec- (1937) from the same manufacturer, had Vedal. It may
tion. For plates of ~1 mm thickness, the cladding layer have been the case that corrosion protection was chosen
is typically between 50 and 100 µm. Al cladding was according to the expected durability of the aircraft. Thus,
systematically employed in German and American Vedal was used to protect civil aircraft (such as D.338),
aircraft parts, but not always in French aircraft parts. expected to be in use for many years, while short-life
Figure 1. SEM-BE images of a

plate from a Messerschmitt
Bf109, evidencing Al cladding:
(a) full view, (b) close-up view.
Insert: main spatial directions
and defined planes of the rolled
products
fighters (such as D.520) had no cladding at all. Among the Dornier 217 fuselage (Figure 2a), which was left for
the plates recovered on British and Italian aircraft, no decades in a chasm in the Pyrenees. However, as shown
cladding was found but they were also strongly altered for the Junkers 88 (Figure 2b), the remnants of which had
and difficult to assess. been used on a farm as the roof of a poultry house, the
loss of material provided observable evidence that the Al
Some of the extruded profiles and forged products (blades)
cladding was no longer effective. On extruded profiles
had been anodised but the thin layer of aluminium oxide
exposed to atmospheric conditions such as those of the
(few microns) is sometimes difficult to distinguish from
Dewoitine D.338 (Figure 2c), the anodisation and paints
a corrosion layer.
originally present had significantly degraded. Pitting
In summary, only a few of the aircraft parts had primer and intergranular corrosion could have quickly led to
and paint as the sole protection against corrosion. Those exfoliation corrosion, as previously reported (Brunet
samples evidence the early awareness of aircraft manufac- et al. 2019). In forged products, once the protection is
turers about the low corrosion resistance of Duralumin. lost, intergranular corrosion occurs, as has been observed
in other wrought products. On artefacts recovered
General observation of the conservation state of from the sea (Figure 2d), corrosion can be severe and
the artefacts the corrosion products often develop a blue colouring
(Rocca et al. 2021).
Most of the artefacts suffered from corrosion due to their
long burial in the ground or sea or their prolonged aban- An understanding of the degradation processes requires
donment in the open air. Figure 2 shows representative the examination of corrosion in relation to the manu-
artefacts and the typically encountered alterations. Gener- facturing process and the original means of protection.
ally, Al-cladded products were more robust to burial or The environment and the history of the aircraft are key
environmental conditions. This was demonstrated by parameters that have to be considered. Depending on
Figure 2. (left) Photographs of

the artefacts. (right) Optical
microscopy images of the
alterations in the aircraft:
(a) Dornier 217 fuselage
(Al cladded), (b) Junkers 88
engine cowl (Al cladded),
(c) Dewoitine D338 stringer
(anodised), (d) propeller blade
recovered from the sea near
Brest (Britany)
the burial environment and the cause of the crash (e.g. present on the surface, while on the inner structure pits
fire), the condition of the original corrosion protection, with elements such as sulfur and phosphorus are seen.
mostly Al cladding, can vary. A case study of the engine Chlorine was also detected within the localised corrosion
cowl of a Junkers 88 is presented below. products. All of these elements indicate interactions with
soil anions. Chlorides are the most aggressive species
Correlation between the alloy manufacturing as they can induce the breakdown of the passive layer,
process, the environment, artefact history and leading to pit formation (localised corrosion).
corrosion forms An image of the pits within the Al cladding is provided
The Junkers 88, a bomber-fighter built in 1943, crashed in Figure 4, including evidence of the evolution of the
in 1944 in southern France after being shot down by a corrosion attack inside the Al-cladded plate (Figure 4b).
British Mosquito. A fire was witnessed after the crash Localised corrosion developed laterally along the surface,
(Collaveri 2016). In 2010, a few fragments were excavated, within the cladding. The depth of the corrosion corre-
including a fragment of an engine cowl. On the outside sponds to the end of copper diffusion – the result of the
plate (plate 1, AlCuMg2 alloy), large concretions are hot rolling process – inside the aluminium cladding. The
present on the surface and only residues of paint remain. resistance of Al cladding to corrosion was discussed in
The inside plate (plate 2, AlCuMg1 alloy) shows material detail by Walton et al. (1953). Pure Al cladding provides
losses, especially on areas involved in the crash. Both good electrochemical protection, even in an aggressive
natural environment. It also confers cathodic protection
plates have undergone macroscopic pitting corrosion,
of the core alloy, such that corrosion will penetrate only
which can develop into exfoliation corrosion.
as far as the core alloy or diffusion zone. Exposure of the
Figure 3 shows the SEM-BE images of the cladded Al cladding to the atmospheric environment (cycles of
Junkers 88 plate 1 in cross-section (transverse plane) humidity) results in the formation of corrosion products
and the corresponding EDS mapping. As shown by the such as aluminium hydroxides and/or sulfates. When
latter (Figure 3b), the concretions (exogenous elements) present in large amounts, they induce cracks in the Al
include localised iron corrosion products. Under the cladding layer, lifting it in some places and reaching the
concretions, thick aluminium corrosion products are bulk alloy in others.
Figure 3. SEM-BE image and

EDS maps of a cross-section
(transverse plane) of the
Junkers 88 (plate 1)
Figure 4. SEM-BE images of a

cross-section (transverse plane)
of the Junkers 88 (plate 1):
(a) pitting corrosion starting
on the Al cladding layer and
(b) corrosion progressing
within the Al-cladded plate
Figure 5. SEM-BE images of

the Junkers 88 (plate 1, rolling
plane): (a) general view,
(b) close-up showing nano-size
precipitates within the grains
In Figure 5, the corrosion has reached the bulk alloy, parts of the aircraft. However, corrosion of these Al-Cu
resulting in both intergranular and intragranular corro- alloys was a well-known issue, which resulted in different
sion. The latter develops in the presence of humidity corrosion protection strategies, most commonly cladding
and is triggered by the intermetallic precipitates located with pure Al for rolled products, anodisation or simply
mainly at the grain boundaries and by nano-size precip- an anti-corrosion primer for some extruded products.
itates inside the grains (Figure 5b). These precipitates
As a result, more than 70 years later, depending on the
suggest exposure to heat, either while the aircraft was
materials, in particular the original protective coating,
in use (proximity to the engine) or during the fire that
and on the environment at the aircraft crash site, the
occurred after the crash.
alterations in the structural alloys are highly variable but
The metallurgical observations demonstrate that an often include pitting and exfoliation corrosion. This study
aggressive environment, including the presence of chlo- focussed more specifically on the particular forms of
rine and sulfur, favours the degradation of the Al cladding corrosion that affect Al-cladded alloys. Aluminium-based
layer. Once the bulk alloy is reached, severe corrosion is alloys, even those that were protected, may further
initiated due to the intrinsic corrosion susceptibility of degrade after their removal from aggressive environments
the alloy, accelerated by cycles of humidity. (marine or soil), as corrosion can continue due to expo-
sure to cycles of humidity, which can promote exfoliation
The inside plate of the Junkers 88 (plate 2, AlCuMg1
corrosion. Thus, future work in the PROCRAFT project
alloy) also exhibits precipitation at the grain bounda-
will focus on the cleaning and stabilisation of represent-
ries and the same corrosion patterns as seen in plate 1.
ative artefacts. The low amount of chlorine elements
Although it was expected that the alloy with the higher
detected within the corroded structure will allow for the
Mg content (plate 1, AlCuMg2) would be less resistant
refinement of current research areas within the project.
to corrosion, due to the increased formation of S-type
In contrast to the dechlorination treatments used for iron
precipitates (DeRose et al. 2013), this was not the main
or copper-based metals, restoration and stabilisation will
determinant of corrosion susceptibility in this case.
require surface protection adapted to indoor or outdoor
Rather, the heat-induced precipitation at grain bounda-
environmental conditions. Innovative protective coatings
ries (present in both plates) played the larger role.
will be developed that take into account the surface condi-
Complementary investigations are still needed, including tions, i.e. alloy with or without Al cladding, the corroded
transmission electron microscopy imaging to identify the metal and the painted surfaces. They will also be tailored
nature of the nano-size precipitates and understand their to the surface state while complying with cultural heritage
role in the intragranular corrosion attack. requirements regarding material compatibility and the
preservation of the original surface aspect.
Conclusion
Acknowledgements
During WWII, different types of aluminium alloys were
used by aircraft manufacturers. In most cases, Duralumin This research was carried out as part of the PROCRAFT
or Super Duralumin (with a higher proportion of Mg) project, within the JPICH Conservation and Protection
was employed for both structural and non-structural Call, supported by the following national funding organi-
sations: Agence Nationale de la Recherche (ANR, France), T. Suter, T. Hack, and R.A. Adey, 23–38. Boston, MA:
Ministry of Universities and Research (MUR, Italy) and WIT Press.
Ministry of Education, Youth and Sports (MEYS, Czech Dix, E.H.J. 1927. ALCLAD: A new corrosion resistant
Republic). The authors would like to warmly thank the aluminum product. Technical note no. 259. Washington,
active participation of associations Aérocherche, ABSA- DC: National Advisory Committee for Aeronautics.
3945, Ailes Anciennes Toulouse, Aérothèque, Aéroscope
Guillemin, J. 1946. L’évolution de la construction aéro-
in France and Air Club & Fun and Volandia in Italy as
nautique de 1939–1946. Revue de l’Aluminium et de ses
well as museums Aeroscopia and Air and Space Museum
applications 127: 339–54.
Le Bourget (France).
High Duty Alloys. 1948. HIDUMINIUM high duty alloys.
Slough: High Duty Alloys.
Notes
Ouissi, T. 2020. Evolution of light alloys in aeronautics:
1
https://www.heritageresearch-hub.eu/project/procraft/
The case of Duralumin from its discovery to the end
of WWII. Nacelles. Past and Present of Aeronautics and
References
Space 8. http://revues.univ-tlse2.fr/pum/nacelles/index.
Alderliesten, R. 2018. Introduction to aerospace structures php?id=923 (accessed 14 March 2022).
and materials. Delft: Delft University of Technology.
Ouissi, T., G. Collaveri, Ph. Sciau, J.M. Olivier, and M.
https://doi.org/10.5074/T.2018.003
Brunet. 2019. Comparison of aluminum alloys from
ALCOA aluminum and its alloys. 1935. Pittsburgh, PA: aircraft of four nations involved in the WWII conflict
Aluminum Company of America. https://archive.org/ using multiscale analyses and archival study. Heritage
details/AlcoaAluminumAndItsAlloys_960 (accessed 14 2(4): 2784–801.
March 2022). Pubellier, M. 1929. L’emploi rationnel du Duralumin.
Böhle, F. 1956. Leichtmetalle. Berlin, Heidelberg: Springer. Revue de l’Aluminium et de ses applications 34: 931–48.
Brunet, M., L. Robbiola, P. Sciau, and M. Drieux-Da- Rocca, E., J. Tardelli, and F. Mirambet. 2021. 60 years’
guerre. 2019. Preserving aviation heritage: The need for marine corrosion of aluminium alloy 24S (2024) from
an approach based on material analysis. In Metal 2019: an historic aircraft wreck site: Implications for conser-
Proceedings of the Interim Meeting of the ICOM-CC vation. Corrosion Engineering, Science and Technology
Metals Working Group, Neuchâtel, 2–6 September 2019, 56(3): 279–88.
eds. C. Chemello, L. Brambilla, and E. Joseph, 425–31. Walton, C.J., D.O. Sprowls, and J.A. Nock. 1953. Resist-
Paris: International Council of Museums–Committee ance of aluminum alloys to weathering. Corrosion 9(10):
for Conservation (ICOM-CC) and Haute Ecole Arc 345–58.
Conservation-restauration (HE-Arc CR).
Authors
Chevigny, R. 1954. L’aluminium et ses alliages. Revue de
l’Aluminium 211: 175–9. Magali Brunet is a CNRS research scientist in mate-
rials science and since 2015 a member of the Cultural
Collaveri, G. 2016. JU88. Aerocherche.fr website. https://
and Industrial Heritage Materials team of the CEMES.
aerocherche.fr/index.php/articles-sur-les-avions/alle-
She studies the manufacturing processes and alterations
mands/ju88 (accessed 14 March 2022).
of historical aluminium alloys, using physicochemical
Davis, J.R. and S.L. Semiatin. 1989. ASM metals hand- analysis techniques. She holds a PhD in microelectronics
book, vol. 14: Forming and forging. Materials Park, OH: engineering from University College Cork, Ireland
AMS International. (2003), and a master’s degree in conservation and
DeRose, J.A., A. Balkowiec, J. Michalski, T. Suter, K.J. restoration of cultural heritage from Paris 1 – Sorbonne
University, France (2015).
Kurzydlowski, and P. Schmutz. 2013. Microscopic and
macroscopic characterisation of an aerospace aluminium Luc Robbiola is a CNRS research engineer in cultural
alloy (AA2024). In Aluminium alloy corrosion of aircraft materials science. He holds a PhD in metallurgy (1990)
structures – Modelling and simulation, eds. J.A. DeRose, and a master’s degree in conservation and restoration
of cultural heritage from Paris 1 – Sorbonne University protection in the field of cultural heritage, the corrosion
(1985). He is presently working at Toulouse University inhibition of metallic alloys in different environments,
in the TRACES laboratory. rebar corrosion behaviour in traditional and innovative
concretes and the corrosion behaviour of biomedical
Christophe Deshayes is a CNRS assistant engineer in
and industrial alloys produced by selective laser melting.
materials science. He has been working at the CEMES
since 1999. His expertise is on scanning electron micros- Jaromír Fišer has been an associate professor in machine
copy and mechanical testing. and process control in the Faculty of Mechanical Engi-
neering, Czech Technical University in Prague, since
Elena Bernardi has been an associate professor at the
2021. He received a PhD in technical cybernetics in 2004.
University of Bologna, Italy, since 2019. She received an
His main research interests are the analysis and synthesis
advanced degree in industrial chemistry in 2001 and a
of time-delay systems and industrial process control,
PhD in industrial chemistry in 2005. Her research and
including microclimate control in heritage buildings.
teaching activities are in the field of chemistry as applied
to the environment and for cultural heritage. The focus Tomáš Vyhlídal has been a professor in machine and
of her research is the effects of the environment and process control in the Faculty of Mechanical Engineering
pollutants on materials as well as on conservation of (FME), Czech Technical University in Prague, since 2012.
cultural heritage. He received a PhD in technical cybernetics in 2003. Since
2019 he has been the FME’s head of the Department of
Carla Martini has been an associate professor at the
Instrumentation and Control Engineering. His research
interests include mathematical modelling and the control
advanced degree in industrial chemistry in 1993 and a
of distributed parameter systems and processes, including
PhD in metallurgical engineering in 1996. Her research
the microclimate of heritage objects.
and teaching activities are in the field of metallurgy with
a focus on the surface engineering of metallic materials Jane Echinard is head of Arc’Antique, a conservation
and the conservation of cultural heritage metals. aboratory in Nantes, France, specialised in archaeological
artefacts from both marine and terrestrial environments.
Cristina Chiavari has been associate professor at the
She has graduate degrees from the Sorbonne University
in metal conservation (2005) and preventive conserva-
advanced degree in physics in 1997 and a PhD in electro-
tion (2012).
chemical engineering in 2003. Her research and teaching
activities are in the field of chemistry as applied to the Elodie Guilminot is a materials engineer (Polytech
environment and cultural heritage. Her research focuses Nantes, France, 1996) and in 2000 received a PhD in
on both the effects of the environment on materials and electrochemistry (Institut National Polytechnique de
the conservation of cultural heritage. Grenoble, France) for research on the conservation of
waterlogged wood/metal composites. She joined the
Andrea Balbo has been an associate professor at the
Arc’Antique laboratory (Nantes, France) as conservation
Engineering Department of the University of Ferrara,
scientist in 2006 to conduct and coordinate research
Italy, since 2020. He conducts research in the field of
projects. Her research interests include the corrosion of
materials science at the Corrosion and Metallurgy Study
metals and the development of restoration treatments.
Centre ‘Aldo Daccò’. His current research activity is
focussed on the corrosion phenomena of metal alloys and
its inhibition by means of different classes of substances
and coatings relevant to industrial applications and the
protection of cultural heritage.
Cecilia Monticelli is an associate professor in materials

science and technology in the Engineering Department of
the University of Ferrara, Italy, where she teaches mate-
rials science and technology and corrosion and protection
of metals. Her recent scientific interests include corrosion
DiscoveryMat: A Freeware Electrochemical Tool for the In

Situ Analysis of Aluminum Alloys in the Cooling Systems of
Historical Vehicles
Elodie Granget* Brice Chalançon Christian Degrigny
Haute École Arc Conservation- Musée National de l’Automobile de Haute École Arc Conservation-restauration,
restauration, HES-SO University of Mulhouse – Collection Schlumpf HES-SO University of Applied Sciences and
Applied Sciences and Arts Western Mulhouse, France Arts Western Switzerland
Switzerland [email protected] Neuchâtel, Switzerland
Neuchâtel, Switzerland [email protected]
[email protected] *Author for correspondence
Abstract
Corrosion of aluminum alloys is frequently observed in the enabling the precise identification of the problematic aluminum
cooling systems of functional vehicles dating from 1920 to 1940 alloys in real conditions directly on the vehicles. The results of
in the Musée National de l’Automobile in Mulhouse, France. these analyses showed that the corroded parts are all made of
This paper shows how DiscoveryMat, a freeware analytical Al-Cu(-Si) alloys, whereas the most recent and stable parts are
tool, has been successfully used for the in situ analysis of the made of Al-Si alloys.
materials concerned. The tool is based on the monitoring of the
electrochemical behavior of metals in different test solutions Keywords
and relies on a database to propose possible alloy compositions. DiscoveryMat, freeware analytical tool, alloy identification,
The analysis of a corpus of detached parts representative of aluminum, car, cooling system
this problem has made it possible to enrich the database, thus
Introduction
The Musée National de l’Automobile de Mulhouse’s Condition report on the studied cars
(MNAM) collection numbers over 600 vehicles, most
The purpose of the cooling system is to remove excess
of which are technically functional cars. However, only
heat from the engine. A coolant flows around the heated
a few (80) can actually be driven. They benefit from an
parts through the jacketed walls of the engine block (cast
annual maintenance service and at least one annual drive
steel) and the cylinder head (cast aluminum alloy). It
around the museum’s circuit. A condition file attached
to each vehicle is updated with information about the flows out of the engine through a system of pipes (cast
driving conditions and observations made during main- aluminum or copper alloy) and hoses (rubber) that
tenance. This allows for a reassessment of the vehicle’s connect it to the radiator, where the liquid is cooled down
functional status and, if deemed necessary, whether it by ambient air and sometimes by a fan (Figure 1). Once
should be immobilized. cooled, it returns back to the engine, where the cycle
starts again (Kennedy 1913). This coolant was originally
During these inspections, the conservation team noticed water but has since been replaced by commercial coolants
a recurrent problem with the cooling system of some (Excell –35°C [Diframa] or MB 325.0 [Mercedes Benz]),
early-20th-century cars. Aluminum alloy parts were
a mix of water, antifreeze, and other additives such as
corroding rapidly, compromising the tightness of the
corrosion inhibitors.
system. This problem was occurring on vehicles dating
from 1920 to 1940. The MNAM approached the Haute A condition report was carried out on a selection of
École Arc Conservation-restauration (HE-Arc CR), in French manufactured cars that are currently or were
Neuchâtel, Switzerland, and its research unit (UR-Arc formerly functional, focusing on the aluminum parts of
CR) to collaborate on understanding the problem. their cooling systems (Table 1).
Figure 1. Simplified illustration of an engine, with the cooling system Figure 2. Corrosion issues on a water pump (Bugatti T30) and at the
in blue junction of a pipe and a hose (Bugatti T41)
Table 1. Different sets of French manufactured car parts dating from

1921 to 1939: (1) test set and (2) validation set for DiscoveryMat
BRAND MODEL YEAR CODE PARTS SET

pipe
water pump
Type 37 1926 B37 gasket
camshaft
Bugatti
Detached part
block
1. Test set
1928– pipes
Type 41 B41
1933 water pumps
Unknown ~1930 Bx pipe
Hispano-Suiza Type H6B 1938 H6B water pump
Panhard & Dynamic water pump
1936 PDJ
Levassor c. Junior piston
pipes
water pump
2. Validation set
Attached parts
Bugatti Type 30 1922 B30

camshaft
block
pipe
Simca Gordini Type 5 1937 S5 cylinder
head
Figure 3. Metal heavily corroded under the hose of a Bugatti T28

Corrosion was observed at the junction between
aluminum alloy pipes and rubber hoses on most of these
cars. As this assemblage is often imperfectly tight, it is
Aim of the project
possible that it creates a confined zone of moisture at the
interface between the two materials. Local differences in The aim of this research project was to identify the
oxygen and ion(s) concentration are generated, inducing different aluminum alloys in the cooling systems of the
an environment favorable to the start of corrosion. A set of samples. This characterization could easily be done
gelatinous and bulky product forms, probably composed in situ with a portable X-ray fluorescence (XRF) system.
of aluminum hydroxide, and further compromises the However, this type of device is expensive and not easily
tightness of the system (Figure 2). In severe cases, the accessible for most museums. Within the scope of this
rubber cracks and the hoses must be replaced if the study, we wanted to test DiscoveryMat (Haute École Arc
vehicle is to remain functional. When disassembled, 2021), an analytical tool and free software developed by
it can be seen that the same product has developed on UR-Arc CR that provides qualitative analyses of heritage
both the outer and inner surfaces of the pipe and has metals developing thin corrosion layers. The fact that
sometimes consumed the metal significantly (Figure 3). it is also portable, low-cost, and easy to use provides
DISCOVERYMAT: A FREEWARE ELECTROCHEMICAL TOOL FOR THE IN SITU ANALYSIS OF ALUMINUM ALLOYS IN THE COOLING SYSTEMS OF HISTORICAL VEHICLES 89
institutions that acquire it and are trained in its use with As the database of aluminum alloys is not very extensive
new expertise (Degrigny et al. 2018). (only 42 entries, mostly wrought alloys), leading to prob-
ably poor results, it was essential to enrich it with alloys
Method from the automotive industry from between 1920 and
1940 in order to improve the relevance of the analysis.
DiscoveryMat is designed to monitor the corrosion
This was carried out in two steps: first, electrochemical
potential (Ecorr) of a metal over time performed in three
measurements were performed on a first set of detached
different solutions (Evian mineral water, pH = 7.2;
car parts, which are easy to handle and enable the tool’s
potassium nitrate 1 wt%, pH = 5.9; sodium sesquicar-
applicability to be tested. They were completed with
bonate 1 wt%, pH = 9.5). It then compares the plots
quantitative XRF values (three-point measurement with
obtained with those already entered into the database
a Niton XL3t XRF handheld analyzer, 50 kV, Ag anode,
of known composition determined by XRF. An algo-
Thermo Fisher Scientific) to form new entries in the data-
rithm computes the distance of similarity (d) between
base. Measurements were then carried out on a second
the plots of the material studied and those of all the
set of parts still attached to two cars, an approach more
entries in the database. It takes into consideration the
in line with the in situ analyses envisaged by the MNAM,
difference in the slope and curvature of the plots for
to validate the relevance of the enriched database.
the first 10 minutes (the initiation phase) and in the
values of the potential at the end of the measurement
Results
(the development phase) (Degrigny et al. 2010). The
smaller the d, the more relevant the match. For this The preliminary results on the detached parts show that,
study, a satisfactory match was established for d < 2000 even though d almost always exceeds the 2000 threshold,
(Granget 2020). The results are presented in the form it is often possible to identify the main elements in the
of a table of database entries classified in increasing alloy. The parts seem to be made of either aluminum-sil-
values of d, as well as superimpositions of the plots of icon alloys (Al-Si) or aluminum alloys containing silicon
the material studied and those of each database entry. (Si), copper (Cu), and sometimes zinc (Zn). According
This makes it possible to appreciate the greater or to XRF analyses, most alloys are indeed from the Al-Si
lesser similarity of the shape of the plots and to discard or Al-Cu(-Si) alloy families (Table 2).
irrelevant proposals.
DiscoveryMat measurements were made in a second
Previous work has refined the measurement protocol phase on parts still attached to two vehicles: a Simca
(Degrigny et al. 2010) in order to obtain reproducible Gordini 5 and a Bugatti Type 30. The plots were compared
and interpretable plots on heritage objects of various to those of the database, completed or not by the new
copper (Degrigny et al. 2018), silver (Fontaine 2012), or entries from the detached parts. It is clear in Table 3 that
aluminum alloys (Despland 2018). the updated database allowed better identification of the
Table 2. Identification of detached parts of H6B, PDJ, B37, and Bx using DiscoveryMat and comparison with XRF analyses; d highlighted in gray
if > 2000
DiscoveryMat XRF
Composition d
H6B a Al, Si, Cu, Zn 1319 Al 90.6, Cu 6.5, Sn 1.7, Fe 0.7, Si 0.5
H6B b Al, Si 2008 Al 88.3, Si 10.1, Cu 0.8, Fe 0.6, Zn 0.1, Pb 0.1
PDJ a Ø 7178 Al 86.9, Cu 8.9, Ni 2, Fe 1.4, Si 0.5, Pb 0.3
PDJ b Ø 5579 Al 89.2, Cu 7.6, Ni 1.8, Si 0.7, Fe 0.7
PDJ c Ø 6851 Al 76.9, Si 12.6, Cu 10, Fe 0.5
B37 a Al, Si, Cu, Zn 4136 Al 85.3, Cu 11, Si 1.3, Zn 1.1, Fe 1.1, Pb 0.2
B37 b Al, Si, Cu, Zn 2085 Al 90, Cu 7.4, Fe 1, Zn 0.8, Si 0.7, Pb 0.1
B37 d Al, Si, Cu, Zn 3897 Al 89.6, Cu 6.7, Si 1.8, Fe 1, Zn 0.8, Pb 0.1
B37 e Al, Si 2504 Al 82.7, Si 13.4, Cu 2, Ni 0.8, Fe 0.8, Mn 0.2, Zn 0.1
Bx Al, Si, Cu, Zn 1146 Al 83.2, Cu 9.9, Zn 3.2, Si 2.1, Fe 1.3, Pb 0.3
Table 3. Identification of attached parts of S5 and B30 using DiscoveryMat, with the database enriched, or not, with the entries from detached parts
and XRF analyses; d highlighted in gray if > 2000
Database without Database with

detached parts detached parts
Composition d Composition d XRF
S5 a Al, Zn, Cu, Si 2935 Al, Zn, Cu, Si 2935 Al 73.8, Si 21.3, Cu 2.8, Fe 1.3, Zn 0.3, Mn 0.3, Pb 0.2, Sn 0.1
S5 b Al, Si, Zn, Cu 4217 Al, Si, Cu, Zn 1390 Al 71.6, Si 12.9, Cu 9, Zn 4, Fe 1.3, Ni 0.5, Mn 0.3, Pb 0.2, Sn 0.2
B30 a Al, Cu, Si, (Zn or Mg) 4796 Al, Cu 8.4, Si 2.4, Fe 1.3, Zn 0.6, Pb 0.1, Ni 0.1 2651 Al 84, Cu 11.2, Si 2.3, Fe 1.2, Zn 1, Sn 0.2, Pb 0.1
B30 b Al, Zn, Cu, Si, Pb 4228 Al, Cu 10, Si 1–2, Fe 0.5, Zn 1187 Al 75.5, Cu 17.6, Zn 2.4,Si 2, Fe 1.5, Pb 0.4, Sn 0.6
B30 c Al, Si, Cu, Zn 6543 Al, Cu 11, Si 1.3, Zn 1.1, Fe 1.1, Pb 0.2 2436 Al 82.9, Cu 11.9, Si 2.3, Fe 1.4, Zn 1, Sn 0.3, Pb 0.2
B30 d Al, Cu 3284 Al, Cu 7–12, Si 1–2, Zn 0.5, Fe 1073 Al 77.5, Cu 16.5, Si 2.4, Zn 1.5, Fe 1.1, Sn 0.5, Pb 0.5
B30 e Al, Cu, Mg 2383 Al, Cu 7–12, Si 1–2, Zn 0.5, Fe 1438 Al 85.9, Cu 9.4, Si 1.7, Fe 1.8, Zn 0.9, Sn 0.2, Pb 0.1
B30 f Al, Si, Cu, Zn, Fe 6160 Al, Cu 11, Si 1.3, Zn 1.1, Fe 1.1, Pb 0.2 1870 Al 86.4, Cu10.2, Si1.5, Fe1, Zn0.7, Sn0.1, Pb0.1
alloys in most cases. Al-Cu(-Si) alloy families were accu- These results are corroborated with what can be found in
rately found and a more relevant idea of the proportions historical and technical literature. Indeed, in the 1910s,
for the main elements was given. The better performance Al-Cu alloys such as Duralumin became widely used
of DiscoveryMat for these alloys is due to the 11 new (L’Aluminium français 1924) for the production of both
entries of similar composition added to the database wrought and cast car parts, because they were lighter than
from the first set of detached parts. However, there was steel and also because the presence of copper improved
no improvement for the part “S5a,” a hypereutectic Al-Si the mechanical strength of the metal. However, it quickly
alloy that is still not represented in the database. became apparent that these alloys had poor corrosion
resistance. When used in casting, they produced porous
Discussion parts with a high chance of breaking when released from
According to the XRF analyses, the parts are composed the mold (Guillemot 1933). This may explain why, during
of three main types of alloys from this period: hypereu- the 1910s and 1920s, in parallel with the search for a better
tectic Al-Si alloys, eutectic Al-Si alloys with or without aluminum alloy composition for casting, efforts were made
Cu, and Al-Cu alloys ranging from 4 to 12% Cu and to further lighten steel for engine castings (Bodet 1931).
containing Si, Zn, Ni, or Sn. Moreover, all corroded parts Therefore, it is not surprising to find that the Al-Cu parts
are made of Al-Cu alloys or contain large proportions from the cars were heavily corroded. These problematic
of Cu (Figure 4). alloys can now correctly be identified by DiscoveryMat.
Figure 4. Proportions of Cu and

Si in the alloys of the studied
parts
DISCOVERYMAT: A FREEWARE ELECTROCHEMICAL TOOL FOR THE IN SITU ANALYSIS OF ALUMINUM ALLOYS IN THE COOLING SYSTEMS OF HISTORICAL VEHICLES 91
In the 1920s, the family of Al-Si alloys, namely Alpax, internship was funded by the Swiss-European Mobility
eutectic Si = 13 wt% (L’Aluminium français 1926), and Program grant. They would also like to thank HE-Arc
hypereutectic, Si > 13 wt% (L’Aluminium français 1924), CR and Microcity in La Chaux-de-Fonds, Switzerland,
was created. Silicon conferred aluminum mechanical for providing access to facilities and equipment, as well
properties similar to those provided by copper but as the supervision required to conduct this study. Finally,
better corrosion resistance and casting properties. From they would like to acknowledge Mr. Christian Vargel†,
the second half of the 1920s, they slowly found their an expert in aluminum corrosion, and Mr. Claude Riss,
place in the automotive industry, more specifically in from the Institute for the History of Aluminum, for their
the casting of more complex engine parts subject to expertise during the condition assessment.
friction or thermal stresses, such as pistons, cylinder
heads, and cooling systems’ parts (Guillemot 1933). This References
information validates the results from the XRF analyses, L’Aluminium français. 1924. L’Aluminium dans la carros-
showing that all the latest and more stable parts from the serie automobile. La revue de l’Aluminium 3: 47–50.
selected cars were made of eutectic and hypersilicated
L’Aluminium français. 1924. L’Aluminium dans le châssis
Al-Si alloys. For example, the Hispano-Suiza water pump
automobile. La revue de l’Aluminium 3: 35–46.
had two parts (H6B a & b), one of which (HB6B b) was
surprisingly not corroded. The analyses showed that this L’Aluminium français. 1926. L’Alpax. La revue de l’Alu-
part was made of Al-Si alloy (Talbe 2); according to the minium 15: 240–47.
conservator, it was probably replaced a few years ago. A Bodet, J. 1931. L’emploi des métaux légers rend plus
precise identification of these two alloy families is not économiques les moyens de transports. La Science et la
yet possible with DiscoveryMat but could be achieved vie 39(166): 333.
by adding more entries to the database.
Degrigny, C., G. Guibert, S. Ramseyer, and G. Rapp,
and A. Tarchini. 2010. Use of Ecorr vs time plots for the
Conclusion
qualitative analysis of metallic elements from scientific
This study demonstrates the suitability of DiscoveryMat and technical objects: The SPAMT Test Project. Journal
as an in situ analytical tool for the qualitative identifi- of Solid State Electrochemistry 14(3): 425–35.
cation of aluminum alloys. With the new entries added Degrigny, C., E. Menart, and G. Erny. 2018. Easy-to-use,
to DiscoveryMat’s database, it was possible to accurately low-cost electrochemical open-source hardware to
identify the Al-Cu alloy family as well as some secondary analyze heritage metals: Possibilities and limits. Current
elements of materials of similar composition. Topics in Electrochemistry 20: 15–23.
The cars in the set of samples studied have aluminum Despland, C. 2018. Identification des alliages d’alu-
alloys in their cooling systems that are representative of minium d’objets ethnographiques et caractérisation de
the transition period between 1920 and 1940: Al-Cu(-Si) leurs produits de corrosion : analyse par voie électro-
was used in older vehicles, whereas Al-Si was slowly chimique et par fluorescence X. Bachelor’s thesis, He-Arc
introduced during the 1930s. All the heavily altered Conservation-restauration, Switzerland.
parts were made of Al-Cu(-Si) alloys, which have lower
Fontaine, C. 2012. Analyse qualitative par voie élec-
corrosion resistance.
trochimique des alliages à base d’argent : application
In addition to alloy identification, further study of the de l’outil CLAMTEC à ce type de matériau. Bachelor’s
corrosion process that occurs at the interface between thesis, HE-Arc Conservation-restauration, Switzerland.
rubber and metal is necessary to better understand and Granget, E. 2020. Corrosion des alliages d’aluminium des
prevent this phenomenon. circuits de refroidissement à eau de véhicules en contexte
patrimonial. Master’s thesis, HE-Arc Conservation-res-
Acknowledgments tauration, Switzerland.
The authors would like to thank the Musée National Guillemot, L. 1933. Congrès mondial de Fonderie (Paris,
de l’Automobile de Mulhouse for hosting the main 13–18 septembre 1932). Bulletin de la société d’encour-
author during completion of her master’s thesis. This agement pour l’industrie nationale, 310–20. Paris: BSPI.
Haute École Arc. 2021. DiscoveryMat [online]. Available

at http://discoverymat.he-arc.ch:8080/ (accessed 28
October 2021).
Kennedy, R. 1913. The book of the motor car: A compre-
hensive and authoritative guide on the care, management,
maintenance, and construction of the motor car and motor
cycle, vol. 2, 1st ed. London: Caxton Publishing Company.
Authors
Elodie Granget graduated in 2020 from the master’s
program at HE-Arc CR in Neuchâtel, Switzerland, where
she specialized in metal conservation. Since 2021, she
has been working as a research assistant at HE-Arc CR.
Brice Chalançon graduated in mechanical engineering

from Polytech Orleans, France, in 2008, and completed
his master’s in conservation of technical, scientific, and
horology objects at HE-Arc CR in 2019. Since 2008, he
has been director of the workshop at the Musée National
de l’Automobile de Mulhouse, France.
Christian Degrigny received a PhD in analytical

chemistry from the University of Paris IV, France, in
1990. Since 2004, he has taught and conducted applied
research projects at HE-Arc CR and has specialized in
electrochemical diagnosis and treatment of historic and
archeological metals.
COATINGS, CONSOLIDANTS &
CORROSION INHIBITORS
A Comparison of the Long-Term Outdoor Performance
of Two Modern Paint Coating Systems and a Traditional
Lead-based Paint Applied to Historic Wrought Iron
Peter Meehan* Nicola Emmerson David Watkinson
School of History, Archaeology and School of History, Archaeology and School of History, Archaeology and
Religion, Cardiff University Religion, Cardiff University Religion, Cardiff University
Cardiff, UK Cardiff, UK Cardiff, UK
Abstract
A real-time, five-year outdoor study compares the performance underwent minor but visually undetectable changes and an
of three paint systems (traditional oil-based, acrylic, epoxy epoxy/acrylic coating remained unchanged. The oxide layer in
resin) applied to naturally corroded wrought iron, with their St3 controlled adherence of the coatings.
surfaces prepared to Swedish Standard Sa 2.5 or Steel Surface
Preparation Standard St3. Interim data at two years reports Keywords
on gloss, colour change, pull-off testing and electrochemical protective coatings, wrought iron, electrochemical impedance
impedance spectroscopy measurements. Traditional oil-based spectroscopy (EIS), colorimetry, pull-off testing, gloss
paint exhibited significant visual change, a modified alkyl paint
Introduction
Conservation of historic ironwork used commercially have performance characteristics that
require adherence to defined standards to guarantee
Wrought iron corrodes readily in the atmosphere to
their service life. These normally involve specific surface
produce permeable, non-protective corrosion products
preparation and coating application methods, as well as
that laminate and spall, a process enhanced by regular
defining the necessary environmental parameters for a
wet and dry climate events (Hoerlé et al. 2004). Coatings
successful coating application. Fulfilling much of this
are commonly used to control corrosion, but since their
guidance within historic contexts is difficult due to the
limited lifespan makes this a costly process, research in
corroded nature of the material, limitations imposed by
the field aims to identify the best-performing coatings
differing ethical and aesthetic standards and the cash-
to support cost/benefit decisions. Decision-making in
poor state of the sector.
the selection of coatings, surface preparation methods
and application protocols within heritage is mostly Ethical constraints and numerous variables make
based on best practice and well-established techniques designing experimental research to replicate the reality
(Blackney and Martin 1998). Quantitative evidence of of conservation practice difficult to achieve, and compro-
coating performance is more often found in commercial mises may need to be sought. A major advance would
or industrial research. Unfortunately, rationales and be to identify the ‘weak link(s)’ in the variable chain, i.e.
acceptable methods and outcomes in these sectors are essentially identify which variable(s) dictate(s) coating
often governed by ethical and aesthetic decision-making performance the most and, if possible, develop this
processes that differ from those used in historic contexts. process to provide a rank order. This would facilitate
informed decisions when balancing factors like ethics
Conservation practices for coating historic iron may use
and aesthetics against longevity and cost/benefit.
commercial standards alongside methodologies such
as flame cleaning, where variable control lies with the While real-time testing and a representative sample
individual (Emmerson et al. 2019). This does not align set are essential to underpin quantitative prediction of
with the needs of coating manufacturers. Many coatings coating lifespan, these factors can only offer an indication
A COMPARISON OF THE LONG-TERM OUTDOOR PERFORMANCE OF TWO MODERN PAINT COATING SYSTEMS AND A TRADITIONAL LEAD-BASED PAINT APPLIED TO HISTORIC WROUGHT IRON 95
of coating performance in the environment adopted for This preliminary paper reports on the results of the first
the test procedure. Some quantitative methods such as 24 months of outdoor exposure of the samples. Oxygen
electrochemical impedance spectroscopy (EIS) (Cano consumption method (Matthiesen 2013) and results are
et al. 2010) can provide useful comparative data, but not reported as it is too early for meaningful data to be
this cannot be extrapolated to provide predictive insight presented.
into service life performance due to the essential but
unrealistic test conditions and an absence of real-time Method
testing. Visual change, sometimes scaled according to
Sample material
international standards, is useful, but the qualitative data
this produces may limit insight into the relative extent of Samples were produced from two rolled wrought iron
failure and the precise prediction of coating performance. sheets from a 19th-century gasometer. Although wrought
Following the concept of assigning value to both quan- iron is inhomogeneous by nature (Dillmann et al. 2004),
titative and qualitative indicators of performance, the taking samples from just two rolled plates minimised the
study reported here employs a mixture of quantitative and risk of compositional differences. The sheets had thick-
qualitative assessment techniques to produce an integrated nesses of 4 and 6 mm and were cut using a hydraulic guil-
insight into coating performance. Real-time testing of lotine to avoid microstructural changes from heat-gen-
historic wrought iron, with surfaces prepared according to erating cutting processes. Corrosion during exterior
two commonly used techniques within historic wrought exposure has generated corrosion products including
iron preservation, is used to study the performance of three magnetite, goethite and lepidocrocite on the surface as
coatings that represent traditional and modern approaches identified by x-ray diffraction (Emmerson et al. 2019).
within the sector. A single-pack, oil-based alkyd and a
two-pack epoxy resin/acrylic urethane are compared with Surface preparation
a lead-based linseed oil system traditionally used for the The samples were prepared to two levels of cleanliness
long-term protection of iron. Their performance is assessed prior to coating (Table 1). A total of 96 samples were
using a range of criteria including their longevity, ease of cleaned to Swedish Standard Sa 2.5 (ISO 8501-1:2007;
use, reversibility and health and safety issues. Hopefully, International Organization for Standardization 2007)
the results will allow a reasoned assessment of their perfor- using abrasive blast cleaning to achieve a near-white
mance, so that the conservator can make informed choices blast-cleaned surface (Joint Surface Preparation Standard
about the materials and methods to use when painting SSPC-SP 10; The Society for Protective Coatings 2007);
historic ironwork. 144 samples were cleaned to a Steel Surface Preparation
Standard St3 (ISO 8504-3:2018), achieved by thorough
Aims and objectives
hand and power tool cleaning to leave the metal surface
Aim:
• To determine the long-term, real-time performance
of two modern protective coating systems and one
traditional system for exterior ironwork.
Objectives:
• To coat samples of historic iron and control samples
of glass with one of three coating systems.
• To assess qualitatively the application of the coatings to
historic wrought iron relative to conservation practices.
• To expose the coated samples to outdoor conditions
for a period of five years.
• To assess the visual and protective properties of the
coating systems by colorimetry, gloss measurement,
pull-off testing, EIS and oxygen consumption at Figure 1. Sample surfaces uncleaned, cleaned to Sa 2.5 and St3 and
12-month intervals throughout the study period. following application of the three coating systems
96 ICOM-CC | METAL 2022 | HELSINKI, FINLAND COATINGS, CONSOLIDANTS & CORROSION INHIBITORS
Table 1. Details of all samples as prepared for the range of analysis Characterisation of coatings as applied
techniques
Following application, the coating thickness, gloss and
Surface
No. of Dimensions Coating Analysis colorimetry measurements were recorded for each
preparation
samples (mm) system methods
standard coating. A control sample of the coating systems is
8 Sa 2.5 stored in the dark throughout the test period to allow
Oxygen
8 40 × 40 × 6 St3 for comparison with samples undergoing outdoor
consumption
8 Glass
exposure.
40 Sa 2.5
1
40 60 × 60 × 4 St3 EIS; gloss
Outdoor exposure
40 Glass
5 Sa 2.5 Colorimetry; Samples are mounted between strips of extruded ‘H’
150 × 150 × 4
5 St3 pull-off testing section PVC fixed to a stainless steel frame angled at
8 Sa 2.5 60° to the horizontal (Figure 2). Each sample frame
Oxygen
8 40 × 40 × 6 St3
consumption is fixed off the ground in a south-facing direction.
8 Glass The samples are exposed in a location whose condi-
40 Sa 2.5 tions are between a low (C2) and medium (C3) risk
2
40 60 × 60 × 4 St3 EIS; gloss environment category (BS EN ISO 12944-2:2017;
40 Glass International Organization for Standardization 2017)
5 Sa 2.5 Colorimetry; on the edge of a small inland town. Onset HOBO data
150 × 150 × 4
5 St3 pull-off testing
loggers record air temperature, relative humidity (RH),
8 St3 Oxygen
40 × 40 × 6 surface wetness and cumulative light exposure in real
8 Glass consumption
time throughout the outdoor sample exposure period.
8 St3
60 × 60 × 4 3 EIS; gloss This data will be collated to relate weather patterns to
8 Glass
coating performance.
5 Sa 2.5 Colorimetry;
150 × 150 × 4
pull-off testing
with a sheen appearance. Control samples of float glass

cut to the same dimensions, oil gilded and coated in
the same manner as the wrought iron samples allow for
examination of changes in the coatings over time inde-
pendent of the wrought iron substrate.
Coating
Sample coupons were brush painted with one of the three Figure 2. Samples mounted on their exposure frame with the
paint systems chosen (Table 2). environmental monitoring equipment fitted
Table 2. Details of the coating systems as applied to the samples
Coating No. Minimum dry film

Manufacturer Type Coating
system coats thickness (µm)
Kem-Kromik 489 zinc phosphate primer 2 75
Sherwin-Williams Protective
1 Alkyd Sherwin L654 micaceous iron oxide midcoat 1 75
and Marine Coatings
Kem-Kromik 530 gloss topcoat 2 30
Macropoxy L425 zinc phosphate primer 2 75
Epoxy Macropoxy 267 high-solid micaceous iron
2 Sherwin-Williams 1 125
oxide midcoat
Acrylic urethane Acrolon C237 sheen finish topcoat 2 50
Lead-based linseed oil Red lead primer 2
3 Traditional
(under licence) Lead carbonate finish coat 3
Assessment of coating performance reference electrode were inserted into the top. The circuit
was completed by connecting the sample terminal to form
Coating performance, both protective and aesthetic,
the working electrode. Synthetic rain was chosen as the
is assessed annually using oxygen consumption in a
electrolyte to match the real-time exposure environment
fixed RH, EIS, pull-off adhesion testing, colorimetry,
more closely. This was developed for the assessment of
gloss measurement and digital photography. As EIS and
copper alloys and steels in outdoor environments (Letardi
pull-off testing are destructive techniques, these samples
et al. 2016). It contains several soluble salts and is used at
(Table 1) are not returned to outdoor exposure. Oxygen
a 10× concentration. The analysis was carried out using
consumption samples are returned to outdoor exposure
an Ametek/Princeton Applied Research PARSTAT 3000A
after measurement, being remeasured each year.
potentiostat/galvanostat instrument, with the acquired
Electrical impedance spectroscopy data being processed using Rhd Instruments RelaxIS 3
impedance analysis software.
EIS can be used to characterise the behaviour of an
electrical system by applying a small electrical excitation Pull-off adhesion testing
signal (sinusoidal voltage) to the system and measuring
Adhesion of the coating systems was examined by pull-off
the response (current) created (Loveday et al. 2004,
testing according to BS EN ISO 4624:2016 (British Stand-
Orazem and Tribollet 2017). The frequency of the signal
ards Institution 2016). This measures the force required
is varied, and the resulting responses can be analysed
to pull-off a specified diameter of coating using hydraulic
using electrical methods (Cogger and Evans 1999). EIS
pressure. The technique records the highest pressure
has been used as a tool to test varying coating types since
required to detach all or part of the paint coating from
the mid-20th century and has recently been applied to
the substrate surface. Coatings might fail at the substrate
the field of heritage science (Amirudin and Thierry 1995,
surface or within/between coating layers.
Cano et al. 2010, Ramírez Barat et al. 2019). EIS spectra
are produced either showing the impedance Z (the ratio Each test area was lightly cleaned and abraded using a
of voltage over current) and phase angle F, known as a 3M Scotchbrite pad (7447) before degreasing with indus-
Bode plot, or in polar coordinates with real and imaginary trial methylated spirits. A 20 mm aluminium dolly was
elements (Nyquist plot). These plots provide information bonded to the paint surface at each test site using Araldite
on the performance of a coating, with impedance values 2021-1, a two-part methacrylate adhesive. Eight dollies
reducing as a coating begins to fail. were adhered to each sample and left for 24 hours for the
resin bond to form fully. The test area was separated from
One corner of each EIS sample was tapped with an M4
thread to fit an electrical connection post. The sample
was placed on a Gamry Instruments PTC1 paint test cell
(Figure 3). The cell was filled with 40 mL of electrolyte
before a graphite counter electrode and saturated calomel
Figure 4. A 150 × 150 mm sample coupon being prepared for pull-off

Figure 3. Painted coupon mounted in the test cell measurements
the surrounding adhesive using a cutting tool (Figure 4) measure the gloss level of a coating by quantifying the
and adhesion tested using a DeFelsko PosiTest AT-A reflectance of the surface. A sample mount was used to
pull-off adhesion tester. ensure the gloss measurement could be repeated in the
same position every 12 months; the gloss of each sample
Colorimetry
was measured using a Rhopoint Novo-Gloss Trio 20/60/85
To assess colour change, 36 colorimeter measurements glossmeter and repeated after rotating the sample by 90°.
were made on each large sample using a Konica Minolta
CM-700d portable sphere-type spectrophotometer every Results
12 months (Figure 5). Results are processed using Konica
Minolta SpectraMagic NX software. Colorimetry meas- Visible change
urements are compared with the previous measurement
12 months before and with the control samples stored in Neither the alkyd system nor the epoxy system samples
the dark (Gilchrist and Nobbs 2017). show any visible change after 24 months of outdoor
exposure. The lead/linseed oil system, however, exhibits
distinct change including chalking, shrinkage and
cracking of the finish coat (Figure 6).
Figure 5. The Konica Minolta CM-700d colorimeter instrument and a

sample with overlaid template Figure 6. Lead-painted sample before exposure (left) and after
24 months in outdoor exposure (right) showing chalking and cracking
Gloss of the finish coat
The gloss of a surface relates to its level of reflectivity and,

Electrochemical impedance spectroscopy
for paints, is defined under ISO 2813:2014/ASTM D532
(Chadwick and Kentridge 2015; International Organiza- Representative EIS Nyquist plots are presented for all
tion for Standardization 2014). The standard specifies a coating systems and surface preparation methods at 12,
method of using three geometries of 20°, 60° and 85° to 24 and 36 months (Figure 7).
Figure 7. Typical Nyquist

plots for the three different
coating types after 12-, 24- and
36-month exposure
Pull-off adhesion testing as the curing of the coating systems could not be guar-
anteed to be complete.
The results of pull-off testing at 12 and 24 months are
given in Figure 8 and Table 3. The average pull-off values,
Colorimetry
their range and the mode of failure of the system are
provided. Pull-off testing was not carried out at 0 months The average colorimeter values as measured for samples
of each coating system at 0 and 24 months of outdoor
exposure are given in Table 4. The differences between the
values at 0 and 24 months are used to produce the dE*,
dL*, da* and db* values using CIE 2000, which express
the change in colour over the exposure time.
Gloss
The average gloss values for samples of each coating type
exposed for 0, 12 and 24 months are given in Table 5.
Discussion
Visible change
After two years of outdoor exposure there have been some
Figure 8. Boxplot of the pull-off values for samples coated with the three
measurable and observable changes to the paint coating
coating systems prepared to Sa 2.5 and St3 showing the range of values
attained. The box represents the interquartile range, the horizontal line systems applied to the wrought iron samples. One of the
within the box denotes the median, and the upper and lower whiskers clearest signs that a coating may be beginning to lose its
show the maximum and minimum values protective properties is a change in its appearance. This
Table 3. Average pull-off values and modes of failure for samples coated with the three coating systems at 12 months
Surface Exposure Pull-off values (MPa)

System Mode of failure
preparation (months) Average Range
1 (Alkyd) Sa 2.5 12 5.9 4.2 Adhesion failure at primer/metal interface
Sa 2.5 24 5.9 3.1 Adhesion failure at primer/metal interface
St3 12 4.0 3.2 Cohesion failure within corrosion layer
2 (Epoxy) Sa 2.5 12 5.2 2.7 Adhesion failure at topcoat/midcoat interface
Sa 2.5 24 2.8 2.3 Adhesion failure at topcoat/midcoat interface
3 (Lead) St3 12 2.8 1.4 Adhesion failure at primer/topcoat interface
St3 24 2.0 0.9 Adhesion failure at primer/topcoat interface
Table 4. Colorimetry values expressed in the L*a*b* colour space with average values for samples at 24 months of exposure compared to values for
samples at 0 months of exposure to produce dE*, dL*, da* and db* values as a measure of change. The threshold for visible change is dE*ab ≥ 1.5
System (exposure
dE*ab(D65) L*(D65) a*(D65) b*(D65) dL*(D65) da*(D65) db*(D65)
months)
1 Alkyd (0) - 89.77 0.43 8.70 - - -
1 Alkyd (24) 1.91 (visible change) 89.57 0.81 6.93 −0.20 (darker) 0.39 (more red) −1.76 (less yellow)
2 Epoxy (0) - 90.62 1.78 7.29 - - -
2 Epoxy (24) 0.69 (no visible change) 90.00 1.74 7.47 −0.62 (darker) −0.04 (less red) 0.18 (more yellow)
3 Lead (0) - 65.14 −2.34 7.92 - - -
3 Lead (24) 10.96 (visible change) 70.48 0.78 −1.04 5.34 (lighter) 3.12 (more red) −8.96 (less yellow)
Table 5. Gloss values (GU) at 20°, 60° and 80° for samples of each coating system and the influence of surface preparation
coating type exposed outdoors for 0, 12 and 24 months
level. After 12 months the alkyd (System 1) and epoxy
Exposure Average gloss units (GU) resin (System 2) systems on surfaces blasted to Sa 2.5
System
(months) 20° 60° 85° exhibit the greatest bond strength, with overlapping value
0 35.41 75.10 83.32 ranges showing that there is no significant difference
1 (Alkyd) 12 20.21 62.18 74.62 between the two. Adhesive failure between the coatings
24 13.86 51.63 71.27 or between the topcoat and the dolly demonstrates that
0 4.04 9.98 19.63 the adhesive bond of the primer to the substrate metal
2 (Epoxy) 12 4.00 10.60 20.34
exceeds these values. For samples of alkyd and epoxy
24 3.89 10.39 19.96
systems prepared to St3, failure occurred within the
0 2.49 2.88 6.88
compact oxide layers remaining on the metal surface,
3 (Lead) 12 2.64 −1.13 1.47
showing that the weakest point in the protective system
24 2.97 1.45 0.08
is within the oxides themselves.
With lead-painted samples at 12 months, failure occurred

can serve as an early warning that the heritage metal- at the lowest forces. The failure at the interface between
work may be at risk. To date, this is only noticeable on the primer and the degraded finish coat layers is unsur-
the lead-painted sample coupons (Figure 6). Observable prising given the extent of visible damage to the finish
after 12 months, by 24 months the lead-based linseed oil coat, which includes shrinkage and cracking, demon-
coating exhibits dramatic change in the appearance of strating poor adhesion to the undercoat.
its finish coat. Not only has the surface lost much of its
colour, but it has also chalked, shrunk and cracked, likely Colour and gloss level changes
due to loss of binder. Despite the lack of visible change in
the alkyd and epoxy systems, the analytical techniques Both the colorimeter measurements and the gloss meter
applied here reveal that all of the coating types have begun results showed that the paint surfaces have become
to show signs of deterioration. altered. This is exhibited as subtle colour changes
(Table 4) and a gradual reduction in the gloss level of
Electrical impedance spectroscopy measurements the paint surface (Table 5).
EIS measurements taken after 12 months revealed that Colour change for the alkyd-coated samples (System 1)
for both sets of metal samples painted with the alkyd and falls just above the threshold for a visible change, although
epoxy resin systems (Figure 7), the coatings were acting as this was not perceived by the observers. The colorimeter
capacitors, essentially non-conducting, with an electrical values indicate that there has been a slight darkening of
resistance of greater than 108 Ω cm2. This is confirmed by the topcoat, which has become slightly more red and less
creating an equivalent electrical circuit as a simulation. yellow. Changes in the epoxy-coated samples (System 2)
For the lead-painted samples, the response was slightly do not meet the threshold for visibility, but they exhibit
different (Figure 7). Electrical resistance was lower after a slight darkening and have become less red and more
12 months (~106 Ω cm2) and had reduced to < 103 Ω cm2 yellow. The biggest change is seen in the lead/linseed
after 24 months. The performance of the coating was oil-coated samples (System 3), whose change value
deteriorating, acting as both a resistor and capacitor. A exceeds the visible threshold by 7 times and indicates
small amount of current (amps, W) is flowing through that the coating is lighter, more red and less yellow. This
the paint layers, but the coating provides increasing corresponds well to the chalking of the finish coat and
resistance as the frequency (Hz) increases. For the alkyd the appearance of the red undercoat through the cracks
and epoxy coatings, the levels of electrical resistance had in the finish coat.
reduced slightly after 24 months.
Gloss units (GU) are the measurement scale based
on a highly polished black glass standard that gives a
Pull-off tests
reflectance of 100 GU at a specified angle.1 The angle
Examination of the pull-off test results (Figure 8; of measurement is altered according to the level of
Table 3) reveals differences in the strength of each gloss. The measurements taken at 60° revealed that
the alkyd-coated (System 1) surfaces have reduced in References

gloss level from a high (> 70 GU) to a medium gloss
Amirudin, A. and D. Thierry. 1995. Application of elec-
(10–70 GU) level, even within the first 12 months. trochemical impedance spectroscopy to study the degra-
For the epoxy (System 2), the gloss level has largely dation of polymer-coated metals. Progress in Organic
remained unchanged after 24 months (Table 3). The Coatings 26(1): 1–28.
original acrylic urethane finish coats applied gave a
Blackney, K. and B. Martin. 1998. Development of long-
sheen rather than gloss finish. As the original level of
term testing methods to clean and coat architectural
gloss was low (< 10 GU) the most appropriate measure-
wrought ironwork located in a marine environment: The
ment values to use are at 85°. The lead-painted samples
maintenance of railings at the Garrison Church, Ports-
showed significant changes to their gloss levels (Table 5).
mouth. In Metals – English Heritage research transactions.
The initial gloss level was very low (> 10 GU) at a meas-
Research and case studies in architectural conservation,
ured angle of 85°, and this has reduced to almost zero
Volume 1, ed. J.M. Teutonico, 103–16. London: James
after 24 months of exposure.
& James.
Conclusion British Standards Institution. 2016. BS EN ISO 4624:2016:

Paints and varnishes. Pull-off test for adhesion. London:
This paper has reported on preliminary findings of a BSI. www.en-standard.eu/bs-en-iso-4624-2016-paints-
five-year study examining the real-time performance and-varnishes-pull-off-test-for-adhesion/
of three protective coating systems for historic wrought
Cano, E., D. Lafuente, and D.M. Bastidas. 2010. Use of EIS
iron exposed outdoors. Even in the limited exposure
for the evaluation of the protective properties of coatings
period to date, it is possible to identify significant changes
for metallic cultural heritage: A review. Journal of Solid
in the traditional lead coating, which may continue to
State Electrochemistry 14: 381–91.
perform a protective function for the iron substrate but
has completely failed in its aesthetic capacity. Aesthetic Chadwick, A.C. and R.W. Kentridge. 2015. The percep-
changes in the alkyd and epoxy systems are minor, being tion of gloss: A review. Vision Research 109(Part B):
barely or not noticeable. 221–235.
Cogger, N.D. and N.J. Evans. 1999. An introduction to
The findings reveal that preparing the surfaces to St3
electrochemical impedance measurement. Technical
means that with the alkyd and epoxy systems, the
Report No. 6, Part No. BTR006, Issue 05 May 1999.
corrosion product layers themselves are the primary
Hampshire: Solartron Limited.
weakness in the protective system. This has implications
for heritage, where justifying the loss of original mate- Dillmann, P., F. Mazaudier, and S. Hoerlé. 2004. Advances
rial associated with blasting to near-white metal can be in understanding atmospheric corrosion of iron. I. Rust
challenging. characterisation of ancient ferrous artefacts exposed to
indoor atmospheric corrosion. Corrosion Science 46(6):
The initial results from the EIS measurements reveal 1401–29.
that the epoxy resin coatings are the best performing,
Emmerson, N.J., D.E. Watkinson, and J.C. Thunberg.
with the alkyd paints also maintaining a high level of
2019. Flame cleaning of historic wrought iron: Practi-
protection. The lead coatings show the greatest changes
tioner methods and their impact on oxide morpholo-
in their performance.
gies and post-treatment corrosion rates. In Metal 2019:
The next three years will see coating performance chal- Proceedings of the Interim Meeting of the ICOM-CC
lenged further by outdoor exposure, and the mode and Metals Working Group, Neuchatel, 2–6 September 2019,
extent of any breakdown of the coatings will be reported eds. C. Chemello, L. Brambilla, and E. Joseph, 281–87.
in future papers. Paris: International Council of Museums–Committee
for Conservation (ICOM-CC) and Haute Ecole Arc
Notes Conservation-restauration (HE-Arc CR).
1
www.rhopointinstruments.com/faqs/how-is-gloss- Gilchrist, A. and J. Nobbs. 2017. Colorimetry, theory.
measured/ In Encyclopaedia of spectroscopy and spectrometry, 3rd
ed, eds. J.C. Lindon, G.E. Tranter, and D.W. Koppenaal, L. Brambilla, and E. Joseph, 83–91. Paris: International
328–33. London: Academic Press. Council of Museums–Committee for Conservation
Hoerlé, S., F. Mazaudier, Ph. Dillmann, and G. Santarini. (ICOM-CC) and Haute Ecole Arc Conservation-restau-
2004. Advances in understanding atmospheric corrosion ration (HE-Arc CR).
of iron. II. Mechanistic modelling of wet–dry cycles. The Society for Protective Coatings/NACE International.
Corrosion Science 46(6): 1431–65. 2007. Joint surface preparation standard SSPC-SP 10/
International Organization for Standardization. 2007. ISO NACE No. 2. Near-white metal blast cleaning. Pittsburgh:
8501-1:2007. Preparation of steel substrates before applica- SSPC.
tion of paints and related products – Visual assessment of
surface cleanliness – Part 1: Rust grades and preparation Authors
grades of uncoated steel substrates and of steel substrates Peter Meehan ACR works in private practice special-
after overall removal of previous coatings. Geneva: ISO. ising in the conservation of outdoor historic ironwork.
www.iso.org/standard/43426.html
Nicola Emmerson lectures in conservation science
International Organization for Standardization. 2007. at Cardiff University, researching the management of
ISO 2813:2014: Paints and varnishes – Determination of corrosion prevention strategies for metallic cultural
gloss value at 20°, 60° and 85°. Geneva: ISO. www.iso. heritage artefacts.
org/standard/56807.html
David Watkinson teaches and researches conservation
International Organization for Standardization. 2017. ISO
at Cardiff University. His work on the corrosion rate of
12944-2:2017: Paints and varnishes – Corrosion protec-
iron underpins the preservation of Brunel’s iron steam-
tion of steel structures by protective paint systems – Part
ship SS Great Britain.
2: Classification of environments. Geneva: ISO. www.iso.
org/standard/64834.html
International Organization for Standardization. 2018.
ISO 8504-3:2018: Preparation of steel substrates before
application of paints and related products — Surface
preparation methods — Part 3: Hand- and power-tool
cleaning. Geneva: ISO. www.iso.org/standard/66696.html
Loveday, D., P. Peterson, and B. Rodgers. 2004. Evaluation
of organic coatings with electrical impedance spectros-
copy. Part 2: Application of EIS to coatings. JCT Coatings
Tech 1(10): 88–93.
Matthiesen, H. 2013. Oxygen monitoring in the corro-
sion and preservation of metallic heritage artefacts. In
Corrosion and conservation of cultural heritage metallic
artefacts, eds. Ph. Dillmann, D. Watkinson, E. Angelini,
and A. Adriaens, 368–91. Woodhead Publishing.
Orazem, M.E. and B. Tribollet. 2017. Electrochemical
impedance spectroscopy, 2nd ed. The Electrochemical
Society Series. Hoboken: John Wiley & Sons.
Ramírez Barat, B., P. Letardi, and E. Cano. 2019. An
overview of the use of EIS measurements for the assess-
ment of patinas and coatings in the conservation of
metallic cultural heritage. In Metal 2019: Proceedings
of the Interim Meeting of the ICOM-CC Metals Working
Group, Neuchatel, 2–6 September 2019, eds. C. Chemello,
Keris Blade Protection – A Comparison of Traditional and

Contemporary Coatings
Janine Meier
Historisches Museum Basel
Basel, Switzerland
[email protected]
Abstract
Keris daggers are weapons and spiritual objects of unique cultural alloy steel coupons. The coupons were then exposed to a stress
significance from the Indonesian Archipelago. They challenge test to assess protective effectiveness. The results of this research
conventional conservation efforts due to their composition, suggest that conservation strategies should include the use of
fabrication techniques, and, in particular, the spiritual value a traditional mixture of natural oils (coconut oil, sandalwood
attributed to the traditional coating on the blade, which is oil, and ylang-ylang) to protect keris in museum environments.
applied during ceremonial processes. This study presents a
comparison of contemporary and traditional coatings for keris Keywords
blades. Each coating was tested on multilayer steel coupons keris, kris, Indonesia, dagger, coating, traditional, etching,
reflecting keris blade properties (pamor) and standardized low- corrosion
Introduction
The keris originates from the Indonesian Archipelago,
where it serves as a weapon, a holy object in ritual cere-
monies, and a personal symbol of protection, making it
crucial to cultural identity (it has been listed as UNESCO
intangible cultural heritage since 2005; UNESCO Office,
Jakarta 2005) (Figure 1). In Indonesia, spirituality is
present everywhere. In contrast to western culture,
spirituality is not seen as something that is directly
linked to religion. As religion changed over time, the
myths and traditions surrounding keris remained the
same.1 A keris is composed of a sheath, handle, and
blade fabricated of different alloys and various organic
materials. Keris blades are made from pamor, a tradi-
tional Indonesian multilayered steel, whose pattern
is emphasized by etching the surface with lime juice
and arsenic sulfide. The etching is renewed and a new
coating applied to the blade (Weihrauch 1996) during a
traditional annual ceremony. Most keris are not treated Figure 1. Four keris from the MKB. © JM, HE-Arc/MKB 2020
during their time in museum collections, leaving only

residues of coating material or coatings that are no longer affects the intangible value, described in Indonesia as the
protective (Figure 2). In the traditional Indonesian loss of contrast between the layers of the pamor, a matt
worldview, the coating not only serves as protection surface, and the absence of odor (Kerner 1995). In this
for the blade material, but it is also believed to provide study, it is suggested that the protective coating layer of
spiritual nourishment. The loss of the nourishment the keris blade should be viewed as an integral part of the
with experts and collectors from various institutions. The

cyclical conditions during stress tests were selected based
on the literature and evaluation of the climatic conditions
in the museums and countries of origin.
Experimental setup
The following coupons were prepared: Six different coat-
ings (2. to 7.) were tested on coupons of three different
Figure 2. Keris (IIc 2118) from the MKB under visible light (VIS) and under substrates (P, S, and K; see following section) as well as
ultraviolet fluorescence (UVF) revealing the synthetic oil coating (FTIR on uncoated coupons as controls (1.). As a stress test,
analyzed). © JM, HE-Arc/MKB 2020 all coupons were stored in varying climatic conditions,
whereas a reference group (X) was stored in stable condi-
object. Multiple studies have assessed protective coatings
tions with low relative humidity (RH). Furthermore, the
(Dalewicz-Kitto 2016, Brüggerhoff 2001), but traditional
test included pamor coupons with residues from etching
coating materials such as natural oils and multilayer steel
(8.) and without etching (9.) (Figure 3).
substrates have rarely been included. This study focuses
on the specific case of protecting keris blades in western
Coupons
collections with the aim of guiding the development
of an integrated conservation strategy specific to their Three different substrates were selected: pamor multi-
cultural significance. layer steel coupons (P), produced and etched with
the traditional recipe of lime juice and arsenic sulfide
Methods (Kerner 1995); low-alloy steel (S) (DC04 B (EN 10130));
and pre-corroded low-alloy steel (K) (corroded for two
A collection of 144 keris at the Museum der Kulturen
weeks by daily cycles of room temperature, 50 °C and
Basel (MKB) was examined. Of these, 13 were analyzed
100% RH).
using X-ray fluorescence (XRF) and Fourier transform
infrared (FTIR) spectroscopy to determine the composi-
Coatings
tion of the coating residues and blade materials. A stress
test was performed to compare the protective capability Three contemporary and three traditional coatings
of the coatings. Coatings and substrates were selected were compared. The former, representing commonly
based on the literature, analysis, and communication used museum coatings, consisted of a microcrystalline
Figure 3. Coupon groups

for stress test: uncoated (1),
contemporary coated (2–4),
traditionally coated (5–7) on
pamor (P), low-alloy steel (S),
and pre-corroded low-alloy
steel (K) substrate. Groups 8
and 9 comprise pamor coupons.
The X group represents the
reference coupons not included
in the stress test
KERIS BLADE PROTECTION – A COMPARISON OF TRADITIONAL AND CONTEMPORARY COATINGS 105
wax (2., Cosmoloid H80, 20% in Shellsol T (w/w)), an Stress test

acrylic resin lacquer (3., Paraloid B-44, 20% in acetone
The coupons were exposed, at a slight angle, to a six-week
(w/w)), and paraffin oil (4.). The traditional coatings
stress test in a climatic chamber (Vötsch Typ VC 0020)
included coconut oil (5., 8.2, 9.2), a traditional mixture
with cycles of warm-humid (34 °C and 90% RH) and
(6.) of coconut oil, sandalwood oil, and ylang-ylang
cold-dry (18 °C and 40% RH) conditions with transition
(50%/25%/25% (w/w)), and a synthetic perfumed oil (7.,
phases of 30 minutes every 12 hours.
Kerisöl 2006). Currently, both blends of naturally odorous
oils (6.) and synthetic oils with added fragrances (7.)
Evaluation
are used during ceremonial processes. A single layer of
each coating was applied at room temperature on clean The coatings were evaluated according to four criteria:
(acetone-treated) coupons, using cotton cloths to apply ease of application, protective capacity, aesthetic prop-
the oil and brushes for the wax and acrylic resin, and erties, and maintenance. The coupons were evaluated
they were dried for three weeks. weekly during the stress test to document the develop-
Figure 4. Results after six weeks in

the climatic chamber with exemplary
coupons from each group and
corrosion evaluation by corrosion
degree according to the DIN norm EN
ISO 6270-2 and amount of corrosion
spots
Figure 5. Results after six weeks in the

climatic chamber. Shows the corrosion
development and change of coating
properties
ment of corrosion and visible changes in the coating with

a digital microscope. The corrosion protection was eval-
uated by determining the degree of corrosion according
to DIN EN ISO 6270-2 AHT (Deutsches Institut für
Normung e. V. 2017) and the number of corrosion spots.
Changes in the aesthetic properties were compared by
color-calibrated photographs.
Results
The traditional coconut oil-based coatings (5., 6.)
showed the best anti-corrosion properties across all
three substrates (Figure 4). Cosmoloid H80 (2.) prevented
Figure 6. Corrosion of uncoated coupons S, P, and K (200× magnification)
corrosion the least well across all tested coatings, with
light microscopic inspection revealing that corrosion
developed in small, uncoated areas due to cold applica-
tion. The two synthetic oils—paraffin oil (4.) and Kerisöl
2006 (7.)—achieved good corrosion protection but only
for a limited period of time. These coatings provided
protection for as long as they were visible as a glossy
surface film (Figure 5). The corrosion morphology varies
with different coatings and is highly dependent on the
substrate and degree of corrosion (Figures 6–8).
As expected, the degree of corrosion on the pre-corroded

coupons (K) was higher than on the non-corroded
substrates, and the uncoated coupons corroded more
than the coated coupons. On pamor coupons (P), the Figure 7. Corrosion of Cosmoloid H80-coated coupons S, P, and K (200×
corrosion developed across the two alloys of the multi- magnification)
layer steel. Additional tests indicated that residues of
the etching solution (8.) were not a source of corrosion
acceleration, and the lack of traditional etching (9.) did
not influence the corrosion process.
Most of the contemporary and traditional coatings

tested increased the contrast in the pamor pattern and
the gloss. However, the wax and coconut oil did not
have this effect on the pamor coupons. The spiritually
important odor requirement was only met by the tradi-
tional mixture (6.) and the Kerisöl 2006 (7.). None of the
coatings changed in color during the test. Dust particles
were observed on the sticky surface of the natural and
synthetic oils.
Figure 8. Corrosion of Paraloid B-44-coated coupons S, P, and K (200×
The oil coatings could be evenly applied to the coupons magnification)
except for the coconut oil, which produced nonuniform
Discussion
and thicker coating films. The Paraloid B-44 was chal-
lenging to apply evenly. The Cosmoloid H80 did not The mixtures of natural oil-based coatings provided
cover all of the coupon cavities when applied at room higher corrosion protection, better aesthetic charac-
temperature. teristics, and greater maintenance advantages than the
KERIS BLADE PROTECTION – A COMPARISON OF TRADITIONAL AND CONTEMPORARY COATINGS 107
synthetic oil, wax, or lacquer coatings. In addition,

natural oils are often less toxic and harmful to conser-
vators and the environment.
Due to the irregular surface on keris blades, coating

interventions are not entirely reversible, with residues
remaining on the surface. Consequently, the introduction
of a new material to the object, such as contemporary
coatings, must be carefully considered.
Coatings need to be renewed periodically. Therefore,

monitoring and renewal must be included in the conser-
vation strategy. Oil coatings can be renewed easily with
less time, which simplifies regular maintenance. Natural Figure 10. Filiform corrosion under acrylic coating (FTIR analyzed) on
oils (5., 6.) protect longer than synthetic oils (4., 7.) keris blade (IIc 22217) from the MKB (20× magnification).
and therefore lead to less frequent renewal cycles. The © JM, HE-Arc/MKB 2020
stress test demonstrated the importance of scheduled
reapplication of synthetic and natural oils to prevent intrinsically contradictory. According to the Burra
accelerating corrosion due to dust particles adhering Charter of 2013, which defines ethics in conservation
to the oil coatings (Figure 9). For coconut oil-based and takes into account cultural significance, traditional
coatings, temperature fluctuations should be avoided techniques should be favored if there are no conserva-
to prevent cyclical hardening and liquefying, which tional disadvantages (Australia ICOMOS 2013, Article
results in an uneven coating film. 2.1/4.2). Since keris blades benefit from traditional
coatings, these techniques are recommended.
The corrosion morphology that arises when natural and
synthetic oil coatings lose their protective properties is
Conclusion
equivalent to corrosion observed on keris blades with
traditional coatings. However, the filiform corrosion The results of this work suggest that traditional and
of acrylic coatings (Figure 10) is atypical and distracts natural coating materials should be employed in
aesthetically from the highly valued pamor pattern. museum environments due to their superior protective
Consequently, lacquer coatings are unsuitable for keris properties. The traditional approach used throughout
blades. the Indonesian Archipelago for keris blade preserva-
It is important to recognize that cultural significance tion adheres to ethical standards and historic aesthetic
and the goals of contemporary conservation are not qualities, while also providing an environmentally
conscious and user-friendly conservation approach.
Ideally, objects are protected from dust, monitored
regularly, and receive a periodic reapplication of the
coating. Additionally, preservation with a traditional
coating allows visitors to experience a nourished keris
in a traditional and spiritual sense as a blade with
enhanced contrast in the pamor layers and the intended
luster and odor. The fact that olfaction is an important
part of these objects offers a rather unexplored oppor-
tunity to use the sense of smell as a means to access
those objects. In conclusion, this study demonstrated
the necessity of studying traditional techniques and
Figure 9. Dust particle on pamor coupon with paraffin oil coating (4) and the value natural products can add to contemporary
localized corrosion development (50× magnification) conservation methods.
Acknowledgments fellowship at The Metropolitan Museum of Arts, New

York, and is currently working in a permanent position
The author would like to thank the Haute École Arc
at the Historisches Museum Basel, Switzerland.
Conservation-restauration in Neuchâtel for providing
equipment and expertise, the Museum der Kulturen Basel
for the possibility to study its collection, and all experts
for generously sharing their experiences.
Notes
1
See article “Simpler keris cleansing ritual this year,” The
Straits Times, 21 May 1996, Singapore.
References
Australia ICOMOS. 2013. The Burra Charter 2013. The
Australia ICOMOS Charter for Places of Cultural Signif-
icance. Burwood: Australia ICOMOS Incorporated and
International Council on Monuments and Sites.
Brüggerhoff, S. 2001. Korrosionsschutz für umwelt-
geschädigte Industriedenkmäler aus Eisen und Stahl.
Bochum: Deutsches Bergbau-Museum.
Dalewicz-Kitto, S.J. 2016. Stopping corrosion around the
world: An investigation into protective surface coatings
on arms and armour. Journal of the Institute of Conser-
vation 39(1): 44–56.
Deutsches Institut für Normung e.V. (DIN). 2017. DIN
ISO 6270-2: Paints and varnishes – Determination of
resistance to humidity – Part 2: Condensation (in-cabinet
exposure with heated water reservoir), German version of
EN ISO 6270-2:2018. Berlin: Beuth Verlag.
Kerner, M. 1995. Der Keris im magischen Weltbild. Kirch-
dorf: M. Kerner.
UNESCO Office, Jakarta. 2005. Annual Report 2005.
Jakarta: UNESCO Office, Jakarta.
Weihrauch, A. 1996. Der Keris in Indonesien – Herkunft,
Bedeutung, Typenvielfalt und symbolischer Hintergrund
der Einzelelemente. Master’s thesis, Ethnology Seminar
of the University of Basel, Switzerland.
Author
Janine Meier’s professional education started with
an apprenticeship as a goldsmith. She completed her
master’s degree in conservation and restoration of
scientific and technical objects at the Haute École Arc
Conservation-restauration in Neuchâtel in 2020, after
which she worked at the Museum der Kulturen Basel and
the Bernisches Historisches Museum. She completed a
Preliminary Investigation of the Use of Chitosan-based

Coatings for Outdoor Cultural Heritage Objects
Carlota Carvalho Nádia Silva Eduarda Vieira Patrícia R. Moreira*

Research Centre for Science and Centre for Biotechnology and Fine Research Centre for Science and Research Centre for Science and
Technology of the Arts (CITAR), Chemistry - Associate Laboratory Technology of the Arts (CITAR), Technology of the Arts (CITAR),
Universidade Católica Portuguesa (CBQF), Universidade Católica Universidade Católica Portuguesa Universidade Católica Portuguesa
Porto, Portugal Portuguesa Porto, Portugal Porto, Portugal
[email protected] Porto, Portugal [email protected] [email protected]
Abstract
Outdoor art is continuously exposed to climatic factors and to with water applied with a cotton swab, thus demonstrating
direct contact with the public, both of which pose challenges treatment reversibility. The coating’s wettability and stability
for long-term conservation. This study examined the efficacy of were determined by evaluating the changes in coated specimens
a chitosan-based coating and its applicability for metallic objects in contact with water (surface tension tests) and submerged
such as outdoor sculptures. The chitosan film was applied on for 24 hours. Application of the coating using a paintbrush
stainless steel specimens using three different application resulted in the best adherence to the metal, considering the
methods (paintbrush, spray, and syringe). Each specimen was structural characteristics of the test specimens. No dissolution
microscopically and colorimetrically examined before, during, of the coating occurred from the immersed coated specimens.
and after application of the coating and following aging tests. These properties of the chitosan coating recommend its use for
Removal of the coating from the surface after 48 hours and after outdoor applications.
aging was also tested, using cotton swabs soaked in acetone,
ethanol 96%, and water. The coating was not removable by Keywords
mechanical force alone, which instead was effective together stainless steel, chitosan, coating, sustainability, public art
Introduction
For conservation and restoration professionals, effective
products that are sustainable, nontoxic, and have a low
impact on the user’s health and on the environment are
highly desired. Chitosan, a biopolymer obtained from
the shells of crustacean waste, is soluble in weak acids
but not in water and has the ability to form films that
are biodegradable and nontoxic (Giuliani et al. 2018).
Consequently, chitosan film has been proposed as a
sustainable alternative to coatings that are currently
commercially available (Moreira et al. 2018). Figure 1. Sculpture Eu Espero (1999), by Fernanda Fragateiro (MIECST –
Santo Tirso International Museum of Contemporary Sculpture)
In this study, the applicability and reversibility of a

chitosan coating on metallic stainless-steel specimens
were determined in order to evaluate the applicability of Specimens and coating preparation and application
this method to the sculpture Eu Espero (1999), a public The chitosan film tested in this study consisted of a
work of art by the artist Fernanda Fragateiro that consists low-molecular-weight chitosan solution prepared at a
of a backless stainless-steel bench (Figure 1). concentration of 1% in 1% acetic acid and with a pH of 3.9
(Silva et al. 2018). The coating was applied to 27 stainless were conducted using chitosan-coated PB and S spec-
steel specimens approximately 39 × 36 × 1 mm in size imens, with specimens PB1–PB3 and S1–S3 subjected
using a paintbrush (PB) or syringe (SY) or as a spray (S). to thermal aging, and specimens PB4–PB6 and S4–S6
Six replicates were established for the SY specimens and to UV light aging.
12 each for the S and PB specimens. The coated S and PB
specimens were dried for 24 hours at room temperature, Reversibility tests
and the SY specimens for 24 hours at 30°C.
The reversibility of the chitosan film on the PB, SY, and S
specimens was tested by mechanical removal and through
Photography and optical microscopy solubility tests using swabs soaked in acetone, 96%
Each specimen was imaged before and after coating appli- ethanol, and water, mimicking the method commonly
cation and the aging tests using a Canon EOS M50 camera used to clean outdoor works of art. The reversibility of the
equipped with an EFM 28 mm, f/3.5 macro IS STM lens coating was also examined in specimens subjected to the
together with a Dino-Lite Pro2 AM4000 microscope. aging tests. Surface and color changes were characterized
Image storage and processing were performed using and quantified using a Dino-Lite Pro2 AM4000 digital
DinoCapture 2.0 software, version 1.5.39.C. optical microscope and a Konica Minolta CM-700d
spectrophotometer.
Colorimetric analysis
Wettability and surface tension tests
Color alterations were assessed by colorimetric meas-
urements, performed according to the CIELab method A representative specimen from each coating method
using a Konica Minolta CM-700d spectrophotometer was submerged in 35 mL of water for 24 hours at room
with three sequential readings. The L*, a*, b* CIELab temperature. Reversibility after water absorption was
parameters were recorded before and after application tested by attempting to lift the film using tweezers and
of the chitosan film. The ∆E was calculated as: to remove it with a dry cotton swab. Surface tension tests
were performed by placing a 0.2 mL drop of water using
∆E*= ) a pipette on the age-tested specimens (PB1, S1, PB4, S4).
The RGB color of the uncoated and coated specimens
and of the specimens after removal of the coating was Results and discussion
recorded. The same methodology was applied to evaluate The metallic specimens had a smooth surface but an
the color changes in the specimens in the aging tests. irregular structure, with concave and convex areas. They
were thus representative of outdoor works of art and
Aging tests appropriate for testing the applicability of the chitosan
coating (Figure 2).
Thermal aging was performed by placing the specimens
for 12 days in a 60°C incubator with a relative humidity Application using a paintbrush resulted in better adher-
of 20%–25%. UV light aging consisted of 24-hour ence of the coating to the metallic specimens. Spray
exposures for 12 days using a UV black light source application also resulted in good adherence but in some
(LAMP15TBL, 15 W, 365 nm), with the specimens placed concave areas deposition was visibly excessive. Syringe
in a closed cabinet to create a darkroom. The aging tests application of the chitosan film was deemed ineffective
Figure 2. Application of the coating using a paintbrush (a), a syringe (b), and spray (c)
PRELIMINARY INVESTIGATION OF THE USE OF CHITOSAN-BASED COATINGS FOR OUTDOOR CULTURAL HERITAGE OBJECTS 111
Method of ∆E standard RGB before RGB after Table 1. Difference in the CIELab
∆E parameter (∆E) before (PB, SY, S)
application deviation (SD) coating coating
PB / PB-CH 7.89 0.60 and after (PB-CH, SY-CH, S-CH)
coating, based on the average
SY / SY-CH 8.47 0.55
values of the parameters L*, a*,
S / S-CH 7.32 1.12
and b*
due to excessive deposition of the film in concave areas

and thus large differences in the thickness of the coating
(Figure 3).
The ∆E of the CIELab values calculated before and after

application of the coating showed that the change in L*
(luminosity) accounted for the largest ∆E value. The
colorimetric analysis and conversion of the CIELab Figure 4. Microscope observations of the metal specimens after
mechanical removal of the coating
parameters to RGB values showed that the coated metallic
specimens were opaquer than the uncoated specimens, method), thus demonstrating the good adherence of the
but the visual impact was minimal (Table 1). film to metallic surfaces under dry conditions (Figure 4).
The films could not be removed mechanically without Regardless of the application or aging method, the film
damaging the specimen (independent of the coating could only be removed using circular movements of a
cotton swab moistened in water. For the specimens aged
with UV light (PB4 and S4), the coating removed by this
method had yellowed. This observation was in line with
the CIELab data, in which the value of b* (an indicator
of yellowing) was twofold higher than that before UV
aging, indicative of yellower hues (Tables 2, 3).
Table 2. Difference in the CIELab parameter (∆E) for the uncoated

specimens and after removal of the coating using a cotton swab with
water and mechanical force
Uncoated specimens / Specimens after removal

∆E
of chitosan coating with water
Uncoated specimen / PB1 after removal 3.16
Uncoated specimen / PB4 after removal 2.53
Figure 3. Microscope observations of the metal specimens before (a, b,
c) and after application of the chitosan coating using a paintbrush (d, Uncoated specimen / S1 after removal 1.92
g) or syringe (e, h) or spray (f, i) Uncoated specimen / S4 after removal 1.36
Chitosan-coated specimens after 24h L* L* SD a* a* SD b* b* SD Table 3. Average CIELab values

of the specimens 24 hours after
PB 73.91 0.59 −0.04 0.02 −0.19 0.12
coating and the average CIELab
S 74.27 0.32 −0.08 0.1 0.35 0.25 values of the specimens after
Specimens after 12 days of aging L* L* SD a* a* SD b* b* SD 12 days of aging. Difference in
PB 60°C 69.30 0.56 0.13 0.10 1.21 0.13 the CIELab parameter (∆E) for
PB UV light 69.14 0.10 −0.17 0.13 2.50 0.24 the specimens 24 hours after
coating and for the specimens
S 60°C 69.93 0.33 0.27 0.15 1.18 0.20
subjected to 12 days of aging
S UV light 69.88 0.35 0.12 0.74 2.05 1.65
Chitosan-coated specimens after 24 hours / PB/PB S/S PB/PB S/S
Specimens after 12 days of aging 60°C 60°C UV light UV light
∆E 4.82 4.43 5.48 4.71
Figure 5. Uncoated specimens PB4 (a) and S4 (b); coated specimens PB4 (c) and S (d) after UV aging; removal of the UV-aged coating from specimens PB4
(e) and S4 (f) using a swab soaked in water; specimens PB4 (g) and S4 (h) after removal of the coating using a swab soaked in water and mechanical force
After removal of the coating, the optical and mechanical 24 hours without dissolving, but it could be removed
properties of the metal specimens were the same as before thereafter by wet swabbing (Figure 6).
the coating was applied (Figure 5).
Conclusion
After 24 hours of immersion of the coated specimen
in water, the coating applied by syringe could only be The chitosan coating tested on the metal surfaces was
removed without damaging the metal by using a dry of low aesthetic impact in dry environments and thus
cotton swab and tweezers. The coatings applied by paint- offers a promising approach to the conservation of metal
brush and by spray could not be removed mechanically. outdoor art works. While it was resistant to solubiliza-
The chitosan film applied was also resistant enough to tion in water for at least 24 hours, the coating could be
be submerged in or in contact with water for at least removed using wet swabs in water and the application
Figure 6. Surface tension tests and Dino Lite images of specimens PB1 (a, a.1), S1 (b, b.1), PB4 (c, c.1), and S4 (d, d.1). Wettability tests of specimens
S6, PB6, and SY6 (e). Detail of specimen SY6 during the wettability test (f)
PRELIMINARY INVESTIGATION OF THE USE OF CHITOSAN-BASED COATINGS FOR OUTDOOR CULTURAL HERITAGE OBJECTS 113
of mechanical force, thus demonstrating its reversibility. (ESB–UCP). She is currently a PhD student at ESB–UCP
Further investigation is needed to evaluate the impact of working on the development of new nanomaterials to
UV aging on the coating. prevent biodeterioration of cultural heritage. Her main
areas of interest are microbiology, natural polymers, and
Acknowledgments antimicrobial coatings and nanoparticles.
This work was supported by National Funds through Eduarda Vieira holds a PhD in conservation and
FCT – Fundação para a Ciência e a Tecnologia under the restoration of historical and artistic heritage. She is
projects BIONANOSCULP (PTDC/EPH-PAT/6281/201), currently an assistant professor at the School of Arts of
BIO4MURAL (PTDC/HAR-ARQ/29157/2017), and the Universidade Católica Portuguesa (Porto) and has
UID/Multi/50016/2019. been the director of CITAR since July 2019. Her main
interests are green conservation, conservation of decora-
References tive arts applied to architecture, public art, and heritage
Giuliani, C., M. Pascucci, C. Riccucci, E. Messina, M. and conservation theory.
Salzano de Luna, M. Lavorgna, G.M. Ingo, and G. Di
Carlo. 2018. Chitosan-based coatings for corrosion
protection of copper-based alloys: A promising more
sustainable approach for cultural heritage applications.
Progress in Organic Coatings 122: 138–46.
Moreira, P.R., N. Silva, R.C. Pullar, B.A.E. Ben-Arfa, J.G.
Abreu, F. Henriques, M.M.E. Pintado, and E. Vieira. 2018.
BIONANOSCULP, an ongoing project in biotechnology
applications for preventive conservation of outdoor
sculptures. IOP Conference Series: Materials Science and
Engineering 364: art. 012075.
Silva, N., R.C. Pullar, M.E. Pintado, E. Vieira, and P.R.
Moreira. 2018. Biotechnology for preventive conserva-
tion: Development of bionanomaterials for antimicrobial
coatings of outdoor sculptures. Studies in Conservation
63(suppl. 1): 230–33.
Authors
Carlota Carvalho has a master’s degree in conservation
and restoration of cultural heritage, with a focus on
outdoor sculptures made of inorganic materials. She is
interested in research fields associated with sustainable,
preventive, and green conservation.
Patrícia R. Moreira holds a PhD in biotechnology, is an

assistant professor at Universidade Católica Portuguesa
(UCP), and coordinates the CITAR’s Focus Area of
Heritage, Conservation, and Restoration in UCP. She is
interested in innovations in the field of cultural heritage,
with an emphasis on biodeterioration, sustainability,
circular economy, citizen science, and green conservation.
Nádia Silva has a MSc in microbiology by the Faculty

of Biotechnology, Universidade Católica Portuguesa
Testing the Compressive Strength and Reversibility of
Consolidants Applied to Marine Archaeological Gray Cast
Iron
Erik Farrell* Molly K. McGath Jonathan Echerd
The Mariners’ Museum and Park The Mariners’ Museum and Park East Carolina University
Newport News VA, USA Newport News VA, USA Greenville NC, USA
www.marinersmuseum.org/ www.marinersmuseum.org/ www.ecu.edu/
Abstract
Excavations of the American Civil War ironclad USS Monitor Different methods have been used historically to consolidate
recovered significant quantities of graphitized cast iron, including graphitized cast iron (e.g., microcrystalline wax, various
major engineering components, structural fittings, and two Paraloid formulations). However, there has been comparatively
7,200 kg XI-Inch Dahlgren shell guns. As is common in gray cast little research into the direct physical effects of consolidants
iron objects from marine archaeological environments, these on the strength of the graphite layer. Additionally, many prior
artifacts have undergone graphitic corrosion, compromising methods have little consideration for the size of the object;
their structure. Over the last decade, a variety of consolidants some treatments simply are not possible at the scale required
have been applied to these artifacts at The Mariners’ Museum for Monitor. A study was undertaken to make quantitative
and Park, with varying degrees of success. Due to the high risk comparisons of consolidants for graphitized cast iron, using
of structural damage in graphitized material and the difficulty compressive strength testing to directly measure their effects.
in re-treating large, complex, mechanical objects, conservators The reversibility of effective treatments was also investigated
began investigating additional methods for the conservation of and additional testing carried out on Owatrol oil to confirm its
cast iron. The end goal was to establish an overall preservation suitability for large-scale use.
system for large cast iron objects from Monitor, and specifically
to conserve the Dahlgren shell guns. The initial phase of research Keywords
into the efficacy of selected consolidants on graphitized Owatrol oil, consolidation, graphitized, cast iron, marine
cast iron has been completed and is discussed in this paper. archaeology
Introduction
Upon launching in 1862, USS Monitor became the first objects retaining core metal typically have a 10–20 mm
iron warship in the United States Navy, and the world’s thick graphitized exterior. Cracking, spalling, and surface
first commissioned warship with a rotating gun turret. losses are significant hazards, and similar difficulties have
Although active less than a year before sinking, Monitor been noted at multiple labs (e.g., McCarthy 2002, 213;
revolutionized naval warship design. Archaeological Mardikian et al. 2009, 85; Näsänen et al. 2013, 294–97).
excavations occurred through 2004, and artifacts are Research into the efficacy and reversibility of consolidants
being conserved at The Mariners’ Museum and Park in was carried out with these challenges in mind.
Virginia.
Consolidation
Large objects are inherently limiting; with the available
facilities, health and safety resources, and funding, some Three test materials were used. Archaeological mate-
treatments used on smaller objects are not possible. Cast rial from SS Columbus (1850) was obtained from the
iron objects excavated from Monitor include numerous Maryland Archaeological Conservation Laboratory at
large engineering components and two 7,200 kg Dahlgren Jefferson Patterson Park and Museum (A-series samples).
shell guns, all structurally compromised by graphitic Additional samples (S-series) were synthesized by elec-
corrosion (Selwyn 2004, 110). Small objects from Monitor trolytically de-alloying ASTM A48 Class 40 gray cast iron
can be graphitized through the full thickness, and larger (Figure 1). Vine charcoal (C-series) was used as a third
TESTING THE COMPRESSIVE STRENGTH AND REVERSIBILITY OF CONSOLIDANTS APPLIED TO MARINE ARCHAEOLOGICAL GRAY CAST IRON 115
material. Although less directly analogous to cast iron and the high probability of runaway reactions in epoxies
due to the lack of iron degradation products, charcoal at large scales.
is more physically homogeneous and was included to
Acrysol WS-24 is an aqueous emulsion of an acrylic
better facilitate direct comparisons between consolidants.
copolymer. It has a near-neutral pH of 7.2, a Tg of 40 °C,
and is nominally reversible in strong alkali. Although
a higher Tg was desirable in case of an environmental
failure, WS-24 otherwise met the requirements for a
consolidant and was tested at 5%, 10%, and 25% v/v. A
single sample was also tested in a 1:7:2 ratio by weight
of WS-24 solids:isopropanol:water as a check against
problems in pure aqueous application methods (Phillips
1987, 146).
Various Paraloid acrylics exist; for efficiency, only one

was tested. Based on Tg, B-44 and B-48N were preferred
candidates, and ultimately B-44 was tested due to local
availability. Despite the inherent limitations of solvent-
born applications, 5% w/v B-44 in acetone was tested.
Poligen ES 91009 is an aqueous emulsion of ethylene wax,

identified in the PROMET project as a potential barrier
coating (Argyropoulos et al. 2013, 508). Unfortunately,
Figure 1. Cross-section of a synthetic sample showing graphitized outer Poligen could not be obtained; multiple attempts to
layer and iron core. Image courtesy of The Mariners’ Museum and Park contact US and European distributors went unanswered.
Aquacer 8527, an aqueous emulsion of a modified
Consolidants were identified based on desired properties polyethylene wax with nominally similar properties
for large cast iron objects from Monitor. The goals were to to Poligen, was tested in its place. No previous testing
increase the compressive strength of graphitized cast iron had been carried out, and additional testing would be
after desalination and drying were complete and to test required to determine aging characteristics if the treated
feasible application/reversal methods. The size of some strength was suitable. Aquacer was tested at a 75% v/v
artifacts from Monitor is restrictive; direct immersion concentration in deionized (DI) water.
of the largest objects requires approximately 15,000 L of Owatrol oil is a proprietary alkyd oil that acts as a
solution, making solvents untenably hazardous for direct consolidant, corrosion inhibitor, and vapor barrier. This
application. Several potential aqueous and non-aqueous material has been successfully used on graphitized marine
consolidants were identified based on a literature review: archaeological cast iron, including a large gun from CSS
microcrystalline wax (North 1987, Hamilton 1999), Alabama (Mardikian 2020, 20). Owatrol is theoretically
Acrysol WS-24, Paraloid/Acryloid formulations, Poligen soluble in strong alkali, but prior testing on cuttlefish
ES 91009 (Degrigny 2008, Wolfram et al. 2010), and bone samples indicated some damage may occur during
Owatrol oil (Mardikian 2020). removal (Ternisien 2011).
Microcrystalline wax was not included in the testing
panel. Facilities are not available to heat enough wax
Density changes
to immerse objects the size of Monitor’s Dahlgrens Consolidants were applied by immersion for two weeks,
and cannot be realistically obtained. Its reversibility is followed by desiccation until sample weights stabilized.
also limited; although nominally reversible in boiling In total, 42 samples of each material were used, with each
water, studies show this is not the case (Johnson 1984, consolidant applied to 6 of each sample material and
72). Epoxies and (poly)urethanes likewise were not 6 held as controls. Density was used to track changes,
tested despite historical use (e.g., North 1987, 230–31; derived from direct measurements of mass and dimen-
Hamilton 1999, 71–72) due to their limited reversibility sions (Tables 1, 2).
Table 1. Average density before consolidation Testing

Material* Mean ± SD (g/mL)
Samples were allowed to dry in constant < 10% RH for
A 2.3753 ± 0.2283
approximately 30 days before testing. Compressive force
S 1.8397 ± 0.3603
tests used an Instron 5900R universal testing machine
C 0.2149 ± 0.0146
(Figure 2), with maximum strength determined by the
* Key: A = archaeological material from SS Columbus (1850);
S = synthetic graphitized cast iron; C = carbon (charcoal)
failure load of the sample.
Table 2. Average change in density
Mean ± SD
Material Consolidant
(percent increase)
Control 0
Owatrol 8.62 ± 1.47
5% B-44 0.92 ± 0.13
A 5% WS-24 2.08 ± 0.32
10% WS-24 2.10 ± 0.37
25% WS-24 2.49 ± 0.37
75% Aquacer 2.78 ± 0.33
Control 0
Owatrol 23.49 ± 5.34
Figure 2. Instron
5% B-44 2.59 ± 0.42
5900R universal
S 5% WS-24 4.88 ± 0.74
testing machine.
10% WS-24 5.80 ± 0.65 Image by
25% WS-24 8.44 ± 1.96 Jonathan Echerd,
75% Aquacer 9.33 ± 2.89 East Carolina
University
Control 0
Owatrol 307.09 ± 18.82
Samples were first loaded into the Instron compres-
5% B-44 38.36 ± 4.15
sion testing plates set with a preload of 5 N to create a
C 5% WS-24 4.74 ± 0.82
common starting point for all samples. C- and A-series
10% WS-24 6.49 ± 0.99
25% WS-24 13.71 ± 2.18
were tested directly on the instrument plates, and an
75% Aquacer 39.88 ± 4.59 aluminum fixture plate was used to maintain the same
orientation for S-series samples and prevent slip under
Owatrol consistently provided the greatest increase in load (Figure 3). Load was increased gradually until
density across all sample materials, and Aquacer the failure, recording load, time, and displacement for each
second greatest. case (Figure 4).
In S- and A-series, WS-24 increased density by the third-
largest amount, with greater increases correlating to
greater concentrations of WS-24 in the immersion bath,
followed by Paraloid B-44 with the least increase. In iron
samples, WS-24 solutions de-emulsified; this behavior
is known, and one additional sample was consolidated
in 1:7:2 WS-24 solids:isopropanol:water (Phillips 1987,
146). This composition prevented de-emulsification
but removed the material’s primary benefit (aqueous
application).
In C-series, B-44 had markedly greater uptake into the

sample than any WS-24 application, with low-concentra- Figure 3. Sample in testing setup using fixture plate to maintain
tion WS-24 causing the least increase in density. orientation. Image by Ed Howard, East Carolina University
High variability was experienced, and failure was often

hard to define. Some samples crumbled while load
continued to increase, not meeting the numerical defi-
nition of failure. Failure modes differed; some samples
completely collapsed, while others only cracked along
the leading edge. The degree of variability can be attrib-
uted to multiple factors, including sample geometry and
differences in material degradation.
Five control samples were tested for A- and S-series, and

six for C. Three samples were tested for each consolidant/
material pairing. Consolidant tests were compared to
control groups in a series of independent, two-tailed
t-tests, with results shown in Table 3.
None of the S-series tests show significance; the small

sample size and large standard deviation mask any effects.
Owatrol, B-44, and 10 and 25% WS-24 intuitively appear
effective, but this cannot be stated with confidence.
However, these samples may still be used to support and
reinforce results achieved in other sample materials, and
S-series samples were extremely useful for use in later
reversibility testing due to their easy availability.
Figure 4. Load-deflection curve showing failure, and visual failure of The A-series is more directly useful. Although still
sample. Image by Ed Howard, East Carolina University irregular, these samples are small pieces cut from a single
Table 3. Results
Mean ± standard deviation Significant Effect size

Material Consolidant
(Newtons) p < .05 (Hedge’s g)
Control/none 1190.60 ± 181.50 N/A N/A
Owatrol 2316.67 ± 495.75 Yes 3.49
5% B-44 2248.67 ± 144.14 Yes 6.23
A 5% WS-24 1864.33 ± 541.09 No 1.95
10% WS-24 1356.00 ± 305.32 No 0.72
25% WS-24 1627.50 ± 570.50 No 1.44
75% Aquacer 1574.67 ± 135.34 Yes 1.06
Owatrol 1025.67 ± 453.93 No 0.46
5% B-44 1821.67 ± 761.77 No 1.74
S 5% WS-24 960.33 ± 83.81 No 0.38
10% WS-24 1522.00 ± 519.86 No 1.48
25% WS-24 1250.67 ± 893.97 No 0.68
75% Aquacer 441.00 ± 108.81 No 1.00
Owatrol 86.00 ± 6.38 Yes 9.40
5% B-44 59.67 ± 6.55 Yes 3.72
C 5% WS-24 52.67 ± 7.36 Yes 2.10
10% WS-24 60.33 ± 4.99 Yes 4.40
25% WS-24 56.33 ± 13.89 No 1.77
75% Aquacer 65.67 ± 4.50 Yes 5.91
large fragment of cast iron spall and are somewhat more Based on these results, Owatrol and B-44 were identified
homogeneous as a result. The A-series still has a high as the best candidates for consolidation of graphitized cast
standard deviation, but significant increases in compres- iron. Of these two, only Owatrol was considered feasible
sive strength can be seen in Owatrol, B-44, and Aquacer. both to apply and remove from large objects.
The C-series is more regular in density than either of the

Reversibility
iron samples, but, lacking iron corrosion products, is less
directly analogous to graphitized marine archaeological The reversibility of Paraloid formulations, including
cast iron. However, the lower standard deviation means B-44, is sufficiently known that additional testing was
these tests may be a better measurement of consolidant not considered necessary (e.g., Horie 1987, 106–108;
strength. Down et al. 1996). At the scale of Monitor’s largest objects,
acrylics are irreversible in practice due to the quantity
Bearing these factors in mind, several materials show
of solvent required. The reversibility of Owatrol oil has
acceptable results. Paraloid B-44 shows significant effect
had limited study; work by Ternisien (2011, 56–60) on
in both A-series cast iron and C-series carbon, with the
greatest effect size of any tested material in the A-series consolidated cuttlebone indicated this process was likely
samples. Although not significant in and of themselves, damaging, but additional data was required.
the S-series results support this. However, B-44 is a
solvent-borne treatment; large objects could be treated Testing
using a vacuum bag (limiting the volume of solvent Graphitized cast iron samples for the reversibility study
required), but the treatment would be irreversible in were numbered R1–R15:
practice because solvent volume in reversal cannot be • Samples R1–R7 were impregnated with Owatrol and
easily limited in a similar manner. On smaller objects, aged at ambient temperature (22 °C) and 10% RH
this treatment is feasible. for 43 days.
Owatrol likewise shows significant effect in both A- and • Samples R8 and R9 were control samples with no
C-series tests, with supporting evidence from S-series. Owatrol present, which were taken through the
It has the greatest effect size of any consolidant in the same reversibility testing to differentiate damage
C-series and the largest mean breaking load in the from handling versus damage specifically caused by
A-series, although the high standard of deviation results Owatrol.
in a lower effect size than B-44. Unlike B-44, Owatrol • Samples R10–R15 were impregnated with Owatrol
hydrolyzes in strong alkali, making it theoretically revers- and aged 377 days prior to testing.
ible in aqueous media. Initial testing of Owatrol’s solubility in aqueous NaOH
Aquacer’s effectiveness is ambiguous. In charcoal samples, between 1% and 10% (w/v) showed all concentrations
Aquacer outperforms B-44, and the material is second were effective. The 10% concentration was fastest and was
only to Owatrol in increasing density. In the A-series chosen going forward. The same effect was achieved in
there is a significant increase in resistance to compressive approximately twice the time using 1% (w/v) NaOH and,
force, but the effect size is relatively small in comparison at scales where cost or safety concerns make a 10% solu-
to Owatrol and B-44, indicating a lower magnitude of tion prohibitive, a lower concentration may be preferable.
effect by the consolidant on the samples’ strength. In
Samples were immersed in NaOH until the solution
the S-series, strength is less than the control group,
turned dark brown, then changed into successive
although not to a statistically significant degree. Given
NaOH baths until clear. During this process the solu-
the observed decrease in the S-series and the compara-
tion remained the consistency of water; a clear, viscous
tively minor increase in the A-series, Aquacer could not
gel collected on the object surfaces and in the container;
be recommended without additional testing, although it
and, in some instances, a white, hard, waxy substance
does show some promise.
formed on the object surface. When placed into DI water
WS-24 is unsuitable. All three concentrations failed to to remove NaOH, the solution began discoloring again,
show significant improvements in any cast iron samples, and it was found that alternating NaOH and DI water
and the effect in the charcoal samples was minimal. baths was more effective. Reversal behavior differed
between samples aged 43 and 377 days; younger samples decreasing pH, requiring balance between neutralization
required 42 days in alternating NaOH and DI water baths, to dissociate them from iron and neutralization to the
whereas older samples required 73 days. This increase in point of insolubility. While active rinsing with DI water
reversal time likely reflects more complete crosslinking of after the NaOH bath was rapidly effective on test tokens
the material; although still fully reversible after one year, of uncorroded metal reported by Ternisien (2011), the
this behavior represents a potential long-term concern method of alternating DI water and NaOH baths used
on the scale of years to decades. here was equally effective and requires less infrastructure
at a large scale. Using static baths is likely slower, although
All samples lost mass greater than the amount of Owatrol
differences in the test materials make it impossible to
present, except R13, which retained mass equivalent
directly compare these times.
to 3.7% of the Owatrol. Assuming 100% efficiency
in Owatrol removal (Figure 5), the other samples Preliminary characterization of Owatrol was carried out
lost between 0.5% and 2.7% of their starting mass of using a Bruker Alpha attenuated total reflectance-Fou-
graphitized iron, with an average of 1.2%. The uncon- rier transform infrared (ATR-FTIR) spectrophotometer
solidated control samples lost 1.2% and 1.1%, indicating to collect spectra from 4000–400 cm−1, with 64 scans
that damage from Owatrol removal does not significantly collected at a resolution of 4 cm−1. A background spec-
differ from damage as a result of equivalent handling. trum was collected prior to measurement under the same
Losses presented as slight rounding of the sharp edges conditions. The spectra were interpreted using Bruker’s
of the samples. Structural damage was only noted in R1 OPUS software version 7.8 Build: 7,8,44.
(crushed during FTIR analysis) and R11 (slightly chipped
while handling during a solution change). Results
Figure 5. FTIR spectra of iron sample with Owatrol (red) and after reversal Figure 6. Owatrol (red) matched with a 65% soybean oil alkyd (blue).
(black). FTIR has limited ability to detect Owatrol below the surface Image courtesy of The Mariners’ Museum and Park
layer, and although FTIR and weight loss indicate complete removal of
Owatrol, in practice some residue likely remains. Image courtesy of The Owatrol aged naturally for 50 days at ambient conditions
Mariners’ Museum and Park (approx. 21 °C and 50% RH) matched with Nicodom
library entry 01628 for an alkyd oil made with 65%
Chemical characterizations soybean oil (Figure 6). Some peaks do not match the
soybean oil alkyd, indicating other materials in the
Owatrol oil is an alkyd-based polymer, chemically a
Owatrol alkyd. Mass spectroscopy analysis would be
polyester mixed with glycerol (or other polyols) and
useful to identify the components more clearly (and to
organic acids through ester bonds. Alkyds can be broken
monitor potential formula changes by the manufacturer).
down via saponification of these ester bonds at high pH,
producing alcohol and carboxylate functional groups. Detection of Owatrol in iron samples is complicated
Fatty acid carboxylates bond to iron, particularly at grain by changes to the carbonyl peak in the presence of Fe2+
boundaries; their removal is facilitated through neutral- ions. This is indicative of an interaction between the
water rinsing, which interrupts the bond (Kanicky and carbonyl group and iron cations; the exact nature of
Shah 2002, Lundgren et al. 2007, Lobzenko et al. 2021). this interaction has not been characterized but is likely
The aqueous solubility of fatty acids decreases with indicative of bonding behavior. As a result of this, FTIR
of an iron-Owatrol mixture shows the carbonyl peak solvent-born applications at large scale and is readily
decreases and shifts from a single peak at ~1740 cm−1 to applied in a fitted vacuum bag as demonstrated by
two peaks: one still at ~1740 cm−1 and another closer to Mardikian (pers. comm. 2020). Crucially, the material
1630 cm−1 (Figure 7). This was first seen when looking seems highly reversible in aqueous alkali with minimal
for Owatrol in the consolidated samples, and when damage, an essential criterion at the scale necessitated
the strong carbonyl peak at 1740 cm−1 was not visible, by Monitor’s objects. This combination of properties—
additional testing was performed to evaluate the effect application in a fitted bag to reduce solution volume,
of cation coordination with the carboxylates (Otero et al. good consolidation and strengthening, and aqueous
2014). Other than potentially indicating some bonding reversibility—is utterly invaluable for the treatment
behavior with iron (which would likely contribute to of large, graphitized marine archaeological cast iron
Owatrol’s strength as a consolidant), this peak shift artifacts.
has no direct effect on the treatment. However, the
peak shift must be accounted for when using FTIR to Acknowledgments
detect residual Owatrol in samples before and after Many thanks to Nichole Doub and the Maryland Archae-
reversibility testing. ological Conservation Laboratory at Jefferson Patterson
Park and Museum, Dr. William Howard and the Depart-
ment of Engineering at East Carolina University, and
the entire staff at The Mariners’ Museum and Park for
material support and assistance with this research.
References
Argyropoulos, V., S. Boyatzis, and M. Giannoulaki.
2013. The role of standards in conservation methods for
metals in cultural heritage. In Corrosion and conservation
of cultural heritage metallic artefacts, eds. P. Dillmann,
Figure 7. Owatrol mixed with neutralized 1000 ppm Fe2+ Ricca standard D. Watkinson, E. Angelini, and A. Adriaens, 478–517.
(blue), and Owatrol control (red). Image courtesy of The Mariners’ Cambridge: Woodhead Publishing.
Museum and Park
Degrigny, C. 2008. The search for new and safe materials
for protecting metal objects. In Metals and museums in
Conclusion
the Mediterranean: Protecting, preserving and interpreting,
Although FTIR and changes in sample mass serve to ed. V. Argyropoulos, 179–236. Athens: Technical Educa-
partially characterize the reversal behavior of Owatrol, tional Institution of Athens.
additional testing is required. Several samples have Down, J.L., M.A. MacDonald, J. Tétreault, and R.S.
been retained from each treatment phase, which will be Williams. 1996. Adhesive testing at the Canadian Conser-
sectioned and analyzed with SEM-EDS in a follow-up vation Institute: An evaluation of selected poly(vinyl
study to help characterize penetration of the consolidant acetate) and acrylic adhesives. Studies in Conservation
and reversibility beyond the surface layer. 41(1): 19–44.
Paraloid B-44 demonstrated a broadly equivalent consoli- Hamilton, D. 1999. Methods of conserving archaeological
dative function, and its aging properties are better known. material from underwater sites. College Station: Texas
Unfortunately, the technical and regulatory challenges of A&M University.
large-scale solvent baths to reverse the treatment limit Horie, C.V. 1987. Materials for conservation: Organic
Paraloid’s applicability on Monitor. consolidants, adhesives and coatings. Oxford: Butter-
Based on experimental results and pending future work, worth-Heinemann.
Owatrol oil is recommended for treatment of large, Johnson, R. 1984. Removal of microcrystalline wax
heavily graphitized cast iron from Monitor. Although from archaeological ironwork. Studies in Conservation
the application is non-aqueous, it is less hazardous than 29(suppl. 1): 107–109.
Kanicky, J.R. and D.O. Shah. 2002. Effect of degree, type, Phillips, M.W. 1987. Alkali-soluble acrylic consolidants
and position of unsaturation on the pKa of long-chain for plaster: A preliminary investigation. Studies in Conser-
fatty acids. Journal of Colloid and Interface Science 256(1): vation 32(4): 145–52.
201–207. Selwyn, L. 2004. Metals and corrosion: A handbook for
Lobzenko, I., Y. Shiihara, Y. Umeno, and Y. Todaka. 2021. the conservation professional. Ottawa: CCI.
Adsorption enhancement of a fatty acid on iron surface Ternisien, V. 2011. Elimination of two conservation prod-
with Σ3(1 1 1) grain boundary. Applied Surface Science ucts, Owatrol Oil and Poligen ES 91009, in alkaline solu-
543: 148604. tion. Master’s thesis, Université Paris Sorbonne, France.
Lundgren, S.M., K. Persson, G. Mueller, B. Kronberg, J. Wolfram, J., S. Bruggerhoff, and G. Eggert. 2010. Better
Clarke, M. Chtaib, and P.M. Claesson. 2007. Unsaturated than Paraloid B-72? Testing Poligen waxes as coatings
fatty acids in alkane solution: Adsorption to steel surfaces. for metal objects. In Metal 2010: Proceedings of the
Langmuir 23: 10598–602. Interim Meeting of the ICOM-CC Metals Working Group,
MacLeod, I.D. 1981. Shipwrecks and applied electro- Charleston, 11–15 October 2010, eds. P. Mardikian, C.
chemistry. Journal of Electroanalytical Chemistry 118: Chemello, C. Watters, and P. Hull, 167–77. Charleston:
291–303. Clemson University.
Mardikian, P. 2020. Epave du CSS Alabama – Conserva-

Authors
tion du canon RML Blakely et de son chassis pivotant.
Rapport de recherche, Terra Mare Conservation LLC. Erik Farrell is an objects conservator at The Mari-
https://hal.archives-ouvertes.fr/hal-02495211v2/docu- ners’ Museum and Park (TMMP). He worked previ-
ment (accessed 12 January 2022). ously as an archaeological conservator at TMMP for
USS Monitor (1862) and at North Carolina Department
Mardikian, P., Terra Mare Conservation LLC, personal
of Natural and Cultural Resources for Queen Anne’s
communication, 27 March 2020.
Revenge (1718). He received his master’s degree in
Mardikian, P., M. Drews, N. González, P. de Viviés, and conservation from Durham University.
J.W. Hunter. 2009. Searching for a diamond in the rust:
A review of large-scale archaeological iron conserva- Molly K. McGath, PhD, is a research scientist at The
tion projects, 1980–2006. In Iron, Steel & Steamship Mariners’ Museum and Park. She received her doctorate
Archaeology: Proceedings of 2nd Australian Seminar, ed. in materials science and engineering with a focus on
M. McCarthy, 82–88. Fremantle: Western Australian conservation science and her master’s degree in organic
Museum. chemistry from the University of Arizona. McGath
previously conducted research at the Arizona State
McCarthy, M. 2002. Iron and steamship archaeology: Museum, The Library of Congress, the United States
Success and failure on the SS Xantho. Amsterdam: Kluwer National Park Service’s National Center for Preservation
Academic Publishers. Technology and Training, the Smithsonian Institution’s
Näsänen, L.M.E., N.G. González-Pereyra, S.A. Cretté, and Asian Art museums and Museum Conservation Institute,
P. de Viviés. 2013. The applicability of subcritical fluids the United States National Gallery of Art, and Johns
to the conservation of actively corroding iron artifacts Hopkins University.
of cultural significance. The Journal of Subcritical Fluids
Jonathan Echerd is a mechanical engineering student at
79: 289–98.
East Carolina University, focusing on engineering design.
North, N.A. 1987. Conservation of metals. In Conser- He works as a project engineer and draftsman in Wash-
vation of marine archaeological objects, ed. C. Pearson, ington, North Carolina, and has previous experience
207–52. London: Butterworth. in engine and machine shops specializing in NASCAR
Otero, V., D. Sanches, C. Montagner, M. Vilarigues, L. crate engines.
Carlyle, J.A. Lopes, and M.J. Melo. 2014. Characterization
of metal carboxylates by Raman and infrared spectros- Materials list
copy in works of art. Journal of Raman Spectroscopy Acrysol™ WS-24 (acrylic colloidal dispersion)
45(11–12): 1197–206. Talas: www.talasonline.com/Acrysol-WS-24
Aquacer® 8527 (aqueous emulsion of a modified poly-

ethylene wax)
BYK: www.byk.com/en/products/additives-by-name/
aquacer-8527
ASTM A48 C40 Cast Iron (91.9%–94.2% Fe, 3.25%–3.5% C)

SpeedyMetals: www.speedymetals.com/pc-4273-8388-
58-rd-gray-cast-iron-class-40.aspx
Charcoal (willow charcoal, 7–9 mm diameter)

Amazon: www.amazon.com/MyArTool-Char-
coal-Sketching-Drawing-Approx/dp/B07XRSXBCG
Owatrol® oil (alkyd drying oil)

Owatrol: www.owatrolusa.com/product/owatrol-oil/
Paraloid™ B-44 (methyl methacrylate/ethyl acrylate

copolymer)
Talas: www.talasonline.com/Paraloid-B-44
Calcium Sulfonate: Adsorptive Inhibitors for Metal
Conservation?
Eva Wentland
Independent conservator
Co-founder of InduKon – Network for the Conservation of Industrial Heritage
Berlin, Germany
[email protected]
Abstract
Dewatering fluids are used prior to and during the application a literature review of the functional principles of dewatering
of a corrosion protection system to dry surfaces through water fluids as well as an experimental study of the efficiency of a
displacement and to form water-repellent films. Dewatering range of dewatering fluids and coating systems in preventing
fluids originated in the 1930s and their development was atmospheric corrosion processes in pre-corroded steel samples.
boosted during World War II, as a means to protect equipment The latter tests demonstrated that RC T4/20 and Conrasin
recovered from sunken supply ships. The technology was dewatering formulations are able to wick deeply into joints
subsequently adopted by the electroplating, aircraft and and crevices while hydrophobising the wetted surfaces, thus
pipeline industries as well as by other commercial sectors. conferring good protection of lap joints. In combination with
In Germany, dewatering fluids such as RC T4/20 are used by an oil-based binder and coating, dewatering fluids can offer
conservators, mainly craftsmen, for the conservation of metal effective corrosion protection.
objects. RC T4/20 contains calcium sulfonate as an adsorptive
inhibitor but its efficiency, and that of other dewatering fluids, Keywords
has not been scientifically investigated for industrial heritage metal conservation, industrial heritage, calcium sulfonate,
applications or metal conservation in general. We conducted adsorptive inhibitors, dewatering fluids
Introduction
Industrial heritage structures of iron and steel are often the most stable modification of iron oxyhydroxides. As
subjected to atmospheric corrosion, due to wet-dry an inactive, non-conductive component, inert goethite
cycles that result in the formation of corrosion stratig- is unaffected by wet-dry cycles and its stable crystalline
raphies, mainly consisting of magnetite (Fe3O4), goethite structure does not dissolve in water (Dillmann et al.
(α-FeOOH) and lepidocrocite (γ-FeOOH) (Dillmann 2003, Morcillo et al. 2011).
et al. 2003). Wet-dry cycles comprise three stages. In
phase 1 (wetting), a 60% relative humidity (RH) suffices In industrial heritage structures, uniform and crevice
to produce electrolytes, which then catalyse the reac- corrosion occur during wet-dry cycles. Uniform corro-
tion of lepidocrocite crystals on an existing corrosion sion results from the production of differential aeration
layer with the underlying iron. Lepidocrocite acts as cells on metal surfaces under the same atmospheric
an oxidation reservoir and its reduction balances the conditions, leading to a uniform attack of the surface.
anodic dissolution of iron cations. Phase 2 (wetness) Crevice corrosion is localised and develops in the crevice
begins when the reducible lepidocrocite is depleted geometries formed by a metal surface and any other
and the cathodic reaction shifts towards oxygen reduc- metal or nonmetal surface. Examples are joint interfaces
tion. During phase 3 (drying), lepidocrocite and other of structure connections, the debris resulting from dirt
compounds formed during phase 1 are re-oxidised or corrosion products and failed coatings (Figure 1).
to iron(III) compounds. Every wet-dry cycle causes The anodic dissolution reaction within the crevice takes
further corrosion and modifies the composition of the place within an acidic microenvironment, which strongly
corrosion stratigraphy. Every drying phase includes the catalyses the formation of corrosion products (Patel and
potential formation of the α-structure of inert goethite, Bowman 2018).
Moreover, moisture can readily penetrate the corroded

layer, which is unlikely to dry beneath the coating, and
in the case of corroded crevices can remain entrapped
beneath the coating layer. As crevices seldomly reach the
drying phase (phase 3) of the wet-dry cycle, the formation
of inert goethite is unlikely, and crevices serve as large
oxidation reservoirs.
Figure 1. Crevice corrosion (Patel and Bowman 2018)
Stratmann (1994) suggested that corroded surfaces can
be protected by lessening the reactivity of lepidocrocite
The determining factor for the overall rate and severity
crystals. In laboratory environments, the direct conver-
of atmospheric corrosion processes is the length of time
sion of lepidocrocite to goethite may be possible. For
an electrolyte is present on the surface (Hœrlé et al.
practical applications, however, Stratmann proposed
2004). Permanently wet regions corrode faster and more
the covalent bonding of suitable organic monomers to
severely than regions that are exposed to the atmosphere
inhibit the interaction of moisture with the atmospheric
and therefore dry quickly after wetting. Cooper et al.
pollutants that diffuse through the coating material.
(2004) demonstrated a high degree of water retention
in overlapping regions of pristine steel sheet samples
Dewatering fluids
and even higher water retention within corroded spec-
imens. The same study showed that short periods of In Germany, dewatering fluids have been used in industry
wetting can saturate corroded crevices; however, as the and by conservation craftsmen for several decades. In the
drying phase (phase 3) is very long, corrosion processes following, we present a brief literature review, performed
inside the crevice can actively continue even under to gain insights into the composition of the different
dry ambient conditions due to electrolyte retention. A dewatering fluids and their working principles.
comparison of the severity of atmospheric corrosion
One of the first mentions of dewatering fluids can be
processes clearly showed that the most severe damage to
found in the proceedings of a symposium, held in London
iron and steel structures is caused by crevice corrosion,
in 1950, on temporary corrosion protection. Several
whereas uniform corrosion may promote the formation
of the articles dealt with or mentioned amphipathic
of goethite crystals, thus producing corrosion layers that
substances that strongly increase corrosion protection
over long periods of time may confer a protective effect
when added to coating compositions (Sell 1950). Formu-
(Dillmann et al. 2003).
lations containing these surfactants were referred to as
dewatering fluids and were alleged to remove moisture
Corrosion prevention strategies
from metal surfaces by ‘preferential wetting’ (Hollis
Coatings 1950). In this process, amphipathic molecules interrupt
the bonding of water molecules to the metal surface,
The most common method of corrosion prevention for
allowing those molecules to evaporate and the persisting
industrial heritage structures is the application of polymer
amphipathic molecules to form strong covalent bonds
coatings to produce a layer on the metal surface that
with the metal surface in their place. The mechanism is
acts as a barrier to the external atmosphere. Key aspects
based on the strong electronegativity of the amphipathic
considered to be critical for the success of coatings are
molecule’s polar head group, which exceeds that of water,
their longevity and/or their thickness (Brüggerhoff
thus favouring water displacement.
2007, CONSIST 2008, Considine et al. 2010). However,
wet-dry cycles can proceed beneath the coatings because A comprehensive article dealing with the history of
polymeric materials are permeable to moisture and dewatering fluids appeared in Galvanotechnik, an elec-
oxygen. When moisture diffuses through a coating, the troplating journal, in 1962. According to that article,
underlying corroded surfaces containing lepidocrocite dewatering formulations were in use before World War II,
crystals can continue to corrode, since they are not but their development was accelerated by the increasing
dependent on oxygen access, as explained above. The need of allied troops for corrosion protection of valuable
process worsens as soon as the coating becomes damaged. material salvaged from their sunken cargo ships. In the
CALCIUM SULFONATE: ADSORPTIVE INHIBITORS FOR METAL CONSERVATION? 125
recovery of the material, dewatering fluids offered a more crevices. HRCSA contain calcite platelets. On the one
efficient process than flushing the salvaged structures hand, the platelets prolong the moisture diffusion path
with fresh water, hot air drying and coating. The same and increase film stability. On the other hand, perme-
author compared dewatering fluids to solutions of lanolin ating moisture is actively buffered towards alkalinity,
in mineral turpentine with an additive of a water-dis- thus helping to mitigate acidic microenvironments
placing compound, thus combining drying and coating. beneath the coating. ‘High ratio’ describes the ratio of
The use of dewatering fluids subsequently entered the active calcium sulfonate that successively adsorbs on
private sector, namely in machine workshops, electro- the metal surface (O’Donoghue et al. 2009). HRCSA
plating and other undertakings that require the removal coatings have been recommended by Departments of
of water from metal surfaces after certain production Transportation in some states in the USA for the refur-
stages (Steet 1962). bishment of historic bridges (Patel and Bowman 2018).
A lecture series from 1963 presented the concept of ‘ener- The effectiveness of dewatering fluids comes from their
gy-less’ drying with dewatering fluids, as an alternative ability to wet surfaces entirely while deeply penetrating
to the high energy consumption of drying plants. Rather crevices and cracks, where they displace water and
than the adsorption or evaporation of water, it would leave a protective layer (Figure 2). Their total efficiency
involve the replacement of liquid films (Energielose depends on the molecular structure of the adsorptive
1963). Table 1 lists possible compositions of dewatering inhibitor, the surface charge of the metal and the type of
fluids as proposed by Zoltowski (1973). Anionic addi- electrolyte. They are effective when the protective layers
tives such as sulfonates were shown to have the highest are adherent, insoluble and prevent electrolyte access
efficiency in displacing water (Zoltowski 1973). to the metal surface. The adsorption strength depends
upon the electronegativity of the metal or oxide surface
Table 1. Composition of dewatering fluids (from Zoltowski 1973) and the inhibitor molecule. Water molecules on surfaces
Function Type can be displaced only if the reciprocal forces between the
Solvent Primarily white spirit inhibitor and the surface are greater than those between
Surface-active Amphipathic molecules with a hydrophilic head water and the surface (Revie 2011).
additive and hydrophobic tail; those additives with polar
groups containing sulfur or nitrogen (fatty amines
or sulfonic acid derivatives) were usually most
effective
Film former Mineral oils, lanolin or other with eligible dry film
characteristics
Cooper et al. (2005) examined the penetration behav-

Figure 2. Adsorptive inhibitors (Sastri 2011)
iour and water displacement of corrosion prevention
compounds (CPCs) that contained water-displacing A recent article by Sharma et al. (2019) examined the
agents, and particularly those with wide application in molecular dynamics of the adsorption of surfactant-type
the aircraft industry. In aluminium aircraft construction, inhibitors on metal surfaces. According to the authors,
lap joints require corrosion protection as they are prone the hydrocarbon tail of the inhibitor molecule is typically
to crevice corrosion. In this setting, a CPC must be able 12 to 18 carbon atoms long. While the polar head group
to penetrate vulnerable areas and displace any moisture. establishes the link between inhibitor molecule and
The authors of that study developed a successful proce- metal surface, the hydrophobic tails, pointing towards
dure to assess penetration depth and water displacement. the solution, influence the formation of a self-assembled
monolayer (SAM), which is the key step resulting in a
In 2009, O’Donoghue et al. proposed the use of high-
protective function.
ratio calcium sulfonate alkyds (HRCSA), an engineered
coating material based on calcium sulfonate, as the
Dewatering fluids in conservation
monomeric raw material. HRCSA consist of copoly-
merised calcium sulfonate complexed with calcium In the 1970s, Stefan Lasch-Abendroth, a craftsman and
carbonate. The former has an inhibitory function while conservator, employed methods used for drying processes
the latter neutralises the acidic microenvironment inside in electroplating to the conservation of steel objects. In
collaboration with Kurt Kalle, a self-employed chemist, are discussed. Experiment 1 assessed the longevity of
they developed the product Kadosin T4/20, which is still the corrosion protection effect within lap joints, and
being sold under the designation RC T4/20 by the family experiment 2 the kinetic behaviour of dewatering fluids,
business (RC Kalle). RC T4/20 is a water-displacing by examining their penetration depth and velocity.
agent for the ‘thermal-energy-less’ drying of metal
surfaces (Lasch-Abendroth, personal communication, Method
15 May 2019). Its precise components are unknown,
Eight commercially available products were chosen as
but it is assumed to be a blend of aliphatic and aromatic
test fluids (fluids 1–8, Table 2). Since dewatering fluids
hydrocarbons, with small amounts of lanolin, microcrys-
are mostly designed for short-term protection, they
talline wax and, most importantly, calcium sulfonate.
usually contain only small amounts of binding agent,
Calcium sulfonate is an amphipathic compound that is
making additional protective layers necessary (Bena-
often included as the adsorptive inhibitor in dewatering
vides 2009). Owatrol oil (fluid 9, Table 2) was chosen as
fluids. The resultant forces between metal and calcium
the additional coating material, in accordance with the
sulfonate are stronger than those between metal and
conservation practice determined from interviews. Two
water, resulting in preferential wetting and the forma-
clear conservation coatings commonly recommended
tion of a monomolecular layer of calcium sulfonate on
for outdoor exposure (CONSIST 2008) were tested as
the metal surface. The polar heads of calcium sulfonate
well: OKS 2100, a corrosion-prevention wax, and a
molecules face the metal surface while the nonpolar tails
water-curing one-component polyurethane (fluids 10
point towards the atmosphere, yielding a water-repellent
and 11, Table 2).
surface. With its strong wicking power, calcium sulfonate
can penetrate deeply into crevices, such as those that
Experiment 1
form between gusset plates, and into material capillaries.
The resulting surfaces have a very high water repellence Fluids 1–9 (Table 2) were tested by applying them onto
that hinders electrolyte adsorption, thus preventing samples in incremental layers: (1) two layers of dewa-
atmospheric corrosion. Although RC T4/20 is scarcely tering fluid; (2) two layers of dewatering fluid with a
mentioned in the conservation literature, interviews by topcoat of a 50:50 blend with Owatrol oil; (3) two layers
the author revealed that craftsmen use it to facilitate rust of dewatering fluid, an intermediate layer of a 50:50
removal and to dry surfaces and crevices. After surface blend with Owatrol oil and a topcoat of pure Owatrol
cleaning, practitioners apply several layers of RC T4/20, oil. The samples consisted of pre-corroded Q-Panel
pure at first and then as an additive to Owatrol oil, a blend standard steel sheet substrates (76 × 127 × 0.81 mm).
of natural resinifying oils and alkyds. Each set-up contained three samples (9 samples per
fluid, 78 in total). A lap joint was created by bolting
In conservation practices, there has been very little
together two pre-corroded steel sheets using two
research into the functional principles and general
M8 screws and a torque of 12 Nm. The samples were
performance of dewatering fluids. The above literature
spray-coated in three groups and assembled as shown
review and the following experimental study, examining
in Figure 3. After drying, the coated samples were
the efficiency of a range of dewatering fluids, mostly
weighed to three decimal places and then transported
containing calcium sulfonate, were performed as part
to the Hamburg Port Museum in desiccated packaging,
of the author’s master’s thesis, under the auspices of the
where they were mounted onto a rack in compliance
University of Applied Sciences Berlin. The aim was to
with ISO 8565:2011 (International Organization for
create a base of knowledge for further discussion and
Standardization 2011) (Figure 3). The set-up was placed
research, with a particular focus on the use of dewa-
on a harbour crane, where it remained for 12 months.
tering fluids in the preservation of industrial heritage
At the end of the outdoor exposure cycle the samples
structures exposed to the outdoor environment.
were dismantled. The inner surfaces of the lap joints
were examined for the extent of active corrosion and
Experimental
the size of the effectively hydrophobised surface area
Several experiments were carried out to assess the was established by visual assessment. After thorough
performance of dewatering formulations. Two of these drying and removal of any loose particles, the samples
Table 2. Test fluids
Name Substances Company Selection criteria

1 RC T4/20 Hydrocarbons, wax, presumably RC Kalle In use by conservation craftsmen in Germany since the 1990s
calcium sulfonate
2 Anticorit® DF 6000 Hydrocarbons, inhibitor unknown Fuchs Fluid 3 without binding agent
3 Anticorit® DFW 6301 Hydrocarbons, wax, calcium sulfonate Fuchs Considered by the DBU-project for transparent corrosion
(1<5%) protection in the 1990s (Brüggerhoff 1999)
4 Anticorit® DFO 6301 Hydrocarbons, oil, calcium sulfonate Fuchs Fluid 4 with oil as binding agent
(1<5%)
5 Brunox® Turbo Spray Hydrocarbons, oil, sulfonic acids, Brunox® Representative of common penetrating oils
sodium salts
6 Teroson® VR 610 Hydrocarbons, oil, nonylphenol, Henkel Has been in use in combination with Owatrol oil at the
1,2,4-trimethylbenzol, MoS2 Adhesives Hamburg Port Museum for the maintenance of outdoor
objects
7 Conrasin Hydrocarbons, wax, calcium sulfonate Conrads® Lacke Developed on the basis of RC T4/20
(1<3%)
8 White Spirit Hydrocarbons - Chosen as reference medium without inhibitor compound,
used by Jaßmann to increase penetration depth of Owatrol
oil (Boellinghaus 2021)
9 Owatrol® Oil Hydrocarbons, resinifying natural oils, OWATROL® – Binding agent with considerable penetration strength
alkyds Groupe DURIEU
10 EK-PUR Transparent, water curing one Conrads®-Lacke State of the art transparent conservation medium following
component polyurethane Brüggerhoff 1999 and CONSIST 2008
11 OKS® 2100 Corrosion protection wax containing OKS® State of the art transparent conservation medium following
calcium sulfonate Brüggerhoff 1999 and CONSIST 2008
were weighed again to record the change in mass due

to material loss. The results are presented as the average
value from the three samples in a sample set.
Experiment 2
For manageability, only fluids 1, 2, 6, 7 and 8 (Table 2)
were tested pure and in a 50:50 blend with Owatrol oil.
Fluids 10 and 11 were chosen to enable a direct compar-
ison with established coating materials. Two sets of
samples (A and B) were made from pre-corroded Q-Panel
standard steel sheet substrates (76 × 127 × 0.81 mm)
and acrylic glass (76 × 100 × 5 mm). Each set contained
three samples (6 samples per test fluid, 78 in total) and
the average value was determined. For set A, a lap joint
was created by bolting an acrylic and a pre-corroded
steel sheet together using four M8 screws and a torque
of 16 Nm. The lap joint interface had a crevice depth of
100 mm, a crevice width of 76 mm and a crevice height
ranging from < 50 μm around the bolts to > 300 μm in
border areas with less surface pressure. Set B comprised
two pre-corroded steel sheets with joints prepared as in
set A (Figure 4).
A precision dropper was used to apply 100 μL of test

Figure 3. Set-up for experiment 1: rack on a harbour crane at the fluid at the centre of the crevice mouth of the levelled
Hamburg Port Museum with indicated sample numbering sample. Another 100 μL of test fluid was added after 60
and 120 s. The flow front in set A was traced after 30,
90, 150 and 210 s and after 48 h. Set B was disassembled
after 48 h. The lower steel sheet was then dipped into
demineralised water. The hydrophobised, water-repel-
lent areas were traced and their surface areas recorded.
Set A samples were evaluated by comparing the pene-
tration velocity to the size of the wetted area, and set B
samples by determining the size of the hydrophobised
surface area.
Figure 5. Results of experiment 1: gravimetric sample analyses
Figure 6. Detail
of experiment
1. Lap joint
surfaces of sample
112, RC T4/20 as
dewatering fluid
only: considerable
active corrosion, no
Figure 4. Set-up for experiment 2 hydrophobisation
Figure 7. Detail of
Results and discussion experiment 1. Lap
joint surfaces of
Experiment 1 sample 333, DFW
6301 as dewatering
The main findings were:
fluid with Owatrol
• A free exposure of 12 months was too short a time- oil topcoat: no
frame to allow distinct gravimetric analyses; the values active corrosion,
of all samples varied between +1.84 g and −2.98 g. hydrophobisation of
the entire interface
However, because the standard deviation was surpris-
ingly low, the derived tendencies were still informative
(Figure 5). Figure 8. Detail
of experiment
• Uncoated samples scored worst and had the greatest
1. Lap joint surfaces
material loss, as there was no hydrophobisation and of sample 932,
substantial active corrosion occurred within the lap Owatrol oil with
joint. no dewatering
fluid: some
• There was no noteworthy corrosion protection of active corrosion,
samples treated with dewatering fluid only (Figure 6). hydrophobisation
• Samples coated with fluids 1, 3, 4 and 7, applied using of three quarters of
a 50:50 topcoat, showed partial hydrophobisation the interface
inside the lap joint.

• Products 1, 3 and 7 showed very promising results
Experiment 2
when applied according to layer system (3): as total
hydrophobisation of the lap joints, no active corro- The results of set A and set B differed, perhaps because
sion and only a slight mass increase were observed the smooth surface of the acrylic glass posed less resist-
(Figure 7). Pure Owatrol oil was not able to protect ance to wetting than did those samples with corrosion
the entire joint interface (Figure 8). products on both sides of the crevice.
The main findings were: Conclusion

• For set A, the best results were obtained with fluids 1
Among the tested fluids, the characteristics of RC T4/20
and 7, as they penetrated quickly and deeply into lap
and Conrasin were consistent with those of dewatering
joints, creating a hydrophobising effect on the inner
formulations identified in a literature review. This study
joint surface (Figure 9). showed that an effective dewatering formulation based
• Similar to water only, RC T4/20 was able to wick into on the adsorptive inhibitor calcium sulfonate will:
crevice heights of < 0.05 mm, hydrophobising the • readily wet surfaces and penetrate deeply into occluded
entire inner surface, including the screw connections. regions, reaching critical crevice geometries not
As reported by Müller and Tostmann (2017), critical covered by conventional corrosion protection systems
crevice heights that strongly induced crevice corrosion • displace water and dry the metal
ranged from 0.01 to 0.5 mm. • form a hydrophobic film that tightly adheres to the
• For set B, the hydrophobised surface areas achieved surface on a molecular level, effectively lowering the
with the pure dewatering fluids, namely RC T4/20 reactivity of metal surfaces such that water will not be
(100%) and Conrasin (75%), were considerably larger able to wet the surface.
than those obtained with pure Owatrol oil (70%), By protecting not only the surfaces but also the critical
wax (13%) or single-component polyurethane (0%). interfaces of connecting parts, these formulations will
The results of the 50:50 blends of dewatering fluid allow significant corrosion protection of inevitably
and Owatrol oil were between those of RC T4/20 and corroded historic steel structures. Indeed, dewatering
Owatrol oil (96%) (Figure 10). formulations containing calcium sulfonate can reduce
the reactivity of corroded surfaces containing lepi-
docrocite crystals, as proposed by Stratmann already
in 1994.
To provide long-term corrosion protection, dewatering

fluids must be integrated into a system containing
a coating material. When used in combination with
Owatrol oil, the performance of dewatering fluids is
improved, especially at incremental concentration ratios
that successfully fill up capillaries and crevices. Other
strategies should also be explored.
Figure 9. Results of experiment 2: penetration rate of test fluids, time
vs surface area The application of formulations such as RC T4/20 or
Conrasin on corroded metal surfaces did not result in a
notable visual difference. For indoor exposure, however,
with fluctuating relative humidity and the danger of
condensation, the long-term effectiveness of these
barrier layers remains to be assessed. Finally, calcium
sulfonate-based formulations might also provide unob-
trusive but effective corrosion protection also for more
delicate metal objects.
References
Benavides, S. 2009. Corrosion control in the aerospace
industry. Cambridge: Woodhead Publishing Limited.
Boellinghaus, T., E. Wentland, R. Jassmann, R. Keller,
Figure 10. Results of experiment 2: effective hydrophobisation of lap and A. Wolfrum. 2021. Conservation of the Shaft #1
joint interfaces, size of the surface area headgear at the Tsumeb Mine, Namibia: Corrosion
protection. Studies in Conservation. https://doi.org/10 Morcillo, M., D. de la Fuente, I. Díaz, and H. Cano.
.1080/00393630.2021.2004007 2011. Atmospheric corrosion of mild steel. Revista de
Metalurgía 47(5): 426–44.
Brüggerhoff, S. 1999. Korrosionsschutz für umwelt-
geschädigte Industriedenkmäler aus Eisen und Stahl Müller, K. and K.-H. Tostmann. 2017. Lehrbuch der
(Modellvorhaben), DBU-Az: 06834. Metallkorrosion. Bad Saulgau: Leuze Verlag.
Brüggerhoff, S. 2007. Beschichtungssysteme für Indus- Patel, C. and M.D. Bowman. 2018. Pack rust identifi-
triedenkmale im Spannungsfeld zwischen Ästhetik und cation and mitigation strategies for steel bridges (Joint
Dauerhaftigkeit. industrie-kultur 3: 28–29. Transportation Research Program No. FHWA/IN/
JTRP-2018/16). West Lafayette, IN: Indiana Department
Considine, B., J. Wolfe, K. Posner, and M. Bouchard. 2010.
of Transportation and Purdue University. https://doi.
Conserving outdoor sculpture: The Stark Collection at the
org/10.5703/1288284316788
Getty Center. Los Angeles: Getty Conservation Institute.
O’Donoghue, M., P. Roberts, and V. Datta. 2009. Over-
CONSIST. 2008. Comparison of conservation materials
coating lead-based alkyd paint on steel penstocks:
and strategies for sustainable exploitation of immov-
Practical experience. In Overcoating bridges and other
able industrial cultural heritage made of iron and steel.
structures, a JPCL eResource, 30–38. Pittsburgh, PA:
A project supported by the European Commission,
Technology Publishing Co.
Contract No. 513706.
Revie, R.W., ed. 2011. Uhlig’s Corrosion Handbook, 3rd
Cooper, K.R., Y. Ma, J.P. Wikswo, and R.G. Kelly. 2004. ed. Pennington: John Wiley & Sons.
Simultaneous monitoring of the corrosion activity and
moisture inside aircraft lap joints. Corrosion Engineering, Sastri, V.S. 2011. Green corrosion inhibitors: Theory and
practice. Hoboken, NJ: John Willey & Sons.
Science and Technology 39(4): 339–45.
Sell, G., ed. 1950. Journal of the Institute of Petroleum 36.
Cooper, K.R., K.W. Furrow, G. Feng, and R.G. Kelly. 2005.
Wicking and water displacement by corrosion prevention Sharma, S., X. Ko, Y. Kurapati, H. Singh, and S. Nešić.
compounds in simulated aircraft aluminium alloy lap 2019. Adsorption behaviour of organic corrosion inhib-
joints. Corrosion 61(2): 155–66. itors on metal surfaces. Corrosion 75(1): 90–105.
Dillmann, Ph., F. Mazaudier, and S. Hœrlé. 2003. Steet, W. 1962. Entwässerungsflüssigkeiten und ihre
Advances in understanding atmospheric corrosion of Verwendung in der Galvanotechnik. Galvanotechnik
iron. I. Rust characterisation of ancient ferrous artefacts 53(8): 387–95.
exposed to indoor atmospheric corrosion. Corrosion Stratmann, M. 1994. Wie rostet Eisen und wie kann
Science 46(6): 1401–29. man verrostete Eisenoberflächen vor einem weiteren
‘Energielose Trocknung mit Dewatering Fluids’. 1963. Korrosionsangriff schützen? In AdR-Schriftenreihe zur
Galvanotechnik 54(1): 158–59. Restaurierung und Grabungstechnik, Schutz der Metall
gegen atmosphärische Korrosion, ed. K. Götz, 11–15.
Hœrlé, S., F. Mazaudier, Ph. Dillmann, and G. Santarini.
Hallstadt.
2004. Advances in understanding atmospheric corrosion
of iron. II. Mechanistic modelling of wet-dry cycles. Zoltowski, Z. 1973. The behaviour of dewatering formula-
Corrosion Science 46(6): 1431–65. tions. Anti-Corrosion Methods and Materials 20(3): 18–20.
Hollis, H. 1950. Performance tests on temporary corro- Author

sion preventives. Journal of the Institute of Petroleum
36(320): 495–504. Eva Wentland is a freelance conservator specialised
in metals and modern materials. She is member and
International Organization for Standardization (ISO).
co-founder of InduKon, Network for the Conservation
2011. ISO 8565:2011: Metals and alloys – Atmospheric
of Industrial Heritage. Recent projects were the conser-
corrosion testing – General requirements. Geneva: Inter-
vation of two harbour cranes in the city of Hamburg, as
national Organization for Standardization.
well as the completion of her master thesis with the title:
Lasch-Abendroth, S., Private restorer, Hamburg, personal “Transparent conservation of large heritage objects in
communication, 15 May 2019. outdoor exposure – Dewatering fluids, an investigation
for use in conservation”. In 2017 she received a bache-

lor’s degree in conservation and restoration of modern
materials and technical heritage from the University for
Applied Sciences in Berlin. She has been employed as a
conservator in the Germany Mining Museum and has
worked and completed internships in several museums
and freelance conservation studios, where she has
focussed on composite objects from everyday life or with
an industrial context.
Investigating a Sustainable Alternative: L-Cysteine

as a Non-Toxic Corrosion Inhibitor for Copper Alloy
Conservation
Emilie Tréhu* Dean Sully
Museum of Fine Arts Boston UCL Institute of Archaeology
Boston MA, USA London, UK
Abstract
Treatments consisting of 1-, 6- or 24-hour immersion in 1, 3 BTA treatment. Immersion time had a greater effect than
or 5% (w/v) solutions of L-cysteine in deionised water were concentration, although surface changes increased with both.
compared to a standard benzotriazole (BTA) treatment using Limitations in the experiment offer insights into further research
artificially corroded copper alloy (brass) coupons. The results that could determine L-cysteine’s suitability for conservation.
were assessed by visual observations, gravimetric measurements
and microscopy/spectroscopy and suggested that L-cysteine Keywords
treatments consisting of a higher concentration and longer L-cysteine, benzotriazole (BTA), copper alloys, corrosion
immersion time were more effective than the standard inhibitors, bronze disease
Introduction
The presence of nantokite (copper(I) chloride) beneath by bronze disease) (Abu-Baker et al. 2013, 29; Gravgaard
more stable layers of oxide and carbonate corrosion and van Lanschot 2012, 14, 23).
poses a well-known threat to archaeological copper
Conservation corrosion inhibitors must exhibit high
alloy objects. With moisture, nantokite oxidises to form
inhibition efficiency, long-term stability, safety for
powdery green copper(II) chloride hydroxides, such as
handling and reversibility, while being readily avail-
paratacamite, in a catalytic reaction, eventually causing
able, easy to maintain in storage and relatively low cost
the corrosion of any remaining metal in a process known (Abu-Baker et al. 2013, 230). The amino acid L-cysteine
as bronze disease (Grayburn et al. 2015, 222; Selwyn (HO2CCH(NH2)CH2SH) is non-toxic, biodegradable,
2004, 67). relatively inexpensive and easily produced at high purity
Conservation approaches, such as chloride removal and (El Ibrahimi et al. 2020, 743). Its sulfur atom is a strong
environmental control, are often supplemented by the electron donor, resulting in the enhanced adsorption
use of corrosion inhibitors. Since 1967, benzotriazole of L-cysteine to copper and its alloys and thus in the
(BTA, C6H5N3) has been the primary corrosion inhibitor formation of a stable copper–cysteine complex on the
for copper alloy conservation (Sease 1978, 81–82). BTA object’s surface (Abu-Baker et al. 2013, 230). Additionally,
is popular in conservation due to its established use L-cysteine can act as a zwitterion, allowing a positive
and simple application that does not require removing charge under acidic conditions, such as those in chloride
corrosion pits, in turn increasing chemisorption with the
pre-existing corrosion products or interfere with objects’
copper (Gravgaard and van Lanschot 2012, 15).
metallic composition. The resulting dark-green appear-
ance is generally deemed an acceptable conservation Three previous studies examined L-cysteine as a corro-
surface (Golfomitsou and Merkel 2004, 346, table 1). sion inhibitor for archaeological copper alloys (Gravgaard
However, BTA loses efficacy within five years, is an and van Lanschot 2012, Abu-Baker et al. 2013, Argy-
environmental and health hazard, is more effective on ropoulos et al. 2017), but systematic comparisons with
copper than on its alloys and exhibits reduced efficacy in BTA are lacking. In this study a standard BTA treatment
low-pH environments (i.e. chloride ion-rich pits caused was compared with selected variations in L-cysteine
INVESTIGATING A SUSTAINABLE ALTERNATIVE: L-CYSTEINE AS A NON-TOXIC CORROSION INHIBITOR FOR COPPER ALLOY CONSERVATION 133
treatment to assess the effects of the treatments. The 2006, 392). Following treatment, the coupons were aged
mechanism of inhibition was not investigated. simultaneously in a Memmert ICH260L climate chamber
at 80% RH and 25°C for 42 days (Leygraf and Graedel
Experimental design 2000, 17, as cited in Golfomitsou 2006, 114).
Factors for testing

Data analysis
Due to time and resource limitations, a single BTA treat-
In consultation with Stavroula Golfomitsou (personal
ment was employed based on Golfomitsou (2006) and the
communication, 25 June 2019), due to time and resource
standard traditionally used at the Museum of London: a
limitations it was determined that the most appropriate
1-hour immersion in 3% w/v (0.26 M) BTA in industrial
techniques for evaluating the experimental results for this
methylated spirit (IMS). Without a current standard for
short-term study were visual observations, gravimetric
L-cysteine use in conservation, three concentrations and
measurements and scanning-electron microscopy with
immersion times were assessed (Table 1). Deionised water
energy-dispersive X-ray spectroscopy (SEM-EDS). All
was identified in previous studies as the most efficient
solvent for L-cysteine (Gravgaard and van Lanschot coupons were documented after pre-corrosion, treatment
2012, Abu-Baker et al. 2013, Argyropoulos et al. 2017). and accelerated ageing with photography and observa-
Low (1% w/v, 0.08 M), intermediate (3% w/v, 0.25 M) tion at 52.5× magnification using a Dino-Lite Pro HR
and high (5% w/v, 0.4 M) concentrations were chosen to AM7000/AD7000 5-megapixel digital microscope. The
compare the range of previously tested L-cysteine concen- visual observations evaluated the aesthetic effects of
trations to the standard BTA treatment. As Gravgaard the inhibitors for conservation use. The coupons were
and van Lanschot found no difference in efficacy between weighed after pre-corrosion, after treatment and during/
24-hour and 7-day L-cysteine immersions of bronze after accelerated ageing. An equation of mass change
samples (2012, 15–16), 1-, 6-, and 24-hour immersions over time was used to calculate the percent inhibition
were selected in this study for comparison with the (%I) (Faltermeier 1999, 125), with the standard BTA
standard BTA treatment. treatment representing the control group:
%I
= (control weight change − test weight change) × 100
Table 1. Test parameters
| control weight change |
Test Inhibitor Immersion
Inhibitor Solvent
name concentration time
After accelerated ageing, a representative coupon from
BTA BTA 3% (0.26 M) IMS 1h
A L-cysteine 1% (0.08 M) Deionised water 1h
each test group was chosen based on the Dino-Lite
B L-cysteine 1% (0.08 M) Deionised water 6h
documentation for SEM-EDS analysis, performed
C L-cysteine 1% (0.08 M) Deionised water 24 h using the parameters listed in Table 2. EDS line-scans
D L-cysteine 3% (0.25 M) Deionised water 1h allowed estimation of the distribution of Cu, Zn, Cl, O,
E L-cysteine 3% (0.25 M) Deionised water 6h C, N and S on the coupons’ surfaces, and SEM analysis
F L-cysteine 3% (0.25 M) Deionised water 24 h allowed comparison of the surface morphological features
G L-cysteine 5% (0.4 M) Deionised water 1h (Abu-Baker et al. 2013, 23; Golfomitsou 2006, 128–29).
H L-cysteine 5% (0.4 M) Deionised water 6h
Table 2. SEM-EDS analytical parameters
I L-cysteine 5% (0.4 M) Deionised water 24 h
Carl Zeiss EVO 25 variable pressure SEM with
Instrument electron microprobe and Oxford Instruments
Coupon preparation Aztec X-ray spectrometer
As brass alloys were underrepresented in previous Accelerating voltage 20.00 kV

Data collection time/
studies, in this study 20 × 50 × 1 mm brass coupons Dead time
100 s/35%–40%
(37% Zn/63% Cu) were selected for testing. The coupons Working distance 8–10 mm
were pre-corroded following the Faltermeier procedure Magnification 90–330×
(Faltermeier 1995, 124). Five coupons for each test group Image type Secondary electron (SE)
were immersed in each inhibitor for the selected time, Method used for EDS
Point and ID
then oven-dried for 30 minutes at 50°C (Golfomitsou collection
Results
Visual observation
Figure 1 shows a representative coupon from each test
after pre-corrosion (before treatment). Figure 2 shows
those coupons’ immediate colour change after inhib-
itor treatment (before accelerated ageing) and Figure 3
shows those coupons after 42 days of accelerated ageing.
The change between Figures 1 and 2 is consistent with
a general trend in L-cysteine immersion time across
Figure 3.
Representative
coupons after
accelerated
ageing
all concentrations: a green colour change after 1 hour,

purple areas after 6 hours, and grey/white surfaces after
24 hours. Higher concentrations caused the darkening
of green and purple areas and increased the surface area
and thickness of the purple and white layers. The change
Figure 1. between Figures 2 and 3 shows that 1-hour treatments
Representative produced dark-green and purple mottled surfaces, and
coupons after coupons in the 6-hour treatments were covered in soft
pre-corrosion
bright-green, white and light-blue corrosion products,
whereas the 24-hour treatments retained areas of pink
metal, ringed by light-blue and purple corrosion prod-
ucts. The grey appearance of L-cysteine on bronze
coupons was noted by Abu-Baker et al. (2013, 236) and
Gravgaard and van Lanschot (2012, 19), but the colour
changes in our samples were more purple than grey,
perhaps due to the coupons’ brass composition.
Gravimetric
Table 3 shows the average %I for each L-cysteine group,
calculated against the average change for the BTA group as
the control. The higher the %I recorded for the L-cysteine
test groups, the more effective the treatment compared to
the BTA treatment. Figure 4 depicts the trends in the %I
values shown in Table 3. The %I of the 1-hour immer-
Figure 2.
sions followed a negative trend over time, whereas the
Representative
coupons after 24-hour immersions followed positive trends from day 7
inhibitor post-treatment onwards. A longer-term assessment would
application reveal whether this positive trend is maintained.
Table 3. Average %I of tests A–I compared to BTA
Test name Percent inhibition after treatment (AT) and at each weigh-in day (D)
AT 1D 2D 7D 14D 18D 35D 42D
A −14.29 100.00 66.67 75.00 57.14 53.85 0.00 −9.10
B −171.43 216.66 166.67 125.00 135.71 146.15 166.67 172.73
C −328.57 266.66 222.22 162.50 178.57 184.62 191.67 209.10
D 0.00 116.66 77.78 68.75 14.29 15.38 25.00 18.18
E −57.14 183.33 133.33 106.25 107.14 107.69 91.67 90.91
F −200.00 250.00 200.00 156.25 164.29 169.23 175.00 172.73
G 14.29 133.33 66.67 68.75 28.57 23.08 0.00 9.10
H −57.14 216.66 144.44 100.00 100.00 107.69 100.00 90.91
I −242.86 350.00 288.89 200.00 214.29 223.08 233.33 254.55
SEM-EDS tities of Cu, Cl and O (Figure 6), corresponding to

copper(II) chloride hydroxides such as paratacamite.
Figure 5 shows a representative SEM image from each Extensive EDS scans across the surfaces of the L-cysteine-
coupon’s surface. The EDS analysis of the voluminous treated coupons indicated that the elements similarly
mounds seen on the BTA coupon identified large quan- associated with the copper(II) chloride hydroxides were
Figure 4. Trends in tests A–I’s %I compared to BTA
Figure 6. SEM-SE images and spectrum 10 from BTA coupon as an

example of the voluminous mounds also seen on A, B, D, E and G
Figure 7. SEM-SE images and spectrum 102 from coupon C as example

Figure 5. SEM image from each test’s representative coupon of the small bunches also seen on H, F and I
results, higher concentrations and immersion times may

create a stronger inhibitive film due to the sulfur atom of
L-cysteine complexing with the copper alloy substrate,
with a 24-hour immersion as the most effective. Together,
our findings suggest the efficacy of 24-hour immersion
in L-cysteine in deionised water as a corrosion inhibitor
treatment for copper alloys. While the treatment time is
longer than the standard BTA treatment, L-cysteine offers
the advantage that it is more environmentally sustainable.
In order to assess L-cysteine’s suitability as a routine

conservation treatment, further tests should include a
narrower range of concentrations as well as other alloys
Figure 8. SEM-SE images and spectrum 37 from coupon I as an example
that allow testing for possible dezincification in brass and
of the flat expanses above small bunches also seen on F
the influence of the mobility of other elements. Studies
of L-cysteine’s long-term efficacy under ambient ageing
only similarly visible on tests A, B, D, E and G, corre-
conditions and analyses using a broader range of analyt-
sponding to the voluminous mounds seen in the SEM
ical techniques (XRD, FTIR, XPS, Raman spectroscopy)
images (cf. Figure 5). By contrast, EDS scans of the small
should also be pursued prior to experimenting with
bunches seen in the SEM images of C, F, H and I showed
corrosion surfaces on heritage objects (Golfomitsou
a much lower Cl content and a much higher S content,
2006, 137).
suggesting the presence of copper–cysteine complexes
and the reduced presence of possible copper(II) chloride
Acknowledgements
hydroxides, evidenced by the absence of the voluminous
mounds (Figure 7). The flat expanses above the small This research was undertaken as part of the lead author’s
bunches on tests F and I corresponded to the grey crusts dissertation for the MSc in Conservation for Archae-
seen on those coupons. The corresponding EDS scans ology and Museums at the Institute of Archaeology
showed an even higher S content (Figure 8). (IoA), University College London (UCL), under the
supervision of Dr Dean Sully. Thank you to Dr James
Conclusion Hales and Dr Agnese Benzonelli for facilitating access
to the climate chamber at the IoA and Dr Tom Gregory
After treatment and before ageing, the lower concentra-
for facilitating and assisting with SEM-EDS analysis at
tions and immersion times of L-cysteine caused a colour
the IoA’s Wolfson Archaeological Science Laboratories.
change that included a dark-green surface very similar
Thank you to Helen Ganiaris, Rebecca Lang and Luisa
to that resulting from BTA treatment. The other treat-
Duarte for their feedback during development of the
ments created uneven purple, grey and white surfaces.
experiment and facilitating access to the resources of the
The 24-hour treatments, however, retained more of the
Archaeological Conservation Laboratory at the Museum
coupons’ pre-ageing pink-brass colour. BTA treatments
of London. Thank you to Dr Stavroula Golfomitsou, Mari
are normally followed by a protective coating, such as a
Gravgaard and Susi Pancaldo for sharing their own expe-
Paraloid acrylic lacquer, which could reduce L-cysteine’s
riences and research with copper alloys and L-cysteine,
greying effect to an acceptable dark-brown/grey, thus
which inspired and shaped this project.
mimicking the stable patinas expected as a visual standard
for copper alloys (Gravgaard and van Lanschot 2012,
17–20, figs. 4, 5). The gravimetric results suggested that
References
L-cysteine’s inhibitive efficacy increases with concentra- Abu-Baker, A.N., I.D. MacLeod, R. Sloggett, and R.
tion and immersion time, but that the immersion time Taylor. 2013. A comparative study of salicyaldoxime,
has the greatest effect. The trends in %I over 42 days of cysteine, and benzotriazole as inhibitors for the active
accelerated ageing suggest that a 24-hour immersion of chloride-based corrosion of copper and bronze arti-
L-cysteine is more effective than BTA treatment as an facts. European Scientific Journal 9(33): 228–51. www.
inhibitor for brass coupons. According to the SEM-EDS academia.edu/35914429/A_Comparative_Study_of_
Salicylaldoxime_Cysteine_and_Benzotriazole_as_Inhib- Gravgaard, M. and J. van Lanschot. 2012. Cysteine as a

itors_for_the_Active_Chloride_Based_Corrosion_of_ non-toxic corrosion inhibitor for copper alloys in conser-
Copper_and_Bronze_Artifacts (accessed 12 April 2022). vation. Journal of the Institute of Conservation 35(1):
Argyropoulos, V., S. Mavroforaki, M. Giannoulaki, S.C. 14–24. www.tandfonline.com/doi/full/10.1080/194552
Boyatzis, T. Karabotsos, and A. Zacharopoulou. 2017. 24.2012.681618 (accessed 30 October 2021).
New approaches in stabilizing chloride-contaminated Grayburn, R., M. Dowsett, M. Hand, P. Sabbe, P. Thom-
ancient bronzes using corrosion inhibitors and/or electro- pason, and A. Adriaens. 2015. Tracking the progression
chemical methods to preserve information in the patinas. of bronze disease – A synchrotron X-ray diffraction study
In Artistry in bronze – The Greeks and their legacy, eds. of nantokite hydrolysis. Corrosion Science 91: 220–23.
J. Daehner, K. Lapatin, and A. Spinelli. Los Angeles: J. Sease, C. 1978. Benzotriazole: A review for conservators.
Paul Getty Trust. www.getty.edu/publications/artistry- Studies in Conservation 23(2): 76–85. www.tandfonline.
inbronze/conservation-and-analysis/38-argyropoulos/ com/doi/pdf/10.1179/sic.1978.011?needAccess=true
(accessed 30 October 2021). (accessed 30 October 2021).
El Ibrahimi, B., A. Jmiai, L. Bazzi, and S. El Issami. 2020. Selwyn, L. 2004. Metals and corrosion: A handbook for the
Amino acids and their derivatives as corrosion inhibi- conservation professional. Ottawa: Canadian Conserva-
tors for metals and alloys. Arabian Journal of Chemistry tion Institute.
13(1): 740–71. www.sciencedirect.com/science/article/
pii/S1878535217301430?via%3Dihub (accessed 12 April Authors
2022).
Emilie Tréhu is Assistant Objects Conservator at the
Faltermeier, R.B. 1995. The evaluation of corrosion
Museum of Fine Arts Boston. She completed the MSc in
inhibitors for application to copper and copper alloy
Conservation for Archaeology and Museums at Univer-
archaeological artefacts. PhD dissertation, Univer-
sity College London in 2019.
sity College London, UK. http://discovery.ucl.
ac.uk/1317947/1/307582.pdf (accessed 30 October 2021). Dean Sully is Associate Professor of Heritage Conser-
Faltermeier, R.B. 1999. A corrosion inhibitor test for vation at University College London.
copper-based artifacts. Studies in Conservation 44(2):
121–28. www.tandfonline.com/doi/abs/10.1179/
sic.1999.44.2.121 (accessed 30 October 2021).
Golfomitsou, S. and J.F. Merkel. 2004. Synergistic
effects of corrosion inhibitors for copper and copper
alloy archaeological artefacts. In Metal 04: Proceed-
ings of the International Conference on Metals Conser-
vation, Canberra ACT, 4–8 October 2004, eds. J.
Ashton and D. Hallam, 344–68. Canberra: National
Museum of Australia. https://pdfs.semanticscholar.
org/5536/82fccb266db33b61710b6d24997cbf2b068e.
pdf ?_ga=2.104540739.232566934.1567168789-
1771302291.1567019025 (accessed 30 October 2021).
Golfomitsou, S. 2006. Synergistic effects of additives to
benzotriazole solutions applied as corrosion inhibitors
to archaeological copper and copper alloy artefacts. PhD
dissertation, University College London, UK.
Golfomitsou, S., Senior Lecturer in Conservation at the
Department of Conservation, University of Gothenburg,
Sweden, personal communication, 25 June 2019.
TECHNICAL STUDIES
Sustainable Growth of Mewari Mail Makers: Traditional to
Contemporary Practices
Vandana Singh
The City Palace Museum
Udaipur, Rajasthan, India
[email protected]
Abstract
In the age of cutting-edge technologies, artisanal methods are also adapted as well as issues related to the participation of
gaining increasing appreciation. In India, the community of craftswomen. The collective knowledge gained from the Mewari
karigars (craftsmen and -women) consists mostly of daily-wage mail makers was put into practice as part of an ongoing Arms
earners. This research examines one such group, traditional and Armours Conservation project at the City Palace Museum,
Mewari chain-mail makers. There are no written sources Udaipur, where it will contribute to a better understanding of the
describing the material culture of this community, although armour collections housed in this and other museums.
chain mail was in high demand in the past. Over the years, as
chain mail has become almost exclusively of historical interest, Keywords
chain-mail making is being abandoned, as artisans use their Mewar, mail makers, karigars, baghtar, armour, riveted rings,
skills to make objects of contemporary interest. This research traditional craftsmen, craftswomen
explored how chain-mail-making skills are being sustained but
Introduction
The City Palace Museum, Udaipur, India, houses an arms by Indo-Aryan peoples after 2000 BC (Bryant 2001).
and armour collection from the late 15th to the 18th Armour was prominently mentioned in the Rigveda,
century that consists of complete sets of body armour, the oldest known Indian Vedic Sanskrit text (between
trousers, helmets and armguards of soldiers and cavalry 1500 and 1000 BC) (Jamison and Brereton 2014). Vedic
as well as armour for women and children. As part of a warriors wore armour made of rhinoceros and deer skins
conservation project, the armour was examined scientifi- but some wore metal armour called varman, which is
cally. However, despite the large-scale production of mail described in the Rigveda as sewn armour or a coat of mail
in the past, its provenance is largely unknown and there that covers the whole body. Armour is also mentioned
are no written sources or other information identifying in the Arthashastra, an ancient Indian Sanskrit treatise
the craftsmen who made the armour or the workshop on military strategy from 321–296 BC.
where they worked. Thus, in this study a survey was
Medieval chain mail armour was light and flexible,
conducted in Mewar, Rajasthan, India, where chain mail
balancing the defensive capabilities of the armour with
is still produced by a few craftsmen, mostly from a single
the offensive capabilities of the wearer. By the late 14th
family. Craftsmen who had switched to contemporary
century, this flexible defence was reinforced with rigid
practices were also interviewed. The survey showed that
plates to cover vital areas of the torso. Later on, to cope
the mail produced today is very similar in appearance
with the intense heat, soldiers decreased the weight
to medieval riveted mail and that its makers continue to
of their armour. Eventually, however, due to technical
take pride in their work.
advancements and the increased reliance on archery, most
soldiers, except those in the cavalry, did not wear heavy
Tracing the development of body armour in India
metallic body protection. This development accompanied
Weapons and armour have been prized in India since very a transition from mail and plate armour to a new type
early times. Defensive armour was introduced into India of armour strongly influenced by contemporary Persian
SUSTAINABLE GROWTH OF MEWARI MAIL MAKERS: TRADITIONAL TO CONTEMPORARY PRACTICES 141
designs, in which a ‘four mirrors’ (chahar a’ineh) type of small and specialised workshops in the princely states of
armour, composed of four plates secured by straps, was India, including those where mail-making was performed
worn over a mail shirt (zerah baghtar) (Bottomley and by karigars (craftsmen). In addition to Mewar, many
Bowstead Stallybrass 2000). As muskets became widely mail-making workshops were located in Hyderabad.
used, in the 16th century, soldiers sought protection Sophisticated tools and expertise were not available at
against projectiles, thus favouring closely modelled the local level but were probably concentrated within
plate cuirass over mail armour. Musket-proof cuirasses a more extensive arms-manufacturing workshop. Skill
consisted of two layers of metal plate, one softer than the was acquired in a very structured manner, with strict
other. The outer layer was designed to absorb a bullet’s rules and methods. Mohammad Baig further added
energy, and the thicker, softer inner layer to slow the that, traditionally, wire was drawn manually, a highly
bullet and thus limit its penetration. strenuous task. Challa is a local term for iron rings. Each
challa was annealed many times during the mail-making
According to its construction technique, mail armour
process. The earlier challa had round holes for riveting,
can be divided into three main categories: armour made
whereas later forms had wedge-shaped holes. The ends
of riveted rings, armour made of alternate rows of riveted
of the challa were pierced together or sometimes welded
and solid rings, and armour made of butted rings (Sharma
with heat. This was a slow process requiring precision and
1986). In rivetted iron rings, wire at the end ring is flattened
dedication. At this stage, the challa were joined together
and punched to receive a rivet. Until the 18th century, the
to shape the baghtar (mail shirt). In traditional construc-
majority of Indian mail was characterised by alternate rows
tion, according to Mohammad Baig, the production of
of riveted and solid rings (welded) made with a 4-in-1
a full baghtar from start to finish might take a year. The
pattern (Metcalf 2001). However, by the end of the 18th
initial, laborious production tasks were left to apprentices
century, this stronger mail was increasingly replaced by
and assistants. Then, the most skilled task was done by
more decorative butted mail (Richardson and Bennett
the master craftsman, who would have been supplied
2015), in which the ends of the ring were simply butted
with the necessary rings and rivets.
together, not joined. Butted iron, copper and brass rings
can be arranged in a horizontal chevron or zig-zag pattern.
The addition of brass and copper rings to the traditional
iron structure enabled craftsmen to produce fine mail with
a range of geometrical designs. However, as the insecure
method of fixation was useless in battle, this armour was
mostly made for ceremonial purposes.
The art of armour making

The most important source of information for this
research was the Mirza’s family, a family of traditional mail
makers from Mewar. The title Mirza was conferred upon
the family, who continued to use it when they migrated
to Udaipur 60 years ago from Karauli, Sawai Madhopur,
in Rajasthan. Mirza Ishtakar is the only member of the
family who continues to dedicate himself to assuring Figure 1. Two generations of Mirza’s family: 99-year-old Mirza Mohammad
Baig (right) with his older son (middle) and his younger son Ishtakar (left)
that the tradition in which he was trained grows and
flourishes. All other family members have entered other However, for the last 40 years karigars have used a tool/
professions. The Mirza’s family has been engaged in machine for ring-making and riveting (Figure 2) that also
mail-making for at least 150 years. Ishtakar’s training flattens the ring and punches a hole into it. According
and knowledge came from his father, Mirza Mohammad to the size of the ring, the flattening frame is adjusted.
Baig, who learned it from his father (Figure 1). In talking As demonstrated by Ishtakar, the riveted rings are made
about the history of armour-making in India, 99-year-old by winding a wire on a rod so as to resemble a spring
Mirza Mohammad Baig said that there had been many (Figure 3). After the ring has been made, the link that
142 ICOM-CC | METAL 2022 | HELSINKI, FINLAND TECHNICAL STUDIES
goes through the hole is severed and the ring is slid off A prerequisite for broader social recognition of both
the rod. Next, the burma – an upper attachment of the craftsmen and craftswomen is that they value their own
hole-punching machine – is removed to complete the work. Therefore, the couple was invited to demonstrate
rivet (Figure 4b). their skills in the conservation laboratory at the City Palace
Museum, Udaipur (Singh and Seth 2021). In this more
appreciative environment, they worked alongside conser-
vators in restructuring and repairing damaged mail armour
(Figure 5). For Shahnaaz Beg, in addition to taking pride in
her work, this opportunity gave her a sense of progress in
gaining social acceptance as a skilled craftswoman.
Figure 2. Ring-making tool/machine
Figure 4. Shahnaaz Beg demonstrating the process of making rived

rings using a machine
Figure 3. Wire winding on the rod to form a spring
Women at work
Mirza Ishtakar’s wife Shahnaaz Beg also makes riveted
rings and assists him in other tasks in the workshop
(Figure 4). Normal body armour usually weighs 12–15 kg,
depending on the design or construction. For the 4-in-1 Figure 5. Shahnaaz Beg and Ishtakar repairing lost areas of mail and
pattern, in which each ring passes through four others, plate armour at the arms and armour conservation laboratory, The City
Palace Museum, Udaipur
tens of thousands of repetitions and therefore extraordi-
nary patience are required. Shahnaaz, who learned from
her husband, is highly skilled and has been helping in the Reproduction of armour by contemporary
workshop for 14 years while continuing to maintain the
chain-mail makers
family’s household. Unfortunately, her work in armour- Mirza Mohammad Saeed Ahmad is a close relative of
making receives little social recognition, to the point that Mirza Ishtakar who moved to the city of Meerut, Uttar
when asked what she does she replied: ‘Nothing much. Pradesh (Figure 6). He is no longer involved in traditional
We become soldiers in an invisible army and acquire mail-making but instead makes weapons and armour
skills to survive’. for use by the Historical Medieval Battle Association
(HMBIA), an international sports organisation created

to develop, support and promote HMB, which combines
a true sport with the preservation of historical cultural
heritage.1 Among historical re-enactment groups, espe-
cially those whose focus is on antiquity or the Middle
Ages, mail is commonly used both as practical armour
and as a costume. Reproductions are much less expensive
to procure and to maintain than historical ones and allow
craftsmen such as Mirza Mohammad Saeed Ahmad to
earn more money in less time.
Figure 7. Shri Gopilal Lohar at his workshop
also honoured by the Government of India with its ‘Shilp

Guru’ award, for excellence in traditional craftsmanship.
In a discussion of the evolution of chain mail armour

in India, Gopilal cited Bundi, Alwar, Rampur, Bhinder
and Sirohi as arms manufacturing centres in Medieval
India. Those in Bundi and Alwar included numerous mail
makers. At that time, armour production was expensive,
due to the amount of time and material required to make
a mail shirt. In addition to production by local small
workshops, armour was exported from northwest India
through horse trading with Iran and Afghanistan.
Figure 6. Mirza Mohammad Saeed Ahmad at his workshop. He is a maker
of contemporary armour for the Historical Medieval Battle Association Abbas Ali Sagar, a 79-year-old master tailor by profes-
(HMBIA) for participants in HMB, an international sport
sion, used to visit the conservation laboratory at the City
Shri Gopilal Lohar is 75 years old (Figure 7) and a Palace Museum to translate the Arabic inscriptions on
third-generation craftsman in a family of traditional inlay the weapons. He provided several important observations
blacksmiths well-known particularly for their produc- about the different mail items. For example, chain mail
tion of ornamental weaponry in Udaipur, Rajasthan. is not uniform across the object, as the mail is thick and
Previously, Shri Gopilal was a traditional weapons chunky at the upper chest, finer towards the shoulder,
manufacturer but he now contributes to innovation in and even finer towards the wrist. Cavalry armour is knee
traditional craftsmanship, with a wide-ranging business length and spilt up the front and up the back, allowing
that includes customers in Europe and the US. In the early the wearer to sit astride a horse. The armour at the
19th century, the Maharana of Udaipur was so impressed knee is very fine. A true karigar would create mail with
with the craftsmanship of Gopilal ancestors that he varying size rings to allow for flexibility and a precise fit.
granted them royal patronage and gifted them lands in A properly fitted mail that does not restrict movement
perpetuity. Initiated into the craft by his grandfather and requires a considerable investment of work.
constantly encouraged by his father, Shri Gopilal none-
Conclusion: A treasure of talent at risk
theless distanced himself from the traditional Mewari
form of blacksmithing to become famous for his exclu- Mewari karigars continue to pass along their knowledge
sive blade-making skills. Today, with his son, he runs a and traditions. Most of these craftsmen come from the
workshop equipped with traditional as well as advanced same background or even family. Today, they still take
machinery (Figure 8). His use of technology has allowed pride in their identity as artisans and in being Mewar
him to provide customisation on a large scale. He was karigars, despite having switched to the production
through generations and which make up the cultural

and artistic fabric of India. However, the economic and
human costs of the 2020–2021 pandemic caused many of
these craftsmen to lose their livelihoods, if not their lives.
It is thus all the more important to recognise, protect and
value their skills, which are irreplaceable. An important
lesson of the pandemic is the need for an ecosystem that
can protect and expand this community, as their skills
will contribute to India’s prosperity. Furthermore, the
skills of craftswomen remain invisible, as women are
often considered unskilled and unproductive. Efforts to
ensure the public acknowledgement of women workers
as a skilled society are urgently needed.
Acknowledgements
Incorporation of traditional practices into contempo-
rary practices would not have been possible without
the interest and support of Shriji Arvind Singh Mewar
of Udaipur, Chairman and Managing Trustee of the
Maharana Mewar Charitable Foundation. The author
thanks all karigars of Mewar, Rajasthan, who shared
knowledge about their material culture, and the technical
staff Dhwani Meghani and Srabani Gayen who partici-
pated in the project.
Notes
1
See https://hmbia.info/ (accessed 14 March 2020).
References
Bottomley, A. and H. Bowstead Stallybrass. 2000. Galva-
nised Indian mail. In Royal armouries yearbook 5, 133–38.
Leeds: Royal Armouries Museum.
Figure 8. Use of traditional smelting and forging methods in the

Bryant, E. 2001. The quest for the origins of Vedic culture:
manufacture of weapons and armour at the workshop of Shri Gopilal The Indo-Aryan migration debate. Oxford: Oxford
Lohar University Press.
Jamison, S.W. and J.P. Brereton, transl. 2014. The Rigveda:
of contemporary items using modern practices. Their
The earliest religious poetry of India, 3 vols. Oxford:
knowledge allows them to adapt inherited traditions to
the contemporary context without compromising their Oxford University Press.
pride as karigars. However, as materials and techniques Metcalf, S. 2001. The treatment of mail on an arm guard
evolve, a body of meta knowledge has developed that from the armoury of Shah Shuja: Ethical repair and in
enables this community to quickly adjust their practices situ documentation in miniature. V&A Conservation
to modern times. Journal 37: 4–6.
In India, there are thousands of karigars earning a living Richardson, T. and N. Bennett. 2015. Indian arms and
using traditional techniques that have been passed down armour, 2nd ed. Leeds: Royal Armouries.
Sharma, R.K. 1986. The military system of the Mewar

(Udaipur) State (ca. 800 to 1947 A. D.). Central Asiatic
Journal 30(1/2): 116–40.
Singh, V. and H. Seth. 2021. The synergy of craft, conser-
vation, and cutting-edge technology: Conservation of
arms and armour at the City Palace Museum, Udaipur,
India. In Transcending Boundaries: Integrated Approaches
to Conservation: ICOM-CC 19th Triennial Conference
Preprints, Beijing, 17–21 May 2021, ed. J. Bridgland. Paris:
International Council of Museums.
Author
Vandana Singh is head of the Metal Conservation Labo-
ratory at the City Palace Museum, Udaipur. She is the
owner of the Heritage Preservation Centre Kalasampadaa
and co-founder of the Centre for Art Conservation and
Research Experts, LLP in New Delhi. Her interdiscipli-
nary research focuses on uniting traditional knowledge,
conservation and science, and exploring the material
culture of a traditional metal craftsmen community of
South Asia.
Development of Mexican Bell Documentation Tools and
Methodologies
Jannen Contreras Vargas* Francesc Llop Álvaro

National School of Conservation, Restoration and Museography (ENCRyM), Conservatory of Music “Joaquín Rodrigo”
National Institute of Anthropology and History (INAH) Valencia, Spain
Mexico City, Mexico [email protected]
[email protected]
Francesc Llop I Bayo
Daniela Lira Pacheco Retired anthropologist
National School of Conservation, Restoration and Museography (ENCRyM), Valencia, Spain
National Institute of Anthropology and History (INAH) [email protected]
Mexico City, Mexico
[email protected] Fanny Magaña Nieto
Oaxaca INAH Center, National Institute of Anthropology and History (INAH)
Ángel E. García Abajo Oaxaca, Mexico
National Coordination of Conservation of Cultural Heritage (CNCPC), National [email protected]
Institute of Anthropology and History (INAH) *Author for correspondence
Mexico City, Mexico
[email protected]
Abstract
Mexican bells suffer from theft, melting, recasting, and other system, and, where possible, architectural context. The
forms of destruction, along with a lack of documentation that inventory included a photographic record, sound recording,
can be used to support prosecution. This paper describes the metallographic sampling, and chemical analyses. The selection
process begun in 2015 to develop specific documenting tools criteria were based on accessibility, date of manufacture, and
and methodologies to help preserve Mexican bells, and the first functionality as ringing instruments.
complete documenting exercise performed on 98 bells at the
Historic Center of Oaxaca, Mexico, in 2019. The first step was Keywords
to fill a form with information about each bell’s characteristics bells, bell frames, documentation, information gathering,
and location, sound and use, state of conservation, fastening inventory
Introduction
Smashing historic bells by throwing them from their with the organization of the Seminario Permanente de
towers prior to melting and recasting is a common Conservación de Campanas [Permanent Bell Conser-
but shocking practice in Mexico. Often, broken bells vation Seminar].
are simply left lying on the ground outside buildings.
The current structure of the methodology was strongly
Bells fulfil a variety of relevant social functions and
influenced by specialists such as Francesc Llop I Bayo,
document metallurgical techniques and history, but
an anthropologist and bell ringer at the Cathedral of
the absence of appropriate records makes it easy for
Valencia, Spain, who pioneered bell documentation that
bells to end up stolen, melted, and recast, or otherwise
resulted in the database <campaners.com>. In addition,
lost without trace.
the assistance from Soledad Díaz Martínez of the Cultural
In this context, the development of a complete docu- Heritage Institute of Spain (IPCE) was instrumental in
mentation methodology was inspired by the theft of two the development of conservation data forms and method-
18th-century bells for scrap from Escobedo, Montemo- ologies (Contreras Vargas et al 2015). The first complete
relos, a small town in northern Mexico. The Metallic documentation using these tools was performed on the
Heritage Conservation Laboratory at the National bells from the Historic Center of Oaxaca, Mexico, in
School of Conservation, Restoration, and Museog- 2019. This work describes the tools developed to provide
raphy (ENCRyM) began a rearguard action in 2015 a proper response to this problem.
DEVELOPMENT OF MEXICAN BELL DOCUMENTATION TOOLS AND METHODOLOGIES 147
Mexican bells Documentation

Compared to European countries where bells can be Documentation is a key part of research and conser-
found dating from the 11th century, such as in San vation. Documenting and recording are often wrongly
Lorenzo or Laurentina, the oldest in Spain, bells and confused and assumed to be synonymous, at least in the
the bell-making tradition are relatively new in Mexico. Spanish-speaking world. For this project, documenting
Christian traditions reached Mexico in the 16th century. the information was seen as a complex and constant
Metallographic studies on Mexican bells carried out by task requiring processes for recording, searching, and
our team have demonstrated that the casting quality was compiling, followed by revising and expanding what
already existed (Nagel 2009, 8). Letellier and Eppich
quite different from that in Europe. One of the main
(2011, 119) have defined this process as “the systematic
reasons for this was the constant scarcity and high price
collection of information and archiving of records in
of materials such as tin. Another was the increasing need
order to preserve them for future reference.” The action
for bells by the new Catholic churches, compounded by
of documenting must take into consideration the level
the absence of a structured body of artisans, such as a
of detail, scope, and methods applicable, and the forms
guild of bellfounders, that could provide training, rules,
in which records should be kept and made available.
and impose standards. These factors and the use of inap-
For bells, this data includes observations about their
propriate clappers resulted in bells suffering from early properties, chemical and physical alterations, the various
deterioration, including deformations, fractures, and analytical techniques and their results, and even the
detachment of elements, along with a lack of recognition decision-making process in the event that a conservation
of these elements as part of Mexican cultural heritage. intervention is required (Appelbaum 2007, 394).
The net result was that bells ended up being neglected
and discarded. What and how much to document?
The destruction of bells has also been the result of various The documenting tools developed required an under-
social conflicts, such as wars, during which their metal standing that bell conservation involves the study of
would be melted down to make cannons. Today, the material elements, sound, the meaning and function of
destruction of bells by recasting is conducted by artisan sound, the meaning and social function of the bell, and
bellfounders under the approval of civil and religious the specific language of this kind of cultural heritage,
authorities, who are unaware of the law and wrongly with all its elements, problems, and the way they interact.
believe these actions to be right and even needed. The experiences of Dr. Francesc Llop I Bayo and his
It must be said that recasting is a destructive procedure. inventory of Spanish bells were combined with the
Conservation and Repair of Bells and Bellframes: Code
The crystalline structure and composition of the alloy is
of Practice issued by the UK’s Council for the Care of
lost, as is the surface layer, decorations, and inscriptions,
Churches (2011), whose documenting and conserva-
even the profile and exact size, as often these features are
tion criteria refer to the age and uniqueness of bells and
not analyzed, documented, or accurately reproduced.
their sound. However, since the bell-making tradition in
The lack of records on historic bells, such as their char-
both Spain and the UK differs significantly from that in
acteristics and date of manufacture, etc., has inhibited
Mexico, which has a far shorter history, the documenta-
prosecution of those who have destroyed these unique
tion must also include bells made during the 20th century.
pieces of our material cultural heritage. The lack of The initial scope was to prepare a documenting tool that
records also means that it is impossible to even measure would be accessible to bell ringers, sacristans, and bell
how much has been lost. However, under the legal frame- enthusiasts, so that they could complete a database of bells
work supporting the National Institute of Anthropology that needed to be protected. However, it soon became
and History (INAH), which protects Mexican cultural clear that only conservators could readily document, fill,
heritage, all of the movable cultural property in religious and use the tool, as it included concepts on preventive
buildings from prior to the 20th century constitutes part conservation, deterioration dynamics, and tangible and
of the country’s heritage, including bells. intangible characteristics.
carillon. It also records the type of architectural element

to which it is attached (belfry, campanile, bell tower), the
fastening system—since it is possible for swinging bells
to lose their wooden counterweight (headstock) and
stop working—, and the characteristics of the fastening
system.
Interaction with bell specialists and enthusiasts at confer-

ences during the conservation process allowed the tool to
be expanded with the different names used in the country
for the parts of bells, supports, and architectural features.
Figure 1. Part of the team seen while recording data on the size of the bells
The conservation data forms

The conservation data forms developed since 2015 were
tested in the field and then corrected and improved. After
the 2017 earthquakes in Mexico, it was evident that it was
necessary to record the yokes, headstocks, wheels (even
though Mexican bells usually have no wheels), etc., and
all kind of supports, as well as important information
on the architectural features of bell towers, bell gables,
etc. This was in order to understand how they interacted Figure 3. Two views of the damage seen on bells after the 2017
and to record relevant data to analyze why these architec- earthquake: on the left, a collapsed bell in a bell tower; on the right, the
result of a weakened wooden counterweight
tural elements were damaged and had even failed, thus
helping to prevent recurrence of the same problems as In the absence of chemical analysis in the field, composi-
far as possible. tion is inferred through visual inspection of the metal and
The current form documents the location of the building, corrosion products. Data on the state of conservation of
the location of the bell within the building, and the name the striker/clapper and its fastening system, as well as the
of the person in charge of filling out the form. It includes bell’s fixing, are also recorded, along with the decorations
general measurements and notes on whether the bell is and any engraving on the surface, since this information
swinging or fixed, and whether it is part of a clock or is very important to place it in its historical context.
Figure 2. The conservation data form developed by our team

The second, and last, page is dedicated to recording all tools to selfie sticks and cell phones with acceptable photo
visible damage, such as corrosion, dirt, deformation, quality—as suggested by Dr. Llop I Bayo from the begin-
cracks, fractures, and signs of vandalism. Thus, the data ning—and the photos to whatever could be collected by
as a whole creates a picture of the bell both as a utilitarian two people under regular outdoor lighting conditions.
object and a historical document of the building and the Using these devices, the team recorded different views of
community to which it belongs. the bells, including from the exterior, from a safe place
inside the belfry.
First inventory
The most complete result was the inventory of the bells
at the Historic Center of Oaxaca. This was carried out
with guidance from the staff and students from ENCRyM,
Dr. Francesc Llop I Bayo, musician and bell specialist
Dr. Francesc Llop Álvaro, lecturers and students from
the School of Conservation and Restoration of the West
(ECRO) in Mexico, and different INAH work centers.
The inventory covers 98 bells in 31 buildings, most of
Figure 4. Photographic records of bell exteriors taken with a selfie stick
which are Catholic churches. The inventory includes
photographic records, sound recordings for 61 bells, A total of 98 bells were documented in this fashion
and metallographic sampling and analysis of 47 bells. with good quality photographs that can be used as
Its publication is intended to encourage other groups to legal evidence in the event of any wrongdoing. The
undertake similar documentation in order to create a team is aware that using a drone would also permit safe
national inventory. and secure digital video and still recording, but that is
currently not a possibility.
Basic features
Sound characterization
Primary characterization information was obtained via
direct observation and by using measuring tapes and One of the most important and irrecoverable losses that
laser measurement devices. occurs when bells are recast or removed is their sound—
which is part of the day-to-day life of a population. The
The team was aware that visual assessment is limited
sound of 61 of the 98 bells was recorded. These recordings
by subjectivity, knowledge, skill, and access to the bells.
were carried out under the guidance of Dr. Francesc Llop
However, getting access to churches and bell chambers is
I Bayo using a Tascam DR-44WL recorder and processed
not easy. Obtaining permission can take a considerable
and analyzed using the methodology developed by Dr.
amount of time, also travelling to the location, so conser-
Francesc Llop Álvaro to document the sound of Spanish
vators may not have much time to devote to each bell.
cathedral bells (Llop Álvaro 2017).
This is a critical point, as Mexico has a massive number of
bells. Visual identification of materials and a conservation Analysis of the sound begins by studying the different
assessment are useful guides, but, whenever possible, frequencies that make the complex sound produced by
data from chemical analysis, metallography, or X-ray the bells. The bell profile controls the sound to a great
fluorescence (XRF) spectroscopy should be included. extent. Its distinctive partials––which are audio charac-
teristics, since the basic striking note is just one of the
Photographic record sounds a bell makes when struck––are identified. The
most important partials are the harmonics. Inharmonic
The original plan to take a photographic record involved
partials follow a different rule (Rossing 1984, 440–47).
inspection teams using portable ladders, cameras, tripods,
and even cycloramas in the narrow and high stairwells For the analysis of the recordings, Wavanal software,
of the bell towers. However, the access, security, and created by William A. Hibbert, was used. Hibbert (2008)
safety limitations, and significant quantities of bird and developed this software for his doctoral thesis, but it is
bat guano, quickly reduced the arsenal of photographic important to note that it was designed to analyze tuned
bells, such as those used in carillons or clock bells. Subse- issues were resolved in 2019 by taking metallographic
quent analysis and elaboration of the results was therefore samples for laboratory analysis. The samples provide
essential. By collecting the results of the different partials, stratigraphic evidence of coatings and their thickness
an image of the bell’s acoustic behavior could be created. over the original cast alloys. All of the sectioned samples
were catalogued for future reference and may be useful
in the event of a bell being destroyed, allowing those
responsible to be prosecuted.
During the 2019 program, a total of 47 samples were taken

from the most accessible bells. Sampling was conducted
in the least damaging way possible by making tangential
cuts in inconspicuous areas that had no structural effect
on the bell at the loops (cannons) or at the outer edge of
the lip. For broken bells, the sample was taken from the
fracture area, since this provides the maximum amount
of metallurgical data. Sample sizes were typically 3 or
4 mm long by 1–2 mm wide for bells in good condition
and slightly larger on fractured bells. The samples were
prepared for metallographic analysis and surveyed with
optical and scanning electron microscopy (SEM), while
the chemical analysis was performed by energy-dispersive
X-ray spectroscopy (EDS).
Figure 5. Sound record taken at a bell tower with the help of the church’s
bell ringer
The importance of this data lies in the fact that it allows

researchers to determine the predominant audio struc-
tures of Mexican bells––if that can be said to exist––and,
where possible, help to re-create it in the case of a cracked
or lost bell.
Metallographic characterization Figure 6. Sampling and analysis process: (1) on-site sampling, (2) sample
preparation, (3) ready sample, (4) SEM-EDS analysis, (5) metallographic
During the initial study in 2015, nine of the bells from study under optical microscopy
the Cathedral of Oaxaca were analyzed using noninva-
Due to the composition and metallographic studies
sive analysis techniques, specifically XRF spectroscopy.
carried out on the bells since 2015 (García et al. 2015),
However, this method only characterizes surface concen-
it is clear that the casting quality of most Mexican bells
trations of elements. The bulk composition can differ
is quite different from that in Europe. This finding was
significantly from the surface, especially in objects that
confirmed by the bells in Oaxaca, which differ signifi-
have undergone heat treatment, as was seen in several
cantly in composition from the normal European “bell
of the bells analyzed.
bronze” standard (78% Cu-22% Sn). Of the 56 bells
Furthermore, XRF equipment has important restrictions analyzed, only one had a content of 22% Sn, with the
in terms of air transport and is awkward to work with in average being only 5.5%. One of the alloys, from a bell
the tight and frequently unsafe spaces inside bell towers. made in 1870, was closer to brass than bronze because
These management and occupational safety and health of its zinc content.
Analysis also determined the presence of different surface Conclusion

materials, such as patinas, coatings––such as paintings––,
The documenting tools and methodologies developed
and gilding, on 12% of these bells. This is very impor-
and described in this paper were intended to be a
tant, since these materials are usually removed during
means of protecting bells and their sounds. While they
“conservation” processes, such as abrasive cleaning, as
demonstrated their usefulness, it is necessary to develop
it is commonly, and mistakenly, believed that bells have
other, simpler processes to facilitate the recording and
no relevant surface materials.
protection of as many bells as possible. Carrying out
documentation in places such as belfries involves diffi-
culties and safety hazards due to the dimensions and
obstructions inherent to these spaces, and bird and bat
guano, so simpler strategies––such as the use of cell
phones and selfie sticks over ladders and professional
cameras, sampling instead of using XRF, and taking
simple sound recordings in the bell chamber, leaving
more complex tasks for the laboratory––proved to be
the most useful.
Acknowledgments
We would like to thank María Álvaro, the conservators
Figure 7. Ternary diagram of the Sn-Zn-Pb composition of the bells at Soledad Díaz, Marcela López, Salem Ojeda, Gabriela
the Historic Center of Oaxaca. Copper is the main constituent Peñuelas, David Vega, Inés Mejía, and Urani Correa, our
fellow teachers and students at ECRO, the Alfredo Harp
Ongoing work Helú Oaxaca Foundation (FAHHO), and Ryszard Rodys
The records were compiled during 2019, but the pandemic for their kind support.
stopped much of the project from being completed.
Currently the data is being uploaded onto a single data Notes
system (SUR) at the INAH. However, unfortunately, 1
àgina oficial dels Campaners de la Catedral de València
P
according to Mexican law, only cultural objects prior to [Official page of the Bell Towers of the Cathedral
the 20th century can be added. Since 20th-century bells of Valencia (in Catalan)] (accessed 18 April 2022).
need protection as well, a parallel database has been http://campaners.com/
created. Meanwhile, Dr. Llop I Bayo is adding data on
the bells to the General Inventory of Oaxaca Bells on the References
well-known platform www.campaners.com.1 Appelbaum, B. 2007. Conservation treatment method-
The team is also constructing a conservation-focused ology. Burlington: Butterworth- Heinemann.
catalog with information and relevant conservation Contreras Vargas, J., G. Peñuelas Guerrero, I.M. López,
conclusions for the 23 bells in the main temples of the and D. Pascual Cáceres. 2015. Tañendo problemas: La
Historical Center of Oaxaca: Santo Domingo, Basilica conservación de campanas sonoras en México. Un semi-
of la Soledad, and the Cathedral. This catalog includes nario permanente. CR. Conservación y Restauración 7:
all the collected data, including metallographic analysis 55–61. https://revistas.inah.gob.mx/index.php/cr/article/
and the sounds recorded. view/12029
Future developments in sound recordings will include a Council for the Care of Churches. 2011. The conservation
more in-depth examination of the complex strike note to and repair of bells and bellframes. Code of practice, 2nd
determine what defines the profiles (shapes) of Mexican rev. ed. London: Church House Publishing.
bells in relation to this factor. This can only take place García, A., J. Contreras, G. Peñuelas, and M. López.
once there are enough recorded sounds of Mexican bells 2015. Caracterización de dos campanas de composición
to have a broad and statistically representative sample. y fracturas atípicas. In MetalEspaña 2015. Actas del II
Congreso de Conservación y Restauración del Patrimonio dination Body (CNCPC-INAH) and a lecturer at the
Metálico, Segovia, Real Casa de Moneda, 1–3 October National School of Conservation (ENCRyM) in Mexico.
2015, eds. E. Cano and J. Barrio. Madrid: Universidad He has a degree in chemical sciences, with a specialization
Autónoma de Madrid. in metallurgy, from Complutense University of Madrid
Hibbert, W.A. 2008. The quantification of strike pitch (UCM) in Spain. He has studied historical metallurgy
and pitch shifts in church bells. PhD thesis, Department since 1996 and is a PhD candidate in ancient world
of Design, Development, Environment and Materials, studies (Universidad Autónoma de Madrid, Spain).
Faculty of Mathematics, Computing and Technology, Francesc Llop Álvaro trained as a pianist at the
The Open University, UK. https://doi.org/10.21954/ Conservatori Professional de Música de València (CSM)
ou.ro.0000add2 in Spain and the École Normale de Musique de Paris
Letellier, R. and R. Eppich, eds. 2011. Recording, documen- Alfred Cortot in France. He has a PhD in art history
tation and information management for the conservation from the Universitat de València (UV) in Spain, for
of heritage places, 1st ed. London: Routledge. which he wrote a thesis entitled “Bells in the Hispanic
cathedrals. Analysis, cultural significance, conservation,
Llop Álvaro, F. 2017. Las campanas en las catedrales
and rehabilitation.” Currently, he is a lecturer at the
hispanas. Análisis, significado cultural, conservación y
Conservatori Superior de Música “Joaquín Rodrigo” de
rehabilitación. Doctoral thesis, University of Valencia,
València in Spain.
Spain. https://roderic.uv.es/handle/10550/59835
Nagel, L. 2009. Manual de registro y documentación de Francesc Llop I Bayo has a PhD in social anthropology
bienes culturales. Santiago: Centro de Documentación from Complutense University of Madrid (UCM) in Spain.
de Bienes Patrimoniales. He joined the Dirección General de Patrimonio de la
Generalitat Valenciana in 1988, where he was head of
Rossing, T.D. 1984. The acoustics of bells: Studying the the archaeology and ethnology department from 1991
vibrations of large and small bells helps us understand the to 2001 and head of the museum department from 2002
sounds of one of the world’s oldest musical instruments. until his retirement in 2014. He has been involved in the
American Scientist 72(5): 440–47. conservation of about a hundred bells and bell towers,
including roughly twelve cathedrals. Along with Mari
Authors Carmen Álvaro Muñoz, he has documented thousands
Jannen Contreras Vargas is a conservator with of bell ringers. He manages the website http://campaners.
a degree from the National School of Conservation com with information about bells, bell ringers, and bell
(ENCRyM) in Mexico, an MSc in forensic conservation ringing.
(University of Lincoln, UK), and a PhD in ancient world Fanny Magaña Nieto is a conservator specializing in
studies (Universidad Autónoma de Madrid, Spain). She the conservation of musical instruments. She trained
has been a lecturer of conservation methodology and at the National School of Conservation (ENCRyM) in
applied chemistry at ENCRyM since 2001. Since 2015 Mexico and is coordinator of the conservation depart-
she has organized seminars and conferences on bell ment at the INAH Oaxaca Center (CINAHO). She is a
conservation in Mexico. member of the Mexican Organ Studies Seminar and the
Daniela Lira Pacheco is a conservator with a degree Mexican Bell Conservation Seminar and has overseen
from the National School of Conservation (ENCRyM) in the management of conservation projects for cultural
Mexico, which she completed with a dissertation entitled heritage affected by the 2017 and 2018 earthquakes
“Analysis of intervention possibilities in fractured sounding and the project to catalog bells in the Historic Center
bells.” She is a member and organizer of the seminar on bell of Oaxaca.
conservation, part of the Metallic Heritage Conservation
Laboratory (LCPM), and co-supervisor of ENCRyM’s
Metals Conservation Workshop Seminar.
Ángel E. García Abajo is an associate researcher at

Mexico’s National Cultural Heritage Conservation Coor-
Corrosion Detection by Color Change Using Crowdsourced
Photographs. Preliminary Results of the MIPAC Project
Blanca Ramírez Barat* María Teresa Molina Emilio Cano

National Centre for Metallurgical National Centre for Metallurgical National Centre for Metallurgical
Research, Spanish National Research, Spanish National Research, Spanish National
Research Council (CENIM-CSIC) Research Council (CENIM-CSIC) Research Council (CENIM-CSIC)
Madrid, Spain Madrid, Spain Madrid, Spain
Abstract
One of the most obvious signs of metal corrosion is color change. color change and corrosion extent need to be determined in
This phenomenon could be used to assess the corrosivity of order to explore the possibilities of quantifying the extent of
the indoor environment in museums and collections in other corrosion from the calibrated photographs. This paper presents
institutions. The MIPAC project is therefore developing a the preliminary results of a validation study, based on tests
preventive conservation tool for monitoring the impact of the of a series of copper and silver coupons exposed in museum
indoor environment on cultural heritage materials, based on the showcases for 6, 12, and 18 months. Color and corrosion
color changes associated with deterioration. The changes are extent measured by gravimetry and galvanostatic reduction
assessed by analyzing photographs taken with cell phones and showed the good correlation between some of the colorimetric
provided by museum visitors and staff through an app. A color information and the corrosion thickness. Although the uneven
reference chart was designed, in which metal coupons are used distribution of surface corrosion complicates the establishment
as dosimeters. CIELAB color coordinates are extracted from the of a correlation between corrosion and color change, this issue
images and adjusted using the color references with a calibration can be tackled by a localized evaluation of the corrosion obtained
algorithm. This approach allows the preventive conservation by electrochemical reduction.
of cultural heritage from both a scientific-technological and
a social perspective, since it involves public participation in Keywords
the conservation of museum collections. However, for this preventive conservation, color, monitoring, image analysis,
monitoring approach to be valid, the correlations between corrosion
Introduction
The degradation of metals and other cultural heritage cultural heritage materials, based on the color changes
materials is commonly associated with changes in their associated with metal corrosion. The basis of this tool is
visual appearance, especially their color: silver objects the relationship between color and corrosion, the ability
darken and tend to blacken upon exposure to sulfur to monitor color changes through image analysis, and
gases; copper turns reddish during the first stages of the use of citizen science.
oxidation and then darkens or assumes a greenish The use of metals and the changes in their appearance as
color as corrosion proceeds; iron changes from silvery dosimeters to detect aggressive environmental conditions
to reddish with the formation of rust layers; and lead is well established and is the basis of the well-known Oddy
darkens due to the formation of thin lead oxide layers, test. Colorimetric and spectrophotometric methods have
shifting to whitish as thicker carbonate layers are formed. been used elsewhere to determine the color changes in
As corrosion processes are the result of ambient condi- metals as indicators of corrosion (Tétreault et al. 2003,
tions, the color changes of metal references can serve Lafuente et al. 2013), but a quantitative correlation
as an indicator of corrosive environments. Thus, the between color change and corrosion extent (quantified
aim of the MIPAC project1 is to develop a tool that can as loss of mass or thickness of the base metal, or as the
be used to monitor the impact of the environment on thickness of corrosion layers) has not been established
due to several difficulties (Lafuente 2017). For instance, dinates of the metal dosimeters, which are then related to
Ankersmit (2001) reported uneven tarnishing on flat the extent of corrosion. The robustness of the calibration
silver samples and an unreliable correlation with the method and its ability to adjust for the color differences
percent color change. in cell phone images from different cell phones have
been demonstrated (Barbero-Álvarez et al. 2021b), and
In the MIPAC project, crowdsourced photographs rather
tests with real metal samples conducted. The results
than spectrophotometers are used in color evaluation,
have thus far shown that color changes are detected with
with the aim of establishing a quantitative correlation
a high degree of confidence, as shown by comparisons
between color change and corrosion extent. The feasi-
with direct colorimetric measurements made using a
bility of using cell phone photographs to monitor color
spectrophotometer (Table 1) (Ramírez Barat et al. 2021).
changes was demonstrated by Brigham et al. (2018). The
This allows at least to detect significant changes in the
use of citizen science as a data collection method has two
metal dosimeters that act as indicators of the presence
major advantages: first, it does not require investments
of harmful emissions in a museum showcase or location.
of additional resources by the museum, and second, it
actively involves the public in museum practices and the Table 1. Measured and calculated L*a*b* coordinates for a silver
conservation of museum objects. coupon (from Ramírez Barat et al. 2021)
In the MIPAC’s approach, copper and silver strips acting 12A_Ag Measured Calculated
as dosimeters are mounted as a color reference chart for L* (10°/D65) 72.44 68.45
use in image calibration, as shown in Figure 1 (Ramírez a* (10°/D65) 5.00 4.64
Barat et al. 2021). The “MIPAC panels” are displayed in b* (10°/D65) 26.82 26.55
the locations to be monitored.
An important step to improve the precision and reliability
of the method is to establish a quantitative relationship
between color changes and corrosion extent. Ideally, a
calibration curve between color coordinates and corro-
sion thickness would allow to measure the corrosion
of metals and classify the corrosivity of a museum
environment, using the MIPAC panels as dosimeters.
Accordingly, we examined the correlation between
corrosion extent and color change in copper and silver
coupons exposed to real museum environments for up
to 18 months at the National Museum of Science and
Figure 1. MIPAC panel for the monitoring of color changes of metal
coupons. The color chart was printed using UV resistant inks on acid-free
Technology (MUNCYT) and at the museum of the
paper covered with propylene Chemistry Faculty of the Complutense University, both
located in Madrid.
For image collection, a dedicated app2 was created that
allows for better control of the cell phone camera and the
Experimental
automatic uploading of photographs to the image-pro-
cessing servers. The integration of a gamification scheme To analyze the relationship between color and corro-
in the app promotes user engagement. sion extent, 1 × 5 cm copper and silver coupons were
cut, drilled, sanded with P1200 grit, and mounted on
To extract the color information from the photographs, an
a MIPAC color reference panel (Figure 1). The panels
image calibration system was designed and subsequently
were placed inside four different museum showcases
validated in collaboration with the Visual Telecommuni-
(labeled S1 to S4), two in each of the two aforementioned
cations Applications Research Group (GATV) from the
museums (Figure 2).
Technical University of Madrid (UPM) (Barbero-Álvarez
et al. 2020, Barbero-Álvarez et al. 2021a). MIPAC panels Additional sets consisting of 12 coupons of each metal
are photographed by visitors and the images are then were hung on plastic racks, according to ISO 11844-2
processed and calibrated by applying linear and spatial standard (International Organization for Standardization
calibration algorithms to obtain corrected L*a*b* coor- 2020), and exposed in the same locations. From this racks,
CORROSION DETECTION BY COLOR CHANGE USING CROWDSOURCED PHOTOGRAPHS. PRELIMINARY RESULTS OF THE MIPAC PROJECT 155
Corrosion was measured by gravimetry as mass gain,

with the values then transformed to the thickness of the
corrosion layer by assuming that the composition of the
corrosion layer was silver sulfide (Ag2S) for the silver
coupons and cuprite (Cu2O) for the copper coupons.
The plot of the corrosion layer thickness versus the
colorimetric coordinates L*a*b* is shown in Figure 3.
The most evident feature of both metals is a reduction
in lightness as the thickness of the corrosion layer
increases. For the silver coupons, changes in the a*
axis are minimum, but a tendency toward yellowing is
evident by the increase in b*. The chromaticity changes
Figure 2. The MIPAC panel monitoring inside a showcase at the Faculty in the copper coupons are smaller, as the hues are
of Chemistry of Complutense University, together with a set of copper
and silver coupons on plastic racks (left side) slightly warmer, with a minimum increase of a* and a
moderate increase of b*.
four coupons of each metal were extracted after 6, 12,
and 18 months to measure the color change and quantify
the corrosion extent by gravimetry and galvanostatic
reduction, following the procedures established in ISO
11844-2 standard. For the electrochemical measurements,
a GAMRY Interface 1000 potentiostat was used, together
with a three-electrode cell with a Ag/AgCl reference
electrode, a graphite counter electrode, and deaerated
KCl 0.1 M as the electrolyte, with constant N2 bubbling.
Reduction was done in delimited areas of the coupons
using a vinyl tape mask. The four coupons at each time
point were subjected to gravimetric evaluation, while
reduction was performed on six areas of one coupon for
each showcase and exposure time.
Color was measured in the L*a*b* color space using a

Konica Minolta CM-700d spectrophotometer with an
∅ 6 mm mask, D65 as the illuminant, and an observer
angle of 10°. Six color measurements were performed,
in the lower, central, and upper parts of each side of
the coupon, to obtain an average value per coupon. Six
measurements were also made in each reduced area.
Results and discussion Figure 3. Correlation between corrosion thickness (measured using the
gravimetric method) and the L*a*b* coordinates for silver and copper
The metal coupons showed mild corrosion after expo-
sure in the museum cases, with differences observed As previously indicated, the distribution of corrosion
depending on the location. Visually, the silver coupons is visibly irregular, particularly for silver. In addition to
had a yellowish hue that was clearly darker at the edges the concentration of the corrosion at the borders of the
and in the lower area of the coupon. By contrast, the coupons, a comparison of the average values of L*a*b*
copper coupons underwent only slight changes, not for the lower, center, and upper areas of different sets of
perceivable at first sight, as they remain within their coupons (Table 2) showed that the lower areas are gener-
natural reddish color range. ally darker and yellower (lower L*, higher b*).
The fact that the gravimetric determination of the mass or Table 2. L*a*b* coordinates in the lower, central, and upper areas of
thickness of the corrosion products provides an average exposed silver coupons. The average values for each set of samples at
four different locations (S1–S4) after 6, 12, and 18 months of exposure
value for the whole surface detracts from the accuracy of are shown
the correlation between the color and concentration of the
L* (10°/D65)
corrosion products. Therefore, the extent of corrosion was
sample/
also assessed by galvanostatic reduction, using localized exposure time
bottom σ center σ up σ
areas of the silver coupons. Up to six circular areas of S1 / 6 m 87.01 1.27 87.08 1.17 87.48 1.22
several silver coupons were isolated with tape and galva- S2 / 6 m 91.01 0.88 91.21 0.86 91.26 0.91
nostatically reduced. The corrosion layer thickness was S3 / 6 m 91.37 0.87 92.04 0.53 91.81 0.54
calculated using Faraday’s law, following the ISO 11844-2 S4 / 6 m 88.31 2.46 88.47 2.65 86.25 3.53
standard (ISO 2020). Color was also measured in those S1 / 12 m 84.44 2.44 85.45 2.85 85.48 2.34
areas. The results and a comparison with the gravimetric S2 / 12 m 88.99 1.39 89.85 1.38 89.57 1.59
data are presented in Figure 4. The most evident difference S3 / 12 m 91.49 0.54 90.91 0.50 91.77 0.14
is that the electrochemical techniques indicate a more S4 / 12 m 92.58 1.18 92.71 1.07 92.42 1.74
pronounced color variation with the increasing thickness S1 / 18 m 76.19 5.20 78.81 4.08 78.07 4.33
of the corrosion layer. The reduction of a defined area,
S2 / 18 m 82.38 2.59 85.94 1.56 86.91 1.12
coincident with the area of measured color, increases the
S3 / 18 m 91.21 0.67 91.27 0.46 92.39 0.48
sensitivity of the method. In the gravimetric method,
S4 / 18 m 84.74 1.28 86.93 2.26 86.30 1.44
which provides average values for the whole coupon, the
a* (10°/D65)
higher concentration of corrosion products at the edges
sample/
is reflected in the weight gain but not in the color varia- bottom σ center σ up σ
exposure time
tion. These results show that a localized measurement of S1 / 6 m 0.31 0.37 0.25 0.32 0.14 0.32
color together with electrochemical reduction solves the S2 / 6 m −0.47 0.14 −0.41 0.15 −0.35 0.12
unreliability of color measurements in flat silver samples S3 / 6 m −1.38 0.29 −1.19 0.27 −1.23 0.16
reported by Ankersmit (2001). S4 / 6 m 0.53 1.19 1.09 0.98 2.11 2.09
S1 / 12 m 0.57 0.55 0.41 0.74 0.20 0.62
While the data thus far suggested that the variation of
S2 / 12 m −0.25 0.16 −0.20 0.15 −0.18 0.21
L*a*b* coordinates is more or less linear, the number
S3 / 12 m 0.57 0.11 0.41 0.07 0.20 0.13
of items in the data set at this stage of the project is too
S4 / 12 m −0.25 0.33 −0.20 0.07 −0.18 0.51
small to draw final conclusions.
S1 / 18 m 4.08 3.07 2.50 1.78 2.80 2.25
To test the applicability of the MIPAC system, the color S2 / 18 m 1.19 1.07 0.36 0.58 0.10 0.54
coordinates extracted from a photograph of a real metal S3 / 18 m 0.04 0.12 −0.10 0.04 0.22 0.52
coupon from the MIPAC panel (Table 1) obtained in a S4 / 18 m −0.48 0.43 −0.52 0.68 −0.66 0.31
previous work (Ramírez Barat et al. 2021) are included b* (10°/D65)
in Figure 4, labeled as 12A_Ag. The corrosion layer sample/
bottom σ center σ up σ
thickness of this coupon (part of another project) was exposure time
also calculated by gravimetry and galvanostatic reduction S1 / 6 m 14.41 2.86 14.28 2.56 14.38 2.28
of the whole surface. S2 / 6 m 9.47 1.22 8.94 1.02 8.41 0.61
S3 / 6 m 17.90 3.10 16.37 2.50 17.90 2.57
Although this is only one data point, and its value fell
S4 / 6 m 18.29 5.21 18.50 5.24 22.47 7.69
at the extreme of the calibration range for corrosion
S1 / 12 m 22.09 3.54 21.16 5.25 22.68 4.72
layer thickness, for which only a small number of data
S2 / 12 m 14.96 4.17 12.46 2.56 12.79 1.90
are available, it can be considered a reasonable result.
S3 / 12 m 10.89 0.48 10.69 0.42 11.76 1.21
With the use of gravimetric values, the lightness value
S4 / 12 m 15.06 3.37 13.13 3.26 14.38 4.99
of 12A_Ag was much lower than the general trend (even
S1 / 18 m 39.60 10.84 36.12 7.86 36.43 9.02
compared to the spectrophotometric value of 72.44): a*
S2 / 18 m 30.47 5.06 21.86 1.87 20.31 1.41
was slightly higher while b* fit well. With the galvanos-
S3 / 18 m 11.23 1.89 10.52 2.19 8.81 1.10
tatic results, fitting was satisfactory for the L* and a*
S4 / 18 m 27.66 4.25 23.97 5.91 22.68 3.76
coordinates. For the b* coordinate, however, the value
was expected to be yellower. As the calculated coordi- environments for up to 18 months, showed that for both
nate was close to the true measured one, the apparent metals the amount of corrosion correlated well with the
mismatch cannot be attributed to the calibration method. color changes. While the main change was in lightness,
A possible explanation is the evolution of colors in a good correlation with the chromaticity changes was
silver tarnishing, particularly along the yellow-blue axis. also observed.
Tarnish layers initially have a yellowish color, but they
The problem of the dispersion of the data due to uneven
can later acquire bluish tones and thick layers become
tarnishing can be largely solved by using localized
black. Consequently, the trend in the evolution of the b*
reductions and color measurements in small areas of
coordinate may differ depending on the thickness inter-
the coupons.
vals. This behavior was suggested in previous works in
which colorimetric measurements were used to quantify Further studies, at longer exposure times and in different
silver tarnishing, based on comparisons of reflectance museum environments, are necessary to understand
(in %) with an ideal reflecting diffuser (Ankersmit 2001, the correlations for thicker corrosion layers. This work
Ankersmit et al. 2013). is currently in progress, with more complete results
expected in the coming months.
The good correlation between corrosion and color change

as well as the good results obtained in the calibration of
photographs, as reported in previous work by the authors,
are very promising for the proposed MIPAC system.
Its full implementation will provide a flexible, scalable,
and low-cost solution to the collection of conservation
data in addition to encouraging society’s commitment
to its heritage.
Acknowledgments
This work was funded by the Comunidad de Madrid,
European Regional Development Fund and European
Social Fund, projects MIPAC-CM (Y2018/NMT-4913)
and TOP-HERITAGE (S2018/NMT4372), and Ministerio
de Ciencia, Innovación y Universidades, grant PRE2018-
086667. Development of the methodology was made
possible in part by a collaboration between the National
Museum of Science and Technology (MUNCYT), the
Faculty of Chemistry of the Complutense University of
Madrid, and the Museum of Science and Technology
of Catalonia (MNACTEC), which provided space for
the panels in their scientific-technical collections. The
authors also acknowledge the support of the CSIC
Figure 4. Correlation between corrosion thickness and the L*a*b* Interdisciplinary Thematic Platform “Open Heritage:
coordinates for silver. Comparison of gravimetric and galvanostatic
Research & Society (PTI-PAIS)”, Drs. Iván Díaz and
reduction results. The L*a*b* coordinates of sample 12A_Ag extracted
from an image of a MIPAC panel and then calibrated are also plotted vs. Santiago Fajardo, for their suggestions and support in
corrosion thickness, calculated by gravimetry (12A_Ag G) and following carrying out the galvanostatic reductions, and Once34
galvanostatic reduction (12A_Ag R) Ltd. for printing the color charts.
Conclusion Notes
The results presented in this work, on the tarnishing 1
onitorización por procesado de imagen y ciencia
M
of silver and copper coupons exposed to real museum ciudadana para la conservación de materiales del patri-
monio cultural (MIPAC-CM) (Monitoring by image Lafuente, D. 2017. La contaminación por COV en museos
processing and citizen science for the conservation of y exposiciones: Análisis y efectos sobre el patrimonio
cultural heritage materials) is a research project funded cultural metálico. Límites y conservación preventiva.
by Comunidad de Madrid and the European Union. PhD dissertation, Universidad Autónoma de Madrid,
www.mipac-cm.es/ Spain. http://hdl.handle.net/10486/681000
2
h
ttps://play.google.com/store/apps/details?id=com. Lafuente, D., E. Cano, I. Llorente, A. Crespo, J. Künne,
norrispalmer.mipac and A. Schieweck. 2013. The effects of organic pollutants
on metals in museums: Corrosion products, synergistic
References effects and the influence of climatic parameters. In Metal
Ankersmit, H.A. 2001. Tarnishing of silver: Evaluation by 2013: Proceedings of the Interim Meeting of the ICOM-CC
colour measurements. In Metal 2001: Proceedings of the Metals Working Group, Edinburgh, 16–20 September 2013,
Interim Meeting of the ICOM-CC Metals Working Group, eds. E. Hyslop, V. Gonzalez, L. Troalen, and L. Wilson,
Santiago, Chile, 2–6 April 2001, eds. I. MacLeod and C. 209–14. Edinburgh: Historic Scotland and International
Degrigny, 157–66. Perth: Western Australian Museum. Council of Museums–Committee for Conservation.
Ankersmit, H.A., G. Noble, L. Ridge, D. Stirling, N.H. Ramírez Barat, B., E. Cano, M.T. Molina, M.A. Barbe-
Tennent, and S. Watts. 2013.The protection of silver ro-Álvarez, J.A. Rodrigo, and J.M. Menéndez. 2021.
collections from tarnishing. Studies in Conservation Design and validation of tailored colour reference charts
45(suppl. 1): 7–13. https://doi.org/10.1179/sic.2000.45. for monitoring cultural heritage degradation. Heritage
Supplement-1.7 Science 9(1): art. 41. https://doi.org/10.1186/S40494-
021-00511-6
Barbero-Álvarez, M.A., J.M. Menéndez, and J.A. Rodrigo.
2020. An Adaptive Colour Calibration for Crowd- Tétreault, J., E. Cano, M. van Bommel, D. Scott, M.
sourced Images in Heritage Preservation Science. IEEE Dennis, M., M.-G. Barthés-Labrousse, L. Minel, and
Access, 8, 185093–185111. https://doi.org/10.1109/ L. Robbiola. 2003. Corrosion of copper and lead by
ACCESS.2020.3017390 formaldehyde, formic and acetic acid vapours. Studies
in Conservation 48(4): 237–50. https://doi.org/10.1179/
Barbero-Álvarez, M.A., J.A. Rodrigo, and J.M. Menéndez.
2021a. Self-designed colour chart and a multi-dimen- sic.2003.48.4.237
sional calibration approach for cultural heritage preven-
tive preservation. IEEE Access 9: 138371–138384. https:// Authors
doi.org/10.1109/ACCESS.2021.3119223 Blanca Ramírez Barat obtained a BA in fine arts
Barbero-Álvarez, M.A., J.M. Menéndez, J.A. Rodrigo, (conservation), after which she focused on heritage
B. Ramírez Barat, and E. Cano. 2021b. Assessment of science, graduating with a degree in chemistry (UCM
the robustness of a color monitoring chart calibration University) and obtaining a PhD in materials science
method for crowdsourcing-based preventive conser- and engineering (UC3M University). Since 2013 she has
vation. Applied Sciences 11(21): art. 10067. https://doi. worked at the CENIM-CSIC on the application of electro-
org/10.3390/APP112110067 chemistry to the conservation and diagnosis of metallic
Brigham, R., J. Grau-Bové, A. Rudnicka, M. Cassar, and heritage. She has participated in several national and
M. Strlič. 2018. Crowdsourcing as an analytical method: European research projects, as well as different initiatives
Metrology of smartphone measurements in heritage related to heritage science, including the TechnoHeritage
science. Angewandte Chemie – International Edition network, the National Conservation Research Plan, the
57(25): 7423–7. https://doi.org/10.1002/anie.201801743 CSIC Interdisciplinary Platform Open Heritage: Research
and Society, and the Spanish Node of the European
International Organization for Standardization (ISO).
Research Infrastructure in Heritage Science (ERIHS.es).
2020. ISO 11844-2:2020: Corrosion of metals and alloys
– Classification of low corrosivity of indoor atmospheres María Teresa Molina has a BA in conservation and
– Part 2: Determination of corrosion attack in indoor restoration of cultural heritage and a MSc in science and
atmospheres. Geneva: International Organization for technology of architectural heritage from the University
Standardization. of Granada. She has worked as an R+D+I technician
in the Department of Mineralogy and Petrology of the

University of Granada in topics such as the deterioration
of heritage materials by exposure to environmental and
atmospheric agents. Since 2019, she has been working on
her PhD thesis in materials science and engineering at the
CENIM-CSIC. Her work is focused on the conservation
of metals in scientific-technical heritage.
Emilio Cano is senior scientist at the CENIM-CSIC in

Madrid. He is head of the “Atmospheric Corrosion and
Cultural Heritage Conservation” (CAPAC) research
group, dedicated to studies of corrosion, the protec-
tion of metallic heritage, atmospheric corrosion, and
electrochemical techniques. He is the author of more
than 90 articles and 100 contributions to conferences.
He is a Fellow of the IIC and was assistant coordinator
of the ICOM-CC Metals Working Group (2012–2020).
He is a member of the coordination team of the CSIC
Interdisciplinary Platform “Open Heritage: Research and
Society” and National Coordinator of the Spanish Node
of the European Research Infrastructure for Heritage
Science (E-RIHS).
Reflectance Transformation Imaging Feature Maps for
Visual Documentation of Metal Objects
Amalia Siatou* Gaëtan Le Goïc Laura Brambilla

Haute Ecole Arc Conservation- Imaging and Artificial Vision Haute Ecole Arc Conservation-restauration
restauration (HE-Arc CR), HES-SO Laboratory (ImViA), EA 7535, (HE-Arc CR), HES-SO University of Applied
University of Applied Sciences and University Burgundy-Franche-Comté Sciences and Arts Western Switzerland
Arts Western Switzerland Dijon, France Neuchâtel, Switzerland
Neuchâtel, Switzerland [email protected] [email protected]
[email protected]
Alamin Mansouri Christian Degrigny
Marvin Nurit Imaging and Artificial Vision Haute Ecole Arc Conservation-restauration
Imaging and Artificial Vision Laboratory (ImViA), EA 7535, (HE-Arc CR), HES-SO University of Applied
Laboratory (ImViA), EA 7535, University Burgundy-Franche-Comté Sciences and Arts Western Switzerland
University Burgundy-Franche-Comté Dijon, France Neuchâtel, Switzerland
Dijon, France [email protected] [email protected]
Abstract
Visual documentation plays a key role in cultural heritage parameters were then visualized as feature maps. These maps
preservation, especially in condition assessment, recording, were compared to common RTI visualization methods based
and monitoring of objects. This research aims at detecting and on relighting and studying the surface normal. Examination
documenting surface changes on cultural heritage metal objects and interpretation of the results indicated that feature maps
by employing reflectance transformation imaging (RTI). The can provide an enhanced visualization of surface information
objective is to use technological advances in the application of RTI in comparison to commonly used methods. Geometric maps
for automated visual documentation of historical metal surfaces. mainly contain information about the surface topography,
The proposed methodology examines how new algorithms whereas statistical maps principally provide information related
developed to process raw RTI data can be applied to study surface to changes in the surface reflectance response and thus can
appearance attributes as a function of their angular reflectance. separate differences in the light reflection between a metal
Through visualization of geometric or statistical surface features, substrate and its corrosion products. The combination of the
global characterization of the appearance attributes and extracted information enhances the visualization of an object’s
isolation of information are possible. As an illustration of the condition and allows the creation of accurate cartographies.
methodology developed, zinc printing plates exhibiting localized
corrosion were examined in terms of condition assessment and Keywords
surface cleaning monitoring. Data were acquired by employing reflectance transformation imaging (RTI), condition
monochromatic acquisitions using a dome with fully calibrated assessment, monitoring cleaning, surface documentation,
and motorized functions. Selected features of the surface feature maps
Introduction
The need for conservation documentation is well defined Visual and realistic representation of cultural heritage
in terms of record keeping, monitoring, decision making, objects is mainly achieved by means of imaging tech-
and accessibility for managing and preserving tangible niques that are constantly gaining ground in the field
cultural heritage (Moore 2001, Letellier 2007). Within the (Stanco 2011, Payne 2013, Bentkowska-Kafel 2017). To
general aspects of conservation documentation, condi- this extent, this paper examines the use of reflectance
tion assessment corresponds to recording the physical transformation imaging (RTI), a multi-light image
state of an object, at a moment in time, in a planned and collection (MLIC) technique, as a visualization tool
methodological way (European Committee for Stand- for the condition assessment of historical metal objects
ardization 2019). It is common practice to accompany and its possible application in mapping conservation
written condition reports with visual documentation. treatments.
REFLECTANCE TRANSFORMATION IMAGING FEATURE MAPS FOR VISUAL DOCUMENTATION OF METAL OBJECTS 161
State of the art RTI cannot replace photogrammetric techniques, it can

provide invaluable information for the interpretation of
RTI has been introduced as an easy-to-use, nonin-
failure analysis.
vasive, portable, image-based technique that creates
photorealistic renderings of surface textures. The basis
Research focus
of RTI lies in the work of Malzbender et al. (2001), who
created an image-based methodology using polynomial The current project proposes a new methodology based
texture mapping (PTM) in order to increase photoreal- on the evaluation of surface features visualized through
istic rendering of textured surfaces and overcome the feature maps, with particular interest in documenting
restrictions and complexities in the application of bump surface information in an accurate and automated way.
mapping. Since then, RTI has become of interest in The concept assumes that metal corrosion causes changes
cultural heritage-related applications, mostly due to the in physical properties that also manifest as changes in
free software made available by the nonprofit corporation appearance, such as topographic alteration (material loss
Cultural Heritage Imaging (2021). or gain), texture differentiation (i.e., change in porosity,
crystal formation on the surface, localized corrosion
In the field of cultural heritage, RTI is being used to
forms), color alteration (one of the most common char-
reveal the artistic and technological characteristics of
acteristics of metal corrosion), and loss of metal gloss.
objects or artworks by enhancing epigraphs and reliefs on
Furthermore, documentation on cleaning treatments
different materials (Mudge 2008, Mudge 2010, Historic
was also examined, since it involved direct changes in
England 2018, Mytum 2018). Gradually, the possibilities
surface appearance.
of RTI, beyond its use as an object representation tool,
indicate its usefulness as a scientific tool. Seen from
Methodology
different perspectives, various scientific teams have
developed methodologies or pipelines for documenting The experiment took place at the Imaging and Artificial
not only object surfaces but also conservation treatments Vision Laboratory (ImViA) at the University of Burgundy
(Kotula 2013, Manrique Tamayo 2013, Ono et al. 2017, Franche-Comte in Dijon, France. It involved the use of a
Boute 2018, Rosseti 2019) or for revealing underlying dome-based system that provided controllable functions,
information with the integrated use of spectral imaging ensuring reproducibility of acquisitions and giving access
(Giachetti 2017, Ono 2017) and to map defects with clear to different forms of data processing, i.e., surface texture,
topographic changes (Manfredi 2014, Corregidor 2020). generating maps or extracting 3D surface parameters,
and quantifying results.
Manfredi et al. (2013; 2014) were the first to introduce a
methodology for measuring changes by computing the
statistical limits of the surface and comparing the surface
normal before and after damage, therefore creating The system consisted of a custom-made dome equipped
damage maps. Corregidor et al. (2020) introduced an with a single light source (cold white LED, 5000 K) that
image-processing methodology that combines RTI was adjusted to a motorized system with the ability
specular enhancement and edge detection algorithms to to move in a circular axis and rotate at 360° and thus
isolate and document topographic defects (i.e., scratches) covering a hemispherical space denominated in φ (eleva-
on a set of similar coins. tion) and θ (azimuth) angles (Figure 1a). An industrial,
monochromatic camera with a CMOS sensor (Sony
With regard to the literature on cultural heritage metal
IMX304, resolution 4112(H) × 3008 (V)) was placed at
objects, little study has been done on documenting the
the top of the dome and a motorized objective lens was
condition of metal surfaces (Corregidor 2020) or moni-
attached to the camera. A user interface allowed the
toring conservation treatments (Kotula 2013, Rosseti
control and setting of experimental parameters.
2019, Min 2021). In most cases, the ability of RTI as a tool
for enhancing details is validated. However, the benefit of The object under investigation was placed in the center
using RTI as a visual inspection tool for failure analysis of the dome orthogonal to the camera, and a stack of
of industrial metal surfaces was analyzed by Coules et al. stereo-photometric images from different light positions
(2019). In their paper, they demonstrate that even though was acquired (Figure 1a). After capturing the images, the
appearance of the surfaces was modeled using custom- of mathematical functions related to the geometric or
made software. This software provided functions for statistical attributes of a surface and is visualized through
controlling the dome but was also equipped with visu- feature maps. These maps correspond to the per-pixel
alization tools and exploitable data formats. analysis of the surface in a global way (the entire surface
selected) and are able to characterize its topography/
texture and appearance.
Feature maps can be categorized as geometric and statis-

tical. In the former, information is derived from isolating
the directional data corresponding to normal map or
from exploiting the curvature properties (calculation
of the change rate of a curve from a straight line). In
the latter, the calculation is made by applying statistical
Figure 1. (a) Simplified schematic representation of RTI acquisitions using functions to the stack of multi-light data. This provides
a dome system with one light source and (b) an image of the system used
information on the distribution, variability, or asymmetry
Two visualization methods, providing different types of the reflectance response of the examined surface
of information at pixel level, were evaluated (Figure 2). (Figure 3).
The first involved fitting models, which are used to

provide continuous reconstruction of the angular light
and thus provide multi-light illumination of the exam-
ined surface. The methodological approach for fitting
models is well described in the literature (Malzbender
2001, Pitard 2017, Pintus 2018, Cultural Heritage Imaging
2021). Here, fitting models are examined as part of
the common approach for data visualization through Figure 3. Categorization of geometric and statistical features
image relighting and extraction of the surface normal.

For this project, a fitting model called discrete modal Application in a case study
decomposition (DMD) was used, as it has shown more Two aspects of conservation documentation were exam-
accurate visual reconstruction on specular surfaces than ined: condition assessments and documentation of
polynomial texture mapping (PTM) and hemispherical cleaning treatments. The condition assessment aims
harmonics (HSH) (Thomas 2019). at recording the visible changes on metal surfaces as a
result of corrosion phenomena. The documentation of
cleaning treatments aims at recording the changes to
the surface as a result of the treatment. The commonly
used approach of relighting and studying the surface
normal is compared to the proposed methodology of
using geometric and statistical feature maps.
The case of zinc printing plates (Figure 4) was selected.

These objects encompassed certain characteristic features
in terms of manufacturing and decoration, including
Figure 2. Schematic representation of the visualization of RTI-derived
decorative engraving and different textures. In particular
information
they consisted of a zinc plate (identified by X-ray fluo-
The second approach was based on feature extraction rescence, analyses not presented in this paper) with a
from the examined surface in the form of feature maps. mat surface finish and the characteristic grayish color of
The features of an image are distinct or repeatable char- zinc. They were engraved and the lower reliefs presented
acteristics/attributes that can be measured. For RTI, data a rougher texture covered by a dark layer, most probably
feature extraction is possible through the application related to residues of ink from the printing process.
In regard to corrosion, they presented clearly visible

signs of localized corrosion in the form of white spots
scattered over the surface (Figure 4) that corresponded
to crystalline agglomerates of zinc carbonates (char-
acterized by Fourier transform infrared spectroscopy,
analyses not presented in this paper). These spots
created localized changes in the surface appearance,
observed as a white color with a rough texture and
crystalline topography that differed greatly from the Figure 5. Condition assessment by common forms of visualization:
metal substrate. (a) image, (b) surface normal, (c) relighting
or perpendicular (black) to the viewpoint (positioning

of the camera) and therefore simulated the height differ-
ences on the surface. As regards local observations, the
corrosion products were accurately visualized in terms
of topography. The particular corrosion products did not
present any specific directionality and extended across
the top of the metal surface.
The statistical maps (Figure 6b) provided interesting

Figure 4. Photograph of the object with (a) the characteristic white corrosion information. The results were similar for both
corrosion spots and (b, c) different magnifications of the examined area magnifications, but the higher magnification presented
more details on the extent of the corrosion products
For the condition assessment, surface corrosion charac-
terization was studied at two different magnifications: one on the metal surfaces. The characteristic white spots
corresponding to the general appearance of the surface were clearly recorded, since they presented completely
(global) and the other concentrated on one corrosion spot different reflectance responses from the metal substrate.
(local). For the cleaning treatment, a similar surface was The corrosion reflectance response was characterized by
studied before and after mechanical cleaning. negative skewness and a low coefficient of variation (CoF)
that could be explained by the fact that these spots, in
Results relation to the surrounding metal surface, were white and
mat (negative skewness) but also presented lower vari-
Condition assessment
ation in reflectivity (low CoF) for the different lighting
The selected object has different appearance properties positions due to their non-specular texture. In parallel,
related to both manufacturing and decorative details, as the manufacturing/decorative characteristics that create
well as the presence of white corrosion spots (Figure 5a). different reflection responses were well depicted in the
Relighting (Figure 5c) provided information on the statistical maps. For example, the dark areas of the lower
surface topography, manufacturing, and condition, which reliefs showed higher skewness and CoF due to the dark
were better depicted depending on the lighting angle. In color of the surface and the varying specular reflectivity
parallel, the surface normal (Figure 5b) was difficult to for the different light positions, respectively.
interpret as it incorporated all of the topographic infor-
mation in a single visualization.
In the general observation of the surface, the geometric

maps (Figure 6a) were mostly able to characterize tech-
nological features related to the manufacturing and
decorative characteristics of the selected printing plate,
while defining the directionality of these features. For Figure 6. Condition assessment using geometric and statistical feature
example, the “Dz” map depicted areas parallel (white) maps under different magnifications
Overall, the feature maps were able to separate infor-

mation and depict the characteristics of the surface in
terms of condition documentation. In certain cases,
especially for statistical feature maps, the visualization
can be enhanced by selecting the appropriate color
scale. In Figure 7, the difference in the visualization
between the grayscale and color scale is demonstrated.
For the color scale, a three-color transition from blue to
yellow to red was selected, illustrating lower to higher
values, respectively. In the particular case, the color
Figure 8. Documenting cleaning treatments: RTI relightable images at
scale on the mean map depicts the white corrosion spots the same lighting angles before (top) and after (bottom) cleaning
varying from yellow to dark red, which corresponds to
lighter-colored areas. The green areas correspond to Both the representative geometric and statistical maps
the metal where the metallic grayish color was retained showed the effect of cleaning in regard to surface changes
and the blue areas to the lower level of the darker metal in topography and visual appearance, respectively
background. (Figure 9). The statistical maps, in particular, were better
able to document the reinstatement of the metal surface
and the extent of the remaining corrosion products. In
both mean and standard deviation maps, the areas with
higher values, representing the areas with corrosion prod-
ucts, showed values closer to that of the metal substrate
after cleaning (moving from red-yellow to light blue).
Furthermore, the remaining corrosion products had
lower values than prior to cleaning (shifting from red
to yellow), corresponding to partial surface cleaning in
the relevant areas.
Figure 7. Enhancement of the visualization by selecting the appropriate

color map. Example of the mean feature map
Monitoring cleaning
A similar surface, from a different printing plate, was
cleaned mechanically with the aim to remove the most
Figure 9. Documenting cleaning treatments: geometric (left) and
friable corrosion products. The effect of the cleaning statistical (right) feature maps before (top) and after (bottom) cleaning
process was documented with the proposed visualiza-
tion methods. Acquisitions before and after cleaning Discussion and conclusion
were constant (same number of acquisitions and light
Even though RTI has been used in the field of cultural
positions, same exposure time, same object position) to
heritage for almost 20 years, there is still room for devel-
ensure reproducibility of the results.
opment in many different aspects. Current developments
Surface normal and image relighting were able to docu- that can lead to the use of RTI as a documentation tool
ment the surface change (Figure 8). The surface normal are still at an experimental stage and require expertise to
recorded the surface topography, but, as stated above, apply and interpret the results. The proposed method-
the information provided was mixed in one single visu- ology, although still under development, is anticipated to
alization. With respect to relighting, the surface change provide a user-friendly interface for visualization-based
was visible, but no important information was extracted conservation documentation. Nevertheless, expertise will
from the different lighting angles. be required to interpret the results.
From the case study examined, it is evident that RTI References

feature maps can provide more enhanced visualization
Bentkowska-Kafel, A. and L. MacDonald, eds.
and isolation of information than common RTI relighting 2017. Digital techniques for documenting and preserving
and the observation of surface normal. These maps can cultural heritage. Yorkshire: Arc Humanities Press.
document technological features and reveal information https://doi.org/10.2307/j.ctt1xp3w16
on corrosion phenomena that provoke alterations in
Boute, R., M. Hupkes, N. Kollaard, S. Wouda, K. Seymour,
surface appearance and therefore automate surface condi-
and L. ten Wolde. 2018. Revisiting reflectance transfor-
tion mapping. When applied as a tool for documenting
mation imaging (RTI): A tool for monitoring and evalu-
conservation treatments, they can accurately record the
ating conservation treatments. In IOP Conference Series:
surface change and help monitor conservation treatments
Materials Science and Engineering 364(1): art. 012060.
by creating accurate cartographies.
doi: 10.1088/1757-899X/364/1/012060.
As a general observation, geometric maps are linked to Corregidor, V., R. Dias, N. Catarino, C. Cruz, L.C. Alves,
information related to the directionality of the surface and J. Cruz. 2020. Arduino-controlled Reflectance
and are thus able to easily isolate technological and Transformation Imaging to the study of cultural heritage
decorative characteristics. When the appropriate scale objects. SN Applied Science 2(9): art. 1586. https://doi.
is selected, corrosion products can be studied in terms org/10.1007/s42452-020-03343-4
of topography and the alterations they can cause to the
Coules, H.E., P.J. Orrock, and C.E. Seow, C. 2019. Reflec-
metal substrate.
tance Transformation Imaging as a tool for engineering
In parallel, statistical maps are directly related to the failure analysis. Engineering Failure Analysis 105: 1006–
surface reflectance response, allowing differences in 17. https://doi.org/10.1016/j.engfailanal.2019.07.037
the appearance attributes to be distinguished between Cultural Heritage Imaging (CHI). Reflectance trans-
corrosion products and metal surfaces. Further exam- formation imaging (RTI). What is it? http://cultural-
ination of the representation of the statistical maps heritageimaging.org/Technologies/RTI/ (accessed 20
can lead to correlations in the behavior of the surface December 2021).
and thus help to improve understanding of corrosion
European Committee for Standardization. 2019. EN
formations or document the extent of a conservation
15898:2019: Conservation of cultural heritage – Main
treatment.
general terms and definitions.
The proposed methodology shows great potential for the Giachetti, A., I.M. Ciortan, C. Daffara, R. Pintus, and E.
application of RTI data processing methodologies as a Gobbetti. 2017. Multispectral RTI analysis of heteroge-
tool for accurate and automated condition assessment neous artworks. In Eurographics Workshop on Graphics
monitoring of conservation treatments in historical and Cultural Heritage, Proceedings, eds. T. Schreck, T.
metal objects. Weyrich, R. Sablatnig, and B. Stular, 19–28. Geneve: The
Future work will involve quantifying the results and Eurographics Association. http://dx.doi.org/10.2312/
isolating selected features based on second-order statis- gch.20171288
tical analysis and data segmentation, respectively. These Historic England. 2018. Multi-light imaging: Highlight-re-
will provide access to more accurate surface documenta- flectance transformation imaging (H-RTI) for cultural
tion and automation of monitoring processes. heritage. Swindon: Historic England.
Kotoula, E. and M. Kyranoudi. 2013. Study of ancient
Acknowledgments Greek and Roman coins using Reflectance Transforma-
This research was conducted within the CHANGE tion Imaging. E-Conservation 25: 74–88.
(Cultural Heritage Analysis for New Generations) Letellier, R., W. Schmid, and F. LeBlanc. 2007. Recording,
project, which has received funding from the European documentation, and information management for the
Union’s Horizon 2020 research and innovation program conservation of heritage places: Guiding principles. Los
under the Marie Skłodowska-Curie grant agreement Angeles: Getty Conservation Institute. http://hdl.handle.
No. 813789. net/10020/gci_pubs/recordim
Malzbender, T., D. Gelb, and H. Wolters. 2001. Poly- photography-based digital imaging techniques for
nomial texture maps. In SIGGRAPH ’01: Proceedings museums. In VAST: Proceedings of the 11th International
of the 28th Annual Conference on Computer Graphics Symposium on Virtual Reality, Archaeology and Intelli-
and Interactive Techniques, August 2021, 519–28. New gent Cultural Heritage – Short and Project Papers, eds.
York: Association for Computing Machinery. https:// A. Artusi, M. Joly-Parvex, G. Lucet, A. Ribes, and D.
doi.org/10.1145/383259.383320 Pitzalis, 111–37. Geneve: The Eurographics Association.
Manfredi, M., G. Williamson, D. Kronkright, E. Doehne, http://dx.doi.org/10.2312/PE/VAST/VAST10S/111-137
M. Jacobs, E. Marengo, and G. Bearman. 2013. Measuring Mytum, H. and J.R. Peterson. 2018. The application of
changes in cultural heritage objects with Reflectance reflectance transformation imaging (RTI) in historical
Transformation Imaging. In Proceedings of the 2013 archaeology. Historical Archaeology 52: 489–503. https://
Digital Heritage International Congress (DigitalHeritage), doi.org/10.1007/s41636-018-0107-x
vol. 2, 189–92. Piscataway, NJ: Institute of Electrical Ono, T., S. Matsuda, and Y. Mizuochi. 2017. Development
and Electronics Engineers. doi: 10.1109/DigitalHer- of a multispectral RTI system to evaluate varnish cleaning.
itage.2013.6743730. In Linking Past and Future. ICOM-CC 18th Triennial
Manfredi, M., Bearman, G., Williamson, D. Kronkright, Conference Preprints, Copenhagen, 4–8 September 2017,
E. Doehne, M. Jacobs, and E. Marengo. 2014. A new ed. J. Bridgland, art. 0205. Paris: International Council
quantitative method for the non-invasive documentation of Museums. Available at https://www.icom-cc-publica-
of morphological damage in paintings using RTI surface tions-online.org/
normals. Sensors (Basel, Switzerland) 14(7): 12271–84. Payne, E.M. 2013. Imaging techniques in conservation.
https://doi.org/10.3390/s140712271 Journal of Conservation and Museum Studies 10(2):
Manrique Tamayo, S.N., J. Valcárcel Andrés, and J. Osca 17–29. http://doi.org/10.5334/jcms.1021201
Pons. 2013. Applications of reflectance transformation Pintus, R., T. Dulecha, A. Jaspe, A. Giachetti, I. Ciortan,
imaging for documentation and surface analysis in and E. Gobbetti. 2018. Objective and subjective evalua-
conservation. International Journal of Conservation tion of virtual relighting from reflectance transformation
Science 4: 535–48. imaging data. In Eurographics Workshop on Graphics
Min, J., S. Jeong, K. Park, Y. Choi, D. Lee, J. Ahn, D. Har, and Cultural Heritage (2018), eds. R. Sablatnig and M.
and S. Ahn. 2021. Reflectance transformation imaging Wimmer, 87–96. Geneve: The Eurographics Association.
for documenting changes through treatment of Joseon https://doi.org/10.2312/gch.20181344
dynasty coins. Heritage Science 9: art. 105. https://doi. Pitard, G., G. Le Goïc, A. Mansouri, H. Favrelière, S.-F.
org/10.1186/s40494-021-00584-3 Désage, S. Samper, and M. Pillet. 2017. Discrete modal
Moore, M. 2001. Conservation documentation and the decomposition: A new approach for the reflectance
implications of digitisation. Journal of Conservation modeling and rendering of real surfaces. Machine Vision
and Museum Studies 7: 6–10. http://doi.org/10.5334/ and Applications 28(1): 607–21. https://doi.org/10.1007/
jcms.7012 s00138-017-0856-0
Mudge, M., T. Malzbender, A. Chalmers, R. Scopigno, J. Rossetti, L., J.G. Aubert, N. Stephant, N. Mélard, and E.
Davis, O. Wang, P. Gunawardane, M. Ashley, M. Doerr, Guilminot. 2019. Study and conservation of lead curse
A. Proença, and J. Barbosa. 2008. Image-based empirical tablets. In Metal 2019: Proceedings of the Interim Meeting
information acquisition, scientific reliability, and long- of the ICOM-CC Metals Working Group, Neuchâtel, 2–6
term digital preservation for the natural sciences and September 2019, eds. C. Chemello, L. Brambilla, and E.
cultural heritage. In Eurographics 2008 – Tutorials, eds. Joseph, 242–9. Paris: International Council of Museums–
M. Roussou and J. Leigh. Geneve: The Eurographics Committee for Conservation (ICOM-CC) and Haute
Association. http://dx.doi.org/10.2312/egt.20081050 Ecole Arc Conservation-restauration (HE-Arc CR).
Mudge, M., C. Schroer, G. Earl, K. Martinez, H. Pagi, Stanco, F., S. Battiato, and G. Gallo. 2011. Digital imaging
C. Toler-Franklin, S. Rusinkiewicz, G. Palma, M. for cultural heritage preservation: Analysis, restoration
Wachowiak, M. Ashley, N. Matthews, T. Noble, and M. and reconstruction of ancient artworks. Boca Raton:
Dellepiane. 2010. Principles and practices of robust, CRC Press.
Zendagui, A., J.B. Thomas, G. Le Goic, Y. Castro, M.

Nurit, A. Mansouri, and M. Pedersen. 2019. Quality
assessment of reconstruction and relighting from RTI
images: Application to manufactured surfaces. In SITIS
2019: Proceedings of the 15th International Conference
on Signal-Image Technology & Internet-Based Systems,
Sorrento, Italy, 26–29 November 2019, eds. K. Yetongnon,
A. Dipanda, G.S. di Baja, L. Gallo, and R. Chbeir, 746–53.
Piscataway, NJ: IEEE. doi: 10.1109/SITIS.2019.00121.
Authors
Amalia Siatou is a professional conservator-restorer
with a master’s in chemistry and material science (2013).
Her studies and professional career have been devoted
to conservation-restoration practices, scientific analysis,
and evaluation of different cultural heritage materials.
Since 2019, she has been a research assistant at HE-Arc
CR and a PhD candidate at the ImViA Laboratory at the
University of Bourgogne.
Marvin Nurit is a PhD candidate in computer science

at the ImViA Laboratory working on the NAPS project
(ANR-17-CE10-0005).
Gaëtan Le Goïc is an assistant professor at the University

Burgundy-Franche-Comté and a member of the ImViA
Laboratory.
Alamin Mansouri has been a professor in computer

science and imaging at the University Burgundy-Franche-
Comté since 2006. He received his PhD in computer
vision in 2005 from the University of Burgundy. His
current research is focused on computational modelling
of visual perception through imaging.
Laura Brambilla has a PhD in chemical sciences from

the Università degli Studi di Milano, Italy. Since April
2013, she has been part of the R&D team at HE-Arc
CR, first as a scientific collaborator and, from 2017, as
a professor. She is currently project leader of various
different projects.
Christian Degrigny has a PhD in analytical chemistry

from the University of Paris IV on the stabilization of
submerged aluminum remains. He has since specialized
in the application of electrochemistry in conservation
science and the use of imaging techniques by conser-
vation professionals. He is currently a professor at
HE-Arc CR.
Revealing Degradation Patterns: Imaging Techniques
for the Study of Metal Soap Formation on Painted Metal
Objects
Silvia Russo* Jean-Baptiste Thomas
Haute Ecole Arc Conservation-restauration (HE-Arc CR), HES-SO University of Norwegian University of Science and Technology (NTNU)
Applied Sciences and Arts Western Switzerland Gjøvik, Norway
Laboratory of Technologies for Heritage Materials (LATHEMA), University of [email protected]
Neuchâtel
Neuchâtel, Switzerland Edith Joseph
[email protected] Haute Ecole Arc Conservation-restauration (HE-Arc CR), HES-SO University of
Applied Sciences and Arts Western Switzerland
Laura Brambilla Laboratory of Technologies for Heritage Materials (LATHEMA), University of
Haute Ecole Arc Conservation-restauration (HE-Arc CR), HES-SO University of Neuchâtel
Applied Sciences and Arts Western Switzerland Neuchâtel, Switzerland
Neuchâtel, Switzerland [email protected]
Abstract
In attempting to document the degradation processes month in order to instigate the formation of metal soaps. Their
occurring on cultural heritage objects, imaging-based reaction was then monitored by means of µ-FTIR. The chemical
analytical techniques present many advantages, as they maps showed an increasing trend over time, elucidating some
provide spatial and spectral information and allow the aspects and differences in the mechanism of formation of
simultaneous investigation of the chemical and morphological the organic salts for the two metal substrates. Additionally,
characteristics of a sample. This study presents a protocol the samples were analysed using two hyperspectral cameras,
based on chemical imaging – Fourier transform infrared operating in the visible-near infrared and short-wave infrared
microspectroscopy (μ-FTIR) and hyperspectral imaging (HSI) spectral range. The appropriateness of the two cameras in the
– aimed at monitoring the formation of metal soaps on investigation of metal soaps, and the effect of the thickness of
model metal coupons. Oil-painted metal supports are in fact the coating on the data obtained, is discussed here.
not immune to degradation due to metal soaps formation,
a phenomenon that affects all oil-painted surfaces from the Keywords
initial curing of the paint film. Copper and zinc sheets were metal soaps, metal carboxylates, imaging, painted metals,
coated with cold-pressed linseed oil and artificially aged for one metal conservation
Introduction
Metals, particularly copper, lead, iron and zinc, have Metal soaps are a class of water-insoluble organic salts
historically been common supports for painted artworks (carboxylates) that form ubiquitously on artworks
due to their unique technical and aesthetic properties whenever a reactive metal ion is in close contact with
in comparison to canvas and wood (Pavlopoulou and fatty acids, which typically come from an oil binder
Watkinson 2006, Terenzi et al. 2006). Some exam- (Hermans et al. 2016). Many scientific studies have
ples include oil-painted metal artworks popularised been conducted in the past couple of decades to inves-
during the 16th and 17th centuries by old masters like tigate the technical aspects of the formation of such
Rembrandt and Rubens, and later in the 20th century by compounds, from the initial chemical reaction (Baij et
contemporary artists such as Frida Kahlo, Frank Stella al. 2018; Casadio et al. 2019, 47; Garrappa et al. 2020)
and Alexander Calder. It is worth mentioning that the to the aggregation and migration mechanism of the
concept of oil-painted metals is not limited to polychrome crystalline species towards the surface of the painting
metal objects but also encompasses metalwork artefacts (Herman 2016) and the subsequent implications for
covered with waxes, lacquers and oils as protective or the preservation of the object (Gabrieli et al. 2017,
finishing coatings (Hughes 1993). Banti et al. 2018).
REVEALING DEGRADATION PATTERNS: IMAGING TECHNIQUES FOR THE STUDY OF METAL SOAP FORMATION ON PAINTED METAL OBJECTS 169
Even though it would be theoretically possible to understanding of the degradation pathways as the reac-
transfer the knowledge derived from such studies to tions progress. Specifically, artworks can be sequentially
any oil-painted object, different substrates present unique monitored by obtaining spectra of the same region of
physicochemical characteristics that must be taken into interest to generate a time series of chemical images.
consideration (Campos et al. 2016). When dealing with
Some considerations on the use of hyperspectral imaging
the formation of metal soaps within a traditional pictorial
(HSI) devices for the study of oil-painted metals are also
film, the analytical focus is centred on the interaction
presented. HSI techniques are a set of non-invasive,
between the metal ions of the mineral pigments and
non-destructive techniques based on the acquisition
the binder in which they are embedded (Casadio et al.
of separate images at narrow intervals of wavelengths
2019). However, the presence of a metal support acts as
(< 10 nm), called bands, that can then be analysed as
a competing source of metal ions, in which case another
a function of the chemical composition of an object
type of interaction must be identified between the paint
(Amigo 2020). Although HSI systems are not always
layer as a whole and the metal substrate. At a microscopic
readily available, they have proven successful in the
level, such organic-inorganic reactions occur between the
field of cultural heritage to identify pigments or binders,
fatty acids in the oil binder and the metal species present
or to monitor conservation treatments. Specifically,
in the support, and they are influenced by both internal
pigments have a strong characteristic signal in the
and external factors that affect the interface between the
visible and near-infrared range (VNIR, 400–1050 nm),
metal substrate and the oil paint. This adds a layer of
whereas binders typically have a strong spectral signa-
complexity to the study of the oil-metal system.
ture in the short-wave infrared spectral range (SWIR,
Applying the most common analytical techniques used 1050–2500 nm) that makes their identification possible
in the study of metal soaps found in paintings on canvas (Cucci et al. 2019, Amato et al. 2021).
would imply sampling and embedding in a suitable resin
The search for diagnostic signals of the presence of metal
as a cross-section (Kaszowska et al. 2013; Keune and
soaps in reflectance spectra acquired in the VNIR and
Boon 2013; Hermans et al. 2018; Casadio et al. 2019, 195;
SWIR spectral regions is described here. Such indicators
Ortiz Miranda et al. 2020). In the case of painted metals,
could manifest themselves in changes in the shape or
the paint alone would not give exhaustive information
intensity of the signal that could imply a chemical vari-
to shed light on the system’s complexity, as the reaction
ation due to the formation of metal soaps.
of interest occurs at the interface between the metal
and the paint layer (Pavlopoulou and Watkinson 2006).
However, sampling the metal support together with the
paint raises both technical and ethical concerns, as it can Sample preparation
provoke irreversible damage.
Model samples were prepared for artificial ageing in trip-
The approach for a suitable elucidation of the degra- licates as 60 × 60 × 1 mm hair-brushed copper and zinc
dation processes occurring on such objects must then sheets. The hair-brushed finishing – an industrial process
involve the use of non-destructive techniques (NDT) which consists in utilising an abrasive material to create
for the investigation of the interface between the metal unidirectional lines giving a satin finish – was chosen
and the paint. A combination of well-chosen NDT can to mimic the artists’ common practice of scratching the
provide the in-depth information necessary to carry out surface of the metal prior to painting to increase the grip
conservation-restoration treatments (Cucci et al. 2019, of the paint on the metallic support. The samples were
Hennen Rodriguez et al. 2020). degreased with absolute ethanol (VWR Chemicals) and
coated with ~80 µL of cold-pressed linseed oil (Kremer
For this study, Fourier transform infrared microspec-
Pigmente). The linseed oil was applied on the metal
troscopy (µ-FTIR) was used to investigate the painted
surface with the aid of a rectangular painter’s spatula.
metal model samples. This analytical technique can be
used both to characterise metal soaps as well as generate
Coating thickness evaluation
chemical maps to understand the relative distribution of
the carboxylates (Gabrieli et al. 2017), allowing compar- The low affinity of oil and metal for one another did not
isons between different regions of an artwork and an allow homogeneous coatings to be obtained, despite the
standardisation of the samples’ preparation and coating geometry was standard at 45°/0° and the focus was set
procedure. The samples presented evident differences at 30 cm, with a field of view of 16°, yielding a pixel size
in coating thickness, a feature whose effect is intended of approximately 50 µm. The model samples and 99%
to be taken into consideration in this study. The inho- white diffuse reflectance standard (Spectralon) were
mogeneities of the oil coating were measured using the placed in-plane on a flat translational stage with mounted
coating thickness gauge Surfix by Phynix, resulting in broadband spectral coverage tungsten-halogen lamps at
ranges of thickness from 2 to 15 µm for the copper and 45° and the camera at 90°. Acquisitions were performed
from 3 to 25 µm for the zinc samples. in the dark with the white standard playing the role of a
wide spectral range Lambertian reflector with uniform
Accelerated ageing response, to correct for the reflectance of the light source.
The coated samples were cured overnight at 80 °C then For the extraction of the spectra, an area of analysis
artificially aged at 80 °C and 80% relative humidity in of 1000 × 1000 µm2 (20 × 20 px2) was chosen as a
a laboratory oven for 30 days. The triplets were kept good compromise between signal quality (S/N ratio)
in a sealed box at constant relative humidity using a and occurrence of representative regions of interest of
glycerol-water emulsion (50% w/w) (Forney and Brandi comparable size within the sample. The bundled OEM
1992) and monitored by means of µ-FTIR at spaced time software was used to implement dark current correction
intervals during 30 days. The samples were then left at and calibrate the sensor in the pre-processing stage of the
room temperature and relative humidity for an additional dataset. HyspexRadV2.0 software was used to convert
11 months. The levels of relative humidity and tempera- the raw data into radiance pixel values. The processing
ture were monitored using a PeakTech 5185 datalogger of the data to obtain interpretable and comparable rela-
equipped with an immersion probe. tive reflectance spectra was carried out using the ImageJ
open-source software. After re-slicing of the datacube at
Fourier transform infrared microspectroscopy the top avoiding interpolation, the data acquired both in
(µ-FTIR) the VNIR and SWIR ranges were corrected for illumi-
nance, i.e. the dataset was divided by the radiance values
Fourier transform infrared spectra were collected using
of the Spectralon white diffuse reflectance standard, after
a Thermo Scientific Nicolet iN10 MX infrared micro-
stacking, re-slicing and adjusting its size to that of the
scope equipped with a motorised stage. For each sample,
samples (Babini et al. 2021).
single scan spectra were acquired in reflectance mode
(650–4000 cm−1) with an aperture of 150 × 150 µm2,
step of 100 µm and at a resolution of 4 cm−1 on three
different areas (7500 × 1800 μm2). Spectra elaboration Fourier transform infrared microspectroscopy
was achieved with Omnic proprietary software and then
According to previous studies (Hermans et al. 2019
visualised using the ggplot2 package in the R free soft-
and 2021), several coordination geometries of the most
ware environment for statistical computing and graphics
common zinc carboxylates can be detected by FTIR
(www.r-project.org/). Baseline, normalisation and atmos-
spectroscopy in the range of 1525–1650 cm−1 (Table 1).
pheric corrections were performed to remove residual
signatures of atmospheric CO2 and water in the spectra. By observing this spectral region, the formation of metal
Chemical heatmaps were generated in the spectral range soaps at an early stage was monitored on the zinc and
of 1525–1650 cm−1 to cover the main vibrational bands copper samples during a period of 30 days (Figure 1). It
corresponding to metal soaps (νaCOO−). The results were can be seen that the evolution over time of the intensity
comparable in all three areas analysed, hence only one of the signal of the characteristic νaCOO− vibrational
of the regions of interest will be discussed in this work. bands for the copper and zinc carboxylates increased.
Here, it was not possible to distinguish between different

Hyperspectral imaging (HSI)
amorphous and crystalline forms of copper and zinc
Twelve-month-aged samples were analysed with push- metal soaps, as only the signal of the general ionomer
broom HySpex VNIR-1800 and SWIR-384 hyperspec- structure is visible (Hermans et al. 2016). However, from
tral cameras by Norsk Elektro Optikk. The acquisition the qualitative monitoring using heatmaps, important
Table 1. Major asymmetric stretching vibrational bands for the different metal carboxylates and coordination geometries in the range of
1525–1650 cm−1
Zinc Copper Vibrational mode References

Palmitate 1539–1536 s 1586 s
Stearate 1540–1539 s 1586 s
Oleate 1547 vs, 1527 vs 1588–1584 vs Izzo et al. 2021;
Crystalline νas[COO−]
Azelate 1556–1535 vs 1588 vs Hermans et al. 2019 and 2021
Type A 1524 s, 1545 s, 1590 w
2D coordination polymer
Type B 1536 s
Oxo 1590 s
Amorphous νas[COO−] Izzo et al. 2021
Chain 1544 s, 1565 w, 1630 s
vs = very strong signal; s = strong signal; w = weak signal
Figure 2. Copper and zinc coupons coated with linseed oil and respective
analysed regions of interest (a, b). The dashed line separates areas
showing higher (darker, > 10 µm) and lower (lighter, < 10 µm) coating
thickness. The time series of µ-FTIR chemical images are obtained in the
range of 1525–1650 cm−1 after drying over night at 80 °C (c, d), 10 days
(e, f) and 30 days (g, h). Intensity range goes from red (maximum) to
blue (minimum). The scale bar indicates 1000 µm
It must be noted that the detection of metal soaps is also

a function of the thickness of the oil, which could impede
their detection when values are higher. For this reason,
the metal soaps have a higher signal in thinner layers of
Figure 1. FTIR spectra for linseed oil-coated (a) copper and (b) zinc
samples in the time intervals considered. The grey area indicates the
the coating (Figure 2a, 2b) for both substrates.
vibrational bands related to the metal carboxylates
Additionally, due to the lack of distinct features on the
coated surfaces, the reproducibility of the selected areas of
analysis was not always achieved when monitoring over
features were observed (Figure 2): in both substrates
time. In particular, the process of placing the samples
investigated, the intensity of the signal of the metal soaps
on the stage manually was bound to introduce errors
increased in the time window considered, confirming in consecutive analyses. This can be observed in the
what was observed by inspection of the point spectra. resulting heatmaps.
Additionally, differences in the reactivity of the two
substrates could be highlighted, with zinc having higher Hyperspectral imaging
reactivity compared to copper. This can be seen in the From the analysis of the dataset, it was not possible to
occurrence of red areas in the heatmaps (Figure 2c, 2d), identify spectral features for the presence of metal soaps
and it is supported by previous knowledge of the reac- in the VNIR and SWIR spectral ranges, since these did
tivity of the two ions (Greenwood and Earnshaw 1997). not show characteristic signals in this range of wave-
lengths. Nevertheless, from the analysis of the spectra,

relevant observations could be obtained, highlighting
differences in the substrate and coating responses in the
two ranges of wavelengths considered. In particular in
the VNIR spectral range, it was possible to identify the
characteristic reflectance spectra of the copper and zinc
bare metals (Sárosi et al. 2010). As the thickness of the
coating increased, the relative reflectance values, shape
and position of the peaks of the metal support changed.
For the copper substrate, the presence of the coating
influenced the shape of the signal and its intensity, but
it affected its position only marginally (Figure 3).
Figure 4. VNIR HSI analysis of a zinc sample coated with a non-

homogeneous layer of linseed oil. Average relative reflectance spectra
obtained from three 20 × 20 px2 areas for low (1–5 µm), medium (5–10
µm) and high (10–25 µm) coating thicknesses, and respective standard
deviations, are shown
drying oil was observed (Amato et al. 2021), with relative

reflectance values increasing with the thickness of the oil
coating (Figure 5).
The SWIR camera could therefore be a complementary

tool to the VNIR camera and could be employed for a
direct estimation of the thickness of the oil.
Figure 3. VNIR HSI analysis of a copper sample coated with a non-

homogeneous layer of linseed oil. Average relative reflectance spectra
obtained from three 20 × 20 px2 areas for low (1–5 µm), medium (5–10
µm) and high (10–25 µm) coating thicknesses, and respective standard
deviations, are shown
For the zinc, where the colour of the metal is significantly

different from that of the coating, a shift in the main peak
was visible, corresponding to the variation in colour of the
coating from yellowish to reddish (bathochromic shift)
as the thickness increases (Figure 4). This suggests that
VNIR hyperspectral imaging could be used to determine Figure 5. SWIR HSI analysis of the copper sample coated with a non-
the thickness of the oil coating indirectly. homogeneous layer of linseed oil (refer to Figure 3). Average relative
reflectance spectra obtained from the three 20 × 20 px2 areas for low
In the SWIR range, where the oil binder has a strong (1–5 µm), medium (5–10 µm) and high (10–25 µm) coating thicknesses,
signature signal, the characteristic spectrum of the and respective standard deviations, are shown
Conclusion References
In this work, the question of whether the selected non-de- Amato, R.S., A. Burnstock, and A. Michelin. 2021. A
structive imaging analytical techniques – namely µ-FTIR preliminary study on the differentiation of linseed and
and HSI – are suitable for the study of metal soaps on poppy oil using principal component analysis methods
painted metal objects was examined. In particular, the applied to fibre optics reflectance spectroscopy and
focus of the work concerned the ability of such analyt- diffuse reflectance imaging spectroscopy. Sensors 20(24):
ical techniques to allow monitoring of the formation of art. 7125.
metal soaps. Micro-FTIR proved to be determinant in Amigo, J.M. 2020. Hyperspectral and multispectral
both the detection and the monitoring of metal soaps imaging: Setting the scene. In Data handling in science
on the copper and zinc model samples. In particular, it and technology, ed. J.M. Amigo, 3–16. Amsterdam:
was shown that metal soaps can be detected immediately Elsevier.
after the application of the oil coating. The intensity of
Babini, A., S. George, and J.Y. Hardeberg. 2021. Hyper-
the monitored νCOO− vibrational bands corresponding spectral imaging workflow for the acquisition and anal-
to the different metal soap forms increased in the time ysis of stained-glass panels. In Proceedings SPIE 11784,
frame considered. By utilising this technique, the differ- Optics for Arts, Architecture, and Archeology VIII, 21–26
ences in the reactivity of the two substrates could also June 2021, Online Only, Germany, 117841F (22 June
be highlighted. 2021).
The detection of metal soaps using HSI cameras in the Baij, L., J.J. Hermans, K. Keune, and P. Iedema. 2018.
VNIR and SWIR spectral ranges was not successful Time-dependent ATR-FTIR spectroscopic studies on
due to the absence of identifiable spectral features in fatty acid diffusion and the formation of metal soaps in
these regions. However, this technique raised ques- oil paint model systems. Angerwandte Chemie 57(25):
tions with regard to the substrates and the thickness 7351–4.
of the oil coating that could be extended to different
Banti, D., J. La Nasa, A. Lluveras Tenorio, F. Modugno,
applications in the field of cultural heritage, e.g. the
K.J. van den Berg, J. Lee, B. Ormsby, A. Burnstock, and
study of unpigmented oil-coated technical objects. For
I. Bonaduce. 2018. A molecular study of modern oil
the purpose of this paper, the focus was on qualitative paintings: Investigating the role of dicarboxylic acids in
monitoring, but this could readily be extended to a full the water sensitivity of modern oil paints. RSC Advances
statistical analysis. 8: 6001–12.
In future research, the mid-wave infrared (MWIR) Campos, M.F., L. Carlyle, and M.J. Melo. 2016. Beneath
range should be explored for the HSI analysis of metal the surface: Paint losses due to zinc soap formation
soaps, as MIR reflectance spectra are reported for these within ground layers from a late 19th century oil painting.
compounds. Poster presented at the conference Metal Soaps in Art,
Amsterdam, Rijksmuseum, Netherlands, 14–15 March
Acknowledgements 2016.
The authors would like to thank the team at the Norwe- Casadio, F., K. Keune, P. Noble, A. Van Loon, E. Hendriks,
gian University of Science and Technology–Norwegian S.A. Centeno, and G. Osmond, eds. 2019. Metal soaps in
Colour and Visual Computing Laboratory, particularly art. Conservation and research. Cham: Springer.
the fellow PhD students Agnese Babini, Jan Cutajar and Cucci, C., E.K. Webb, A. Casini, M. Ginanni, E. Prandi,
Federico Grillini, for offering to acquire the hyperspectral and L. Stefani. 2019. Short-wave infrared reflectance
images when free circulation between countries was not hyperspectral imaging for painting investigations: A
permitted. methodological study. Journal of the American Institute
This project has received funding from the Euro- for Conservation 58(1–2): 16–36.
pean Union’s Horizon 2020 research and innovation Forney, C.F. and D.G. Brandi. 1992. Control of humidity
programme under the Marie Skłodowska-Curie grant in small controlled-environment chambers using glycer-
agreement No. 813789. ol-water solutions. HortTechnology 2(1): 52–4.
Gabrieli, F., F. Rosi, A. Vichi, L. Cartechini, L. Pensabene tion of painting materials and their degradation products
Buemi, S.G. Kazarian, and C. Miliani. 2017. Revealing in paint cross sections. Vibrational Spectroscopy 65: 1–11.
the nature and distribution of metal carboxylates in Keune, K. and J.J. Boon. 2013. Analytical imaging studies
Jackson Pollock’s Alchemy (1947) by micro-attenuated of cross-sections of paintings affected by lead soap aggre-
total reflection FT-IR spectroscopic imaging. Analytical gate formation. Studies in Conservation 52(3): 161–76.
Chemistry 89(2): 1283–9.
Ortiz Miranda, A.S., D. Kronkright, and M. Walton.
Garrappa, S., E. Kočí, S. Švarcová, P. Bezdička, and D. 2020. The influence of commercial primed canvases in
Hradil. 2020. Initial stages of metal soaps formation in the manifestation of metal soaps protrusions in Georgia
model paints: The role of humidity. Microchemical Journal O’Keeffe’s oil paintings. Heritage Science 8: art. 107.
156: art. 104842.
Pavlopoulou, L.-C. and D. Watkinson. 2006. The degra-
Greenwood, N.N. and A. Earnshaw. 1997. Chemistry of
dation of oil painted copper surfaces. Studies in Conser-
the elements, 2nd ed. Oxford: Butterworth-Heinemann.
vation 51(suppl. 1): 55–65.
Hennen Rodriguez, S., C.R. Appoloni, P.H.O.V. Campos,
Sárosi, Z., W. Knapp, A. Kunz, and K. Wegener. 2010.
B. Gonçalves, E.A.M. Kajiya, R. Molari, M.A. Rizzutto,
Evaluation of reflectivity of metal parts by a ther-
and C. Winter. 2020. Non-destructive and portable anal-
mo-camera. In InfraMation 2010 Proceedings: Infrared
yses helping the study and conservation of a Saraceni
Camera Applications Conference, Las Vegas, NV, USA,
copper plate painting in the São Paulo Museum of Art.
8–12 November 2010, 475–86. Zurich: ETH-IWF.
Microchemical Journal 155: art. 104787.
Terenzi, M.G., F. Ferrucci, and M.L. Amadori. 2006.
Hermans, J., K. Keune, A. van Loon, and P. Iedema.
Dipinti su rame: storia, tecnica, fenomeni di degrado,
2016. The crystallization of metal soaps and fatty acids
diagnostica, indicazioni per la conservazione e il restauro,
in oil paint model systems. Physical Chemistry Chemical
2nd ed., coll. Collana i Talenti. Padua: Il Prato.
Physics 18: 10896–905.
Hermans, J., G. Osmond, A. van Loon, P. Iedema, R. Authors
Chapman, J. Drennan, and K. Keune. 2018. Electron
Silvia Russo has a BSc in chemistry from Sapienza
microscopy imaging of zinc soaps nucleation in oil paint.
University of Rome (Italy, 2015) and an MSc in science
Microscopy and Microanalysis 24(3): 318–22.
and technologies for the conservation and restoration
Hermans, J.J., L. Baij, M. Koenis, K. Keune, P.D. Iedema, of cultural heritage (European Master Programme in
and S. Woutersen. 2019. 2D-IR spectroscopy for oil paint ARCHMAT (2018). She developed her expertise in the
conservation: Elucidating the water-sensitive structure of study of metal objects and their degradation processes at
zinc carboxylate clusters in ionomers. Science Advances the Soprintendenza Archeologia, Belle Arti e Paesaggio
5(6): 21 June 2019. delle Marche (Italy, 2016), The British Museum (UK,
Hermans, J., L. Zuidgeest, P. Iedema, S. Woutersen, and 2017) and the University of L’Aquila (Italy, 2019). She
K. Keune. 2021. The kinetics of metal soap crystallization is currently an MSCA ITN-CHANGE Fellow (ESR11)
in oil polymers. Physical Chemistry Chemical Physics 23: working on non-invasive analytical protocols to study
22589–600. painted metal artefacts and detect early-stage degrada-
tion.
Hughes, R. 1993. Artificial patination. In Metal plating
and patination: Cultural, technical and historical develop- Laura Brambilla has a PhD in chemical sciences from
ments, eds. S. La Niece and P. Craddock, 1–17. Oxford: the Università degli studi di Milano, Italy. Since April
Butterworth-Heinemann. 2013, she has been part of the R&D team at HE-Arc CR,
Izzo, F.C., M. Kratter, A. Nevin, and E. Zendri. 2021. first as a scientific collaborator and then as a professor
A critical review on the analysis of metal soaps in oil from 2017. She is currently leader of various projects
paintings. ChemistryOpen 10: 904–21. and was one of the organisers of the ICOM-CC Metal
2019 conference.
Kaszowska, Z., K. Malek, M. Pańczyk, and A. Mikołajska.
2013. A joint application of ATR-FTIR and SEM imaging Jean-Baptiste Thomas received his BSc in applied
with high spatial resolution: Identification and distribu- physics (2004) and MSc in optics, image and vision
(2006) from the Université Jean Monnet in France. He

completed his PhD from the Université de Bourgogne
in 2009. Since 2010, he is been an associate professor at
the Université de Bourgogne. Between 2016 and 2021, he
was on sabbatical at NTNU as a researcher and then as
an associate professor. He has worked extensively on the
development of spectral imaging systems using spectral
filter array technology. Since 2016, he has worked on
understanding material appearance and measuring it by
using imaging systems.
Edith Joseph is an assistant professor at the University

of Neuchâtel and University of Applied Sciences Haute
Ecole Arc Conservation-Restauration in Neuchâtel, Swit-
zerland. She holds a PhD in environmental and heritage
chemistry from the University of Bologna, Italy. Her main
field of research is the application of green chemistry for
the preservation of artistic and archaeological objects. She
has published around 60 papers on analytical chemistry
and conservation science in international journals and
books.
Monitoring of Metal Sculptures and Their Environment in
Art Museums in Sweden
Stavroula Golfomitsou* Mariateresa Pullano Eva Nygårds Karin Hermerén

Department of Conservation, University Gothenburg Museum of Art Gothenburg Museum of Art Moderna Museet
of Gothenburg Gothenburg, Sweden Gothenburg, Sweden Stockholm, Sweden
Gothenburg, Sweden [email protected] [email protected] [email protected]
[email protected]
Malin Borin Michaela Florescu *Author for correspondence
Elyse Canosa Gothenburg Museum of Art Moderna Museet
Swedish National Heritage Board Gothenburg, Sweden Stockholm, Sweden
Visby, Sweden [email protected] [email protected]
[email protected]
Abstract
During recent discussions related to carbon emission, there has sampling devices and X-ray fluorescence. Elemental analysis
been little consideration of how such measures might impact showed the presence of sulfur and chlorine as well as selective
cultural heritage collections. This paper examines the efforts to zinc corrosion. Among the detected pollutants was formic acid.
monitor the conditions of selected metal sculptures from Moderna All tools were used noninvasively on site. If these same processes
Museet in Stockholm and the Gothenburg Museum of Art in are used intermittently but consistently over the coming years,
Gothenburg, Sweden. The selected sculptures, which included when combined with visual documentation they will shed light
significant artworks by Picasso, Fautrier, and others, differed in on how display environments change over time and how a
their surface characteristics and were displayed under different particular environment can impact a collection.
environmental conditions. Our aim was to link the environment
and microenvironment of these artworks to the condition of the Keywords
metal alloys, in order to manage and prevent potential corrosion. pollutants, corrosion, air quality monitoring, contemporary art,
The methods used included pollutant monitoring using passive Sweden, X-ray fluorescence (XRF)
Introduction
Artworks made of metal alloys have a prominent place actively work to reduce their carbon footprint and the
in many museum collections. In Sweden, those on open use of heating, ventilation, and air-conditioning (HVAC)
display include works from artists such as Henry Moore, systems, which can contribute to high greenhouse gas
Degas, Rodin, and Picasso. The corrosion of metal alloys emissions. Although there is ongoing research to develop
is often inevitable, even indoors. Metal sculptures are low-carbon emission systems, it may well take years
often displayed along with artworks made of organic before museums can afford to replace older systems. The
materials that are considered sensitive, and climatic need to reduce carbon emissions has been addressed by
parameters are adjusted to levels appropriate to these museum professionals, and conservation can contribute
objects. These choices may adversely impact metals; for by allowing greater fluctuations in humidity and temper-
example, higher humidity could increase the corrosion ature (see National Museum Directors’ Council 2014, IIC
rate of metallic sculptures. Although a number of studies and ICOM-CC 2014). However, there is little knowledge
have determined the corrosion rates of alloys exposed on the tolerance threshold of the materials on display,
outdoors, the long-term effects of indoor environments and these measures could lead to irreparable damage.
have been less well studied.
A determination of the impact of local environmental
Sweden has an ambitious strategic plan to achieve zero conditions on Swedish collections becomes paramount
net emissions of greenhouse gases by 2045 (Ministry to ensure the longevity of the artworks on display while
of the Infrastructure 2020). Implementation includes working toward developing environmentally sustainable
an action plan for cultural heritage. Museums should strategies.
MONITORING OF METAL SCULPTURES AND THEIR ENVIRONMENT IN ART MUSEUMS IN SWEDEN 177
This paper discusses issues regarding the metal alloy all located on the main entrance floor. The building is
artworks displayed in two Swedish museums using an divided into three controlled climatic zones—the exhibi-
analytical approach to characterize the alloys and their tion area, the temporary exhibition space, and the storage
surrounding environments. The focus is primarily on space—with the latter on the lowest floor of the building.
the development of a methodology aimed at preventing The air-conditioning system is remotely controlled. In
corrosion phenomena. As the project is currently in its the exhibition area, the relative humidity (RH) varies
early stages, it has concentrated on the choice of artworks yearly between 47 and 53% and the temperature between
and a characterization of their environment. Analyses of 18 and 22°C. The maximum 24 h variation in RH and
the alloy composition were prioritized over determina- temperature is ± 3% and ± 2°C. There is currently no
tions of the corrosion products and were carried out in monitoring or control of pollutants.
situ. The results, combined with visual documentation,
Metals are present in a variety of forms and representing
aided in: (i) identifying synergistic relationships between
all families of alloys, from traditional (ferrous and copper
metal substrate corrosion and the deterioration of surface
alloys, lead, silver) to modern (aluminum, stainless
coatings; (ii) linking the object’s history to current
conditions; (iii) correlating atmospheric environmental steel, and metallic coatings such as nickel and chrome).
conditions to degradation pathways; and (iv) comparing In some artworks, metals are the sole or main material
artworks by the same artists or of similar composition and in others they have been used in combination with
displayed in different environments. The project will other inorganic (other metals, concrete, stone, glass,
also provide museums with information about their ceramic) or organic (wood, textiles, synthetic materials,
ambient environment that will inform future climate paints, leather, straw and plant fibers, food) materials.
and air-quality decisions. These composite artworks can be difficult to preserve,
particularly as Moderna Museet encourages a playful and
Study locations experimental approach to displaying its art.
The selected artworks included those made of copper Gothenburg Museum of Art
and iron alloys and displayed under different conditions
in two different museums located in urban-coastal The Gothenburg Museum of Art building was inaugu-
environments: Moderna Museet in Stockholm and the rated in 1923 and was partially extended to gain more
Gothenburg Museum of Art in Gothenburg. exhibition space in 1968, with a new entrance added
in 1996. The collection includes Western art from the
Moderna Museet 15th century until today, with an emphasis on Nordic
art. The museum owns a unique collection specialized
Moderna Museet is a Swedish government-funded in Nordic art from the turn of the 20th century. The
museum responsible for the national collection of 70,000 artworks collection comprises 4,300 paintings, 900
modern and contemporary art. Opened in 1958, the sculptures, installations, and multimedia, around 61,000
museum has one of Europe’s finest and most well-known
works of art on paper, 900 photographs, and some 2,300
collections of 20th- and 21st-century art, comprising
children’s book illustrations.
more than 130,000 works in various media, including
some 1,800 sculptures and installations. Originally The museum has a history of climate issues. It was built
dominated by three categories—Swedish and Nordic art, with poor insulation and included a souterrain construc-
French-oriented modernism, and American art from the tion that exposes it to extra moisture. In 1996, the new
1950s and 1960s—the collection has been extended to entrance, connected to the Sculpture Hall by an open
include key international artworks, in line with current staircase, resulted in higher particle levels.
societal issues.
The climate system was installed in 2008 and includes
Moderna Museet’s building dates to 1998 and was reno- separate systems for high-security exhibition rooms and
vated in 2004. It is situated on Skeppsholmen, a small the collection. Some of the rooms housing the collection
island in the center of Stockholm. Exhibition halls for contain north-facing walls and no radiators, while in
the permanent collection cover 2,226 m2, divided into others there are south-facing windows and radiators.
19 rooms, grouped in three blocks and a corridor and For this reason, the climate is kept at level A, with a
RH of 50% ± 15% and a temperature that is set to 20°C

but can range from 18°C in winter to 24°C in summer.
In 2008, air filters were installed that currently include
class F9 filters.
The historical climate data are limited, as only single

meters were used. Periodic data show a RH range of
30 to over 70%, with repeated mold issues until 2008.
In the exhibition rooms relevant to this study’s selected
sculptures, there are now fewer humidity issues.
Artworks
The artworks selected for this study are highly important
Figure 2. Jean
to the museums’ collections and their preservation is of
Fautrier, La
particular concern. Grandetete,
1943, 33 cm,
Moderna Museet patinated copper
alloy, wood
The selected objects include a group of cast bronze sculp-
tures from Pablo Picasso (Femme aux bras croisés, 1948;
Nu, 1945; Femme debout, 1948; Le Bras, 1959) (Figure 1),
Jean Fautrier (La Grande Tête, 1943) (Figures 2, 3), and
Germaine Richier (L’Hydre, 1954). Additionally, analysis
and observations were performed on a steel and stainless
steel sculpture by a young Swedish artist, Hanny Kamali
(Freddie Gray, 2019) (Figures 4, 5).
The cast bronze sculptures are in one of the exhibition

rooms while the steel and stainless steel sculpture is
Figure 3. Jean Fautrier, La Grandetete, detail of a regularly monitored

area
Figure 1. Pablo Figure 4. Hanni

Picasso, Le Kamaly, Freddie
Bras, 1959, Gray, 2016–2018,
58 × 18 × 16 cm, 235 × 170 × 190 cm,
patinated copper steel and stainless
alloy steel
Figure 6. Pablo Picasso, Madame Fernande Olivier, 1906, 35.7 × 24.8 ×

Figure 5. Hanni
24.4 cm, panitated copper alloy
Kamaly, Freddie
Gray, detail of a
corroded area
displayed in the corridor leading to the exhibition rooms.

This corridor is exposed to a high level of visitor traffic
and is close to the main museum entrance. The sculptures
are dusted regularly.
Gothenburg Museum of Art

The two bronze sculptures analyzed were Pablo Picasso’s
Madame Fernande Olivier (Figure 6), acquired in 1951,
Figure 7. Marino Marini, Rider, 1952, 108 × 96 × 37 cm, copper alloy
and Marino Marini’s Rider, acquired in 1952 (Figure 7).
With a combination of distinct facial features and a more
unfinished and abstract part, Picasso’s Madame Fernande article described Marini’s working methods as casting
Olivier foreshadows an important shift in the artist’s and molding at the same time, reworking the raw cast
oeuvre, from his Rose Period (1904–1906) to his Cubist with sculptor’s tools and color, and then using plaster
phase. The sculpture dates from 1906 and was made after techniques, bringing the opaque, neutral, white mass
the Gósol period, but it is uncertain when it was cast. seemingly to life. The result is an immediate and tension-
filled effect that makes a deep impression on the viewer
The sculpture is likely one of the early sets of casts signed (Kuhn 1944). These details are reflected in Rider, the
by the artist. Subsequent unsigned casts were made in surface of which has traces of the sand from the casting
1959. In the acquisition record, the sculpture is described and hints of gold, blue, and red.
as bronze, with a dark-brown patina. Picasso used gauze
or net in the still wet plaster to achieve the porous texture Both sculptures have been on permanent display for
in the metal surface. almost 70 years, Rider in the Sculpture Hall, located at
the top of the staircase, where levels of particulates are
The Italian sculptor Marino Marini, one of modernism’s
high, and Madame Fernande Olivier in the French Gallery
foreground figures, referenced Etruscan heritage and
on the 6th floor, where the environment is more stable.
classic art in his works. Rider, from 1945, was made in
connection with his time spent in Switzerland and was
Methods
inspired by a 13th-century equestrian sculpture in the
cathedral of Bamberg, Bavaria. The monumental form The composition and condition of the sculptures were
in combination with the treatment of the surface, using investigated using visual observation and X-ray fluo-
scratches, chisels, and crayons, gives the sculpture a rescence (XRF). Pollutant levels were measured in two
timeless appearance that seems to refer to archaeo- different areas of each museum, each of which contains
logical antique finds. In Basel’s National-Zeitung, an at least one of the sculptures discussed. The areas were
selected because their displayed objects have suffered Table 1. Detection limits of pollutants measured by passive sampling
devices
corrosion or due to their corrosion potential. At Moderna
Museet, air quality was measured in the corridor leading Pollutant Detection limit (μg/m3)
to the exhibition rooms (containing Hanny Kamali’s CH3COOH 1.0
Freddie Gray) and in the exhibition room containing the HCOOH 1.0
Picasso, Fautrier, and Richier sculptures. The corridor SO2 0.2
is near the main entrance of the building while the HCl 0.3
exhibition room is shielded by walls. At the Gothenburg

Museum of Art, air quality was measured in the Sculp- both the original materials used to create each sculpture
ture Hall containing Marini’s Rider and in the French and the identification of possible corrosion products
Gallery on the 6th floor. It should be noted that yearly resulting from atmospheric pollutant exposure. The
visitor traffic is higher in Moderna Museet than in the XRF device consisted of an XG Elio portable, noncon-
Gothenburg Museum of Art. tact XRF spectrometer with a rhodium tube. For each
measurement, a voltage of 40 keV, a current of 20 μA, and
Pollutant monitoring a measurement time of 40 s were applied. The head of
the spectrometer was placed close to the surface and the
The intermittent monitoring of pollutants allows a
measurement was made with the aid of alignment lasers.
better understanding of the microenvironments in these
The medium between the object and XRF head was air
museums. Monitoring has thus far been performed using
(no light detection element was possible). At least five
passive sampling devices from the Swedish Environ-
different points were measured on each sculpture, with
mental Research Institute (IVL). The devices are compact
each point measured in duplicate or triplicate.
plastic cylinders approximately 3 cm in diameter and
1 cm thick and containing adsorbent materials. They
Results
require no pumps and allow for the diffusive capture
of pollutants through a mesh on the top of the device. Climate in Stockholm and Gothenburg
As they are quiet, small, and unobtrusive, these passive Climate data were collected by the authors from the
sampling devices are ideal for museum environments. Stockholms Luft- och Bulleranalys (SLB) (www.slb.nu/
Their exposure times can last between a few hours and slbanalys/) for Stockholm and the Swedish Meteorolog-
several weeks. Longer exposure times allow for lower ical and Hydrological Institute (SMHI) (www.smhi.se/q/
limits of detection, which is necessary for cultural Stockholm/2673730) for Gothenburg. The climate in
heritage environments, where low levels of pollutants both cities is considered mild. Average winter temper-
over time can cause collection damage. The samplers atures annually vary from −14 to 5°C (coldest month:
used in this study collect acid gases, namely acetic acid February) and average summer temperatures from 6 to
(CH3COOH), formic acid (HCOOH), sulfur dioxide 19°C (hottest month: August). It rains year-round but the
(SO2), and hydrochloric acid (HCl), and were exposed intensity and the amount vary, with May and September
for 4 weeks (8 September–7 October 2021, at the Goth- as the wettest months and April as the driest month. The
enburg Museum of Art; 18 October–3 November 2021, largest deviation occurs during the winter, although in
at Moderna Museet). The samplers were either placed some years a large deviation can also occur in summer.
directly on a flat surface (e.g., ground, platform) with The average RH is very high, averaging between 60 and
the mesh facing up or were adhered vertically to a wall 85%, with lower levels in the summer but with greater
using putty. After exposure, the samplers were returned fluctuations.
to the IVL for analysis using ion chromatography. The
Both Stockholm and Gothenburg are port cities and
detection limits for each measured pollutant over a
therefore receive a large volume of shipping traffic. Based
4-week exposure time are given in Table 1.
on pollutant data collected from a weather station near
the Gothenburg Museum of Art for the past 10 years,
Compositional analysis of the artworks (XRF)
average nitrogen dioxide (NO2) levels in this area, with
The elemental composition of the selected sculptures was the exception of 2020, were 30.9 ± 6.6 μg/m3, which is
determined using XRF, which provided information on higher than the suggested concentration range for general
cultural heritage collections of 3.8–19.1 μg/m3 (Grzywacz devices placed in Moderna Museet were, unfortunately,
2006). In 2020, because of the COVID pandemic, the unexpectedly removed during the original measurement
levels declined to an average of 17.94 μg/m3. Outdoor period and were not recovered. New samplers were placed
annual PM10 levels have been relatively constant, even in on 18 October 2021, but the data could not be analyzed
2020, with an average of 18.9 ± 2.3 μg/m3 (SMHI 2021). in time for this publication.
According to data from the past 10 years, NO2 levels Thus, as a comparison, in 2019 Canosa and Norrehed
in Stockholm have fluctuated, with annual averages of (2019) measured indoor pollutants at the Nationalmu-
12 ± 11 μg/m3 (SMHI 2021, SLB 2021). Concentrations seum, located near Moderna Museet. These measure-
are highest in February and March (highest measured ments, performed during a period of renovation (between
value: 113 μg/m3) and lowest during the summer months 18 January and 15 February 2019), showed indoor
(lowest measured value: below sensor detection levels). average values of < 0.2 μg/m3 for SO2, 14.7–42.3 μg/m3
Conversely, ozone concentrations are highest between for acetic acid, 28.4–36.2 μg/m3 for formic acid, and
May and August (highest measured value: 145 μg/m3) and < 1.0–1.5 μg/m3 for ozone. The high concentrations may
lowest during the winter months (lowest measured value have been the result of the renovation environment and
0.2 μg/m3), with a 10-year average of 51.4 ± 19.5 μg/m3. the use of fresh construction and decoration materials
Average SO2 levels in both cities over the last 10 years that were off-gassing.
have been low (0.4–1.5, with a winter average maximum
of 2.9 μg/m3). However, SO2 levels have been steadily X-ray fluorescence
declining, with a dramatic decrease over the last 30 years Figures 8 and 9 show representative spectra of each
due to environmental regulations, good public transport, artwork. All of the artworks are composed of bronze
and the widespread use of bicycles. This is consistent (evidenced by the presence of copper and low amounts
with many European and North and South American of tin) containing zinc and iron. Additionally, all objects
countries, where SO2 levels have similarly decreased were determined to contain low amounts of copper
since the 1970s (Klimont et al. 2013). corrosion products, based on the presence of sulfur and
chlorine peaks. Traces of chromium were detected in the
Indoor air quality Richier sculpture (MR), Fautrier sculpture (MF), and
Data from the IVL passive samplers exposed for 4 weeks Moderna Museet Picasso sculpture (MP), and traces of
showed that the indoor concentrations of SO2, acetic nickel, silicon, and potentially titanium in the leaded
acid, and HCl were below the instrument’s detection bronze sculpture by Marini (GM). Unlabeled peaks are
limits. While the levels of formic acid were detectable,
they were still below the suggested ranges for sensitive
cultural heritage collections (Table 2). In general, the
indoor air quality measured between 8 September and
7 October 2021 was good. However, the passive sampling
Table 2. Pollutants measured by passive sampling devices at the

Gothenburg Museum of Art, 9 September–7 October 2021. All values
are in μg/m3
Suggested Suggested
Compound GBG GBG range for range for
measured location 1 location 2 sensitive general
collectionsA collectionsA
SO2
< 0.2 < 0.2 0.11–1.1 1.1–5.3
HCOOH 6.4 6.4 < 9.6 9.6–38.2
CH COOH
3
<1.0 < 1.0 < 12.5 99.6–697.2
Figure 8. XRF spectra of the analyzed sculptures. MR: Moderna Museet,
HCl < 0.3 < 0.3 NA NA
Richier; MF: Moderna Museet, Fautrier; MP: Moderna Museet, Picasso;
A: Data from Grzywacz (2006), converted from ppb to μg/m3 GP: Gothenburg Museum of Art, Picasso; GM: Gothenburg Museum of
NA: Not applicable, no current data found in the literature Art, Marini
lacking. Each artwork displays different patinas, ranging

from uniform, highly polished, and homogeneous to
non-uniform and rough, due partially to their manufac-
turing and finishes and partly to the selective corrosion
that has developed over their lifetime.
Corrosion rates are often linked to the type and level

of pollutants as well as to temperature and humidity.
Copper- and iron-based alloys are susceptible to corro-
Figure 9. Details of the XRF peaks observed in the spectra from Figure sion under conditions such as those found in both
8. The left side shows the peaks found in the energy range of 1.5–8 keV Moderna Museet and the Gothenburg Museum of Art.
and the tin Ka peak at 22–28 keV
The high levels and wide fluctuations of humidity in
combination with NO2, SO2, O3, HCl, Cl2, reduced sulfur
from the rhodium target in the XRF or from argon in compounds, and organic acids could result in increased
the surrounding atmosphere. rates of corrosion (Leygraf et al. 2016, 87). Therefore,
The XRF spectra from three different measurement the monitoring of environmental parameters, such as
points on the Marini sculpture at the Gothenburg those discussed here, can help conservators determine
Museum of Art are shown in Figure 10. According to whether action should be taken to prevent corrosion.
the spectra, areas showing white surface accretions The cost of monitoring pollutants is lower than that
contain increased relative amounts of zinc, tin, and iron, of conservation treatments, and intermittent meas-
indicative of the selective corrosion of zinc (i.e., dezinci- urements may suffice. As conservation moves towards
fication). Additionally, the XRF spectra for the Kamali preventive measures, pollutant monitoring could be a
sculpture at Moderna Museet showed the presence of two significant step in this direction. As demonstrated by
different iron alloys: a galvanized stainless steel threaded the sculptures included in this study, leaded bronzes
rod bar that is attached to a body of plain steel (Figure 5). such as used in the Marino Marini sculpture can
corrode in the presence of organic acid gases, including
those of acetic acid and formic acid (Grzywacz and
Tennent 1994, Tétreault et al. 2003, Grzywacz 2006).
Copper alloys can corrode in the presence of common
atmospheric gases, especially sulfur dioxide and chlo-
rine-containing compounds (e.g., HCl and Cl2) (Rice
et al. 1981), both of which will be high given the prox-
imity of both museums to the sea and to port traffic.
While the pollutant concentrations measured in this
first monitoring campaign were lower than suggested
for sensitive collections (see Table 2), they may fluc-
tuate, both seasonally and over time. Additionally, the
environmental data collected (SLB 2021, SMHI 2021)
showed a significant decrease in pollutants during the
pandemic. Therefore, future measurement campaigns
Figure 10. XRF of different areas of Marino Marini’s Rider. (A) left hip of
the horse, dark green patina; (B) left side of the human torso; (C) left side will be performed.
of the human torso in an area with significant white accretions
COVID reduced the exposure of the air quality samplers,
as a number of museums in Sweden remained closed for a
Discussion
significant period of time and, once opened, their visitor
Different surface finishes during a sculpture’s manufac- numbers were much lower. Exposure levels increased
ture may result in different corrosion rates. Conserva- once the museums were back to relative normalcy, with
tion also impacts corrosion, but sufficient information the first samplers exposed in September, albeit before
regarding previous treatments of the studied objects is governmental restrictions were lifted (on 29 September
2021). The next set of samplers will be placed in late default/files/static/6972-2014-iic-icom-cc-environmen-

December and then in the spring and summer. tal-guidelines.pdf (accessed 10 October 2021).
In the next steps of the project, high-definition photo- Klimont, Z., S.J. Smith, and J. Cofala. 2013. The last
graphs of the selected areas will be made under consistent decade of global anthropogenic sulfur dioxide: 2000–2011
lighting conditions. Fourier transform infrared (FTIR) emissions. Environmental Research Letters 8(1): art.
spectroscopy using contactless nondestructive reflection 014003. doi: 10.1088/1748-9326/8/1/014003.
will be employed in situ to detect traces of varnish or Kuhn, H. 1944. Vier ausländische Bildhauer im Basler
wax used in the past. Other analytical methods, such Kunstmuseum. National-Zeitung 12, December 1944.
as Raman spectroscopy, could be used to determine the
Leygraf, C., I. Odnevall Wallinder, J. Tidblad, and T.
composition of newly formed corrosion products (espe-
Graedel. 2016. Atmospheric corrosion, 2nd ed. Hoboken,
cially those representing active corrosion).
NJ: John Wiley & Sons.
Conclusion The Ministry of Infrastructure (Sweden). 2020. Sweden’s

Integrated National Energy and Climate Plan. https://
As museums align with wider environmental goals and ec.europa.eu/energy/sites/ener/files/documents/se_
conservation treatments become less frequent than in the final_necp_main_en.pdf (accessed 15 October 2021).
past, measures aimed at preventing corrosion become
essential tools in the conservation of metal sculptures. National Museum Directors’ Council (NMDC). 2014.
Environmental monitoring, including the monitoring of Environmental sustainability – Reducing museums’ carbon
humidity, temperature, and pollutants, can be a signifi- footprint. Bizot Green Protocol. Available at www.national-
cant indicator of potential corrosion and other forms of museums.org.uk/what-we-do/contributing-sector/
deterioration. Related areas of research include descrip- environmental-conditions/ (accessed 10 October 2021).
tors based on high-definition imaging. Most museums Rice, D.W., P. Peterson, E.B. Rigby, P.B.P. Phipps, R.J.
monitor RH and temperature but not pollutants. This Cappell, and R. Tremoureux. 1981. Atmospheric corro-
project explores practicable methods and novel, low-cost sion of copper and silver. Journal of The Electrochemical
tools, such as passive samplers, to develop protocols for Society 128(2): 275–84.
monitoring artworks vulnerable to corrosion and other Stockholms Luft- och Bulleranalys (SMB). 2021. www.
forms of environmental damage. slb.nu/slbanalys/ (accessed 1 October 2021).
References Swedish Meteorological and Hydrological Insti-

t u t e ( S M H I ) . 2 0 2 1 . w w w. s m h i . s e / q / S t o c k -
Canosa, E. and S. Norrehed. 2019. Strategies for pollutant holm/2673730 (accessed 1 October 2021).
monitoring in museum environments. Project report.
Visby: Swedish National Heritage Board (Riksan- Tétreault, J., E. Cano, M. van Bommel, D.A. Scott,
tikvarieämbetet). doi: 10.13140/RG.2.2.24172.00640. M. Dennis, M.-G. Barthés-Labrousse, L. Minel, and
L. Robbiola. 2003. Corrosion of copper and lead by
Grzywacz, C.M. 2006. Monitoring for gaseous pollut- formaldehyde, formic and acetic acid vapours, Studies
ants in museum environments. Los Angeles, CA: Getty in Conservation 48(4): 237–50. https://doi.org/10.1179/
Publications. sic.2003.48.4.237
Grzywacz., C.M. and N.H. Tennent. 1994. Pollution
monitoring in storage and display cabinets: Carbonyl Authors
pollutant levels in relation to artifact deterioration.
Stavroula Golfomitsou (BA, PhD, FIIC) is senior
Studies in Conservation 39: 164–70.
lecturer at the Department of Conservation, University
International Institute for Conservation of Historic of Gothenburg, Sweden. Previously, she was a lecturer in
and Artistic Works (IIC) and International Council of conservation studies at UCL Qatar, where she set up and
Museums – Committee for Conservation (ICOM-CC). coordinated the MSc program in Conservation Studies.
2014. Environmental guidelines – IIC and ICOM-CC
declaration. Available at www.iiconservation.org/sites/
Discovery of a 12th-Century Enamelled Reliquary Pendant:
Elemental Analysis and Content Visualisation Using Prompt
Gamma Neutron Activation Analysis and Neutron Tomography
Matthias Heinzel* Eschly Kluge Burkhard Schillinger Christian Stieghorst
Leibniz Research Institute Institute for Nuclear Physics (IKP), Heinz Maier-Leibnitz Zentrum Heinz Maier-Leibnitz Zentrum
for Archaeology, Römisch- University of Cologne (MLZ), Technical University of (MLZ), Technical University of
Germanisches Zentralmuseum Köln, Germany Munich (TUM) Munich (TUM)
Mainz (RGZM) [email protected] Garching, Germany Garching, Germany
Mainz, Germany www.ikp.uni-koeln.de/ [email protected] [email protected]
[email protected] https://mlz-garching.de/ https://mlz-garching.de/
https://web.rgzm.de/ Dorothee Kemper
German Society for Studies in Art *Author for correspondence
History
Berlin, Germany
[email protected]
www.dvfk-berlin.de
Abstract
In October 2008, a heavily corroded pendant was found during latter consisted of small packages of tiny particles of bone
an excavation campaign in the historic district of Mainz, covered by textile and wrapped with a thin thread to keep
Germany, in an ancient dump. The reliquary pendant, made them together. Neutron imaging allowed exact measurements
of gilded copper, is in the shape of a quatrefoil, also referred of the textiles and bones. The construction of the container
to as a phylacterion, and dates from the late 12th century. and especially the closure mechanism were also clearly visible.
Its front and verso were enamelled using the technique of With the aid of infrared spectroscopy, a small sample of wax
émail champlevé and its several compartments include figural was identified as beeswax. The analysis of a small fragment of
representations. The side section is also gilded and enamelled fibre which had survived in the pendant’s suspension eyelet
with cross-hatched decoration. Two different techniques were revealed that it was made of silk. This object is one of only
applied in the analysis of the pendant’s surface. An elemental four known phylacteria of this type from the late 12th- and
analysis was done using micro-X-ray fluorescence, and the glass early 13th-century Hildesheim workshop in Lower Saxony.
and pigments of the enamels were investigated using Raman The other three are housed in Boston, Rome and Halberstadt.
spectroscopy. The carbonaceous content of five reliquary This reliquary pendant is a rare example of an object found in
packages became visible after neutron imaging, performed context from controlled archaeological excavations.
at the Research Neutron Source Heinz Maier-Leibnitz
(FRM II), Technical University of Munich in Garching, Germany. Keywords
A combination of a position-sensitive prompt gamma-ray reliquary, phylacterion, neutron imaging, prompt gamma-ray
neutron activation analysis and neutron tomography was used neutron activation imaging (PGAI), (micro-X-ray fluorescence)
to reveal the reliquary’s internal structures and contents. The μ-XRF, émail champlevé, Hildesheim
Introduction
A heavily corroded reliquary pendant was found at an
ancient dump in October 2008, in the historic district
of Mainz, Germany. The dump was rectangular, approx-
imately 3 × 3 m in size, and had been backfilled in the
mid-14th century. The pendant was already relatively
old before it ended up in the dump, where it stayed for
over 600 years.
The interdisciplinary analysis of the pendant included

Figure 1. Pendant before its restoration and radiography of its side
art-historical, material-scientific and manufacturing and front (photograph by Matthias Heinzel, radiographs by Stephan
perspectives aimed at reconstructing the pendant’s manu- Patscher, RGZM)
DISCOVERY OF A 12TH-CENTURY ENAMELLED RELIQUARY PENDANT: ELEMENTAL ANALYSIS AND CONTENT VISUALISATION 185
facture, the composition of its enamels, the iconography

and the possible origin. Neutron tomography in combi-
nation with position-sensitive prompt gamma neutron
activation imaging (PGAI-NT) was applied to the object
for elemental analysis and content visualisation.
Condition and conservation treatment

The pendant was covered with a 3- to 5-mm-thick layer of
green and blue copper corrosion products with embedded
soil constituents. An initial radiograph of the pendant
performed at the Leibniz Research Institute for Archae-
ology (RGZM) revealed the exact shape of the quatrefoil
and several figural representations (Figure 1). The copper Figure 2. Digital microscopy: one of the five fibre fragments from the
eyelet, as seen at 200× magnification (by Matthias Heinzel with a
corrosion products were removed mechanically using Keyence VHX-6000 digital microscope)
diamond-tipped grinding tools and fine mechanical instru-
ments. During this process, the original enamel and gilded
copper surface were uncovered. As the entire base material
was already heavily mineralised, the gilding had detached
from the base metal in many areas or lay exposed on top.
In addition, some areas of the enamel were chipped

or brittle and the entire object was relatively unstable.
During the restoration process, conducted in 2009 in
the workshop of the Directorate of State Archaeology in
Mainz, exposed areas were stabilised with epoxy resin Figure 3. Front, side and verso of the pendant after restoration
(Araldite 20/20) to avoid a loss of substance. The epoxy (photographs by Sabine Steidl, RGZM)
resin was of high transparency, with extensive non-yel-
lowing properties, high adhesive strength, and the same the pendant began when it was shown to art historian
light refraction as glass. Due to its low viscosity, it is very Dr Dorothee Kemper during a conference in Hildesheim
well suited for infiltration bonding. At the time of the on the famous Welandus or Henry reliquary from the
restoration, 13 years ago, it was one of the best avail- Louvre in Paris. She immediately recognised the special
able options and thus far no change is discernible. The nature of the Mainz reliquary pendant and encouraged
stabilisation had no influence on any of the subsequent further exploration.
analytical methods. During the discovery, a cord-like
fragment of fibre was found inside the pendant’s suspen- Description
sion eyelet. The surrounding copper corrosion products
The rare pendant, also known as a phylacterion, is quat-
preserved this fragment, due to their biocidal effect. This
refoil shaped with a central square (Kemper 2020). It is
1.5-mm-long fragment could not be preserved in situ and
made of gilded copper. An eyelet is visible at the top,
was separately packaged for further analysis (Figure 2).
suggesting that the object was worn around the neck on
After its restoration, the pendant had the following a ribbon or chain. This object consists of a container and
dimensions: length 67 mm (with bail), 57 mm without a closure which are held together by a visible rivet. The
bail; width 57 mm, and thickness 11–12 mm. The front and verso were enamelled using the technique of
suspension is ~10 mm in diameter. The object’s total émail champlevé and are decorated with several figural
mass is 70.5 g. Conservation work on the pendant representations (Figure 3). All figures carry a nimbus
required ~500 hours in total. The restored pendant was and are raised from the background, engraved and
kept in storage at the Directorate of State Archaeology gilded. The background fields, the halos, the frame of
in Mainz until the end of 2016, when the first author the central field and the internal drawings of the figures
started a new job at the RGZM. In 2017, research on are enamelled. The colour spectrum of the enamelled
areas is very narrow and varies on the front and verso. Results of the μ-XRF (in weight %)
The square field on the front is green and framed by blue
• Substrate metal is copper: Cu 99.2%/Pb 0.4%/
semicircles, with the halo of the central figure being blue
Ag 0.2%/Sn 0.2%
and the halos of the surrounding representations being
• Gilding was applied using fire gilding: approx.
white. On the back side, the colours are reversed: the
Au 83%/Hg 17%
square field is blue and the semicircle fields are turquoise
green, whereby the halo of the central figure is white and The pendant consists of pure copper, with only traces of
the halos of the surrounding figures are blue. other elements, most likely reflecting impurities from
the ore that was used. Fire gilding was evidenced by the
The representations in the upper semicircle segment of
high content of mercury.
both sides, where the riveting of the two components is
located, were constructed with the placement of the rivet There are four different opaque enamel colours on this
having been taken into account, such that the figures on object: white, blue, green and turquoise green. Due to
the front and verso tilt their heads to the right to make its fragility, the enamel was measured under normal
room for the rivet. The side section of the pendant is atmosphere and not in a vacuum, as is usual for glass
cross-hatched, except the closure. The lozenges are gilded and enamel. Thus, a quantitative determination of the
and the lines are enamelled. glass composition was not possible, although it can be
concluded that all colours were made from soda glass
Optical examination and the green glass is a leaded glass.
High-resolution digital photos were taken for documen- The other three enamel colours contain lead, with a
tation and further research. The cord-like fragment, few percent of lead oxide. Calcium antimonate was the
consisting of five thread fragments, from the area of opacifying agent in the opaque enamels (white, blue and
the suspension eyelet was examined in photos obtained turquoise). The antimony content of the green glass is
using a Keyence digital microscope with reflected and too low to have created opaque enamel. The colours
transmitted light. During the restoration process, a soft, of the enamels were derived as follows: blue by copper
waxy substance was discovered on the side containing a and cobalt, green by copper and lead, and turquoise by
representation of Mary, in the upper semicircle segment copper. The turquoise enamel also contains tin and zinc.
with the rivet. In this area, where the closure is also Presumably, a bronze alloy that was on hand was added
located, the surface of the pendant is heavily damaged, here instead of pure copper, a common procedure and
most likely due to the strong degradation that occurred indicative that the goldsmith used what he had at hand.
during its time in the ground. A sample of this waxy
substance was taken for analysis. Results of FTIR of the waxy substance
Transmitted light microscopy of the fibre sample revealed Infrared spectroscopy clearly identified the waxy
the fibre to be silk. The cord consisted of at least five substance as beeswax (Figure 4).
strands of thread, each of them z-twisted. It is the first
evidence of material on which such pendants were most
likely carried.
Technological studies
A radiograph (front and side) was made to gain insights
into the construction of the pendant. Non-destructive
and non-invasive methods, specifically, micro-X-ray
fluorescence (μ-XRF) and Raman spectroscopy, were
combined to obtain information on the composition of
the metals and enamels (type of glass, colourants and
opacifiers). The waxy substance was investigated by Figure 4. FTIR spectra of beeswax (top) and a sample of the waxy
infrared spectroscopy (FTIR). substance from the reliquary pendant
The construction of the quatrefoil pendant suggested that stemmed from worshiped saints. To secure the fragments
it contained relics, but these were not initially visible on within the textile, thin threads were wrapped around the
the X-rays taken at the RGZM. In X-ray images, organic woven textile wraps. Some packets were more sack-like,
and inorganic materials, such as textiles and bones, are with one thread closing it on top. Usually, such packages
optically ‘outshone’ by the surrounding metal and enamel were labelled with the specific saint’s name, written on
and are thus not recognisable. attached parchment strips known as authentic or cedula.
However, in the present case, no signs of parchment strips
For X-ray-complementary, non-destructive structural
were seen on the NT images. It cannot be ruled out that,
and elemental analyses, the reliquary was investigated after being buried for more than 600 years, any parch-
at the research reactor FRM-II of the Heinz Maier-Lei- ments underwent either partial or complete decay. Since
bnitz Institute of the Technical University of Munich at the reliquary’s riveting is strong and tight, the subsequent
Garching, using NT of different spatial resolution and removal of the strips is unlikely, as opening the reliquary
PS-PGAA. The absorption behaviour of neutrons is would have required great force, of which there are no
nearly opposite of that of X-rays. Free of electrical charge, signs. It may be that a list of the reliquaries contained in
neutrons penetrate deeper into matter, specifically metals. this pendant was stored separately.
Conversely, their stronger absorption by hydrogen results
in a high imaging contrast in biological material. NT was With the aid of NT and the highly valued interpreta-
used to specifically visualise the structure of the organic tions of the reconstructions, the structural properties
content and allowed the identification of five distinctive of the reliquaries could be determined in detail. For
specimens showing high grey-value contrasts, it was
packages (Figure 5). The investigation methods have
possible to measure the thread spacing or density, with
no implications for possible carbon dating and DNA
a mean square error of ± 0.3 threads. Since the wrapped
extraction. However, the latter are not possible because
reliquaries were stuffed into the pendant from the top,
the reliquary pendant cannot be opened without causing
most of their fabric is heavily warped. Thus, many threads
its destruction.
within the observable fabric parts were either extremely
In general, reliquary packages of the Middle Ages stretched apart or compressed, which in some specimens
consisted of bone fragments wrapped in linen or silk. hindered determinations of the horizontal and vertical
In most cases these fragments were claimed to have thread densities (Figure 6).
Figure 5. Neutron tomography of the pendant with its five reliquary Figure 6. Detailed structures of the reliquary packages (textile and
packages splinters) as seen on a NT-slice
Details of the five reliquary packages along the vertical The thinly threaded, denser textile of bags 1, 2, 3 and 5 is
symmetry axis of the pendant from top to bottom include most likely silk, and the thicker, less dense textile of bag
the following: 4 and within bag 1 is probably linen. In the NT recon-
structions, all structural components of the pendant were
1. Packaging: a very fine, loosely lying outer textile
clearly visualised. This allowed the complex construction
(vertical and horizontal thread spacing of 0.33 mm
of the closure (later shown to be opaque to X-rays) to
and 0.26 mm; vertical and horizontal thread densi-
ties of 30/cm and 39/cm), with an estimated spatial be investigated in detail. The chiselled immersions of
volume of ~2 cm3, encloses a relatively tight-fitting the émail champlevé were clearly visible in the NT slices
and coarsely threaded textile, whose nominal thread (Figure 7).
characteristics could not be determined. Contents:
an irregular tetrahedral-like splinter with maximum
dimensions of 7.2 × 5.3 × 5.1 mm.
2. Packaging: a loosely packed thinner textile (vertical

and horizontal thread spacing of 0.35 mm and
0.30 mm; vertical and horizontal thread density of
29/cm and 33.9/cm) with an estimated volume of
~6 cm3. Contents: two fragment plates which appear
to be of bone cortex and adjacent bone sponge tissue.
Maximum dimensions: one is 15 × 8.8 × 8.6 mm, and
the other 2.77 × 1.8 × 2.4 mm.
3. Packaging: loose bundle of fine textile (thread

spacing of ~0.34 mm or 39.5 threads per cm)
with an estimated volume of 2 cm3. The bundle is
wrapped with a single string. Contents: four small
Figure 7. Chiselled and engraved immersions of the émail champlevé
plates within the upper part. Maximum dimen- seen on the cross-sectional NT-slice
sions: 1. 10.8 × 1.0 × 7.0 mm; 2. 4.6 × 0.7 × 1.8 mm;
3. 2.6 × 0.5 × 1.0 mm; 4. 3.4 × 0.8 × 2.7 mm. The elemental compositions were determined using
PS-PGAA, in which two depth layers with 14 measure-
4. Packaging: tightly folded, thick textile (vertical and
ment positions each and an additional inter-layer and
horizontal thread spacing of 0.4 mm or 25 threads per
inter-row position at the object’s closure were analysed
cm) wrapped with a string. Estimated volume: ~3 cm3.
(Figure 8). The chosen rasterization covered all reliquary
Contents: some small and many very tiny fragments.
bags, with a total of 21 measurement positions (along the
Maximum dimensions of the most prominent frag-
pendant’s vertical axis): eight other measurements served
ments: 1. 2.2 × 0.9 × 2.5 mm; 2. 3.1 × 0.8 × 1.8 mm;
as the lateral reference. PGAI-NT is the combination of
3. 2.6 × 1.1 × 1.8 mm.
PS-PGAA imaging and NT.
5. Packaging: loosely packed bag of thinner textile
The PGAA results tensor is visualised in three dimen-
(thread spacing ~0.34 mm or 29.5/cm thread density)
sions via voxels embedded in the reconstructed object
and an estimated volume of ~4 cm3. Contents: two to
three crossed fragment plates. Maximum dimensions:
1. 4.0 × 0.6 × 5.0 mm; 2. 2.3 × 1.2 × 7.4 mm. Three to
four objects other than bone fragments are unevenly
distributed within the bag. Compared to the observed
bone fragments, those objects are highly homoge-
neous, spherically symmetrical, with much smoother
surfaces and a mean diameter of ~1.3 mm. Their grey
values, however, are identical to that of the bone frag-
ments. Their nature and material remain unknown. Figure 8. Reliquary PGAI-NT positions with 29 measurement points
volume on NT. The different elements are represented by

means of different colours and their relative abundance
is encoded via different colour intensities (Explanatory
notes: E. Kluge et al., www.ati.ac.at/~saagas26/css/files/
SAAGAS_26_paper_22.pdf, accessed 30 June 2020) . In
the analysis of the pendant, the elemental composition
of six positions was noteworthy. Compared to the refer-
ence positions and surface materials, these positions
contained significant amounts of hydrogen, calcium and
potassium. Manganese and iron were also found at a few
points (Figure 9).
Figure 9. PGAI-NT of the elements calcium, hydrogen, potassium/iron Figure 10. Exploded drawing of the construction components (drawing
and oxygen (from left to right) by Vera Kassühlke, RGZM)
The additional hydrogen content along the vertical axis

supported the presence of biological material. The rela- The closure was made separately and consists of a
tively high hydrogen content at the closure fits well with rectangular, ~10-mm-long frame made by bending a
the identified beeswax within this area. The increased 1.1-mm-thick copper sheet. The frame was then firmly
calcium content along the vertical axis together with soldered to a lid made from an ~1.6-mm-thick copper
the amounts of hydrogen found was a strong elemental sheet. The lid is slightly larger than the opening, so that
indication of the presence of bone material. Some of the it overlaps. In addition, there is a sturdy eyelet, made
hydrogen may have stemmed from the textiles. Based of 4 mm copper wire, with an oval cross-section on the
on the structural and elemental analyses obtained with outside of the cover plate. The wire ends of the eyelet,
PGAI-NT, it is extremely likely that the objects within forged into narrow sheet-metal strips, lead through the
bags 2, 4 and 5 are indeed biological bone material. The centre of the lid into the inside of the frame. The eyelet,
plate in package 2 is also likely to be bone, based on including the sheet-metal strips, is soldered to the cover
the visually prominent sponge structures in the cortex plate. The entire closure element is constructed like a
tissue. As the same certainty cannot be applied to the drawer that fits exactly into the existing opening of the
contents of bags 1 and 3, these objects were labelled container. A rivet pin connects this closure element to the
simply as fragments. In the analysis of the components main container. Like all other components, the rivet pin
of the pendant’s frontal and verso plates, the results of is made of copper. It has a round cross-section of 2.5 mm.
the PGAA coincided, as expected, with those of X-ray
fluorescence spectroscopy. Decoration
The container is enamelled and gilded; the closure area,
Manufacturing technique
the figural depictions and the outlines are gilded. The
The pendant consists of a bottom and a top sheet side area is cross-hatched and also gilded and enamelled.
of copper, each ~1.5–2.4 mm thick. The frame is The manufacturing process of the pictorial surfaces is
made of four equally sized circle segments, formed as follows: After the hollows were completed they were
of ~1.9-mm-thick and 8-mm-wide copper sheets. A filled with the appropriately coloured enamel powders
rhombic copper peg is located in the intersections of each and then heated. This process was most likely repeated
petal and forms the corners of the central square frame. several times until all hollows were sufficiently filled with
All of these components are soldered together, forming enamel. The entire surface was then polished (Brepohl
the quatrefoil shape (Figure 10). 1999). The thickness of the enamel varies between 0.5
and 0.9 mm. Next, almost all of the raised metal areas, pendant was shown to be made of gilded copper and was
i.e., the figural representations and outlines of the front enamelled using the technique of émail champlevé. The
and verso, as well as the diamond areas of the lateral four different enamel colourswere analysed by μ-XRF and
cross-hatched section, were fire-gilded and polished. Raman spectroscopy. Examination of a thread fragment
The areas of the closure that would later be visible on the from the area of the suspension eyelet revealed silk, thus
outside were also fire-gilded. Finally, the relic packages providing the first evidence of the material on which such
were placed inside and the container was tightly closed pendants were most likely carried. A sample of a waxy
with the aid of the rivet. The beeswax found in the area substance could be identified as beeswax by infrared
of the closure could have served as a sealing compound spectroscopy. NT was used to visualise the contents of
to additionally secure the closure of the reliquary and/ five relic packages and measurements of the textiles and
or to firmly connect the container with the closure and individual splinters were obtained. The material thickness
thus prevent it from wobbling. of the metal parts and the enamel fields was similarly
determined. PS-PGAA allowed an elemental analysis of
Historical classification the pendant’s contents. This reliquary pendant is a rare
example of such an object from a controlled excavation.
Based on several criteria, the Mainz phylacterion can It was only through the collaboration of archaeologists,
be assigned to a group of smaller, enamelled quatrefoil art historians, chemists, neutron researchers, a nuclear
phylacteria from the second half of the 12th century physicist, restorers and goldsmiths that it was possible
(Kemper 2020). These phylacteria, now kept in Boston, to arrive at these detailed results.
Rome and Halberstadt, have approximately the same
dimensions, basic shape, manufacturing technique, Analytical details/methods
pictorial motifs, colour spectra and internal division
• Micro-XRF was conducted by Ms Sonngard Hart-
(Boston, Museum of Fine Arts, Inv. no. 55.466; Rome,
mann, RGZM. Instrument: Eagle III spectrometer
Vatican Museums, Hall of Addresses, Inv. no. 64548;
(Röntgenanalytik, Taunusstein), rhodium source,
Halberstadt, Dommuseum, Inv. no. 27). The pictorial
maximal power at 40 kV and 1 mA, 25 μm titan filter,
motifs of this group are limited to themes of Christ, Mary
beam diameter 0.3 mm, vacuum, the surface was
and the Crucifixion (with accompanying figures). They
analysed under clean conditions but without further
vary slightly in design and type but all are characterised
surface preparation.
by a generally widespread iconography that did not have
• Raman spectroscopy was conducted by Dr Marlène
to convey any specific references to the relic content, the
Aubin, RGZM. Instrument: LabRam HR800 (Horiba
place of origin/destination, or the bearer. With these char-
Jobin Yvon company), laser 532 nm, 1800 lines/mm
acteristics, they coincide with productions of Hildesheim
grating, 50 × long focal lens objective.
enamel workshops from around the mid-12th to the
• FTIR-microscopy was conducted by Ms Sonngard
mid-13th century, where items for liturgical or religious
Hartmann and Mr Matthias Heinzel, RGZM. Instru-
use were serially produced.
ment: Thermo Scientific FTIR spectrometer. Measured
with an ATR crystal (diamond).
Conclusion • NT was conducted by Dr Burkhard Schillinger at
The quatrefoil shaped reliquary pendant dates to the FRM II, Garching, TUM. Instrument: ANTARES
last third of the 12th century and most likely came (https://mlz-garching.de/antares) (27 October 2021).
from a workshop in Hildesheim, Lower Saxony. This is • PS-PGAA was conducted by Mr Eschly Kluge and
supported by its shape and by other characteristics found Dr Christian Stieghorst at FRM II, Garching, TUM.
in comparable objects. The object presented here is one Instrument: PGAA (https://mlz-garching.de/pgaa)
of only four phylacteria of this type known today from (27 October 2021).
the Hildesheim workshop; the other three are housed in
Halberstadt, Boston and Rome. This artefact was made References
to contain reliquary packages, which after completion Brepohl, E. 1999. Theophilus Presbyter und das mittelalter-
of the container were inserted and the reliquary then liche Handwerk, Buch III, Kap. LIIII, 136-140. Köln:
tightly closed. Through different modes of analysis, the Bölhau.
Kemper, D. 2020. Die Hildesheimer Emailarbeiten des

12. und 13. Jahrhunderts. Regensburg: Schnell & Steiner.
Author
Matthias Heinzel is a conservator responsible for the
precious-metal conservation workshop at the Leibniz
Research Institute for Archaeology, Römisch-German-
isches Zentralmuseum Mainz, Germany. He is a trained
goldsmith and a conservator of archaeological objects.
ANALYTICAL INVESTIGATIONS
Sulfur, the ‘Enemy of Copper’: Replication of Sulfurous
Efflorescence on Copper-based Heritage Materials with
Elemental Sulfur
Charlotte F. Kuhn-Wawrzinek Thomas Schleid
TEMA GmbH Institute for Inorganic Chemistry, University of Stuttgart
Kusterdingen, Germany Stuttgart, Germany
Gerhard Eggert* *Author for correspondence

State Academy of Art and Design
Stuttgart, Germany
[email protected]
Abstract
Copper containing visible three-dimensional sulfurous composition – crystalline copper sulfides, oxides and (rarer)
efflorescence is known in Europe as ‘black spots’ and in the sulfates, and amorphous material – was detected and analysed
USA as ‘brown fuzzies’. These are found on many heritage using Raman microscopy, energy-dispersive X-ray (EDX) analysis
materials containing copper exposed to traces of sulfurous and X-ray powder diffraction (XRPD). For the first time, typical
gases. A systematic investigation was carried out to get a ‘black spot’ corrosion was reproduced in laboratory experiments.
better understanding of the different factors that influence the The laboratory results aligned well with analysis of ‘black spot’
underlying corrosion process. The low sublimation pressure corrosion samples from objects. A systematic correlation
of sulfur in closed containers was utilised to generate typical between colour, morphology and chemical composition of the
efflorescence on copper as well as copper alloys with different efflorescence was found and the systematic influence of the
surface properties at varying relative humidity (RH) and parameters of RH, temperature, alloy type and surface properties
temperature. The stability of chalcocite and covellite, often found was proven. The stability of covellite and chalcocite decreases
as corrosion products in ‘black spots’, was investigated by exposing with increasing RH, which increases the corrosive effect of
powder samples to different RH. Laboratory experiments were covellite on metallic copper when in direct contact.
complemented by surveying more than 300 affected objects
from 11 different collections. The laboratory-grown ‘black spots’, Keywords
powder samples and corroded objects were studied using light anilite, black spots, chalcocite, copper sulfide, covellite,
microscopy and scanning electron microscopy (SEM) and their efflorescence, elemental sulfur, spionkopite
Introduction
The Latin word sulp(h)ur is related to the Sanskrit
s(h)ulbari, a composite of shulba (copper) and ari
(enemy) (Böhtlingk and Roth 1875, col. 269; Monier-Wil-
liams 1899, 1084; Vira 1950, XVI, XXI). The name ‘enemy
of copper’ might preserve ancient chemical knowledge
on the reactivity of sulfur with copper, which is the
background for our study.
The characteristic feature of ‘black spots’ is their striking

three-dimensional growth. Weichert et al. (2004) defined
their morphological nomenclature as ‘trees, bunches and
cauliflower’. They can easily be rubbed off an object’s
surface due to their loose cohesion. Contrary to what
the common term ‘black spots’ suggests, they do not Figure 1. ‘Black spots’ and ‘brown fuzzies’ in the eye of a classical Zeus
only appear in black but also brown, grey and irides- figure
SULFUR, THE ‘ENEMY OF COPPER’: REPLICATION OF SULFUROUS EFFLORESCENCE ON COPPER-BASED HERITAGE MATERIALS WITH ELEMENTAL SULFUR 195
cent colours. In the USA, the term ‘brown fuzzies’ is products and the source of the pollutant. The forma-
commonly used to describe the phenomenon (Eggert tion of sulfurous efflorescence needs volatile sulfur
2021) (Figure 1). sources. There is no proof at all in the literature that the
ubiquitous pollutant sulfur dioxide can produce sulfide
Conservator Gustav Rosenberg (1878–1941) of the
phases, but it might have a synergistic role in corrosion
National Museum of Denmark first described black
processes. Instead, gases containing reduced sulfur (i.e.
spots on bronzes in 1913 in his unpublished working
sulfur with oxidation number −II, −I or 0) have been
notes (Madsen 1985). The first public mention of black
considered in the literature: hydrogen sulfide (H2S),
spots was in a recorded conference discussion remark by
carbonyl sulfide (COS), carbon disulfide (CS2), dimethyl
Werner Wimmel in 1972 (Kirchner 1973, 47). Unpub-
sulfide (CH3SCH3), dimethyl disulfide (CH3SSCH3) and
lished microprobe analyses from 1973 in the archives
elemental sulfur (references in Kuhn-Wawrzinek 2020,
of Antikenmuseum Berlin (Peltz 2021, 541 ff.) detected
22). Cases where the emitting material has been identified
copper, sulfur and possibly oxygen and assumed the
are quite rare, mostly rubber vulcanised with sulfur or
existence of copper sulfide. In the first article on the
wool (emitting, e.g., COS) used in display cases. They can
subject, Madsen (1977) detected covellite by X-ray
easily be identified by the Oddy test or the iodine-azide
powder diffraction (XRPD) and assumed it was due to a
test (Thickett and Lee 2004). Objects from anaerobic
microbiological attack, since a cladosporium fungus was
sites may contain elemental sulfur or unstable sulfides
found on a bronze. Later, this was refuted by Oddy and
(e.g. pyrite) and are known to cause efflorescence (Eggert
Meeks (1982) and Eggert and Sobottka-Braun (1999).
and Sobottka-Braun 1999). Some brands of plasticine are
Black spots have been observed on a wide range of objects formulated with elemental sulfur and may cause black
of different dating, provenance and material composition. spots when used as mounting material (Eggert 2006).
Weichert et al. (2004) and Eggert et al. (2004) reported
With the aim of reproducing the efflorescence in its
black spot efflorescence on copper alloys (including
varying morphologies, a systematic laboratory study
pewter with only 1% Cu), corrosion products, minerals
was designed (Kuhn-Wawrzinek 2020) to investigate the
and pigments. The sulfurous efflorescence consists of
influence of the parameters of time, RH, temperature,
either various crystalline sulfides and rarer sulfates as
copper alloy types, surface finish and minerals on the
oxidation products, or they are amorphous. Although
corrosion process. To ensure that the results from the
they were first discovered in conservation over a century
laboratory experiments could be translated into practice,
ago, no systematic investigations into the corrosion
the morphologies and chemical composition of labora-
process and the stability of the corrosion products
tory-grown corrosion were compared to original objects
have yet been undertaken, as it has not been possible
and samples. Additional experiments were designed to
to reproduce the corrosion products with analogous
investigate the stability and corrosivity of typical black-
morphologies in the laboratory so far.
spot corrosion products on copper to gain information
Thermodynamically, copper sulfides are not stable in on whether and how affected objects should be treated.
the presence of oxygen, therefore sulfates are preferred
(Gow et al. 2016, figs. 1–6). Copper sulfides can also Experimental
transform into other stoichiometries (Ramdohr 1980).
Experiments with metal coupons
Pewter samples (1%–10% Cu) were exposed to hydrogen
sulfide as part of the work of Weichert et al. (2004) and Metal coupons of pure copper, as well as the copper
initially showed only a little tarnish. After 12 years of alloys CuSn8 and CuZn10, with a 2 cm edge length were
storage in polyethylene (PE) bags without the presence exposed in desiccators in the presence of elemental sulfur
of an external sulfur source in our office in Stuttgart, at different temperatures and RH.
they were re-examined and showed significant amounts
To run more than 200 experiments with different param-
of brown efflorescence – a clear indication that black
eter combinations over considerable time (up to 60 weeks)
spots are not stable.
excluded the use of climatised gassing cabinets (e.g. using
Most previous publications present individual case hydrogen or carbonyl sulfide as a sulfur source). As the
studies and focus on the identification of the corrosion pollutant concentration is adjusted by the sublimation
196 ICOM-CC | METAL 2022 | HELSINKI, FINLAND ANALYTICAL INVESTIGATIONS
pressure of sulfur (1.38 ppb at 20 °C, 16.98 ppb at 40 °C), in direct contact to find out if the sulfides generated
it naturally resembles values that can also be expected in corrosive gases that affected the pure copper.
practice. For a slow corrosion process, no depletion of
All test series were run for 32 weeks and samples were
reactants in the chamber is to be expected. Gas diffusion
analysed after 1, 2, 4, 8, 16 and 32 weeks.
is fast enough and no constant gas stream with a definite
stream of pollutant is needed. Synthesised chalcocite and covellite as end products of
the series of sulfides were chosen for the study.
To adjust RH, saturated solutions of salts (analytical grade)
in beakers were used according to Greenspan (1977): To adjust RH in closed containers, saturated salt solutions
NaOH for 9%, MgCl2∙6H2O for 33%, Mg(NO3)2∙6H2O were used as described above, using NaCl for 75% RH.
for 54% and KNO3 for 95% at 20 °C (the values were
slightly lower for 40 °C). Investigation of museum samples
For experiments at 20 °C, the desiccators were placed in To check the relevance of laboratory results, they
a temperature-controlled laboratory. For experiments at were compared with black spots that had occurred
40 °C, the desiccators were placed in a heating cabinet. in collections. With public calls for samples and the
Temperatures as well as RH were measured with a data- support of many colleagues, it was possible to docu-
logger and found to be constant in a range of ± 2 °C and ment and sample more than 300 affected objects in
± 3%, respectively. eleven museums (Kuhn-Wawrzinek 2020) in Göttingen,
Metal coupons were lightly abraded using a glass bristle Frankfurt, Nuremberg, Cologne, Leipzig (Germany),
brush then cleaned with demineralised water followed Lund (Sweden), Samos (Greece), Zurich (Switzerland),
by acetone and air drying. Amsterdam (The Netherlands), Thorsminde (Denmark)
and Fremantle (Australia).
In one test series, surfaces of copper coupons were sprin-
kled with powders of malachite (Cu2(OH)2CO3), cuprite Analytical methods
(Cu2O) and tenorite (CuO) (10 mg each) to study their
stability and influence on copper corrosion. Scanning electron microscopy (SEM; Zeiss EVO 60) was
used to visualise the efflorescence with high magnifica-
A further series included copper coupons that were tion and depth resolution. Combined with energy-dis-
cleaned and heated before the experiment to form a persive X-ray analysis (SEM-EDX; Bruker X Flash 6130
uniform oxide layer on the surface to investigate its detector), elemental information could be obtained.
protective properties.
Micro-Raman spectra of efflorescence as well as single
All test series were run for 30 weeks. Additional series particles in powder samples were recorded using a
were run with coupons at 20 °C for 45 and 60 weeks due Renishaw inVia Raman spectrometer (grating 1800 L/mm,
to their slower corrosion rate. resolution 1 cm−1, He-Ne laser (632.8 nm)) equipped with
Blind tests were run for every series without the pres- a Leica DMLM microscope and a RenCam CCD detector.
ence of elemental sulfur. Test series were carried out in Sulfides low in copper (up to spionkopite, Table 1) contain
triplicate to check for reproducibility. disulfide units which can easily be detected by their
intensive stretching vibration in the Raman spectrum.
Stability of copper sulfides Sulfates as oxidation products can be recognised by the
typical S–O vibrations of the anion.
Powder samples of pure covellite, chalcocite and copper
were exposed in closed containers at 20 °C and at different Selected corrosion samples from objects and laboratory
RH levels of 33, 54, 75 and 95% to examine their stability experiments as well as the test series with powder samples
depending on RH. Additionally, 1:1 mixtures of covellite were analysed by XRPD (Stadi P diffractometer with PSD
with copper and chalcocite with copper were exposed detector, Mo Kα radiation with a Ge monochromator) for
to find out whether the sulfides had a corrosive effect their composition of crystalline phases. Beside covellite
on copper. Another experiment exposed covellite or (CuS) and chalcocite (Cu2S), a number of other copper
chalcocite to copper in the same container without being sulfides with Cu to S ratios in between have been differ-
entiated in the last decades as minerals (Table 1) and can

now be distinguished by precision XRPD.
Table 1. Copper sulfide minerals and their composition
Mineral name Formula Cu:S ratio

Chalcocite1 Cu2S 2.00:1
Djurleite 1
Cu31S16 1.94:1
Roxbyite Cu58S32 1.81:1
Digenite1 Cu9S5 1.80:1
Anilite 2
Cu7S4 1.75:1
Geerite Cu8S5 1.60:1
Spionkopite 2
Cu39S28 1.39:1
Yarrowite 1
Cu9S8 1.13:1
Covellite1 CuS 1.00:1 Figure 2.
1
Identified in the literature in black spots Spionkopite on
2
Newly identified in this study in black spots Cu at 9% RH and
40 °C after 30
weeks
Corrosion rates on metal coupons were qualitatively esti-
mated by a visual assessment of the amount of corrosion
product formed. Conversion in powder samples could
be estimated semi-quantitatively by comparing their
XRPD data.
Materials from collections that were suspected to be the

source of a sulfurous pollutant were tested using the
Oddy test.
Results and discussion of laboratory

experiments
Experiments with metal coupons

Figure 3.
RH is the decisive factor influencing which compounds
Yarrowite (and
form on copper as well as the tested alloys. possibly covellite)
on Cu at 33% RH
At lower RH, sulfides lower in copper were detected on and 40 °C after 30
copper coupons. weeks
At 40 °C, only spionkopite was detected at 9% RH

(Figure 2), yarrowite (and possibly covellite) at 33% RH At 40 °C and 95% RH, a thicker cuprite layer initially
(Figure 3) and both anilite and spionkopite at 54% RH formed and caused a less evenly distributed growth of
(Figure 4). the sulfurous efflorescence, which predominated along
Between 9 and 54% RH, the corrosion rate increased the edges of the coupons. Djurleite was the primary
slightly. Corrosion was quite evenly distributed over the product and formed a very fine light-brown bush-like
efflorescence which could easily be recognised as needles
surface but more prevalent at the edge of the coupon,
and dendrites in the SEM (Figure 5a, 5b). The finer the
which is an expected reaction with gas phase trans-
dendritic units, the lighter the macroscopic colour.
port of a reactant. Covellite, yarrowite and spionkopite
formed (blue-)black platelets. Anilite occurred in net-like Over time, djurleite transformed into tenorite (CuO) and
branched structures which appeared as brown to grey copper sulfides with disulfide bonds, as could be detected
efflorescence under low magnification. at the top of the thickening needles. This process could
bushes. Green and colourless particles could be differen-

tiated on the surfaces of the efflorescence by microscopic
investigation, due to sulfate-producing oxidation.
Experiments at 20 °C showed similar results, but the

vapour pressure of sulfur is lower, and the corrosion
rate is, of course, significantly lower. At 20 °C and 95%
RH, the transformation of djurleite to sulfates was not
observed within 30 weeks.
Bronze (8% Sn) and brass (10% Zn) showed similar

results to pure copper, with a slight tendency towards
sulfides lower in copper. The amount of corrosion prod-
ucts formed was lower and the crystal size was smaller.
Figure 4. Anilite They were less homogeneously distributed over the metal
and spionkopite surface and tended more towards net-like intergrowth. At
on Cu at 54% RH
and 40 °C after 30
95% RH, tiny green spots of brochantite (Cu4(OH)6(SO4))
weeks occurred on the brass. Oxidation of djurleite to sulfates
seemed to be faster on brass; the bushy efflorescence
turned mainly grey within 30 weeks at 40 °C. No corro-
sion products containing tin or zinc were detected.
Powders of malachite (Cu2(OH)2CO3), cuprite (Cu2O)

and tenorite (CuO) that were sprinkled on copper
coupons proved to be more stable than the metal itself
when exposed to sulfur vapours. Severe conditions (40 °C,
95% RH, 30 weeks) produced some brown discolouration
on the malachite and cuprite powders, which showed
tenorite as well as a disulfide vibration in the Raman
spectrum. The tenorite remained stable.
Figure 5. Heating a copper coupon prior to the experiment to

Corrosion on create a uniform oxidation layer of tenorite and cuprite
Cu at 95% RH
caused a very limited and spotty growth of the sulfurous
and 40 °C after
30 weeks: (a) efflorescence at 54% RH and both temperatures. Under
bushes along these conditions, the oxidation layer had protective
the edge of a properties allowing the sulfurous corrosion to form only
coupon, (b) light-
at defect points. At 95% RH and both temperatures, the
brown bushes of
djurleite, (c) dark- sulfurous efflorescence was also less evenly distributed
brown bushes than in experiments without prior oxidation of the
of djurleite coupons’ surface. However, the protective ability of the
with secondary
reaction to
oxidation layer was less effective at the higher RH.
tenorite and
All experiments showed good reproducibility. Three
sulfur-rich copper
sulfides at the morphological types of efflorescence could be differ-
needle tops entiated by macroscopic and microscopic examination.
The morphologically characteristic growth of copper
also be observed macroscopically and microscopically as sulfides could be traced back to motifs within their
a darkening of the efflorescence’s colour by the formation crystal structures: sulfides low in copper with disulfide
of black particles (Figure 5a, 5c). During further expo- units (covellite (CuS), yarrowite (Cu1.13S) and spionkopite
sure, this turned into an overall greyish aspect of the (Cu1.39S)) appeared as plate-like crystals of (blue-)black
colour. Covellite formed two-dimensional hexagonal

layers which favoured platelet growth. Anilite (Cu1.75S)
and digenite (Cu1.8S) formed branched net structures
which appeared as brown to grey efflorescence. In the
copper rich sulfides, djurleite (Cu1.94S) and chalcocite
(Cu2S), Cu-S layers were interrupted. The resulting
one-dimensional bands favoured needle and dendritic
growth. Such efflorescence occurred in very fine light-
brown bushes (needles and dendrites in the SEM).
Stability of copper sulfides

Figure 6. Roman
No changes could be detected on pure copper samples at fibula with dark-
all RH levels over 32 weeks. The presence of chalcocite brown bushes
and very dark to
or covellite without direct contact also had no impact on iridescent surface
the stability of the copper powder. due to secondary
reactions on
Only at 95% RH did covellite develop some sulfates copper-rich
(Raman: chalcanthite (Cu(SO4)∙5H2O) and antlerite copper sulfides:
(Cu3(OH)4SO4)). The conversion rate was very slow and (a) fibula,
(b) iridescent
the amount of sulfates was below the detection limit for area, (c) SEM
XRPD after 32 weeks. micrograph
showing bushes
Pure chalcocite reacted from 33% RH onwards to with a dense layer
djurleite (XRPD) and tenorite (Raman) which could be of secondary
observed in the formation of a dark tarnish on the particle reaction products
on the bushes’
surfaces. The reaction rate increased significantly with surface

increasing RH.
In mixtures, chalcocite had no effect on copper powder

was also identified as chalcocite or djurleite. Darker
when in direct contact at any RH level, whereas covel-
particles or layers on top of the bush-like formations
lite did react with copper powder. The reaction rate
were identified as tenorite and sulfides with disulfide
increased significantly with increasing RH, forming
units (Raman). In one corrosion sample, the copper
cuprite (Raman, XRPD) and brochantite (Raman).
sulfate antlerite was detected as well as djurleite (XRPD).
Investigation of museum samples Eighty-nine objects had blue-black efflorescence
Efflorescence found on museum objects can roughly be consisting of plate-like units (Figure 8). They consisted
divided into three categories similar to those found in of sulfur-rich sulfides (disulfide vibration in the Raman
the laboratory experiments. Sometimes different forms spectrum), as found in corrosion from laboratory exper-
are found side by side on the same object. iments at 9 and 33% RH. In one sample, antlerite was
detected, indicating oxidation to basic sulfate.
Among the objects studied, 205 exhibited brown, bush-
like efflorescence often intergrown to larger areas and Sixty-five objects had brown to grey efflorescence,
sometimes with (blue-)black or iridescent spots. Light- showing quite coarse dendritic to net-like structures
brown bushes consisted of very fine dendritic units, under the microscope. They showed no vibrations in
while darker-brown bushes had coarser particles at the the Raman spectrum, while XRPD pointed to spionko-
top. In very dark, black or iridescent areas, a continuous pite and digenite. Fine plate-like units in between the
layer had formed on the bushes’ surface (Figure 6). The dendrites consisted of disulfides (Figure 9). All in all, this
light-brown efflorescence looked similar to that found corrosion was similar to the products formed at 54% RH
in laboratory experiments at 95% RH (Figure 7) and in the laboratory.
Figure 9. Grey
efflorescence on
a candelabrum
with dendritic-
branched
structure: (a) grey
Figure 7. Light-
efflorescence, (b)
brown bushes
SEM micrograph
on a metal
fragment with
of efflorescence
characteristic
dendritic
structures: (a)
detail, (b) SEM growth of efflorescence found on museum objects.
micrograph of The systematic influence of the RH, temperature, alloy
light-brown composition and surface condition parameters on the
bush, (c) SEM
micrograph of
corrosion process could be proven.
dendritic units in
RH was the decisive factor in growing the different
light-brown bush
morphologies of brown, grey or blue-black efflorescence
which consist of different compounds with different
Cu:S ratios.
The initially formed sulfides can undergo secondary

reactions and thereby promote further corrosion of the
metal itself, especially at high RH.
Based on the results, guidance for best practice is as

follows:
• Remove copper sulfides from objects whenever
possible.
• Store copper-containing objects at low temperature
Figure 8. Blue-black efflorescence at cracks in the corrosion layer of a and low RH.
bronze sword • Test storage and display materials for sulfur emissions
(Oddy or iodine-azide test) and look for other sources
Conclusion and outlook of sulfur.
The sublimation pressure of up to 17 ppb at 40 °C was Despite extensive experimental effort, open questions
sufficient to grow significant amounts of sulfur-induced still remain. The methodology developed in this work
efflorescence on copper-containing alloys and copper can now be used to study further issues related to the
in 30–60 weeks, making it possible to simulate the slow black-spot corrosion process.
More research is needed on remedial conservation like Canberra, 4–8 October 2004, eds. J. Ashton and D. Hallam,
the protective effect of lacquer coatings, which need to 142–8. Canberra: National Museum of Australia.
be flawless or spots can often grow in the defects.
Gow, R.N., H. Huang, and C. Young. 2016. Utility of
The authors came across cases where silver rich in mass-balanced EH-pH diagrams I – Applications of
copper formed mixed silver-copper sulfides with similar Gibbs’ phase rule. Minerals & Metallurgical Processing
dendritic morphologies. But what about other alloys such 33(2): 58–67.
as lead, a common compound in ancient copper alloys? Greenspan, L. 1977. Humidity fixed points of binary
And what is the nature of the amorphous material in the saturated aqueous solutions. Journal of Research of the
spots which cannot be identified by XRPD? In the Raman
National Bureau of Standards – A. Physics and Chemistry
spectra, no evidence was found for any intermediate
81A(1): 89–96.
S–O compound between sulfide (S2−) and sulfate (SO42−).
Much more sophisticated analytical methods will be Kirchner, N. 1973. Probleme der Konservierung und
needed to study the oxidation state of sulfur (e.g. X-ray Aufbewahrung von Benin-Bronzen. Arbeitsblätter für
absorption near-edge spectroscopy (XANES)). Restauratoren 6(1), Gr. 2: 40–8.
While black spots and brown fuzzies are certainly infre- Kuhn-Wawrzinek, C. 2020. Korrosion von Kulturgut
quent, they are present today in museums around the aus Kupferwerkstoffen in Gegenwart von elemen-
world. Conservators must be aware of them so that the tarem Schwefel in der Gasphase. PhD dissertation,
‘enemy of copper’ can be identified and prevented. Staatliche Akademie der Bildenden Künste Stuttgart,
Germany. http://archiv.ub.uni-heidelberg.de/artdok/
Acknowledgements volltexte/2020/7019
The authors would like to thank Baden-Württemberg Madsen, H.B. 1977. Mikrobiologisk angreb på bronzerne
for providing Charlotte F. Kuhn-Wawrzinek with a fra Budsene brønden. Meddelelser om Konservering 2(9):
graduate stipend, Christoph Krekel for providing access 265–70.
to Raman and SEM-EDX, Rainer Niewa for providing Madsen, H.B. 1985. Schwarze Fleckbildung auf Bronzen.
access to XRPD and Uwe Peltz for indicating useful In Archäologische Bronzen. Antike Kunst – Moderne
earlier research. Technik, ed. H. Born, 104–9. Berlin: Dietrich Reimer.
Monier-Williams, M. 1899. A Sanskrit-English dictionary,
References
new ed. (repr. 1960). Oxford: Clarendon Press.
Böhtlingk, O. and R. Roth. 1875. Sanskrit-Wörterbuch,
Peltz, U. 2021. Aes Aeternum. Die Restaurierungsgeschichte
vol. VII. St. Petersburg: Kaiserliche Akademie der
der Bronzen aus Berlins Antikensammlung. Mainz:
Wissenschaften, col. 269.
RGZM.
Eggert, G. 2006. Plastiline: Another unsuspected danger
in display causing black spots on bronzes. Beiträge zur Ramdohr, P. 1980. The ore minerals and their intergrowths,
Erhaltung von Kunst- und Kulturgut 2/2006: 112–16. 2nd ed. Oxford: Pergamon Press.
Eggert, G. 2021. Copper and Bronze in Art and the search Thickett, D. and L.R. Lee. 2004. Selection of materials
for rare corrosion products. Preprints, 2021030523. for the storage or display of museum objects, Occasional
https://www.preprints.org/manuscript/202103.0523/v1 Paper 111, 2nd rev. ed. London: The British Museum.
Eggert, G. and U. Sobottka-Braun. 1999. Black spots on Vira, R. 1950. The consolidated great English-Indian
bronzes and elemental sulphur. In ICOM-CC 12th Trien- dictionary of technical terms. Nagpur: International
nial Meeting Preprints, Lyon, 29 August–3 September 1999, Academy of Indian Culture.
ed. J. Bridgland, 823–7. London: Earthscan Ltd. Available Weichert, M., G. Eggert, A.M. Jones, and H.A. Anker-
at https://www.icom-cc-publications-online.org/ smit. 2004. Trees, bunches, cauliflowers – A closer look
Eggert, G., M. Weichert, H. Euler, and B. Barbier. 2004. at sulphurous corrosion on copper alloys and minerals
Some news about ‘Black Spots’. In Metal 04: Proceedings (‘black spots’). In Metal 04: Proceedings of the Interna-
of the International Conference on Metals Conservation, tional Conference on Metals Conservation, Canberra, 4–8
October 2004, eds. J. Ashton and D. Hallam, 149–59.

Canberra: National Museum of Australia.
Authors
Charlotte F. Kuhn-Wawrzinek has a degree in objects’
conservation (2010) and a PhD (2020) from the Stuttgart
State Academy of Art and Design. She has worked as a
conservator, conservation scientist and lecturer focusing
on the preservation and corrosion of metal artefacts,
historic pigments and chemistry in conservation. Since
2016, she has worked in various strategic roles in the IT
industry.
Gerhard Eggert has a degree in chemistry (1982)

and a PhD (1986) from the University of Bonn, where
he conducted experimental work in electrochemistry.
Between 1985 and 1998 he was head of the Conserva-
tion Department at Rheinisches Landesmuseum Bonn,
dealing mainly with archaeological finds. From 1998
to 2019 he was Chair of Objects’ Conservation at the
Stuttgart State Academy of Art and Design. His research
interests include the manufacture, degradation and
conservation of inorganic objects.
Thomas Schleid has a degree in chemistry (1984) and

a PhD (1988) from the Justus Liebig University, Gießen,
where he conducted experimental work in inorganic
chemistry. In 1993, he obtained a postdoctoral quali-
fication in inorganic chemistry from the University of
Hanover. Since 1995, he has been Chair of Inorganic
Solid State Chemistry, University of Stuttgart and director
of the Institute of Inorganic Chemistry, with research
interests in multi-anionic solid-state rare-earth metal
compounds.
The Role of Patina on Archaeological Copper Alloy Coins in
the Outbreak and Progression of Bronze Disease
Johanna C. Thunberg* Nicola J. Emmerson Ildikó Harsányi Mark Lewis

Cardiff University Cardiff University Centre for Energy Research National Roman Legion Museum
Cardiff, UK Cardiff, UK Budapest, Hungary Caerleon, UK
[email protected] [email protected] [email protected] [email protected]
David E. Watkinson Zoltan Kis Zsolt Kasztovszky *Author for correspondence

Cardiff University Centre for Energy Research Centre for Energy Research
Cardiff, UK Budapest, Hungary Budapest, Hungary
Abstract
Archaeological copper alloy objects form unique corrosion transformation imaging and visually using descriptors reflecting
structures according to their manufacture and burial Type I, Type II and ‘bronze disease’ surfaces. When subjected
environment. Categorisation has identified protective Type I to high relative humidity, none of the coins exhibited bronze
structures and chloride-containing Type II structures which may disease despite containing chlorides and showing characteristics
lead to bronze disease. Using visual examination to determine of Type II structures and bronze disease. The results highlight the
which of these exist on objects dictates conservation decisions; difficulty of determining risk and taking informed decisions for
unfortunately, there is limited evidence linking surface features the management of archaeological copper alloy objects.
directly to corrosion risk. In this study, the chloride content
of 39 archaeological copper alloy coins was determined non- Keywords
destructively using prompt gamma-ray activation analysis. Their bronze disease, corrosion monitoring, management, decision-
surfaces were examined through Raman spectroscopy, reflective making, conservation, characterisation
Introduction
Archaeological copper alloy objects often occur in large The rapid and voluminous growth of Cu2(OH)3Cl can
numbers in heritage collections. The primary concern disrupt overlying corrosion layers, destroying surface
with these objects is the onset of bronze disease. This details and forming a disfiguring growth on the surface.
is an active, chloride-driven corrosion process that can The ability of the process to continuously form CuCl (4)
result in the complete destruction of an object (Scott means that the destruction will continue until the supply
2002). Bronze disease develops from cuprous chloride of oxygen, atmospheric moisture or copper is supressed
(CuCl), which forms adjacent to the metal core during (MacLeod 1981).
burial when chloride ions interact with cuprous ions (1)
(Wang et al. 2006): 2HCl + 2Cu → 2CuCl + H2(4)
Cu+ + Cl− → CuCl + e− (1) As a result of the significant risks to objects suffering
from bronze disease, conservation efforts are focused on
If pH < 5, cuprous oxide (Cu2O) can also form CuCl
preventing post-excavation processes from occurring.
through (2) (Grayburn et al. 2015):
Cu2O + 2H+ + 2Cl ←→ 2CuCl + H2O(2) Management of archaeological copper alloys
In a post-excavation environment, CuCl can interact The tools to evaluate the risk of bronze disease within
with atmospheric oxygen and moisture to form copper practical or ethical remits are limited. Thorough inves-
trihydroxychloride isomers (Cu2(OH)3Cl) (3) (Dowsett tigation of the burial and post-excavation environments,
et al. 2012): and the nature of the alloy, manufacturing processes
4CuCl + O2 + 4H2O → 2Cu2(OH)3Cl + 2H+ + Cl−(3) and corrosion stratigraphy can inform conservation
strategies, but this information is rarely available. The Cosano et al. 2018). Table 1 demonstrates the descriptors
location of CuCl next to the metal surface means that most frequently encountered in literature to describe
it cannot be identified without destructive sampling. Type I and Type II patinas, alongside descriptors used
Due to the complex formation of corrosion products, to identify bronze disease (Table 1). Several descriptors
surface analysis offers limited insight unless Cu2(OH)3Cl to identify bronze disease overlap those with a Type II
is detected, yet at this point the object is likely to have patina, indicating that these may be linked.
already incurred damage. In practice, visual inspection is
While the terminology around archaeological copper
often employed for assessing the condition of copper alloy
alloys is well established, the link between the descrip-
objects. The success of visual inspection is dependent on
tors in Table 1 and the stability of an object has not been
factors such as understanding of corrosion processes,
investigated in detail. When copper alloy objects in
experience, colour perception and method of assessment.
the Burrell collection were examined in a recent study,
Guidance to identify corrosion risk can standardise and
only 6 of the 64 objects suspected to suffer from bronze
improve the accuracy of assessments.
disease after visual inspection were shown to be actively
Archaeological copper alloys display complex corro- corroding (Bryan 2021). This suggests that the accuracy
sion profiles as a result of extrinsic factors, such as the of descriptors currently used in practice to identify
chemo-physical properties of the burial environment, and objects affected by bronze disease is low.
intrinsic factors, such as the nature of the alloy and manu-
As object appearance informs conservation deci-
facturing processes (Ingo et al. 2019, Leygraf et al. 2019).
sion-making, it is critical to have a good understanding
The appearance of copper alloy objects can therefore differ
of the information that can be determined visually from
widely. Despite this, it has been suggested that copper alloys
corroded surfaces. This pilot-study explores the link
can fall within two types of corrosion structures that create
between existing vocabulary to describe copper alloy
distinguishable features, commonly recognised as Type I
objects and the outbreak of bronze disease.
and Type II structures (Robbiola et al. 1998). A literature
review of conservation and archaeometallurgical publica-
Aim and objectives
tions identified over 50 sources applying Type I or Type II
structures to characterise objects, suggesting they are often This study aims to investigate the relationship between
used for this reason. A Type I structure has a two-layered the appearance of archaeological copper alloy objects,
profile that replicates the metal surface (‘original surface’) the presence of chlorides and the risk of bronze disease.
without a significant change in volume. A Type II structure This is achieved by:
is defined by a porous, three-layer profile rich in chloride • characterising the surfaces of 39 Roman archaeological
ions next to the metal and containing no detail from the copper alloy coins by visual and instrumental analysis;
original surface. As a result of their corrosion history, • establishing the chloride content of the coins by
Type I and Type II structures could display features that prompt gamma-ray activation analysis (PGAA);
are visually distinguishable by their textures and colours. • exposing the coins to high humidity environments
Being able to separate these surfaces by appearance is understood to trigger bronze disease and monitoring
potentially important for conservation practice, as a Type I corrosion using reflectance transformation imaging
structure is believed to have protective properties, while (RTI) and oxygen consumption.
a Type II structure is thought unstable due to its chloride
content and by allowing ion transport (Payer et al. 1995, Method
Robbiola et al. 1998).
In this study, 39 archaeological copper alloy coins
Examination of guidelines and literature reveals the (IP01–IP40) were examined (Figure 1). These were
existence of jargon in conservation practice to describe metal detectorist finds donated to Cardiff University
archaeological copper alloy surfaces and bronze disease, Conservation Department for research. The coins varied
with many terms overlapping or being closely connected between 6 and 27 mm in diameter and weighed between
to the terminology used by Robbiola et al. (1998) to 0.88 and 4.96 g. X-radiography demonstrated that all
define Type I and Type II patinas (e.g. Wadsak et al. of the coins had a metallic core. The detection site and
2000, Constantinides et al. 2002, Bozzini et al. 2017, context are unknown, and the coins are therefore deemed
THE ROLE OF PATINA ON ARCHAEOLOGICAL COPPER ALLOY COINS IN THE OUTBREAK AND PROGRESSION OF BRONZE DISEASE 205
Table 1. Example of descriptors used in literature to characterise archaeological copper alloys
Type I Type II Bronze disease

Smooth Powdery Powdery
Even Buboes Dusty
Uniform/monochrome Warts Patches
Shiny Limpets Flakes
Lustrous Coarse/rough Blisters
Hard Pits Warts
Dense Crevices Pustules
Pleasant effect Crust Mounds
Coherent Loss of material (incl. spalling, breakup of overlying layers) Loss of material
Adherent Local outbreaks/attack Voluminous
Compact Patches Local outbreaks/attack
Texture
Detail from original surface Polychrome Spots

Spots Cracking
Cracking Non-adhering, loose
Soft Pits
Non-adhering, loose
Underweight
Lamellar plates
Flakes/delamination
Holes
Porous
Hard to nearly hard
No detail from original surface
Green Green Green
Red Light green Blue-green
Brown Light blue Pale green
Colour
Grey Bright green Bright green

Blue Blue-green Light green
Black
Red
Brown
Figure 1. Images of all coins

examined. Images are not
to scale. Length of diameter
at widest point is given in
mm in parentheses
of low archaeological value. The coins were mechanically active corrosion for 37 days. A 100% relative humidity
cleaned to remove soil products. (RH) environment was created by enclosing three beakers
with cotton wool saturated with water (total 450 mL) in
Establishing chloride content a Stewart Gastronorm 13l polypropylene box.
The bulk chloride content of the coins was determined The oxygen consumption of five coins containing the
non-destructively using PGAA at Budapest Neutron highest chloride content (IP05, IP11, IP39, IP20 and
Centre, Hungary. PGAA can detect bulk chloride through IP40) was measured individually to produce a quanti-
irradiation from a cold-neutron beam which detects tative measure of their corrosion rates. The method has
characteristic gamma-photons emitted in (n, γ) reac- been reported in detail in an open access publication
tions. The neutrons can penetrate several centimetres (Emmerson et al. 2021).
into a material, facilitating determination of the bulk
A 250 mL Mason Ball jar was used to create an airtight
concentration of elements in small objects. The distribu-
environment for oxygen measurements containing
tion of chlorides was mapped in three coins (IP05, IP39
a two-point PreSens GmbH SP-PSt3-NAU-D5-YOP
and IP40) using the NIPS-NORMA (neutron-induced
ruthenium oxygen sensor spot adhered to the interior
prompt gamma-ray spectroscopy and neutron optics and
wall using Radio Spares RTV silicone rubber compound.
radiography for material analysis) station.
Every vessel was assigned a Madgetech RHTemp101A
datalogger (± 1.1% RH at 80% RH) to monitor the
Surface examination
internal RH. Silica gel (160 g) conditioned to 80% RH
All coins were recorded using a Broncolor Scope D50 was enclosed in the vessel to maintain a high humidity
RTI dome equipped with 48 white (5000 K ± 75 K) LED environment for the test duration. A relative humidity
lights. Images were processed using Truvis Authentica of 80% RH was chosen as a suitable environment to
software. Images were captured with a Nikon D5600 with detect activity related to bronze disease as this is known
a Nikon AF-S Micro-Nikkor 105 mm lens. to produce rapid corrosion rates (Thickett 2016). A
PreSens GmbH POF-L2.5-1SMA 505 nm fibre sensor
Raman spectroscopy was employed to characterise
connected to a Fibox 4 oxygen meter was used as the
surface corrosion products with a B&W Tek iRaman
luminescence source. The coins were monitored for their
Plus equipped with a 785 nm (resolution ~3.5 cm−1
oxygen consumption for 44 days.
max 30 mW) and 532 nm (resolution ~4.5 cm−1 max
340 mW) probe at ×50 magnification (laser spot size To investigate the protective properties of corrosion
42 µm). Integration time ranged from 2 to 60 sec and 1 structures, coins IP05, IP11, IP20, IP39 and IP40 were
to 100 acquisitions. Laser intensity ranged from 1 to 5% thereafter scored through the corrosion profile and down
of maximum laser output. Twenty spectra of each coin to the metal surface using a scalpel, revealing the full
were taken at sites of different colours and textures to corrosion profile of the coin and potentially exposing
account for the surface heterogeneity of the coins. chloride sites (Figure 2). Their oxygen consumption was
recorded for 140 days. Detailed mapping of chlorides
Three trained conservators with practical experience
using NIPS-NORMA on IP05, IP11 and IP39 determined
of working with archaeological copper alloys examined
the location of scoring down the middle of the obverse
the surface of each coin at ×4 magnification under a
and reverse of the coins (Figure 2). All coins were visually
Just Normlicht XL colour viewing light with a D65 LED
assessed at the end of oxygen consumption to control
light source (colour temperature of 6500 K) at a 70 cm
for any changes.
working distance. The descriptors in Table 1 were used
to characterise the surface of each coin during exami-
Results
nation. All descriptors that were applicable to describe
the surface of the coins were selected. PGAA/NIPS-NORMA analysis
The chloride content in all of the coins is expressed as a
Corrosion monitoring
ratio of chloride to copper in elemental weight. All of the
Coins IP01–IP40 were subjected to a saturated humidity coins (IP01–IP40) contained chloride, demonstrating a
environment and monitored for any physical signs of copper to chloride ratio between 2.65 × 10−4 and 1.46 × 10−2
(Figure 3). Neutron imaging of IP05, IP11 and IP39

demonstrated that chloride was localised within the coin.
Surface examination
During examination, different descriptors were
employed by each observer to characterise a coin. No
two coins had identical sets of descriptors, and descrip-
tors for Type I, Type II and bronze disease were used
to describe each coin (Figure 3). The five coins with
the highest chloride/copper ratio (IP05, IP11, IP20,
IP39, IP40) and those with the lowest chloride/copper
ratio (IP01, IP24, IP25, IP30, IP36) were characterised
by many overlapping descriptors, with no apparent
distinguishing features to suggest one set would contain
more chlorides (Figure 4).
Figure 2. IP05, IP11, IP20, IP39 and IP40 after scoring on the obverse
and reverse Eighteen coins displayed some preservation of the orig-
inal surface detail. These were both within the range of
the highest (IP05, IP20, IP39, IP40) and lowest (IP01,
IP25) chloride-containing coins, demonstrating no
Figure 3. The number of Type I patina, Type II patina and bronze disease Figure 4. Descriptors used for the coins containing the lowest chloride
characteristics identified in each coin by at least one of the three assessors. content (IP01, IP24, IP25, IP30 and IP36) and highest chloride content
The chloride/copper ratio of each coin as measured by PGAA is included (IP05, IP11, IP20, IP39 and IP40)
obvious relationship between the preservation of the Corrosion monitoring

original surface detail and chloride content (Figure 3).
None of the coins showed signs of active corrosion after
Some descriptors were used consistently throughout exposure to 100% RH. Coins IP05, IP11, IP39, IP20 and
the dataset, whereas some were never used to describe IP40 did not consume any oxygen during exposure to
surface features (Table 2). 80% RH. Scoring the coins to damage the patina did not
affect their oxygen consumption rate, and the coins did
Table 2. Descriptors applied to all coin surfaces and descriptors not
not develop any visible signs of bronze disease.
used to characterise coin surfaces
Descriptors used for all coins Descriptors not used Discussion

Hard Porous
Dense Dusty Establishing the presence of chloride cannot indicate
Coherent Warts whether bronze disease will occur. None of the coins
Adherent Pustules investigated developed new signs of active corrosion
Buboes when subjected to environments known to promote the
Limpets process, despite a known chloride content. Compro-
Holes mising the patina and exposing chloride sites in five of
Pleasant effect the coins with the highest chloride content (IP05, IP11,
IP39 and IP40) did not result in the breakout of bronze
Detection of corrosion compounds using Raman spec-
disease. This makes it unlikely that any CuCl was exposed
troscopy was difficult, probably due to poor crystallinity
during the study.
of the corrosion products and significant fluorescence
masking the Raman signal. Raman detected the presence It is possible that the chlorides mapped by PGAA and
of malachite (Cu2CO3(OH)2) and cerussite (PbCO3) NIPS-NORMA are not present as CuCl but located
(Figure 5). Soil components included quartz, calcite, elsewhere in the corrosion products. Copper trihy-
feldspar and the TiO2 isomers rutile and anatase. droxychlorides (Cu2(OH)3Cl) may already be present
in the patina, and other copper chloride compounds,
such as connellite (Cu 19 Cl 4 (SO 4 )(OH) 32 ∙3(H 2 O)),
sampleite (NaCaCu5(PO4)4Cl∙5(H2O)) and calumetite
(Cu(OH, Cl)2∙2(H2O)), have been found on archaeological
copper alloys (Scott 2002, Giovanelli et al. 2010, Muros
and Scott 2018). Other alloying chloride compounds have
been identified, such as mendipite (Pb3Cl2O2), pyromor-
phite (Pb(PO4)3Cl) and laurionite (PbCl(OH)) in leaded
copper alloys (Ingo et al. 2006, Bernard and Joiret 2009,
Mezzi et al. 2012) and AgCl in silvered coins (Fabrizi
et al. 2019). None of these compounds were identified
using Raman spectroscopy, but due to the heterogeneity
of the surface, poor crystallinity of compounds and the
limit of laser penetration, these compounds could still be
present. Chlorides could also be present as a soil product
and integrated on the surface. A second, follow-up study
will examine the coins in more detail to locate chlorides
throughout the corrosion profiles.
Surface examination
The surface examination of each coin using descriptors
Figure 5. Raman spectra of (a) malachite (Cu2CO3(OH)2) and (b) cerussite highlights the subjectivity of visual examination and the
(PbCO3) difficulty of standardising vocabulary to guide object
inspection. Several challenges occurred during exam- Implications for conservation practice
ination resulting in an unclear relationship between
Identifying chloride in archaeological copper alloy corro-
surface features and chloride content, further limiting
sion structures is not a determining factor in identifying
the ability to correlate the appearance of objects to risk.
risk of bronze disease. Consequently, visual examination
These challenges included:
and identification of surface features that are suggestive
• Interpretation of key descriptors of a chloride-containing corrosion structure may not
A range of descriptors were not used by assessors due be a reliable indicator that bronze disease will occur.
to the lack of references illustrating the descriptors, Complexity of surfaces, ambiguity in interpretation of
making them difficult to interpret. Others lacked a descriptors and the occurrence of multiple features on
contrasting reference point, such as ‘soft’ or ‘under- the surface of one object will further complicate deci-
weight’. The challenge of using descriptors to char- sion-making.
acterise a surface is demonstrated in the range of
Some key features that were not used by the assessors in
terminology used by the assessors to characterise the
this study could be indicators of bronze disease (Table 2)
same objects.
but may only provide retrospective evidence for ongoing
• Applicability of descriptors for each object active corrosion when damage has already occurred,
Interpretation of object surfaces is further compli- rather than detect risk. As the outbreak of bronze disease
cated by the presence of Type I, Type II and bronze may not occur immediately after excavation, active
disease descriptors on all examined surfaces. While bronze disease can proceed unhindered unless objects are
this reflects the heterogeneity of archaeological copper monitored regularly. Conversely, the cautious approach
alloy surfaces, it suggests that there is a limit to the of bulk treatment or environmental control of objects
usefulness of these descriptors for determining risk. (possibly unnecessarily) to mitigate the risk of bronze
This is illustrated in Figure 6, which shows the simi- disease can be costly in time, materials and potential
larity of a stable surface (IP03) and the surface of a exposure to harmful materials and may also result in
coin from a different sample set on which active bronze unwanted surface alterations.
disease has been recorded.
Conclusion
• This study highlights the limitations of using visual
examination and surface analysis to determine whether
an object is at risk of developing bronze disease, which
normally dictates management policy and treatment
decisions.
• Using written guidance for visual analysis is complex,
and it is difficult to determine the success of using
descriptors.
• Some terminology may be more useful for determining
Figure 6. IP03 (∅ 10.6 mm) and an actively corroding coin (∅ 25.9 mm) the risk of active bronze disease, but a more precise
from outside this sample set showing similarity of surface features
language is necessary to determine risk.
• Non-destructive analysis techniques to determine the
• Insufficient terminology to reflect the complexity of presence of chlorides are of limited use in assigning
some surfaces risk, as they do not serve as an indicator for bronze
Despite the range of terminology used to distinguish disease.
Type I, Type II and bronze disease surfaces, the • The cost-benefit rationale for treatment or imple-
descriptors did not reflect the complexity of each coin. menting environmental control is affected when
In particular, colour may not be a reliable indicator corrosion risk cannot be evidenced.
owing to the colourant properties of Cu(II), which • Ongoing monitoring to detect changes in copper
may produce a range of products similar in colour alloys on display or in storage is an important but
(Leygraf et al. 2019), as illustrated in Figure 6. often monumental task.
The sample set of coins are of similar size, dimension Fabrizi, L., F. Di Turo, L. Medeghini, M. Di Fazio, F.
and corrosion thicknesses and cannot provide a complete Catalli, and C. De Vito. 2019. The application of non-de-
picture of the different corrosion surfaces associated structive techniques for the study of corrosion patinas of
with bronze disease. A further larger-scale study will ten Roman silver coins: The case of the medieval Grosso
include objects of different shapes and contexts to gain Romanino. Microchemical Journal 145: 419–27.
a more complete understanding of appearance/bronze Giovanelli, G., L. D’Urzo, G. Maggiulli, S. Natali, C.
disease relationships. A follow-up study is underway to Pagliara, I. Sgura, and B. Bozzini. 2010. Cathodic chlo-
determine the chloride location throughout the corrosion ride extraction treatment of a late Bronze Age artifact
profiles of the coins. affected by bronze disease in room-temperature ionic-
liquid 1-ethyl-3-methylimidazolium bis(trifluorometh-
Acknowledgements anesylfonyl)imide (EMI-TFSI). Journal of Solid State
The authors wish to thank the AHRC-funded SWW DTP Electrochemistry 14: 479–94.
consortium and IPERION CH for supporting this project. Grayburn, R., M. Dowsett, M. Hand, P.-I. Sabbe, P.
Thompson, and A. Adriaens. 2015. Tracking the progres-
References sion of bronze disease. Corrosion Science 91: 220–3.
Bernard, M.C. and S. Joiret. 2009. Understanding corro- Ingo, G.M., T. De Caro, C. Ruccucci, E. Angelini, S.
sion of ancient metals for the conservation of cultural Grassini, S. Balbi, P. Bernardini, D. Salvi, L. Bousselmi,
heritage. Electrochimica Acta 54: 5199–205. A. Cilingiroglu, M. Gener, V.K. Gouda, O. Al Jarrah, S.
Bozzini, B., B. Alemán, M. Amati, M. Boniardi, V. Khosroff, Z. Mahdjoub, Z. Al Saad, W. El-Saddik, and
Caramia, G. Giovannelli, L. Gregoratti, and M. Kazemian P. Vassilou. 2006. Large scale investigation of chemical
Abyaneh. 2017. Novel insight into bronze disease gained composition, structure and corrosion mechanism of
by synchrotron-based photoelectron spectro-microscopy, bronze archaeological artefacts from Mediterranean
in support of electrochemical treatment strategies. Studies basin. Applied Physics A 83: 513–20.
in Conservation 62(8): 465–73. Ingo, G.M., C. Riccucci, G. Guida, M. Albini, C. Giuliani,
Bryan, B.N. 2021. Copper alloy objects suspected of and G. Di Carlo. 2019. Rebuilding of the burial environ-
bronze disease – The Burrell ‘Bronzes’. In Studies in ment from the chemical biography of archaeological
archaeological conservation, eds. C. Caple and V. Garlick, copper-based artifacts. ACS Omega 4: 11103–11.
115–23. London: Routledge. Leygraf, C., T. Chang, G. Herting, and I. Odnevall Wall-
Constantinides, I., A. Adriaens, and F. Adams. 2002. inder. 2019. The origin and evolution of copper patina
Surface characterization of artificial corrosion layers on colour. Corrosion Science 157: 337–46.
copper alloy reference materials. Applied Surface Science Macleod, I. 1981. Bronze disease: An electrochemical
189: 90–101. explanation. ICCM Bulletin 7(1): 16–26.
Cosano, D., D. Esquivel, L.D. Mateos, F. Quesada, C. Mezzi, A., E. Angelini, C. Riccucci, S. Grassini, T. De
Jiménez-Sanchidrián, and J.R. Ruiz. 2018. Spectroscopic Caro, F. Faraldo, and P. Bernardini. 2012. Micro-struc-
analysis of corrosion products in a bronze cauldron from tural and micro-chemical composition of bronze artefacts
the Late Iberian Iron Age. Spectrochimica Acta Part A: from Tharros (Western Sardinia, Italy). Surface and
Molecular and Biomolecular Spectroscopy 205: 489–96. Interface Analysis 44: 958–62.
Dowsett, M., A. Adriaens, C. Martin, and L. Bouchenoire. Muros, V. and D. Scott. 2018. The occurrence of
2012. The use of synchrotron X-rays to observe copper brochantite on archaeological bronzes: A case study
corrosion in real time. Analytical Chemistry 84: 4866–72. from Lofkend, Albania. Studies in Conservation 63(2):
E m m e r s o n , N . J . , J . H .  S e i f e r t , a n d  D . E . 113–25.
Watkinson. 2021. Refining the use of oxygen consump- Payer, J.H., G. Ball, B.I. Rickett, and H.S. Kim. 1995.
tion as a proxy corrosion rate measure for archaeological Role of transport properties in corrosion product
and historic iron. European Physical Journal Plus 135: growth. Materials Science and Engineering: A 198 (1–2):
art. 546. 91–102.
Robbiola, L., J.-M. Blengino, and C. Fiaud. 1998. Zsolt Kasztovszky is a senior fellow scientist at the
Morphology and mechanisms of formation of natural Centre for Energy Research, focusing on nuclear chem-
patinas on archaeological Cu-Sn alloys. Corrosion istry and applications of PGAA in archaeometry and
Science 40(12): 2091–111. material sciences.
Scott, D.A. 2002. Copper and bronze in art: Corrosion, Mark Lewis is a senior curator at the National Roman
colorants, conservation. Los Angeles, CA: The Getty Legion Museum, National Museum Wales and an expert
Conservation Institute. in Roman archaeology.
Thickett, D. 2016. Critical relative humidity levels and
carbonyl pollution concentrations for archaeological
copper alloys. In Metal 2016: Proceedings of the Interim
Meeting of the ICOM-CC Metals Working Group, 26–30
September 2016, eds. R. Menon, C. Chemello, and A.
Pandya, 180–7. New Delhi: International Council of
Museums-Committee for Conservation (ICOM-CC) and
Indira Gandhi National Centre for the Arts (IGNCA).
Wadsak, M., I. Constantinides, G. Vittiglio, A. Adriaens,
K. Janssens, M. Schreiner, F.C. Adams, P. Brunella, and M.
Wuttmann. 2000. Multianalytical study of patina formed
on archaeological metal objects from Bliesbruck-Rein-
heim. Microchimica Acta 133: 159–64.
Wang, J., C. Xu, and G. Lv. 2006. Formation processes
of CuCl and regenerated Cu crystals on bronze surfaces
in neutral and acidic media. Applied Surface Science
252(18): 6294–303.
Authors
Johanna C. Thunberg teaches conservation and is
undertaking a PhD at Cardiff University. Her research
focuses on storage protocols for archaeological metals
and identification of instability in copper alloy artefacts.
David E. Watkinson teaches and researches conser-

vation at Cardiff University. His work on the corrosion
rate of iron underpins the preservation of Brunel’s iron
steamship SS Great Britain.
Nicola J. Emmerson lectures in conservation science

at Cardiff University and researches the management
of corrosion prevention strategies for metallic cultural
heritage artefacts.
Zoltan Kis is a senior fellow scientist at the Centre for

Energy Research, specialising in gamma spectroscopy,
PGAA and imaging.
Ildikó Harsányi is a researcher at the Centre for Energy

Research. She has published extensively on the use of
PGAA to examine archaeological materials.
Non-Invasive Analysis: The Mirage and the Reality
Peter Northover
University of Oxford (retired)
Abingdon, Oxfordshire, UK
[email protected]
Abstract
This paper aims to seek a balance between non-invasive and from metallographic samples. In the final example, non-
sample-based analyses. To do so, it presents three case studies: invasive analysis would have completed the picture but was
a medieval sculpture, the Pisa Griffin; copper fastenings from not practical at the time. The two approaches should be explored
wooden warships from the time of the Napoleonic Wars; and together and an appropriate analysis strategy reached without
a study of the bronze artefacts from a Late Bronze Age village preconceptions.
in England that had been destroyed by fire. In the first case,
a correct understanding of the Griffin could not be obtained Keywords
without taking metal samples, and for the copper it was non- non-invasive analysis, macrostructure, microstructure,
invasive neutron diffraction that amplified the conclusions sampling, metallography, interpretation
Introduction
The central premise of this paper is that contained within penetration and, with the exception of some proton
the material of every metal artefact is a wealth of infor- instruments, are more or less blind to surface and
mation for the archaeologist, the historian, the curator, sub-surface alterations such as corrosion, oxidation, and
the conservator, and the metallurgist. Everything that plating; obtaining a composition unaffected by these
has happened to it since the metal left the furnace can factors requires some intervention.
leave traces which can be recovered by an appropriate
The composition of a metal artefact only tells part of
metallographic method. Traces of one event may over-
its story: macro- and microstructure are as important.
write another but, very often, a very complete biography
These pose another problem, since observation of
is possible, one that is enhanced when it is linked to the
the microstructures of metals requires a flat, polished
documentary record, and that can have a telling impact
surface, either on a cut sample or in situ on an object
on both scholarship and broader outreach.
(or on a thin foil for the rare occasions when transmis-
At the start of, and throughout, any project, be it a single sion electron microscopy (TEM) might be relevant to a
museum or gallery exhibit, or the assemblage of metal- project). A cut sample is also often the most effective and
work from an excavation, there should be a conversation efficient way of displaying the complexities of corrosion
among all these parties to review all the possibilities. Two to both the archaeologist and the conservator, and this
questions have to be addressed. The first is how much of is especially true for archaeological material. Advances
that information needs to be retrieved, bearing in mind in computed tomography (CT) have great potential in
that if the artefact is correctly conserved, unrecovered this area as micro-CT can map inclusions and corrosion
information will still be retained. The second question in three dimensions in suitable sections, but expense
is how to do this, and the cry is always for non-invasive is always an issue. Other whole-object techniques for
analysis, which can be difficult since the information composition and structure do exist, especially those
is distributed throughout the artefact. Non-sampling using neutron beams, but they are limited in resolu-
methods of compositional analysis, even using heavy, tion and can be extremely time consuming. They are,
high-energy particles such as 3 MeV protons (proton though, still the best way of addressing some important
induced X-ray emission (PIXE)), have limited depth questions over whole objects.
NON-INVASIVE ANALYSIS: THE MIRAGE AND THE REALITY 213
Case studies
Three case studies are presented to demonstrate different
combinations of non- and minimally invasive approaches
to characterisation. In the first two the mix has evolved
over the continuing life of the projects, greatly extending
the scope of the initial conversations. The third project is
of much shorter duration to a planned design and is now
in the course of publication. All three projects provide
scope for further development, the only limitation being
funding. The projects are:
• the Pisa Griffin, a medieval piece of public art, where
much could be done non-invasively but sampling was
essential to solve certain problems;
Figure 1. 3D scan of the Pisa Griffin
• copper fastenings from Royal Navy wooden warships
from the late 18th and early 19th centuries, where
12th century. Its early history is unknown and the first
most questions could be answered from small samples,
direct evidence of its presence in Pisa is not until the 15th
although there was one major aspect which required
century, by which time it was mounted on the apex of the
a non-invasive solution;
east pediment of Pisa Cathedral. It is reasonable, though,
• the metalwork from the pile-dwelling Late Bronze Age
to believe that it was placed there when the cathedral was
village at Must Farm, near Peterborough, England,
completed in the early 12th century. How it came to Pisa
which was destroyed by fire a few months after it was
is also unknown, but acquisition during a Pisan raid on
built, c. 850 BC. For the questions asked, the project
Almeria in 1089 or on the Balearic Islands in 1113–15 is
was completely dependent on metallographic samples;
the one opportunity for extensive non-invasive anal- very possible. A date of this order is supported by a radi-
ysis, the mapping of sulfide corrosion, was not practical ocarbon date from organic material found in one wing
within the budget. (Calcagnile 2018); it should be pointed out that the wings
are separate castings riveted in place and are not a pair.
The first two projects evolved from a rather basic design
over an extended period of time, and they are still The relevance of the Griffin to the discussion in this paper
evolving with the incorporation of new techniques as lies in its present location and access to it. The Griffin was
so much is still unknown. The third used a much more removed from the cathedral roof in 1828 and exhibited in
detailed design based on previous experience of Bronze the Camposanto, protected from the rain but still exposed
Age metalwork assemblages. In all cases, electron probe to a marine atmosphere. In 1986 the Griffin and its tall
microanalysis (EPMA) with wavelength dispersive spec- plinth were moved to the new Opera del Duomo Museum.
trometry (WDS) and optical metallography formed the In its present position, access along one side is difficult.
analytical core of the project. Another consideration is a large crack originating from a
cold shut in its back, which gave great cause for concern
The Pisa Griffin after the Griffin was transported to an exhibition in Doha
in 2008. The decision was made not to move it again, which
The Pisa Griffin is a large copper alloy figure of a griffin,
prevented its transport for radiography or a CT scan. In
a mythical beast with the head, breast, and wings of an
addition, it was also decided that it was not possible to
eagle and the body of a lion, and all analysed components
take radiographs in situ safely. Thus, to understand the
were cast in a copper-tin-zinc alloy with a modest lead
construction and condition of the Griffin, it was necessary
content of 1%–2%. The Griffin measures 1.07 m in height
to depend on photography, endoscopy, and a 3D scan of the
and 0.90 m in length (Figure 1). It has been the subject of
exterior and as much of the interior as could be reached.
extensive research over a long period, culminating in a
symposium in 2013, published in 2018 (Contadini 2018). When compositional analysis was first agreed in the early
It was most probably cast in Al-Andalus (Andalusia), in 1990s, portable analytical techniques such as X-ray fluo-
Moorish Spain, at the end of the 11th or beginning of the rescence (XRF) or laser-induced breakdown spectroscopy
(LIBS) were not available. Sampling was permitted in five

locations: a sample was cut from the finning inside the
body via a ventral opening and two were drilled from
Figure 3.
each wing. The remaining two samples were drilled from Metallography
a copper alloy vessel that had been discovered cast into of sample
the interior of the body, with the aim of determining from the
interior of the
whether it was cast integrally with the body or as a retrofit
Griffin (field of
(Figure 2). It is proposed that the vessel was part of an view: 100 µm)

apparatus for producing sound, in other words to make
the beast roar (Camber and Contadini 2018).
Figure 4a. Plot of tin against zinc for Pisa Griffin samples
Figure 2. The vessel in the interior of the Griffin
Pisa is close to the sea and metallography of the cut sample

(Figure 3) clearly showed the effects of centuries of expo-
sure. Here, EPMA allows the selection of corrosion-free
areas for analysis coupled with detection limits down
to 100 ppm for impurities and trace elements. Eighteen
elements were analysed, typically over 30 × 50 µm rasters,
with bivariate plots made for a number of element pairs,
for example tin vs zinc (Figure 4a) and iron vs antimony
(Figure 4b) (Northover 2018). From these data it is clear
that the two wings are not a pair and that they and the Figure 4b. Plot of iron against antimony for Pisa Griffin samples
vessel differ from the alloy of the body, showing that the
vessel was not cast integrally with the body.
The interior of the body and the vessel were later analysed
by both XRF and LIBS on uncleaned surfaces (Loren-
zetti et al. 2018). The results for the principal alloying
elements, zinc, and tin (Figure 4c) were broadly similar
for the body but probably reflect segregation in a large
casting, with the EPMA sample from the interior coming
from metal penetrating a crack in the core and solidifying
before the other areas analysed. Differences will also
arise from varying and, in the case of XRF and LIBS,
unknown effects of corrosion. For the vessel there was a
large difference: the drilled EPMA samples had corroded Figure 4c. Comparative analyses of tin and zinc for the Pisa Griffin
material discarded, and higher zinc contents and lower

tin contents relative to the corroded surface are to be
expected (Figure 4c). Also, in the EPMA data the impu-
rity pattern in the body with, for example, a lower iron
and higher antimony concentration confirms its differ-
entiation from the vessel. We must conclude that, when
understanding the composition and construction of an
object is crucial to understanding its history, appropriate
sample-based analysis is essential.
Figure 5a. Copper bolt from HMS Impregnable
Copper fastenings
The beginning of this project was a post-excavation
study of copper bolts from the wrecks of two Royal Navy
ships, the 96-gun HMS Impregnable (completed 1786,
lost 1799) and the frigate HMS Pomone (completed
1895, lost 1811). Cut samples were taken for optical
metallography and EPMA (Figures 5a–c) (Northover
et al. 2015), with part of the work on Pomone being tele-
vised. It quickly became apparent that the copper used
by the Royal Navy was very uniform in character, with
arsenic, bismuth, silver, and lead being the principal
Figure 5b. Detail of bolt showing Government broad arrow and Forbes
impurities, and all the bismuth and part of the arsenic works stamp
being present as oxide inclusions. The copper came
from mines in Cornwall and on the island of Anglesey
in north-west Wales, the same copper that was used for
the British copper coinage from 1797 and exported to
the Philadelphia Mint in the United States of America
for striking ‘Liberty Head’ cents. It also became apparent
that the generally accepted history of the use of copper
fastenings by the Royal Navy was wrong. The outcome
has been continuing work to understand the metallurgy
of the bolts and to correlate this with documentary
research to create a better history.
From the first voyage in 1761–63 of HMS Alarm, the first

Royal Navy vessel to be copper sheathed as a protection
Figure 5c. Microstructure of copper bolt from Impregnable
against the shipworm (Teredo navalis), the galvanic corro-
sion caused by the contact between copper sheathing
Scotland and housed in the mansion Forbes built for his
and iron fastenings in seawater was recognised. As an
retirement. In conjunction with Admiralty records, this
experiment, the 14-gun HMS Swallow, built in 1769, was
previously unexplored archive offers us a very complete
ordered to be copper fastened; despite its loss in 1776, the
paper trail for this history.
experiment was deemed a success and, at the beginning of
1777, six vessels up to 32 guns were ordered to be copper From 1777 onwards, the size of vessels that it was deemed
fastened. The bolts were forged with a trip hammer and a to be safe to bolt with copper increased until, in 1783,
swage and were delivered by William Forbes, the Navy’s it was ordered that all new construction, including the
copper contractor. Miraculously, when Forbes retired largest battleships, should be copper fastened. This was
in 1793 he took his business archive with him, and it quickly followed by the order to re-fasten all existing
is now preserved by Falkirk Community Archives in iron-fastened ships with copper. To achieve this, the
production of copper bolts had to be rapidly expanded preparation for EBSD as well. One of the first results
and, since forging was too slow, rolling mills with grooved was to identify a wire texture in a bolt from Pomone,
rolls were introduced and three separate patents for hot implying that it had been produced following William
and cold rolling taken out in 1783. Collins’s patent of 1783, which used grooved rolls to
grip a copper bar to pull it through a die (Figure 6a–b).
Characterisation of the copper continued to be by
(Texture describes the distribution of preferred orienta-
EPMA and optical microscopy followed up by other
tions of grains produced when metal is worked and heat
techniques in order to answer questions raised by the
treated, with different textures correlated with different
metallography. In particular, it was necessary to explain
processes.) Thus, although it was not possible to assign
microstructures which showed elongated grains, but no
a bolt to a source by its composition, some at least could
slip traces or deformation twins could be etched. The
be linked with a maker through their texture.
suggestion was made that electron backscatter diffrac-
tion (EBSD) could assist (Northover and Northover Copper bolts were produced in a large range of sizes, up
2012); fortuitously, the routine of polishing and etching to 60 mm in diameter and 4 m or more in length and, as
used for the metallography proved to be the ideal noted above, produced by forging, rolling, or drawing,
or, indeed, a combination of these processes. Add to this
the stresses incurred in driving the bolts, the workings of
a ship’s hull at sea, and the breaking up of the hull when
wrecked, all of which add complexity to the pattern of
textures in a bolt. To map this using cut samples would
be unnecessarily destructive, but the same results can
be achieved using neutron diffraction techniques. This
work is being carried out at the ENGIN-X beamline at
the ISIS neutron source at Harwell, England. A complete
bolt, held by a robot, can be moved through a range of
positions and orientations with crystallographic data
collected at each location; the analysed volume at each
step might be a 4 mm cube. Copper is a good absorber
of neutrons and so there is a limit to the path length
through the copper that can usefully be used; data
collection is slow, and it may take 24 hours per bolt. A
bolt from HMS Impregnable in position in the Cyberman
robot is shown in Figure 7a and typical texture results in
Figure 7b (Malamud et al. 2016, Malamud et al. 2017).
Despite this slow rate, neutron diffraction is seen as the
only realistic approach.
So far only bolts produced after the introduction of

rolling/drawing in 1783 have been studied, but a bolt
from HMS Crocodile, one of a batch of six copper-fas-
Figure 6a. Drawing from specification of William Collins’s patent No. tened ships ordered in 1777, and certain to be forged, is
1388, 2 October 1783 (image from Sheffield City Library)
now available. Work will also be done on bolts used by
the merchant fleet and those used on French ships of the
same period, mainly using technology exported from
Britain; an application for beam time will be available
shortly (to date, all applications for this project have been
successful). Where good cross-sections of bolts are avail-
able, micro-hardness maps can also be made but, so far,
Figure 6b. Wire texture data from a Collins bolt they have proved challenging to interpret and correlate
pile-dwelling settlement at Must Farm had several objec-

tives, namely, to determine:
• the alloys used and possible correlations with the
types of object;
• the impurity patterns and the evidence they provide
for the origins of the metal;
• the results of lead isotope analysis of selected pieces;
• the quality of manufacture;
• given the short period of occupation, whether there
was any evidence that particular tools or weapons
were new, or if these had been brought to the site by
the settlers;
• the nature of a container of scrap metal;
• the condition of the objects on the day of the fire;
• the impact of the fire on each object and how this
Figure 7a. Bolt
correlated with the forensic investigation of the fire;
from Impregnable
at ENGIN-X and

• the different states of corrosion of the bronze and their
relationship with the different environments in which
they were left after the fire.
It was agreed that all objects, where their condition

allowed, would be analysed by EPMA on cut samples,
which would then be available for metallography.
Sampling and sample preparation used the same methods
as for the two projects already described; samples for
lead isotope analysis were drilled from within the cut
made with the saw.
The significance of the site lies in the way in which it

was abandoned with a full range of domestic equip-
ment – bronze, ceramic, wood, and other materials
were left behind, meaning we can learn what the typical
possessions of a household were. It should be noted that
potentially ‘elite’ objects such as swords and sheet bronze
Figure 7b. Typical texture data from ENGIN-X
vessels were present only as fragments, suggesting that the
settlement itself was not of particularly high status. Also
of great importance is the exact dating of the site through
with the texture data. One precious asset is that two sister dendrochronology and radiocarbon to the middle of
ships of Pomone are still afloat, HMS Trincomalee and the 9th century BC, because such precise dates for Late
HMS Unicorn. Remarkably, Unicorn was never taken into Bronze Age metalwork in Britain are scarce.
service and, apart from being towed from the Thames to
The selection of alloys was typical for the British Late
Dundee in Scotland, it has never been to sea. It is to be
Bronze Age, with leaded bronze for cast objects and
hoped that we can access some of the bolts and so exclude
unleaded bronze for sheet, the lead often being residual
some variables from our texture studies.
from scrap bronze. Tin contents were mainly 7%–11%
with a few around 3% tin, and some specialist tools such
Must Farm
as chisels with 12% tin or more. The impurity patterns
The post-excavation project design for the metallurgical were those common in eastern England at the time
study of the bronze artefacts from the Late Bronze Age (arsenic, antimony, nickel, and silver), with much of the
metal imported from the Continent and lead added in

England. The scrap bronze found in a wooden container
had just the same characteristics and was, presumably,
from objects broken in use, or just found, and collected
to exchange for new bronze. None of the Must Farm finds
appeared new to the site.
The metallographic study was keenly awaited: previous

studies of fire-damaged ancient metalwork were from
cremations and none from an extensive fire such as that
which engulfed Must Farm. The author has developed
a system for tabulating metallographic data so as to give
a simple overview of an assemblage and to allow some
Figure 8a. Partially melted medium tin leaded bronze bucket fitting
variables to be treated quantitatively. The system was
(scale bar: 20 µm)
pioneered for a large metalworker’s hoard from Arbedo,
Ticino, Switzerland (Northover 1998), and extended to
fire-damaged bronze from the Late Bronze Age/Early
Iron Age cemetery at Bischofshofen-Pestfriedhof, Austria
(Northover 2009), and a Late Bronze Age weapon assem-
blage in Norfolk (Bridgford and Northover 2020). This
was also done for Must Farm and it was found that the
most severely heat-affected microstructures correlated
strongly with the hotspots found in the forensic study of
the fire. Although the excavation is now in the course of
publication, there is still more work to be done, especially
with understanding partially melted structures (e.g.
Figures 8a–b). Sadly, the detailed analysis and micros-
copy needed cannot be funded even from a generous Figure 8b. Partially melted and quenched low tin bronze from cast bronze
excavation budget. vessel (scale bar: 50 µm)
All the analysis from Must Farm so far has been carried
out on cut samples; a typical sample size is 2–3 mm across,
although useful results can be obtained from a sample of
less than 1 mm diameter. Nonetheless, there is one area
where a non-invasive, surface analysis would have been
of great benefit, and that is in mapping the occurrence of
sulfide corrosion, given the areas of waterlogged, anaer-
obic conditions across the site. Not every bronze need
be analysed, only sufficient to characterise the different
types of patina so that sulfide corrosion can be identi-
fied visually and mapped photographically. Because the
sulfur K X-ray lines are of low energy, the most effective Figure 9. An example of the patination of bronze tools from Must Farm
approach would be either XRF or PIXE with a helium
flush or, for preference, a vacuum chamber. If such an Discussion
experiment were funded it would offer a requirement and
To achieve the greatest benefit from the input of the
a boon to conservators, the requirement being to ensure
as accurately as possible the surface appearance and the metallurgist to the study and interpretation of the arte-
boon to offer a simple approach to understanding objects facts discussed in each of these case studies, it is very clear
with sulfide corrosion. that the optimum approach to such study is a combina-
tion of non-invasive and sample-based analyses. We must to the project and of the methods needed to obtain it,
now consider how this impacts the decision-making of with no a priori assumptions about the use of either
the conservator and curator. analytical approach. The outcome must be an appropriate
balance between non-invasive and invasive techniques,
Perhaps an antipathy to sampling has stemmed from
however minimalist the latter might be. Those should be
scholars in the past advocating samples that were unnec-
the keywords in the future: appropriate analysis.
essarily large. It may be true that the bigger the sample,
the greater the possible return in information, but there
References
is a law of diminishing returns. Perhaps those analysing
ancient metalwork did themselves a disservice by a Bridgford, S.D. and J.P. Northover. 2020. Three Bronze Age
concentration on provenance using drilled samples such weapon assemblages from Norfolk, East Anglian Archae-
that asking for a metallographic sample was seen as a step ology 171. Norwich: Norfolk Historic Environment.
too far. It might also be said that the information to be Calcagnile, L. 2018. Risultati delle datazioni con il
obtained from the present-day range of metallographic radiocarbonio. In The Pisa Griffin and the Mari-Cha
techniques was never adequately explained. This is where Lion. Metalwork, art, and technology in the medieval
the conversations advocated in the introduction are so Islamicate Mediterranean, ed. A. Contadini, 485–88.
important, and where the metallurgical input must be Pisa: Pacini Editore.
supported by a demonstration of typical sample sizes
and sites, and the range of information to be gained from Camber, R. and A. Contadini. 2018. Acoustic automata.
combining current metallographic techniques, not least In The Pisa Griffin and the Mari-Cha Lion. Metalwork,
the pattern and extent of the corrosion. art, and technology in the medieval Islamicate Mediter-
ranean, ed. A. Contadini, 63–76. Pisa: Pacini Editore.
It can be argued that invasive sampling, however minimal,
Contadini, A., ed. 2018. The Pisa Griffin and the
can be avoided because methods still being developed
Mari-Cha Lion. Metalwork, art, and technology in the
will allow essential data to be collected non-invasively.
medieval Islamicate Mediterranean. Pisa: Pacini Editore.
This is unlikely, but even if it were not, there are still
two constraints: the laws of physics and cost. An impor- Lorenzetti, G., E. Grifoni, S. Legnaioli, S. Pagnotta,
tant consequence of the former is a lack of resolution, and V. Palleschi. 2018. Analysis of the alloy of the Pisa
both spatial and spectral, and extended time scales for griffin. In The Pisa Griffin and the Mari-Cha Lion.
collecting data. In contrast, a cut sample can provide a Metalwork, art, and technology in the medieval Islam-
quantitative analysis for, say, 18 elements at a detection icate Mediterranean, ed. A. Contadini, 105–12. Pisa:
limit down to 100 g/g and optical micrographs for GBP Pacini Editore.
50 per sample, with multiple samples each day. The same Malamud, F., S.M. Northover, J. James, J.P. Northover, and
sample can then be preserved for future use, for example J. Kelleher. 2016. Texture analysis of Napoleonic War era
for stable isotope analysis by laser ablation inductively copper bolts. Applied Physics A 122: 276–90.
coupled plasma mass spectrometry, eliminating the need
for further analysis. Malamud, F., S.M. Northover, J. James, S. Nneji, J.P.
Northover, and J. Kelleher. 2017. Napoleonic War era
From the non-invasive perspective, making CT and 3D copper bolts. Journal of Applied Crystallography 50(5):
scans more accessible to museums and archaeology would 1369–75.
be enormously helpful, especially in imaging the mechan-
Northover, J.P. 1998. Analysis of copper alloy metal-
ical state of an object (as with the Griffin), mapping
work from Arbedo TI, Annex 1. In Der Depotfund von
corrosion and particle distributions, and guiding sample
Arbedo TI (Antiqua), ed. M.P. Schindler, 289–315. Basel:
selection. Neutron imaging and diffraction methods are
Schweizerische Gesellschaft für Ur- und Frühgeschichte.
also powerful, as seen in the discussion of the copper
bolts, but none of these can contribute the compositional Northover, J.P. 2009. Analysis and metallography of
data at the level required to inform the conservator, the copper alloy metalwork. In Das spätbronze und frühhall-
curator, the archaeologist, and the metallurgist. This stattzeitliche Gräberfeld von Bischofshofen-Pestfriedhof,
brings us back to the start of these arguments: the need eds. A. Lippert and P. Stadler. Bonn: Verlag Dr Rudolf
for a proper definition of what is required by each party Habelt GmbH.
Northover, J.P. 2018. Technical analysis of the Mari-Cha

Lion and the Pisa Griffin. In The Pisa Griffin and the
Mari-Cha Lion. Metalwork, art, and technology in the
medieval Islamicate Mediterranean, ed. A. Contadini,
85–105. Pisa: Pacini Editore.
Northover, J.P., S.M. Northover, J. James, and F. Malamud.
2015. Analysis and metallography of copper and brass
components from the Alum Bay wreck sites. In The
maritime archaeology of Alum Bay. Two shipwrecks on
the north-west coast of the Isle of Wight, England, BAR
British Series 608, eds. J. Satchell and J. Whitewright.
Oxford: British Archaeological Reports.
Northover, S.M. and J.P. Northover. 2012. Applications
of electron backscatter diffraction in archaeology. In
Historical technology, materials and conservation: SEM
and microanalysis, eds. N. Meeks, C. Cartwright, A. Meek,
and A. Mongiatti, 76–85. London: Archetype Publications
and The British Museum.
Author
Peter Northover was born in Oxford and grew up in
the Isle of Wight. He read metallurgy at the University
of Oxford, having worked at the metallurgy lab of West-
land Aircraft at Cowes, Isle of Wight, between school
and university. He graduated in 1970 and went on to a
DPhil in metallurgy with a thesis on stress corrosion and
hydrogen embrittlement in controlled transformation
stainless steels. Since 1974 he has pursued the applications
of metallurgy in archaeology, specialising in non-ferrous
metals, producing well over 200 publications. Since
retiring from the University of Oxford in 2014 he has
specialised in copper and copper alloys of the industrial
revolution, especially for ships and railways.
An Analytical Study of the Corrosion Behavior and
Microstructural Properties of a Group of Copper Alloy
Artifacts from the Khirbet Yajuz Archaeological Site, Jordan
Ahmad N. Abu-Baker* Lutfi A. Khalil
Department of Conservation and Management of Cultural Resources, Faculty Department of Cultural Resources Management and Conservation, Faculty of
of Archaeology and Anthropology, Yarmouk University Archaeology and Tourism, The University of Jordan
Irbid, Jordan Amman, Jordan
Abstract
In this study, the composition, corrosion, and microstructure the manufacture of the artifacts. The uncommon presence
of a group of four copper alloy artifacts excavated from the of phosphate corrosion products was attributed to the
Byzantine cemetery at the Khirbet Yajuz archaeological site, phosphate species released by the hydrolysis of hydroxyapatite
Jordan, were investigated. Scanning electron microscopy (Ca 5(PO 4) 3OH) from the skeletal materials in the adjacent
coupled with energy-dispersive X-ray spectroscopy (SEM- bodies buried in the cemetery. The resulting formation of
EDX) analyses of a necklace, a bracelet, a paten, and a spatula insoluble phosphate corrosion products on the surface of
showed that they were made of quaternary copper-zinc- the artifacts improved their stability and reduced the risk of
tin-lead alloys. X-ray diffraction (XRD) analysis revealed the activating the cuprous chlorides present in the internal pits
presence of cuprite (Cu2O), cornetite (Cu3PO4(OH)3), sampleite and the dezincification of these artifacts in the uncontrolled
(NaCaCu 5(PO 4) 4Cl·5H 2O), spencerite (Zn 4(PO 4) 2(OH) 2·3H 2O), environment of their storage area. The conservation options
pyromorphite (Pb 5(PO 4) 3Cl)), malachite (CuCO 3·Cu(OH) 2), for the artifacts include direct intervention using corrosion
atacamite/paratacamite (α- and γ-Cu2(OH)3Cl), and cotunnite inhibitors as well as preventive conservation measures, such
(PbCl2) as the corrosion products. SEM-EDX and EDX elemental as by controlling the relative humidity.
mapping were conducted to determine the corrosion
stratigraphy and elemental distribution in the corrosion Keywords
layers and original alloy. The metallographic examination Khirbet Yajuz, copper alloy, corrosion, microstructure, copper
showed various degrees of working and heat treatment in phosphates, zinc phosphates, lead phosphates
Introduction
Archaeological copper alloy artifacts form an important corrosion processes, the exposure conditions (e.g., burial,
part of cultural heritage collections. Analytical investi- exhibition, or storage), and metallurgical aspects of the
gation of these artifacts is essential to understand their studied artifacts. The diverse range of corrosion mecha-
chemical, physical, and mechanical characteristics. Iden- nisms and products associated with copper alloy artifacts
tification of their corrosion products and the processes emphasizes the importance of comprehensive analytical
involved in their formation can shed light on the archae- investigation to elucidate the particular characteristics of
ological context and determine the preservation state the artifacts with respect to their context, which in turn
and conservation needs. The conservation literature will allow a determination of their conservation needs.
includes several studies that investigated archaeological As an example, a multi-analytical investigation of a group
copper alloy artifacts and identified their composition of four copper alloy artifacts excavated from the graves
and corrosion characteristics using various analytical of the Byzantine cemetery at the Khirbet Yajuz archae-
techniques (Fabrizi et al. 1989, Abu-Baker 2008, Ingo ological site in Jordan were investigated. The aim was to
et al. 2010, Oudbashi et al. 2016, Papadopoulou et al. identify the composition, corrosion, and microstructural
2016, Abu-Baker and Al-Qudah 2017, Fan et al. 2020, properties of the artifacts and thus their archaeological
Abu-Baker et al. 2021). The results of those studies context and corrosion processes in order to develop
have helped to demonstrate the link between long-term appropriate preservation and conservation strategies.
Figure 1. The copper alloy artifacts under investigation: (a) necklace, (b) bracelet, (c) paten, and (d) spatula
Materials and methods The corrosion products of the artifacts were analyzed by
X-ray diffraction (XRD) analysis using a Rigaku ultima
The artifacts that were the focus of this study consisted
IV X-ray diffractometer operating at 40 kV and 40 mA.
of a necklace, a bracelet, a paten, and a spatula. They are
A small amount of the corrosion products was scratched
representative of dozens of similar artifacts excavated
off from the surface of each artifact, then mixed and
from the Khirbet Yajuz archaeological site, which is
dispersed in ethanol and spread on a microscopic glass
located on a hill situated about 11 km northwest of the
slide. After evaporation of the ethanol, the slide was
center of Amman, Jordan. The site was inhabited in
placed in the XRD instrument for analysis. The X-ray
the Roman, Byzantine, Umayyad, and Abbasid periods, was generated from an anticathode copper (Cu) tube
as indicated by the architectural and material finds with a wavelength CuKα= 1.54178 Å. The analysis range
discovered during excavations (Khalil 1998). The arti- was 5°–95° 2θ and the scan speed was 3° per minute.
facts are heterogeneously corroded, broken, and have
missing parts (Figure 1). They have yet to be analytically The embedded cross-sections were also examined using
assessed since their discovery and placement in the an Optika IM-3MET inverted metallurgical microscope
storage area of the University of Jordan Archaeological to investigate their corrosion profile. They were then
Museum (UJAM) in Amman some 25 years ago. The etched with alcoholic ferric chloride to investigate their
fluctuating relative humidity and other corrosive agents microstructures using bright-field and cross-polarized
in the uncontrolled environment of the storage area light microscopy (PLM) to determine the respective
have had an adverse effect on the chemical stability and manufacturing techniques (Scott 1992, 72). The photo-
physical integrity of the artifacts. It is expected that the micrographs were taken using an Optika C-P8 digital
analytical investigation will enable their preservation camera attachment.
and conservation.
Scanning electron microscopy (SEM) coupled with ener-
gy-dispersive X-ray spectroscopy (EDX) spot analysis Alloy compositions and microstructures
and X-ray elemental mapping were used to study the SEM-EDX analysis of the necklace’s internal alloy showed
corrosion stratigraphy and elemental distribution in the a quaternary Cu-Zn-Sn-Pb alloy. The SEM image and
original alloys and corrosion layers. One cross-section EDX elemental mapping showed a homogeneous distri-
was taken from each artifact and prepared according to bution of Zn and Sn in the Cu-rich α-phase, with Pb
the procedure described in Abu-Baker et al. (2021). A appearing as separate globules randomly distributed in
FEI-Quanta 600 scanning electron microscope coupled the alloy (Figure 2a). On the PLM image of the etched
with an energy-dispersive X-ray spectroscopy analyzer internal alloy of the necklace, recrystallized equiaxed
was used for this analysis. Wet mode at 0.5 Torr H2O and α-grains together with annealing twins and white lead
an electron beam of 30 kV were employed. The samples globules were seen (Figure 2b). The repeated cycles of
were gold-coated before their analysis to avoid charging working and annealing were finished by an annealing
effects at the corroded edges of the cross-sections. process, as indicated by the presence of straight annealing
AN ANALYTICAL STUDY OF THE CORROSION BEHAVIOR AND MICROSTRUCTURAL PROPERTIES OF A GROUP OF COPPER ALLOY ARTIFACTS FROM THE KHIRBET YAJUZ ARCHAEOLOGICAL SITE, JORDAN 223
twins; however, coring in the copper-rich α-phase was also apparent on the EDX elemental mapping image. The
not completely removed by the annealing process (Scott zinc and tin had completely solubilized in the copper,
and Schwab 2019, 154). resulting in their homogeneous distribution in the
alloy (Figure 3a). The PLM image of the etched internal
According to the SEM-EDX analysis, the bracelet’s alloy showed recrystallized equiaxed α-grains and white
internal alloy is a quaternary Cu-Zn-Sn-Pb alloy. The lead globules (Figure 3b). The small grains suggested
SEM image showed air porosity and white lead globules, repeated cycles of working and annealing and a final
Figure 2. SEM-EDX, EDX elemental mapping (a), and PLM (b) of the Figure 3. SEM-EDX, EDX elemental mapping (a), and PLM (b) of the
internal alloy of the necklace internal alloy of the bracelet
annealing process, as indicated by the presence of straight Corrosion characteristics

annealing twins; however, as in the necklace, coring in
XRD analysis
the copper-rich α-phase was not completely removed
by the annealing process (Scott and Schwab 2019, 154). The XRD analysis of the four artifacts revealed
SEM-EDX analysis of the paten’s internal alloy was cuprite (Cu 2 O) and copper phosphate minerals,
consistent with a quaternary Cu-Sn-Zn-Pb alloy. The such as cornetite (Cu 3(PO 4)(OH) 3) and sampleite
EDX elemental mapping images showed the even distri- (NaCaCu5(PO4)4Cl·5H2O), as the main corrosion prod-
bution of zinc and tin in the copper matrix, with a small
amount of lead present as elongated, chloride-corroded
globules (Figure 4a). The bright-field image of the
etched internal alloy revealed a partially recrystallized
microstructure with some annealing twins and lead
globules; however, coring remained from the cast alloy,
which together with the weakly defined grain structure
indicated that the paten had not been extensively worked
or annealed. The presence of slip lines and intergran-
ular cracks suggested cold working, probably to finish
the surface decorations, without sufficient annealing
(Figure 4b) (Scott and Schwab 2019, 154).
The SEM-EDX analysis of the spatula’s internal alloy
indicated a quaternary Cu-Zn-Sn-Pb alloy. The SEM
image together with EDX elemental mapping showed
spherical and elongated lead globules distributed in the
alloy, and zinc and tin that were completely solubilized in
the copper and distributed homogeneously (Figure 5a).
The partially recrystallized grains appearing under
bright-field illumination suggested that the cast artifact
had been annealed for a short time at high temperature
or for a longer time at a moderate temperature. The
presence of slightly curved annealing twins and a series
of slip lines indicated cold working without sufficient
annealing, to give the artifact its final shape (Figure 5b)
(Giannossa et al. 2013; Scott and Schwab 2019, 154).
The four artifacts are made of quaternary Cu-Zn-Sn-Pb
alloys. As decorative and cosmetic objects, the presence
of large amounts of zinc, particularly in the necklace and
bracelet, give the alloy an attractive golden color. In the
3rd century AD, quaternary Cu-Zn-Sn-Pb alloy (modern
leaded gunmetal) was used regularly and was made by
mixing pure brass, freshly made using a cementation
process, with scrap bronze, which decreased the amount
of zinc from 22%–28% to around 13% and added tin
into the composition of the alloy. Lead was either added
deliberately to reduce the melting point and improve the
fluidity and casting properties of the alloy or due to the
use of copper ores containing lead minerals. The small
amount of tin increased the hardness and durability of Figure 4. SEM-EDX, EDX elemental mapping (a), and bright-field image
the alloy (Craddock 1978, Thornton 2007). (b) of the internal alloy of the paten
ucts. Significant amounts of the zinc hydroxy phosphate

mineral spencerite (Zn4(PO4)2(OH)2·3H2O) were detected
on the necklace, bracelet, and paten, and the lead minerals
pyromorphite (Pb5(PO4)3Cl) and cotunnite (PbCl2) on
the spatula. The basic copper chlorides atacamite/parat-
acamite (α/γ-Cu2(OH)3Cl) are also present in significant
amounts on the bracelet and the spatula, while malachite
Figure 6. XRD analysis of the corrosion products: (a) necklace, (b) bracelet,
(c) paten, and (d) spatula
(CuCO3·Cu(OH)2) is a minor corrosion product found

on all of the artifacts (Figure 6).
Cuprite is the first corrosion product to form in the pres-

ence of oxygen and moisture, and the predominant oxide
of copper that appears over a wide range of conditions
(Scott 2002, 82). Malachite forms over cuprite, due to the
dissolution and precipitation of the oxide as copper(II)
salts; its presence indicates a high carbonate content
at a pH of 7.5 or higher (Eary 1999). The presence of
atacamite/paratacamite and the absence of nantokite
(CuCl) in the XRD results suggested that the initially
formed nantokite in the burial soil underwent oxidative
hydrolysis, with the formation of atacamite/paratacamite
under the uncontrolled storage conditions. The artifacts
have either been stabilized by this reaction or are in a
metastable state if nantokite is present deeper in the alloy
such that it was not collected in the samples obtained
for XRD analysis (Scott 1990). SEM-EDX analysis of
the cross-sections will help to distinguish between these
possibilities.
Cornetite (Cu 3(PO 4)(OH) 3) was identified on four

artifacts and sampleite (NaCaCu5(PO4)4Cl·5H2O) on
three. The phosphoric species causing the formation of
these copper minerals in the burial environment was
hydroxyapatite (Ca5(PO4)3OH), which is the main inor-
ganic material of bones and teeth. The decomposition
of the adjacent buried bodies in the graves where the
artifacts were found resulted in the release of phosphate
ions, which induced the formation of copper phosphate
minerals. The reaction leading to the formation of
Figure 5. SEM-EDX, EDX elemental mapping (a), and bright-field image cornetite by cupric ions in contact with hydroxyapatite
(b) of the internal alloy of the spatula and ground water is shown in Equation (1):
Ca5(PO4)3OH + 6Cu2+ + 5H2O → 5Ca2+ + and faster than Cu+1 ions, which have an ionic radius
2Cu3(PO4)(OH)3 + H2PO4− + 3 H+ (1) of 0.096 nm (Ravichandran and Rajendran 2005, Ingo
et al. 2010, Papadopoulou et al. 2016). The PLM of the
According to the equation, H2PO4− and H+ resulting
unetched cross-section shows the blue-green corrosion
from this reaction will cause the formation of cornetite
layer, which mainly contains cornetite and sampleite,
(or other copper hydroxy phosphate minerals). The
over a thin white band that is probably the spencerite
presence of sampleite can be attributed to the soluble
sodium and chloride ions in the ground water of the
burial environment, which react with the cupric and
phosphate ions to form this mineral (Fabrizi et al. 1989;
Bruckner et al. 1993; Scott 2002, 243; Fan et al. 2020). The
hydrogen phosphate ions resulting from the hydrolysis
of hydroxyapatite also react with the released zinc ions
and form spencerite (Zn4(PO4)2(OH)2·3H2O) (Chen et al.
2016). The formation of compact zinc phosphate minerals
hindered Cu(I) dissolution and improved the corrosion
resistance of the alloy (Yohai et al. 2011).
Pyromorphite (Pb5(PO4)3Cl) is a stable lead mineral

found in a wide range of geochemical environments.
The Pb(H2PO4)2 and PbHPO4 formed by the reaction
of Pb2+ with the H2PO4− and HPO42− ions resulting from
the hydrolysis of hydroxyapatite can be converted to
pyromorphite at pH values above 2.7 and 4.7, respectively
(Nriagu 1973, Fan et al. 2020). In addition to pyromor-
phite, the XRD analysis revealed cotunnite (PbCl2) in
the spatula. Cotunnite is a stable lead corrosion product
under acidic and high-chloride-containing microenvi-
ronments (Selwyn 2004, 122).
SEM-EDX, EDX mapping, and PLM
SEM-EDX analysis and EDX elemental mapping of the

necklace’s corroded edge showed an external layer that
mainly contains Cu, with variable amounts of P, C, O,
Zn, Pb, Sn, Cl, and Fe (Figure 7a, area a). Accordingly,
cornetite and sampleite predominate, while the layer
in Figure 7 marked (b) is richer in Zn, indicative of the
presence of spencerite, and the thin band marked (c)
contains Cu, C, and O, i.e., the elements of malachite
detected in the XRD analysis. The internal layer marked
(d) has a higher O content, attributable to the presence of
cuprite, while the EDX results of the white area marked
(e) are consistent with lead or lead minerals. Finally, the
internal corrosion pits marked (f) are zinc depleted and
rich in Cu and Cl, which suggests a dissolution of the
more anodic metal, zinc, from the alloy and the pres-
ence of cuprous chloride corrosion cells. The corrosion
process is controlled by diffusion such that anodic Zn+2 Figure 7. SEM-EDX, EDX elemental mapping (a) and PLM (b) of the
ions, with an ionic radius of 0.070 nm, diffuse easier corroded edge of the necklace
detected by XRD and SEM-EDX. This is followed by minerals of copper, zinc, and lead, as detected in the XRD
red-brown cuprite and a lead-rich white layer. The dark analysis. Similar to the corrosion process of the necklace,
red areas of internal dealloying were zinc depleted by the the internal corrosion area has been zinc depleted and is
selective removal of the more anodic metal in the alloy, rich in products, reflecting the chloride-based corrosion
i.e., dezincification (Figure 7b). of Cu (area b) or Sn (area c), both of which are susceptible
SEM-EDX analysis and EDX elemental mapping of the to oxidative hydrolysis. On the bracelet’s unetched PLM
bracelet’s external corroded edge showed the presence of image, the external blue-green and white corrosion prod-
Cu, Zn, Pb, C, O, P, and Cl (Figure 8a, area a), indicative ucts were analytically identified as cornetite, sampleite,
of the presence of mixed oxide, carbonates, and phosphate and spencerite intermixed with atacamite/paratacamite
and malachite, followed by a red-brown cuprite layer and
lead-rich white spots (Figure 8b).
SEM-EDX analysis and EDX elemental mapping of the

paten’s external corroded area showed the presence of
Cu, Zn, P, Pb, O, C, and Cl (Figure 9a), corresponding
to the cuprite, cornetite, sampleite, spencerite, and
pyromorphite identified by XRD. On EDX elemental
mapping, the zinc can be seen to have diffused and
subsequently formed precipitated minerals in the external
phosphate-rich corrosion layer, while the zinc-depleted
internal layer is rich in Cu and Cl, which form active
corrosion cells, as discussed above. The paten’s unetched
PLM image shows the blue-green, white, and red-brown
layers of the previously identified corrosion products
(Figure 9b).
The SEM-EDX analysis and EDX elemental mapping

of the spatula’s external corroded edge showed large
amounts of Cu, O, Cl, and C, and thus the presence
of atacamite/paratacamite and malachite (Figure 10a,
area a). The significant amounts of Pb, O, and P present in
the second corrosion layer (area b) indicate the presence
of cuprite, cornetite, and pyromorphite. In Figure 10, area
(c) is richer in Sn, which suggests that tin(IV) oxide forms
a primary thin corrosion layer underneath the secondary
copper corrosion products. Consistent with the EDX
elemental mapping, the internal intergranular corrosion
is Cu depleted and rich in Pb and Cl (area d), which
suggests a decuprification of the copper-rich α-phase
in an oxygenated burial microenvironment as well as
the presence of lead chloride minerals such as cotun-
nite (PbCl2), as detected by XRD analysis. The spatula’s
unetched PLM image shows stratified blue-green, white,
red-brown, and gray zones of the identified corrosion
products. These are followed by a strong intergranular
corrosion zone that extends to the internal alloy, due to
Figure 8. SEM-EDX, EDX elemental mapping (a), and PLM (b) of the the decuprification that accompanied the chloride-based
corroded edge of the bracelet corrosion of copper (Figure 10b).
The metastable conditions of the artifacts indicated the caused the formation of copper phosphate minerals
beneficial effects of the copper phosphate corrosion that are stable and insoluble over a broad pH range, in
products, despite the presence of chloride ions in the contrast to the copper chloride minerals responsible for
burial soil and the poor environmental conditions in the bronze disease and thus active corrosion. In the storage
storage area. Phosphate ions can displace the less elec-
tronegative chloride ions critical to the active propagation
of corrosion pits (Lytle and White 2014). Therefore, in
the burial environment, the influx of negatively charged
phosphate anions near the corroding copper anode
Figure 9. SEM-EDX, EDX elemental mapping (a), and PLM (b) of the Figure 10. SEM-EDX, EDX elemental mapping (a), and PLM (b) of the
corroded edge of the paten corroded edge of the spatula
area, the insoluble copper phosphate corrosion products in Physics Research Section B: Beam Interactions with
likely prevented the diffusion of oxygen, water, and Materials and Atoms 502: 73–79.
corrosive anions into the corrosion pits or the nearby Bruckner, S., G. Lusvardi, L. Menabue, and M. Saladini.
cathodic region of the corrosion cell, which reduced the
1993. Effect of Cu2+ ion on the structural stability of
risk of activating the cuprous chlorides present under
synthetic hydroxyapatite. Journal of Materials Chem-
the external corrosion layers. Similarly, the formation
istry 3(7): 715–19.
of the zinc phosphate spencerite seems to have reduced
corrosion resulting from dezincification in the zinc-rich Chen, Y., W. Zhang, M.F. Maitz, M. Chen, H. Zhang, J.
artifacts. Mao, Y. Zhao, N. Huang, and G. Wan. 2016. Compar-
ative corrosion behavior of Zn with Fe and Mg in the
Conclusion course of immersion degradation in phosphate buffered
saline. Corrosion Science 111: 541–55.
The multi-analytical investigation of copper alloy arti-
facts from the Khirbet Yajuz allowed their compositional Craddock, P.T. 1978. The composition of the copper
and microstructural properties to be determined. The alloys used by the Greek, Etruscan and Roman civili-
results of the chemical and microscopic investigations zations: 3. The origins and early use of brass. Journal of
showed the effects of the alloy’s composition and burial Archaeological Science 5(1): 1–16.
environment on the types of corrosion products. The Eary, L.E. 1999. Geochemical and equilibrium trends
formation of the insoluble phosphate corrosion products in mine pit lakes. Applied Geochemistry 14(8): 963–87.
on the surface of the artifacts increased their stability
Fabrizi, M., H. Ganiaris, S. Tarling, and D. Scott. 1989.
and reduced the risk of activating the cuprous chlorides
The occurrence of sampleite, a complex copper phos-
present under the external corrosion layers and the
phate, as a corrosion product on copper alloy objects from
dezincification of these artifacts in the environmentally
Memphis, Egypt. Studies in Conservation 34(1): 45–51.
uncontrolled storage area. Possible conservation options
for the artifacts to ensure their long-term stability include Fan, X., Q. Wang, and Y. Wang. 2020. Non-destructive
direct intervention using corrosion inhibitors or preven- in situ Raman spectroscopic investigation of corrosion
tive conservation measures such as by controlling the products on the bronze dagger-axes from Yujiaba site in
relative humidity to a stable value of < 40%. Chongqing, China. Archaeological and Anthropological
Sciences 12(4): 1–10.
Acknowledgments Giannossa, L.C., S. Loperfido, M. Caggese, G.E. De Bene-
This research was supported by a grant from the Gerda detto, R. Laviano, L. Sabbatini, and A. Mangone. 2013. A
Henkel Foundation. systematic characterization of fibulae from Italy: From
chemical composition to microstructure and corrosion
References processes. New Journal of Chemistry 37(4): 1238–125.
Abu-Baker, A. 2008. Investigating the corrosion and Ingo, G.M., A. Çilingiroğlu, F. Faraldi, C. Riccucci,
microstructure of five copper-based archaeological arte- M.P. Casaletto, A. Erdem, and A. Batmaz. 2010. The
facts from Tell el-Ajjul. AICCM Bulletin 31(1): 87–96. bronze shields found at the Ayanis fortress (Van region,
Turkey): Manufacturing techniques and corrosion
Abu-Baker, A. and M.A. Al-Qudah. 2017. A novel
phenomena. Applied Physics A 100(3): 793–800.
dioxime compound for protecting copper in neutral
chloride solutions and to treat bronze disease in archae- Khalil, L. 1998. University of Jordan excavations at
ological artefacts. AICCM Bulletin 38(2): 94–102. Khirbet Yajuz. Annual of the Department of Antiquities
of Jordan 42: 457–72.
Abu-Baker, A., L.A. Khalil, and T. Al-Gonmeen. 2021.
A multi-analytical exploration of the chemical composi- Lytle, D.A. and C.P. White. 2014. The effect of phosphate
tion, microstructural properties and corrosion inhibiting on the properties of copper drinking water pipes expe-
treatment for an archaeological brass censer from Umm riencing localized corrosion. Journal of Failure Analysis
Zuwaytinah, Amman. Nuclear Instruments and Methods and Prevention 14(2): 203–19.
Nriagu, J.O. 1973. Lead orthophosphates—II. Stability of Lutfi A. Khalil is a professor at the School of Archaeology
cholopyromophite at 25°C. Geochimica et Cosmochimica and Tourism at the University of Jordan in Amman,
Acta 37(3): 367–77. Jordan. He is a former dean of the Institute of Archae-
Oudbashi, O., A. Hasanpour, and P. Davami. 2016. Inves- ology at the University of Jordan.
tigation on corrosion stratigraphy and morphology in
some Iron Age bronze alloys vessels by OM, XRD and
SEM–EDS methods. Applied Physics A 122: 262.
Papadopoulou, O., P. Vassiliou, S. Grassini, E. Angelini,
and V. Gouda. 2016. Soil‐induced corrosion of ancient
Roman brass – A case study. Materials and Corro-
sion 67(2): 160–69.
Ravichandran, R. and N. Rajendran. 2005. Influence of
benzotriazole derivatives on the dezincification of 65–35
brass in sodium chloride. Applied Surface Science 239(2):
182–92.
Scott, D. 1990. Bronze disease: A review of some chemical
problems and the role of relative humidity. Journal of
the American Institute for Conservation 29(2): 193–206.
Scott, D. 1992. Metallography and microstructure of
ancient and historic metals. Los Angeles: The Getty
Conservation Institute.
Scott, D. 2002. Copper and bronze in art: Corrosion,
colorants, conservation. Los Angeles: The Getty Conser-
vation Institute.
Scott, D. and R. Schwab. 2019. Metallography in archae-
ology and art. Cham: Springer.
Selwyn, L. 2004. Metals and corrosion: A handbook for
the conservation professional. Ottawa: Canadian Conser-
vation Institute.
Thornton, C.P. 2007. Of brass and bronze in prehistoric
Southwest Asia. In Metals and mines: Studies in archae-
ometallurgy, eds. S. La Niece, D. Hook, and P. Craddock,
123–35. London: Archetype Publications in association
with the British Museum.
Yohai, L., W.H. Schreiner, M. Vázquez, and M.B. Valcarce.
2011. Surface characterization of copper, zinc and brass
in contact with tap water inhibited with phosphate
ions. Applied Surface Science 257(23): 10089–95.
Authors
Ahmad N. Abu-Baker is the vice-dean of the Faculty of
Archaeology and Anthropology at Yarmouk University
in Irbid, Jordan. He is an associate professor of metals
conservation at the Department of Conservation and
Management of Cultural Resources.
Corrosion Analysis and Assessment of a Collection of
Archaeological Copper Alloy Objects from Sagzabad,
Northern Iran
Omid Oudbashi* Sepehr Bahadori Ahmad Aliyari
Department of Conservation of Department of Conservation of Department of Archaeology,
Cultural and Historical Properties, Cultural and Historical Properties, University of Tehran
Art University of Isfahan Art University of Isfahan Tehran, Iran
Isfahan, Iran Isfahan, Iran [email protected]
Department of Scientific Research, [email protected] *Author for correspondence
The Metropolitan Museum of Art
New York NY, USA
[email protected]
Abstract
Among the most important objects excavated from the microscopy with energy-dispersive X-ray spectroscopy. The
archaeological site of Sagzabad in northern Iran is a collection results revealed active corrosion or bronze disease due to the
of metal objects that includes a significant number of copper burial of the objects in Cl−containing soil, which caused the
alloy objects. The objects were excavated during archaeological formation of copper oxide and copper chloride compounds.
activities at the site that spanned approximately 50 years. They A lack of environmentally controlled storage has exacerbated
are now stored at the Institute of Archaeology, University of the corrosion processes over several decades, leading to a
Tehran. The condition of the copper alloy objects is not ideal, transformation of the remnant metallic structure into copper
and evidence of active corrosion is visible on some of them. corrosion products, and in some cases, the formation of new
The environmental conditions appropriate for the storage copper chloride powdery compounds on the surface. Based on
and long-term preservation of these objects have not been the condition of the objects, a preservation strategy consisting
established. The aim of this study was to determine the of preventive conservation measures for the collection is
conservation state of the copper alloy objects using analytical proposed.
tools. In addition to an assessment of the storage/display
environment, microscopic observations were used to determine Keywords
the condition of the objects. Eight samples were selected for archaeological copper alloys, corrosion, bronze disease,
microanalytical study by metallography and scanning electron preventive conservation
Introduction
The archaeological site of Sagzabad, also known as Tappeh of Tehran, Zagheh was identified as a Neolithic village
Sagzabad, is located in the Qazvin Plain in northern dating to the 6th and 5th millennia BC, while Qabristan
Iran (Figure 1a). This plain is a very important region was a settlement from the Chalcolithic period (5th and
of prehistoric archaeological research in Iran, as it thus 4th millennia BC) that was abandoned in the early
far includes 30 archaeological sites containing archae- 3rd millennium BC. Sagzabad is a large settlement that
ologically significant remains (Fazeli et al. 2005). The includes graveyards and dates from the Late Bronze Age
Sagzabad site, together with Zagheh (Tappeh Zagheh) and Iron Age (Negahban 1973 and 1977).
and Qabristan (Tappeh Qabristan), was one of the most
important settlements in this region. At all three, the During the excavation of Sagzabad a collection of metal
stratigraphic layers form a sequence ranging from the objects consisting of more than 200 copper alloy and a
Neolithic to the Iron Age (6th to 1st millennium BC). few iron objects was recovered. The collection includes
The sites are located close to each other, less than 3 km weapons, tools and decorative objects made from copper,
along a straight line between Qabristan and Zagheh, with tin bronze and arsenical copper (Mortazavi et al. 2011,
Sagzabad located in between (Figure 1b, 1c). Based on the Oudbashi et al. 2021). All objects were recovered in
findings of archaeological excavations that began in the the 1970s during archaeological excavations at the site
1970s by the Institute of Archaeology at the University directed by E.O. Negahban.
Figure 1. (a) Map of Iran

and the location of Qazvin
Plain in north-central Iran.
(b) Schematic view of the
location of three important sites
in the Qazvin Plain: Zagheh,
Qabristan and Sagzabad
(Negahban 1977). (c) Aerial
view of the region showing the
location of the archaeological
sites in the Qazvin Plain (image
credit: map.google.com)
However, the objects have been stored under less-than- metallic substrate has been completely converted into
ideal conditions, in storage enclosures such as boxes corrosion products (Figure 2b–e). This shows that the
and plastic bags (Figure 2a). A few objects are displayed burial environment was corrosive and that the storage/
in showcases in the small museum of the Institute. No display environment at the Institute was not appropriate
climate control system is installed in the storage and for the long-term curation of the collection. Thus, the
display spaces and no monitoring process was ever estab- authors initiated a multi-analytical study to characterise
lished to evaluate the objects. Evidence of active corrosion the corrosion morphology and processes and to assess
is visible on most of the objects and in some cases the the conservation conditions.

To study the corrosion morphology and assess the
condition of the objects, eight representative objects were
selected from the collection: two arrowheads, a ring, a
rod, a pin, a vessel, a fragment of an unknown object and
a broken pipe (Figure 3, Table 1).
Microscopic and microanalytical investigations were

conducted using a small sample (~2 mm thick) cut
from each object using a jeweller’s saw. The sample
was mounted in a two-part epoxy resin, and ground
Figure 2. (a) Copper alloy objects excavated from Sagzabad. The objects
are stored in plastic bags, untreated and lacking preventive conservation
treatments. (b–e) Different groups of objects have been placed in a
single bag and show active corrosion, pale-green corrosion products Figure 3. The eight copper alloy objects analysed in this study. The
and spots on the surfaces of the objects. Some parts of objects are objects are from the Sagzabad collection, belonging to the Institute of
completely corroded Archaeology, University of Tehran
CORROSION ANALYSIS AND ASSESSMENT OF A COLLECTION OF ARCHAEOLOGICAL COPPER ALLOY OBJECTS FROM SAGZABAD, NORTHERN IRAN 233
Table 1. Results of SEM-EDS of all phases marked in Figures 6 and 7 (SEM-BSE micrographs), reported in wt%
Object type Analysis Cu O Cl S Sn As Fe Pb Mg Al Si P K Ca

SAG-02 Arrowhead A 100 – – – – – – – – – – – – –
B 79 7.9 13 – – – – – – – – – – –
C 97 2.8 – – – – – – – – – – – –
D 36 25 1.5 – – – 3.8 – 1.3 4.7 25 – 1.3 1.5
E 83 1.0 16 – – – – – – – – – – –
SAG-04 Ring A 67 13 3.1 3.1 – – 0.4 – – 1.7 9.8 – 0.91 0.64
B 69 8.2 – 8.8 5.0 7.3 1.8 – – – – – – –
C 73 15 2.0 0.87 2.1 5.9 0.97 – – – – – – –
D 58 12 6.7 1.9 8.7 9.7 3.4 – – – – – – –
E 74 9.1 12 0.46 0.48 3.6 0.48 – – – – – – –
F 35 5.5 4.5 – 8.1 18 3.21 26 – – – – – –
G 18 10 – 1.9 46 15 7.78 – – – 0.9 – – –
SAG-05 Arrowhead A 100 – – – – – – – – – – – – –
B 97 3.2 – – – – – – – – – – – –
C 76 – 24 – – – – – – – – – – –
D 97 2.9 0.35 – – – – – – – – – – –
E 81 7.6 11 – – – – – – – – – – –
SAG-06 Fragmented rod A 8.3 41 0.19 2.3 – – 2.0 – 0.58 3.1 19 – 1.2 22
B 79 9.4 12 – – – – – – – – – – –
C 95 3.3 1.7 – – – – – – – – – – –
D 78 8.9 13 – – – – – – – – – – –
E 41 9.0 – – 1.0 33 – 16 – – – – – –
F 58 9.4 9.3 – 19 2.4 0.98 0.7 – – 0.29 – 0.02 0.66
SAG-07 Fragment A 81 6.9 12 – – – – – – – – – – –
B 84 3.6 1.7 – 9.7 – 1.5 – – – – – – –
C 16 9.4 0.05 – 71 – 3.9 – – – – – – –
SAG-08 Fragment of vessel A 90 – – – 9.7 – – – – – – – – –
B 88 1.2 – – 11 – – – – – – – – –
C 89 – – – 11 – – – – – – – – –
D 91 – – – 8.5 – – – – – – – – –
E 86 – – – 14 – – – – – – – – –
F 74 2.7 – – 23 – – – – – – – – –
G 92 – – – 8.5 – – – – – – – – –
SAG-09 Pipe A 96 – – – 3.7 – – – – – – – – –
B 76 – 24 – – – – – – – – – – –
C 69 1.6 17 – 12 – – – – – – – – –
D 97 3.2 – – – – – – – – – – – –
SAG-10 Pin A 100 – – – – – – – – – – – – –
B 97 3.1 0.2 – – – – – – – – – – –
C 66 14 3.4 – – – – – – – 5.9 5.9 0.87 3.5
and polished with abrasive paper and diamond paste, The cross-sections were gold-coated for scanning elec-
respectively, to obtain a smooth and suitable surface. The tron microscopy coupled with energy dispersive X-ray
mounted cross-sections were initially observed without spectroscopy (SEM-EDS). The microstructure of each
etching and studied with low magnification optical sample was observed at different magnifications under
microscopy (OM), using a Zeiss reflected and trans- backscattered (BSE) mode. The chemical composition
mitted light microscope (model Primotech) with plane of the layers and small phases was determined by EDS.
polarised light (PPL) and crossed polarised light (XPL), All available phases were analysed using an area analysis
at the Central Laboratory of Art University of Isfahan. to obtain accurate chemical data from the metallic and
corroded phases. The analysis was conducted using a surface, referred to as the ‘limitos’ (Bertholon 2000), is
Philips XL30 scanning electron microscope, equipped visible between the corrosion layers [samples SAG-04
with a BSE detector and EDAX silicon drift detector (ring), SAG-06 (rod fragment) and SAG-07 (uniden-
(HV: 20 kv and spot: 5) in the SEM laboratory of the tified fragment)].
Faculty of Materials Engineering, Isfahan University • Partially corroded objects in which some parts of the
of Technology. metallic structure are preserved underneath layers
of corrosion. The corrosion layers vary in thickness
Results and discussion and no uniform corrosion is visible on the surface of
the respective objects [samples SAG-02 (arrowhead),
Optical microscopy
SAG-05 (arrowhead), SAG-08 (vessel fragment),
Figures 4 and 5 show PPL and XPL micrographs of a SAG-09 (pipe), SAG-10 (pin)].
cross-section of each object. Complex corrosion crusts
The corrosion crust is made up of corrosion layers with
can be clearly seen on the surfaces, and some objects
a wide range of colours, from greyish-white to red,
are completely corroded, consisting solely of corrosion
green and sometimes blue, thus revealing the different
products. The objects were accordingly classified in two
corrosion products that formed during burial and after
main groups based on their corrosion stratigraphy:
excavation.
• Completely corroded objects, in which the metallic
substrate has been completely converted into corrosion The microstructure and corrosion morphology of sample
products with different layers in various shades and SAG-08 are of particular interest (Figure 5). The object
colours. Some evidence of the limit of the original is partially corroded and a very thin layer of corrosion
Figure 4. Optical microscopy images of cross-sections from samples Figure 5. Optical microscopy images of cross-sections of samples SAG-07,
SAG-02, SAG-04, SAG-05 and SAG-06, showing the limit of the original SAG-08, SAG-09 and SAG-10, showing the limit of the original surface,
surface, under plane polarised light (PPL) and crossed polarised light (XPL) under PPL and XPL
covers the surface, whilst the metallic structure is well The SEM-BSE micrographs showed differences in the
preserved. A partial needle-like grain microstructure is corrosion layers between the two groups (Figure 6),
visible, with numerous spread metallic islands (phases). consistent with the differences in the chemical composi-
The corrosion rate is significantly lower than that of tion of the layers. Table 1 presents the EDS results of the
the other objects in this study, although the needle-like areas analysed, as shown in the SEM-BSE micrographs
matrix is corroded in some areas and is visible due to in Figure 6. Copper, oxygen and chlorine are the most
internal corrosion of the metallic structure, thus revealing frequent major constituents. Tin, arsenic and lead were
the grain shapes. detected in the internal corrosion layers and phases of
samples SAG-04, SAG-09, SAG-06, SAG-08 and SAG-07.
SEM-EDS Soil elements, such as Si, Al and Mg, were detected based
on the composition of the external corrosion layers.
SEM-EDS revealed several aspects of the morphology and
The microstructure of the completely corroded samples
mechanisms of corrosion. An analysis of the alloy compo-
includes a tin-rich phase in the centre of the cross-sec-
sition of five samples (SAG-02, SAG-05, SAG-08, SAG-09,
tions of samples SAG-04/G and SAG-06/F. This observa-
SAG-10) by semi-quantitative EDS (Table 1) showed
tion implies that both objects are made of tin bronze and
impure copper in three samples (SAG-02, SAG-05,
contain a central tin-rich phase, as previously observed
SAG-10) and tin bronze in the other two samples (SAG-08
in other archaeological tin bronze objects (Oudbashi and
and SAG-09). The concentration of tin is 3.7 wt% in
Emami 2010, Oudbashi 2015). The use of tin bronze in
sample SAG-09 but > 10 wt% in sample SAG-08.
Iran dates to the 3rd millennium BC (Late and Middle
Bronze Age) but it became widespread during the 2nd
and 1st millennia BC (Late Bronze Age and Iron Age)
(Pigott 2004, Oudbashi et al. 2021).
The results of the chemical analysis allowed two types of

corrosion layers to be distinguished: (i) corrosion layers
composed of copper and oxygen and (ii) corrosion layers
composed of copper and chlorine; with the additional pres-
ence of small amounts of oxygen and chlorine in all layers.
Accordingly, copper oxides and copper chlorides are the
main corrosion products, with their formation occurring
either during burial or during storage after excavation.
Figure 7. SEM-BSE micrograph of sample SAG-08 under higher

Figure 6. Scanning electron microscopy in backscattered mode (SEM- magnification shows the matrix and islands scattered in the
BSE) of cross-sections from all eight samples, including the marked microstructure. The chemical composition of the marked phases is
analysed phases reported in Table 1 reported in Table 1
SAG-08 was subjected to further analysis as its micro- corrosion layers, the main corrosion products are copper
structure differs in several respects from the other oxides (probably cuprite) and copper trihydroxychlorides
samples, as described above. The SEM-BSE micrograph (probably atacamite and paratacamite), visible as the
obtained at high magnification (200 ×) (Figure 7) shows red and green tonalities, respectively, seen in the OM
a microstructure similar to a typical so-called beta bronze micrographs (Figures 4, 5).
or white bronze (Scott 1991), which are high-tin bronze
The presence of a tin-rich corroded core in samples
alloys with a tin content of ~20 wt%. The alloy is heat-
SAG-04 and SAG-06 (and probably in sample SAG-07,
treated and then quenched to retain the less brittle beta
as observed in its OM micrograph) may be due to the
phase, which can be more easily worked and avoids the
selective dissolution of copper, thereby increasing the
risk of embrittlement (Scott 1991). This technology was
concentration of tin in the interior of the corroded
already well known ~1000 years ago in Iran, India and
structure. This phenomenon, known as copper leaching
the Far East (Korea) (Park and Gordon 2007, Oudbashi
in tin bronzes, has been observed in many archaeological
et al. 2017, Chattopadhyay et al. 2020). The earliest
objects buried in highly corrosive environments (Scott
examples of beta bronzes are from the Sassanid empire
2002, Oudbashi et al. 2010, Oudbashi 2015). The leaching
in Iran (3rd to 7th centuries AD) (Oudbashi et al. 2017)
of copper during burial followed by the internal oxidation
but the technique was widespread in the Iranian Plateau
of tin in the centre of the object corresponds to the type
during the Islamic period (10th to 14th centuries AD)
II corrosion of tin bronzes, as described in the classifi-
(Melikian-Chirvani 1974). Nevertheless, the chemical
cation system proposed by Robbiola et al. (1998). Those
analysis of the phases scattered in the microstructure of
authors together with Piccardo et al. (2007) proposed a
SAG-08 showed that the tin concentration is lower than
corrosion morphology classification system based on the
in beta bronzes. Specifically, while high-tin bronze objects
shape and sequence of the corrosion layers, the presence
have a tin concentration of ~20 wt%, the tin content of
or absence of the original surface and the thickness and
sample SAG-08 is ~10 wt% (Table 1). Thus, the discovery
chemical composition of the corrosion layers. According
of object SAG-08 in the Sagzabad prehistoric site raises
to this classification system, two types of corrosion are
many questions that remain to be explored.
visible in the tin bronze objects studied:
• Type I: This superficial, two-layered corrosion
Corrosion mechanisms and assessment
morphology is characteristic of alkaline and moder-
The analyses of the corrosion mechanisms, corrosion ately corrosive soil burial environments. The original
morphology and stratigraphy of the eight copper alloy shape (or original surface) of the artefact is retained,
objects provided several insights into the corrosion of and macroscopic traces of polishing or use may be
archaeological copper alloy objects. According to the present.
EDS results, two objects are made of tin bronze albeit • Type II: The corrosion morphology involves three
with differences in their tin concentrations (SAG-08 and layers and is characteristic of corrosive chloride-con-
SAG-09), and three objects are made of impure copper taining soil burial environments. The corrosion crust
(SAG-02, SAG-05, SAG-10). For the three completely is coarse (rough) and much thicker than in type I,
corroded objects, analyses of their corrosion layers possibly transforming the artefact into a multi-layered
showed the presence of tin-rich layers, suggesting bulk material without any remaining metal.
that they were originally made of tin bronze (SAG-04,
In the Sagzabad copper alloy there is a predominance of
SAG-06, SAG-07). The corrosion rate of some objects
type II, although, in some cases, there is evidence of the
led to the complete conversion of the metallic structure
original surface in the corrosion products.
into corrosion products. In addition, in nearly all of the
samples the high concentrations of chlorine were detected While the corrosion process may be partially influenced
in some of the corrosion layers. Together, these obser- by the original composition of the alloy, the analytical
vations indicate that the main corrosion mechanism in results and literature reports (for example, Robbiola et al.
the eight objects is bronze disease, an active corrosion 1998, Scott 2002, Piccardo et al. 2007) indicated that the
process driven by the presence of soluble chloride salts corrosion morphology and mechanisms of the sampled
and humidity in the burial environment (Scott 2002, objects were strongly influenced by the presence of chlo-
Selwyn 2006). Based on the chemical composition of the ride ions in the soil, leading to the formation of bronze
disease, completely corroded objects and considerable or even inside the storage boxes. However, due to budget
copper leaching during burial. Nevertheless, some objects limitations, mechanical methods of controlling the RH
are not heavily corroded, and a significant amount of their may not be possible in the short term.
metallic structure has been preserved under the corrosion
A plan to initiate ongoing monitoring of the collection is
crust. The corrosion layers in these objects (SAG-02,
needed. Conservators should also assess the condition of
SAG-09, SAG-10) are composed of copper oxides and
all materials in storage (including the metal collection) as
copper chlorides, likely reflecting the variable chloride
a prerequisite to deciding the best method for the long-
concentrations in different levels or areas of the soil and
term preventive conservation of the collection. Finally, an
thus the different corrosion rates, as observed at another
interventional conservation approach by conservators is
Iranian site (Haft Tappeh, southwestern Iran) with high
needed, including stabilisation treatments of the objects
chloride concentrations in the soil (Oudbashi 2015).
such as by desalination and inhibition methods. Indeed,
Consequently, both known corrosion morphologies,
this study of the copper alloy objects from a collection in
type I and type II, were observed in the Sagzabad copper
Iran demonstrates the importance of having conservators
alloy objects (Robbiola et al. 1998, Piccardo et al. 2007). at the field excavation sites and at museums, to evaluate
the corrosion/deterioration processes after excavations
Conservation strategy for the collection and during long-term storage.
The presence of bronze disease places the objects at high
risk of further corrosion, a situation made worse by the Conclusion
fact that, currently, no suitable storage conditions are In this study, the corrosion mechanisms and morpholo-
available at the Institute of Archaeology. Furthermore, gies of eight copper alloy objects excavated from the site
since their excavation, the majority of the copper alloy of Sagzabad in northern Iran were investigated using
objects have not been treated by conservators. Only a few microscopy and microanalytical methods. The interac-
objects are in display cases, and the majority of the copper tion between the aggressive burial environment and the
alloy objects are in storage (Figure 2). The archaeological copper alloy objects led to the formation of a variable
metal collection must be preserved in an environment range of corrosion morphologies, mainly the selective
with low relative humidity (RH) to control active corro- dissolution of copper in the tin bronze objects. Partially
sion. The corrosion processes affecting some of the thin corrosion crusts are observed in three objects. Most
objects have continued since their excavation or have even of the objects have undergone type II corrosion (i.e.
been exacerbated during their storage or display at the complete corrosion of the object) and bronze disease,
Institute over the last 50 years. Unfortunately, there are but there are also a few cases of type I corrosion. As the
no recorded data from the storage/display environment main corrosion products are copper oxides and chlorides,
of the collection and a monitoring strategy has yet to be the soil environment of the site was likely to have been
established. The institute is located in the north of Tehran, corrosive. This is supported by the high concentration
which is humid during winter (average RH between 50 of chlorine in the corrosion layers of all objects and the
and 80%) but much dryer in summer (RH between 15 significant amounts of soluble chlorides, inducers of
and 40%). The pale-green powdery corrosion products bronze disease. To address the active corrosion process
visible on the surface of some objects may be copper and the unsuitable storage/display conditions at the Insti-
trihydroxychlorides, formed after exposure to the high tute of Archaeology, University of Tehran, a preventive
RH in the storage or display environment. conservation plan, including proper storage cabinets
and boxes, as well as a method to control the RH in the
A preventive conservation plan that includes controlling
storage and display areas, were proposed.
the RH to < 45% (Scott 1990) is necessary together with
a storage method in which the objects (currently stored
Acknowledgements
together in old plastic enclosures) are stored separately
in appropriate boxes and cabinets made of neutral (such The authors are grateful to the reviewers for their
as polyethylene) or buffered materials. The RH can be comments and edits. This study was undertaken with
controlled by installing a dehumidification system or by the financial support of the Deputy of Research and
using silica gel or Art-Sorb boxes in the storage cabinets Technology, Art University of Isfahan, Iran.
References Park, J.-S. and R.B. Gordon. 2007. Traditions and tran-
sitions in Korean bronze technology. Journal of Archae-
Bertholon, R. 2000. La limite de la surface d’origine
ological Science 34(12): 1991–2002.
des objets métalliques archéologiques. Caractérisation,
localisation et approche des mécanismes de conservation. Piccardo, P., B. Mille, and L. Robbiola. 2007. Tin and
PhD dissertation, Université Panthéon-Sorbonne – Paris copper oxides in corroded archaeological bronzes. In
I, France. Corrosion of metallic heritage artefacts: Investigation,
conservation and prediction for long-term behaviour, eds.
Chattopadhyay, P.K., P.K. Datta, and B. Maji. 2020.
P. Dillmann, G. Béranger, P. Piccardo, and H. Matthiesen,
Forging technology of high-tin bronzes in ancient Bengal.
European Federation of Corrosion Publication 48,
Current Science 118(11): 1822–31.
239–62. Cambridge: Woodhead Publishing.
Fazeli, H., E.H. Wong, and D.T. Potts. 2005. The Qazvin
Pigott, V.C. 2004. On the importance of Iran in the study
Plain revisited: A reappraisal of the chronology of of prehistoric copper-base metallurgy. In Persia’s ancient
northwestern Central Plateau, Iran, in the 6th to 4th splendour, mining, handicraft and archaeology, eds. T.
millennium BC. Ancient Near Eastern Studies 42: 3–82. Stöllner, R. Slotta, and A. Vatandoust, 28–43. Bochum:
Melikian-Chirvani, A.S. 1974. The white bronzes of early Deutsches Bergbau-Museum.
Islamic Iran. Metropolitan Museum Journal 9: 123–51. Robbiola, L., J.M. Blengino, and C. Fiaud. 1998.
Mortazavi, M., A. Salehi Kakhki, M.A. Colozar, and H. Morphology and mechanisms of formation of natural
Tala’i. 2011. Preliminary metallurgical investigation of patinas on archaeological Cu-Sn alloys. Corrosion Science
copper-based artifacts at Tepe Sagzabad in Qazvin Plain, 40(12): 2083–111.
Iran (1500–800 BC). Iranian Journal of Archaeological Scott, D.A. 1990. Bronze disease: A review of some chem-
Studies 1: 49–59. ical problems and the role of relative humidity. Journal of
Negahban, E.O. 1977. Preliminary report of Qazvin the American Institute for Conservation 29(2): 193–206.
expedition: Excavation of Zaghe, Qabristan, Sagzabad Scott, D.A. 1991. Metallography and microstructure
(1971–72). Marlik 2: 26–44 (in Persian). of ancient and historic metals. Los Angeles, CA: Getty
Negahban E.O. 1973. Preliminary report on the excava- Conservation Institute.
tion at Sagzabad, 1970 season. Marlik 1: 1–9 (in Persian). Scott, D.A. 2002. Copper and bronze in art: Corrosion,
colorants, conservation. Los Angeles, CA: Getty Conser-
Oudbashi, O. 2015. Multianalytical study of corrosion
vation Institute.
layers in some archaeological copper alloy artefacts.
Surface and Interface Analysis 47: 1133–47. Selwyn, L.S. 2006. Corrosion of metal artifacts in buried
environments. In ASM handbook volume 13C. Corrosion:
Oudbashi, O. and S.M. Emami. 2010. A note on the
Environments and industries, eds. S.D. Cramer and B.S.
corrosion morphology of some Middle Elamite copper
Covino, Jr., 306–22. Materials Park, OH: ASM Interna-
alloy artefacts from Haft Tappeh, South-West Iran. Studies
tional.
in Conservation 55(1): 20–25.
Oudbashi, O., A. Hasanpour, A. Jahanpoor, and Z. Authors
Rahjoo. 2017. Microscopic and microanalytical study
Omid Oudbashi has a BSc, MA and PhD in conserva-
on Sasanian metal objects from Western Iran: A case
tion of cultural and historical properties from the Art
study. STAR: Science & Technology of Archaeological
University of Isfahan, Iran, and is currently associate
Research 37: 194–205.
professor in the Department of Conservation of Cultural
Oudbashi, O., M. Mehofer, S. Bahadori, A. Aliyari, and and Historical Properties at the same university and
H. Tala’i. 2021. The emergence and spread of tin bronze senior scientist in the Department of Scientific Research
alloying in prehistoric Iran – The LBA metallurgy in Sagz- at the Metropolitan Museum of Art, USA. His research
abad, Northern Iran. In Proceeding of the 5th International interests are the deterioration of archaeological objects,
Conference “Archaeometallurgy in Europe” (AIE2019), especially the corrosion mechanism of ancient metals in
19–21 June 2019, Miskolc, Hungary, eds. B. Török and A. terrestrial, marine and outdoor environments; preventive
Giumlia-Mair, 253–70. Drémil-Lafage: Editions Mergoil. and archaeological conservation; archaeometallurgy and
ancient metalworking; metallography of ancient and

historic metals; and ancient techniques and methods of
production and decoration in other inorganic materials.
Sepehr Bahadori has completed a BSc in conserva-

tion of cultural and historical properties and a MSc in
archaeometry of inorganic materials at the Art University
of Isfahan, Iran. His research interests include the study
of archaeometallurgical remains, such as prehistoric
metals, ores and slags, using multi-analytical methods,
such as chemical analysis and metallography, to obtain
knowledge related to the technology and production
methods used by ancient craftsmen.
Ahmad Aliyari received a MA in archaeology from the

University of Tehran and a PhD in prehistoric archae-
ology from the Art University of Isfahan. He is currently
a chief researcher in the Institute of Archaeology at the
University of Tehran. He previously held the positions
of vice director, head of the Research Department and
head of the Research Archaeological Museum at the
Institute and was a key collaborator on several Institute
projects. He has participated as a researcher and member
of several archaeological teams excavating in the Qazvin
Plain and the Tehran Plain in Iran, and the Tepe Sanjar-
Shah in Tajikistan.
The Effect of Synthesis Routes on the Ability of
Akaganeite to Corrode Iron
Nicola J. Emmerson* Kimberly Roche Johanna C. Thunberg

Cardiff University Warren Lasch Conservation Center Cardiff University
Cardiff, UK Charleston SC, USA Cardiff, UK
David E. Watkinson Jerrod H. Seifert *Author for correspondence

Cardiff University Cardiff University
Cardiff, UK Cardiff, UK
Abstract
Akaganeite (βFeOOH) occludes chloride within its crystal structure containing a mixture of βFeOOH and iron powder. The oxidation
and adsorbs it onto its surface during its formation. βFeOOH was rate of iron was directly related to the mass% of adsorbed
synthesised by solid-state and aqueous precipitation routes. The chloride removed by the aqueous wash. This contributes to an
amount of adsorbed chloride removed by a single aqueous wash understanding of both iron corrosion and the development of
was measured. The impact of the synthesised βFeOOH on the conservation treatments.
oxidation rate of iron was determined using a remote oxygen
sensing technique to record oxygen concentrations within a Keywords
sealed vessel, internally controlled to 80% relative humidity and iron, corrosion, archaeological, akaganeite, chloride
Introduction
Conservation practitioners devise and apply treatments results are transferable to conservation practice. This
to fulfil perceived goals and require quantitative and research considers these questions for the experimental
qualitative research to evidence whether these are use of synthesised akaganeite (βFeOOH) to parallel the
attained. The research design dictates the value of its role of naturally formed βFeOOH in the corrosion and
outcomes to conservation practice and, if this design treatment of archaeological iron.
is flawed, it can foster misinterpretation and increase
risk in conservation practice. Within any experimental Archaeological iron: The corrosion problem
design, the sample introduces variables that impact on Chloride plays a major role in the post-excavation corro-
data quality, range, reproducibility and relevance to the sion of archaeological iron. Turgoose’s (1982a) corrosion
question being addressed. While analogue samples can model for buried archaeological iron reports Cl− ions
provide reproducibility, they are often problematic as accumulating at the metal surface to provide charge
they seek to replicate outcomes that may have taken years balance for anodically generated Fe2+. Above this, there
or centuries to achieve, such as patinas and corrosion is a dense corrosion product layer (DPL) that normally
product layers. In order to study corrosion or decay comprises a goethite, magnetite and maghemite matrix,
processes, research may require synthesis of compounds overlaid by an altered porous layer of iron corrosion
that are produced naturally on objects. While this often products and soil particles (Reguer et al. 2006, Neff et al.
seems straightforward to achieve, questions remain 2007). Localisation of chloride occurs at anodes on the
regarding whether the synthesis routes produce similar metal, due to strips of magnetite that conduct electrons
chemistry and morphology to the compounds formed on into the DPL where ingress of moisture and oxygen
objects, whether formation routes for analogues produce support the cathode reaction (Neff et al. 2007). Highly
the same end product, how well an analogue represents mobile Cl− ions from the soil pass through the corrosion
contextual reality and, consequently, whether research matrix and associate with Fe2+ ions at the anode. They
THE EFFECT OF SYNTHESIS ROUTES ON THE ABILITY OF AKAGANEITE TO CORRODE IRON 241
can exist in solution and Reguer et al. (2006) identified upwards (Watkinson and Lewis 2005, Watkinson et al.
the β-Fe2(OH)3Cl iron hydroxy-chloride ferrous phase 2019). The prevalence and form of chloride in βFeOOH
occurring next to the metal surface, postulating that it is will impact on its role in corrosion processes, as will the
a precursor for the formation of the oxidised ferric phase access of moisture and oxygen to its location. Preventing
β-FeOOH, which is detected above it. Turgoose (1982b) its formation post-excavation is essential, since its growth
identified β-FeOOH on archaeological iron post-exca- at the metal surface cracks overlying corrosion product
vation, attributing it to oxidation of a Fe2+/Cl− solution layers, which then spall off and destroy the heritage value
at the metal surface by atmospheric oxygen and fresh of an object.
oxidation of iron with chloride as an electrolyte. Iron
oxidation generates more Fe2+ ions and their hydrol- Controlling corrosion of chloride-infested iron
ysis lowers pH in the thin moisture film which makes
An obvious and proven option for non-interventive
βFeOOH formation thermodynamically favourable in the
action to control corrosion is desiccation (Turgoose
prevailing high Cl− ion concentration (Turgoose 1982b,
1982b, Watkinson et al. 2019, Thunberg et al. 2021).
Rémazeilles and Refait 2007).
Desalination to remove electrolytes has been the main-
βFeOOH occludes chloride in its crystal structure and stay of interventive conservation treatments for over a
adsorbs it onto its surface (Mackay 1962). The overall century, yet its effectiveness has long been questioned.
chloride content can vary, but substantial amounts are Determining an end point for desalination treatments
often surface adsorbed. Reguer et al. (2009) produced involves monitoring chloride release; when outward
βFeOOH with 12%mass chloride, attributing 6.7%mass diffusion of chloride ceases, measured as concentra-
of this to being within the crystal structure as determined tion of chloride in the wash bath, this is considered a
by X-ray diffraction (XRD) structural analysis. They proxy measurement that all chloride is removed. To be
reduced chloride content to 4.5%mass after 10 washes in successful, desalination will need to remove all soluble
deionised water, which removed the adsorbed chloride chlorides, adsorbed chloride from βFeOOH and access
and reduced the chloride in the crystal structure by 20%. deep-seated chloride beneath corrosion layers and in pits.
The mobility of the adsorbed chloride in water and water This has not been shown to be successful, as chloride
vapour makes it readily available for corrosion of adja- is detected within archaeological iron post-treatment
cent iron (Turgoose 1982b, Watkinson and Lewis 2005, (Rimmer et al. 2012). While this may be occluded in the
Thickett and Odlyha 2013, Watkinson and Emmerson βFeOOH where it does not support iron corrosion, anec-
2016). Lowering the amount of adsorbed chloride present dotal reporting of post-treatment instability of washed
on βFeOOH by aqueous washing treatments reduces iron and detection of large quantities of fresh βFeOOH
the extent and rate of corrosion it produces on iron on fragmenting objects indicates not all chloride remains
(Watkinson and Emmerson 2016). While the residual in this inaccessible form. Following desalination, it
chloride in the crystal structure of βFeOOH is unavailable is realistic to expect iron to remain unstable but have
to support corrosion, it remains a risk if transformations enhanced stability compared to its untreated state and
occur and it is released. to corrode at a significantly slower rate.
βFeOOH has been identified on archaeological objects

Impact of βFeOOH on the corrosion rate of iron
(Thickett and Odlyha 2013) many times post-excavation
and poses a major corrosion threat as it and any ferrous Initial research used βFeOOH, produced by solid-state
chloride present are hygroscopic, hence their corrosion synthesis involving mixing iron powder with FeCl24H2O
threat is a function of prevailing relative humidity (RH) at 92% RH, to investigate its impact on the corrosion
(Watkinson and Lewis 2005, Watkinson et al. 2019). rate of iron at low humidity (12%–35%) (Watkinson and
For βFeOOH mixed with iron powder, corrosion ceases Lewis 2005). Since oxidation of iron was recorded by
below 15% RH (Watkinson and Lewis 2005, Thickett mass gain, which records both oxidation and hydration,
and Odlyha 2013) and for ferrous chloride below 20% this prevented production of an oxidation rate graph. A
RH (Turgoose 1982b, Watkinson and Lewis 2005). For remote sensing technique for quantitative measurement
both compounds, it is very slow below 30% RH then of oxygen concentration in a body of air (Matthiesen
rises, finally escalating exponentially from 60% RH 2007, Matthiesen and Wonsyld 2010) has been adopted to
provide a standardised reproducible experimental design Objectives:

for determining oxygen within an enclosed volume of air • To synthesise βFeOOH by solid-state and precipitation
(Emmerson et al. 2021). This has been used to study the techniques, and then wash these in cold water.
impact of washing on synthesised βFeOOH (Watkinson • To record the oxidation rate of iron mixed with
and Emmerson 2016) and corrosion rates of archaeolog- synthesised βFeOOH using oxygen consumption and
ical iron as a function of ambient RH (Watkinson et al. relate this to mass% of chloride removed during the
2016, Watkinson et al. 2019) and is utilised in this study. wash process.
The data generated when using synthesised βFeOOH

Method
to study corrosion rates is likely to be influenced by
the synthesis process. Solid-state (Turgoose 1982b) and βFeOOH was prepared either by precipitation or solid-
precipitation (Mackay 1962, Reguer et al. 2009) methods state synthesis (see Table 1). These were either past
are reported synthesis routes, but each can be expected to or newly synthesised samples with the synthesis year
produce varied chloride content, crystal shape and crystal recorded in the sample labelling.
size in the end product. Crystal morphology influences the
For all βFeOOH samples, synthesis products were iden-
amount of surface-adsorbed chloride, as does washing to
tified immediately prior to this experimental study using
refine the synthesis product, although this may not always
a PANalytical X’Pert Pro (CuΚα) X-ray powder diffrac-
be carried out. Amounts of occluded chloride can also
tometer and the data was interpreted using PANalytical
vary. Post-synthesis purification washing may be either
X’Pert HighScore software. Imaging of βFeOOH crystals
aqueous or anhydrous and it aims to remove excess reac-
was carried out using a CamScan Maxim 2040 scanning
tants, specifically chloride compounds. Both the solvent
electron microscope (SEM) equipped with Oxford
used and the number of washes will influence the amount
Instruments energy-dispersive X-ray spectrometer using
of adsorbed chloride on βFeOOH, which is normally
secondary electron imaging. All βFeOOH sample mass
reported as a mass percentage (mass%) of the βFeOOH.
was recorded using a Sartorius Secura 225D-1S balance
Aim: (readability 0.01 mg). Washing involved placing 0.05 g
• To compare how differing synthesis methods for of each sample individually in beakers, adding 10 mL of
βFeOOH influence its impact on the corrosion rate deionised water and then stirring continuously for 2 h
of iron and consider how this informs understanding at room temperature. This will not remove all adsorbed
of post-excavation corrosion and treatment of archae- chloride but offers a guide to the amount of chloride
ological iron. available for corrosion processes on each βFeOOH
Table 1. Synthesis methods for βFeOOH samples used in experimental work
Sample-synthesis Post-synthesis
Synthesis method
year washing
(P1-2021) Precipitation from solution Acetone
3 L of 0.1 M FeCl36H2O solution in deionised water heated to 70°C for 48 h. Filtered in a Buchner funnel
(P2-2021) under vacuum. Rinsed with acetone x3. Ground with pestle and mortar for 10 minutes.
(S1-1983) Solid-state transformation Acetone
Iron powder (99.0% pure) mixed with equal mass FeCl24H2O spread in petri dish enclosed in desiccator
with beaker containing saturated solution of Na2CO3 to produce 92% ± 5 RH.
(S2-2000) Solid-state transformation None
25 g iron powder (Aldrich GPR™ 99.0% pure) mixed with 25 g FeCl24H2O (BDH Chemicals 99.0% pure)
spread in petri dish enclosed in desiccator with beaker containing saturated solution of Na2CO3 to produce
92% ± 5 RH. No post-synthesis washing.
(S3-2014) Solid-state transformation Acetone
100 g iron powder (Aldrich GPR™ 99.0% pure) mixed with 100 g FeCl24H2O (Sigma > 99% pure) spread
in petri dish enclosed in desiccator with beaker containing saturated solution of Na2CO3 to produce
92% ± 5 RH. Humidity cabinet aerated regularly to ingress oxygen for 4 months. Residual FeCl24H2O
washed off with acetone using a Buchner funnel. Wash volumes unknown.
NB: for solid-state, desiccator opened for short period of time daily to ingress oxygen over unspecified
period of months. This avoids danger of depleting oxygen to the point where magnetite (Fe3O4) is the
thermodynamically favoured product.
sample. Using a Microman CP M1000E pipette, 500 µL Results and discussion

(5 µL max systematic error and 3 µL max random error)
of sample solution was removed from each wash sample Composition
solution and chloride concentration was measured XRD confirmed the samples as comprising only
using a Sherwood Model 926 MKIII industrial chloride βFeOOH except for S2-2000, which contained a small
analyser (accuracy: ± 3 mg/L at 200 mg/L). This was amount of iron oxide hydroxide (αFeOOH, reference
scaled up to determine the βFeOOH mass% chloride pattern 01-076-7158 5) (Figure 1; Table 2). Where there
for each synthesis method. The reproducibility of the are mixtures of compounds and fluorescence, detection
analyser was measured using 177 ppm NaCl solution, limits are high, at potentially 15%. The βFeOOH match
before and after measuring a sample. Each solution was in all cases was reference pattern 00-042-1315.
sampled three times for reproducibility.
The βFeOOH samples were synthesised at differing times
To measure the impact of βFeOOH on the oxidation rate over a 38-year period (Table 1) and were later stored
of iron, 0.05 g of synthesised βFeOOH was mixed with either in sealed screw-top glass jars or, in the case of
0.05 g of Aldrich GPR 99.0% pure iron powder, ground for S1-1982, a clip-top plastic container. XRD performed
1 min in an agate mortar and then transferred to a plastic
weigh boat that was placed in a Mason Ball reaction vessel
Table 2. X-ray diffraction data and mass percentage chloride removed
with an airtight screw-top lid containing a Ruthenium
by washing in cold deionised water
oxygen sensor spot and 160 g of silica gel preconditioned
to 80% RH (Emmerson et al. 2021). Since this was to be Sample XRD Washing extraction
(mass%)
a comparative rate test, choosing a single, fixed high RH P1-2021 Reference pattern: 00-042-1315 1.5%
allowed comparison of the corrosion rate between the Iron oxide chloride
various βFeOOH powders and produced results quickly. Empirical formula: Cl1.3Fe8H9.7O16
Reaction vessels were placed in a Binder KBF720 climatic P2-2021 Reference pattern: 00-042-1315 2.0%
Iron oxide chloride
chamber controlled to 50% RH and 20°C. A fibre-optic
S1-1983 Reference pattern: 00-042-1315 2.6%
cable attached to a PreSens Oxy-1 SMA oxygen meter Iron oxide chloride
was fixed to a support, placed against the sensor spot and S2-2000 Reference pattern: 00-042-1315 5.9%
the fluorescence of the spot was recorded every 5 min in Iron oxide chloride
Reference pattern: 01-076-7158 5
the dark interior of the climatic chamber (Emmerson Iron oxide hydroxide FeO(OH)
et al. 2021). Quenching of fluorescence is proportional S3-2014 Reference pattern: 00-042-1315 3.7%
to oxygen pressure in the reaction vessel. Iron oxide chloride
Figure 1. βFeOOH samples: X-ray diffraction data; left to right – (top) P1-2021, P2-2021; (centre) S1-1983, S2-2000; (bottom) S3-2014
in 2021 confirmed all samples contained βFeOOH and synthesis used in the study reported here, Mackay
S2-2000 included an undetermined amount of αFeOOH, (1962) identified elongated crystals in naturally formed
which is a change from the original XRD data collected in βFeOOH and showed overall that crystal shapes for
2000 which identified only βFeOOH (Table 2; Figure 1). βFeOOH differ according to synthesis method and
Gilberg and Seeley (1981) reported on the thermody- time. Rod-shaped 1 µm-length crystals were identified
namic instability of βFeOOH relative to αFeOOH but in archaeological βFeOOH, which formed post-excava-
cited no examples. Thickett and Odlyha (2013) reported tion by transformation of β-Fe2(OH)3Cl on an iron bar
atmospheric transformation of synthesised βFeOOH to from a marine context (Bayle et al. 2016). The shape of
αFeOOH occurring over a 6-month period, with higher the βFeOOH crystals is important, since surface area
RH producing more transformation and the presence of influences adsorption of chloride.
ethanoic acid vapour also promoting transformation.
They detected similar transformations on βFeOOH
samples formed on archaeological iron on display in
showcases. The cause of the XRD assay discrepancy
for S2-2000 between the year 2000 and 2021 cannot be
readily evidenced. Time cannot account for it, as the
older S1-1983 sample comprised only βFeOOH at its
time of synthesis and this has remained unchanged,
although S2-2000 differs in that it was not washed to
purify it post-synthesis (Watkinson and Lewis 2005).
Potentially, the post-synthesis environment and proce-
dures for washing post-synthesis may have influenced
the transformation.
Unwashed βFeOOH has already been shown to corrode Figure 2. SEM images of the crystal structures of the βFeOOH samples
iron (Turgoose 1982b, Watkinson and Lewis 2005, synthesised by precipitation and the solid-state reaction (4000×
magnification)
Thickett and Odlyha 2013) and its transformation
increases its ability to corrode iron by releasing chloride Adsorbed chloride on the βFeOOH, measured as the
from within its crystal structure. Similarly, if βFeOOH on %mass chloride removed by single cold water washing,
archaeological iron has had its adsorbed chloride mostly ranged from 5.9% to 1.5%mass (Table 2). βFeOOH
or entirely removed by desalination treatments, transfor- samples formed by precipitation had low mass% of surface
mation will reactivate its corrosion threat by releasing adsorbed chloride, which differed (2 and 1.5 mass%)
residual occluded chloride. At present, the transformation despite their similar synthesis (Table 1). As might be
of corrosion products on washed archaeological iron does expected, due to it being unwashed following synthesis,
not seem to have been studied. S2-2000 had the most adsorbed chloride (5.9 mass%),
while the acetone-washed S1-1983 and S3-2014 had
Crystal structure and adsorbed chloride 2.6 and 3.7 mass% chloride, respectively. Differences
in crystal shape and/or acetone wash procedures could
Precipitated βFeOOH (P1-2021) produced smaller
explain why the solid-state synthesised βFeOOH had
and more rounded crystal clusters compared with the
more adsorbed chloride or, for S2-2000, transformation
larger and more elongated crystals in the solid-state
to αFeOOH released occluded chloride.
synthesis samples (S1-1982, S-2000, S3-2014) (Figure 2).
This outcome tends to align with reported shapes The short wash time employed here has been shown
and comparative sizes of crystals for precipitated and to remove a significant amount of adsorbed chloride
archaeological samples. Somatoidal crystals have been (Watkinson and Emmerson 2016). This identifies the
reported for the exact same precipitation synthesis high mobility of the adsorbed chloride and leads to the
(García et al. 2004) and Bayle et al. (2016) identified 0.5 expectation that damp conditions will readily mobilise
µm rice-shaped crystals in their precipitated βFeOOH. it, which was confirmed by the oxygen consumption
While no crystal shape data is reported for solid-state corrosion rate tests carried out here (Figure 3). Once
βFeOOH receives a single cold water wash, its impact on

the corrosion rate of iron falls significantly (Watkinson
and Emmerson 2016).
Corrosion of iron by βFeOOH

Reaction of all the βFeOOH samples with iron, recorded
as oxygen consumption, was initially fast (Figure 3),
aided by the large surface area of the iron powder and
mobility of the adsorbed chloride at 80% RH. This
created an electrolyte at the interface between the
iron powder and the βFeOOH, producing Fe2+ ions
which hydrolysed, lowering the pH. According to the Figure 4. Oxygen consumed by βFeOOH/iron powder mixtures over the
ratios of Fe2+ and Cl− and the prevailing pH, different 16-hour logging time as a function of chloride removed by aqueous
washing tests (mbar converted to mg, see Emmerson et al. 2021)
forms of FeOOH are produced: high concentrations
of Fe2+ and Cl− favour production of only βFeOOH;
Context for conservation practice
intermediate concentrations of Fe2+ favour production
of αFeOOH; and low Fe 2+ concentrations produce When carrying out experimental studies using synthe-
γFeOOH (Rémazeilles and Refait 2007). Mixtures of sised βFeOOH, extrapolating data quantitatively to
these compounds might be expected to occur in the archaeological iron contexts is challenging. Quantifying
corrosion formed here, since pH is expected to vary as the risk of corrosion of archaeological iron as a rate factor
Fe2+ is converted to corrosion products and hydrolysis is of synthetic βFeOOH is not possible, as the amount of
limited. Iron is also depleted, and the crusts of corrosion βFeOOH on objects, its adsorbed chloride mass and its
formed influence oxygen access to the remaining iron, accessibility to oxygen and moisture are all unknown.
influencing oxygen access and hence its consumption For investigating treatments aimed at removing adsorbed
and the corrosion rate (Figure 3). Plotting the amount chloride from βFeOOH, using synthesised βFeOOH can
of adsorbed chloride removed from the βFeOOH by offer insights into the impact of treatment parameters like
the cold water wash against the total amount of oxygen solution composition, concentration, temperature and
consumed by unwashed βFeOOH/iron powder mixtures time on chloride removal and βFeOOH transformation.
over a 16-hour period, at which point reactions either
cease or are significantly slowed, revealed a linear rela- Conclusion
tionship (Figure 4). As might be expected, the greater
The corrosion risk to iron due to the presence of βFeOOH
the amount of adsorbed chloride, the faster and more
is a function of the mass of chloride adsorbed on the
extensive the corrosion of the iron.
βFeOOH surface. This can differ appreciably according
to the conditions in which βFeOOH forms and may
change with time if transformations occur that release
occluded chloride. Since solid-state formation routes are
expected to occur on iron corroding in high humidity, the
mass% of surface adsorbed chloride is likely to be high,
as compared to βFeOOH formed from ionic solutions.
In designing experimental studies using synthesised
βFeOOH, its synthesis route should be taken into account
when extrapolating results to practical contexts within
conservation.
Figure 3. Oxygen consumption data for βFeOOH samples. Since the References
volume of all components of the reaction system in each case are
identical, the mbar oxygen consumed can be used to compare rates Bayle, M., P. de Viviés, J.-B. Memet, E. Foy, P. Dillmann,
within this sample set and D. Neff. 2016. Corrosion product transformations in
alkaline baths under pressure and high temperature: The diffraction and absorption spectroscopy. Corrosion
sub-critical stabilisation of marine iron artefacts stored Science 51(12): 2795–802.
under atmospheric conditions. Materials and Corrosion Rémazeilles, C. and P. Refait. 2007. On the formation of
67(2): 190–99. β- FeOOH (akaganéite) in chloride-containing environ-
Emmerson, N.J., J.H. Seifert, and D.E. Watkinson. ments. Corrosion Science 49(2): 844–57.
2021. Refining the use of oxygen consumption as a Rimmer, M., D. Watkinson, and Q. Wang. 2012. The
proxy corrosion rate measure for archaeological and efficiency of chloride extraction from archaeological iron
historic iron. The European Physical Journal Plus 136: objects using deoxygenated alkaline solutions. Studies in
art. 546. Conservation 57(1): 29–41.
García, K.E., A.L. Morales, C.A. Barrero, C.E. Arroyave, Thickett, D. and M. Odlyha. 2013. The formation and
and J.M. Greneche. 2004. Magnetic and crystal struc- transformation of akaganeite. In Metal 2013: Proceedings
ture refinement in akaganeite nanoparticle. Physica B: of the Interim Meeting of the ICOM-CC Metals Working
Condensed Matter 354(1–4): 187–90. Group, Edinburgh, 16–20 September 2013, eds. E. Hyslop,
Gilberg, M.R. and N.J. Seeley. 1981. The identity of V. Gonzalez, L. Troalen, and L. Wilson, 103–109. Edin-
compounds containing chloride ions in marine iron burgh: ICOM-CC and Historic Scotland.
corrosion products: A critical review. Studies in Conser- Thunberg, J., D. Watkinson, and N. Emmerson. 2021.
vation 26(2): 50–56. Desiccated microclimates for heritage metals: Creation
Mackay, A.L. 1962. β-Ferric oxyhydroxide—Akaganéite. and management. Studies in Conservation 66(3): 127–53.
Mineralogical Magazine and Journal of the Mineralogical Turgoose, S. 1982a. The nature of surviving archaeolog-
Society 33(259): 270–80. ical objects. In Conservation of iron, eds. R. Clark and
Matthiesen, H. 2007. A novel method to determine S. Blackshaw, Monograph 53, 1–7. London: National
oxidation rates of heritage materials in vitro and in situ. Maritime Museum.
Studies in Conservation 52(4): 271–80. Turgoose, S. 1982b. Post-excavation changes in iron
Matthiesen, H. and K. Wonsyld. 2010. In situ meas- antiquities. Studies in Conservation 27(3): 97–101.
urement of oxygen consumption to estimate corrosion Watkinson, D. and M. Lewis. 2005. Desiccated storage
rates. Corrosion Science Engineering and Technology of chloride-contaminated archaeological iron objects.
45(5): 350–56. Studies in Conservation 50(4): 241–52.
Neff, D., E. Vega, P. Dillmann, M. Descostes, L. Bellot- Watkinson, D.E. and N.J. Emmerson. 2016. The impact
Gurlet, and G. Béranger. 2007. Contribution of iron of aqueous washing on the ability of βFeOOH to corrode
archaeological artefacts to the estimation of average iron. Environmental Science and Pollution Research Inter-
corrosion rates and the long-term corrosion mechanisms national 24(3): 2138–49.
of low carbon steel. In Corrosion of metallic heritage
Watkinson, D., N. Emmerson, and J. Seifert. 2016.
artefacts: Investigation, conservation and prediction of
Matching display relative humidity to corrosion rate:
long-term behaviour, eds. P. Dillmann, G. Béranger,
Quantitative evidence for marine cast iron cannon balls.
P, Piccardo, and H. Mathiesen, 41–74. Woodhead:
In Metal 2016: Proceedings of the Interim Meeting of the
Cambridge.
ICOM-CC Metals Working Group, New Delhi, 26–30
Reguer, S., P. Dillmann, F. Mirambet, J. Susini, and P. September 2016, eds. R. Menon, C. Chemello, and A.
Lagarde. 2006. Investigation of Cl corrosion products Pandya, 195–202. New Delhi: International Council of
of iron archaeological artefacts using micro-focused Museums-Committee for Conservation (ICOM-CC) and
synchrotron X-ray absorption spectroscopy. Applied Indira Gandhi National Centre for the Arts (IGNCA).
Physics A – Materials Science & Processing 83: 189–93. Watkinson, D.E., M.B. Rimmer, and N.J. Emmerson.
Reguer, S., F. Mirambet, E. Dooryhee, J.-L. Hodeau, P. 2019. The influence of relative humidity and intrinsic
Dillmann, and P. Lagarde. 2009. Structural evidence for chloride on post-excavation corrosion rates of archae-
the desalination of akaganeite in the preservation of iron ological wrought iron. Studies in Conservation 64(8):
archaeological objects, using synchrotron X-ray powder 456–71.
Authors
Nicola J. Emmerson is a reader in conservation at
Cardiff University with research interests in the corro-
sion and conservation of ferrous metals within heritage
contexts.
David E. Watkinson is professor of conservation at

Cardiff University where he teaches conservation of
metallic and inorganic objects and researches the corro-
sion and treatment of metals.
Kimberly Roche graduated from Cardiff University

with an MSc in conservation practice and now works at
the Warren Lasch Conservation Center in Charleston,
South Carolina, USA.
Jerrod H. Seifert gained his PhD on the corrosion and

treatment of marine cast iron at Cardiff University where
he is now an Analytical Laboratory Teaching Associate.
Johanna C. Thunberg is studying for a PhD on copper

alloy corrosion at Cardiff University where she is also
employed full-time to teach conservation.
Analysis of Heterogeneous Tarnish on Silver-based Alloys

Using the Pleco for Local, Controlled Electrolytic Cleaning
Nicola Ricotta* Andrea Cagnini Christian Degrigny

Department of History, Opificio delle Pietre Dure (OPD) Haute Ecole Arc Conservation-
Archaeology, Geography, Fine and Florence, Italy restauration (HE-Arc CR), HES-SO
Performing Arts (SAGAS), University [email protected] University of Applied Sciences and
of Florence www.opificiodellepietredure.it Arts Western Switzerland
Florence, Italy Neuchâtel, Switzerland
www.sagas.unifi.it www.he-arc.ch
Abstract
Artworks made of silver-based alloys tarnish unevenly, silver-based artwork, we demonstrate the potential of the
which considerably compromises their surface appearance Pleco electrolytic pencil to identify the nature of the tarnish.
as well as their historical and artistic value. Mechanical and Artificial coupons mimicking the artwork were employed to
chemical cleaning methods remove silver tarnish but in optimise the use of the Pleco in the local, controlled cleaning
some cases are difficult to control. Electrolytic processes, of metal surfaces.
which are often preferred because they retain some of the
constituent materials, require an appropriate protocol to Keywords
avoid side effects such as overcleaning and black spots. Using silver-alloy, heterogeneous tarnish, Pleco®, artificial coupon,
the representative example of a heterogeneously tarnished linear sweep voltammetry, analysis
Introduction
The tarnishing of silver-based objects is a common based on their reduction peaks and then to define treat-
problem faced by conservation professionals. The metal ment parameters (Degrigny et al. 2016).
surface tarnishes unevenly, with the colour varying from
The Pleco pencil is expected to solve the problem
yellow to black depending on the thickness of the tarnish.
of overcleaning of heterogeneously tarnished silver
In addition, the composition of the alloy, the environ-
surfaces during electrolytic immersion cleaning. Over-
ment and previous conservation interventions influence
cleaning occurs because: (i) the same treatment param-
the nature and appearance of the alteration. The main
eters are applied to the whole metal surface and (ii) the
components of silver tarnish are silver sulfide (Ag2S) and
polarisation of less tarnished areas is similar to that of
silver chloride (AgCl), but the presence of other elements
heavily tarnished ones, causing hydrogen to be released
in the alloy, such as copper, leads to the formation of other
from the former and resulting in excessive cleaning.
compounds (Cu2O, Cu2S, etc.) (Costa 2001).
Furthermore, black spots due to the reduction of copper
Local identification of the silver tarnish is crucial to compounds may appear (Ricotta et al. 2019). The Pleco,
appreciate its heterogeneity and the likely composition however, works selectively and can clean differently
of the underlying metal but also to define appropriate tarnished metal surfaces as appropriate (Degrigny et al.
cleaning protocols. The Pleco electronic pencil was 2015). Nonetheless, it poses a few technical problems
developed in 2016 by the Haute Ecole Arc Conserva- for end users. For example, the identification of small
tion-restauration in Neuchâtel and has proved to be a reduction peaks on LSV plots is difficult due to their
versatile tool for the analysis and local cleaning of silver masking by current fluctuations. These fluctuations
tarnish. Unlike conventional cleaning methods, it also are caused by the pumping system, which heteroge-
allows the reduction of silver tarnish. Linear scanning neously supplies and extracts the electrolyte on the
voltammetry (LSV) is used to identify tarnish compounds metal surface.
ANALYSIS OF HETEROGENEOUS TARNISH ON SILVER-BASED ALLOYS USING THE PLECO FOR LOCAL, CONTROLLED ELECTROLYTIC CLEANING 249
Our approach Table 1. XRF analysis of the different elements of the Valadier cooler
Measuring points Ag% Cu% Trace elements

The objective in this research was to further optimise
1 93.6 ± 0.1 6.4 ± 0.2 Zn
the use of the Pleco as an analytical tool allowing correct Inner upper plate
2 94.1 ± 0.1 5.9 ± 0.2 Zn
tarnish reduction for local, controlled cleaning.
3 94.1 ± 0.1 5.9 ± 0.2 Zn and Fe
Body 4 95.0 ± 0.1 5.0 ± 0.2 Zn and Fe
Valadier case study
5 96.3 ± 0.1 3.7 ± 0.2 Zn and Fe
The Valadier cooler, made by the Roman silversmith 6 90.7 ± 0.1 9.3 ± 0.1 Zn, Fe and Pb
Handle
Giuseppe Valadier in 1810, is part of the collection of 7 90.4 ± 0.1 9.6 ± 0.1 None
Gallerie degli Uffizi, Tesoro dei Granduchi in Flor- 8 92.4 ± 0.1 7.6 ± 0.2 Zn and Pb
Base
ence (Inv. Pitti Art Objects (1911) n. 1838). It depicts 9 90.0 ± 0.1 10.0 ± 0.1 None
mythological scenes (Figure 1a) and consists of a 10 93.3 ± 0.1 6.7 ± 0.2 Zn
Inner bottom plate
11 92.4 ± 0.1 7.6 ± 0.1 Zn
body, handles and a base, all made by casting from a
silver alloy. The object has decorations in relief that
have been chased and both polished and matt surface Preliminary LSV plots performed with the Pleco on
finishes have been applied. Silver-alloy foils have been two discrete, randomly chosen areas (1 and 2), located
embossed to fit inside the cooler and its base. All opposite each other on the surface of the cooler body
elements are secured by a central bolt and nut. Non-in- (Figure 2), showed the presence of different corro-
vasive X-ray fluorescence (XRF) analyses, performed sion products. On one side, only silver compounds
with the portable XRF spectrometer ELIO (XGLab) were detected, and on the other side additional copper
and without any preliminary surface cleaning, revealed compounds were found. This result suggested that the
that the elements of the cooler consist of different composition of the cooler body surface is not uniform. To
silver-copper alloys. Quantitative analysis results were confirm this hypothesis, smoother plots were required.
obtained using bAxil BrightSpec software and a certified Indeed, the high current fluctuations between −1 and
Ag% 92.44 ± 0.12 Cu% 7.59 ± 0.14 standard and were −2V/GC (glassy carbon) prevented a correct visualis-
normalised to 100% (Table 1). ation of any small reduction peaks. However, one such
peak may have been present on the plot of side 1, around
Like many objects in the museum’s storage, the Vala-
−1.6V/GC (Figure 3).
dier cooler is heterogeneously tarnished. As shown in
Figure 1a, some parts of the body, handles and base
are more tarnished (either the hollow parts or chased
backgrounds, Figure 1b), while the polished figures
and other reliefs seem to be only slightly tarnished
(Figure 1c). The differences could be due to intentional
depth effects or to regular maintenance of the object, as
shown by the thick tarnish layers in some grooves that
are more difficult to reach than the embossed surfaces.
This condition demonstrated the importance of a more
selective cleaning, such as is possible using the Pleco.
Figure 2. Electrolytic setup to obtain LSV plots of the body of the Valadier
cooler using the Pleco
Electrochemical study of coupons simulating the

Valadier cooler
To optimise the LSV plots, metal coupons made from
sterling silver (Ag7.5%Cu) plates were used. Some were
Figure 1. Valadier cooler: general view (a) and details of the scenes flat, matt and chased (4 × 2 cm, Figure 4a), and others
depicted on the body (b) with different levels of tarnish (c) were polished and curved (6 × 4 cm, Figure 4b).
Figure 6. Comparison of the LSV plots on side 2 of the Valadier cooler

body and the artificial coupon
Figure 3. LSV plots of two sides of the Valadier cooler body. The different
compounds are identified by the potentials located at the intersection
between the tangents of the slopes of the reduction peaks and the As the artificial silver-based coupons correctly simulated
potential axis. The reductions of AgCl, Cu2O and Cu2S are hypothetical. the electrochemical behaviour of the Valadier cooler, they
Note the possible presence of a peak on the blue plot around −1.6V/GC
were employed to optimise the use of the Pleco.
Optimisation of the Pleco

Several parameters were studied to minimise the current
fluctuations seen in the LSV plots:
• The circulation rates of the diaphragm pumps. If the
extraction flow rate is too high, the Pleco’s micropo-
Figure 4. Metal coupons used to mimic the behaviour of the Valadier rous polyvinyl formal (PVFM) pad dries out, reducing
cooler: matt and chased (a) and polished and curved (b)
fluctuations, whereas if it is too low, the electrolyte
leaks out of the pad and induces more fluctuations.
The metal coupons were artificially aged in a 24 h expo- • A large pad (pad 1 in Figure 7) compressed when
sure to warm hard-boiled eggs, peeled and cut into four inserted into the Pleco nozzle causes smaller current
sections. The coupons, handled without gloves so as to fluctuations but also prevents electrolyte renewal on
induce the formation of fingerprints, quickly tarnished, the metal surface. Conversely, a small pad promotes
acquiring a heterogeneous appearance ranging from electrolyte renewal but also causes electrolyte leakage
yellow to blue (Figure 5). (pad 3 in Figure 7).
• By changing the shape of the pad (from flat, round,
slightly sharp to conical; Figure 7), the contact surface
with the metal surface is reduced, causing larger fluc-
tuations of current that can be reduced by increasing
the extraction flow rate.
Figure 5. Artificial tarnishing of the metal coupons in a desiccator using

warm hard-boiled eggs, peeled and cut into four sections (a), and the
surface of the coupons after a 24 h exposure (b)
LSV plots on the tarnished sterling silver plates (bluer

areas) showed an electrochemical behaviour similar to
that of body side 2 of the Valadier cooler (Figure 6). The
same compounds appeared, although the maxima of the
peaks of the Valadier cooler shifted to more negative
potentials in areas of presumably thicker corrosion Figure 7. Different sizes and shapes of the tested pads, inserted in the
products. nozzle of the Pleco
ANALYSIS OF HETEROGENEOUS TARNISH ON SILVER-BASED ALLOYS USING THE PLECO FOR LOCAL, CONTROLLED ELECTROLYTIC CLEANING 251
LSV plots on the Valadier case study Conclusion

The best LSV plots were obtained with conically shaped The Pleco device and the accompanying LSV plots are a
pad 2 (medium size), which resulted in supply and valuable diagnostic tool, providing critical information
extraction flow rates of 10 and 90, respectively. A total on the tarnishing of silver-based objects. However, the
of nine LSV measurements were performed under these accuracy of the analysis requires an understanding of
conditions on the Valadier cooler body. The studied areas, the effects of the parameters that influence the presence
preliminarily degreased with alcohol and acetone, were of current fluctuations in the LSV plots. The Pleco
randomly selected on sides 1 and 2, considering only the pencil is perfectly adapted to the study of heterogeneous
most tarnished surfaces and different surface finishes. tarnish, but it is invasive, as the measurement point on
Three measurements were made on polished areas and the surface of the object where the reduction reactions
six on chased areas. occurred is visible. However, this disadvantage becomes
secondary if cleaning the object after the diagnostic
On side 1, the tarnish was shown to consist mainly of
steps is planned.
Ag2S and AgCl (?) (Figure 8). On side 2, reduction peaks
of Cu2O (?) and Cu2S (?) were also present (Figure 9).
Acknowledgements
The authors wish to thank Regione Toscana POR FSE
2014–2020 for funding this PhD research project, COST
Action PortASAP – CA16215 for funding the short-term
scientific mission at HE-Arc CR, OPD and HE-Arc CR
for their technical support, and Gallerie degli Uffizi,
Tesoro dei Granduchi for allowing access to their collec-
tion.
Figure 8. LSV plots on different areas of side 1 of the Valadier cooler body References
Costa, V. 2001. The deterioration of silver alloy and some
aspects of their conservation. Studies in Conservation
46(suppl. 1): 18–34.
Degrigny, C., R. Jeanneret, and D. Witschard. 2015.
Local cleaning with the Pleco electrolytic pencil of
tarnished Saint Candide reliquary head at the Treasury
of Saint-Maurice Abbey, Valais (Switzerland). e-Preser-
Figure 9. LSV plots on different areas of side 2 of the Valadier cooler body vation Science 12: 20–27.
Degrigny, C., R. Jeanneret, D. Witschard, C. Baudin, G.
These results suggested that the Valadier cooler body Bussy, and H. Carrel. 2016. A new electrolytic pencil for
had been superficially treated during its production, to the local cleaning of silver tarnish. Studies in Conservation
enrich its silver content. As a result of repeated tarnishing 61(3): 162–73.
and cleaning interventions, part of the enrichment layer
Ricotta, N., C. Degrigny, A. Cagnini, and C. Ortolani.
abraded, exposing the underlying metal. Why one side
2019. Design of a suitable cleaning procedure for the
of the body was apparently differently preserved from
tarnished silver elements of the reliquary bust of Santa
the other is unknown.
Vittoria, Diocesan Museum of Agrigento, Sicily. In Metal
To validate these results, further LSV measurements will 2019: Proceedings of the Interim Meeting of the ICOM-CC
be made, with the measurement areas chosen according Metals Working Group, Neuchâtel, 2–6 September 2019,
to the different chromatic types of surface tarnish eds. C. Chemello, L. Brambilla, and E. Joseph, 261–64.
observed under binocular microscopy. This will allow Paris: International Council of Museums–Committee
mapping of the surface and the assignment of a specific for Conservation (ICOM-CC) and Haute Ecole Arc
composition to each aspect of the surface tarnish. Conservation-restauration (HE-Arc CR).
Authors
Nicola Ricotta is a conservator and PhD student at the
University of Florence, Italy.
Andrea Cagnini is a chemist and professor at the Scuola

di Alta Formazione e di Studio, Opificio delle Pietre
Dure, Florence, Italy.
Christian Degrigny is an electrochemist and professor

at the Haute Ecole Arc Conservation-restauration,
Neuchâtel, Switzerland.
Using Epidemiology to Validate Scientific Results for
Complex Situations
David Thickett
English Heritage
London, UK
[email protected]
Abstract
Science can help in some aspects of improving conservation copper alloy responses. The graphs generated can be used for
practice. Laboratory experiments have elucidated many factors other collections and to carry out similar surveys and analyse
for preventive conservation of archaeological iron and copper the results. The approach requires very large numbers if based on
alloy objects. However, these rely on simple model systems visual observation, but these can be reduced with more sensitive
(powder mixtures) and objects are much more complex in analytical methods. Oxygen depletion has been demonstrated
nature. Epidemiological methods can be readily adapted to to provide useful conclusions using a much smaller sample set.
conservation problems. Analysing the results of two large
criteria-anchored surveys has shown that the experimental Keywords
results do represent a good portion of archaeological iron and iron, copper alloy, surveys, epidemiology, oxygen depletion
Introduction
Burial profoundly changes metal artefacts and can and measured the amount of akaganeite formed or
significantly alter their response to the environment. oxygen depleted to quantify deterioration. Significant
It introduces reactive species, alters the structure and analytical evidence exists that archaeological copper does
dramatically increases porosity and often the fine-scale indeed contain copper chloride (Scott 1990) and iron
porosity responsible for uptake of water vapour and pollu- chloride was analysed in archaeological iron (Thickett
tion. As well as significant differences between objects 2012). Whilst this research provides powerful evidence
from the same archaeological site, burial can cause great to inform preventive conservation strategies, there is
differences throughout an object. a concern that simple model mixtures may not fully
Scientific methods can help understand the impor- represent very complex objects. This has been addressed
tant factors in preventive conservation and treatment. in two ways:
However, this requires near-identical replicates for expo- • Two surveys were undertaken of almost all of the
sure or treatment. Three broad options exist: archaeological iron and copper alloy objects on display
• the use of low-value objects, such as nails; at English Heritage sites and objects in store from
• the use of replicates, which are particularly difficult excavations at those sites.
to produce for archaeological metals; • Measurements were taken of oxygen depletion rates
• exposure of objects only to the same conditions they from a large number of real objects and compared with
will be exposed to in display or storage, which, given the RH measured in other studies on powder mixtures.
changes are slow, mainly requires very sensitive analyt-
Several important criteria for object response studies
ical techniques.
have already been considered within the field of epide-
Two pieces of work on the required relative humidity miology, such as sample size, appropriate statistics,
(RH) and carbonyl pollutant conditions for archaeolog- experimental design, quality of evidence and potential
ical iron and copper alloy have been published (Thickett bias (Kelsey et al. 1996, Fleis et al. 2003, Dean et al. 2021).
and Odlyha 2013, Thickett 2016). These used iron/iron One useful concept in epidemiology is the hierarchy of
chloride and copper/copper chloride powder mixtures methods, which indicates the reliability of different types
of evidence, as summarized in Table 1. Randomised The x scale is essentially the increased number of
controlled studies (comparing two cohorts – groups) are objects responding to the environment. The Fleis
considered the strongest of the unfiltered (experimental) method produces a higher number of objects and,
methods. So-called filtered information is considered to therefore, should be more robust. It uses continuity
be higher quality evidence in epidemiology. However, correction, a process to allow a continuous distribu-
this does not exist at the moment in this field. There are tion (normal) to approximate a discrete distribution
certainly critical knowledge gaps in preventive conser- (binomial). Less sensitive detection, such as visual
vation for a number of materials (Thickett and Lankester observation, will produce a smaller number of objects
2012). The sample size for such unfiltered studies is with observed changes. For example, for cracking
essentially determined by the smallest change detectable. wood, observation by eye may detect increases in crack
Figure 1 shows the required sample size in each cohort length of perhaps 1 mm. More sensitive instrumental
for differing percentages of the exposed object cohort techniques, for example acoustic emission, has been
with observed deterioration (Kelsey et al. 1996, Fleis shown to detect crack increases in wood in the order
et al. 2003). An alpha value of 0.05 (95% certainty that of microns, which means changes in more objects
the two cohorts represent the population from which will be detected. This moves the experiment into the
they are drawn) and a beta value of 0.2 (80% chance of right-hand portion of the graph, requiring less objects
avoiding a false positive result) were used to construct in each cohort. Advances in instrumentation promise
Figure 1. This followed the methodology described in significant improvements over visual observation,
detail in the OpenEpi website (Dean et al. 2021: OpenEpi, hence higher percentages of objects with observable
Version 3, open.source.calculator, SSCohort). Figure 1 deterioration. The curves in Figure 1 can be used in
was generated by feeding several hundred values into other studies, provided the researchers are prepared
the calculator to produce a smooth line. The OpenEpi to accept the alpha and beta value used.
site not only provides extensive background information
but also leads a user through all the necessary steps and The feasibility of cohort studies with visually assessed
decisions to undertake these kinds of studies. survey data of archaeological metals, supplemented
by analysis, was investigated. In the first study, objects
Table 1. Hierarchy of epidemiological methods from the same archaeological site were grouped together.
Information Quality of Different burial environments are known to affect reac-
Study type
type evidence tivity significantly. Conservation treatments can have
Systematic reviews Filtered a significant effect. Although complete conservation
Critically appraised topics Filtered
records from excavation to the present day were not
Critically appraised individual articles Filtered
available, objects from the same site are more likely to
Randomized controlled trials Unfiltered
have had similar treatments than those from different
Cohort studies Unfiltered
sites. The first study investigated archaeological iron.
Case-controlled studies Unfiltered
Cohorts were formed from objects kept at different RH
Background information/expert opinion Unfiltered
levels (5 or 10% bands of maximum RH) and under low
(< 1000) and high (> 1001 µgm−3) concentrations of acetic
acid in a series of showcases.
The complex nature of archaeological iron and copper

artefacts makes measuring deterioration rate difficult
via normal analytical techniques, with several complex
layers of corrosion needing to be identified and quanti-
fied. In many instances oxygen reduction is considered
to be the main cathodic reaction. If it is also the rate
limiting reaction, then measuring oxygen consumption
would indicate the corrosion rate of iron or copper.
Figure 1. Number of objects required for cohort sizes for different Enclosing an object in an impermeable container of
observed reaction rates fixed volume or measured volume, controlling the RH
USING EPIDEMIOLOGY TO VALIDATE SCIENTIFIC RESULTS FOR COMPLEX SITUATIONS 255
and measuring the oxygen content of the air periodically The results were assessed to determine if enough
allows measurement of the oxygen consumption rate objects had been surveyed to produce a statistically
(Thickett 2021). valid randomised controlled study (Kelsey et al. 1996,
Fleis et al. 2003). The sample size for such studies is
Whilst the oxygen depletion technique can provide very determined by the increase in proportion of the exposed
valuable results, there are limits to the atmospheres that sample cohort with the outcome over the unexposed
can be generated and maintained in a very well-sealed sample cohort or, in this instance, the sample cohort
container. The author is unaware of any method to exposed to different conditions. Samples of corrosion
produce mixtures of room-type pollutants, nitrogen were taken from all objects and analysed by Fourier
dioxide, ozone and sulfur dioxide that is suitable for the transform infrared (FTIR) spectroscopy (PerkinElmer
closed environments of this testing. Whilst enclosing 2000 FTIR with Amplif-IR diamond ATR using 32 scans
room air provides an initial concentration, this will and 4 cm−1 resolution) and some by X-ray diffraction
tend towards zero as the gas is consumed in reaction. (XRD) (Phillips 1830/1840).
Any pre-mixed atmospheres would rapidly lose pollu-
As the room environments and showcases for the iron
tion gases as they react with the objects, altering the
were highly variable, the display material had been
concentrations. Additionally, heritage environments
exposed to a wide range of environmental conditions
can be quite complex and fluctuate over wide ranges
(RH and acetic acid concentrations). RH was measured
with time.
with SmartReader SR002 loggers or Meaco radiotelem-
etry sensors with Rotronic HygroClip 2 probes. For
Methods each showcase, changing silica gel or the seasonal RH
pattern leads to one or two maximum RH values per
Surveys
year. These values were recorded over several years.
Archaeological iron Acetic acid concentrations were determined using
diffusion tubes analysed with a Dionex DX600 ion
For each of 31 sites, archaeological iron objects on
chromatograph and AS14 column (Gibson et al. 1997).
display and in low RH storage were surveyed. Just over One measurement (two replicate analyses if within 10%
1,600 objects from 121 showcases and 1,200 objects of each other and repeated if otherwise) was undertaken
in store were investigated. The objects were assessed in each showcase during August or September. Expe-
visually, using a criteria-anchored methodology (Sully rience has shown that these months give the highest
and Suenson-Taylor 1996). Cracking, flaking and acetic acid concentrations in most showcases in rooms
percentage powdering visible on the surface were in the UK without air conditioning. The RH results
assessed, according to four defined categories, as shown were initially considered in 5% bins (e.g. 35%–40%)
in Table 2. All objects were surveyed by the primary of maximum RH measured for two sets of showcases,
author and photographs of representative objects used those above 1001 µg/m3 acetic acid and those below
to improve long-term consistency (Thickett and Odlyha 1000. If a maximum reading was not reached for
2013). Many of the displays had condition photographs half of the instances (each showcase had between 9
taken at the time of installation. These were consulted and 70 maximums), a lower maximum RH was used
to determine if the present damage had occurred during which met that criterion. The OpenEpi website was
the display period. used to undertake t-tests (two-sided, 95% confidence
Table 2. Criteria used for surveys
Iron Copper alloy

Degree, score Corrosion visible on surface Cracking Flaking Corrosion visible on surface
None, 0
Some, 1 1 or 2 spots, < 1% coverage 1 or 2 cracks, < 1 mm total length 1 or 2 flakes, < 1% coverage 1 or 2 spots, < 1% coverage
Medium, 2 3 to 10 spots, < 5% coverage 1 or 2 cracks, < 5 mm total length 3 to 10 flakes, < 5% coverage 3 to 10 spots, < 5% coverage
Heavy, 3 > 10 spots, > 5% coverage Multiple cracks, > 5 mm total length > 10 flakes, > 5% coverage > 10 spots, > 5% coverage
interval) comparing adjacent RH bins. If significant, measured in 10% of the containers with calibrated
these were accepted; if not statistically significant, they iButton (± 2%) temperature and RH data loggers. No
were repeated with the next two bins combined (10%) excursions from the expected conditions were deter-
and then with the next three (15%) until statistically mined. Oxygen concentration was measured with a
significant results were obtained (Dean et al. 2021). PreSens 4 oxygen meter with PreSens Sp-PSt3-NAU-
Objects from the lowest display RH were compared D7-YOP self-adhesive oxygen spots through the glass
to those in store. Stored iron is kept below 16% RH in container walls.
pollution-free (certainly in terms of carboxylic acids)
Tests were undertaken at 30, 35, 40, 45, 50 and 55% RH to
polypropylene boxes.
investigate where the curve appears to be most complex
Archaeological copper alloys in shape. A group of 60 archaeological iron objects were
analysed at each RH value sequentially. A similar-sized
A series of previously reported results for observed
group of archaeological copper artefacts were also
bronze disease on circa 3,800 Egyptian copper alloy
analysed. Higher RH values of 60 and 65% were also used
artefacts were also re-evaluated in these terms (Thickett
for the 54 objects that had previously been exposed to
et al. 2008). The data was reprocessed with the statis-
those RH values on display.
tical methods described. That survey, undertaken two
decades previously, did not use a criteria-anchored
method, which was under development at the time.
Results
Photographs of so-called ‘index objects’ were used to Surveys
define the assigned category boundaries. For some dete-
rioration phenomena, there are issues with visual iden- Archaeological iron
tification under low magnification. Different corrosion Data from a selection of the 121 showcases assessed are
products or materials can look similar and be mistaken shown in Figure 2.
for each other (Argyropoulos, pers. comm., 2007;
Thickett and Pretzel 2010). Samples of all observed
corrosion products were collected and analysed with
FTIR spectroscopy (Nicolet 510PC) using a diamond
cell and beam condenser.
A further similar survey to that for iron was undertaken

on 1,200 objects on display in 84 showcases at 41 sites
and 1,000 objects in store. A number of objects were
assessed with oxygen depletion.
Oxygen depletion
The oxygen depletion rates of objects identified in the Figure 2. Maximum RH values and acetic acid concentrations in
surveys as being representative of the types and degree showcases (numbered 1–17)
of deterioration were measured. Most sites showed
two groups of objects from previous tests (Thickett
This includes the maximum RH and acetic acid values
2021): those that did not react, even at very high RH,
and a spread of the values observed.
and those that did. Only reactive objects were selected.
Objects were selected that had been exposed to at least English Heritage has approximately 85% of its collection
55% RH. The objects were placed in variously sized in store. For this approach, the maximum sample size
sealed glass containers (Quickfit laboratory glassware, was determined by the number of objects on display in
Bernardin or Bocal mason jars, corrosion jars and similar conditions. The number of objects in dry storage
borosilicate glass jars with polypropylene screw tops is almost always significantly larger. Table 3 shows the
blocked with aluminium foil) with glycerol solutions object numbers on display and the calculated required
to control RH (Milner and Dalton 1953). The RH was cohort sizes for 11 selected sites.
Table 3. Selected cohort sizes mined and afterwards the statistical significance between
Percentage Number required Number of archaeological different RH bands.
Site
reacting in cohort iron objects on display
The grey and red lines are experimental data on powder
1 11.18 139 286
2 12.31 122 276
mixtures from Thickett (2012). It was not possible to
3 25.00 46 104 determine errors for the survey results. It would be
4 26.67 41 39 possible to have multiple surveyors assessing the same
5 26.88 41 92 objects, but as this would involve visits to 72 sites over a
6 27.08 41 73 very wide geographic area, this was considered prohib-
7 28.88 37 90 itively expensive. Published research on surveys of
8 29.44 37 38 archaeological iron objects indicated errors in the order
9 28.57 37 72 of 10% (Leese and Bradley 1995). As can be seen, there is
10 29.74 34 64 a very good correlation between the two data sets. Figure
11 66.67 11 14 3 also includes similar data for higher concentrations of
acetic acid (red line and red diamonds).
The FTIR and XRD analyses showed three sites that
Again, a good correlation is shown. There are no survey
contained a high proportion of corrosion products other
points below 40% RH, as these were all older wooden
than akaganeite (goethite and rozenite, iron sulfate).
showcases with relatively high air exchange rates and
As the corrosion products are different, caution had
even large amounts of silica gel, very low RHs cannot
to be exercised when considering these sites, and they
be maintained. Many of these cases have been replaced
were excluded from the analyses. The survey scores in
in the last decade, hence the smaller number of points.
a particular RH and acetic acid band were added to
give the survey value. If a showcase had eight objects Archaeological copper alloys
surveyed as 1, 1, 2, 2, 2, 3, 4, 4, the survey value was the
For the copper alloy artefacts, a much smaller percentage
sum (19 in this case). The survey score was normalised
difference was observed, meaning larger sample sizes
based on the number of objects and multiplied by 100 to
would be required. The number required for a signifi-
give manageable numbers, i.e. 19/8, 237.5. All the show- cant response at the visually observed difference levels
cases in the same acetic acid and maximum RH band is shown in Figure 4.
were added together. The experimentally determined
risk was then set to the survey score at the highest RH
point (80%). Figure 3 shows the statistically significant
points from the survey overlaid with the response line
generated from powder experiments (partial results in
Thickett and Odlyha 2013, full results in Thickett 2012).
The points were first assessed for the number of objects
in the cohort, then the corrosion products were deter-
Figure 4. Number of objects required for cohorts for observed copper

alloy degradation rates
This was generated by entering the percentage of non-cat-

egory 0 objects from each site into the OpenEpi website.
Taking the raw, visually assessed data, the 17 sites
required cohort sizes of between 130 and 911. This is
Figure 3. Experimentally derived reaction risk compared to cumulative an example of the sample sizes required and no further
scores from object assessments for low and high acetic acid environments statistical analyses were undertaken with this data. For
many collections, these numbers would rarely be on curve, but because there is so little available data, it can
display. The effect of acetic acid (and formic acid and only be considered an indication.
formaldehyde) is much smaller than for iron, so more
showcases can be grouped together to provide larger Oxygen depletion
numbers. The types of corrosion analysed for these
Archaeological iron
objects are shown in Figure 5.
Figure 7 shows the results of oxygen depletion meas-
urements on 60 archaeological iron objects. Minimum,
maximum and interquartile ranges are shown. The red
bars are experimental powder results (same of the raw
data can be seen in the grey line in Figure 3, but the graph
is scaled to fit the survey results).
Figure 5. Copper alloy corrosion products determined by site
Many of the objects recorded as having bronze disease

from visual examination (mainly spots of bright green
corrosion) were found by FTIR analysis to have minerals
other than atacamite or paratacamite (labelled as bronze Figure 7. Oxygen depletion rates measured for 60 archaeological iron
objects
disease in Figure 5) on their surfaces. Hence, the propor-
tion reacting with actual bronze disease is lower and the There was a sizeable spread of data similar to the results
required cohort size greater. reported by Watkinson et al. (2019). The ranking of each
individual object compared to all those tested across
The significant results from the second copper alloy
the RH values tended to remain similar. It is reasonably
criteria-anchored survey are shown in Figure 6.
clear that the bulk of the data follows the experimentally
derived data.
Archaeological copper alloys
Figure 8 shows the results for the archaeological copper

alloy objects.
Figure 6. Experimentally derived reaction risk compared to cumulative

scores from object assessments for copper alloys
The blue line is from experimental data on powder

mixtures from Thickett (2012). Only five points were
determined to be statistically significant. They follow Figure 8. Oxygen depletion rates measured for 56 archaeological copper
the basic shape of the experimentally derived response alloy objects
Again, within the range of responses, it appears the determines whether statistically significant differences
experimentally derived bars provide a reasonable fit for can be detected between cohorts, and these numbers are
actual object response. always higher than those from the Fleis equation. Epide-
miology also has potential as a primary research method
Discussion and conclusion and the advantage of considering the full complexity of
real objects in real environments provided methods to
The surveys produced results in agreement with exper-
accurately determine small changes are developed.
imental powder studies, validating the adoption of
their results and enhancing confidence in their use to
References
improve preventive conservation. The iron powder
research elucidated the impact on the reactions of RH, Argyropoulos, V. Technological Educational Institute of
acetic and formic acid and formaldehyde concentration, Athens, Greece, personal communication, 25 February
temperature to a degree, the presence of copper (from 2007.
associated burial), humic acid, goethite and akaganeites Dean, A.G., K.M. Sullivan, and M.M. Soe. OpenEpi:
formed under different RH conditions, and the metal to Open source epidemiologic statistics for public health,
chloride ratio. The impact of short periods of higher RH, version 3.01, updated 2013/04/06. www.openepi.com/
such as when showcases or storage boxes are opened, and (accessed 25 October 2021).
the influence of still and flowing air were also investi-
Fleis, J.L., B. Levin, and M.C. Paik. 2003. Statistical
gated. Many factors remain to be elucidated: the effect
methods for rates and proportions, 3rd ed. New Jersey:
of the metallurgy of the object, the different forms of
John Wiley & Sons.
chloride present, differences in the reactivity and stability
of the akaganeites present, the reactions generating Florida State University. 1995. The Marsaglia random
goethite, and iron carbonate and iron sulfate. Whilst number CDROM including the diehard battery of tests of
the selection of objects for display in each showcase is randomness. Tallahassee: Florida State University.
clearly not random, tests were undertaken that took into Gibson, L.T., B.G. Cooksey, D. Littlejohn, and N.H.
consideration object typology and burial depth using Tennent. 1997. A diffusion tube sampler for the determi-
the diehard test suite (Florida State University 1995) for nation of acetic acid and formic acid vapours in museum
four sites. Over 98% of the p-values were between 0.025 cabinets. Analytica Chimica Acta 341: 11–19.
and 0.975, indicating the randomness tests were passed Kelsey, J.L., A.S. Whittemore, A.S. Evans, and W.D.
at a 0.05 level. Thompson. 1996. Methods in observational epidemiology.
The number of objects on display was found to be a Oxford: Oxford Academic Press.
limiting factor in several instances and especially with Leese, M.N. and S.M. Bradley. 1995. Conservation condi-
the less responsive copper alloy objects. The oxygen tion surveys at the British Museum. In CAA94. Computer
depletion measurements produced more finely graded applications and quantitative methods in archaeology
results and could determine much smaller differences in (BAR International Series 600), eds. J. Huggett and N.
response with smaller numbers of objects. Techniques Ryan, 81–86. Oxford: Tempus Reparatum.
from epidemiology proved extremely useful to the devel-
Milner, C.S. and N.N. Dalton. 1953. Glycerol, 269. New
opment of the study. Together, the methods provided
York: Reinhold Publishing Corporation.
confidence that the experimental corpus of results repre-
sents a good portion of archaeological iron and copper Scott, D.A. 1990. Bronze disease: A review of some chem-
alloy responses. The methods described can readily be ical problems and the role of relative humidity. Journal of
used by other researchers. The OpenEpi website guides the American Institute for Conservation 29(2):193–206.
users through the decisions needed. The cohort number Sully, D. and K. Suenson-Taylor. 1996. A condition survey
requirements shown in Figure 1 give a general indication of glycerol treated freeze-dried leather in long-term
for required sample sizes for such studies. Previous work storage. In Archaeological Conservation and its Conse-
has shown that surveys of small subsets can give fairly quences: Preprints of the Contributions to the Copenhagen
robust estimates for percentage reaction rates for archae- Congress, 26–30 August 1996, eds A. Roy and P. Smith,
ological metal collections. The distribution of the results 177–81. London: International Institute for Conservation.
Thickett, D. 2012. Post excavation changes and preventive chloride on post-excavation corrosion rates of archae-
conservation of archaeological iron. PhD dissertation, ological wrought iron. Studies in Conservation 64(8):
University of London, UK. Available online at www. 456–71.
english-heritage.org.uk/siteassets/home/learn/conser-
vation/collections-advice--guidance/thickettthesisfinal- Author
version.pdf (accessed 25 October 2021).
David Thickett has a degree in natural sciences, a PhD
Thickett, D. 2016. Critical relative humidity levels in archaeological conservation and chemistry and worked
and carbonyl pollution concentrations for archaeo- for two years in industrial ceramics research. He joined
logical copper alloys. In Metal 2016: Proceedings of the the British Museum in 1990, specialising in preventive
Interim Meeting of the ICOM-CC Metals Working Group, conservation and inorganic materials conservation
New Delhi, 26–30 September 2016, eds. R. Menon, C. research. He joined English Heritage in 2003 as a senior
Chemello, and A. Pandya, 0401_24. New Delhi: Interna- conservation scientist, mainly researching preventive
tional Council of Museums-Committee for Conservation conservation. Recent projects have focussed on historic
(ICOM-CC) and Indira Gandhi National Centre for the house environments, acoustic emission, collections
Arts (IGNCA). demography and epidemiology, non-destructive testing,
Thickett, D. 2021. Oxygen depletion testing of metals. microclimate frames and optical coherence tomography.
Heritage 4(3): 2377–89. He sits as a UK expert to the European Standards CEN/
TC 346 (conservation standards) and is a directory board
Thickett, D. and B. Pretzel. 2010. Micro-spectroscopy: A
member of the Infrared and Raman Users Group (IRUG).
powerful tool to understand deterioration. E-Preservation
Science 7: 158–64.
Thickett, D. and P. Lankester. 2012. Critical knowledge
gaps in environmental risk assessment and prioritizing
research. Collections 8(4): 281–96.
Thickett, D. and M. Odlyha. 2013. The formation and
transformation of akaganeite. In Metal 2013: Proceedings
Group, Edinburgh, 16–19 September 2013, eds. E. Hyslop,
V. Gonzalez, L. Troalen, and L. Wilson, 103–109. Edin-
burgh: International Council of Museums-Committee
for Conservation (ICOM-CC) and Historic Scotland.
Thickett, D., S. Lambarth, and P. Wyeth. 2008. Deter-
mining the stability and durability of archaeological
materials. In Art08: 9th International Conference on NDT
of Art, Jerusalem, Israel, 25–30 May 2008. www.ndt.net/
article/art2008/papers/024Thickett.pdf (accessed 25
October 2021).
Thickett, D., V. Vilde, P. Lankester, and E. Richardson.
2017. Using science to assess and predict object response
in historic house environments. In Preventive Conserva-
tion in Historic Houses and Palace Museums: Assessment
Methodologies and Applications. Conference Proceed-
ings, Palace of Versailles and Trianon, 29 November–1
December 2017, ed. N. Francaviglia, 258–70. Milan:
Silvana Editoriale SpA.
Watkinson, D.E., M.B. Rimmer, and N.J. Emmerson.
2019. The influence of relative humidity and intrinsic
Long-Term Provision of Stable Environments for Metals
Conservation
Paul Lankester* Sally Johnson

English Heritage English Heritage
London, UK Silsoe, UK
David Thickett *Author for correspondence

English Heritage
London, UK
[email protected]
Abstract
Preservation of metals often requires dry environments. gel were also investigated. Relevant issues and observations
Several aspects of generating such environments over many are discussed. The work clearly concludes the necessity for
years have been investigated, building on previous research. suitable calibration to achieve accurate low RH readings
The measurement of low relative humidity values, even and the superiority of certain showcase designs in terms of
with electronic probes, requires well-adapted calibration both initial and long-term air exchange rate. The importance
procedures for accurate results. Long-term changes in silica of accelerated ageing to predict long-term performance
gel performance were measured. The provision of suitable in materials was demonstrated and improvements to dry
low air exchange rate showcases and the changes in their environment lifetime in real storage situations using a certain
performance over 13 years were investigated. Pollutant gas type of lockable box were shown.
build-up in showcases over decades were assessed as well as
dehumidifier performance. The performance and ageing of Keywords
Lock & Lock boxes for storing archaeological metals with silica showcases, storage, dehumidifier, pollution
Introduction
Providing suitable environments for long-term metals there are also several technical aspects that can affect
preservation requires dehumidification and control of long-term performance. The actual impact of these has
pollutants. Sometimes concentrations must be very low been investigated.
and careful testing of materials in enclosures is required to
Development of low-cost methods for measuring air
avoid internally generated pollution. Archaeological iron
exchange rates (AERs) and protocols for repeatable
and copper can deteriorate at ambient relative humidity
(RH) values. Research has improved understanding of measurements were critical in providing a pathway to
the processes involved and curatorial debate at English predict, specify and guarantee showcase performance in
Heritage has settled on values of 30% for iron and 35% for the wide range of environments across English Heritage’s
copper while on display, and 16% for both archaeological estate (Calver et al. 2005; Thickett, forthcoming). Lower
metals in storage, to manage losses of value (Thickett air exchange rates help to control RH but increase the
2012, Thickett and Odlyha 2013, Thickett 2016). For concentration of damaging internal pollutants, if present.
preventive conservation to work as a strategy, these condi- A series of simple spreadsheet-based models have been
tions need to be achieved over very long time periods. tested and shown to accurately predict internal RH
Additionally, increasing awareness of anthropological performance based on room values (Thickett 2020b;
climate change means it will be important to predict Thickett, forthcoming). These can be used with data
the carbon footprint of such approaches. Organisation from any space to predict the likely performance (internal
is always critical to success for such endeavours, but RH) of showcases and of using dehumidifiers. Thomson
(1977) developed an approximation for hygrometric have far greater capacity than is required for a single
half-life: showcase, multiple cases can be controlled from a single
dehumidifier. Performance has been assessed with cali-
tH1/2 = 4MHB/n(1)
brated monitoring.
Where tH1/2 is the hygrometric half-life, the time it takes
Oxygen depletion testing has also been developed
for the RH of the case/box to move halfway towards
and tested against long-term changes and can identify
the room RH (days); MH is the moisture-specific buffer
those iron and copper alloy objects that are stable at
reservoir corrected for hysteresis (g/kg/%); B is the buffer
high RH and do not need controlled RH environments
loading (kg/m3); and n is the AER (day−1). See Weintraub
(Thickett 2021).
(2002) for an explanation of the practical use of silica
gels, hysteresis and M and MH. Non-archaeological metals are much less RH sensitive,
but control is still important. The strategies listed above
The models were shown to give sufficiently accurate
have been used in many display and storage situations
results when iterated at 15, 30 or 60-minute intervals
successfully. At suitable RH values for non-archaeolog-
provided that the air is well mixed in the display volume
ical metals, silica gel calculations become more complex
and between any silica gel compartment and the display
because of hysteresis (above 40% RH). They are sufficient
volume. The half-life concept can be used to estimate the
to model showcase performance in particular envi-
demand for dry air to keep a showcase below a certain
ronments (Thickett, forthcoming). In some instances,
RH in a given room. Coupled with capacity information
an additional amount of silica gel has been added to
on dehumidifiers, this can accurately predict the energy
compensate for uncertainty in the value.
use (Thickett 2020b).
Both sorbents and dehumidifiers have been used in Materials testing

different showcase situations to reduce RH. Carbon
Pollutant gases from showcase materials are probably the
footprints have been assessed for silica gel-controlled
most significant risk that showcases pose to metals. Lead,
showcases (Thickett 2020a). Long-term performance
copper alloys, steels and archaeological iron are corroded
has been assessed with carefully calibrated RH meas-
by acetic and formic acid. A full review of the reported
urements.
damage is given in Thickett (2020b). Many methodolo-
gies exist for testing and several are being actively devel-
RH measurement
oped. The Oddy test, an accelerated corrosion test with
Previous work has shown issues with the ageing of metal pieces acting as corrosion sensors, is widely used.
RH001 indicator strips and the fact that they give low Reduction of the test time from the established Oddy
reading after 6 years of use (Thickett 2012). Strips are test is a common theme, with more sensitive analytical
now replaced every 5 years. Electronic probes have better methods postulated to bring improvements. The param-
precision than indicator cards and better accuracy when eters of such tests have been investigated. Experiments
carefully calibrated. Their calibration for low RH has have identified materials that only emit late on in the
been investigated. Oddy test, and the presence of acetic, formic and nitric
acid in these tests has been assessed. Ambient temper-
Changes to showcase AER and silica gel ature tests have been undertaken to examine emission
performance profiles of both acetic and formic acid for these materials
(Thickett et al. 2007).
The two factors likely to degrade dry showcase perfor-
mance have been investigated. The increases in AER
Performance of polymer boxes for dry storage
in varying designs of showcases as they age have been
measured. Changes to silica gel isotherms after prolonged Polyethylene boxes (frequently the Stewart brand or ‘old
use have also been measured with water vapour uptake Stewart’) with dry silica gel have been used for storage of
and dynamic vapour sorption. Cases can be controlled archaeological iron for decades. The boxes changed to
with dehumidifiers, and some results have already been polypropylene in the mid-1990s (‘new Stewart’). Storage
published (Thickett et al. 2007). As many dehumidifiers of archaeological metals in Stewart boxes with silica gel
LONG-TERM PROVISION OF STABLE ENVIRONMENTS FOR METALS CONSERVATION 263
to keep RH values below 16% for several years in 50% mathematically manipulating the curves measured to
dehumidified rooms has been investigated. Newer box fit the pair of RH points.
designs (with locking lids) potentially offer longer silica
gel replacement lifetimes (Thickett and Odlyha 2010). Silica gel
Lock & Lock boxes have been used for a decade. Recent
A series of 35 samples of silica gel (variously sized bags,
research has identified variability in the performance of
but mainly 500 g) were collected at random from those
some brands (Thunberg et al. 2021). Similar tests have
held by English Heritage to map the range of gels pres-
been carried out in actual stores and ageing of the boxes
ently used. The samples were selected according to a
undertaken to indicate future performance. Previous
range of yellowness (a rough indication of number of
work showed the iterated Thomson equation worked
reconditioning cycles) as determined by eye. Five new
well for old-style (polyethylene) Stewart boxes as most
samples were also tested. The mass increase of new and
of the leakage occurred through the seals and not via
old silica gel was measured by drying the gel at 110°C for
permeability through the box body (Thickett 2010). The
48 hours, followed by exposing it in a chamber at 30%
much tighter seals of the newer boxes could change this
RH (generated by a Munters MG50 dehumidifier) and
and experiments were undertaken to estimate the ratio
measuring the mass gain until no further increase was
of water vapour transfer between the two mechanisms.
recorded. Each bag was measured with a Minolta 2600D
Some aspects of ageing of the newer boxes have been
colorimeter and the b* value was used as an indication of
analysed, investigating the change in crystallinity, which
yellowing. There is a general link between the yellowing
is related to permeability (Thickett 2012).
of the cotton bags and the number of times they have
This study aims to address several important questions been reconditioned (heated to 110°C), but as the bags
related to safely maintaining environments for metals: may contain different bleaching agents, this cannot be
• measurement of RH considered a definitive measure (Farzana and Akter
• design of showcases for low AER and how this degrades Smriti 2015). The water uptake measurement gave an
with age indication of the important M value (Weintraub 2002).
• changes in performance of silica gel and whether it Changes in hysteresis may affect performance, with a
needs periodic replacement modified hysteresis corrected for MH value. For low
• long-term performance of dehumidifiers RH situations, this would only occur in heated rooms
• the requirement for accelerated ageing in materials in winter, when lower RHs (< 30 or 35%) can be gener-
testing for damaging pollutant emissions ated. Ten representative silica gels were further tested
• long-term performance of polymer boxes for dry with dynamic vapour sorption (DVS) using a Surface
storage. Measurement Systems Adventure. Initially isotherms at
20°C were generated between 5 and 80% RH and then
Methods MH values were calculated by cycling 5 times between 25
and 30% and then 30 and 35%.
Measuring RH
The performance of electronic probes for RH often Dehumidifiers
changes with time. The RH response of a series of six
The maintenance and failure rates of Munters MG50
Vaisala HUMICAP probes was measured by exposing
dehumidifiers were assessed from records of 34 units
them above glycerol solutions at 5% RH intervals at
used at 13 English Heritage and British Museum sites.
between 0 and 80% in polycarbonate chambers in
For 12 dehumidifiers, the dehumidification capacity was
an oven at 15°C. The RH was checked with two new
assessed by the manufacturers between 6 and 12 years
Rotronic HygroClip II probes manufacturer calibrated
after installation.
at five points with a reported accuracy of 0.8%. Two
Hanwell HumBug loggers were new, two had been used
Showcases
for 3 years in environments between 30 and 70% RH
and two were 6 years old and had been used in similar The air exchange rates of over 300 newly procured show-
environments. The effect of two-point calibrations at cases from six manufacturers were measured using the
33 and 75% and at 20 and 80% RH was assessed by carbon dioxide tracer gas decay method (International
Table 1
Type Dimension range Materials Tabela formatada

(m)
Table 1 le width height
264 ICOM-CC | METAL Type
2022 | HELSINKI, FINLAND n
Dimension range Materials Tabela formatada
ANALYTICAL INVESTIGATIONS
gt
(m)
h
le width height
Bottom-hinged n 0. 0.3 0.3 to 9.4, 11.2 mm
Table 1. Showcase details
desktop gt 4 to 0.6 laminated
h to 2.0 glass, steel,
Bottom-hinged 0. 2.0.3 aluminium,
0.3 to 9.4, 11.2 mm
desktop 4 0to 0.6 Dimension
laminated
gas lift assist
range (m)
Type to 2.0 glass, steel, Materials
2. length
aluminium, width height
0 gas lift assist
Bottom-hinged desktop 0.4 to 2.0 0.3 to 2.0 0.3 to 0.6 9.4, 11.2 mm laminated glass, steel,
aluminium, gas lift assist
Top-hinged 0. 0.3 0.3 to 9.4, 11.2 mm
desktop 6 to 0.4 laminated
Top-hinged 0. to
0.3 0.8 0.3 to 9.4, 11.2 mmsteel,
glass,
desktop 6 1.to 0.4 laminated
aluminium,
to 20.8 glass, steel,
some gas lift
Top-hinged desktop 1. 0.6 to 1.2 aluminium, 0.3 to 0.8
assist
0.3 to 0.4 9.4, 11.2 mm laminated glass, steel,
2 some gas lift
assist aluminium, some gas lift assist
Hinged upright 9.4, 11.2 mm

Hinged upright 9.4, 11.2 mm
laminated
Hinged upright laminated
glass, steel, 9.4, 11.2 mm laminated glass, steel,
glass, steel,
aluminium
aluminium aluminium
Pull-and-slide
Pull-and-slide 9.4, 11.2 mm
9.4, 11.2 mm
Pull-and-slide upright upright
upright laminated
laminated 9.4, 11.2 mm laminated glass, steel,
glass, steel,
glass, steel,
aluminium, aluminium, slide mechanisms top and
aluminium,
slide
slide
mechanisms
bottom
top andmechanisms
bottomtop and
bottom
Organization for Standardization (ISO) 2017). The designs named. Samples (20 g) of each material were enclosed
fell into four main categories, described in Table 1. in 1 L corrosion jars with diffusion tubes (Gibson et al.
1997) and 0.5 mL of water for 7, 14, 21 and 28 days at
To assess air exchange between silica gel trays and the
60°C. The diffusion tubes were extracted with 5 mL of
display volume, initial trials used perforated powder-
18.2 MOhm cm−1 water, filtered to 50 μm and analysed
coated steel baseboards with Baumann Ultra II cotton
for acetate, formate and nitrate ions with a Dionex DX600
fabrics adhered to them. The perforations were 5 mm and
with an AS14A column, 8 mM of sodium carbonate and
8 mm diameter holes, each covering 25% of the baseboard
18 mM of sodium bicarbonate eluent.
area. RH probes were placed in the display volume and
silica gel tray of eight showcases. The number of days it Two of the materials were available as large samples.
took for the RH in the two compartments to achieve a These were tested periodically (1, 3, 5, 8, 10 and 15 years)
3% difference was recorded. with emission tests designed to replicate those under-
taken industrially to methods that became ISO 16000
Having settled on the 8 mm holes, similar measurements
part 6 and 11. The samples were kept in ambient room
were taken on 40 showcases of three different designs.
conditions of 10°C–24°C and 30%–75% RH. A 70 × 70 cm
The air exchange rates were re-measured in 46 showcases sheet of each sample was placed in a stainless steel box
after periods of use of between 2 and 13 years, sometimes (80 × 80 × 40 cm) and kept at 23°C and 50% RH in a
at multiple time periods. Sanyo environmental chamber. Air was fed into the box
using a pump (Cassella) at a rate designed to give an
Pollutants air exchange rate of 0.18/hour. The air in the box was
sampled (120 L at 2 L/min) using the same type of pump
A previously unpublished protocol for the Oddy test
into a potassium carbonate solution and analysed with
examined coupons in a test tube after 7, 14 and 21 days
ion chromatography as previously.
in the oven as well as at the normal end of the test after
28 days. Four materials that were assessed as showing no
Storage boxes
corrosion after 14 or 21 days but observed to cause signif-
icant lead corrosion and, in two instances, copper corro- Two types of polypropylene boxes were investigated:
sion after 28 days were subjected to further testing. As Stewart Sealfresh (1 L) and Lock & Lock Classic (1 L
the materials are no longer available, they have not been polypropylene with a silicone strip seal and locking tabs).
Tests were conducted similar to those described by As can be seen, the response of the probes decreases
Thunberg et al. (2021), with 44 g/L of dry silica gel in the at higher RH values as they age. The new probes are
box and a calibrated RH probe. The boxes were placed perfectly linear. It is important to undertake three-point
in dehumidified storerooms used for archaeological calibrations to detect loss of linearity as the probes age.
metal and kept at between 48 and 53% RH. Twenty-eight The black line shows the effect of recalibrating the 6-year
boxes of each type were tested in three different stores probe B at two points: 30 and 75%. This introduces
for between 6 and 18 months. very large errors (up to 8%) in the low RH region. This
To investigate the amount of water vapour permeation is a combination of the non-linearity of the response
through the box body, the sides and tops of the boxes of the aged logger and the two-point calibration used.
were covered in 3M 425 aluminium tape, leaving the seal Unfortunately, a two-point calibration will not detect
exposed. The RH increase tests were then repeated and non-linearity in the probe response. Several other RH
compared with the same box in its original state. probes commonly used in cultural heritage show similar
behaviour. Three-point calibration should be undertaken
The potential ageing of the Lock & Lock boxes was inves-
over the region of interest (5%–30% or 40% for this
tigated, both by accelerated ageing and by comparing the
application), with the calibration points just bracketing
performance of boxes used in stores for the past decade
the RH region. All previous calibrations (over 3600)
with newly purchased ones.
have shown good linear performance in even very aged
New 1 L Lock & Lock boxes (that had previously had probes in the 5%–40% RH region, which indicates that
their performance measured) were aged at 40, 50, 60 and with careful calibration, accurate measurements are
70°C for 14 days in quadruplicate. The environmental certainly possible.
chamber was run at 50% RH and 500 g of dry silica gel
was placed in the boxes to replicate their use. Before Silica gel
and after ageing, attenuated total reflectance (ATR) with
Fourier transform infrared (FTIR) spectroscopy were Figure 2 shows the mass uptake of the silica gels at 30%
used on a sample cut to expose the full cross-section RH, plotted against the yellowing of their cotton bags.
(Bruker Alpha FTIR with a diamond ATR). The degree
of crystallinity was calculated using the peak ratios at
997, 972 and 917 cm−1 following the method used by
Yakimets et al. (2004). The reaction rate was calculated
as the percentage change in crystallinity per year.

Measuring RH
The measured versus actual RH for the six loggers is
shown in Figure 1.
Figure 2. Silica gel mass uptakes at 30% RH
The mass uptakes cover less than a 1% range, indicating

that M values do not change significantly with ageing
(provided the isotherms have no significant hysteresis up
to 30% RH). There does not appear to be a correlation
between the water mass uptake and yellowing which is
indicative of ageing. This indicates the response of the
gel is not affected by the number of times it has been
reconditioned (likely related to its age).
Figure 1. Calibration data for HUMICAP probes showing measured RH Isotherms for three representative silica gels of the ten
versus actual RH measured are shown in Figure 3.
Figure 3. Silica gel isotherms showing moisture content versus RH

Figure 5. Measured maintenance and failure rates for Munters MG50
dehumidifiers
The isotherms are very similar. There is slight hysteresis
starting around 27% RH. The hysteresis-corrected MH
cation capacity had not fallen below 90% of performance
values were all within 3% of the M value (Figure 4), indi-
cating this would only introduce very slight differences when new in any of the 12 instances measured.
in performance.
Showcases
The AERs as measured just after installation of the
showcases are summarised in Table 2.
As can be seen, the type of design has a significant

impact on performance. There is a large spread of values.
Bottom-hinged desktop showcases have produced
consistently lower AERs. Top-hinged desktop showcases
have the highest measured AERs, probably due to leakage
in the hinge mechanism. Leak detection with Inficon
D-Tek and the UE Systems Ultraprobe 100 equipment
Figure 4. Difference between M and MH for naturally aged silica gels indicated most of the leakage was from the hinge in the
26 instances tested (Calver et al. 2005). The pull-and-slide
The silica gel does not appear to be ageing in a way that
doors produced lower AER in upright showcases than
affects its performance.
in those with hinged doors.
Dehumidifiers The number of days to equilibrate the tested showcases

The vast majority of the units maintain the showcase (for the RH in the showcase to drop within 3% of that in
environment below the desired RH level. Some instances the silica gel compartment) is shown in Table 3.
(5) were observed of the humidistat calibration moving,
Table 2. Measured AER of as-installed showcases
generating higher RHs. All units have on-line monitoring
to allow swift diagnosis of failure, and this is now moni- AER (/day)
Number
tored for increases due to humidistat calibration drift. Type
tested Minimum Lower Upper
Maximum
quartile quartile
The collated failure and maintenance rates for the 34 Top-hinged 61 0.08 1.2 4.5 35
units are shown in Figure 5. desktop
Bottom-hinged 126 0.03 0.12 0.75 2.5
Five of the 34 units failed and were uneconomic to repair. desktop
Sixteen visits were required by the manufacturer, leading Hinged upright 48 0.23 0.81 1.84 12.4
to on-site maintenance or repair and in three instances Pull-and-slide 92 0.02 0.62 1.24 3.6
repair at the manufacturer’s premises. The dehumidifi- upright
Table 3. Effect of hole diameter in metal baseboards on RH equilibration time
T (°C), RH (%) Hole diameter(mm) AER (day-1) Days to equilibrate

Porchester 10–16, 63–91 Case 1 5 0.65 60
Case 2 5 0.45 65
Case 3 8 0.30 <1
Case 4 8 0.13 <1
Pevensey 15–22, 61–89 Case 4 5 0.48 54
Case 5 8 0.57 <1
Dover 11–18, 62–73 Case 6 5 1.45 52
13 – 17, 63-67 Case 6 8 1.45 <1
Rangers 14–19, 45–53 Case 7* 8 0.03 <1
16–23, 41–61 Case 7* 5 0.03 55
20–29, 41–56 Case 7* 8 0.03 <1
21–27, 43–61 Case 7* 5 0.03 54
21–33, 35–61 Case 7* 8 0.18 <1
20–34, 35–61 Case 7* 5 0.18 49
20–22, 40–51 Case 7* 8 0.18 <1
20–22, 38–52 Case 7 5 0.18 61
The 8 mm holes clearly work with very rapid equilibra- values measured just after 10 years after installation were
tion between the silica gel and display volumes for the due to the cases being moved without consultation of the
bottom-hinged desktop showcases and environments technical staff and without being levelled. The pull-and-
shown in Table 3. The 5 mm holes obviously limit air slide upright cases came next. It is difficult to separate
exchange between the two volumes very significantly the drop in long-term performance of the top-hinged
under a variety of temperature conditions. How much desktop cases and the hinged upright cases. The hinged
of a risk this poses depends on what RH the silica gel is upright cases are more closely grouped together, whilst
changed at. the top-hinged desktops show some higher but also lower
increases. It should be noted that the highest increases
Figure 6 shows the percentage increase in AER versus
came from showcases with lower as-installed AERs. It is
the number of years since installation. A log scale was
clear that the type of showcase has a significant impact
used to accommodate the very wide range in values. The
on the AER and particularly its performance after ageing.
markers are hollow for those showcases that were still
Showcase performance can be affected by a number of
within AER specification after ageing and solid for those
factors. On many types of wooden floor, showcases will
that were above specification after ageing.
move from level over time. Repeated knocks from the
public can also cause this. The mass of glass can deform
door supports and repeated opening can damage seals.
Silicone seals shrink with age and can embrittle and creep,
reducing their efficiency. Dust particles trapped in seals
also reduce their performance.
Pollutants
The amounts of acetic, formic and nitric acid emitted
from the four materials in the tests are shown in Figure 7.
All of the materials show no detectable emissions

Figure 6. AER increases versus showcase age (0.001 mg for acetate and formate, 0.005 mg for nitrate)
in the first 14 days of ageing at 60°C. Amounts are
It is clear the bottom-hinged desktop cases retain their expressed as weights as the diffusion coefficients are not
performance better in the long term. The four higher known for these species at 60°C. Two materials emitted
Figure 8. Hygrometric half-lives of storage boxes

Figure 7. Pollution concentrations emitted from materials at 60°C and
near 100% RH after 1, 2, 3 and 4 weeks
damaging acetic and formic acid after 21 days, while all

four emitted acetic, formic or nitric acid after 28 days.
This clearly shows the materials are emitting the gases late
in the test, as no or very low levels of gases were present
after 14 days. Increasing the sensitivity of the detection
system will not detect gases that are not present.
Two of the materials had large enough samples for

chamber emission tests and were tested periodically.
At 1 and 3 years, sample B did not emit any detectable Figure 9. Arrhenius plot of accelerated ageing of polypropylene boxes
formic acid (12 µg/m3). After 5 years, 1,567 µg/m3 was at various temperatures and 50% chamber RH with the rate of decrease
detected. This concentration is quite high and poten- in degree of crystallinity
tially damaging to a range of materials (Norwegian

Institute for Air Research 2012). No acetic acid was polypropylene has some effect but changes the values by
detected from sample A after 1, 3, 5, 8 and 10 years of less than 20%. The measured Lock & Lock box values
natural ageing (22 µg/m3). After 15 years, 1,023 µg/m3 in the actual storerooms are both somewhat higher and
of acetic acid was detected. Under natural ageing, both much less varied than those reported for an environ-
samples showed very delayed emissions of formic and mental chamber (Thunberg et al. 2021) and higher than
acetic acid, with extremely low amounts present in the the Gastronorm boxes. The 10-year-old boxes do appear
early years. This indicates a risk when these materials to have lower hygrometric half-lives, although there is
are used in showcases at ambient RHs. The newly significant overlap in the data. The values with most of the
developed ambient/low temperature tests – BEMMA permeation blocked with 3M 425 tape are significantly
(Bundesanstalt für Materialforschung und -prüfung higher still for unaged boxes and the 10-year-old boxes
n.d.), ISO 16000-9:2006 (ISO 2006) and SMART testing appear to perform to a similar level.
(National Conservation Service n.d.) – do not assess The Arrhenius plot from the accelerated ageing data is
ageing and will certify these materials, which will go shown in Figure 9.
on to pose a risk in several years.
As can be seen, the data follows a solid straight line and
Storage boxes the data from natural ageing also fits reasonably well. The
decrease in crystallinity observed would be expected to
The hygrometric half-lives of the boxes are shown in
increase permeability and may account for some of the
Figure 8.
decreased hygrometric half-life for the naturally aged
The values and spread of data for the Stewart Gastronorm Lock & Lock boxes in Figure 8. The overall performance
boxes are similar to those reported by Thunberg et al. of the box is also influenced by the seal. Experiments are
(2021). Blocking most of the permeation through the ongoing to assess its long-term ageing properties.
Conclusion International Organization for Standardization (ISO).

2006. ISO 16000-9: Indoor air – Part 9: Determina-
The necessity for three-point calibration at lower RH
tion of the emission of volatile organic compounds
values has been shown to generate accurate RH data in
from building products and furnishing — Emission test
dry environments. The design of showcases is impor-
chamber method. Geneva: International Organization
tant for the initial AER but much more so to ensure a
for Standardization.
showcase retains a low AER over time. Bottom-hinged
desktop showcases have been assessed to perform best, International Organization for Standardization (ISO).
followed by upright showcases with pull-and-slide doors. 2017. ISO 12569: Thermal performance of buildings and
The performance of regular silica gel does not appear materials – Determination of specific airflow rate in build-
to degrade over time (although bags will embrittle and ings – Tracer gas dilution method. Geneva: International
eventually burst, requiring replacement). Examples of Organization for Standardization.
emissions from enclosed materials have been measured
National Conservation Service (NCS). n.d. Product
after 3 or 4 weeks at 60°C but only after several years
testing. www.ncs.org.uk/producttesting.php (accessed
under ambient conditions. This confirms the need for
1 November 2021).
accelerated ageing to assess materials for long-term use
with metal collections. One brand of lockable polypro- Norwegian Institute for Air Research (NILU). 2012. The
pylene box has been found to perform somewhat better MEMORI Technology – Innovation for Conservation.
than the commonly used Stewart boxes under actual https://memori.nilu.no/ (accessed 1 November 2021).
day-to-day conditions. Its long-term ageing behaviour Thickett, D. 2012. Post excavation changes and preventive
has begun to be investigated. Arrhenius-type ageing conservation of archaeological iron. PhD dissertation,
showed increased permeability of the polypropylene. University of London, UK. Available at www.english-her-
The seal performance has yet to be studied but is another itage.org.uk/siteassets/home/learn/conservation/collec-
important contributor to the box’s utility. tions-advice--guidance/thickettthesisfinalversion.pdf
(accessed 6 August 2021).
References
Thickett, D. 2016. Critical relative humidity levels
Bundesanstalt für Materialforschung und -prüfung
and carbonyl pollution concentrations for archaeo-
(BAM). n.d. Focus on the low-emissions display case.
logical copper alloys. In Metal 2016: Proceedings of the
BEMMA-Scheme version 10-2020. Description of the
BEMMA investigative and evaluative procedure. https://
New Delhi, 26–30 September 2016, eds. R. Menon, C.
netzwerke.bam.de/Netzwerke/Content/DE/Downloads/
Chemello, and A. Pandya, 180–87. New Delhi: Interna-
Nike/bemma-testprocedures.pdf?__blob=publicationFile
tional Council of Museums-Committee for Conservation
(accessed 1 November 2021).
(ICOM-CC) and Indira Gandhi National Centre for the
Calver, A., A. Holbrook, D. Thickett, and S. Weintraub. Arts (IGNCA).
2005. Simple methods to measure air exchange rates
and detect leaks in display and storage enclosures. In Thickett, D. 2018. Frontiers of preventive conservation.
ICOM-CC 14th Triennial Meeting Preprints, The Hague, Studies in Conservation 63(suppl. 1): 262–7.
12–16 September 2005, ed I. Verger, 597–609. London: Thickett, D. 2020a. Sustainable collections environments.
James & James. Available at www.icom-cc-publica- ECR – Estudos de Conservação e Restauro 11(2020:
tions-online.org/ YOCOCU Conference 2019: 3rd International Conference
Farzana, N. and S. Akter Smriti. 2015. Reflectance value of Green Conservation of Cultural Heritage, Porto, Univer-
and yellowing properties of cotton treated with different sidade Católica Portuguesa). https://doi.org/10.34632/
softeners. International Journal of Current Engineering ecr.2020.n11
and Technology 5(10): 507–11. Thickett, D. 2020b. Specifying air exchange rates for
Gibson, L.T., B.G. Cooksey, D. Littlejohn, and N.H. showcases. In Chemical interactions between cultural
Tennent. 1997. A diffusion tube sampler for the determi- artefacts and indoor environment, eds. M. Adriaens, S.
nation of acetic acid and formic acid vapours in museum Bioletti, and I. Rabin, 25–48. Leuven/The Hague: ACCO.
cabinets. Analytica Chimica Acta 341(1): 11–19. http://hdl.handle.net/1854/LU-8550694
Thickett, D. 2021a. Oxygen depletion testing of metals. tions and carrying out research to improve their care.
Heritage 4(3): 2377–89. Previous research has focused on showcases and pollu-
Thickett, D. Forthcoming. Simple, accessible modelling tion, amongst other areas. Prior to working at English
for showcase performance. In International Conference Heritage, he studied for a PhD in collaboration with
on Museums Environment: Challenges and Opportunities, English Heritage, The National Trust and Historic Royal
14–15 December 2021, Cairo. Palaces. This research looked at the impact of climate
change on historic interiors.
Thickett, D. and M. Odlyha. 2010. Assessment of dry
storage microenvironments for archaeological iron. In The
conservation of archaeological materials: Current trends
and future directions, BAR International Series 2116,
eds. E. Williams and C. Peachey, 187–99. Oxford: BAR
Publishing. https://doi.org/10.30861/9781407306575
Thickett, D. and M. Odlyha. 2013. The formation and
transformation of akaganeite. In Metal 2013: Preprints of
Edinburgh, 16–19 September 2013, eds. E. Hyslop, V.
Gonzalez, L. Troalen, and L. Wilson, 103–109. Edinburgh:
ICOM-CC and Historic Scotland.
Thickett, D., P. Fletcher, A. Calver, and S. Lambarth. 2007.
The effect of air tightness on RH buffering and control.
In Proceedings of Museum Microclimates, Copenhagen,
Denmark, 19–23 November 2007, eds. T. Padfield and
K. Borchersen, 245–52. Copenhagen: National Museum
of Denmark.
Thomson, G. 1977. Stabilisation of RH in exhibition
cases: Hygrometric half-time. Studies in Conservation
22(2): 85–102.
Thunberg, J.C., D.E. Watkinson, and N.J. Emmerson.
2021. Desiccated microclimates for heritage metals:
Creation and management. Studies in Conservation 66(3).
https://doi.org/10.1080/00393630.2020.1799599
Weintraub, S. 2002. Demystifying silica gel. In Objects
Speciality Group Postprints, Volume Nine, 2002, orgs.
V. Greene and P. Griffin, 169–94. Washington, DC:
The American Institute for Conservation of Historic &
Artistic Works.
Yakimets, I., D. Lai, and M. Guigon. 2004. Effect of
photo-oxidation cracks on behaviour of thick poly-
propylene samples. Polymer Degradation and Stability
86(1): 59–67. https://doi.org/10.1016/j.polymdegrad-
stab.2004.01.013
Author
Paul Lankester is a conservation scientist at English
Heritage responsible for caring for our historic collec-
CONSERVATION PRACTICES
The Reanimation of a Renaissance Automaton: Diana and
Stag at the Museum of Fine Arts, Boston
Susanne Gänsicke* Rolf Lang

J. Paul Getty Museum Independent watchmaker and
Los Angeles CA, USA conservator
[email protected] Dresden, Germany
[email protected]
Abstract
The preservation of museum artifacts with a functionality, paper traces the multifaceted history of one such elaborate
whether musical instruments, industrial heritage, or other Trinkspiel, Diana and Stag, housed at the Museum of Fine
works of art, presents numerous challenges. Mechanical devices Arts, Boston. Over time, both its appearance and function
date back to antiquity, were explored in the 12th century by deteriorated. The treatments applied to address the alterations
the Turkish engineer Al-Jazari, and during the Renaissance in of the automaton and the decision to restore its mobility are
Europe generated considerable interest. Among the many types reviewed in the context of other “moving objects” and their
of devices were self-propelled automata that featured opulent, mechanisms.
three-dimensional precious metalwork of mythological scenes
and other figural compositions. During drinking games, these Keywords
apparatuses likely moved across banquette tables, powered automaton, Joachim Fries, gilded silver, Renaissance, reverse
by wind-up mechanisms incorporated into their bases. This engineering, Trinkspiel, watchmaker, wind-up mechanism
Introduction
The preservation of museum artifacts with a function- and engineer Al-Jazari explored and developed moving
ality, whether musical instruments, industrial heritage, machines, such as water clocks, on a large scale, and
or other works of art, presents numerous challenges. though none survived, his book and drawings provide
In the following we present a case study, based on the a glimpse of his ingenious inventions (Al-Jazari 1974).
Diana and Stag automaton by Joachim Fries (Figure 1), After the development of clockworks in Europe in the
to discuss the strategies and resources needed for the 14th century, craftsmen across disciplines used state-
conservation of this and related objects. The article of-the-art technologies to create animated displays and
traces the multifaceted history of this Trinkspiel, or automata. These objects were intended as luxury goods,
dinking game, which is housed at the Museum of Fine entertainment, and for political messaging (Maurice and
Arts, Boston (MFA). Over time, both its appearance and Mayr 1980, Keating 2018, Koeppe 2019).
function deteriorated. The treatments applied to address
Renaissance automata feature opulent, three-dimen-
the alterations and the decision to restore the automaton’s
sional, precious metalwork depicting mythological
mobility are discussed in the context of other “moving
scenes and other figural compositions. Self-propelled
things” and their mechanisms.
apparatuses are said to have navigated banquette tables
Mechanical devices date back to antiquity, although they by wind-up mechanisms incorporated into their bases.
are primarily known from the literature, which is the During ritualized festivities and ceremonies, special
source of the term automaton (Greek: “acting of one’s own drinking vessels and automata, referred to as Trinkspiel,
will”). The most famous surviving ancient apparatus, the featured prominently. These extravagant gold and silver
Antikythera Mechanism, from the 2nd century BC, has machines were highly treasured possessions, along with
been compared to an early computer (Marchant 2009, exotic materials and imports, in German princely collec-
Freeth et al. 2021). In the 12th century, Turkish scholar tions and would have been displayed in the Kunstkammer
THE REANIMATION OF A RENAISSANCE AUTOMATON: DIANA AND STAG AT THE MUSEUM OF FINE ARTS, BOSTON 273
(1544–1631/32), manufactured a group of about 30

Diana and Stag automata (Seelig 1994, 154–9). Those by
Fries were found in eight collections in 1967 (Dietschy
1967), but since then more have appeared, or perhaps
reappeared or circulated.1 At least five of the automata
(in addition to the MFA’s Diana) contain a wind-up
mechanism; but if these are original and functional is not
known to the author.2 Even more spectacular self-pro-
pelled apparatuses and Tischspiele with more complex
moving superstructures are known from the same period
as Fries’ Diana and Stag (Seelig 1994, 137–8).
The desire for European luxury goods from as far as

India was documented by Francisco Pelsaert in the
17th century, in lists of items that were shipped from
England to Asia, which included “art ware inlaid with
gold and gems … new inventions and curiosities such
as have never seen before” (Moreland and Geyl 1925,
25–26).3 A 1620 watercolor from the St. Petersburg album
depicts a fictitious meeting between Jahangir and Shah
Abbas, displaying a table of luxury goods from the East
and West, and a courtier cradling a Diana automaton
(Figures 2, 3) (Ettinghausen 1961, pl. 13). The dates of
the watercolor and the creation of the Diana automata
correspond well and support the notion of the contem-
Figure 1. Diana and Stag automaton, Trinkspiel, MFA 2004.586a-d, porary import of this new invention to the Mughal court.
Joachim Fries (1579–1620), German (Augsburg), about 1610–20. Cast
and chased silver, partially gilded and painted with translucent lacquers, The fascination with moving things, and animation in
iron, wood. Height: 33 cm; width: 24.3 cm; diam. 25.4 cm. Other (base
general, was and remains universal. The repair and treat-
dimensions): 24.3 × 10.2 cm. Museum purchase with funds donated
anonymously and the William Francis Warden Fund, Frank B. Bemis Fund, ment of moving clockworks and automata lie at the inter-
Mary S. and Edward Jackson Holmes Fund, John Lowell Gardner Fund, section of horology and conservation. Multidisciplinary
and by exchange from the Bequest of William A. Coolidge. Photograph approaches are required that combine craftsmanship and
© Museum of Fine Arts, Boston
skills in the fabrication of mechanical gears and springs
with historical connoisseurship, scientific examination,
or Wunderkammer, the forerunner of modern museums. as well as conservation and ethical concerns. However,
The southern German city of Augsburg was a center the conservation literature on automata, and especially
of goldsmithing and production and held a monopoly on Renaissance mechanisms, is limited (Lang 1987;
on these courtly drinking amusements during the Lorenz 2008; Vincent 2016, 75).4 The conservation of a
early 17th century, when a large inventory took shape mechanical bird formed the central element of a novel,
(Seling 1994, 22). Today, the Grünes Gewölbe, Staatliche perhaps loosely based on the Bowes Museum’s Silver
Kunstsammlungen, Dresden, holds the richest collection Swan automaton.5
of these objects (Syndram and Schermer 2004).
The Diana and Stag automaton at the MFA
Diana and Stag was a popular theme, as the goddess of
the hunt was well-suited for festivities that often revolved The superstructure of the MFA’s Diana and Stag autom-
around hunting (Michie 2018). The story of Diana and aton, created by Joachim Fries likely between 1610 and
the stag is told in Ovid’s Metamorphoses, whose alle- 1620, is an example of Augsburg’s finest goldsmithing
gories also influenced other Kunstkammer creations. tradition. A large hollow stag, wearing eardrops and
Three master goldsmiths, Jacob I. Miller (1550–1618), elaborate trappings, leaps upwards from an octagonal
Joachim Fries (1579–1620), and Matthias Walbaum base. A silver figure of Diana, with her quiver, sits side-
274 ICOM-CC | METAL 2022 | HELSINKI, FINLAND CONSERVATION PRACTICES
saddle. Under the stag, there is a group of smaller figures,

including two sizable dogs, both leashed to the goddess,
and several much smaller figures, including a small horse
(its rider has been lost), a tiny hunter, a rabbit, an array
of small lizards and insects, and a flower arrangement.
Their respective sizes bear no relation to reality. Some of
the hollow shapes were likely worked by repoussé, with
sections joined by soldering, and some parts, such as the
openwork, were cut from sheet metal. Other elements
were cast and some of the small animals may have been
life casts (Koeppe 2019, 45–47, 68). Most connections
between the elements are mechanical. The stag, base,
one dog, and selected decorative elements retain what
appears to be the original amalgam gilding. The small
animals are covered with what has been described as
Kaltemail or cold enamel, likely a type of paint (Weinhold
et al. 2018).6 The overall impression is one of rich shapes,
textures, and contrasting materials.
The hollow base contained a spring-based steel mech-

anism, to be wound up using a key; the released spring
would propel the work across a banquet table. The stag
and dogs are hollow and removable, as are their heads,
and would have contained wine during celebrations.
Figure 2. Jahangir entertains Shah Abbas with a display of international
During the Trinkspiel, once the automaton came to a
luxury items, circa 1618–22. Opaque watercolor, gold and ink on paper.
Height: 25 cm; width: 18.3 cm. Album leaf with painting. Attributed to
standstill, the diner closest to it removed an animal head,
Bishandas / Freer Gallery of Art, Smithsonian Institution, Washington, sipped or drank the wine from its body, or perhaps fully
D.C.: Purchase – Charles Lang Freer Endowment, F1942.16a emptied the animal; the large stag was to serve a man,
the smaller dog a woman (Menzhausen 1987).
The Diana and Stag automaton traced a far longer

journey prior to joining the collection of the MFA in
2004. After its creation by Fries, the object entered the
possession of Prince Heinrich the Younger of Reuss,
known as Posthumus (1572–1635), between 1610 and
1620, and was kept in the possession of the family at
Osterstein Castle, Gera, Germany. During World War II,
the castle was completely destroyed in an air raid on
April 6, 1945, and its ruins were leveled by deliberate
detonation in 1962. Diana was presumed lost, along
with other precious collections. While the princely
family was rumored to have hid their treasures in the
garden for fear of looting, it is also possible that the
automaton was among the objects taken from a shelter
Figure 3. Detail with Diana automaton. Jahangir entertains Shah Abbas at the castle when the House of Reuss was dispossessed
with a display of international luxury items, circa 1618–22. Opaque
by the Soviet Army. Nonetheless, some artworks were
watercolor, gold and ink on paper. Height: 25 cm; width: 18.3 cm.
Album leaf with painting. Attributed to Bishandas / Freer Gallery of Art,
discovered during construction work at the beginning of
Smithsonian Institution, Washington, D.C.: Purchase – Charles Lang Freer the 1980s and reappeared in 1985 in a private collection,
Endowment, F1942.16a (detail) allegedly as an inheritance.
Thus, Diana and Stag and the other artworks assumed to According to Lang (1987), the surface of the silver and
have been lost came into possession of the city of Gera, gilded silver was sandy, while the steel mechanism was
which secured the objects in the Museum für Geschichte, not fully recognizable and its condition resembled that
where they were partially conserved and displayed in of an archaeological find. There were some losses to
1987 (Menzhausen 1987, Mues 1987).7 After Germany’s the small figures as well as breaks and damage to the
reunification in 1990, Diana was restituted to the princely silver openwork. The gold surface of the stag exhibited
House of Reuss but then offered in the art market more both blue-green copper corrosion related to its alloy
than a decade later, when it was acquired by the MFA.8 composition, and iron stains, and some of the silver was
blackened. Surface cleaning of the superstructure was
Conservation history carried out with complexing agents, EDTA, and thiourea,
The gap in Diana’s history created by World War II followed by buffing with a wool cloth and jeweler’s rouge
leads to speculations derived from careful examina- and surface-coating with Paraloid B-72. Thus the orig-
tion of the automaton’s condition. Safekeeping during inal aesthetic appearance of the object was restored, but
the war, and afterwards, may have included burial or its function was not. The steel drive was mineralized to
an unsuitably moist environment. After the object’s a degree that prevented cleaning and it was therefore
arrival at the Museum in Gera in the 1980s, it was docu- treated with tannin followed by wax impregnation. From
mented and treated by conservator and watchmaker that point forward, the mechanism remained separated
Rolf Lang (Lang 1987). The corrosion observed on the from its superstructure. Overall, the treatment held up
different metals of the automaton is an excellent illus- well, and in 2004, upon the automaton’s arrival at the
tration of the behavior of precious versus non-precious MFA, both parts were stable and apparently unchanged,
metals in corrosive environments and of the enhanced except for minor tarnish.
galvanic corrosion that occurs when different metals
are in contact. When Lang first saw the automaton, Fries’ Diana and Stag had remained largely unaltered
the gears had been separated from the superstructure since its manufacture, and its shapes and surfaces are
but whether they had been in direct contact during the consistent with the original concept. A small rock
time of possible burial remains unknown. The gold crystal cross was added to a chain in Diana’s hand, which
and silver survived exposure or burial for over 40 years would originally have held a dog leash. When and why
in a corroded yet salvageable condition, but the steel this modification was carried out is unknown. This
wind-up mechanism did not and was mineralized and rare, nearly original condition is important, since most
damaged beyond repair (Figure 4). automata were altered and repaired over time, as demon-
strated by the Metropolitan Museum’s Diana, described
as having undergone alterations in the 19th century
(Seelig 1991, 112).
Re-creation of the wind-up mechanism

Lang had carefully documented the individual parts in
1986 and was able to recognize fragments of toothed
wheels or broken levers. From this information, he
worked out the motor’s construction, calculated the
gear train, and determined the axel distances. Given
his expertise in the conservation of Renaissance objects
and his familiarity with the concepts, techniques, and
design of related mechanical works, Lang was interested
in making a copy of the motor. However, at that time the
project was not pursued and a few years later, when the
Figure 4. Diana and Stag automaton, MFA 2004.586a-d. Deteriorated
steel wind-up mechanism, top and partial side views. Photograph © political landscape in Germany changed dramatically,
Museum of Fine Arts, Boston the automaton left Gera.
The idea of creating a new mechanism was revisited at

the MFA soon after it was realized that Lang was the
first conservator to have examined Diana. This led to
discussions about the possibility of reverse engineering
the motor. The information preserved in the corroded
but relatively intact gears was of critical importance and
its potential was recognized by Lang.
The Renaissance wind-up mechanism may be perma-

nently defunct, but it retains value and information vital
for the interpretation of the entire object. X-radiographs
Figure 6. 3D reconstruction of the mechanism. Graphic © Rolf Lang
taken at the MFA provided new insights into the wind-up
gears that allowed the approximate dimensions of the side
plates for the gears, wheels, drives, spring, and a wooden
spindle to be determined (Figure 5).
Figure 7. Separate, new gear parts prior to assembly. Photograph ©

Rolf Lang
Figure 5. Diana and Stag automaton, MFA 2004.586a-d. X-radiographs

showing gears; top and side views. 120 kV, 3 mA, 2.0 min, December 14,
2013 (for both). Photograph © Museum of Fine Arts, Boston
In 2013, the original, corroded work was shipped to

Lang’s workshop, near Dresden. After detailed study of
the elements, an initial 3D model was created (Figure 6).
Little is known about the original movement of Fries’
Dianas, nor that of the other Diana automata. Rosen-
berg (1922, 92) mentioned a run of 80 cm around a
Figure 8. The wind-up mechanism assembled. Photograph © Rolf Lang
square table for the Basel Diana, also by Fries. Based on
Lang’s calculations and an evaluation of the corroded
steel elements, the gears of the MFA’s Diana only move could be manufactured and positioned similar to the
forward, over a random course of 2.64 m. original (Figure 9). One difference is a larger cut-out in
the new base plate, as the position of the original wheel
New gears of the wheels and levers were made from steel could not be accurately established.
(grade ST35) and hand formed, as was done centuries ago,
although some larger parts were roughly shaped using Diana’s mechanism, although compromised, is fully
laser technology and then processed by hand (Figure 7). preserved without recent alterations, which allowed
The wooden worm was turned on a lathe (Figure 8). a determination of its individual parts and dimen-
Since the base plate, made of fire-gilded copper alloy, sions. This was particularly important since most of the
was available with its original slots and holes, the drive drinking games are now defunct or have been modified.
The automaton, with its new mechanism temporarily

installed inside the base (Figure 10) for demonstration
to MFA staff, was recorded in motion in a video. Today,
an interactive on an iPad at the museum allows visitors
to see Diana and Stag in movement, propelled by Lang’s
new gears.10 What the small screen does not convey is
the rather unruly and wobbly gait of Diana, the rasping
sounds of the works, and the jingling of the dangling
decorations, as also observed by Keating (2018, 212)
in other automata. While Diana scoots across the table
in formidable form, other types of automata flaunt
further animated features: some roll their eyes, shoot
arrows, or flap their wings. Videos of other moving
devices can be found at other museum websites, such
as the Kunstkammer video channel of the Kunsthistor-
isches Museum in Vienna, which features a number of
astounding examples. 11
Figure 9. Diana and Stag automaton, MFA 2004.586a-d. The old

mechanism next to its reconstruction. Photograph © Rolf Lang
Figure 10. Rolf Lang inserts the new mechanism into the automaton in
April 2014. Photograph © Museum of Fine Arts, Boston
Ultimately, the information was applied to create a new
drive almost identical in its technology and design. While Diana is on display, the two mechanisms, old and
new, remain in storage and are available for further study.
The study of the Diana and Stag automaton opens ques-
Preservation also encompasses the full research potential
tions for further research and technical examination. It
of the object. Beyond mechanics, the mineralized element
would be of interest to compare the extant mechanisms
retains memories of recent German history. Its inclusion
and their movement in greater detail to determine the
in future displays, real or virtual, will contribute to reflec-
similarity of the works and their movements.9 In the
tions on the horrors of war or on a particular aspect of
process, more might be learned further about the clock-
the German narrative.12
makers who produced the Trinkspiele, as they generally
did not sign their works (Vincent 2016, Keating 2018).
Conclusion
During the project at the MFA, 3D scanning was not
The MFA’s Diana and Stag was set again into motion with
employed, but it may offer new information about the
a working replica of its original gears, while the Renais-
interior of the base. The development of digital models
sance drive remains highly compromised but preserved
and their movement may provide new possibilities
and available for future examination and interpretation.
for comparing and quantifying the movements of the
In Boston, a video of the animated Diana automaton,
remaining automata.
with its new, temporarily installed wind-up mechanism
and in motion, has become integral to its display in the
Movement and presentation at the MFA
museum’s galleries. Lang brought not only his technical
The Kunstkammer gallery at the MFA features rare watchmaking skills to the project, but also his familiarity
and wonderful objects, including Diana and the Stag. with similar objects as well as a personal connection to
the object’s wartime woes.13 The story of this particular 4

M emoria Technica. https://mechanicalcurios.
object, its demise and resurrection, is deeply linked to com/49278-2/ (accessed 28 November 2021).
the 20th-century history of Germany: World War II, 5
Peter Carey, The chemistry of tears, 2012, New York,
division of the country, and unification. As history has
Alfred A. Knopf. Bowes Museum, Durham, UK: www.
shown, movement of these machines across continents
thebowesmuseum.org.uk/Collection/Explore-The-Col-
is a continuum, as will hopefully be the new knowledge
lection/The-Silver-Swan (accessed 29 November 2021).
gained and its renewed presentation.
6
he polychromy on the small shapes has not been
T
Acknowledgments analyzed to date. It is opaque with color modulations
typical of paint or lacquer, rather than enamel.
At the Museum of Fine Arts, Boston, Curator of Deco-
7
See Gera-Chronik: www.gera-chronik.de/www/gera-
rative Arts Tracy Albainy (†) enthusiastically supported
historie/ (accessed 28 November 2021).
initial discussions in 2004 about recreating the mecha-
nism, after the object was acquired; Richard Newman 8
S old by the princes of Reuss to Rudigier Gallery, Munich
provided radiography. Special thanks to Thomas S. Michie and London. Sold in 2004 by Rudigier Gallery to the
for helping to realize the project, to MFA photographers MFA.
and the digital team, and to Abigail Hykin for assisting 9
ang believes the complexity of the automata would
L
with questions and logistics. demand individual manufacturing rather than mass
production, although some standardized elements may
Notes have been used.
1
ccording to Dietschy (1967), some 20 Diana and
A 10
Museum of Fine Arts, Boston: www.youtube.com/
Stag automata can be attributed to Joachim Fries,
watch?v=EbmhpNaVtts (accessed 28 November 2021).
Walbaum, and Miller. He lists eight Dianas by Fries:
The installation was well and generously received
Basel (Historisches Museum Basel), Darmstadt (Hessis-
by the local press: www.wbur.org/news/2014/06/05/
ches Landesmuseum), Frankfurt am Main (former
diana-mechanical-drinking-game-mfa (accessed 29
collection Freiherr Karl von Rothschild), Hamburg
November 2021).
(Sammlung Udo and Mania Bey), Lugano (Gräfin M.
Batthyany), Mailand (Ambrosiana), Moscow (Kremlin),
11
unsthistorisches Museum Wien: www.khm.at/en/
K
and New York (The Metropolitan Museum of Art), visit/collections/kunstkammer-wien/video-channel/
although the exact location could not be confirmed for (accessed 28 November 2021).
all objects. Since his publication, one more Fries Diana 12
O bjects with recent war damage are featured in
and Stag automaton was added to the collection in museums, for example, in the Neues Museum, Berlin,
Liechtenstein (2010) and to the Museum of Fine Arts, Germany, and the National Museum, Beirut, Lebanon.
Boston (2004). He omitted the Walters Art Gallery 13
he author wishes to acknowledge a deep personal
T
Diana automaton.
connection to this project. Born and raised in West
2
istorisches Museum Basel (Inv. 2005.336); Liechten-
H Berlin, this collaboration would have been beyond
stein, The Princely Collections, Vienna (Inv. No. SI261); imagination for the first half of my life.
Metropolitan Museum of Art, New York (17.190.746);
Sammlung Udo and Mania Bey; and Walters Art References
Gallery, Baltimore (57.923), which is dated to either
Dietschy, K.A. 1967. Der Werthemannsche Hirsch und
1620–1635 or the early 19th century.
seine Verwandten. Historisches Museum Basel, Jahres-
3
n illustrious, precious object, the Gemma Constan-
A berichte: 29–39.
tiniana, combining a large ancient gem in an elaborate
Ettinghausen, R. 1961. Paintings of the sultans and
bejeweled and enameled gilt silver frame (Rijksmuseum,
emperors of India in American collections. New Delhi:
Leiden, GS-11096), was in Peter Paul Ruben’s collection
in 1620. It was likely destined for sale to the Mughal Lalit Kala Akademi.
Court but was shipwrecked near the coast of Australia Freeth, T., D. Higgon, and A. Dacanalis, L. MacDonald,
in 1629 (Henderson 2016, 18–31). M. Georgakopoulou, and A. Wojcik. 2021. A model of
the cosmos in the ancient Greek Antikythera Mech- Seelig, L. 1994. Willkommensgruβ und Willkommen-
anism. Scientific Reports 11: art. 5821. https://doi. strunk. Figurliche Trinkgefäβe – Trinkspiele – Tischauto-
org/10.1038/s41598-021-84310-w maten. In Silber und Gold: Augsburger Goldschmiedekunst
Henderson, G. 2016. The Dutch East India Company’s für die Höfe Europas, eds. R. Baumstark, H. Seling, L.
ship Batavia. Lost in 1629. In Swallowed by the Sea: The Seelig, U. Arnold, and Bayerisches Nationalmuseum,
Story of Australia’s Shipwrecks, 18–31. Canberra: National 137–73. München: Hirmer.
Library of Australia. Seling, H. 1994. Einleitung. In Silber und Gold: Augs-
Al-Jazari, I.R. 1974. The book of knowledge of ingenious burger Goldschmiedekunst für die Höfe Europas, eds. R.
mechanical sevices, trans. D.R. Hill. Dordrecht/Boston: Baumstark, H. Seling, L. Seelig, U. Arnold, and Bayer-
D. Reidel Publishing Company. isches Nationalmuseum, 17–31. München: Hirmer.
Keating, J. 2018. Animating empire: Automata, the Holy Syndram, D. and A. Schermer. 2004. Princely splendor:
Roman Empire, and The Early Modern World. University The Dresden Court 1580–1620. New York/Dresden: The
Park: The Pennsylvania State University Press. Metropolitan Museum of Art/Staatliche Kunstsammlu-
ngen Dresden.
Koeppe, W. 2019. Making marvels: Science and splendor
at the courts of Europe. New York: The Metropolitan Vincent, C. 2016. Diana, goddess of the showering arrows:
Museum of Art. A seventeenth-century German automaton clock. Yale
University Art Gallery Bulletin, European Art: 75–87.
Lang, R. 1987. Die Restaurierung des Geraer Diana-Au-
tomaten. Neue Museumskunde 30(4): 292–4. Weinhold, U., T. Witting, E. Begov, C. Herm, S. Hoblyn,
R. Richter, M. Willert, and Staatliche Kunstsammlungen
Lorenz, K. 2008. Les automates du XIXème siècle:
Dresden. 2018. Natürlich Bemalt: Farbfassungen auf
Réflexion sur la conservation préventive et la restaura-
Goldschmiedearbeiten des 16. bis 18. Jahrhunderts am
tion. Coré: Conservation et Restauration du Patrimoine
Dresdner Hof. Dresden: Sandstein Verlag.
Culturel 20(2008): 15–21.
Marchant, J. 2009. Decoding the heavens. Solving the Authors
mystery of the world’s first computer. London: Windmill
Books. Susanne Gänsicke, Head of Antiquities Conservation,
J. Paul Getty Museum, Los Angeles, USA, graduated in
Maurice, K. and O. Mayr. 1980. Die Welt als Uhr: Deutsche archaeological conservation (Römisch Germanisches
Uhren und Automaten, 1550–1650. München: Bayerisches Zentralmuseum, Mainz, Germany, 1987). She was objects
Nationalmuseum.
conservator at the Museum of Fine Arts, Boston (1990–
Menzhausen, J. 1987. Der Geraer Diana-Automat von 2016), where she remains a project member of “The auloi
Joachim Fries. Neue Museumskunde 30(4): 290–91. from Meroë” (funded by the European Research Council).
Michie, T.S. 2018. Hunter or hunted? A drinking game of Her research interests include manufacturing techniques
chance. In Animal-shaped vessels from the ancient world: of ancient and historic metalwork, and the intersection
Feasting with gods, heroes, and kings, ed. S. Ebbinghaus, of traditional craftsmanship and conservation. In 2016,
178–9. Cambridge, MA: Harvard Art Museums. she received an individual grant for fieldwork on metal
preservation in Nepal (Asian Cultural Council, NY).
Moreland, W.H. and P. Geyl, transl. 1925. Jahangir’s India:
She co-authored, with Yvonne Markowitz, Looking at
The Remonstrantie of Francisco Pelsaert. Cambridge: W.
Jewelry (2019).
Heffer & Sons Inc.
Rolf Lang, master watchmaker in private practice (since
Mues, S. 1987. Wertvolle Kunstschätze an die Museen
2012), is developing a high-quality watch line, Canaletto,
der Stadt Gera übergeben. Neue Museumskunde 4: 287–9.
and exquisite art objects. He holds a degree from the
Rosenberg, M. 1922. Der Goldschmiede Merkzeichen R3. University of Applied Sciences for Finance, Gotha, and
Frankfurt am Main: Frankfurter Verlagsanstalt. a diploma in conservation (1985) from the Deutsches
Seelig, L. 1991. Die Gruppe der Diana auf dem Hirsch in Museum, Berlin, GDR. In 1981, he founded the Watch
der Walters Art Gallery. The Journal of the Walters Art Restorers working group in Dresden. Lang served at
Gallery 49/50: 107–18. the State Mathematical and Physical Salon, Dresden, as
clock conservator (1979–1987) and as head conservator

(1987–1989). In 1989, he founded a conservation studio
and continued to develop watches. In 2003–2007 he was
operations manager/senior developer for H. Moser &
Cie, Schaffhausen.
Building an Ironclad System: A Quarter-Century of
Innovation in the Conservation of USS Monitor
Erik Farrell Lesley Haines Molly K. McGath William Hoffman*

The Mariners’ Museum and Park The Mariners’ Museum and Park The Mariners’ Museum and Park The Mariners’ Museum and Park
Newport News VA, USA Newport News VA, USA Newport News VA, USA Newport News VA, USA
www.marinersmuseum.org www.marinersmuseum.org www.marinersmuseum.org www.marinersmuseum.org
Elsa Sangouard Laurie King *Author for correspondence

The Mariners’ Museum and Park The Mariners’ Museum and Park
Newport News VA, USA Newport News VA, USA
[email protected]. [email protected]
www.marinersmuseum.org www.marinersmuseum.org
Abstract
In 1987, The Mariners’ Museum and Park partnered with the was completed in less than three years, it took time to build out
National Oceanic and Atmospheric Administration (NOAA) to internal capabilities as the conservation staff gained experience
be the official repository of artifacts raised from the United working with, or at times even within, the artifacts. This work
States’ first National Marine Sanctuary. Starting in the late 1990s, allowed conservators to fully understand the artifacts’ needs,
archaeologists from NOAA, partnering with the US Navy, began which created pathways to tailor the laboratory system and make
a major effort to recover artifacts from the wreck of USS Monitor the project more efficient. While the conservation of USS Monitor
(1862), the US Navy’s first commissioned ironclad. Objects is a massive endeavor from every conceivable angle, it has been
retrieved from the vessel encompassed nearly the entire engine made possible through continued adaptation by conservators and
room and its iconic revolving gun turret. With the arrival of supporting staff at the Museum. The impacts of this effort have
Monitor’s turret in 2002, the Museum held over 208 tonnes of improved all aspects of the project, including documentation,
archaeological material and had to build a conservation program health and safety, space and scale needs, treatment and research,
from the ground up. The huge scale of the project initially work planning, expertise, public engagement, and funding
necessitated the use of materials on hand, rather than an ideal challenges. Looking at the project from these perspectives, this
facility and equipment. Objects were stored outside in tanks paper focuses on sharing nearly 25 years of growth, change, and
as there was not a dedicated conservation space large enough learning undertaken to conserve artifacts from USS Monitor.
to contain them. Following a multi-year capital campaign, the
Batten Conservation Complex (BCC) officially opened along Keywords
with an artifact exhibition space in 2007, having been designed marine archaeological conservation, long-term project,
utilizing available conservation knowledge. Although the BCC industrial archaeology, shipwreck, conservation laboratory
Introduction
To date, the USS Monitor Project is the largest marine the facilitation of long-term project administration, all
archaeological metals recovery and conservation initi- while remaining publicly engaged over multiple decades.
ative ever undertaken. Due to the scale and complexity
of the retrieved artifacts, personnel at The Mariners’ Background
Museum and Park have had to chart a relatively unknown
course over the last 20 years. Not only have staff had to On December 31, 1862, USS Monitor was caught in a
build a robust conservation program from scratch, they storm and sank 16 nautical miles off the coast of Cape
did so before and during the construction of a conserva- Hatteras, North Carolina, in over 70 meters of water.
tion facility. This paper highlights experiences gained and The vessel sank stern first and turned over such that its
lessons learned on how the Museum has dealt with space revolutionary 104-tonne revolving gun turret separated
and physical considerations, conservation documenta- and became pinned under the ship on the seafloor. The
tion, research and treatment, health and safety issues, and remains of the vessel were discovered in 1973, confirmed
in 1974, and in 1975 the wreck fell under the jurisdiction Space and physical considerations
of the National Oceanic and Atmospheric Administration
From the onset of excavations, space and physical consid-
(NOAA) with the establishment of the Monitor National
erations became the primary challenge. Transferring
Marine Sanctuary.
objects to the Museum required coordination between
Archaeological investigations and recovery of small several institutions and the use of multiple ships, cranes,
objects were conducted throughout the late 1970s and and flatbed trucks. However, communication was not
early 1980s, and in 1987 The Mariners’ Museum and always clear on what was arriving and when, making it
Park in Newport News, Virginia, was designated as difficult to plan how to best receive materials. Storage
the repository for the management and curation of tanks were acquired as needed; larger assemblies like the
Monitor-related materials. turret required custom tanks from the beginning, but
By the late 1990s, NOAA archaeologists determined many smaller objects were stored in converted dump-
that Monitor was in an advanced state of deterioration sters as they were large, watertight, and easily available
and planned a major recovery effort in partnership with (Figure 2). It was clear that this ad hoc situation was not
the US Navy to retrieve the ironclad’s most significant ideal and that dedicated facilities were required.
artifacts (Broadwater 2012). Due to the turret resting
underneath the aft portion of the vessel, a multiphase
plan was developed to retrieve the engineering section
first, followed by the turret (Figure 1).
Figure 2. Early image of the conservation facilities, with large dumpsters

in the foreground used for the wet storage of engineering components
and the large tank in the background containing the turret. Courtesy of
The Mariners’ Museum and Park
The Batten Conservation Complex

Figure 1. Detail of the recovery plan for USS Monitor. Courtesy of NOAA,
Monitor Collection Completed in 2007 as part of a USD 31 million capital
Between 1998 and 2001 nearly the entirety of Monitor’s campaign, The Batten Conservation Complex (BCC) is a
engine room was recovered and transferred to the 1,850-square-meter facility consisting of a large industrial
Museum via barge. Objects included the propeller and space for wet object treatment (the wet laboratory), a
nine-meter shaft, skeg assembly (supported propeller clean laboratory for final treatment of small to midsized
and rudder), the 24-tonne steam propulsion engine artifacts, and a dedicated collections storage area. Perma-
and its supporting bulkheads, hull fragments and other nently positioned within the wet laboratory were seven
structural members, four additional steam engines, steam oversized tanks housing the turret, propulsion engine,
lines, plumbing, stairs, railings, the propulsion engine steam condenser, and the turret’s two Dahlgren guns
throttle and reversing wheel, engine room flooring, and and their carriages (Figure 3). All tanks were installed
the steam condenser. In 2002, structural elements over the before the walls and roof were built. Integrated trench
turret were separated and the massive object was raised and plumbing systems were constructed into the floor,
using a two-part lifting system. With the turret’s arrival, allowing the movement of storage solutions to and from
the Museum became home to approximately 20 percent the artifact tanks and acting as an emergency water
of USS Monitor. capture and leak/overflow detection system. To move
BUILDING AN IRONCLAD SYSTEM: A QUARTER-CENTURY OF INNOVATION IN THE CONSERVATION OF USS MONITOR 283
Figure 3. View of the wet laboratory with the square engine tank in
the foreground, the octagonal turret tank in the background, and the
yellow bridge crane above. Courtesy of The Mariners’ Museum and Park
artifacts and materials around, an overhead bridge crane

was installed with the capacity to lift up to 41 tonnes.
Space limitations
Although originally intended to house many more large
tanks, once staff occupied the facility, open space was
quickly consumed for other purposes. Due to a short Figure 4. Large objects being removed from their tank via a contracted
planning and construction timeline, compounded by crane for mechanical cleaning. Courtesy of The Mariners’ Museum and
the inability to fully know how the conservation project Park
would unfold, the long-term functionality of the labo-

ratory could not be vetted. Over time, areas within the
wet laboratory were carved out for a fabrication shop,
X-radiography booth, additional equipment, and spaces
to physically work on artifacts.
As a consequence of limited space, some artifact tanks

had to remain outside. This required additional fund-
raising to build an exterior storage, treatment, and work
area. However, the exterior facility did not include an
overhead crane, so conservation staff must contract a
local lifting and rigging company to move objects in
and out of treatment tanks (Figure 4). Therefore, crane
rental fees and a detailed schedule of object treatment
have had to be built into the yearly budget.
Another major factor affecting laboratory space has been

composite object disassembly.
As described by Krop and Nordgren (2011), a key step

in the treatment of Monitor’s artifacts is disassembling
them, treating the pieces individually, and ultimately reas-
sembling them, if feasible, after conservation. While the
Figure 5. (A) propulsion engine after deconcretion; (B) rack storage of
24-tonne propulsion engine would never be described as small components removed from larger engineering assemblies in the
small, it is at least compact (Figure 5A). However, disas- steel tanks behind. Courtesy of The Mariners’ Museum and Park
sembling it exponentially increases its footprint. As a result, and electrolysis, with quarterly reports compiled by the
there is not enough room to simultaneously disassemble conservation department’s scientist and reviewed by the
and treat all objects. For example, Monitor’s condenser conservation team to guide future treatment decisions.
could not be disassembled until one gun carriage was
Composite artifact disassembly has been fully docu-
completed and its tank emptied, which was itself a complex,
mented using photographs, diagrams, and written notes
multiyear disassembly process requiring multiple treat-
(Figure 6). The collections manager has often assisted
ment tanks and containers for storing component parts
with disassembly, assigning newly removed component
(Figure 5B). Good space management is key, which in turn
parts with catalogue numbers and recording their exact
necessitates time management and organization.
orientation in consultation with the conservators. This
Spatial limitations have also affected long-term artifact process ensures that composite objects can be accurately
storage and display. At the time of construction, these reassembled later. Institutional knowledge alone is
areas faced the same unknowns, and the intended collec- unreliable; even without staff changes, it is unlikely that
tion storage area has since been consumed by growing people will remember exactly what they did years before.
numbers of microclimate cabinets and Marvelseal-lined
crates. Compounding the challenge, this space lost much
of its initial footprint after construction with the instal-
lation of a photography studio and walk-in refrigerator.
As the USS Monitor exhibition space was built long before

the start of major conservation work, it was impossible
to accurately plan for final artifact display. Consequently,
the Museum has been faced with the challenges of floor
loading in multilevel galleries, providing accessibility for
heavy lifting and rigging equipment to move artifacts,
and maintaining microclimates in aesthetically pleasing
display designs.
Project documentation
Figure 6. Diagram of fragments from a wool coat found in the turret with
As with any archaeological conservation endeavor, overall each assigned an individual object number. Courtesy of The Mariners’
Museum and Park
project, treatment, and artifact documentation is critical.
This can be challenging with a short-term project, but
even more so when working with a complex assemblage Photodocumentation
over multiple decades. Accurate records for numerous
Evolution in documentation techniques is common
small artifacts and complex composites must be main-
during lengthy conservation projects and this cannot
tained while dealing with staff turnover and technological
be more evident than with photographic documenta-
advances. This has been made easier by the addition of
tion. When the Monitor project began, film, slides, and
a dedicated collections manager who has maintained
early digital techniques were utilized. This has shifted to
object records and coordinated with federal personnel,
exclusively digital photography alongside 3D modeling.
allowing conservators to focus on treatment.
When possible, pre- and post-treatment photography
has been conducted utilizing professional, in-studio
Treatment documentation
photographic equipment. To capture treatment steps
Although conservators within the BCC are assigned in progress, smartphones have often been employed,
individual objects, overall laboratory treatment docu- protecting more expensive cameras from dirt and debris.
mentation has been a group effort. As of 2021, the wet Where image quality has been paramount, especially
laboratory houses 29 tanks containing either a single during larger treatment operations, a museum photogra-
large artifact assembly or batches of smaller objects. pher has been employed to document the work. This has
Shared responsibilities include tracking desalination allowed conservators to focus on the work at hand and
created easy pathways to provide images for marketing

and social media.
Research and treatments

Due to the scale and complexity of Monitor artifacts,
conservators have faced a variety of treatment challenges,
which were unpredictable at the onset of the project.
When a new conundrum has presented itself, often
there have been limited comparative reference examples.
Therefore, conservators have invested extensive time
and resources in modifying existing treatment methods
or developing new ones. The culmination of these Figure 7. Conservator Erik Farrell cleaning the barrel of USS Monitor’s
processes has improved laboratory efficiency, increased starboard Dahlgren gun utilizing a bespoke drilling system. Courtesy of
understanding of the stabilization needs of artifacts, and The Mariners’ Museum and Park
made future challenges more approachable. The nature of
this work also highlights the need for quality, adaptable Artifact disassembly
research skills to be emphasized in early conservation Artifact disassembly has been by far the project’s greatest
training. While suitable techniques can often be found
challenge. Disassembly has primarily been associated
in other conservation specialties and from industry,
with large mechanical components, having required
creativity is required in their application.
method and technique development for very specific
purposes. These include custom-made machinery to
Scale
separate materials, like hardened rubber gaskets holding
Although mechanically removing corrosion on small iron degraded metal elements together. Equipment was
artifacts is relatively straightforward, cleaning a six-meter designed utilizing elevated water temperatures to soften
diameter, three-meter-tall gun turret has been a much the gaskets, allowing conservators to take the components
more difficult and time-consuming task. To address apart without damage (Hoffman and Spohn 2021).
this, conservators adapted dry ice blasting technology
However, artifact disassembly has not always been
and established cleaning parameters for wrought iron
feasible, creating practical and ethical dilemmas regarding
(Hoffman and King 2019). Not only has the technique
non-separable composite objects. There is still no ideal
allowed more rapid cleaning of large surfaces, it has also
enabled easier and more effective removal of components, way to treat wooden objects with embedded iron elements
such as bolts and screws. Improved cleaning on fastener without having to “choose” one material over the other.
heads has allowed for a better grip with extraction tools, Therefore, multiple years of research have been devoted
minimizing damage. to corrosion inhibitors to protect metallic parts while
treating organic constituents (Sangouard et al. 2015,
A similar challenge of scale occurred when removing Sangouard 2019).
concretion within the two four-meter-long, seven-tonne
Dahlgren guns (Figure 7). While the boring of marine External expertise
archaeological artillery has been undertaken in the past,
these were the largest caliber guns to be cleaned to date Given the Monitor project’s industrial nature, some tasks,
(Farrell 2020). Therefore, after several years of designing such as re-supporting the turret, required external tech-
and adapting previous boring systems, and at a cost of nical expertise. The two-part lifting/supporting apparatus
approximately $25,000 (USD), BCC conservators part- used to raise and transport the turret in 2002 was never
nered with several companies to build a custom machine. intended as a storage mount, and a replacement was not
Importantly, the new equipment has been designed to funded in the initial planning. Consequently, access to the
accommodate multiple-sized guns, allowing its use on exterior of the turret’s roof was severely limited. When it
other artillery within the Museum’s collection and any became clear that a purposely designed support system
future archaeological projects (Farrell 2022). was needed, and that the conservation team did not have
the expertise or capability to lift the 104-tonne object,

they approached the local shipbuilding industry in 2011.
Having limited dedicated funding for large-scale projects,

conservators have often relied on in-kind or at-cost
support, which can take time to set in motion. For the
turret, the re-support endeavor took eight years of onsite
consultations, email threads, and executive meetings
to arrange. Ultimately, three independent shipyards
came together in 2019 to lift the turret using four
81-tonne-capacity bottle jacks. The old supports (Figure
8A) were removed and replaced by compact pedestal
stands, allowing conservation to progress (Figure 8B).
Figure 9. (A) Pallets of NaOH required for the turret tank; (B) custom
pouring device being used to add NaOH into the propulsion engine
tank. Courtesy of The Mariners’ Museum and Park
solution, conservation staff has to restrict access to the

laboratory, wear full-body Tyvek suits and full-face
respirators, and utilize an in-house fabricated pouring
apparatus to add chemical to the tank over approximately
three hours (Figure 9B).
The disposal of old treatment solutions has been arranged

by working with the regional wastewater treatment
provider and an independent environmental testing labo-
ratory. The testing laboratory conducts certified analyses
Figure 8. (A) Turret sitting on part of the recovery structure from 2002–
of treatment solutions for heavy metals: cadmium, chro-
2019; (B) current pedestal support system. Courtesy of The Mariners’ mium, copper, lead, nickel, zinc, and iron, and organic
Museum and Park nitrogen and phosphorus. The wastewater treatment
provider determines whether these solutions may be
Health and safety added to general wastewater streams and at what rate
this can be done, if the solutions must be further diluted,
Due to the scope of the USS Monitor Project, maintaining
or if specialized chemical disposal must be contracted.
proper health and safety measures has been critical.
Treatment tank capacities range from 75 liters for small
Public engagement
artifacts to over 350,000 liters for the gun turret, meaning
that staff must work in confined spaces, wear specialized One of the few knowns at the onset of the USS Monitor
respiration equipment, climb up ladders and scaffolding, project was that it would be a long-term endeavor.
and handle large amounts of chemicals. For example, a Explaining the process and length of treatment is not
1% (w/w) solution for the turret tank requires 3.5 tonnes always straightforward to communicate to broad audi-
of sodium hydroxide (NaOH) (Figure 9A). To mix this ences, but engagement with the public has been critical
to ensuring that the project remains relevant. Therefore, years to include Monitor collection storage and the clean
a large amount of staff time across multiple museum laboratory. Although tours impact productivity, because
departments has been dedicated to developing and some work must temporarily halted due to health and
executing public outreach programs, and these outreach safety considerations, they provide a fuller view of the
efforts continue to evolve. conservation process. This helps visitors, potential
partners, and donors understand why conservation can
Viewing windows take months, years, or even decades, and why a large
laboratory is necessary. Over time, experiences gained
The BCC was purposely built with large viewing windows
from these tours led the conservation staff to develop a
into the wet laboratory that enable the public to watch
range of conservation-based tabletop activities that they
conservation treatments unfold (Figure 10). The turret, can take into the galleries. These hands-on activities
propulsion engine, guns, and carriage tanks are visible have provided the public with insight into what goes on
through these windows, and the conservation process within the laboratory spaces when behind-the-scenes
is explained and supported by videos and regularly programming has been unavailable.
updated signage with content reflecting current activities.
Additionally, conservation staff have been encouraged to Virtual content
spend time in this part of the gallery answering visitor
questions (Figure 10A). Positive feedback about visual Since not everyone can physically visit the Museum,
access into the wet laboratory led the Museum to renovate investments have been made supporting virtual content.
the clean laboratory and install viewing windows in 2020. As part of the BCC’s construction, web cameras were
This provides the public with a more comprehensive view installed within the wet laboratory, aimed at several
of the overall conservation process. large artifacts, and are accessible via the Museum’s
website. Over time, the conservation department’s web
presence has expanded to include regular blogs about
ongoing conservation projects and the status of treat-
ments. In addition, the conservation team has worked
with digital content personnel at the Museum and
NOAA to develop material for Facebook, Instagram,
YouTube, and Sketchfab, reaching audiences outside
the local area and encouraging guests to keep up with
the conservation work.
Administration
From an administrative standpoint, the primary chal-
lenge to Monitor’s conservation has been the longevity
of the project and the lack of a clearly defined end date.
This has a direct impact on developing effective work
plans and maintaining financial support and project
personnel.
Figure 10. (A) Visitors viewing Monitor objects in treatment through Project funding
windows; (B) behind-the-scenes tour of the wet laboratory. Courtesy
To date, the total operating cost for the USS Monitor
of The Mariners’ Museum and Park
conservation project is over USD 15 million, with
major conservation efforts expected to last another
Behind-the-scenes and in-gallery programming
15–20 years. Financial support has come from local,
Following the construction of the BCC, the Museum state, and federal institutions (including dedicated funds
began providing behind-the-scenes tours of the wet from NOAA), multiple public and private grants, and
laboratory (Figure 10B). This has expanded over the many individual donors and philanthropists. This has
only been possible through a robust development and effort, rotating on and off the staff for both personal and
grant writing program, as well as institutional leadership professional reasons.
that continues to fundraise. Without sustained effort to
In addition to core department staff, volunteers have
maintain long-term financial support, the conservation
played a critical role in advancing the project, with many
project could not be completed.
bringing with them valuable professional expertise in
engineering and chemistry.
Work plan structure
Furthermore, to help facilitate training in marine archae-
As noted throughout this paper, the speed at which the
ological conservation, The Mariners’ Museum has hosted
overall project advances has been based on experience
several interns and fellows, having supplemented their
gained in working with the artifacts, development and classroom education with opportunities for hands-on
implementation of new conservation techniques, acqui- training, research, and publications.
sition of new laboratory equipment, and identification
and remediation of safety issues. Factors outside the Conclusion
conservation department’s control have also affected
the project timeline, including institutional leadership The key to the success of the Monitor conservation effort
changes, economic downturns, additional funding chal- has been the ability to adapt over time in association with
lenges, a world-wide pandemic, and staffing changes sustained support (monetary and programmatic) from
resulting in the loss of project knowledge. As a result, The Mariners’ Museum, NOAA, and other partners.
the utilization of a narrowly defined long-range (e.g., From the onset, there were many unknown treatment
five-year) conservation plan has been found impractical. challenges, ethical considerations, and engineering
Planning has shifted towards structured twelve-month hurdles to overcome. Most importantly, the conservation
team has had the freedom to face these unknowns without
work plans targeted at different parts of the collection
the pressure of arbitrary treatment deadlines. This has
each year with the overarching goal of advancing the
empowered them to take the time required to fully vet
treatment of all objects within the collection at the same
each situation, ultimately facilitating the advancement of
relative rate. These yearly plans have defined goals around
conservation knowledge within the BCC that can then
which a budget (both monetary and personnel time) is
be shared with the greater conservation community.
built and which can be more easily supported with private
donor funding and grant initiatives. In addition, utilizing
References
an annual planning system provides greater flexibility
to pivot and re-evaluate priorities and goals, enabling Broadwater, J.D. 2012. USS Monitor: A historic ship
changes to be made when necessary. completes its final voyage. College Station, Texas: Texas
A&M University Press.
Personnel Farrell, E. 2020. A review of selected drilling methods
used to de-concrete cast iron, smoothbore, marine
As highlighted above, staffing changes have presented a
archaeological artillery. Journal of the Institute of Conser-
challenge to maintaining a consistent operation of the
vation 43(2): 142–61.
conservation effort. Due to the complexity of Monitor’s
artifact assemblage, extensive archaeological and historic Farrell, E. 2022. At the core of the problem: A bespoke
background information, and limited availability of method used to clean the bores of USS Monitor’s XI-Inch
marine archaeology-specific conservation training at Dahlgren Shell Guns. Journal of the American Institute
the graduate level, it can take several years for new staff for Conservation. doi: 10.1080/01971360.2022.2031458.
to become fully proficient. As a result, the loss of a staff Hoffman, W. and L. King. 2019. Equipment identification
member has always had an impact on project planning and the development of dry-ice blasting parameters for
and ongoing initiatives. Hired personnel dedicated to the cleaning archaeological wrought iron, copper alloys,
project at any given time has fluctuated from a low of 2 and gray cast iron. In Metal 2019: Proceedings of the
to a high of 11, directly driving the amount and type of Interim Meeting of the ICOM-CC Metals Working Group,
work accomplished from year to year. In total, 33 indi- Neuchâtel, 2–6 September 2019, eds. C. Chemello, L.
viduals have worked on the USS Monitor conservation Brambilla, and E. Joseph, 265–72. Paris: International
Council of Museums–Committee for Conservation Laurie King is an archaeological conservator within the
(ICOM-CC) and Haute Ecole Arc Conservation-restau- Batten Conservation Complex at The Mariners’ Museum
ration (HE-Arc CR). and Park, where she conserves artifacts recovered from
Hoffman, W. and R. Spohn. 2021. Hot tub time machine: the wreck of the Civil War ironclad USS Monitor. Laurie
A heated water system for artifact disassembly and treat- received her MSc in conservation practice from Cardiff
ment. In Proceedings of the Objects Specialty Group and University in Wales. Previously, she has worked in archae-
Objects and Archaeology Discussion Group Sessions 46th ological conservation at the Çatalhöyük Research Project
Annual Meeting, Houston, 31 May–2 June 2018, eds. K. in Turkey and at the Colonial Williamsburg Foundation.
Dodson and M. Wilcop, 74–93. Washington DC: Amer- Molly K. McGath, PhD, is a research scientist at The
ican Institute for Conservation. Mariners’ Museum and Park. She received her doctorate
Krop, D. and E. Nordgren. 2011. Disassembly of USS in materials science and engineering with a focus in
Monitor’s complex mechanical components. In Metal conservation science and her master’s degree in organic
2010: Proceedings of the Interim Meeting of the ICOM-CC chemistry from the University of Arizona. McGath
Metals Working Group, Charleston, 11–15 October 2010, previously conducted research at the Arizona State
eds. P. Mardikian, C. Chemello, C. Watters, and P. Hull, Museum, The Library of Congress, the United States
70–78. Charleston: Clemson University. National Park Service’s National Center for Preservation
Technology and Training, the Smithsonian Institution’s
Sangouard, E. 2019. When wood and iron are inextricably
Asian Art museums and Museum Conservation Institute,
intertwined: The treatment of a gun carriage from the
the United States National Gallery of Art, and Johns
USS Monitor. AIC News 44(3): 1–10.
Hopkins University.
Sangouard, E., E. Nordgren, R. Spohn, K. Brunke, and D.
Krop. 2015. Evaluation of sodium nitrite as a corrosion William Hoffman received his master’s degree in art
inhibitor for USS Monitor artifacts. Studies in Conserva- conservation from Queen’s University in Kingston,
tion 60(4): 253–66. Ontario, specializing in the conservation of objects. He
has worked at The Mariners’ Museum and Park since
Authors 2009 and presently holds the position of Director of
Conservation and Chief Conservator.
Erik Farrell is an objects conservator at The Mari-
ners’ Museum and Park (TMMP). He worked previ-
ously as an archaeological conservator at TMMP for
USS Monitor and at North Carolina Department of
Natural and Cultural Resources (NCDNCR) for Queen
Anne’s Revenge. He received his master’s degree in conser-
vation from Durham University.
Elsa Sangouard is an archaeological object conservator

at The Mariners’ Museum and Park and was previously
employed at the Swiss National Museum. She received
a conservation degree from Haute Ecole Arc Conser-
vation-restauration in Neuchâtel, Switzerland, and a
master’s degree in archaeology from the University of
Nantes, France. She has been assistant coordinator for
the ICOM-CC WOAM working group since 2017.
Lesley Haines is an archaeological conservator at The

Mariners’ Museum and Park. She works predominantly
with waterlogged materials recovered from USS Monitor.
She holds an MSc in conservation practice from Cardiff
University and a MA in maritime archaeology from the
University of Bristol.
Conservation and Technical Study of Ten Objects in the
Thiriyaya Collection at the Department of Archaeology in
Sri Lanka
Kasthuri Arachchilage Anusha Kasthuri* Anuradha Fernando
Department of Archaeology Department of Archaeology
Colombo, Sri Lanka Colombo, Sri Lanka
Diveesha Ruckmal Athukorala *Author for correspondence

Department of Archaeology
Colombo, Sri Lanka
[email protected]
Abstract
This paper presents a case study of ten copper alloy icons from objects. Scanning electron microscopy was used to examine
the Thiriyaya Collection of the Department of Archaeology in Sri one sample. The technical assessment of the collection resulted
Lanka. These icons include images of Buddha, bodhisattvas, and in a better understanding of the surface morphology, past burial
Tara. In 1983, the Thiriyaya Collection was discovered in a ruined environments, post-discovery damage, and the effects of a lack of
meditation hall at the Thiriyaya Girikanda temple, built during the conservation treatments. Our methodology provides a protocol
first political regime of the Anuradhapura Kingdom (5th century for studying the specific metallurgy, casting technology, and
BC–10th century AD) and thought to be the first Buddhist stupa morphology of similar copper alloy objects from archaeological
on the east coast of Sri Lanka. The technical investigation of sites throughout Sri Lanka and elsewhere.
the icons included X-radiography combined with a detailed
visual examination of the surfaces using a binocular microscope. Keywords
This approach allowed identification of the manufacturing Sri Lanka, Thiriyaya, bronze, radiography, cast, Buddha,
methods and materials as well as the current condition of the bodhisattva, Tara
Historical background
The historical framework of Sri Lanka was divided by passed away in 483 BC at the age of 80 (Geiger 2014). In
Deraniyagala (1992) into three major periods: Early Historic the 6th century BC, Thapassu and Bhalluka, considered
(500 BC–AD 300), Middle Historic (AD 300–AD 1200), to be the first disciples of the Buddhist faith, visited Sri
and Late Historic (AD 1200–AD 1815). Within this Lanka as merchants. According to legend, they worshiped
framework, the Anuradhapura Kingdom, which began in the dwelling of Bodhisattva Avalokiteshvara, located
in the 5th century BC, was the first political regime in at the Girikanda in the village of Thiriyaya. The site was
Sri Lanka and the longest stable kingdom in Sri Lankan later converted to a shrine with a hair relic of Buddha
history. However, it was the official arrival of Buddhism in placed inside (Lakdusingha 1990). The shrine consists
Ceylon (now Sri Lanka), in the 3rd century BC, that rein- of a typical classical Sri Lankan Buddhist-style circular
forced the culture, education, arts, and social status of the relic house (vatadage), image houses, and other objects
people and allowed the country to flourish. The spread of diverse type from various phases of the Anuradhapura
of Buddhism was accompanied by the construction of period (Lakdusingha 1990).
many Buddhist monasteries, including stupas, shrines,
The many cast metal figures of bodhisattvas and Buddha
meditation halls, and image houses (pilimageya), all of
images discovered throughout Buddhist sites in Sri
which became the principal types of Buddhist religious
Lanka, including the ones at the Thiriyaya site, located
architecture in Sri Lanka.
69 miles from Anuradhapura, attest to the high-level
Before the official arrival of Buddhism in Sri Lanka, production and distribution of these objects. Thiriyaya is
Lord Buddha visited the country three times before he of particular interest because of its large trove of standing
CONSERVATION AND TECHNICAL STUDY OF TEN OBJECTS IN THE THIRIYAYA COLLECTION AT THE DEPARTMENT OF ARCHAEOLOGY IN SRI LANKA 291
and seated bodhisattvas and seated Buddhas, found in The corrosion products on all of the objects were sampled
1983 under the paving stones of a ruined meditation hall and tested for chloride ions using the silver nitrate
north of the temple. protocol described by Selwyn (2018).
The Thiriyaya Collection X-radiography

Lakdusingha (1990, 347) recorded that the Thiriyaya X-radiography (200 kV for hollow-cast objects and
trove contains 31 images of Buddha, 11 of bodhisattvas, 280 kV for solid-cast objects) provided insights into the
and 3 of Tara (a consort of Bodhisattva Avalokiteshvara). manufacturing methods and materials as well as the
The collection from the Thiriyaya monastery dates current condition of the icons.
to different times during the Anuradhapura period.
Schroeder (1990) dated all of the Buddha, bodhisattva, Scanning electron microscopy
and Tara objects to the 7th–10th century and proposed Scanning electron microscopy (SEM) of Buddha figure
the simultaneous production of ascetic- and bejew- CCL/10/19 proved to be exceptionally informative as
eled-style bodhisattva figures. According to Schroeder it revealed much about the casting technology and the
(1990), bodhisattva and Tara figures can be dated to the choices of materials used in the manufacture of the statue.
7th–8th century, although some Buddha figures might
be older. Results
This research was a first step in the study of the specific Visual examination and chloride test
metallurgy, casting technology, and surface texture of the
Thiriyaya Collection. The results are relevant to related The objects of the Thiriyaya Collection are chemically
archaeological examples of copper alloys from historic and aesthetically distinct from those in other collec-
sites throughout Sri Lanka. Our methodology can serve as tions of the DOA, in terms of their surface textures and
a guide to the conservation treatment of similar objects. morphology. Unfortunately, the original appearance of
the objects at the time of their discovery is unknown.
Technical study The Thiriyaya bronzes differ in their types and degrees of
This paper presents a case study of ten copper alloy corrosion as well as in their surface morphology. Most of
icons from the Thiriyaya Collection of the Department the surfaces are covered with light-green to olive-colored
of Archaeology (DOA) in Sri Lanka. The ten icons, corrosion, indicative of bronze disease on copper-tin alloy
comprising 6 images of Buddha, 2 of bodhisattvas, and objects. In addition, a white crystalline corrosion is present
2 of Tara, have been stored together with the rest of the on some of the objects. Observations from 2012–2013
collection at the conservation laboratory in the DOA. The included “warty” surfaces, pitting, cracked crust-like
collection was excavated in 1983 and each object has since corrosion products, and, on some of the objects, compact
been stored in individual plastic bags (possibly polyvinyl patinas. A fair amount of surface details is still visible on
chloride) with a brown cardboard label enclosed inside most of the objects, despite the presence of corrosion. The
each one and an additional brown cardboard label on corrosion products vary from quite loose and powdery to
crystalline. The corrosion beneath the powdery layers is
the exterior. The bags are tied shut with cotton string.
waxy looking. The latter was not observed on the three
Buddha figures (CCL/10/19, CCL/10/43, CCL/10/34),
Methodology
which are more severely corroded.
Visual and microscopic investigation and
The Thiriyaya Collection has not yet been subjected to
chloride test
an elemental analysis to determine the metal compo-
Technical features and corrosion products were identi- sition of the objects using SEM and energy-disper-
fied without the aid of magnification. The fabrication sive X-ray (SEM-EDX) or X-ray fluorescence (XRF).
techniques, identity of the materials, and details of the According to Reedy and Harlacher (2007), bronzes of
surface decoration were determined with the aid of a the Anuradhapura period of Sri Lanka often had a high
binocular microscope. lead content, with a tin content ranging from 4 to 17%
Table 1. The investigated objects of the Thiriyaya Collection
Dimensions: lnvestigation method Condition Clues of

Casting
Object ID Date Material/1 Material/2 height, X-radiograpy previous
method CI− test SEM** Surface Structure
width (cm) kV/mAs* treatments
Seated Buddha CCL/10/19 AD 800–900 Cu alloy barely 14.0, 11.5 hollow Positive Three layers; 120/6/60 physical damage, poor cleaning and
discernable core:clay:metal light-green restoration with
central corrosion/previous modern resins
armature restoration
Seated Buddha CCL/10/21 ? Cu alloy – 16.0, 12.5 solid Positive NO 220/4/120 front thin light-green sound mechanical &
corrosion/back thick chemical cleaning
corrosion
Seated Buddha CCL/10/24 ? Cu alloy copper 9.0, 8.0 hollow Negative NO 140/6/50 mostly coherent and sound mechanical
chaplets compact cleaning
Seated Buddha CCL/10/25 AD 800–900 Cu alloy iron 12.7, 11.0 hollow Negative NO 150/6/60 mostly coherent sound mechanical
chaplets and compact, little cleaning
brownish iron
corrosion
Seated Buddha CCL/10/34 AD 650–700 Cu alloy iron 14.0, 11.0 hollow Positive NO 150/6/60 light-green warts medium mechanical &
chaplets and granular texture chemical cleaning
Seated Buddha CCL/10/43 AD 800–900 Cu alloy – 10.0, 9.5 solid Positive NO 220/4/120 bronze disease/deep medium mechanical
pitting cleaning
Standing CCL/10/20 AD 700–800 Cu alloy iron 20.5, 5.7 hollow Slightly NO 140/6/50 Patches of light-green uneven mechanical
bodhisattva complex positive powdery corrosion/ metal wall cleaning
armature green patina
Seated CCL/10/23 AD 800–900 Cu alloy – 6.0, 5.5 solid Positive NO 160/5/90 patches of light-green sound mechanical
Bodhisattva powdery corrosion/ cleaning
gray color patina
Standing Tara CCL/10/33 AD 700–800 Cu alloy – 16.0, 4.5 solid Positive NO 160/5/110 White to light-green poor mechanical &
goddess powdery/crystal/ chemical cleaning
waxy corrosion
Standing Tara CCL/10/37 AD 700–800 Cu alloy – 20.0, 5.5 solid Slightly NO 200/5/120 patches of light-green sound mechanical
positive powdery corrosion, cleaning
dark-green compact
patina
* kV (kilovoltage); mAs (milliampere-seconds)
**SEM: 15.0 kV, 14.1 × 37 mm (surface cracks); 15.0 kV, 14.9 × 30 mm (cross-section)
and a slightly lower lead content of 1%–16% (Reedy and of the figures were cast, two major types of production
Harlacher 2007). Craddock and Hook (2007) determined can be recognized: solid cast and hollow cast by direct
that 8th- to 10th-century Sri Lankan castings of Tara or indirect casting methods. The direct hollow lost-wax
(2 objects), bodhisattva (6 objects), and a number of male production of a figure usually begins with the modeling
and female figures were made of leaded bronze (Pb range of a refractory core, generally a mixture of sand, clay, and
0.03%–33.7%), with a variable tin content (1.7%–7.6%). organic material. The core is then either clad with wax
sheets or molten wax is brushed onto it. Additional wax
Casting technology is applied to create solid details, such as hair, feet, hands,
fingers, and sometimes the head or other ornaments
Of the ten statues, five were solid cast and five hollow (Schorsch et al. 2019).
cast. The solid-cast statues consist of two Buddha
figures (CCL/10/21 and CCL/10/43), one bodhisattva A requirement of the hollow lost-wax method is that the
figure (CCL/10/23), and two figures representing Tara core within the investment (also referred to as the outer
(CCL/10/33 and CCL/10/37). The hollow-cast objects mold) is supported, to assure that it does not move after the
wax is melted out. This is achieved by inserting metal or
comprise four Buddha figures (CCL/10/19, CCL/10/24,
wooden core pins through the wax layer into the core that
CCL/10/25, and CCL/10/34) and one bodhisattva figure
are then secured in the investment. Generally, organic core
(CCL/10/20).
pins (if used) are burnt out during casting, leaving holes in
While a preliminary examination of the surface textures the casting that are commonly closed post-casting, either
together with other casting evidence suggested that all by plugging or by patching. According to Schorsch et al.
(2019), in some cases the core is built around the metal The metal substance has been reduced by use, age, and
armature, which extends into the investment to support burial, as seen towards the proper-left hand shoulder,
the core during casting. The core may also be extended the proper-left leg, and underneath the statue. The intact
through the wax layer in one or more areas such that it core is black and very compact. The clay core was built
engages the investment and is thus stabilized. Filled or up layer by layer; once it had hardened in the kiln, the
unfilled undercuts, cast-in repairs, remains of sprues, and skin of the burned clay core was covered with a fine clay
core pins are expressive and naturalistic in direct or indirect layer before the wax was applied and modeled. Three
lost-wax casting (Strahan 2010). It is generally believed layers (metal, unburnt clay, and burnt clay) are visually
that historic bronze statuary was in most cases verti- evident in a cross-section of the broken fragments. The
cally cast, usually upside down. Kasthuri (2016a, 2016b) SEM image of the cross-section also indicates the pres-
suggested that horizontal casting configurations were used ence of three layers.
to produce most deities in the Buddhist pantheon in Sri
Voids in the metal and the central armature are visible
Lanka, such as the lokapala guardian deities. However,
in the frontal radiograph of CCL/10/19, though they
the Buddha and bodhisattva figures were invariably cast
are barely discernable. The core pin and cast-in repairs
in vertical positions (Kasthuri 2016a, 2016b). In vertical
are not discernable due to modern repairs and the poor
position casting, sprues, undercuts, and projections are
observed under the feet or around the head, while major condition of the object. The core may have been built
casting porosity at the site of gates indicates the orientation around the metal armature, which extended into the
of the statue during horizontal casting (Kasthuri 2016a, investment and supported the core during casting. The
2016b). During the Sri Lankan Historic periods, small statue’s deterioration can be attributed to its inherent
Buddha and bodhisattva figures were cast predominantly weaknesses as well as its previous and post-burial
in the vertical position. In this study, details of the casting environments. The modern repairs can be seen on the
methods were revealed by the presence of chaplets, internal statue’s head, hand, back, and underside. It is unknown
armatures, and core cavities as seen by X-radiography.
The surface appearance and casting techniques of each

statue are as follows:
Seated Buddha (CCL/10/19)
The 8th- to 9th-century hollow-cast seated Buddha has

an intact core but is not structurally sound, and active
bronze disease is present on the surface. The surface
morphology of the figure’s metal is unusual. Several crusts
of corrosion have cracked or flaked off. Figure 1a shows
both how the cracks have spread through the surface and
their depth (Figure 1).
Figure 1. Scanning electronic microscopy of the cracked crust (15.0 kV,

14.1 × 37 mm), cross-section (15.0 kV, 14.9 × 30 mm) of a broken piece Figure 2. The corrosion products and evidence of modern repairs.
from CCL/10/19: (a) outer metal layer, (b) middle clay layer, (c) carbon- Enhanced radiography of voids in the metal and the central armature
rich inner core layer. © DOA of the Buddha statue (CCL/10/19). © DOA
when these repairs were made, but they were most

likely done at the conservation workshop of the DOA
(Figure 2).
A 6th-century solid-cast Buddha (1982:21) housed at

the Metropolitan Museum of Art (MMA) similarly
demonstrates later historical repairs unrelated to the
original casting. The casting flaws of the MMA’s figure
have been repaired with iron fills, which may have been
introduced during the object’s use.
The front surface of the seated Buddha is covered with a

Figure 3. Buddha (CCL/10/24). Digitally enhanced radiographs; side by
thin layer of loose white to light-green corrosion product. side projecting chaplets. © DOA
The corrosion layer on the back is considerably thicker
and more compact than that on the front, and chloride Seated Buddha (CCL/10/25)
ions were detected. The light-gray metallic surface of
The surface of this Buddha was less corroded, more
the statue is characteristic of a high rather than a low
coherent, and compact. No chloride ions were detected.
tin content in the copper alloy. The Buddha, which is
The smooth silvery color of the surface of the body indi-
structurally sound, appears to have been solid-cast in
cated a high-tin copper alloy. However, unlike the body,
a single piece. The form and the surface-relief details
the forearms are a light-green color. This discontinuity
were achieved with tools used to work the wax model
suggests that the forearms were cast separately from
before casting (for example, the hair is modeled onto the
the body. However, in the radiograph these features are
wax layer). Also, pre-formed wax ears and a siraspatha
indistinguishable and appear opaque and dense. Tool
(ornament on top of the head) were adhered to the wax
marks on the body, including the hair, and iron core
body. The base of the sculpture is closed; cast-in repairs
pins nonetheless indicate the use of the hollow lost-wax
and the filed-off remains of an extension appear at the
casting method (Figure 4).
bottom of the base edge.
Buddha image (CCL/10/24)
A well-adhered and cohesive silvery color patina is

present on the majority of the surface of this seated
Buddha. The figure’s color indicates a high tin content
in the alloy. No chloride ions were detected.
In the initial study of this hollow lost-wax seated

Buddha, the right-side nipple was identified as part of
a copper core support. A radiograph taken at an angle
revealed a projecting nail from the reverse side, through
the core. The same radiograph showed one other
pin, a square-shaped plug at the chest that projected
into the center from the front. The circular primary Figure 4. Buddha (CCL/10/25). The different texture of the forearms;
cast-in fills can be seen on the reverse side of the hip iron chaplet. © DOA
and the end of the belly. Core supports made of iron
were noted, as were copper core pins. The radiographs
revealed a rectangular-shaped, core-pin hole or repair This hollow-cast seated Buddha exhibits light-green
on the left shoulder and a primary sprue on the neck, corrosion products, “warts,” and “lumps,” and a cracked
all indicating the use of the lost-wax hollow casting crust where chloride ions were detected. The crust had
technique (Figure 3). cracked along the surface, generally in a granular shape.
The deep and unfilled rectangular undercut seen on Standing bodhisattva (CCL/10/20)
the underside suggested the location of the main sprue.
This standing bodhisattva is less decorated than either
Radiographs also revealed the locations of cast-in repairs.
the other works in this study or the MMA’s bodhisattva
Deep undercut filled and unfilled core supports indicated
(1994.509). The figure does, however, show the addition
the use of the lost-wax casting process (Figure 5).
of the upavita (sacred thread), making it an ascetic
type of bodhisattva (Schroeder 1990). Patches of light-
green corrosion products cover most of the surface of
both sides of the icon. The chloride test was slightly
positive. An interesting technical finding was the use
of a complex vertical armature consisting of at least
five irregular iron strips joining and extending from
four places. However, it seems that structurally it is not
required by the statue. The vertical armature extends
from the head down through the body and terminates
above the knees. In the head, it is only vaguely seen in
the radiograph, due to the thickness of the cast wall.
Each joint in the armature has what appears to be a
knob, inserted pin, or clamp, most probably meant
to resist any slippage of the horizontal arms. The end
of the first section of the armature is connected to an
Figure 5. Buddha statue (CCL/10/34). “Warts” and cracked crust on the additional piece of metal support, probably serving as
metal; deep undercut and cast-in repairs and core pins. © DOA a kind of anchor. The end of the third section of the
armature is completed by a coiled strip and clamp. The
Seated Buddha (CCL/10/43) third and fourth sections were joined with an inserted
pin. An additional piece of metal supports the last
This solid-cast seated Buddha is structurally sound.
connection (Figure 7).
The physical damage on the left arm was presumed to
be a casting flaw. Chloride ions were detected, and the
object has suffered pitting corrosion due to active bronze
disease, which has disturbed the original appearance of
the fine patina. The figure has a closed base with casting
repairs that suggest the main sprue at the underside,
illustrating vertical casting (Figure 6).
Figure 7. Bodhisattva (CCL/10/20). Digitally enhanced radiographs; a

complex armature; cast-in repairs. © DOA
By comparison, the armature of the MMA’s bodhisattva

has both vertical and horizontal parts, with a continuous
twisted horizontal wire running through the shoulder
and the arm (Kasthuri 2016). Bodhisattva CCL/10/20
has three main core cavities, at the head, chest, and
lower body. Two rectangular chaplets are located at the
Figure 6. Buddha statue (CCL/10/43). Pitting corrosion. © DOA right thigh, and two are visible below the right shoulder.
Seated bodhisattva (CCL/10/23)
Patches of white corrosion products are present on the

surface of the bodhisattva figure, particularly near the
base of the figure and generally on the reverse side.
Chloride ions were detected.
The object is a solid cast and has fine details. The loca-
tion of one of the sprues is revealed by an area on the
underside where metal has been removed and filed,
leaving a bump on the surface. The top of the head
(jatamakuta), the ornament on the back of the head,
and the hip offered locations for a core extension during
casting (Figure 8).
Figure 9. Tara statue (CCL/10/33). The overlying corrosion products

conceal its details; casting repairs. © DOA
to the highest point of the mold in the horizontal casting

position (most likely face-up). The repair was made by
casting new metal directly over the first casting. Cast-in
repairs are on the knees, with one on the ankle. One
calf is substantially thicker than the others.
Tara figure (CCL/10/37)
The patches of white corrosion products seen on the

surface of this Tara figure tested slightly positive for
Figure 8. Bodhisattva (CCL/10/23). Possible location of core extensions
chloride ions, particularly those on the front of the figure
and on the reverse.
Tara figure (CCL/10/33)
The radiograph of the standing Tara showed opaque areas
The original surface of this Tara figure was entirely
of the solid lost-wax cast, with engraved lines in the waist.
concealed by an overlying powdery white to pale-green
The discontinuity and sharp lines on the radiograph
corrosion product, indicating active bronze disease.
indicated that a pre-formed wax head was attached to
This was confirmed by the detection of chloride ions.
the wax body before casting (Figure 10).
The detail on the surface is partially intact, including a
precious stone on the necklace (obscured by the corro-
sion products) and engraved lines on the robe and head
(representing hair).
The solid-cast Tara partially retains details from its

presumed original form. The casting flaws below the
proper-right knee, from the groin through the prop-
er-left thigh to the right thigh, were cast-in repaired.
These repairs are visible as opaque (white) patches on
the radiographs (Figure 9).
The precious stone at the neck may have been used to

hide the sprue left in the metal after casting. The area of Figure 10. Tara statue (CCL/10/37). Digitally enhanced radiographs;
the casting damage to the leg would have corresponded engraved lines in the waist; a pre-formed applique. © DOA
The uniform texture throughout the figure as seen on Buddha figure CCL/10/19 were very informative and
the radiographs shows that the metal was poured all at illustrated cracks in the corrosion layers.
once during casting. A square runner can be seen on the
Most sculptures discussed herein were made with the
right and left elbows. The garland and legs may also have
lost-wax casting method, although the technique used
served as pouring gates and runners.
in CCL/10/19 could not be determined. Several features,
such as undercut details, tool marks in the hair, chaplets,
Conservation treatment
and the use of wax appliques, indicate the use of the
The conservation treatment included a stability assess- direct lost-wax process. The statues may have been cast
ment, mechanical cleaning, and chemical stabilization. vertically (indicated by casting repairs on the underside
Excess burial soil, old restorations, and unstable corrosion or around the head), except in the case of the two Tara
products were removed mechanically. Accretions and figures, in which the wax channels and cast-in repairs are
active corrosion products were chemically stabilized. The located on the lower part of their bodies. Further evidence
manufacturing process used for each bronze determined of horizontal casting methods on the Tara figures includes
the stabilization methods. For example, the immersion the casting porosity through the garlands and legs. Arma-
method was employed for the solid-cast figures, and the tures were used in varying thicknesses, sizes, and shapes
brush application method for the hollow-cast figures. in the hollow-cast sculptures. This study also revealed
that armatures ranged from solid metal rods, strips, nails,
Surface cleaning was completed to remove dirt and
and dowels to coiled wires. Some parts of the armatures
unstable corrosion products. Stable accretions of corro-
were only faintly visible in the radiographs since iron in
sion products were left untouched, as they may hold
archaeological contexts can corrode quickly. A variation
information about the chemical nature of the alloy, the
in the materials used to hold the core in place was found
object’s production, and the interactions of the object
in this collection. The X-radiographs showed that the
with the burial environment. Cleaning was done using a
chaplets remaining in situ are made of metal and are
combination of cotton swabs and ethanol, with mechan-
circular or rectangular (CCL/10/24 and CCL/10/25), with
ical cleaning carried out using a scalpel, dental picks, and
the organic core pins burnt out during casting, leaving
a glass bristle brush under the binocular microscope.
holes in some cases. Some of these holes were closed by
Although several methods to remove chloride ions from
plugging with copper or a copper alloy (CCL/10/24) or
the bronzes were considered, the author(s) finally decided
by cast-in patching.
upon benzotriazole (BTA; 3% by weight) in ethanol.
Incralac1 lacquer in toluene was then applied as a coating The deterioration of the objects is a consequence of their
to the BTA-treated objects (Figure 10). manufacturing techniques, materials, and casting flaws,
as well as their previous and post-burial environments.
After conservation, the objects were sent to the
For example, the post-excavation storage of the objects
Anuradhapura Archaeological Museum to be placed
in plastic bags could have enhanced their corrosion and
in a display funded by the US Ambassador’s Fund for
worsened their condition. As is widely known, polyvinyl
Cultural Preservation.
chloride releases hydrochloric acid, which can promote
the deterioration of lead-containing metal and other
Discussion
alloys. Additionally, the brown cardboard labels are
Visual inspection and radiography were fundamental likely high in lignin, which could have released acids
tools to determine the conservation treatments appro- harmful to the object in the long term. The formic and
priate to each of the described objects. The technical acetic acids released by wood products may cause severe
information will be shared with archaeologists to recon- damage to lead-containing alloys and other metals,
struct the history of bronze casting in Sri Lanka. The usually resulting in the formation of a white crystalline
described icons, while heavily corroded, with areas of loss growth attributable to formaldehyde (Hatchfield and
as well as later repairs, have now been stabilized. As illus- Carpenter 1986). Traces of previous treatments and
trated in this study, the corrosion and later, inappropriate restoration, performed many years ago, were seen on
repairs often hid the full extent of the underlying damage some figures. The acidic environment of the storeroom,
and technical details. The SEM images of the hollow-cast the previous treatments, and the contaminants from
long-term burials could together have accelerated the Notes

degradation of the bronzes, resulting in the observed 1
I ncralac contains benzotriazole and therefore does not
corrosion products. However, given the lack of conser-
degrade or discolor in UV light and does not flake off
vation records, questions remain regarding the effects
(Ogburn et al. 1977, 176).
of the previously used chemicals and procedures as
well as those of treatment on corrosion development.
References
Conclusion Deraniyagala, S. 1992. The prehistory of Sri Lanka: An
ecological perspective, 2nd ed. Colombo: Archaeological
A comparative investigation of the hollow and solid sculp-
Department.
tures demonstrated that the works were cast using the
lost-wax technique, with vertical and horizontal casting Geiger, W., transl. 2014. The Mahavamsa, or the great
positions. A complex armature was found in one figure chronicle of Ceylon, English transl. Colombo: Buddhist
in this study. The objects have suffered varying degrees Cultural Centre.
and types of corrosion, including pitting corrosion and Hatchfield, P.B. and J.M. Carpenter. 1986. The problem
the formation of warty corrosion products, but more of formaldehyde in museum collections. International
compact patinas were also found. The degradation states Journal of Museum Management and Curatorship 5(2):
of the objects likely reflected not only their burial envi- 183–8.
ronments but also the post-excavation storage conditions.
Kasthuri, K.A.A. 2016a. Radiographic investigation of
The unique appearance of these sculptures in terms of
Sri Lankan bronze statues. In Metal 2016: Proceedings
their surface textures, the white to light-green corrosion
products, and the results of the chloride test all point to
Group, New Delhi, 26–30 September 2016, eds. R. Menon,
the presence of active chloride corrosion products. After
chemical stabilization with BTA, the figures acquired the C. Chemello, and A. Pandya. New Delhi: International
dark-green color typical of copper alloy metal surfaces. Council of Museums-Committee for Conservation and
An examination of the alloy composition used to manu- Indira Gandhi National Center for the Arts.
facture these objects, including the copper alloys of the Kasthuri, K.A.A. 2016b. Preliminary investigation of Sri
CCL/10/43, CCL/10/34, and CCL/10/19 Buddha statues, Lankan copper-alloy statues. Science and Technology of
would thus be of interest to clarify the pitting corrosion Archaeological Research 2(2): 159–76.
and the unusual granular cracked crust that flaked off
Lakdusingha, S. 1990. Thiriyaya Bodhisattva. Ancient
their surfaces. A metallographic analysis would also be
Ceylon.
informative regarding the statues’ thermal history and
composition. Ogburn, F., E. Passaglia, H.C. Burnett, J. Kruger, and
L.M. Picklesimer. 1977. Restoration of large gilded
Acknowledgments statues using various electrochemical and metallurgical
techniques. In Corrosion and metal artifacts – A dialogue
We are grateful to conservation staff members at the
between conservators and archaeologists and corrosion
Department of Archaeology, including Reshani Dhar-
scientists, eds. B.F. Brown, H.C. Burnett, W.T. Chase, M.
mawardhana, Mithila Diyalagoda, Saduni Koshila,
Goodway, J. Kruger, and M. Pourbaix. Washington, DC:
and Dinushi Sadamali. We are also grateful to Jerry
U.S. Department of Commerce and National Bureau of
Podany (retired Senior Conservator of Antiquities,
Standards.
J. Paul Getty Museum) for his generous support for the
revision. Director General of Archaeology Prof. Anura Reedy, C. and S. Harlacher. 2007. Elemental composition
Manathunga, Palitha Herath, Anura Jayathilaka, Sanjaya of Sri Lankan bronzes: Technological style and change.
Perera at the Atomic Energy Authority Sri Lanka, Prof. In Scientific Research on the Sculptural Arts of Asia:
Harsha Rajapakse, and Sam Groveman are gratefully Proceedings of the Third Forbes Symposium at the Freer
thanked for providing expertise and the use of the SEM Gallery of Art, eds. J.G. Douglas, P. Jett, and J. Winter.
facility at the Medgar Evers College, City University of Washington, DC: Archetype Publications with the Freer
New York. Gallery of Art.
Schorsch, D., L. Becker, and F. Carò. 2019. Enlightened

technology: Casting divinity in the Gupta age. Arts of
Asia 49(2): 131–43.
Schroeder, V.U. 1990. Buddhist sculptures of Sri Lanka.
Hong Kong: Visual Dharma Publications Ltd.
Selwyn, L. 2018. How to test for chloride ions in iron
treatment solutions using silver nitrate. Canadian Conser-
vation Institute (CCI) Notes 4/5. https://www.canada.ca/
en/conservation-institute/services/conservation-pres-
ervation-publications/canadian-conservation-insti-
tute-notes/test-chloride-ions-iron-treatment-silver-ni-
trate.html (accessed March 2022).
Strahan, D. 2010. Piece-mold casting: A Chinese tradi-
tion for fourth-and fifth-century bronze Buddha images.
Metropolitan Museum Studies in Art, Science, and Tech-
nology 1/2010: 133–55.
Authors
Kasthuri Arachchilage Anusha Kasthuri joined
the Central Cultural Fund in Sri Lanka as an artifact
conservator in 1993. Since 2005, she has worked at the
Department of Archaeology of Sri Lanka as an Archaeo-
logical Research Officer and Conservator. She is currently
pursuing a PhD in Near East Archaeology at the Freie
Universität Berlin, Germany.
Diveesha Rukmal Athukorala is an archaeological

research officer and conservator at the Department of
Archaeology in Sri Lanka since 2005. She holds a Master
of Philosophy in Archaeology from the Postgraduate
Institute of Archaeology, University of Kelaniya, Sri
Lanka. She completed her postgraduate diploma in
architectural conservation and monuments and sites
at the University of Moratuwa, Sri Lanka, in 2009. She
received her conservation training from the National
Research Institute Korea.
Anuradha Fernando has worked at the Department

of Archaeology in Sri Lanka since 2010, having been
promoted to Assistant Conservator in 2019. She received
her conservation training from the National Research
Institute of Cultural Heritage, Korea.
Conservation of Copper and Copper Alloy Objects: A

Training Programme Conducted at the Mehrangarh
Museum Trust, Jodhpur
Sahitya Raja Sunayana Rathore*
Mehrangarh Museum Trust Mehrangarh Museum Trust
Jodhpur, India Jodhpur, India
Abstract
The Mehrangarh Museum Trust (MMT) conducted a six- conservation needs and treatment requirements. The training
day in-house training programme for the conservation of programme not only benefitted the in-house conservators and
copper and copper alloy objects held at the Mehrangarh Art junior conservation professionals but also helped in forming
Conservation Centre. The instructor invited was Dr Vandana interdisciplinary interactions with the curatorial department
Singh, Managing Director of the Centre for Art Conservation and and creating awareness for the conservation and preservation
Research Experts (CARE) in New Delhi. Theoretical and practical of heirlooms belonging to private stakeholders. The training
sessions were taught on different subjects, which included programme also encouraged capacity building, innovative
the ethics and principles related to metal conservation, the thinking, creation of a strong ethical base, reflective practices
nature of cupreous objects, scientific analysis, and chemical and involvement with science which are needed to establish
and mechanical conservation treatments. The training standards of competence in the field of metal conservation in
programme also included a session on X-ray fluorescence India. This collaborative training programme was supported by
analysis performed on 46 selected museum objects to identify Tata Trusts, under its Art Conservation Initiative, which selected
their elemental composition and the use of the interpreted MMT as one of its five zonal partners.
data for the preparation of a conservation treatment proposal.
The programme’s learning outcomes are a more nuanced Keywords
understanding of the material characteristics and technology metal conservation, training programme, capacity building,
of metals, especially copper-based objects, and their X-ray fluorescence (XRF)
Introduction
The Mehrangarh Museum Trust (MMT), situated inside of cultural heritage. The museum collection nurtures the
the Mehrangarh Fort, was established in 1972 by His artistic and historic endeavours of Marwar-Jodhpur, and
Highness Maharaja Gaj Singh II, the 36th custodian of the the fort encompasses a platform for learning, research
region of Marwar-Jodhpur (the south-western region of and the revitalisation of age-old traditions.
the state of Rajasthan). The Trust is a generous patron of
the arts, music and a lively centre of academic study. The The fort and the museum
fort and the museum are a centre of excellence in the field
The fort, at over 500 yards in length, is also one of the
largest in India, with its walls towering over the city
of Jodhpur, named after Rao Jodha (r. 1438–89) who
built the fort and established the Rathore kingdom of
Marwar-Jodhpur. Mehrangarh means ‘fort of the sun’
and is a reference to the Rathore clan’s mythical descent
and lineage from the Hindu sun god Surya (Jasol et al.
2018). The fort is not only a centre of culture and history
Figure 1. Aerial view of the Mehrangarh Fort towering over the city of but also a home to the royal family, a former military base
Jodhpur and, with its many temples and shrines, a place of worship
CONSERVATION OF COPPER AND COPPER ALLOY OBJECTS: A TRAINING PROGRAMME CONDUCTED AT THE MEHRANGARH MUSEUM TRUST, JODHPUR 301
(Jasol et al. 2018). The Trust undertakes the architectural on the art and history of copper and copper alloys in
conservation of the Mehrangarh Fort through a conser- Rajasthan, and their interaction with the curators and
vation master plan. While the fort is its top priority, the private collectors will aid in the learning and review of
Trust is also actively involved in the preservation and preventive conservation techniques for metal objects.
restoration of other heritage structures such as Nagaur
Fort, Soor Sagar Palace, the Mandore Cenotaphs and the About the teaching expert
Step Wells of Jodhpur.1 The conservation and restoration
The instructor invited to conduct the training was Dr
of historical objects and museum artefacts are conducted
Vandana Singh, whose experience and expertise are
by the Mehrangarh Art Conservation Centre (MACC),
applicable to the requirements of the training programme.
established in 1996. The conservation centre has expertise
She has been the Managing Director of the Centre for Art
in dealing with a wide spectrum of materials, such as
Conservation and Research Experts (CARE) in New Delhi
paper, paintings, textiles and metal artefacts. Further-
since 2016. Her research focuses on the scientific analysis
more, the centre also undertakes projects on preventive
of metallic heritage artefacts to develop a deeper under-
conservation, display and ideal storage parameters in
standing of the history, craftsmanship, provenance, dete-
collaboration with the museum curatorial department.
rioration process and originality of objects using advanced
material characterisation techniques (Singh 2014).
In-house training programme on the conserva-
tion of cupreous objects at MMT The instructor developed the curriculum, theoretical
information, conservation treatments for the practical
In 2019, MMT was selected as one of the five zonal
sessions and the portable X-ray fluorescence analysis
partner institutes for the north-western region under the
(pXRF).
Tata Trusts Art Conservation Initiative.2 One of the objec-
tives of this initiative is to perform a field survey which
Museum objects chosen for the pXRF analysis and
involves conducting condition assessments of collections
practical treatments
at private or public museums and historic houses in India,
provide preventive conservation training and prepare The museum houses an impressive number of arte-
proposals for the conservation of these collections. facts – firearms, arms and armour dating from the
MMT houses a wide variety of metallic heritage, which 16th–20th centuries (Elgood 2017, Elgood 2021); utility
reflects the rich artistic traditions of the royal dynasties of objects dating from the 20th–21st centuries; textiles and
Rajasthan and Indian courtly life (Jasol et al. 2018). This costumes dating from the 18th–20th centuries; and deco-
heritage is also reflected in the collections of decorative rative art dating from the 17th–20th centuries (jewellery,
art and household and utilitarian objects in the historic furniture, palanquins and interior décor) belonging to
houses of Rajasthan. The survey revealed that most of the Marwari culture in Jodhpur – and showcases the
the private collectors were unaware of the value, origin, opulence of the royal family, who have created, provided
provenance and design characteristics of the metal objects and received majestic creations from all over the world
in their possession. These objects also lacked care and (Jasol et al. 2018).
maintenance, leading to deterioration and corrosion,
After the assessment of the metal collection with the help
most notably in the case of copper and copper alloys.
of the curatorial department, a total of 46 objects were
The zonal centres also conduct training programmes for chosen, which consisted of utility and decorated house-
the in-house conservators, both junior and senior, private hold objects such as storage boxes, vessels, lamps, orna-
collectors and the curators. The training programme mental plates, ornamental animal jewellery and vintage
facilitates capacity building, upskilling, formulation of toys. These objects were also composite, made of different
innovative and novel procedures for preventive conser- metals (soldering, plating and finishing) and materials –
vation, and remedial treatments. The MACC decided for example the metallic pieces on the animal jewellery
to conduct a metal conservation training programme are linked together by a textile material. The objects were
in order to learn and research copper and copper alloy selected predominantly according to visual observation,
objects. The programme also helped conservators and state of deterioration (by observing the coloured patina
conservation scientists to gain comprehensive knowledge and corrosion layers) and previous documentation. The
metal composition of at least 40 objects was uncertain, Group discussions were held throughout the programme
hence the choice of pXRF analysis for the identification at the end of every practical session. The instructor raised
of the elemental characteristics. Three utility objects questions on the treatments performed, choice of tools
(storage containers), a decorative plate and a toy car were and materials, and the outcome of the conservation
selected for the remedial conservation treatments. The treatments.
decision-making for the treatment proposals was a core
objective of the training programme. Classroom exercises
Under the guidance of the instructor, the participants
Syllabus and training schedule
performed photo-documentation and a condition assess-
The teaching and communication methods were both ment for five objects: UT/32 (storage vessel), UT/15
theoretical and practical. The instructor began with an (storage vessel), UT/254 A-B (storage container), UT/463
interactive session on the nature of cupreous objects (toy car) and UT/468 (decorative plate) (Figure 3).
which included alloy compositions, casting techniques,
chasing and finishing techniques (Figure 2). She also
highlighted the importance of understanding the envi-
ronmental influence on the cupreous objects, and infor-
mation was provided on the factors of deterioration,
such as the effect of moisture and relative humidity on
the formation of corrosion. Case studies on scientific
examination and analytical techniques such as optical
microscopy and XRF were provided. The instructor
also explained the use of scanning electron microscopy
Figure 3. The five objects selected for the practical conservation
(SEM) and X-radiography to understand corrosion treatments: UT/32 (storage vessel), UT/15 (storage vessel), UT/254 A-B
chemistry. Format samples of a condition assessment (storage container), UT/463 (toy car) and UT/468 (decorative plate)
report and a conservation treatment record specific to
metal objects were also provided. The rest of the subject For the first time in Mehrangarh, XRF analysis was
matter included: performed on the 46 objects. The participants learned
• mechanical and chemical cleaning techniques; about the working principles, importance of non-destruc-
• aesthetic and ethical considerations while performing tive analysis and the parameters required for selecting the
cleaning treatments; spots for analysis. The handheld XRF analyser3 identified
• safety protocols while using chemical reagents and the elements present and their concentrations. Using
solvents; fundamental parameter (FP) algorithms, concentrations
• conservation methods for treating bronze disease; were calculated as mass fractions normalised to 100%
• protective coatings. to avoid systematic errors in the geometric factors used
when forward calculating X-ray intensities (Elam et al.
2002, Elam et al. 2004). These FP methods are commonly
referred to as ‘standardless’, because day-to-day measure-
ment of standards is not required to measure concentra-
tions in a sample (Table 1; Figure 4). The pXRF analysis
of the five objects revealed the major and trace elements.
Analysis of object UT/463 revealed that the toy car was
made of a binary alloy of copper (Cu-Zn) or brass.
Conservation treatment proposals were drafted incor-

porating the pXRF results, and a USB microscope was
used to investigate the deterioration process.
Figure 2. The instructor, Dr Vandana Singh, presenting a theoretical These portable and easily accessible microscopes were
session used to view (at 50× and 100× magnification) the
Table 1. Results of X-ray fluorescence (XRF) measurements (in weight %) on object UT/463
ID
Sample
Acc. No. location Fe Ni Cu Zn Ag Pb Co Cd
position
file #
Roof 122 0.07 - 64.03 35.71 - 0.19 - -
UT/463 Wheel 123 0.28 0.02 65.64 33.8 0.03 0.25 0.01 0.07
Bonnet 124 0.07 0.04 62.23 37.2 - 0.33 0.01 0.12
Figure 4. Image showing the sampling spots for pXRF analysis from Figure 6. Microscopic examination of object UT/468 to view the blue-
object UT/463 coloured corrosion product
Figure 7. Microscopic examination of object UT/254 A-B to view the

green-coloured corrosion product
which they suggested could be chlorides, carbonates,

Figure 5. Microscopic examination of object UT/463 to document the acetates, sulfates, oleates or sulfides.
corrosion and patina layers
The causes and the factors of deterioration were discussed,
such as the influence of relative humidity, atmospheric
different layers present on the surface, such as dust
acidic gases, the organic constituents, mainly fatty acids
and dirt depositions, and the oxide patina. The storage
and proteins (Gebreselassie et al. 2016, Chaudhari et al.
vessels and boxes also had organic depositions, since
2018), and the reaction of moisture, which leads to
they were used for storing food. Grease accumulation
the formation of corrosion products. The instructor
and oily surfaces could be seen on object UT/15, a
explained that further analysis was required to identify
storage vessel locally known as thumbi/lota (in the
the corrosion products, for which techniques such as
Marwari language), which was used to store ghee
Raman spectroscopy and SEM-EDS could be performed.
(clarified butter) and buttermilk (Pena-Serna and
Other damages identified included abrasions, white
Restrepo-Betancur 2020). Object UT/468 showed bright
accretions and adhesive labels.
blue crystals in the microscopic images (Figure 6), and
object UT/254 A-B revealed a thick layer of sea-green After identification of the conservation problems, the
coloured crystals (Figure 7) which made the surface instructor explained that the objectives of cleaning were
look uneven and irregular, and different from the to contribute to the long-term stabilisation of the object,
smooth, brown-coloured layer which did not disfigure not destroy the naturally occurring protective patina
the objects. Participants provided their observations and remove the corrosive particles affecting the original
and deductions about the copper corrosion products, layers of the object. For example, tinning (locally called
kalai) was done on the insides of object UT/254 A-B,

including the lid, since the vessel was a roti daan, i.e. it
was used to store Indian bread. According to tradition,
kalai work was performed to prevent food poisoning from
the reaction of copper with organic acids. The bottom
rim and the spout were soldered with tin in the case of
object UT/32, showcasing the traditional metallic surface
coating and soldering techniques. Most of the animal
jewellery was coated with a layer of silver or gold, which
was confirmed by pXRF.
The conservation treatments started with dry brushing

of the object to remove the dust and dirt deposition. For
the removal of accumulated grease, oil and wax resi-
dues, cotton swabs dipped in chemical solvents, such as
acetone, white spirit or ethanol, were used in the presence
of a fume hood. Mechanical cleaning using a micromotor
equipped with brass bristle brushes and abrasives, such as
fine calcium carbonate and pumice powder, was tested.
A slurry of the abrasives was prepared using deionised
water (Ases Chemical Works, Jodhpur; Figure 8). The
slurry of pumice powder was applied using a cotton swab
on objects UT/254 A-B, UT/463, UT/32 and UT/15 and
was effective in the removal of the blue-green corrosion
products. However, caution was advised while using the
Figure 9. After the application of microcrystalline wax as a protective
abrasives. Gentle movements are required, since they can coating, the instructor buffs the wax using a muslin cloth to achieve a
act as stripping agents. uniform coating
and UT/32, after which a coating of Renaissance wax

was buffed using muslin cloth (Figure 9).
The training programme ended with the evaluation of

the conservation treatments applied on the five selected
objects. Group and individual discussions were held
on the topics of gel-based cleaning treatments, such as
agar and gellan gums (Giraud et al. 2021), and the use
of Incralac and its durability for the conservation of
outdoor brass objects.
Figure 8. The conservator performing mechanical cleaning of object Learning outcomes of the programme
UT/463 with a bristle brass brush attached to a micromotor tool
The intensive six-day training course provided an excel-
For the protective coating, microcrystalline wax (Renais- lent opportunity for the young and senior conservators,
sance wax) and Paraloid B-72 (CTS Europe) were the curators, stakeholders from private sectors and volunteers
preferred choices, as these create a barrier between the to gain comprehensive knowledge and hands-on expe-
object and any environmental impacts and slow down rience in the conservation and scientific examination of
the ongoing deterioration of the metal substrate. Paraloid copper-based artefacts. The conservation team created
B-72 was prepared by dissolving 2% w/v in acetone and a condition assessment reports and specific conservation
single layer was applied on objects UT/468, UT/254 A-B treatment proposals for cupreous objects. The reports
included the curatorial aspects, such as the artistic and The in-house conservators performed the conservation
historical details, and the scientific details focusing on treatments on the remaining cupreous objects in accord-
possible chemical testing and analysis required to identify ance with the techniques learned on the programme. A
the corrosion products. joint proposal was presented with the curatorial depart-
ment to incorporate passive air samplers for pollutant
The use of XRF led to a eureka moment. The gold-
monitoring and for understanding the most significant
en-hued elephant bell, which was listed as a brass
pollutants responsible for copper corrosion in order to
ornament, was actually made of silver gilt (Figure
improve storage and environmental control for cupreous
10). Similarly, the animal ornamentation (neckpieces;
objects. The success of the training programme opened
Figure 11) and a decorated bedstand were incorrectly
a pathway to plan and execute a series of object-specific
listed as brass objects, whereas, vice versa, objects which
and metal-specific conservation training programmes
were thought to be made of silver, for example the deco-
for artefacts, such as swords and firearms, silver objects,
rated Paan daan or betel nut box, were actually made of
armour, and the processes of sword-making and hilt
tin-coated copper and zinc. The museum has prepared
designing.
documentation reports for the 46 objects under the
instructor’s guidance and their material characteristics
Conclusion
are now fully known.
Strong ethics, reflection, science, preventive measures and
advocacy are all part of the modern conservation remit.
The power of collaboration is astounding and can create
partnerships that establish standards of competence in
the field of art conservation.
Acknowledgements
The authors would like to express their appreciation to
His Highness Maharaja Gaj Singh II, managing trustee
of the Mehrangarh Museum Trust, and Baijilal Sahiba
Shivranjani Rajye, CEO of the Mehrangarh Museum
Trust, for their full support throughout the project.
We are forever grateful to our previous director, Karni
Singh Jasol, for his encouragement, and we would like
Figure 10.
The in-house to thank the instructor, Dr Vandana Singh, for her
conservator guidance and in-depth knowledge. Our thanks also go
performing a to the conservation team at the Mehrangarh Art Conser-
pXRF analysis
vation Centre – Mr Sunil Laghate, head of department,
on the
elephant bell Mr Vikram Rathore, the manager, the conservators Mr

Surender Singh, Mr Achal Shekhawat, Mr Ajay Vyas,
Figure 11. Course participants

identifying a sampling spot
on the elephant necklace for
XRF analysis and a technician
performing the analysis on
the verso of the medallion. An
average of four sampling spots
were analysed
Mr Harender Singh, Ms Nisha Poonia and Ms Komal Elgood, R. 2021. The Maharaja of Jodhpur’s guns. New
Joshi, the mount section supervisor, Mr Zakir Belim, Delhi: Niyogi Books.
the senior computer operator, Mr Dilip Dave, and Gebreselassie, N., R.K. Abrahamsen, F. Beyene, F. Abay,
the senior attendant Mr Dhirendra Singh – for their and J.A. Narvhus. 2016. Chemical composition of natu-
enthusiastic participation and support. We extend our rally fermented buttermilk. International Journal of Dairy
thanks to the Tata Trusts Art Conservation Initiative Technology 69(2): 200–208. https://doi.org/10.1111/1471-
team for their insights. 0307.12236
Giraud, T., A. Gomez, S. Lemoine, C. Pelé-Meziani, A.
Notes
Raimon, and E. Guilminot. 2021. Use of gels for the
1
C onservation of heritage structures performed by cleaning of archaeological metals. Case study of silver-
MMT: www.mehrangarh.org/research-conservation/ plated copper alloy coins. Journal of Cultural Heritage
architectural-conservation-projects/ 52: 73–83. https://doi.org/10.1016/j.culher.2021.08.014
2
he Tata Trusts is a charitable organisation duly regis-
T Jasol, K.S., P.A. Andrews, R. Elgood, C. Glynn, and A.D.
tered under Maharashtra Public Trusts Act, 1950. The Jhala. 2018. Peacock in the desert: The royal arts of Jodhpur,
Tata Trusts Art Conservation Initiative proposes to India. New Haven: Yale University Press.
establish, develop and strengthen art conservation
Pena-Serna, C. and L.F. Restrepo-Betancur. 2020. Chem-
centres in five zonal locations across India and create
ical, physicochemical, microbiological and sensory char-
a cohort of trained conservators. www.tatatrusts.org/
acterization of cow and buffalo ghee. Food Science and
our-work/arts-and-culture/conservation/art-conser-
Technology 40(suppl. 2). https://doi.org/10.1590/fst.32219
vation-initiative
Singh, V. 2014. Linking heritage conservation with
3
he XRF analysis was performed using a handheld
T
cultural revitalisation. In Building Strong Culture through
portable XRF spectrometer belonging to the company
Conservation: ICOM-CC 17th Triennial Conference
Elvatech LTD, Ukraine. The model was a Prospector 2
Preprints, Melbourne, 15–19 September 2014, ed. J.
and the elemental concentrations were provided by the
Bridgland, art. 0906, 6 pp. Paris: International Council
analysis algorithms based on ‘standardless’ fundamental
of Museums. Available at www.icom-cc-publications-on-
parameters and empirical (regression) calibration. The
line.org/
software used was Windows-based and compatible with
ElvaX™ PC software. The X-ray was used for a maximum
Authors
of 15 s at a voltage of 40 kV, and an SDD X-ray detector
with a maximum energy resolution of < 165 eV. Sahitya Raja is a conservation scientist currently
working as a conservator-restorer at the Mehrangarh
References Museum Trust in Jodhpur. She has an MSc in conser-
vation and restoration of cultural heritage from the
Chaudhari, N., S. Balakrishnan, and A.M. Patel. 2018.
University of Bologna and completed her experimental
Buttermilk: An unrevealed nutraceutical. Interna-
thesis at the University of the Basque Country, which
tional Journal of Fermented Foods 7(2): 113–17. doi:
focused on the use of multi-analytical instrumentation
10.30954/2321-712X.12.2018.6.
for the conservation and restoration of the Punta Begoña
Elam, W.T., B.D. Ravel, and J.R. Sieber. 2002. A new Galerias, a historic building located in Getxo in Spain.
atomic database for X-ray spectroscopic calculations. Her ongoing projects include the spectroscopic analysis
Radiation Physics and Chemistry 63(2): 121–8. of the wall paintings at Mehrangarh Fort and the conser-
Elam, W.T., R. Shen, B. Scruggs, and J. Nicolosi. 2004. vation of the metallic heritage of Rajasthan.
Accuracy of standardless FP analysis of bulk and thin
Sunayana Rathore, PhD, is currently deputy curator
film samples using a new atomic database. Advances in of the Mehrangarh Museum Trust in Jodhpur. She has
X-Ray Analysis 47. worked in the areas of creative exhibition planning,
Elgood, R. 2017. Rajput arms and armour: The Rathores design and display and preventive care and storage of
and their armoury at Jodhpur Fort, vol. I. New Delhi: museum objects. She has a master’s degree in heritage
Niyogi Books. management with a specialisation in conservation from
Sophia College. She qualified the National Eligibility Test

(NET) by the University Grants Commission (UGC)
for a lecturer’s position in museology and conservation
in 2004 and was awarded a Junior and Senior Research
fellowship by UGC for her PhD in museology from the
National Museum Institute. A native of Rajasthan, she
is interested in preserving its tangible and intangible
heritage.
Keeping it Kosher: The Care and Treatment of the Judaica
Collection at the Israel Museum, Jerusalem
Jessica Lewinsky* Hadas Seri Maya Delano Irit Lev Beyth

The Israel Museum, Jerusalem The Israel Museum, Jerusalem The Israel Museum, Jerusalem The Israel Museum, Jerusalem
Jerusalem, Israel Jerusalem, Israel Jerusalem, Israel Jerusalem, Israel
www.imj.org.il/en www.imj.org.il/en www.imj.org.il/en www.imj.org.il/en
Abstract
The Metals and Organic Materials Conservation Laboratory their symbolism and reinforces the codes that convey a deeper
serves all departments at the Israel Museum in Jerusalem. As understanding of the Jewish culture. (iii) The historical repairs,
Israel’s foremost cultural institution, the museum stewards a marks of usage, inherent flaws, wear, and secondary use of
remarkable encyclopedic collection. This paper outlines the these unique objects explain their story and value. (iv) Added
bespoke criteria used for the treatment and study of metal materials and marks of usage that properly convey the values,
objects in the Jewish Art and Life wing (JAAL). These criteria uses, narratives, and timelines associated with the objects must
guide the conservators’ decisions for each task or challenge, be dealt with, respected, and preserved. (v) Many historical
and differ from those used for other collections, such as pieces sensitivities converge in one object, and it is important to
from the contemporary art collection and archaeological validate these different narratives and explore them further. This
metals. They take into account the scope of the collection, the paper is an invitation to continually reexamine conservation
needs of the objects, and the needs of the visitors. Through practices and raise awareness of how to deal with specific
them, conservators can also become aware of how their own collections. Having clear principles, guidelines, and criteria is a
identities and trajectories can affect how and what objects are way to ensure that the conservators’ perspective enriches the
treated. Five different customized criteria are identified that relationship with material culture.
directly affect the conservation care and treatment of JAAL’s
living collection: (i) Life in a Jewish country implies that every Keywords
year object treatment, display, and research is influenced by the conservation, Judaica, historiographic metal collections,
Jewish calendar. (ii) The aesthetics of the objects contributes to museum preservation
Introduction
The Israel Museum in Jerusalem (IMJ), established in
1965, is Israel’s foremost cultural institution respon-
sible for stewarding a unique encyclopedic collection.
Its comprehensive holdings cover thousands of years,
regions, themes, and materials. The collection is taken
care of, studied, and exhibited in four wings: Archaeology,
Fine Arts, Youth Art Education, and Jewish Art and Life.
This paper addresses the conservation criteria tailored to

suit the treatment and study of the Jewish Art and Life
wing. It focuses on the extensive metals collection and
how the conservation criteria differ from those used on
objects in the other museum wings.
Figure 1. A general overview of the Hanukkah lamp wall at the JAAL
The “Jack, Joseph, and Morton Mandel Wing for Jewish galleries, The Israel Museum, Jerusalem. Photo by Elie Posner. © The
Art and Life” (JAAL) (Figure 1) includes over 30,000 items Israel Museum, Jerusalem
KEEPING IT KOSHER: THE CARE AND TREATMENT OF THE JUDAICA COLLECTION AT THE ISRAEL MUSEUM, JERUSALEM 309
in storage, archive, display, and on-site in four original greatly influenced the creation of the JAAL’s tailor-made
reconstructed synagogue interiors. The collection and its conservation criteria.
comprehensive display explore and reflect the depth and
beauty of Jewish material culture and creativity. It illustrates Bespoke criteria
the aesthetic and stylistic influences of other cultures in
The bespoke criteria stem from the team’s professional
communities and places where Jews once lived as well as
conservation principles. Over the years, it was noticed
present, living communities. The objects are imbued with
that the approach would differ if the object to be treated
our values, points of view, a religious spirit, zeitgeist, and
was archaeological, an artwork, or an object from the
multiple memories. They showcase a vivid cultural tapestry
JAAL collection, especially since there is an effort to
weaving together the individual and the communal, the
showcase the values and sentiments the objects elicit as
sacred and the secular, the awe-inducing and the mundane,
fully as possible.
and the heritage of the past and creative innovations of the
present. The display enlightens visitors about the Jewish The criteria used are collection-, object-, and people-
culture, but also provides tools and introspective moments driven.
for any visitor to address their own identity.
In the first case (collection-driven), the wide variety and
The Metals and Organic Materials Conservation Labo- unique study opportunities this collection provides are
ratory studies and cares for the objects from the JAAL taken into account and this serves as a form of measure-
collection. This work is done in the knowledge that the ment and comparison. The criteria consider the wide
conservation treatment and understanding of an object scope of materials and periods, and attempt to unify the
can bring forth and clarify a lot of information as well as objects from the same perspective during conservation.
reassess the values associated with the collection. The sheer volume of objects at the JAAL wing facilitates
Currently, the team is composed of four Jewish Israeli the identification of patterns and comparisons in material
female conservators. Each comes from a different Jewish culture. The types of materials, periods, shapes, and main
background, community, and level of religious practice. and secondary uses are taken into account to compre-
Being both immersed in the culture as well as treating hend when an object is unique, out of the ordinary, or
the objects has enriched the conservators’ practice; there reveals a clear distinction between the awe-inspiring
is an awareness of the pros and cons. On the one hand, and the mundane.
there is an innate understanding of the objects and their In the second case (object-driven), decisions are based on
uses/traditions. On the other, the conservators can be the particular needs of the object and take into account
captives of their own perceptions, unable to look at the historical background, context, aesthetic tastes, and
objects or situations inertly. Rather than treating ethno- materials. The object is treated rather than just the mate-
graphic objects from other cultures based on reading rial. The conservators’ point of view can enrich many
and hearing about them from an expert, for example, of the other tasks carried out in the museum, such as
the conservators actually experience them in one way research, display, and education. The approach is inclu-
or another. Talking to experts involves a double-sided
sive and considers many of the cultural and historical
exchange in which first-hand accounts can be recounted,
values and external factors that affected the item and
shaping the perception and treatment of the object. This
prompted its re-appreciation, display, and narrative
open conversation is welcome where it serves to pursue
within the museum.
as many of the museum’s aims as possible.
In the third case (people-driven), the end goal of the
The discussions amongst the conservators and curators
conservation process is to share and display the object
are in-depth, and it is imperative that the decision-making
with visitors and communities.
process and all its considerations are explained. There
is an awareness of bias and “professional deformations.” These criteria have facilitated the decision on which
It is always personal in some respects. Individual expe- treatments to apply, bringing awareness to many of the
riences and familiarity with Judaica objects, combined underlying assumptions the conservators face. It is impor-
with knowledge of past and present methods, philoso- tant to note that many of the objects in the collection
phies, and a dynamic ever-developing profession, has would be used in religious ceremonies in their original
context in synagogues and Jewish homes. However, in the In general, polishing metal objects is performed only
museum, they are detached from their ritual functions. when absolutely necessary, as the conservators are deeply
The question of “holiness” is a complex one and not the aware that this action changes the object’s surface and
topic of this paper. The museum objects are awe-inspiring patina and also removes original material. Conser-
but not sanctified or holy, in the sense that Jewish law vation training and knowledge does not stipulate the
(Halakha) limits their functions and imposes restrictions requirement to exhibit these objects in line with their
as to how they can be displayed. The gender, religious true aesthetics. That is why, more often than not, unlike
affiliation, or belief of the conservator do not affect the metal objects from other collections at the IMJ wings,
museum’s practices. The primary exception in Jewish Judaica is regularly polished.
law regards the written word. However, even there, the
To prolong the life of objects while maintaining their
objects exhibited are those that are no longer deemed to
pristine appearance, routine procedures have been set up
be kosher for use as prayer.
to delay patina formation and extend periods between
polishing:
Living within the Jewish context and calendar • Periodic surveys to document changes in object
As a Jewish state, Israel adheres to the Jewish calendar condition with follow-up measures for unstable items,
along with the Gregorian one. This fact is essential to treating only what is necessary.
our working rhythm in the laboratory, affecting many • Application of microcrystalline wax coatings after
of the treatment decisions, priorities, and timeframes. treatment to postpone re-tarnishing.
The regular use of objects in special displays relating • Storage of polished objects in transparent airtight,
to Jewish holidays reinserts the objects into circulation anti-tarnish polypropylene bags embedded with
every year and makes the collection a truly living one. The copper nanoparticles (Figure 2). The bags are marked
treatment and display contribute to the reinterpretation with the date they were heat-sealed for monitoring.
of meaning, traditions, and questions regarding Jewish • Monitoring of the environmental conditions (relative
identity. There is an awareness of how being immersed humidity, temperature, and air quality) by the preven-
in the culture changes understanding of these ritual and tive conservator. By reviewing this data, very stable
traditional objects. environments have been created that help care for the
collections throughout the museum.
Before the celebrations for Hanukkah, for example, it has
become customary to treat large numbers of Hanukkah Other aesthetic considerations go beyond how polished
lamps. These lamps are used in many exhibits in and or shiny a metal object looks. When possible and conven-
outside the museum and commemorate, celebrate, ques- ient, objects that can be dismantled are taken apart to
tion, interpret, and internalize the values of this holiday. allow access for cleaning, ensuring mechanisms work
One such lamp is the one lit every year by the president
of Israel. For this lamp, separation babettes (sconces)
were built that allow the lamp to be used as intended
and not just displayed. Furthermore, attention was given
to its function as a symbol of state and diplomacy. The
object must look pristine, stable, and regal for any photo
request with emissaries and prime ministers.
Aesthetics
Most Judaica metal objects are highly polished for use in
Jewish life and tradition, and are also kept in the same
condition in daily life in Jewish homes. A well-polished
set of Shabbat candlesticks, for example, is also a symbol Figure 2. Judaica objects on storage shelves, some sealed in
of status, respect, Jewish values, and even good house- polypropylene anti-tarnish bags, Jerusalem, 2021. Photo by Hadas
keeping. Seri. © The Israel Museum, Jerusalem
and preserving their kinetic properties and alignment.

This has led sometimes to the storage of the polluting
elements separately to delay the corrosion process or
chemical reactions. An added advantage of doing so is that
hidden hallmarks, original assembly instruction marks,
and other details and surprises have often been found that
have contributed to knowledge about the objects and their
correct display based on taste at the time of their creation.
For example, an alteration to a late-19th-century Russian
tin Hanukkah lamp defaced by adherents of iconoclastic
ideologies was only noticed while being examined in the
conservation laboratory (Figure 3).
Figure 3. Defaced late-19th-century Russian cast lead and tin Hanukkah

lamp, accession number B08.0790. Photo by Elie Posner. © The Israel
Museum, Jerusalem
Historical repairs, marks of usage, wear, and

secondary use
Many Judaica objects in the IMJ’s collection were in heavy
use in their respective communities for generations. As
such, it is common for the objects to have undergone
different kinds of repairs, had new elements added, and
become strange hybrids. Many of these additions and
omissions are part of the story and values of the object.
They might indicate how the object was created, used,
repaired, or adapted during its lifetime.
Historical repairs are not altered, since they denote the

care and relationship that existed with their users. That
is to say great efforts were taken to salvage and repair
Figure 4. Torah shield with changeable plaques by Cornelius Poppe,

Augsburg, Germany, 1695–1700, accession number B50.02.0227.
Silver, repoussé, engraved, and partly gilt. Received through the Jewish
Restitution Successor Organization (JRSO). (a) Front, (b) back and detail
of back. Photo by Hadas Seri. © Israel Museum, Jerusalem
rather than discard. The museum, therefore, only treats There are many challenges associated with objects given
the structural instabilities with reversible procedures. a secondary use. Beyond recycling and reuse consider-
Missing elements are usually replaced using different ations, the mindset associated with their creation needs
materials to give the impression of the object’s full shape to be understood. A case in point are the flower vases
or by displaying a drawing or photograph of what the created by artisans at Bezalel, the first school of arts and
object would have looked like originally. For example, crafts founded in Jerusalem in 1905. Student artisans
Torah shields (Tas Torah) (Figure 4a, 4b) often have pressed, engraved, and stamped brass bomb shells and
different removable plaques to indicate holy days in the embellished them with silver and copper damascene
Jewish year. Years of placing and removing these plaques (Figure 6). The transformation of the ammunition shells
during the lifetime of the object in the community creates suggests hope, as metal was a scarce commodity and
deep scratches and dents. These marks are left to indicate material. Time and money were invested in this trans-
the high usage of the object as well as give an idea of the formation as a call for peace after the military conflicts.
hidden mechanism. By leaving these marks, the curator By leaving the shell shape recognizable in the flower vase,
can better tell the story of these Jewish communities. the artisans were making a statement about laying down
Another example is a pair of Torah finials (rimonim). weapons and ensuring the people and the land had an
During the Simchat Torah celebration, the Torah is opportunity to flourish.
removed from the Ark in the synagogue and marched
ceremoniously along the streets of the town. This is an
extremely joyous occasion. The Torah passes from man
to man during dancing in a usually very crowded space.
Often, metal adornments on the finials (Figure 5) fall
off, break, get deformed, etc. This damage is frequent in
our finial collection and more often than not the dents,
losses, and deformations are not repaired.
Figure 6. Artillery shell

adorned with images
of holy sites in Israel by
Avraham Asher Baradon,
Bezalel School of Arts
and Crafts, Jerusalem,
1920, accession number
B11.1189. Brass, pressed,
engraved, and stamped
with silver and copper
damascene work.
Photo by Elie Posner.
© The Israel Museum,
Jerusalem
Composite and mixed-media objects by use

Historical objects consist of a wide range of organic and
inorganic materials. Often, they are fragile because of this
combination, and at times show a more advanced state
Figure 5. Torah finials surmounted by double-headed eagles by of degradation. Moreover, many of their materials were
Fredericus Becker senior, Pressburg, Austria-Hungary, 1813, accession not used during the creation of the object but instead
number B50.02.0706 (a–b). Silver, repoussé, pierced, chased, and
incorporated over time by its use within the community,
machine turned. Received through JRSO. The double-headed eagle on
one finial and most of the bells are missing. Photo by Yair Hovav. © The converting them into mixed media. These materials have
Israel Museum, Jerusalem now become an integral part of the object and therefore
cannot be dissociated from it, even when this combina- Historical sensitivities
tion of materials sometimes creates a new problem for
As mentioned, the IMJ in its encyclopedic universal
the conservation of the object.
vision explores and celebrates the experiences, accom-
Examples of composite or historical mixed-media objects plishments, richness, and spirit of Jewish identity.
include books and parchment scrolls with metal cases, Pogroms, persecution, ostracism, genocide, and the
mezuzot, amulets, dresses with charms, and metallic Holocaust are only one aspect of Jewish history and
ornaments. Considerations have to be given to how identity. Objects that have survived traumatic historical
materials included in mixed-media objects are handled, events have a much more complicated relationship with
such as feathers in kits to search for leavened bread the both the people who experienced the events and the
night before Passover or candlesticks with wax or oil visitors who experience the displayed objects in the
remnants, information currently being gathered for museum. The objects have narrative and symbolic qual-
research at the JAAL wing. ities that facilitate ethical and critical relations between
Since each material has specific requirements, it is a museumgoers and traumatic histories. The museum’s
preventive conservation challenge to maintain stable job is to make certain that any intervention prioritizes
storage and display conditions. Decision-making is their stabilization but does not erase or diminish the
supported by prioritizing the likelihood of risk to any marks conveying their story.
one of the elements. That is to say, a risk-based approach It is essential to remark that treating these objects requires
informs the treatment. For example, in the case of a conservators to be mentally prepared and to have a
Havdalah set (Figure 7)––which includes two handheld certain mindset. Personal family histories, experiences,
spice containers stored in a silver ornamented, textile-lined and sensitivities affect not only the treatment of the object
box, probably a secondary use of the box––, many acetate but also the trajectory of our staff. Some staff members
corrosion products staining the lining were found during find it harrowing and draining to work on objects that
treatment and considered to be the result of the odorous stem from trauma, while others perceive it as a fulfilling
cloves kept inside the spice containers. The cloves were honor. It is an opportunity to remark on our collective
removed from the containers and placed inside poly- history and to address injustices in society.
ethylene zip-top bags and returned to the interior of the
containers. The complexity of these objects makes each The “Kiddush cup from Munich” (Figure 8) in Germany
treatment a mental exercise that takes into account many is a testimony to the vandalism and hatred that took place
aspects that are both material and intangible. across Germany during the Kristallnacht pogrom. This
object, as well as two others displayed in the same vitrine,
was left disfigured, discolored, warped, broken, and
melted. From a conservation point of view, they were left
mostly untouched. Their display acknowledges hatred,
past and present, and invites meaningful discussion.
Marks that are often documented and examined in the

conservation lab are those on objects brought to Israel
by Jewish refugees. These objects were smuggled out
of places ridden with discrimination and antisemitism
that ultimately eradicated peoples’ communities in the
Diaspora. The damages we see on these objects represent
how emotional and difficult forced immigration can
be. When these objects are brought to the workbench it
can be heartbreaking, because they reveal the struggles
Figure 7. Two Havdalah spice boxes, unknown provenance, 1877, and sacrifices made not only to escape but also to bring
accession number B97.0445 (a–b). One is dismantled for cleaning with
the cloves contained inside in the center. The silver container in the lower
these remnants of a life left behind. Deciding to bring a
part of the photo probably had another purpose originally. Photo Jessica family heirloom, rather than another piece of bread, is
Lewinsky. © The Israel Museum, Jerusalem a difficult choice to make.
Conclusion
The Metals and Organic Materials Conservation Labo-
ratory provides its services to all departments of the
Israel Museum in Jerusalem regardless of the cultures
or periods they study. The care and treatment of objects
contribute to the permanence, dissemination, display,
study, and enjoyment of the collections. The proliferation
of metal objects in the collections reflects important
aspects of the world of material culture.
This paper outlines the bespoke criteria created for the

treatment and study of metal objects in the Jewish Art
and Life wing. The criteria used help to decide how to
approach each task or challenge. These guidelines are
collection-driven, object-driven, and people-driven.
Collection-driven because it is a unified collection.
Viewing the wide variety of objects together is a unique
study opportunity that serves as a reference and compar-
ison. Object-driven because the item is treated rather
than the material. The guidelines are based on what the
objects convey and mean and their inherent character.
Figure 8. Kiddush cup that survived the Kristallnacht pogrom in 1938, Lastly, people-driven because the conservation process
Munich, Germany, 20th century, accession number B50.04.2556, is inscribed in Jewish society. Although neutrality
133/040. Silver, repoussé. Photo by Laura Lachman on deposit from Yad
is the goal, the conservator’s point of view can both
Vashem (The Holocaust Martyrs and Heroes Remembrance Authority).
© The Israel Museum, Jerusalem enrich the understanding of the object and be limited
by personal bias.
The museum is a caretaker, memorial, and repository Five different customized criteria were identified that
for many “orphaned” objects. Many of the objects with directly affect the conservation care and treatment of
gaps in their provenance, origin, and ownership are this living collection. Life in a Jewish country affects
connected to the Holocaust. During World War II, the timeframe of when objects are treated, based on the
Jews were subjected to mass extermination and their Jewish calendar used by the state. As a result, objects are
personal objects, sacred objects, books, and art collections put back into circulation every year. Via their treatment,
were confiscated, stolen, re-appropriated, hidden, and they are appreciated, reinterpreted, and experienced by
discarded. After the war, the US army created special visitors.
units in charge of discovering, collecting, cataloging,
The objects in the museum have certain aesthetics, and
and returning these objects to individuals, communities,
their appearance is one part of their symbolism and the
and Jewish institutions. Only a minority of objects were
codes that convey a wider understanding of the past.
identified and restored to their rightful owners, and most
By properly displaying these objects, the collection is
were given to museums around the world, including
represented and shared in ideal condition, regardless of
the IMJ, for safekeeping. The IMJ takes an active part
whether it is exhibited in the galleries, catalogs, or on
in the restitution process worldwide. Every once in a
social media.
while the museum has the honor to repatriate objects to
their rightful heirs. In 2016, a modest pewter jug with The material complexity of some objects prompts outside-
personal initials was returned to its owners following the-box thinking to create custom-made solutions that
lengthy research and due diligence. The conservators are guarantee the dissemination of the values, ideas, and
closely involved in identifying materials, stamps, taking materials to future generations. At times, added materials,
detailed condition reports, and handling instructions that marks of usage, and compositions tell a new story about
help return these objects safely to their rightful owners. an object’s use, history, and associated values.
Recounting history is a responsibility the museum takes Land of Israel. The concept, whose coining occurred in
seriously. It is imperative to be aware that many stories similar terms to “Americana,” initially referred to the
and historical sensitivities converge in one object, and collected materials produced at the Bezalel Academy
it is important to validate these different narratives. An of Art. However, today, it includes the paraphernalia
inviting environment must be created where topics can produced since the 1920s.
be discussed openly.
Judaica: Historical, literary, ritual, or artistic objects
The different experiences, awareness, and knowledge relating to Jewish life and the customs of the Jewish faith.
gained over the years by the members of the conservation
Kosher: A word in Hebrew that literally means “fit” and
laboratory helped to form these bespoke criteria. In a
refers to the preparation of food and utensils following the
practical sense, it has also led to routine conservation
requirements of Jewish law. The term is nowadays used
treatments for the Judaica collection. These include more widely as a synonym of “above-board,” acceptable,
sealing silver objects in anti-tarnish bags, sometimes genuine, or legitimate.
adding a protective coat to objects on loan, and making
sure an object is displayed in a polished manner, since Mezuzah (plural mezuzot): A piece of parchment
this is how it would have been presented in its original inscribed with religious texts and affixed in a case to the
context. Likewise, it also entails leaving damages on the doorpost of Jewish homes. Written on the parchment is
object when justified, since their presence enhances the the Jewish prayer Shema Yisrael.
historic and symbolic value of the object. Rimonim (Torah finials): Pair of ornaments inserted at
One of the most valuable tools a museum has is the the top ends of the rollers of a Torah scroll. Very often
dialogue between curators and colleagues. Every day, they are made of precious metals and adorned with bells.
it is apparent how our own identities and trajectories Shabbat: The Jewish day of rest. It begins at sundown
affect which objects are treated and how. The aim is not on Friday and ends at sundown on Saturday when a new
to become creatures of habit but to question constantly week begins. Shabbat is a day of abstention and prayer
the practices used. Every conservation treatment involves for observant Jews.
much consideration and dialogue. The focus is placed on
Simchat Torah: A Jewish holiday that celebrates the
important questions such as the best possible treatment,
conclusion of the annual cycle of public Torah readings
the methods inherited from previous conservators at the
and marks the beginning of a new one.
IMJ, established traditions at the lab, and whether our
identity as Jews, Israelis, and conservators changes with Tas Torah (Torah shield): An ornament, usually
every object we treat. composed of precious metal suspended by chains on
the front of the Torah case.
Glossary
Torah: The compilation of the first five books of the
Babette (sconce): Part of a candlestick where the candle Hebrew Bible: Genesis, Exodus, Leviticus, Numbers,
is attached. and Deuteronomy.
Hanukkah: The Festival of Lights, a Jewish celebration Yom Hashoah: “Holocaust and Heroism Remembrance
commemorating the re-dedication of the Second Temple Day” known colloquially as Yom Hashoah is observed
in Jerusalem in the 2nd century BCE. as Israel’s day of commemoration for the approximately
six million Jews murdered by the Nazi regime during
Havdalah: A Jewish religious ceremony that marks the
World War II. The day focuses on learning about Jewish
symbolic end of Shabbat and ushers in the new week. The
heroism in the face of inhumanity as well as exploring
ritual involves lighting a special candle with several wicks,
the roots of racism, genocides, and antisemitism. Unlike
blessing a cup of wine, and smelling sweet spices. Shabbat
the International Holocaust Day commemorated on
concludes with a verbal announcement that separates it
January 27, Yom Hashoah is held on the 27th day of Nisan
from the mundane weekday that follows.
(which falls in April or May), unless the 27th would be
Israeliana: A term that refers in a broad sense to the set adjacent to the Jewish Sabbath, in which case the date
of things that are specifically identified with Israel or the is shifted by a day.
Acknowledgments
We would like to thank the staff at the JAAL wing: Chief
Curator Dr. Rachel Sarfati, Miki Joelson, Gioia Perugia,
and Sharon Weisser.
Authors
Jessica Lewinsky is a preventive conservation officer
and object conservator at the Metals and Organic Mate-
rials Conservation Laboratory at the Israel Museum,
Jerusalem. She has a BSc in art and heritage conservation
(2010) from the Escuela de Conservación y Restauración
de Occidente in Guadalajara, Jalisco, Mexico, and an MA
in art theory and policy (2016) from Bezalel Academy
of Art and Design in Jerusalem, Israel.
Hadas Seri is an object conservator at the Metals and

Organic Materials Conservation Laboratory at the Israel
Museum, Jerusalem. She has an MA in art conservation,
objects stream (2010), from Queen’s University, King-
ston, Ontario, Canada, and an MA in art history and
BSc in chemistry and art history, both from the Hebrew
University, Jerusalem, Israel.
Maya Delano is object conservator at the Metals and

Organic Materials Conservation Laboratory at the Israel
Museum, Jerusalem. She completed art and archaeolog-
ical conservation studies in 2012 at Istituto per l’Arte e il
Restauro, Palazzo Spinelli, Florence, Italy, and has a BA
in art and archaeology (2010) from Tel Aviv University.
Irit Lev Beyth has been head of Metals and Organic

Materials Conservation at the Israel Museum, Jerusalem
since 2015, where she has been an object conservator
since 1998. She has a BA in museum studies (1989)
from Baylor University, Texas, USA, and an MA in
art conservation, objects stream (1994), from Queen’s
University, Kingston, Ontario, Canada.
Commercially Available Vacuum Chambers as an

Alternative in the Deoxygenated Desalination Treatment
of Archaeological Iron
Aleksi Pienimäki
Kiruna Centre for Conservation of Cultural Property
Kiruna, Sweden
[email protected]
Abstract
This study examined the feasibility of commercially available weekly using a luminescent dissolved oxygen probe, did not
vacuum chambers in the deoxygenated desalination treatment significantly differ between the NaOHvac and ASamb treatments,
of archaeological iron objects. Stainless steel chambers with but the oxygen concentrations in the ASvac treatment were
acrylic lids and silicone gaskets were shown to perform lower. However, care should be taken when interpreting the
well when used with alkaline desalination solutions. The results, as DO levels alone provide no information about the
performance of the chambers was then tested by desalinating reaction types or the dynamics of chloride extraction.
samples of terrestrial archaeological iron nails in alkaline sulfite
(ASvac: 0.5 M NaOH + 0.05 M Na2SO3) and sodium hydroxide Keywords
(NaOHvac: 0.5 M) solutions. The control treatments were carried archaeological iron, desalination, dissolved oxygen, partial
out using similar solutions in non-vacuum polypropene boxes vacuum
(ASamb and NaOHamb). Dissolved oxygen (DO) levels, measured
Introduction
The post-excavation corrosion of archaeological iron chambers has become commercially available in recent
objects is a major issue in the conservation of archae- decades. As a simple potential method of deoxygena-
ological metals. This type of corrosion is driven by the tion, they present an attractive option for desalination
aggressive ions, especially chloride, trapped inside the treatments.
corrosion layers (Turgoose 1985, Selwyn et al. 1999).
This study examined the feasibility of these chambers in
Accordingly, the removal of these ions through desal-
deoxygenating the desalination solutions used for archae-
ination is a significant part of the routine workflow in
ological iron. Evaluating the effectiveness of desalination
many conservation laboratories.
treatments is complicated, as discussed in detail elsewhere
Desalination is usually carried out by immersion in alka- (Watkinson 2010). In this study, the treatment solutions
line fluids. Deoxygenated treatment solutions have been were compared based on dissolved oxygen (DO) levels,
employed since the mid-1970s, after the introduction of according to two basic assumptions:
the alkaline sulfite (AS) treatment (North and Pearson
1. A dilute sulfite concentration of 0.05 M (Schmidt-Ott
1975). During subsequent decades, the effectiveness of
and Oswald 2006) was shown to perform as well as
this treatment was attributed to the presence of sulfite
the more concentrated version originally suggested
ions, which function as oxygen scavengers (Gilberg and
(North and Pearson 1975).
Seeley 1982, Turgoose 1985). Physical deoxygenation of
the treatment solutions followed, with promising results 2. In this setting, the sulfite ion has no other function
(Al-Zahrani 1999, Rimmer 2010). except as an oxygen scavenger (Gilberg and Seeley
1982, Turgoose 1985, Watkinson et al. 2013).
Laboratory-grade vacuum chambers have been used to
deoxygenate treatment solutions (Schmutzler 2012), but If the DO levels achieved with NaOHvac and ASamb are
these are expensive and their use is restricted in routine comparable, then the desalination performance should
treatment. However, an increasing assortment of vacuum be similar as well. Nonetheless, it must be noted that
DO levels do not provide insight into the reactions taking

place during treatment and the exact role of sulfite cannot
be defined with certainty.
The test material

Chambers supplied by VacuumChambers.eu were used
(Figure 1).
Figure 2. Example of the one-litre IKEA 365+ polypropene box with a

silicone gasket and locking lid, used in non-vacuum control batches in
the study
set at 0.8 L. The object weight:solution volume ratio was

about 1:5 in all batches.
Table 1. The treatment batches used in the study. d = treatment

duration in days, n = number of DO measurements carried out on
treatment batch
Batch Solution Start Finish d n

1 ASvac
11.3.2021 30.8.2021 162 19
9 ASvac 11.3.2021 30.8.2021 163 18
14 ASvac
11.3.2021 30.8.2021 162 19
4 NaOH vac
11.3.2021 21.9.2021 193 22
8 NaOHvac 11.3.2021 11.1.2022 291 34
11 NaOH vac
11.3.2021 18.1.2022 298 35
7 ASamb 11.3.2021 30.8.2021 162 19
12 AS amb
11.3.2021 28.7.2021 130 18
13 AS amb
11.3.2021 30.8.2021 162 19
6 NaOHamb 11.3.2021 21.9.2021 193 22
10 NaOH amb
11.3.2021 21.9.2021 174 22
16 NaOHamb 11.3.2021 23.11.2021 248 31
Figure 1. The one-litre vacuum chamber (from VacuumChambers.eu)

Methodology
used in the study
The tested solutions were AS (0.5 M NaOH + 0.05 M
The test chambers had a capacity of 1 L but chambers Na2SO3) and sodium hydroxide (0.5 M NaOH). The
with capacities as high as hundreds of litres are available. sodium hydroxide concentrations in desalination treat-
ments usually range between 0.1 and 0.5 M. A concen-
The performance of the vacuum chambers was tested
tration of 0.5 M was chosen as some studies stress the
against a control group of non-vacuum polypropylene
importance of maintaining a high hydroxide ion concen-
boxes with locking lids (Figure 2).
tration (Pelé et al. 2010, Schmutzler 2012, Coelho et al.
Test specimens consisted of 96 dry archaeological nails 2014). The original concentration for sulfite (North and
from a terrestrial site. These were divided into 12 batches Pearson 1975) was 0.5 M, but concentrations as low as
of eight nails, each batch with an average mass of about 0.05 M were subsequently recommended (Schmidt-Ott
160 g. Six randomly chosen batches were treated under and Oswald 2006) and shown to suffice for the deoxygen-
partial vacuum, while six were used as non-vacuum ation of solutions in closed containers. A concentration
controls (Table 1). The treatment solution volume was of 0.05 M was therefore chosen for this study.
COMMERCIALLY AVAILABLE VACUUM CHAMBERS AS AN ALTERNATIVE IN THE DEOXYGENATED DESALINATION TREATMENT OF ARCHAEOLOGICAL IRON 319
The partial vacuum inside the chambers reached an the DO levels in the ASvac treatment were the lowest
average of 10 kPa, with a fluctuation of less than ± 5 kPa. (Figure 3, Table 2)
Average leakage was < 10 kPa during the week-long
measurement interval. All treatments were carried out Discussion
at ambient temperature.
Care needs to be taken in the interpretation of DO
The treatments were completed between March 2021 readings, as they do not provide insight into the reac-
and January 2022. DO levels were determined weekly tions taking place during treatment. Further research is
using a luminescence dissolved oxygen probe (Hach needed on the differences between physical and chemical
Lange LDO101 probe connected to a Hach Lange deoxygenation methods.
HQ440D laboratory meter), except during a 4-week
period between 2 and 29 August 2021. Schmutzler (2012, 64–78) investigated the role of sulfite,
by applying it to physically deoxygenated solutions,
The stability criterion was set to 0.05 mg/L/min. When
but obtained mixed results. Although no clear pattern
this criterion was met, the measurement usually drifted
emerged, the results questioned the expectation that
towards a lower value, but this drift could be ignored.
deoxygenated solutions outperform oxygenated ones.
Note that the treatment chambers had to be opened and
the solutions exposed to atmospheric oxygen during Al-Zahrani (1999, 137–8) compared the desalination
measurement. efficiency of deionised water deoxygenated using sulfite
A positive systematic error in the measurements was with deionised water deoxygenated using nitrogen and
likely. Nonetheless, the achieved precision and the large found that the latter was more efficient in removing
number of readings allowed a meaningful comparison chlorides. The reason for the difference in performance
of the solutions. could not be identified, but the author suggested that
physical and chemical methods of deoxygenation are
Results not necessarily comparable.
There was no statistically significant difference in the DO Coelho et al. (2014) tested dilute and concentrated AS
levels in the ASamb and NaOHvac treatments. However, on marine archaeological cast iron and found out that
the concentrated solution outperformed the dilute one.
How far these results can be generalised and whether the
observed difference could be attributed to the sulfite or
hydroxide ion remain uncertain.
Rimmer (2010, 81) reported lower, more consistent

DO levels in ASamb treatments than were measured in
this study. This could be due to differences in treatment
vessels or the measurement method.
Deoxygenation is usually assessed qualitatively rather

than quantitatively. This study does not support this
approach, as DO levels over a wide range could be
measured.
Figure 3. Box and whisker graph of DO levels measured in the test solutions
Conclusion
Table 2. Statistical summary of the DO measurements of the test
solutions. All values in mg/L
Judging by the DO levels, NaOHvac was comparable to
Solution n Average Median SD ASamb. Thus, no difference is to be expected between the
AS amb
56 0.78 0.62 0.50 use of a vacuum chamber versus dilute sodium sulfite
NaOHamb 75 6.14 5.69 0.99 as a method of deoxygenation, assuming that the sulfite
ASvac
56 0.45 0.46 0.17 ion does not serve other purposes than that of an oxygen
NaOHvac 91 0.83 0.78 0.35
scavenger.
If the deoxygenation levels achieved with ASamb are logical Iron Conservation Colloquium 2010 Extended
deemed sufficient, NaOHvac is a valid substitute for this Abstracts, Stuttgart, 24–26 June 2010, eds. G. Eggert and
solution. If the aim is to reach consistently low DO levels, B. Schmutzler, session 4. Stuttgart: State Academy of Art
better results are achieved using ASvac. and Design Stuttgart.
The sulfite ion often complicates chloride determinations. Rimmer, M.B. 2010. Investigating the treatment of chlo-
Thus, physical deoxygenation simplifies both chloride ride-infested archaeological iron objects. PhD disserta-
determination and the disposal of the used solution tion, Cardiff University, UK.
(North 1987; Wang et al. 2008; Scott and Eggert 2009, Scott, D.A. and G. Eggert. 2009. Iron and steel in art:
141). The use of vacuum chambers is simpler than deox- Corrosion, colorants, conservation. London: Archetype
ygenation with nitrogen. Publications Ltd.
Significant differences in the performance of ASamb versus Schmutzler, B. 2012. Rettung vor dem Rost: Die Weiter-
NaOHvac could arise despite the comparable DO levels entwicklung der Eisenentsalzung nach der Alkali-Sulfit-
measured in these solutions. Methode zur Erhaltung großer Fundmengen. Rahden:
Verlag Marie Leidorf Lmt.
Acknowledgements Schmidt-Ott, K. and N. Oswald. 2006. Neues zur
The author would like to thank his supervisor, Stavroula Eisenentsalzung mit alkalischem Sulfit. Beiträge zur
Golfomitsou, from the University of Gothenburg for Erhaltung von Kunst- und Kulturgut 2: 126–34.
her advice and support and John Lagerstedt from the Selwyn, L.S., P.J. Sirois, and V. Argyropoulos. 1999. The
Finnish Heritage Agency/Archaeological Field Services corrosion of excavated archaeological iron with details
for providing the test material. This work was financially on weeping and akaganéite. Studies in Conservation
supported by the Märta, Gunnar and Arvid Bothéns Trust. 44(4): 217–32.
Turgoose, S. 1985. The corrosion of archaeological iron
References
during burial and treatment. Studies in Conservation
Al-Zahrani, A.N. 1999. Chloride ion removal from 30(1): 13–18.
archaeological iron and ß-FeOOH. PhD dissertation,
Wang, Q., S. Dove, F. Shearman, and M. Smirniou. 2008.
University of Wales, UK.
Evaluation of methods of chloride ion concentration
Coelho, J.C., C.M. Oliveira, M.D. Carvalho, and I.T.E. determination and effectiveness of desalination treat-
Fonseca. 2014. The efficiency of electrochemical methods ments using sodium hydroxide and alkaline sulphite
for the removal of chloride ions from iron marine solutions. The Conservator 31(1): 67–74.
archaeological objects: A comparative study. Materials
Watkinson, D. 2010. Measuring effectiveness of washing
and Corrosion 65(1): 38–44.
methods for corrosion control of archaeological iron:
Gilberg, M.R. and N.J. Seeley. 1982. The alkaline sodium problems and challenges. Corrosion Engineering, Science
sulphite reduction process for archaeological iron: A and Technology 45(5): 400–406.
closer look. Studies in Conservation 27(4): 180–4.
Watkinson, D., M.B. Rimmer, and F. Kergourlay. 2013.
North, N.A. 1987. Conservation of metals. In Conser- Alkaline desalination techniques for archaeological
vation of marine archaeological objects, ed. C. Pearson, iron. In Corrosion and conservation of cultural heritage
207–52. Bodmin: Butterworth & Co. Ltd. metallic artefacts, eds. P. Dillmann, D. Watkinson, E.
North, N.A. and C. Pearson 1975. Alkaline sulfite Angelini, and A. Adriaens, 407–33. Cambridge: Wood-
reduction treatment of marine iron. In ICOM-CC 4th head Publishing Ltd.
Triennial Meeting Preprints, Venice, 13–18 October 1975,
75133-1-75133-14. Rotterdam: Bouwcentrum (Impr.) Author
[for] the International Council of Museums. Available Aleksi Pienimäki has a BA in archaeology from the
at www.icom-cc-publications-online.org/ University of Helsinki and a BA in objects conserva-
Pelé, C., S. Lemoine, and E. Guilminot. 2010. Evolution tion from Helsinki Metropolia University of Applied
of pH in the solutions of dechlorination. In Archaeo- Sciences. Since 2017, he has been enrolled in the Master’s
COMMERCIALLY AVAILABLE VACUUM CHAMBERS AS AN ALTERNATIVE IN THE DEOXYGENATED DESALINATION TREATMENT OF ARCHAEOLOGICAL IRON 321
Programme in Conservation at the University of Goth-

enburg. Since 2019, he has been employed as a conser-
vator of archaeological metals at the Kiruna Centre for
Conservation of Cultural Property.
Materials list
Sodium hydroxide (AnalaR NORMAPUR®)
Sodium sulfite anhydrous (AnalaR NORMAPUR®)
VWR International/Avantor
www.vwr.com
Hach Lange HQ440D laboratory meter

Hach Lange LDO101 probe
Hach Lange GmbH
www.hach.com
IKEA 365+ Food container with lid, rectangular/plastic 1.0 L

IKEA Group
www.ikea.com
Vacuum chamber 1.0 L, stainless steel VC0918SS

VacuumChambers.eu
www.vacuumchambers.eu
POSTER SESSION
324 ICOM-CC | METAL 2022 | HELSINKI, FINLAND POSTER SESSION
Subcritical Fluids in the Conservation of Archaeological

Iron Objects: From Experiments to Permanent Practice
Aki Arponen* Anna Hyppönen Petri Kilpeläinen Kalle Kaipanen

The National Museum of Finland The National Museum of Finland Natural Resources Institute Finland Natural Resources Institute Finland
Finnish Heritage Agency Finnish Heritage Agency (Luke) (Luke)
Helsinki, Finland Helsinki, Finland Helsinki, Finland Helsinki, Finland
Subcritical fluids are an efficient approach to the conser- stainless steel wire to stainless steel chains. The chains
vation of archaeological iron objects but their use has were attached to a steel plate that was lifted into the
been limited by the costs of the required device. In 2020, tank. Again, processing did not damage the objects,
the National Museum of Finland began negotiations with hematite was easily removed and no corrosion occurred.
the Natural Resources Institute Finland (Luke) to test Sewing the pouches for each object, however, was time
the subcritical processing of archaeological iron objects consuming and the mesh broke down easily when it
using the Institute’s devices. Testing was conducted with caught on a sharp edge. Moreover, as the mesh is non-re-
the aim of developing the method so that it could be cyclable, the method is both costly and non-ecological.
introduced on a permanent basis. The first processing The third processing test was conducted in February 2022
test was performed in February 2021 with a small device using the same device. However, the objects were not
and 11 objects. Each object was placed in stainless steel placed in mesh but in stainless steel baskets that in turn
mesh and then transferred to a 3-litre tank. After 10 were placed on stainless steel shelves in the tank. Some
days of processing, all items were undamaged and there of the baskets were custom made for this purpose, and
was no evident activation of corrosion during the obser- others were bought ready-made. As in the previous tests,
vation period of several weeks. Hematite accumulated no problems emerged with respect to the objects or the
on the surface of the objects, but it was easily removed results, but processing with the baskets proved to be both
with glass-bead micro-air abrasion. Use of a stainless time saving and object friendly. The National Museum
steel mesh, however, is not advised because the folded of Finland will continue developing this subcritical
rigid mesh may harm the objects and the sharp edges of processing method in collaboration with Luke until the
the mesh caused injuries to the handler. In the second end of 2023. The goal in 2024 is for the Finnish Heritage
processing test, in November 2021, a larger and more Agency’s Collection and Conservation Centre to acquire
advanced device was used. The 300-litre tank held 70 its own device such that the subcritical processing of
objects, each sewn into a PTFE mesh and attached with archaeological iron objects will be a permanent practice.
ICOM-CC | METAL 2022 | HELSINKI, FINLAND POSTER SESSION 325
When Corrosion is Useful: Absolute Dating of Metals by

the Radiocarbon Method
Lucile Beck* Cyrielle Messager

LMC14 National Laboratory, Laboratory of Climate and Environmental Sciences LMC14 National Laboratory, Laboratory of Climate and Environmental Sciences
(LSCE), French Alternative Energies and Atomic Energy Commission (CEA) (LSCE), French Alternative Energies and Atomic Energy Commission (CEA)
Saclay, France Saclay, France
The radiocarbon dating method is usually applied to radiocarbon method. In order to better understand
organic materials – wood, charcoal, bone, etc. – which the incorporation mechanism of 14C in cerussite and
absorb radioactive carbon through photosynthesis hydrocerussite, lead carbonates were produced by the
or ingestion. However, a few studies have shown that corrosion process under various monitored experi-
radiocarbon dating can also reveal the age of inorganic mental conditions. Lead carbonates were obtained using
manufactured materials. Lead carbonates cerussite two types of acid (vinegar containing 14C vs 14C free
(PbCO3) and hydrocerussite (2PbCO3Pb (OH)2) were acetic acid) and three types of CO2 sources (atmospheric
used as white pigment or cosmetics from the 4th century air, horse manure containing 14C and 14C free fossil
BC to the 20th century. They were formed by the corro- CO2 gas). Twelve different conditions were tested and
sion of metallic lead by vinegar and horse manure. These the results show the production of lead acetates when
organic materials release carbon dioxide (CO2), and it CO2 is absent and the production of cerussite when
has been recently demonstrated that this CO2 has been CO2 is present. Cerussite incorporated 14C when horse
incorporated into the corrosion products, meaning that manure was used. These results confirm the action of
the carbonate function of the lead carbonates carries a vinegar/acetic acid as a precursor and horse manure
14
C signature corresponding to the organic matters used. as a reagent to produce carbonates carrying 14C, which
Therefore, lead white pigment and cosmetics produced allows the absolute dating of lead carbonates using the
as a result of this corrosion process can be dated by the radiocarbon method.
Corrosion, Communication, and Comparison:

Collaborative Conservation Approaches Towards Two
Medieval Axes from Norfolk, England
Kate Berlewen* Reed Hudson
Royal Albert Memorial Museum University College London
Exeter, UK London, UK
Two Medieval axes from Norfolk, England, required notes, and written records, were utilised to communicate
stabilisation before display due to active corrosion and the progression of decision-making and treatments.
illegibility of features of interest. A uniform final appear- Documentation and research were shared through
ance was desired, both to clarify the nature of the objects the file-sharing platforms Dropbox, WeTransfer, and
and because both axes would likely be displayed together. WhatsApp. These methods also helped mitigate working
This posed challenges as the two axes, both casual finds patterns being broken by lockdowns, providing continuity
from different sites, had undergone diverse historic through well-documented thought processes and stages
treatments and been stored for varying lengths of time, of treatment. The techniques and materials used included:
leading to very different states of conservation. Further (i) mechanical removal of previous conservation mate-
challenges were encountered, as each object was worked rials and corrosion products through air abrasion and
on as an individual student project and the treatments hand tools; (ii) reconstruction using Paraloid B-48N
took place over a year of lockdowns with intermittent (methyl methacrylate and butyl acrylate copolymer) in
laboratory access. Three key aims were identified for the acetone; (iii) application of supportive gap fills using
treatments, which needed to respond sensitively to the iron powder, glass micro balloons, and fumed silica with
structure and distribution of corrosion products, features Paraloid B-48N in industrial methylated spirits/acetone;
of interest, and stability of each object: (i) the removal of (iv) passivation with tannic acid (C76H52O46) in deionised
corrosion products allowing legibility of the axe typology water; and (v) coating with low concentration Paraloid
and features of interest, including pseudomorphic corro- B-44 (methyl methacrylate and ethyl acrylate copolymer)
sion patterns; (ii) appropriate stabilisation and protection in toluene, bulked with fumed silica as a mattifying
of the objects from further corrosion; and (iii) a final agent. While both axes were conserved using a different
cohesive appearance of the colour and reflectivity of combination of these techniques and materials tailored
any coatings or passivation films on both objects, to aid to their specific needs, the collaborative decision-making
their legibility. A process of collaborative working was process ensured a cohesive final appearance and state
adopted between the students, making use of in-person of conservation for the axes as a pair. The flexible and
laboratory work whenever possible. When it was not cooperative approach also ensured the adaptability and
possible to work together, recorded video updates and resilience of the project, enabling each student to work
shared documentation, including photography, research when it was safe and accessible to them.
Atmospheric Particulate Matter: How to Include it in

Artificial Ageing?
Elena Bernardi* Cecilia Velino Ivano Vassura

Department of Industrial Department of Industrial Department of Industrial
Chemistry, University of Bologna Chemistry, University of Bologna Chemistry, University of Bologna
Bologna, Italy Bologna, Italy Bologna, Italy
Carla Martini Cristina Chiavari *Author for correspondence

Department of Industrial Department of Cultural Heritage,
Engineering, University of Bologna University of Bologna
Bologna, Italy Bologna, Italy
Corrosion phenomena occurring on outdoor-exposed analyse the effect of PM on bronze specimens. To obtain
metallic artefacts have been the subject of several ambient PM-induced corrosion patterns to validate the
studies, but relatively few of them have investigated or complete artificial ageing procedure (from PM formu-
attempted to artificially reproduce the specific effects lation to ageing), ambient PM was collected through an
induced by atmospheric particulate matter (PM). Beside ad hoc accelerated sampling system on bronze substrates
natural ageing, reliable artificial ageing procedures are (Cu-5Sn-5Zn-5Pb). The ionic composition of the PM
key to corrosion mechanism studies, as well as to repro- was determined and the specimens were exposed to
duce representative corroded substrates and complex accelerated ageing in a climatic chamber (simulating
environmental conditions to assess the effectiveness daily variations of T, RH% and UVA irradiation,
of protective products. PM, unlike gaseous pollutants according to the cycle previously defined). Two seasonal
and rain, is not often included in accelerated ageing campaigns were performed to collect the widest range
methods for cultural heritage materials, and generally, of PM composition/concentration. The specimens were
when considered, only single salts or a simple mix of periodically observed by optical microscopy and, at
salts are used. This is due to several problematic aspects the end of the ageing process, were characterised using
related to appropriate PM sampling and effective repro- different techniques (optical microscopy, field emission
ducibility of real PM compositions. In order to develop gun-scanning electron microscopy, energy-dispersive
new coatings, the lack of a suitable ageing procedure X-ray spectroscopy, infrared spectroscopy, Raman
that includes PM (and its potentially synergic inter- spectroscopy and UV-Vis spectrophotometry for colour
action with UV) can produce discrepancies between analysis). The corrosion products were then removed
the predicted behaviour of protective treatments in to determine their metal content as well as mass loss
traditional ageing tests and real-world performance: and changes in the surface profile. The qualitative and
coatings with successful results under laboratory condi- quantitative results identified the main relationships
tions fail under real conditions. To solve these issues, between the ambient PM composition/concentration
several actions were carried out: a wide literature survey and surface evolution. The results obtained so far are
followed by data elaboration and a preliminary formula- very promising for the development of a protocol for
tion of synthetic PM; the set-up of an accelerated ageing reproducible accelerated ageing tests that include PM
cycle in a climatic chamber and of a methodology to as a corrosive factor.
Study and Restoration of the Trajan’s Column

Electrotypes at the National Museum of Archaeology of
Saint-Germain-en-Laye
Isabel Bonora Andujar* Antoine Amarger Anouk Molineri
Musée d’Archéologie nationale Independent conservator Independent conservator
Saint-Germain-en-Laye, France Paris, France Paris, France
Julia Jouet Brittany Branche *Author for correspondence

Independent conservator Independent conservator
Paris, France Vitry-sur-Seine, France
In 1861, the French Emperor Napoleon III obtained analysis of the present elements by portable X-ray fluo-
permission from Pope Pius IX to make a plaster replica rescence, and identification of subsequent interventions),
of Trajan’s Column. A few months later, the 450 molds their state of conservation, and the establishment of a
were exhibited in Paris at the Palace of Industry, where, restoration protocol allowing the surface and structure of
for the first time, all of the bas-reliefs on the column, the electrotypes to be treated and their presentation in an
which is almost 40 meters high, could be admired and exhibition space. Because of their varied state of conser-
studied. Concerned about the durability of the molds, vation, the project was quite complex. The bas-reliefs have
and in order to erect a similar column to that of Napo- very different surface aspects, which are linked to the
leon I in Place Vendôme, the emperor sent them to the conditions of conservation as well as the manufacturing
industrialist Léopold Oudry to make copper electrotypes, process. Some of the electrotypes are particularly fragile,
during which process the molds were mostly destroyed. because they are cracked, fragmented, and sometimes
In 1864, the electrotypes were successfully exhibited at incomplete. To establish a protocol adapted to the casts
the Louvre Museum, and, in 1872, were permanently as a whole, and so that the results were representative,
transferred to the National Museum of Archaeology in the study phase was carried out on eight casts in various
Saint-Germain-en-Laye, where the first section has been states of conservation. As far as the structural aspect was
exhibited in the castle moat ever since. Electrotyping, an concerned, consolidation tests were carried out to define
innovative technique for its time, was used in the field of the most suitable materials to meet the criteria of revers-
statuary for only a short time. To study this little-known ibility, support compatibility, durability, shaping, and
technique, it was necessary to collect information on the mechanical strength. Tests focused on modern synthetic
manufacturing parameters, analyze the observed altera- materials to create removable reinforced resin shells. For
tions, and develop treatment recommendations. At the the surface, the choice of most suitable treatment was
request of the museum, and while keeping in mind the dependent on future conservation conditions and had to
desire to continue to exhibit all of the electrotypes, the consider the problems of mass treatment and structural
first phase of study consisted of a technical analysis of fragilities. To meet these constraints, selective cleaning
the pieces (a description of the manufacturing process, tests, by spraying plant abrasives, were first carried out.
Acoustic Emission Techniques for the Detection and

Monitoring of Corrosion Phenomena
Laura Brambilla* Alejandro Roda-Buch

Haute Ecole Arc Conservation-restauration (HE-Arc CR), HES-SO University of Tribology and Interfacial Chemistry Group, École polytechnique fédérale de
Applied Sciences and Arts Western Switzerland Lausanne (EPFL)
Neuchâtel, Switzerland Lausanne, Switzerland
Brice Chalançon Stefano Mischler

Association de Gestion du Musée National de l’Automobile Tribology and Interfacial Chemistry Group, École polytechnique fédérale de
Mulhouse, France Lausanne (EPFL)
[email protected] Lausanne, Switzerland
[email protected]
In the field of cultural heritage, the use of non-inva- the wet environment of the cooling system. To confirm
sive techniques to diagnose and monitor artefacts is of this hypothesis, an experimental setup, mimicking the
utmost importance in order to avoid surface changes cooling system under investigation, was developed. An
and damages to the materials. However, detection of electrochemical cell was assembled using specific glue to
ongoing corrosion phenomena in closed environments mount PVC tubes on iron coupons, and AE sensors were
is still a challenge. This poster presents the results installed on the external part of the coupons. A classic
from applying acoustic emission (AE) techniques to three-electrode electrochemical cell was then created
non-invasive detection and monitoring of ongoing on the internal part of the coupons. This setup allowed
corrosion inside the cooling systems of historical vehicle the correlation between electrochemical measurements,
engines. The study was conducted within the framework typically used for corrosion monitoring, to be assessed
of the ACUME_HV project. The main objective was to invasively and for non-invasive AE measurements to be
develop a technique for monitoring engine mechanisms taken. The iron coupons were initially pre-corroded.
during their reactivation. As part of the monitoring According to the protocol, the corrosion layers were
process, an engine was initially emptied of its cooling first left to dry and subsequently the cell was filled
liquid and lubricants for preliminary measurements. again with electrolyte, in order to simulate the process
An inspection of the cooling system, made of cast iron, observed previously in the engine’s cooling system. The
using an endoscope revealed the presence of extended correlation of electrochemical and AE measurements
corrosion and thick patinas. The cooling liquid was revealed that a new ongoing corrosion process did
then re-introduced into the system and AE sensors, indeed exist and therefore it was possible to conclude
placed in contact with the external surface of the engine, that AE was a promising method, even in the case of
recorded intense and frequent hits over a 24-hour closed systems that cannot be accessed for inspection
period. The origin of these signals was hypothesised with standard tools or by any other monitoring methods
to be related to the onset of a new corrosion process in known (spectroscopic, electrochemical, etc.).
The Bronze Hand of Prêles: Protecting a Valuable

Archaeological Object Using 3D Techniques
Sabine Brechbühl
Archaeological Service Bern
Bern, Switzerland
[email protected]
For conservators, dealing with a unique object of high importance were established, its documentation was
value and renown presents numerous challenges. completed with 3D scanning, which provided a true-
However, 3D techniques can be used to reduce the many to-scale backup of the shape and colour of the hand
stresses to which such objects are frequently exposed. In that could be used for a multitude of purposes. For
2017, metal detector searches resulted in the discovery example, a 3D print was made that was used in defining
of ‘the bronze hand of Prêles’, which was subsequently and planning interventions as well as in discussions of
brought to the Archaeological Service of Bern. This typological details and the manufacturing process. In
object, which dates to the Middle Bronze Age, is the addition, the necessary mountings for tomographic
oldest metallic representation of a human body part analyses, for sampling the metal or for exhibition could
thus far known in Europe. The hand is a massive casted be produced without ever having to touch the original.
bronze object partly plated with a decorated gold foil. The 3D data were also used to plan the packing of
Given its high value for archaeological research, it is in the object. Ethafoam was milled according to the 3D
steady demand for analyses, exhibition, documentation data and two shells with the exact shape of the hand
and study. However, climatic changes, manipulation, were produced, thus providing accurate support and
transport and mountings are major sources of stress to cushioning from all sides. Currently, tests are being
archaeological objects and endanger their preservation conducted to produce a high-quality copy of the hand.
and integrity. These risks must be balanced against the A high-resolution scan and print will be manually
need to examine the objects to determine their cultural coloured and gilded. Although the precision is not
and historical meaning. A method that allows essential yet at quite the same level as that of moulds realised
activities while minimising the exposure and manipu- with silicone rubber, the 3D print nonetheless avoids
lation of fragile objects is therefore needed. In the case subjecting the original object to possible contamination
of the bronze hand, after its age, provenance and hence and other potential sources of damage.
Corrosion of Strings on Musical Instruments
Vera de Bruyn-Ouboter* Andreas Erbe Erlend Flø Gustad

Ringve Music Museum/Rockheim, Department of Materials Science Department of Materials Science
The Museums of Southern and Engineering, Norwegian and Engineering, Norwegian
Trøndelag (MiST) University of Science and University of Science and
Trondheim, Norway Technology (NTNU) Technology (NTNU)
[email protected] Trondheim, Norway Trondheim, Norway
https://ringve.no/en [email protected] [email protected]
https://mist.no/en www.ntnu.edu/ima *Author for correspondence
In the collections at the Ringve Music Museum, Trond- presence of long-chain carboxylic acids on the metal
heim, Norway, a range of stringed musical instruments surface. The characteristic absorbance in IR increased
exhibit contact corrosion. This affects the metal strings over time. The decomposition of cellulose through
of stringed instruments as well as keyboard instruments reaction with oxygen can in part explain the presence of
such as pianos. The critical point is where a metal string carboxylates but not the presence of all detected signals.
is attached to the sounding body. When in contact with Interestingly, in contrast to the samples in contact with
other materials, such as felt, leather or plastic, the string the fabric, lab-exposed reference samples that had not
tends to corrode and crack. This type of corrosion has been in contact with the textile fabric showed clearly
been observed over the course of several years. Simple visible signs of corrosion. The same metal sheets were
attempts have been made to isolate the materials from one exposed inside a square piano (ca. 1825, Scandinavia)
another, but the problem reoccurs. The relative humidity for 30 days. Surface analysis also indicated the pres-
and temperature in storage are controlled and the air ence of carboxylates in these samples, but with shorter
quality has been analysed for the presence of formic hydrocarbon chains compared to the samples exposed
acid, acetic acid, nitrogen dioxide and ozone, but the in lab conditions. XPS analysis indicated the presence of
amounts are lower than critical threshold values. In labo- bromine, possibly stemming from the textile dye or other
ratory model experiments, low-alloyed steel sheets were textile treatment. Scanning electron microscopy with
cleaned in sodium hydroxide and exposed for 30 days energy-dispersive X-ray analysis was used to investigate
in contact with cotton and wool piano textile fabric. the composition of the piano strings. The results shed
Analysis of the surface by grazing incidence infrared (IR) more light on the challenges in understanding contact
reflectance absorbance spectroscopy (time-dependent), degradation processes at the atomic-scale and nanoscale,
X-ray photoelectron spectroscopy (XPS) and secondary and in developing appropriate conservation-compatible
ion mass spectrometry showed typical signals for the mitigation strategies.
Do the Lubricating Oils Used in Scientific-Technological

Objects Protect Metals Against Corrosion?
Emilio Cano* Blanca Ramírez Barat Joaquina Leal María Teresa Molina
National Centre for Metallurgical National Centre for Metallurgical National Museum of Science and National Centre for Metallurgical
Research, Spanish National Research, Spanish National Technology (MUNCYT) Research, Spanish National
Research Council (CENIM–CSIC) Research Council (CENIM–CSIC) Alcobendas, Spain Research Council (CENIM–CSIC)
Madrid, Spain Madrid, Spain [email protected] Madrid, Spain
One of the challenges in the conservation of scientific work (Metal 04), refinements to their methodology for
and technological heritage objects is their use and performing EIS were introduced. Working electrodes
operation, which distinguishes them from works of were prepared with the rods of each metal, embedded
artistic heritage and complicates the establishment of in epoxy resin, attached to a copper wire for electrical
conservation criteria with respect to their maintenance, contact, and the flat cross-section left exposed. This side
restoration and valorisation. Moreover, these objects are was dipped in the oils chosen for testing and positioned
often machinery made up of several materials, including horizontally in the electrochemical cell. Once the oils
those related to their operation and mechanisms, such as had been evaluated by EIS to determine their protective
greases, waxes and lubricating oils, whose preservation, properties, they were further analysed by external reflec-
renewal or elimination must be decided upon. The aim tion Fourier transform infrared spectroscopy (ER-FTIR).
of this work was to evaluate whether the lubricating oils This methodology resulted in more homogeneous and
in scientific-technological objects, given their ageing and reproducible oil layers. The differences in the protective
lack of use, have a corrosive or protective effect on the properties of the oils were large, with variations in the
objects’ metals. In this work, inspired by David Hallam’s impedance modulus at low frequencies of several orders
research, several original oils were evaluated using elec- of magnitude, depending on the composition of the
trochemical impedance spectroscopy (EIS). Samples of sample and its contamination from previous use. Two
lubricating oils were taken from different objects from the new oils used in preservation/restoration were examined
National Museum of Science and Technology (Madrid). as well: of note, one of them had a high protective capacity
The objects have not been in use since the end of their and the other none at all. Evidence of the contamination
service life and still retain their original materials. Steel and ageing of the oils was obtained using ER-FTIR, which
and brass were selected as the test metals in accordance revealed peaks indicative of oxidation, nitration and
with their abundance in the collection of the museum. In water. The presence of these impurities has been shown
light of the difficulties reported by Hallam et al. in their to affect the reproducibility of EIS measurements.
The Bluish Green and the Greenish Blue
Mariana Cardoso
Independent conservator
Lisbon, Portugal
[email protected]
Is it blue? Is it green? Is it grey? I see red! I see brown! Poster Session, all the coloured cards will be added to the
No, it’s reddish brown. Or maybe, brownish red. Perhaps poster, finally completing it. The purpose of the game
yellowish brown. I think it’s opaque. No, dull. Normally is to highlight the fact that the description of corrosion
it’s translucent. But can be also vitreous. It’s homogeneous, products in many publications and other texts can
but sometimes heterogeneous. I know it as a deep colour. vary widely. With the contributions of a large number
Heavily saturated. Almost black. I heard it is pale. I heard of professionals, the game is therefore a good exercise
it is light. My poster will be designed as a simple board in developing a consensus description of corrosion
game allowing the participation of many conference products based on colour and other features. Diver-
attendees. Rectangular sheets of paper will be distrib- gences and similarities in colour perception and in the
uted during the conference and coloured pencils will colour-based descriptions of the most common copper
be made available. The attendees will be asked to draw corrosion products will also be revealed. Participation
the colour of certain corrosion products (e.g. nantokite, in the game will encourage discussions of the work
atacamite, cuprite and paratacamite) and to provide a experiences of colleagues in the areas of copper objects
written description of the colour underneath. At the conservation and related topics.
Heritage Biota and Bronze Patina Composition:

A Correlation?
Cristina Chiavari* Carla Martini Federica Costantini

University of Bologna University of Bologna University of Bologna
Bologna, Italy Bologna, Italy Bologna, Italy
Andrea Timoncini* Elena Bernardi *Author for correspondence

University of Bologna University of Bologna
Bologna, Italy Bologna, Italy
Epilithic bacteria play a fundamental role in the conser- The aims of the present study were to: (i) characterise the
vation of cultural heritage materials in the outdoor bacterial community on bronze statues by conducting
environment. Bacterial communities cause stone and DNA analysis based on the 16S RNA gene, and specifi-
metal degradation by the chemical dissolution/precip- cally using samples obtained from sheltered/unsheltered
itation of substrate ions, biochemical redox reactions areas on the surfaces of monuments in three Italian cities
as well as organic and inorganic acid release, resulting (Rimini, Ravenna and Bologna), in order to generate a
in discolouration of the surface and physical alteration. bacterial library; (ii) determine the correlations between
However, in some cases, bacteria exert bioprotective chemical profiles and the biological characteristics
actions. For bronze statues in non-buried conditions, of the patina; (iii) obtain insights into the relation
few studies have examined the associated biota or the between climate and heritage biota. The next step of
possible correlation between the chemical profile of the this research will consist of laboratory studies of the
patina (i.e. pale green, black crusts) and the bacterial degradation of bronze fostered by biota communities
community. Moreover, the changing environment, such and climate change and investigations of the underlying
as decreased SO2 concentrations, relative enrichments mechanisms. This will be pursued through accelerated
of NOx, O3 and particulate matter, and, more generally, ageing tests, including the use of artificial acid rains, the
local variations in multi-pollutants combined with composition of which is formulated through updated
climate change effects, creates new, mostly aggressive environmental monitoring data, and bacterial commu-
scenarios that impact the decay of cultural heritage mate- nities selected on the basis of data obtained from actual
rials and therefore their bacterial communities as well. statues (bacterial library).
Testing Silver Lacquers: What about Agateen #27?
Gerhard Eggert* Gregory Dale Smith Michael J. Samide

State Academy of Art and Design Indianapolis Museum of Art at Butler University
Stuttgart, Germany Newfields Indianapolis IN, USA
[email protected] Indianapolis IN, USA [email protected]
Materials for storage and display are nowadays routinely cyanide on brown sites and AgCN·2AgNO3 on black
Oddy-tested to avoid corrosive emissions harming objects sites, the latter supported by scanning electron micros-
in closed cases. Conservation materials (adhesives, copy–energy-dispersive X-ray spectroscopy (SEM-EDX)
consolidants, coatings) which are permanently attached elemental mapping. The unusual reaction might have
to objects must of course meet the same standard. Cellu- been driven by the extreme insolubility of silver cyanide
lose nitrate products fail the Oddy test, which has been and the availability of reactive carbon and nitrogen
confirmed for the British silver lacquer Frigilene. Lead,
compounds from the decomposition of cellulose nitrate:
copper and the silver coupon were clearly corroded, and,
Ring-CH2-O-NO2 → Ring· ·CH2-O· ·NO2 → Ring· + CH2=O + ·NO2
surprisingly, a cyanide compound (AgCN·2AgNO3) was
detected by Raman spectroscopy and powder diffrac- CH2O + 3NO2 + 3 Ag → AgCN·2AgNO3 + H2O
tion in tests performed in Stuttgart and presented at In addition to the Oddy test, Agateen #27 was analysed
Metal 2019 (Neuchâtel, 2–6 September 2019). In the by direct thermal desorption–gas chromatography–mass
US, Agateen #27 is frequently used by conservators. spectrometry (GC–MS) for volatile emissions. A number
However, as it is not on sale in or shipped to Germany, it of compounds of concern were detected: NO, acetic
could not be included in the Stuttgart tests. A test at the
acid, formaldehyde and other aldehydes, and various
Indianapolis Museum of Art at Newfields yielded extreme
alkylnitrates. Although no unequivocal evidence for
corrosion and discolouration on all coupons, and the test
the formation of cyanide on lacquered silver objects was
beaker exploded a few days into the run. In a repetition
found, it also cannot be ruled out. The release of NOx
experiment in milder conditions (40 instead of 60 °C)
and its effect on other sensitive materials alone is reason
using only 0.5 g fully dried Agateen film (4 years old), a
yellow-brown tarnish with black spots on the silver foil enough to discard cellulose nitrate products as conser-
was analysed after 159 days of exposure. The formation vation materials. Unfortunately, if silver lacquering is
of at least two cyanide compounds could be confirmed needed, there is no good replacement available today with
with Fourier transform infrared (FTIR) spectroscopy a stable material that has all the advantages of cellulose
and Raman spectroscopy by the distinctive C≡N stretch nitrate coatings. New research with modern coatings is
vibrations between 2000 and 2200 cm−1: an unknown urgently needed.
Possibilities for Implementing Aesthetic Unity,

Reversibility and Distinctiveness: Two Conservation Case
Studies
Eszter Tóth
Hungarian National Museum
Budapest, Hungary
[email protected]
Two prominent weapons from the armoury collection in smithing techniques. The engravings were emphasized
the Hungarian National Museum were subject to conser- with black paint rather than black substrate infilling to
vation before display in the permanent exhibition to make the replaced pieces more distinguishable. The other
present the evolution of weaponry in Hungarian history. weapon, a sword that is presumed to have belonged to
Conservation was undertaken using traditional principles the Hungarian king Matthias Corvinus (1458–1490), is
(distinctiveness, aesthetic unity, reversibility, minimum presumably a copy from the 17th century, but has a Latin
intervention) and was used for educational purposes as text etched on the blade that refers to the ruler. The grip
part of the Applied Arts Object Conservation Program is decorated with a gilded copper filigree braid consisting
organized jointly by the Hungarian National Museum and of twisted and straight wires. Partially broken and incom-
the Hungarian University of Fine Arts. The first weapon plete, this filigree decoration has diminished the aesthetic
was a wheellock pistol from Nürnberg (16th century) unity. However, in the absence of comparable examples,
with gilded steel parts and a wooden body decorated the reconstruction plan was developed by museologists.
with engraved inlays made of mother-of-pearl and bone. The plan seemed to be in jeopardy due to the fact that
Originally, both inlay materials were carved and filled the wood grip was originally covered with a textile, so
in with a black substance. The missing inlays were made fixing the new wires caused further difficulties. A crack
of the same material in accordance with the originals. on the wooden part of the grip served as a fixing point
Gluing of the supplements without any decoration was for the complementing patinated copper wire used as a
not considered an acceptable solution, since their plain supplement for the heavily worn gilded copper filigree.
white surfaces were conspicuous and confusing from an The copper wires were affixed only to the wooden infill
aesthetic point of view. After studying the decoration of the grip with epoxy resin, while the original broken
pattern on the original parts and consulting museologists, filigree pieces were joined without adhesive and merely
the supplements were engraved using traditional gold- stitched under the undamaged braids.
Supercritical Carbon Dioxide: Cleaning and Stabilization Tests

for Copper Alloys and Iron Keys from a Scientific Collection of
the Centro Hospitalar Conde de Ferreira (Porto, Portugal)
Catarina Figueiredo* Carlo Bottaini Inês Silva Eduarda Vieira
School of Arts, Universidade HERCULES Laboratory, Centre for Biotechnology and Fine Research Centre for Science and
Católica Portuguesa Universidade de Évora Chemistry – Associate Laboratory Technology of the Arts (CITAR),
Porto, Portugal CityUMacau Chair in Sustainable (CBQF), Faculty of Biotechnology, School of Arts, Universidade
[email protected] Heritage, Universidade de Évora 8 Universidade Católica Portuguesa Católica Portuguesa
Évora, Portugal Porto, Portugal Porto, Portugal
Nuno Camarneiro [email protected] [email protected] [email protected]
Research Centre for Science and
Technology of the Arts (CITAR), Rui Bordalo Marta Duarte *Author for correspondence
School of Arts, Universidade Research Centre for Science and Centre for Biotechnology and Fine
Católica Portuguesa Technology of the Arts (CITAR), Chemistry – Associate Laboratory
Porto, Portugal School of Arts, Universidade (CBQF), Faculty of Biotechnology,
[email protected] Católica Portuguesa Universidade Católica Portuguesa
Porto, Portugal Porto, Portugal
An increasing number of eco-friendly nanotechnologies Portable X-ray fluorescence was carried out before
for the conservation of cultural heritage materials are and after scCO2 treatment to determine the elemental
becoming available. These include supercritical carbon composition of the metal alloys and the effectiveness
dioxide (scCO2), which has proven to be a viable green, of the cleaning process. Before the objects were treated,
non-flammable, and ecological method. While previ- scCO2 was tested on eight samples of similar material
ously tested and now widely used in the pharmaceutical and degradation pathologies. Comparative experiments
and food industries, scCO2 has recently been applied were performed with and without the addition of a
in the conservation of cultural heritage objects, and co-solvent to scCO2. The co-solvents included ethanol,
specifically for their cleaning, consolidation, and stabili- hydrogen peroxide, ethylenediaminetetraacetic acid
zation. Experience has shown that scCO2 is a promising (EDTA), and sodium hexametaphosphate. Different
alternative to conventional toxic organic solvents for temperatures, reaction times, and pressures were eval-
the removal of organic materials, such as oils, grease uated. Cleaning efficiency was assessed by the visual
deposits, and active corrosion products. In this study, examination of dirt removal, material stabilization,
we tested and applied scCO2 as a cleaning and stabiliza- oxidation stain removal, fiber mechanical strength,
tion agent on two coils of copper and silk textile wires and dye loss. The results revealed that scCO2 without
used in the electrical induction mechanism of a faradic the addition of a co-solvent did not have a significant
machine and on an iron key from a set of handcuffs. impact on the samples compared to when it was applied
Both objects come from a set of psychiatric instruments with ethanol or EDTA, which as co-solvents improved
and equipment dating from the 19th to the mid-20th cleaning and stabilized the corrosion products. The
century that now form part of the scientific collection respective formulations proved to be suitable alter-
of the Centro Hospitalar Conde de Ferreira, which natives for the treatment of metal artifacts as well as
belongs to Museu da Misericórdia do Porto, Portugal. objects composed of different materials.
A Simple Treatment System for the Deaerated

Desalination of Archaeological Iron
Jacob Frydendahl* Trine Wiinblad Delphine Neff Henning Matthiesen

The Royal Danish Academy – The National Museum of Denmark LAPA NIMBE – IRAMAT, CEA/ The National Museum of Denmark
Conservation Lyngby, Denmark CNRS/U P Saclay, UMR 3685 Lyngby, Denmark
Copenhagen, Denmark [email protected] Gif-sur-Yvette, France [email protected]
Charlotte Carré *Author for correspondence
Lars Brock Andersen LAPA NIMBE – IRAMAT, CEA/ Philippe Dillmann
Centre for Preservation of Cultural CNRS/U P Saclay, UMR 3685 LAPA NIMBE – IRAMAT, CEA/
Heritage Gif-sur-Yvette, France CNRS/U P Saclay, UMR 3685
Vejle, Denmark [email protected] Gif-sur-Yvette, France
Deaerated alkaline desalination is a well-established, N2 flow rates, thus minimizing both the cost and the
highly effective treatment for archaeological iron as environmental impact of treatment. A similar system
it prevents further corrosion and induces the transfor small-scale treatment can easily be constructed as
formation of chlorinated phases while avoiding the well, using the materials on hand at most conservation
precipitation of green rust. Deaeration is most often workshops. The method is not limited to the use of
achieved by the addition of the oxygen scavenger alkaline solutions, as treatment can also be performed
sodium sulfite. However, the use of sulfite has raised at neutral pH if the right combination of solution
concerns because of the potentially harmful effects of and gas is used. Chloride release during treatment in
sulfate remnants in the treated objects. The problem can 0.5 M NaOH and 0.01 M NaHCO3 was measured for
be avoided by purging with inert gas instead, but this 12 samples of archaeological iron excavated from a
requires a specialized treatment set-up, which until now chlorinated terrestrial soil environment as well as for
has limited its use. We developed a simple, inexpensive, samples of pure akaganeite. In addition, selected samples
and robust treatment method that relies on oxygen were assessed by scanning electron microscopy with
displacement by N2 gas and allows the simultaneous energy-dispersive spectroscopy, Raman spectroscopy,
treatment of several objects in separate containers. and total chloride measurements. The results suggest
Monitoring over the last 1.5 years has shown that stable that a very high chloride extraction efficiency can be
anoxic conditions can be obtained even at very low reached using our gas-based method.
Metals Conservation Teaching at University Level in

Greece and the Cultural Vision of EYDAP S.A.
Maria Giannoulaki* Christina Panagiotopoulou Sofia Tzimopoulou Vasilike Argyropoulos

Department of Conservation Division of Communication Division of Communication Department of Conservation
of Antiquities and Works of Art, and Corporate Affairs, EYDAP’s and Corporate Affairs, EYDAP’s of Antiquities and Works of Art,
University of West Attica Historical Archive, EYDAP S.A. Historical Archive, EYDAP S.A. University of West Attica
Athens, Greece Athens, Greece Athens, Greece Athens, Greece
www.uniwa.gr www.eydap.gr www.eydap.gr www.uniwa.gr
This study looked at the integration of the conservation Marathon Museum. The objects are mainly made of
of a significant historic metals collection with training at iron and copper alloys and other metals. Many of them
university level through collaboration between the Metals are composite metal objects, comprising metal and
Conservation Laboratory and Research Laboratory inorganic materials, such as glass, or organic materials,
for the Study & Conservation of Ancient and Modern such as wood, leather, fabric, plastic, or paper. Some have
Cultural Properties (CULTLAB) at the University of coatings or are painted. They are important documents
West Attica and the Historical Archive of the Athens of technological history and, along with photographic
Water Supply and Sewerage Company (EYDAP S.A.). archives, a snapshot of the period, where a prosperous
Through its aim to improve the lives of individuals and and self-sufficient settlement for workers grew up around
to aid the development and prosperity of society, EYDAP the museum. For the exhibition to mark the 90th anni-
supports initiatives and actions based on social engage- versary of the building of the dam, EYDAP assigned
ment and the promotion of culture. Its historical archive CULTLAB the project to conserve 64 selected objects.
is responsible for compiling, storing, managing, and A joint decision was made to integrate the conservation
promoting a collection of tools and machinery used for of the collection with the educational work of the Metals
the construction of the Marathon Dam, a technological Conservation Laboratory, giving students the opportu-
achievement in the 1920s that serves as a major symbol nity to develop practical skills, scientific documentation,
of the modernization of Greece. These objects, which are and decision-making in the aim to limit corrosion and
of various types, such as tools and machinery, lighting preserve and highlight the technological information and
and heating equipment, measuring instruments, pumps, values of the objects as museum exhibits and witnesses
typewriters, and calculators, are exhibited at EYDAP’s of technological achievement and progress.
A Lacquer to Dye For! Exploring a UV Fluorescent Additive

for Coating Applications in Silver Conservation
Rosie Grayburn* Lauren Fair

Winterthur Museum, Garden & Library Winterthur Museum, Garden & Library
University of Delaware Art Conservation Department University of Delaware Art Conservation Department
Winterthur DE, USA Winterthur DE, USA
The conservation and scientific analysis departments camera with a 20 mm lens, a Peca 918 filter, a UV filter
at Winterthur Museum have developed a silver coating (similar to the Kodak Wratten 2E gel filter), a 30 s
program over 40+ years. Our last phase of research integration time, and a 365 nm UV light source. The
occurred in 2016–2019. Building on research from empirical data clearly showed that the fluorescence
colleagues at MFA Houston and Colonial Williamsburg, wanes after exposure under ambient conditions. More-
we refined our Agateen Lacquer #27 recipe—unmodi- over, there was a clear shift in the visible spectrum, as
fied, non-plasticized, high-molecular-weight cellulose the coating became more yellow in select areas of the
nitrate derived from cotton linters and dissolved in tray after one year of ambient exposure. To understand
butyl acetate, butanol, toluene, ethanol, and isopropanol the dye-coating interaction at a molecular level, a more
(manufactured by Agate Lacquer Tri-Nat, LLC and avail- controlled experiment was designed using mock-ups.
able from manufacturer and at conservation suppliers, Thin films of Agateen containing a range of Coumarin
including Talas)—by the addition of the UV-fluores- 6 concentrations were applied to glass slides by brush in
cent dye Coumarin 6 (3-(2-benzothiazolyl)-N,N-di- the same manner that the coatings would be applied to
ethylumbelliferylamine), CAS # 38215-36-0, available objects: two coats with a 50 min drying time in between.
from Sigma Aldrich). When applied to silver, the dye The films were aged at 70°C and 55% RH for 400 h.
contained in the lacquer is not detectable in visible light Absorbance and fluorescence spectra were collected
but fluoresces bright yellow-green at long-wave UV. before and after thermal aging to gauge the lifetime of
The fluorescence enables conservators to immediately the useful optical properties of the dye. Fourier trans-
assess the coating application process, by observing form infrared spectra were obtained to gauge polymer
lacunae, thin spots, or other flaws. Nonetheless, it was deterioration patterns. The glass transition temperature
important to understand the long-term effects of this was also measured before and after aging to assess the
additive, whose properties were desirable for only a effect of the dye on film porosity and to understand
small duration of the coating lifetime. Fluorescence whether the dye would act as a plasticizer over time.
images of a coated silver tray were captured soon after The aim of this research was to determine the long-term
treatment and again after one year under ambient aesthetic and performance effects of application-focused
conditions. Images were obtained using a Nikon D850 additives for silver lacquer.
Port-Jeanne-d’Arc: Is Conservation of an Isolated Site

Possible?
Marie Grima
Terres Australes et Antarctiques Françaises (TAAF)
Saint-Pierre, La Réunion, France
[email protected]
Port-Jeanne-d’Arc was the only French whaling station (ii) the lack of continuity in the heritage missions within
and it remains one of the rare examples of industrial the French Southern and Antarctic Lands (TAAF)
seal activity in the sub-Antarctic. The aim of a mission complicates research and follow-up; (iii) the station is
undertaken between October 2020 and April 2021 located on a natural reserve classified as an UNESCO
was to diagnose the condition of Port-Jeanne-d’Arc’s World Heritage Site, which raises the question whether
metallic structure. Located in Kerguelen, in the French this isolated site, with limited visitor numbers (less
sub-Antarctic, Port-Jeanne-d’Arc was built in 1908 by than 50 tourists per year), should or can be enhanced,
the Norwegian company Storm, Bull & Co., on the as large-scale operations would imply significant logis-
orders of France’s Bossières brothers. From the station, tics and thus the need for efforts to ensure a minimum
sea mammals were hunted, mainly elephant seals but environmental impact; finally, (iv) the remoteness of the
also whales. The oil thus produced was used for street territory makes it difficult to implement conservation
lighting in Europe and North America. However, the operations. Promotion of the heritage of the Austral
overall yield of the project was poor and the station was Islands may require that the conservation of the respec-
closed in 1926. In conservation terms, Port-Jeanne- tive materials be abandoned. We thus recommend that
d’Arc, despite the uniqueness of its heritage, raises Port-Jeanne-d’Arc be examined in all its aspects before
several issues, some of which overlap with those of its final disappearance, including through the use of
classic, industrial heritage: (i) the site has typical conser- alternative tools such as 3D scanning. Specific lines of
vation problems, including its destruction due to the research have been initiated, including investigations
natural degradation of its materials, accelerated by the of the behaviour of puddled iron, a material extensively
aggressive environment, and to looting and vandalism; used in the base construction of the station.
Evaluation of the Impact of Dechlorination Treatments on

the Organic Parts of Composite Objects
Elodie Guilminot* Sylvie Labroche
Arc’Antique Laboratory, Grand patrimoine de Loire-Atlantique (GPLA) Arc’Antique Laboratory, Grand patrimoine de Loire-Atlantique (GPLA)
Nantes, France Nantes, France
Charlène Pelé-Meziani* *Author for correspondence

Arc’Antique Laboratory, Grand patrimoine de Loire-Atlantique (GPLA)
Nantes, France
[email protected]
Ferrous objects can be effectively stabilised by chemical analysis showed a degradation of carbohydrates at neutral
or electrochemical dechlorination. However, the treat- and slightly alkaline pH (KNO3, Na2SO3, sesquicarbonate).
ment of objects composed of both ferrous and organic Lignin degradation started at pH > 11.5 and increased
elements is challenging, because chemicals may adversely in higher-pH solutions (NaOH/Na2SO3 from the first
affect the latter. In this study, the impact of a range of month and NH4OH, NH2OH more visible 9 months
dechlorination treatments on bone, wood, horn, ivory later). Leather: After one month, samples in solutions with
and leather, as representative organic materials, was pH > 13 (NH2OH and NaOH/Na2SO3) had degraded and
assessed. Two samples and one reference sample per degradation continued during the entire 9-month period.
material were immersed in one of six chemical solu- Samples immersed in NaOH/Na2SO3 completely jellified.
tions selected through bibliographical research and/ In sesquicarbonate and KNO3, both the volume and
or following experimental treatments: NaOH/Na2SO3 delamination increased. FTIR analysis showed a degra-
(20 g/L–10 g/L, pH 13.6), Na2SO3 (0,5% w/w, pH 9.1), dation of amide I and II bonds at pH < 12; at higher pH
sodium sesquicarbonate (5.6 g/L–4.5 g/L, pH 10.0), (> 12), the polysaccharide bonds underwent degradation.
NH2OH (2.5% v/v, pH 13.1), KNO3 (1.0% w/w, pH 6.7) and Bone: There was no visible degradation except for samples
NH4OH (1.0 mol/L, pH 11.7). The samples were analysed immersed in NaOH/Na2SO3, which became greyer than
by Fourier transform infrared (FTIR) spectroscopy after 4 the reference. FTIR analysis revealed bone mineral
and 37 weeks of immersion. The initial results ruled out degradation, especially in Na2SO3. Degradation became
the use of certain chemical solutions, based on the visible apparent within the first month and continued thereafter.
appearance of the samples. For example, horn and leather Horn: Samples immersed in NaOH/Na2SO3 became
immersed in NaOH/Na2SO3 became gelatinous such that completely gelatinous within the first month whereas
FTIR analysis was impossible. By contrast, the appearance samples immersed in NH2OH became soft and grey. In
of the same materials after 1 month in NH2OH (pH 13) solutions with a pH > 12, FTIR analysis showed degrada-
was nearly unchanged. FTIR analyses showed a marked tion of the cysteine/cystic acid bonds. Lower-pH solutions
decrease only in the intensity of bonds corresponding to chemically impacted the amide bonds, especially Na2SO3
bands at 1080 cm−1 (attributed to dC-H, C-H3ro/?C-N (80%). Ivory: The colour and mechanical strain of samples
amide I/polysaccharide bonds). For the other samples, immersed for 1 month remained stable. However, samples
despite bacterial growth in the solutions, there was no immersed in NaOH/Na2SO3 were completely delaminated
visible alteration in their appearance or their resistance and bleached. FTIR analysis showed stable ratios, indi-
to mechanical handling. Based on these experiments cating similar level of organic and mineral degradation.
and the FTIR data, a table summarising the treatments, Except for those immersed in sesquicarbonate, all samples
parameters of interest (pH, volume, solution composition) displayed dark stains that could not be removed by rinsing.
and the results for each organic material was established. FTIR analysis showed the chemical stability of the mineral
The information can be cross-referenced with data on part, with a significant increase in the ratio of amide bonds
the dechlorination of an object’s ferrous parts. Wood: in samples immersed in KNO3 and NH2OH. The amide
No visible degradation in colour or mechanical strain bonds of samples immersed in Na2SO3, sesquicarbonate
occurred, irrespective of the treatment. However, FTIR or NH4OH were slightly degraded.
Historical Repairs on Ancient Chinese Bronzes:

Identification and Conservation Issues
Ke-Shiuan (Tracy) Han

Rijksmuseum
Amsterdam, The Netherlands
[email protected]
This project examined the historical restoration of the reproduction of missing elements, and the use of
Chinese bronzes to identify conservation risks and the patinas resembling corrosion layers and incrustations –
types of intervention that can improve preservation, referred to in various expert texts from the 12th to the
contributing to the historical discussion on the conser- 18th century. Traditional restoration materials consisted
vation of ancient Chinese bronzes in international of organic binding media, salts, and metal infills. Cast
collections. Since the late 19th century, ancient Chinese bronze sheet composed of 80% Cu, 15% Sn, and 5% Pb
bronzes dating from 2000–221 BCE have been collected was therefore artificially corroded and the reconstructed
and exported around the world, and they reveal signif- surface visually characterized and documented, followed
icant variations in conservation. What explains this? If by artificial aging in a climate chamber. X-ray diffraction
materials added to the original object during repair are (XRD) and Fourier transform infrared (FTIR) spectros-
unstable, to what extent is removal appropriate? Tradi- copy were conducted to understand the material property
tional restoration techniques for ancient Chinese bronzes changes. Aging the reconstructed samples artificially
have been used since antiquity and were particularly allowed changes in the material to be observed and
popular in the 15th–19th century. These treatments, such the preservation issues caused by old treatments to be
as surface patination, often altered the original appear- determined. The data collected will be used to create a
ance of the object, reflecting changing historical styles database and a reference framework within which to
and tastes in restoration in China. The challenge facing recognize alterations in ancient Chinese bronzes and
conservators who now deal with these bronzes is that examine the intentions behind the treatment of these
no technical research on traditional Chinese restoration historical objects. Furthermore, it will help to identify
practices has been carried out. This gap in knowledge broader conservation risks within the collections where
makes it difficult to treat these objects, as there exists no intervention can improve preservation. More generally,
reference framework for preservation or intervention. the project will contribute to the historical discussion on
The research therefore aimed to examine traditional the conservation of ancient Chinese bronzes in interna-
restoration techniques – including rejoining methods, tional collections.
Microclimates for Marine Archaeological Iron Artillery:

Three Case Studies in Treatment, Storage, and Display
Laurie King* Erik Farrell

The Mariners’ Museum and Park The Mariners’ Museum and Park
Newport News VA, USA Newport News VA, USA
The unstable nature of marine archaeological iron is of dehumidifier to quickly dry artillery after desalination
primary concern to its conservation. It often contains treatment. The second case study focuses on storage,
trapped chloride species even after desalination, driving where a maintenance environment is required that can
future corrosion. Limiting this corrosion process maintain low RH in a large air volume with minimal
within an aerobic atmosphere requires desiccation to active input. A method used at The Mariners’ Museum
a relative humidity (RH) of less than 12%. Small iron and Park in Virginia, USA, involved lining a wooden
objects can be sealed in anything from Tupperware crate with Marvelseal to create a storage environment
to barrier films, using silica gel to create a desiccated that would experience minimal environmental change,
microclimate. Artillery, however, is large, heavy, and using silica gel to control RH. The third case study
difficult to contain, and the hollow bore can develop a revolves around display, where environmental condi-
different microclimate than the exterior surface. These tions are similar to those for storage but with added
challenges impact the conservation of marine archaeo- aesthetic requirements to accommodate exhibit design
logical iron artillery during interventive treatment, both decisions. A system developed by QAR Lab illustrates
in storage and while on display. This poster illustrates a unique solution to microclimate desiccation in cases
conservation problems at these three different stages of where exhibit designs impose significant constraints.
artifact care and conservation, providing a case study The outcome was the creation of segmented tubes that
for each. In active treatment, environmental control is can hold enough desiccant to control the microclimate
vital during drying; a longer time at higher RH increases and be easily inserted into artillery barrels with limited
the formation of akaganeite, leading to less desirable space in front of the muzzle. In demonstrating how these
long-term outcomes. A system implemented at the three solutions have evolved iteratively, this poster is
Queen Anne’s Revenge Conservation Laboratory (QAR intended to provide ideas for solutions and to spark a
Lab) in North Carolina, USA, used a Lexan box, large broader discussion on how archaeological artillery is
enough to hold artillery, and a recirculating ducted conserved and maintained.
Results of Research on 17th–19th-Century Coffin

Portraits in the Collection of the National Museum in
Krakow
Maria Labut* Julio del Hoyo-Meléndez Paulina Krupska Maria Goryl
National Museum in Krakow National Museum in Krakow National Museum in Krakow Faculty of Conservation and
Krakow, Poland Krakow, Poland Krakow, Poland Restoration of Works of Art,
[email protected] [email protected] [email protected] Academy of Fine Arts
Krakow, Poland
Aldona Stępień Michał Obarzanowski Katarzyna Stefańczyk [email protected]
National Museum in Krakow National Museum in Krakow Research and Development Centre
Krakow, Poland Krakow, Poland for Renovation and Conservation *Author for correspondence
[email protected] [email protected] Nysa, Poland
[email protected]
The project’s main goal was to research the techniques to have been used from the 17th to the 19th century
and technology used to create the coffin portraits in when the portraits were made. The pigments identified
the National Museum in Krakow’s collection. Coffin included lead white and orpiment. Other pigments
portraits are a unique example of Polish cultural that help to establish a potential date of production,
heritage popular in the 17th and 18th centuries. They such as lead-tin yellow, Prussian blue and Ba and Zn
were common amongst the Polish nobility who cele- white, were also identified. Fourier transform infrared
brated funerals with great ceremony and spectacle and spectroscopy was employed to identify the organic
at which portraits were used to evoke the impression compounds present in the portraits and confirmed
of the deceased’s presence. They were painted very the presence of oil and proteinaceous components as
realistically on a metal sheet and attached to the shorter the paint binders. The composition of the metal alloys
side of the coffin, so they were usually hexagonal or used for the supports was identified by portable XRF
octagonal in shape. The coffin portraits were studied spectrometry, which revealed that four of them were
with a combination of analytical tools. Visual examina- made of copper and the remaining seven of binary
tion by VIS, IR, UV and RTG camera and microscopic alloys such as Sn-Pb, Cu-Zn and Ag-Cu. Voltam-
observation of cross-sections allowed their stratigraphy metry of immobilised particles was used to investigate
to be established. The cross-sections were additionally corrosion samples at the nanogram level to discern
examined using scanning electron microscopy with between regions with different types of alterations. The
energy-dispersive X-ray spectroscopy. The paint layers identification of corrosion products and knowledge of
were examined using micro and macro X-ray fluores- the range of secondary layers from later restorations
cence (XRF) spectrometry. The analysis confirmed the enabled the selection of the most appropriate conser-
presence of elements associated with pigments known vation materials.
Examination of an Unusual Apulian-Corinthian Helmet

Using X-Ray and Computed Tomography
Manuel Leroux* Elsa Lambert

Centre for Research and Restoration of the Museums of France (C2RMF) Centre for Research and Restoration of the Museums of France (C2RMF)
Paris, France Paris, France
A preliminary study was conducted at the Centre for rescence and micro gas chromatography combined with
Research and Restoration of the Museums of France mass spectrometry, provided a precise understanding of
(C2RMF) on a partially preserved Apulian-Corinthian the previous restoration. Many original fragments were
helmet with a surviving inlaid eye belonging to the Musée included in a large modern polyester putty reconstruc-
départemental des Antiquités of Rouen in Normandy, tion incorporating large and small fragments as a filling.
France. Computed tomography (CT) and X-radiog- In some instances, modern brass powder was mixed
raphy were used to help solve a mystery surrounding the with the polyester filling causing corrosion, probably
helmet. A comparison with similar bronze helmets and due to the presence of chlorides in the mixture. The CT
prometopidia (armor that protected a horse’s forehead) cross-sectional images of the helmet allowed a better
with inlaid bone eyes indicates that the Rouen helmet understanding of the multilayered restoration work but
was made in southern Italy during the 5th century also highlighted its original fabrication. CT examination
BC. However, the curious shape of the helmet has no also helped understand how the surviving polychromed
known equivalent. The examination and analysis of eye was attached as well as the method of assembly of
this unusual artifact revealed that the original shape the crest using rivets. The objective of this poster is to
had been extensively altered and reinterpreted during highlight the critical role that examination techniques
restoration in 1965. The combination of X-ray and CT routinely used at C2RMF play in the interpretation of
examination, supported by scanning macro X-ray fluo- artifacts.
Diagnostic, Restoration and Maintenance of Outdoor Bronze

Artworks: Investigating Patinas in Marine Environments and
Their Stabilisation with Low Environmental Impact Treatments
Paola Letardi* Giulia Monari
Institute for the Study of Anthropic Impacts and Sustainability in the Marine Institute for the Study of Anthropic Impacts and Sustainability in the Marine
Environment, National Research Council (IAS–CNR) Environment, National Research Council (IAS–CNR)
Genoa, Italy Genoa, Italy
The conservation of outdoor bronze artworks is chal- gated in detail. Thus, with the goal of more successful
lenging because of the need to consider the highly conservation treatments of outdoor bronze monuments,
complex system of alloy+patina/coating/environment. we conducted multi-analytical NDT measurements of
Despite the numerous reports on bronze conservation marine-weathered coupons and then compared the
published in the scientific literature, in most cases results to those obtained with several bronze artworks,
the studies have been limited to model surfaces and mainly from the 19th and 20th centuries, in the Stag-
laboratory-based investigations. Among the issues that lieno Monumental Cemetery (Genoa, Italy) and Certosa
remain to be addressed in detail are: (i) the corrosion Monumental Cemetery (Bologna, Italy). Our primary
of outdoor bronze surfaces in a changing environment, objective was to better define the variability ranges of
especially the marine environment, with respect to relevant parameters and the similarities of the surface
bronze disease; (ii) the value of the different surface morphologies of the coupons and those of the outdoor
products, in order to better evaluate the effectiveness metal artworks exposed to different environments.
of different cleaning methods; and (iii) the morpho- Moreover, we consider the applications of the avail-
logical properties of corrosion products with respect able NDT techniques to better define good practices
to the behaviour of protective treatments (inhibi- for outdoor bronze conservation. Particular attention
tors, coatings). Insights into the complex interactions was paid to the identification of nantokite, a promoter
underlining these issues can only be obtained through of bronze disease, and its uneven distribution within
carefully designed experiments. In heritage science, the patinas. We also discuss the results of preliminary tests
multi-analytical in situ approach offered by non-de- directed at the in situ characterisation of traditional
structive testing (NDT) allows the properties of heritage and innovative treatments of patinated bronze surfaces
objects, such as visual appearance, chemical composi- aimed at inhibiting the cyclic corrosion linked to the
tion, corrosion rate and surface texture, to be investi- presence of CuCl.
The Curious Case of Storage at the Petrie Museum of

Egyptian Archaeology
Graeme McArthur* Merry Chow

University College London University College London
London, UK London, UK
Following safe storage during the Second World War, the materials used to build the storage cupboards. Fourier
the Petrie Museum’s collection was moved into a former transform infrared (FTIR) analysis was thus carried out
stable on the University College London campus. Storage to identify the plastic used to make the drawers; although
for non-display objects consisted of purpose-built it is certainly cellulose based, an exact match was not
wooden cupboards containing wooden drawers. Most found. Samples of corrosion products analysed with
of the collection, including metals, remains there to FTIR and X-ray diffraction (XRD) suggested that the blue
this day. Very recently it was noticed that the base of corrosion product was copper chloride acetate, known to
most of the drawers is made from a brown plastic that form on copper artefacts in the presence of acetic acid.
probably originates from the 1950s. This immediately Wooden storage often produces acetic acid, but this tends
raised concerns about its stability and the potential for to diminish over time, so there were concerns that the
off-gassing. To begin to investigate this problem, an Eltek plastic was the most likely cause. However, samples of
pollutant monitor was placed inside one of the cupboards. plastic in a sealed polythene bag produced no changes
It read high levels of volatile organic compounds that in colour to an A-D strip, whereas samples of wood
were not present in the air outside. As part of this inves- produced a clear change. From this it was possible to
tigation, a condition survey of the 7,000+ metal objects infer that, despite the age of the cupboards, the wood
was carried out. Although most of the collection is in was still off-gassing. A dedicated metals storage area has
stable condition, some corrosion products of an unusual now been created with Polstore steel cupboards and door
appearance were observed. On some bronze objects, seals to allow some control of the environment via silica
powdery blue corrosion was forming, while others lying gel. This will allow unstable metal artefacts to be stored
flat in Plastazote cut-outs were found to have a different at low levels of relative humidity in a cupboard made
corrosion colour on each side. This evidence all pointed from inert materials, removing them from the vicinity
towards the metal objects potentially being affected by of inappropriate storage materials.
Comparing Simple Portable XRF with PIXE Analysis for

Archaeological Metal Objects: Suitability, Advantages and
Drawbacks
Eva Menart* Žiga Šmit
National Museum of Slovenia Jožef Stefan Institute
Jožef Stefan Institute Faculty of Mathematics and Physics, University of Ljubljana (retired)
Ljubljana, Slovenia Ljubljana, Slovenia
Portable X-ray fluorescence spectrometry (pXRF) has results obtained from pXRF and PIXE match very well
been adopted by many heritage institutions and conser- in terms of both the identification and quantification of
vation laboratories, due to its non-invasive nature, different elements. For the main alloy components, the
simplicity and relatively low price, at least compared differences were within reasonable uncertainty ranges,
to more sophisticated analytical instruments. As a thus demonstrating that the simpler XRF can safely be
semi-quantitative tool for alloy identification, pXRF used to determine the elemental concentrations of the
is of particular interest for conservation purposes. main metal elements. An important advantage of PIXE
However, more in-depth studies are often left to much is its more accurate determination of low-concentration
more complicated scientific instrumentation, such as or trace elements as well as its lower detection limits
proton-induced X-ray emission (PIXE) or even the more (and smaller measurement spots). While the commer-
invasive laser ablation, coupled with inductively coupled cial software of the portable XRF instrument allows
plasma mass spectrometry. To investigate the semi-quan- the determination of element-mass concentrations
titative nature of pXRF, a series of analyses was carried below 0.1%, in these cases the data must be treated
out, first on standard materials and then on real archae- with caution and confirmed by the source data, i.e. the
ological objects, to compare the results obtained using spectra. Although the XRF instruments are generally
pXRF (Hitachi X-MET 8000 and imbedded software for simple to use, a correct interpretation of the data requires
spectrum analysis) and PIXE (Tandetron accelerator, a certain level of expertise. When investigating metal
custom-/in-house-made software). Commercial XRF objects, pXRF and PIXE data, when analysed correctly,
instruments commonly include a series of calibrations are comparable and both are accurate enough to avoid
for the user to select according to the samples. Our the need for destructive analysis. However, in trace
experiments showed that if the calibration is selected analysis (e.g. to determine provenance), the use of at
correctly (i.e. without excluding certain elements), the least PIXE is recommended.
Methods of Marking Small Museum Objects: A Joint

Project of the National Museum in Warsaw and the
National Institute of Cultural Heritage in Poland
Anna Mistewicz* Justyna Kwiatkowska
National Museum in Warsaw National Museum in Warsaw
Warsaw, Poland Warsaw, Poland
In 2020, the National Museum in Warsaw (NMW) and on large collections made up of thousands of objects is a
the National Institute of Cultural Heritage (NICH) in time-consuming practice, the proposed method should
Poland initiated a joint project entitled ‘Analysis of also be easy to apply and inexpensive. Five marking
methods of marking artifacts and museum objects with methods have so far been analyzed: microparticles
the use of microtechnology’. Its aim is to analyze the use (microholograms), RFID passive tags, mixtures of phos-
of micromarkings to label objects from the collection of phors invisible to the naked eye that have a particular
the NMW that cannot be numbered using traditional glow on exposure to ultraviolet light, QR codes, and
methods, due to their size, surface texture, or decora- data matrix codes. Several meetings with specialists
tion. This problem mostly concerns small metal objects from different sectors have taken place. Oddy tests of
such as coins, which comprise roughly one-fourth of the some tags and microparticles in resin were conducted
NMW’s collection. The long-term aim of the project to evaluate the risk of corrosion initiation by the chosen
is to devise an effective and safe method of marking markers on heritage materials. The microparticle method
all museum and heritage objects. The chosen method (which may include the addition of subsequent infor-
should enable clear identification and aid in verifying mation, although there is a limit to the total number of
an object’s provenance in case of theft or loss. Conse- digits) is readable only with the use of a magnifying glass
quently, it should include an accession number or at least or microscope. As this method was used experimentally
the museum’s name or a distinguishable owner’s mark. 15 years ago for marking ecclesiastical artifacts against
Desirable features of the marking method are small size, theft, its legibility, stability, and corrosion resistance were
legibility (detectability), reversibility, durability (long examined on site. Laboratory tests of methods to remove
life), and stability under changing environmental condi- this type of numbering were also conducted. The poster
tions as well as the ability to encode specific information will present the state of progress of the project and the
within the marking. Since applying accession numbers preliminary results.
Influence of the Application Methodology on the

Performance of Coatings for the Protection of Metallic
Heritage Objects
María Teresa Molina* Blanca Ramírez Barat Emilio Cano
National Centre for Metallurgical National Centre for Metallurgical National Centre for Metallurgical
Research, Spanish National Research, Spanish National Research, Spanish National
Research Council (CENIM–CSIC) Research Council (CENIM–CSIC) Research Council (CENIM–CSIC)
Madrid, Spain Madrid, Spain Madrid, Spain
Acrylic resins and waxes are still the most widely used samples to UVA radiation and condensation cycles. The
coatings for the protection of metallic heritage objects. corrosion protection of the coatings during accelerated
However, a literature review shows that the majority of ageing was characterised using electrochemical imped-
the research on the performance of coatings for metal ance spectroscopy and the coating efficacies compared
conservation has not paid sufficient attention to the using external reflection Fourier transform infrared
application methodology, making comparisons of the spectroscopy. The results showed that the high volatility
results across studies unreliable. In the COMPACT of acetone complicated a homogeneous application of
project, the protective capacity and stability of tradi- the coating on the sample, producing waves consistent
tional coatings for bare steel, i.e. Paraloid B-72, Incralac with solvent evaporation. The number of layers and the
and Cosmolloid Wax, were evaluated. These coatings thickness of the coating were not directly related. The
have been extensively studied over time, but application drying time of the coating was found to be one of the
variables such as drying time, coating thickness, number most important variables affecting its protective ability:
of layers, solvent used and application technique have After accelerated ageing, for coated steels left to dry for
been largely ignored. In this project, the influence of only 3 days, their impedance modulus at low frequencies
these variables on the corrosion protective capacity of was reduced by a much larger extent than that of samples
the coating and the physico-chemical changes resulting dried between 2 and 5 weeks. Infrared analyses showed
from artificial ageing were studied. Brush was selected chemical changes in the molecular bonds during the
as the method of application and acetone, xylene and ageing process in most samples. Our results demonstrate
white spirit as the solvents. Different numbers of layers the relevance of controlling the application parameters
were applied to bare steel samples, with each layer when comparing the performances of different coatings
allowed to dry for different times and the thickness for the conservation of metals. The protection offered
of the layer thus obtained then determined. The fully by different coatings depends not only on the product,
coated samples were left to dry for 3 days to 5 weeks. substrate and environmental conditions but also on the
Accelerated ageing tests were carried by exposing the application methodology.
Archaeological Tinned-Copper Objects: A Study on

Corrosion and Cleaning Treatments
Mohammad Mortazavi* Abbas Abed-Esfahani

Art University of Isfahan Islamic Azad University – Khorasgan Branch
Isfahan, Islamic Republic of Iran Isfahan, Islamic Republic of Iran
www.aui.ac.ir www.khuisf.ac.ir
Hamid Reza Bakhshandehfard *Author for correspondence

Art University of Isfahan
Isfahan, Islamic Republic of Iran
[email protected]
www.aui.ac.ir
The tinning of copper and copper-based objects has a effects of chemical reagents commonly used to clean
long history. This surface treatment, which is responsible and remove copper corrosion from tinned surfaces
for the silvery appearance of the treated objects, forms a were also investigated. For this purpose, an artificial
corrosion-resistant layer and has been a common practice tin oxide was electrochemically formed on the surface
since antiquity. The long-term burial of tinned-copper of experimental tinned-copper samples. The effects of
artifacts can lead to the formation of copper and tin various cleaning reagents—disodium ethylene diamine
corrosion compounds, which reflect the dissolution of tetraacetic acid (Na2-EDTA), sodium tripolyphosphate
copper followed by the precipitation of the corrosion (STPP), citric acid, formic acid, and alkaline Rochelle
products and their formation as a thick layer on the salt—on the electrochemically produced oxide layer were
surface. In fact, coverage of the surface with corrosion then compared. The amounts of dissolved tin that accu-
products disguises the fact that the object had been mulated in the cleaning reagents were determined using
tinned. However, cleaning can also dissolve the tin atomic absorption spectroscopy. The results showed that
layer, resulting in the loss of an important part of the tinned surfaces were dissolved by all of the tested chem-
object’s identity and of insights into the manufacturing ical solutions, but to varying degrees. Tin dissolution
process. Consequently, cleaning archaeological artifacts was highest in alkaline Rochelle salt. Although formic
made using a tinned-copper technique and covered acid was the least aggressive cleaning reagent, it also has
with corrosion products is a complicated task. In this the potential to damage the tin oxide layer beneath the
study, the corrosion process of historical tinned-copper copper corrosion products and therefore must be used
objects was examined using scanning electron micros- with caution. Mechanical cleaning, which may also cause
copy with energy-dispersive X-ray spectroscopy. The disruption on the surface, is also discussed.
Archaeometallurgical Investigations of a Late Iron Age

Helmet: An Interdisciplinary Approach
Nataša Nemeček* Lucija Grahek Peter Fajfar

National Museum of Slovenia Research Centre of the Slovenian Faculty of Natural Sciences and
Ljubljana, Slovenia Academy of Sciences and Arts Engineering, University of Ljubljana
[email protected] Ljubljana, Slovenia Ljubljana, Slovenia
Tomaž Lazar
National Museum of Slovenia Marijan Nečemer Borut Žužek
Ljubljana, Slovenia Jožef Stefan Institute Institute of Metals and Technology
[email protected] Ljubljana, Slovenia Ljubljana, Slovenia
In early 2020, a bronze helmet was found in Podzemelj, lographic analyses. Their microstructures showed that
southeastern Slovenia, at the newly discovered cemetery the helmet was made primarily by cold-working. A closer
Pezdirčeva njiva. The Slovenian-type Negova helmet investigation of the samples confirmed that the metal
dates to the Late Hallstatt period and Late Iron Age is heavily corroded. The stratigraphy of the corrosion
(500–200 BC) and was ritually damaged before its deposi- layers was studied in detail using optical microscopy
tion. It therefore ranks among the most important archae- analyses and scanning electron microscopy with ener-
ological finds in Slovenia in recent years. The decision to gy-dispersive spectroscopy. The presence of intergranular
submit the helmet to in-depth analyses was dictated by corrosion below the original surface dictated the choice
two factors. First, a better understanding of the helmet’s of conservation treatment, specifically, consolidation with
composition and state of preservation was needed in Paraloid B-72. The initial condition of the artefact was
order to determine suitable conservation techniques for documented using 3D scanning, which proved instru-
treating the heavily corroded find. Second, it was hoped mental over the course of subsequent conservation work.
that a metallurgical examination of the helmet would The applications of the resulting 3D model included the
reveal its technological make-up, thus providing valuable manufacture of a closely fitting internal frame enabling
insight into the craftsmanship and skills of the Iron Age the safe long-term display of the helmet. Identification
metalworkers and into the manufacturing processes they of the mechanisms and morphology of the corrosion of
employed. Our investigation began with non-destructive archaeological bronze has proved invaluable in helping
X-ray radiography and X-ray fluorescence analyses. The conservators to characterise deterioration processes and
latter revealed a high percentage of tin on the surface of take suitable steps to prevent their progression. Once
the helmet, consistent with the selective dissolution of the conservation work is finished, the helmet will be
copper from the alloy. Several samples detached from included in the new permanent exhibition in the Bela
various sections of the helmet were submitted to metal- Krajina Museum in Metlika.
The Care and Maintenance of Heritage Sample Collections
Peter Northover* Vanessa Cheel

University of Oxford (retired) University of Oxford
Oxford, UK Oxford, UK
A century of metallurgical research in cultural heritage, tion techniques this is now much less of a problem.) The
from the Chalcolithic to contemporary times, has collections with which we are at present concerned were
amassed a collection of many thousands of samples deposited with the intention of being loaned for re-use,
taken for compositional analysis and metallography. As and it is our responsibility to ensure that the methods
new or enhanced methods of analysis and microscopy used are appropriate, typically repolishing and etching
develop and become available, these samples provide for optical and electron microscopy and for electron
an enormous resource for future research and to backscatter diffraction, microhardness measurements
protect the original objects from further sampling. The and laser ablation. These techniques have a minimal
Archives and Collections Committee of the Historical impact on the sample and the mounts are otherwise
Metallurgy Society (HMS) currently curates a small conserved as received. Sometimes the mounts them-
number of such heritage sample collections left to us selves are deteriorating. Sixty-year-old epoxy does not
by pioneer researchers in the area. This poster aims to always survive well and then the samples are re-mounted
provide an idea of the practical and ethical questions in the same orientations as before. The availability and
we have had to address along the way and to describe conditions of access to a sample, together with a cata-
approaches to the management of sample collections in logue, are available on the HMS website. The existing
the future. The first and biggest question is the extent collections have primarily been used for metallography
to which new interventions should be allowed. Where and compositional data is often lacking. The documen-
a collection is sufficiently historically important, for tation for future accession will include metallurgical
example the samples prepared in the 1860s by Henry as well as archaeological data so researchers can more
Clifton Sorby, new interventions may be ruled out and easily find relevant material. Beyond that, it is our aim
the samples preserved as they are. At the other extreme, to develop an area on our website where the metadata
samples have been consumed for lead isotope analysis in for other sample collections will be published, ensuring
order to avoid re-sampling an object. (With laser abla- that their value is properly realised.
Monitoring the Effect of Humidity on the Tarnishing of

Silver 0.800 by Means of Electrochemical Noise
Janette Ortíz-Corona Francisco Javier Rodríguez-Gómez*

National Autonomous University of Mexico (UNAM) Department of Metallurgical Engineering, Faculty of Chemistry, National
Mexico City, Mexico Autonomous University of Mexico (UNAM)
[email protected] México City, Mexico
[email protected]
Esteban Miguel García-Ochoa
Corrosion Research Center, Autonomous University of Campeche *Author for correspondence
Campeche, Mexico
[email protected]
Historical silver artifacts made with silver-copper alloys in silver artifacts, silver 0.800, was monitored for
alloys tend to tarnish—or corrode—in environments 48 hours by EN and under controlled atmospheres. To
that contain sulfides, causing changes in aesthetic simulate the controlled atmospheres, a desiccator and
appearance and sometimes losses. This represents a standard solutions were used to generate RH of 40 and
problem in their conservation and restoration, since to 90%. Electrochemical atmospheric corrosion monitors
perform an intervention, the corrosion products and (EACM) made of silver plates were used as an electro-
their corrosion mechanisms must be identified. Electro- chemical cell in an arrangement of three nominally
chemical techniques that have proven their effectiveness identical electrodes. The surfaces of the EACMs were
are required to safeguard the properties, aesthetics, etched and then contaminated with a 10% v/v solution
and original surfaces of these artifacts. Electrochem- of (NH4)2S to form the corrosion or tarnishing layer. To
ical noise (EN) is a nondestructive and nonintrusive analyze the time series, a trend removal was performed
technique that has been widely used to understand and the data were interpreted by direct examination,
the interfacial behavior of electrochemical systems, statistical analysis, and recurrence of EN. Finally, direct
as well as to study real surfaces in situ and corrosion analysis showed that the transients in most signals were
mechanisms. However, it is little known in the field of bidirectional with high frequency, irregular, and low
conservation and restoration. Therefore, the purpose amplitude oscillations. The increase in humidity favored
of this work was to use EN to monitor the influence greater noise resistance (Rn) and adsorption of H2O and
of relative humidity (RH) on the tarnishing process in OH− on the surface, which caused the growth of the
silver and to study it in order to identify the corrosion corrosion layer (Ag2S and Cu2S) and thus the formation
mechanism. The tarnishing of one of the most common of a pseudo-passive layer that protected the surface.
Investigating Precious Artefacts Using Non-destructive FIB

on FEG-SEM: A Case Study of 10th-Century Hollow Gilded
Copper Alloy Pendants (Gombiky) from Prague Castle
Estelle Ottenwelter* Luc Robbiola
Czech Academy of Sciences, Institute of Archaeology TRACES Laboratory (CNRS UMR5608), Université de Toulouse
Prague, Czech Republic Toulouse, France
Claudie Josse *Author for correspondence

Centre de Microcaractérisation Raimond Castaing (CNRS UAR 3623)
Toulouse, France
[email protected]
Focussed ion beam (FIB) mounted on a scanning elec- soft layers. The technique is therefore particularly
tron microscope with a field emission gun (FEG-SEM) useful for observing and characterising gilded artefacts.
offers new perspectives in studies of soft, multilayered After surface milling, detailed structural information
metal artefacts. In addition to conventional observa- can be obtained at the nanometre scale and used in
tion, the system allows cross-sections to be made in situ, further analyses. Applied here on hollow gilded copper
that is, within the SEM chamber, using the FIB to mill alloy pendants (known as gombiky) from Prague Castle,
the selected surface and thus obtain a local cross-section dated from the 10th century AD, FIB on FEG-SEM
up to 100 to 200 μm wide and 10 to 20 μm deep. This yielded new and accurate information regarding the
new method of investigation is of particular interest thickness, porosity, metallurgical structure and general
in the study of precious cultural heritage artefacts, for condition of these complex artefacts. The method aims
which sampling is often problematic. Unlike classical to provide new parameters for drawing comparisons
metallography, the accurate cross-sections obtained between fine products produced by highly skilled
with FIB on FEG-SEM avoid damage to multilay- jewellers and coarse imitations made by less-experi-
ered materials, as ionic ablation does not deform the enced craftsmen.
The Enemy Within: Dezincification and Cyclic Corrosion on

a Brass Patinated Equestrian Sculpture
Maria João Pereira* Sara Fragoso Rui Silva Alexandra P. Rodrigues

NOVA School of Science and LAQV–CHARM and CENIMAT/ CENIMAT/i3N, Department of Department of Conservation
Technology, NOVA University i3N, Department of Conservation Materials Science, NOVA School and Restoration, NOVA School of
Lisbon and Restoration, NOVA School of of Science and Technology, NOVA Science and Technology, NOVA
Caparica, Portugal Science and Technology, NOVA University Lisbon University Lisbon
[email protected] University Lisbon Caparica, Portugal Caparica, Portugal
Caparica, Portugal [email protected] [email protected]
In this work, a brass equestrian sculpture (Cu 62.2/ which resulted in an aggravated active corrosion process.
Zn 36.5/Sn 1.3 wt%), in a poor state of preservation due Further progression of this process risked causing struc-
to localized cyclic corrosion, was assessed with the aim tural instability that would jeopardize the handling and
of developing a strategy for its conservation. Considering appreciation of the object. It was thus necessary to design
the lack of information regarding the origin, the date a methodology of investigative cleaning, both mechan-
of production, and the history of this object, a stylis- ical and chemical, of the surface and a pre-stabilization
tic-based contextualization was carried out together protocol that followed the criteria of minimum inter-
with a literature review of dezincification processes in vention, so as not to compromise the surface decoration
chlorine-containing media. To prepare a proposal suited while also promoting the stabilization of the corrosion
to the conservation needs of the object, physical and layers using a corrosion inhibitor. A methodological
chemical methods of analysis were used, which provided proposal of controlled internal desalination to achieve
essential information regarding the sculpture’s constitu- chemical stability without affecting the decoration was
tion and composition, corrosion and damage problems, also elaborated. As the interior of the sculpture was
and their respective causes. The surface examination was inaccessible, a small hole was drilled in a low-visibility
performed by optical microscopy using a stereoscopic area to allow the extraction of a sample of the ceramic
magnifying glass; structural and microstructural analyses material used to make the hollow space during the casting
were carried out, respectively, using digital radiography process. Examination of the sample using SEM-EDS
and metallography, and compositional studies using and XRD showed that this material was the origin of
energy-dispersive X-ray fluorescence, scanning electron the NaCl contamination. For the desalination process, a
microscopy and energy-dispersive X-ray spectroscopy second hole was drilled near the first one, with thin tubes
(SEM-EDS), and X-ray diffraction (XRD). The results of placed in each one to enable the inflow and outflow of
the investigation showed that different parts of the brass deionized water and therefore the internal removal of the
sculpture were cast by sand-casting, subsequently joined salts. This approach also allowed the desalination process
by brazing and then patinated in brown. The presence to be monitored without compromising the decorative
of chloride inside the object was determined to be the patina of the object. The stable packaging of the object
main cause of dezincification and autocatalytic corrosion, was ensured as well.
Standardizing Acid Digestion Methodology to Determine

the Chloride Content of Ferrous Corrosion Samples
Kimberly Roche* Johanna Rivera Lisa Arslaner Stephanie Crette

Warren Lasch Conservation Center, Warren Lasch Conservation Center, Warren Lasch Conservation Center, Warren Lasch Conservation Center,
Clemson University Clemson University Clemson University Clemson University
North Charleston SC, USA North Charleston SC, USA North Charleston SC, USA North Charleston SC, USA
www.clemson.edu/wlcc www.clemson.edu/wlcc www.clemson.edu/wlcc www.clemson.edu/wlcc
Acid digestion is a critical step in determining desalina- slightly punctured polyethylene film); temperature
tion treatment progress and endpoints for conservation (20°C–22°C and 35°C–40°C); time (3, 7, 10, and 14 days);
practitioners. This method quantifies residual chloride and phase separation method (no filtration, Celite
in the corrosion layers of ferrous artifacts in contrast filtration aid with qualitative filter paper, Whatman 42
to indirect chloride concentrations extracted from quantitative filter paper, Whatman 542 quantitative filter
desalination solutions. The analytical protocol for this paper, and centrifugation). Trials were run using sodium
technique varies by institution and suffers from a lack of chloride (NaCl) and synthesized solid-state akaganeite
standardization. Sample controls and materials are rarely (b-FeOOH) as the analytes. Three sample replicates and
reported, making the protocols hard to reproduce with two procedural blanks were used to test each variable.
precision. Conservators often rely upon small quantities The results of the NaCl trials indicated that the covering
of corrosion powders to analyze chloride content, and method had a negligible effect (less than 0.3% difference)
any chloride contamination in the procedure will have on chloride readings. Heated NaCl solutions resulted in
large implications on data output and interpretation. greater chloride loss (4.5%) compared to unheated solu-
This study aims to standardize an analytical protocol tions. Nonetheless, heating was found to be necessary,
for determining chloride content in iron corrosion as heated samples of akaganeite resulted in 25% greater
samples through acid digestion, with a focus on elimi- chloride concentrations, indicating that akaganeite was
nating or significantly reducing chloride contamination not fully digested at room temperature after 7 days. The
during each stage. The protocol was divided into five chloride yield in the akaganeite samples was highest after
stages as follows: (1) sample collection; (2) digestion 3 days of heating and declined with increased heating
in concentrated nitric acid; (3) neutralization using time. The greatest source of chloride contamination and
sodium hydroxide; (4) phase separation/filtration; and yield inconsistencies occurred at the separation stage of
(5) chloride analysis with ion chromatography. After a the method. An analytical protocol for acid digestion
literature review determined the most common protocol is proposed that produces the highest theoretical yield
variations, the variables tested in this study were: the of chloride, the least contamination in the blanks, and
covering method (watch glass over a flask sealed with greater precision among the analytes.
Treatment of Two Fanghu Vessels (1st Century BC):

Conservation Issues of Archaeological Painted Bronze
Cristina Scarrone* Marco Demmelbauer

Independent conservator Centro Conservazione e Restauro “La Venaria Reale” and SUSCOR, University of
Genoa, Italy Turin
[email protected] Venaria Reale, Italy
[email protected]
Dominique Maria Scalarone
Department of Chemistry, University of Turin Tommaso Poli
Turin, Italy Department of Chemistry, University of Turin
[email protected] Turin, Italy
[email protected]
Franca Varallo
Department of Historical Studies, University of Turin *Author for correspondence
Turin, Italy
[email protected]
The conservation of two Chinese fanghu vessels in cleaning of the metal surface was a compromise between
painted bronze (1st century BC) from the MAO Museum conservation and aesthetic requirements. The removal of
of Oriental Art in Turin raised issues concerning archae- unstable corrosion products revealed the metal surface
ological bronze painted with tempera but showing a loss or a layer of cuprite, the colour of which was, unfor-
of cohesion and the extensive presence of corrosion prod- tunately, very close to that of the paint film. This layer
ucts. The opaque paint layer, the type of pigments and the was treated mechanically using an ultrasonic scaler to
corrosion products prohibited the use of the materials and lower the tone and matt the surface, so that it would not
methods normally adopted for archaeological artefacts. compete with the decoration. Cleaning was alternated
This study focussed on three fundamental phases of the with retouching measures. Traces of the tempera detach-
conservation intervention: pre-consolidation, cleaning ments were visible owing to the surrounding corrosion
and retouching. For the pre-consolidation of the paint products, which had stuck to the surface forming an
film, a study was carried out to identify a consolidating outline of the area. Therefore, retouching was carried
agent that could be applied to metal while leaving the out before the corrosion products were removed. After
characteristics of the film unchanged, in particular its several tests, it became clear that traditional retouching
opacity and colour, and at the same time provide adhe- techniques, such as veiling or dotting, were not suitable
sion and cohesion. The materials investigated were the as they enhanced the lack of homogeneity of the artefacts.
acrylic resin Paraloid B-72, cellulose ethers (Klucel E Ultimately, an ultraviolet fluorescent pigment (Kremer
and Klucel G), the aliphatic resin Regalrez 1126 and the Pigmente #56200), dispersed in a mimetic colour made
urea-aldehyde resin Laropal A81. All were subjected to a with mineral pigments, was used together with a 5%
mechanical strength test, optical and autoptic evaluations solution of Paraloid B-67 in ligroin. The fluorescent
and photochemical ageing. The best-performing solution pigment is imperceptible at visible light such that it can
was Paraloid B-72 at 5% in methyl ethyl ketone, with be used to create tone-in-tone additions that respect the
a dosage of 2 drops on a surface of 1 cm2. The level of recognition criteria of the intervention.
Investigating Three Types of Dry Cell Battery

Deterioration
Meredith Sweeney
Smithsonian National Air and Space Museum
Washington, DC, USA
[email protected]
Electric cells and batteries are found in the collec- dry cell batteries within the NASM collection where
tions of nearly every history museum, but there is no these common condition issues occur. The case study
universal guidance on how to approach these poten- objects include zinc-carbon WW2 American and Japa-
tially hazardous objects. A 2021 survey of heritage nese batteries and a nickel-cadmium battery used in
professionals indicated that half of the respondents the Apollo 11 cameras. Each case study utilizes various
immediately discard batteries when they enter collec- analytical techniques. X-radiography is used to deter-
tions and/or when they show signs of deterioration. mine corrosion location, the loss of internal material,
Following an inspection of 32 batteries in the Smithso- and the source of any structural deformation. Ultraviolet
nian National Air and Space Museum (NASM) collec- photography, portable X-ray fluorescence, and Fourier
tion, three recurring condition issues became apparent: transform infrared spectroscopy with attenuated total
corrosion occurring within the cell, corrosion occurring reflectance (FTIR-ATR) aid in characterizing the signs
on the external sides of the cell, and electrolyte leakage. of degradation and differentiating between corrosion
Determining the location of corrosion indicates whether products and leaking electrolyte. This is supported by a
its development was influenced by the electrolyte or by discussion of dry cell battery construction and common
external environmental factors. Alkaline electrolyte sources of deterioration. Treatment rationale is briefly
leakage can mimic the appearance of zinc corrosion explored with the aim of informing collection stew-
(a common material found within dry cell batteries), ardship practices. The goal of the poster is to inform
and differentiating between the two is an imperative conservators on how to characterize and analyse some
step in determining stability and designing treatment forms of battery deterioration and to illustrate that signs
strategies. This poster discusses three case studies of of degradation do not necessitate disposal.
The Complexity of Developing a Restoration Plan for an

18th-Century Cast Iron Fountain
Júlia Tauber
Landesamt für Denkmalpflege im Regierungspräsidium Stuttgart
Baden Württemberg, Germany
[email protected]
Stuttgart’s beautiful cast iron fountain known as the and structural engineers is required to complete this
Marktbrunnen, or Thouret Brunnen, is situated in the complex project. Prior to the restoration, the fountain was
city’s Market Square (Marktplatz). The basin consists of digitised using structure-from-motion photogrammetry.
16 relief plates that vary in their design. The plates were Preliminary investigations included an analysis of the
cast in 1714, with the middle column added in 1804. coating using an optical microscope, energy-dispersive
However, as the Marktplatz is currently under renovation X-ray spectroscopy and Fourier-transform infrared
to raise its entire surface to the same level, the fountain spectroscopy. Trials for removing the current coating are
must be raised 45 cm higher than its current position. complete and alternative coating materials have under-
This manoeuvre has necessitated its restoration. The gone adhesive tensile strength tests to determine the most
fountain has already been moved five times in its history, suitable coating methodology. X-ray imaging was used
including its display in three different squares. Yet moving to obtain evidence of graphitic corrosion of the plates.
and restoring the fountain once more is a complex As the relief plates forming the basin are buried 1.5 m
process, as the latest move, in the 1970s, compromised underground, their restoration in situ was enabled by
its exterior. Specifically, the fountain was cleaned by air excavating the installation such that it could be raised in
abrasion and coated with a zinc primer, which caused the one piece using hydraulic jacks. This poster describes the
removal of the cast skin, thus exposing the grey cast iron decision-making process employed in the conservation,
plates and increasing the risk of corrosion. Construction which included a methodical approach to investigations,
of the square is ongoing and is due to be completed by careful planning and collaboration between disciplines.
the end of 2022. This has created pressure to rapidly find Successful restoration of the Marktbrunnen will retain
solutions for the fountain’s restoration. Collaboration and enhance the fountain’s natural beauty, allowing its
between conservators, analytical scientists, architects further appreciation for many years.
Testing Gap Fillers for Archaeological Metal
Simeon Tsvetkov
AOC Archaeology Group
Loanhead, UK
[email protected]
The headplate of a 4th-century fibula was excavated from B100 SY100, Butvar B-98, and Aquazol 200) were selected
the island of Funen in Southern Denmark. It is an inter- for the experiments. Each one was mixed with different
esting find because of its rarity in the local context, and solvents (ethanol, acetone, and toluene), as well as in
because its design and characteristics suggest a foreign mixtures of them, after which microballoons were added
influence or origin. It has a gold leaf decoration with to create a putty-like gap filler. Two concentrations of each
granulated design, as well as two semi-spherical blue glass adhesive/solvent pairing – a low one (20%) and a high one
beads and silver wire around the edges. The object was (40%) – were conducted in order to see how the choice
severely damaged during burial and some of its structural of solvent and concentration would affect the properties
and decorative elements had broken or fallen off. Whilst of the filler. The two experiments aimed to evaluate the
a lot of the fragments could simply be adhered back into amount of shrinkage, shape retention, adhesive strength,
their place, the reattachment of the two glass beads was and working properties of each sample. Several general
more complex due to the fact that the original slots are trends were observed. The 20% concentration samples of
partially destroyed and the back plate is fragmented. It each adhesive shrank less, retained their shape better, and
was decided that the best way to reattach the beads would were easier to work with, but this was counterbalanced by
be to reconstruct the holes and back plate using a gap the loss of adhesive strength. Paraloid B-72 was the only
filler. Since there is no general consensus in the conser- adhesive that managed to preserve sufficient strength in
vation literature as to the best material for gap filling of low concentrations and was therefore deemed the most
archaeological metal, two experiments were conducted suitable choice for our case. A 20% solution of Paraloid
to determine the most suitable adhesive/solvent combi- B-72 in acetone mixed with microballoons was used to
nation (from the ones readily available in the lab) to use reconstruct the destroyed parts of the fibula and reattach
on the fibula. Four adhesives (Paraloid B-72, Vinnapas the glass beads with good results.
The ‘Hallmarks on (Dutch) Silver’ Project:

Where Are We Now?
Ellen van Bork* Robert Erdmann Tamar Davidowitz

Rijksmuseum Rijksmuseum Rijksmuseum
Amsterdam, The Netherlands Amsterdam, The Netherlands Amsterdam, The Netherlands
www.rijksmuseum.nl www.rijksmuseum.nl www.rijksmuseum.nl
The available information regarding (Dutch) hallmarks users can upload their own images of hallmarks, was
on silver is limited to printed books containing short designed. Hallmarks found on those objects can then be
descriptions or drawings of known marks, an approach linked to known hallmarks, including their metadata,
requiring fundamental contextual knowledge from thus aiding in the attribution and dating of silver works
the user and access to specific publications. To this of art. The aims of the second phase of the project,
end, funding was sought to create an automated image launched on 1 November 2021, are: to complete the
recognition system, easily accessible to users, that photography of all hallmarks in the Rijksmuseum’s
would provide a comprehensive and efficient solution. collection and to simultaneously characterise the
In January 2019, the ‘Hallmarks on Silver’ project was objects’ alloy composition by gathering X-ray fluores-
launched. In its first phase, a ground-truth database cence (XRF) data (acquired in situ with a handheld
containing 483 hallmarks from the Rijksmuseum collec- XRF analyser, Olympus Delta). The XRF results will be
tion was used to realise an automated methodology entered into the Rijksmuseum’s silver alloy composition
for identifying hallmarks from digital photographs. A database, thereby placing them in a larger context and
convolutional neural network was trained to classify enabling comparative analyses. This phase will result in
images from the database and to report the known complete hallmark and composition datasets, a singular
identities of those hallmarks. In the inference phase, accomplishment for such a vast collection. Subsequent
new marks were introduced and the network gave an phases of the project will focus on the continuation and
estimate of the possible identities of the hallmark, along initiation of collaborations with other museums and
with estimates of the certainty of those predictions. institutes, in an effort to expand the datasets and link
The proof-of-concept network had an overall top-1 them to the relevant metadata. The final aim is to make
accuracy exceeding 99%. To make this tool accessible the online portal, and the ever-expanding information
to the public, a prototype of an online portal, in which it contains, available to the public.
Metallotheek: A Collaborative Exploration of Bronze

Patination
Heleen van Santen* Laura van Santen

Museumssenteret i Hordaland Royal Academy of Art, The Hague
Salhus, Norway The Hague, The Netherlands
www.muho.no www.kabk.nl
The concept of ‘patina’ in metal conservation is essen- and design communication; the foundry worker’s prac-
tial but often remains enigmatic, and the line between tical approach to problem-solving, material knowledge
corrosion and patina can be difficult to discern. Pati- and experience; and the metals conservator’s chemical
nation as a finishing technique can be more concretely knowledge of metals and corrosion, and experience
defined as a practice involving recipes and techniques with setting-up and documenting experiments. The
to attain a coloured and textured corrosion product on work parameters, i.e. choice of patination recipes and
metal. Practical experience with patination can therefore patination techniques, were decided as a team. This
offer insights into the wider concept of patina, allowing resulted in 81 sample patinas and a wealth of experience.
for clearer definition of the term and a better reading To communicate this visually interesting experiment to
of patinas on objects. The many variables affecting future makers and the public, a permanent exhibition
patina formation prevent an exact documentation of of the project was built at the art foundry and a digital
the patination technique. The conservator’s instinct publication created, calling attention to the fascinating
to work systematically and repeatably interferes with overlap between materials research and art in the hope
an organic approach to learning the craft. Under- of inspiring others to build on the research. As a metals
standing patination and how the technique is applied conservator, the project provided practical knowledge
requires the technical expertise of foundry workers, about a finishing technique that is essential to the field,
craftspeople and artists. Equally important is under- allowing an eye for patinas to be cultivated and a clearer
standing their priorities, decision-making processes and vocabulary for discussing them. This knowledge has
ways of working. The Metallotheek project sought to improved the authors’ understanding of the objects
combine the expertise of a designer, a foundry worker they work with and has informed their conservation
and a metals conservator to expand their knowledge practice. For more information about this project, please
of patination. This symbiosis brought together the visit the following website: www.makeeindhoven.nl/
designer’s eye for aesthetics and focus on applicability knowledgebase/metal/metallotheek/
ICOM-CC | METAL 2022 | HELSINKI, FINLAND 365
Author index
A de Bruyn-Ouboter, V. . . . . . . . . . . . . 331 Han, K.-S. (Tracy) . . . . . . . . . . . . . . . . 343
Abed-Esfahani, A. . . . . . . . . . . . . . . . . 352 Degrigny, C. . . . . . . . . . . . . 87, 160, 248 Harsányi, I. . . . . . . . . . . . . . . . . . . . . . . . . . . 203
Abu-Baker, A.N. .. . . . . . . . . . . . . . . . . . 221 del Hoyo-Meléndez, J. . . . . . . . . . . . 345 Heinzel, M. .. . . . . . . . . . . . . . . . . . . . . . . . . 184
Aliyari, A . . . . . . . . . . . . . . . . . . . . . . . . . . . . 231 Delano, M. . . . . . . . . . . . . . . . . . . . . . . . . . . 308 Hermerén, K. . . . . . . . . . . . . . . . . . . . . . . . 176
Amarger, A. . . . . . . . . . . . . . . . . . . . . . . . . . 328 Demmelbauer, M. .. . . . . . . . . . . . . . . . 359 Hoffman, W. . . . . . . . . . . . . . . . . . . . . . . . . 281
Andersen, L.B. . . . . . . . . . . . . . . . . 16, 338 Deshayes, C. . . . . . . . . . . . . . . . . . . . . . . . . . 78 Hokoi, S. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
Anusha Kasthuri, K.A. . . . . . . . . . . 290 Dillmann, P. .. . . . . . . . . . . . . . . . . . . 16, 338 Horelick, L. .. . . . . . . . . . . . . . . . . . . . . . . . . . 56
Argyropoulos, V. . . . . . . . . . . . . . . . . . . 339 Duarte, M. .. . . . . . . . . . . . . . . . . . . . . . . . . . 337 Hudson, R. . . . . . . . . . . . . . . . . . . . . . . . . . . 326
Arponen, A. . . . . . . . . . . . . . . . . . . . . . . . . 324 Dupuy, V. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25 Hyppönen, A. . . . . . . . . . . . . . . . . . . . . . . 324
Arslaner, L. . . . . . . . . . . . . . . . . . . . . . . . . . . 358
E J
B Echerd, J. .. . . . . . . . . . . . . . . . . . . . . . . . . . . . 114 James, S. .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
Bahadori, S. . . . . . . . . . . . . . . . . . . . . . . . . . 231 Johnson, S. . . . . . . . . . . . . . . . . . . . . . . . . . . 261
Echinard, J. . . . . . . . . . . . . . . . . . . . . . . . . . . . 78
Bakhshandehfard, H.R. . . . . . . . . . 352 Joseph, E. . . . . . . . . . . . . . . . . . . 25, 35, 168
Eggert, G. . . . . . . . . . . . . . . . . . . . . . 194, 335
Balbo, A. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78
Emmerson, N.J. . . . . . . . . 94, 203, 240 Josse, C. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 356
Beck, L. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 325
Erbe, A. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 331 Jouet, J. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 328
Berlewen, K. . . . . . . . . . . . . . . . . . . . . . . . . 326
Erdmann, R. . . . . . . . . . . . . . . . . . . . . . . . . 363
Bernardi, E. . . . . . . . . . . . . . 78, 327, 334 K
Beyth, I.L. . . . . . . . . . . . . . . . . . . . . . . . . . . . 308 F Kaipanen, K. . . . . . . . . . . . . . . . . . . . . . . . . 324
Bonora Andujar, I. . . . . . . . . . . . . . . . . 328 Kasztovszky, Z. .. . . . . . . . . . . . . . . . . . . . 203
Fair, L. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 340
Bordalo, R. . . . . . . . . . . . . . . . . . . . . . . . . . . 337 Kemper, D. . . . . . . . . . . . . . . . . . . . . . . . . . . 184
Fajfar, P. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 353
Borin, M. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 176 Khalil, L.A. . . . . . . . . . . . . . . . . . . . . . . . . . . 221
Farrell, E. . . . . . . . . . . . . . . . 114, 281, 334
Bottaini, C. . . . . . . . . . . . . . . . . . . . . . . . . . . 337
Fernando, A. .. . . . . . . . . . . . . . . . . . . . . . . 290 Kilpeläinen, P. . . . . . . . . . . . . . . . . . . . . . . 324
Brambilla, L. . . . . . . . . . . . 160, 168, 329
Figueiredo, C. . . . . . . . . . . . . . . . . . . . . . . 337 King, L. . . . . . . . . . . . . . . . . . . . . . . . . 281, 344
Branche, B. . . . . . . . . . . . . . . . . . . . . . . . . . . 328
Fišer, J. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78 Kis, Z. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 203
Brechbühl, S. . . . . . . . . . . . . . . . . . . . . . . . 330
Florescu, M. .. . . . . . . . . . . . . . . . . . . . . . . . 176 Kluge, E. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 184
Brunet, M. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78
Fragoso, S. . . . . . . . . . . . . . . . . . . . . . . . . . . . 357 Kohdzuma, Y. . . . . . . . . . . . . . . . . . . . . . . . 40
C Frydendahl, J. . . . . . . . . . . . . . . . . . 16, 338 Krupska, P. . . . . . . . . . . . . . . . . . . . . . . . . . . 345
Cagnini, A. . . . . . . . . . . . . . . . . . . . . . . . . . . 248 Kuhn-Wawrzinek, C.F. . . . . . . . . . . 194
Camarneiro, N. . . . . . . . . . . . . . . . . . . . . 337 G Kwiatkowska, J. . . . . . . . . . . . . . . . . . . . . 350
Campos, B. . . . . . . . . . . . . . . . . . . . . . . . . . . . 50 Ganesan, S. . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
García Abajo, Á.E. . . . . . . . . . . . . . . . . 146 L
Cano, E. . . . . . . . . . . . . . . . . . 153, 332, 351
García-Ochoa, E.M. . . . . . . . . . . . . . . 355 Labroche, S. . . . . . . . . . . . . . . . . . . . . . . . . . 342
Canosa, E. . . . . . . . . . . . . . . . . . . . . . . . . . . . 176
Giannoulaki, M. . . . . . . . . . . . . . . . . . . . 339 Labut, M. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 345
Cardoso, M. .. . . . . . . . . . . . . . . . . . . . . . . . 333
Carré, C. . . . . . . . . . . . . . . . . . . . . . . . . 16, 338 Golfomitsou, S. . . . . . . . . . . . . . . . . . . . . 176 Lambert, E. .. . . . . . . . . . . . . . . . . . . . . . . . . 346
Carvalho, C. . . . . . . . . . . . . . . . . . . . . . . . . 109 Gonçalves, M. . . . . . . . . . . . . . . . . . . . . . . . 50 Lang, R. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 272
Chalançon, B. . . . . . . . . . . . . . . . . . 87, 329 Goryl, M. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 345 Lankester, P. .. . . . . . . . . . . . . . . . . . . . . . . . 261
Cheel, V. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 354 Grahek, L. . . . . . . . . . . . . . . . . . . . . . . . . . . . 353 Lazar, T. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 353
Chiavari, C. . . . . . . . . . . . . . 78, 327, 334 Granget, E. . . . . . . . . . . . . . . . . . . . . . . . . . . . 87 Le Goïc, G. . . . . . . . . . . . . . . . . . . . . . . . . . . 160
Chow, M. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 348 Grayburn, R. .. . . . . . . . . . . . . . . . . . . . . . . 340 Leal, J. .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 332
Contreras Vargas, J. .. . . . . . . . . . . . . . 146 Grima, M. . . . . . . . . . . . . . . . . . . . . . . . . . . . 341 Leroux, M. . . . . . . . . . . . . . . . . . . . . . . . . . . 346
Costantini, F. . . . . . . . . . . . . . . . . . . . . . . . 334 Guilminot, E. . . . . . . . . . . . . . 25, 78, 342 Letardi, P. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 347
Crette, S. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 358 Gustad, E.F. . . . . . . . . . . . . . . . . . . . . . . . . . 331 Lewinsky, J. . . . . . . . . . . . . . . . . . . . . . . . . . 308
Cuvillier, L. .. . . . . . . . . . . . . . . . . . . . . . . . . . 25 Gänsicke, S. . . . . . . . . . . . . . . . . . . . . . . . . . 272 Lewis, M. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 203
Lira Pacheco, D. . . . . . . . . . . . . . . . . . . . 146
D H Llop I Bayo, F. . . . . . . . . . . . . . . . . . . . . . . 146
Davidowitz, T. .. . . . . . . . . . . . . . . . . . . . . 363 Haines, L. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 281 Llop Álvaro, F. .. . . . . . . . . . . . . . . . . . . . . 146
366 ICOM-CC | METAL 2022 | HELSINKI, FINLAND
M Ricotta, N. . . . . . . . . . . . . . . . . . . . . . . . . . . . 248 Velino, C. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 327

MacLeod, I.D. . . . . . . . . . . . . . . . . . . . . . . . 71 Rivera, J. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 358 Vieira, E. . . . . . . . . . . . . . . . . . 50, 109, 337
Magaña Nieto, F. . . . . . . . . . . . . . . . . . . 146 Robbiola, L. . . . . . . . . . . . . . . . . . . . . 78, 356 Vyhlídal, T. . . . . . . . . . . . . . . . . . . . . . . . . . . . 78
Mansouri, A. . . . . . . . . . . . . . . . . . . . . . . . 160 Roche, K. . . . . . . . . . . . . . . . . . . . . . 240, 358
Martini, C. . . . . . . . . . . . . . . 78, 327, 334 Roda-Buch, A. . . . . . . . . . . . . . . . . . . . . . 329 W
Matthiesen, H. . . . . . . . . . . . . . . . . 16, 338 Rodrigues, A.P. .. . . . . . . . . . . . . . . . . . . 357 Wakiya, S. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
McArthur, G. . . . . . . . . . . . . . . . . . . . . . . . 348 Rodriguez, A.E. . . . . . . . . . . . . . . . . . . . . . 65 Watkinson, D. . . . . . . . . . . 94, 203, 240
McGath, M.K. . . . . . . . . . . . . . . . 114, 281 Rodríguez-Gómez, F.J. . . . . . . . . . . 355 Wentland, E. . . . . . . . . . . . . . . . . . . . . . . . . 123
Meehan, P. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94 Ruckmal Athukorala, D. . . . . . . . . 290 Wiinblad, T. .. . . . . . . . . . . . . . . . . . . 16, 338
Meier, J. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103 Russo, S. .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 168 Wilcox, K. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
Menart, E. . . . . . . . . . . . . . . . . . . . . . . . . . . . 349
Messager, C. . . . . . . . . . . . . . . . . . . . . . . . . 325 S X
Mischler, S. . . . . . . . . . . . . . . . . . . . . . . . . . . 329 Samide, M.J. . . . . . . . . . . . . . . . . . . . . . . . . 335 Xie, H. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
Mistewicz, A. . . . . . . . . . . . . . . . . . . . . . . . 350 Sangouard, E. .. . . . . . . . . . . . . . . . . . . . . . 281
Scalarone, D.M. .. . . . . . . . . . . . . . . . . . . 359 Y
Molina, M.T. . . . . . . . . . . . 153, 332, 351
Scarrone, C. . . . . . . . . . . . . . . . . . . . . . . . . . 359 Yanagida, A. . . . . . . . . . . . . . . . . . . . . . . . . . 40
Molineri, A. . . . . . . . . . . . . . . . . . . . . . . . . . 328
Schillinger, B. . . . . . . . . . . . . . . . . . . . . . . . 184 Yasui, H. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
Monachon, M. . . . . . . . . . . . . . . . . . . . . . . 35
Monari, G. .. . . . . . . . . . . . . . . . . . . . . . . . . . 347 Schleid, T. . . . . . . . . . . . . . . . . . . . . . . . . . . . 194
Seifert, J.H. . . . . . . . . . . . . . . . . . . . . . . . . . . 240 Z
Monticelli, C. . . . . . . . . . . . . . . . . . . . . . . . . 78
Seri, H. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 308 Žužek, B. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 353
Moreira, P.R. . . . . . . . . . . . . . . . . . . . 50, 109
Mortazavi, M. . . . . . . . . . . . . . . . . . . . . . . 352 Siatou, A. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 160
Silva, I. .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 337
N Silva, N. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109
Neff, D. . . . . . . . . . . . . . . . . . . . . . . . . . . 16, 338 Silva, R. .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 357
Nemeček, N. . . . . . . . . . . . . . . . . . . . . . . . . 353 Singh, V. .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 140
Nečemer, M. . . . . . . . . . . . . . . . . . . . . . . . . 353 Šmit, Z. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 349
Northover, P. .. . . . . . . . . . . . . . . . 212, 354 Smith, G.D. . . . . . . . . . . . . . . . . . . . . . . . . . 335
Nurit, M. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 160 Stefańczyk, K. . . . . . . . . . . . . . . . . . . . . . . 345
Nygårds, E. . . . . . . . . . . . . . . . . . . . . . . . . . . 176 Stieghorst, C. . . . . . . . . . . . . . . . . . . . . . . . 184
Stępień, A. . . . . . . . . . . . . . . . . . . . . . . . . . . . 345
O Sully, D. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 132
Obarzanowski, M. . . . . . . . . . . . . . . . . 345 Sweeney, M. .. . . . . . . . . . . . . . . . . . . 56, 360
Ogura, D. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
Ortíz-Corona, J. . . . . . . . . . . . . . . . . . . . 355 T
Ottenwelter, E. . . . . . . . . . . . . . . . . . . . . . 356 Takatori, N. . . . . . . . . . . . . . . . . . . . . . . . . . . 40
Oudbashi, O. . . . . . . . . . . . . . . . . . . . . . . . 231 Tauber, J. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 361
Thickett, D. . . . . . . . . . . . . . . . . . . 253, 261
P Thomas, J.-B. . . . . . . . . . . . . . . . . . . . . . . . 168
Panagiotopoulou, C. . . . . . . . . . . . . . 339 Thunberg, J.C. . . . . . . . . . . . . . . 203, 240
Passaretti, A. . . . . . . . . . . . . . . . . . . . . . . . . . 25 Timoncini, A. . . . . . . . . . . . . . . . . . . . . . . 334
Pelé-Meziani, C. . . . . . . . . . . . . . . . . . . . 342 Tréhu, E. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 132
Pereira, M.J. . . . . . . . . . . . . . . . . . . . . . . . . . 357 Tsvetkov, S. . . . . . . . . . . . . . . . . . . . . . . . . . . 362
Pienimäki, A. . . . . . . . . . . . . . . . . . . . . . . . 317 Tzimopoulou, S. . . . . . . . . . . . . . . . . . . . 339
Poli, T. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 359 Tóth, E. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 336
Pullano, M. .. . . . . . . . . . . . . . . . . . . . . . . . . 176
V
R van Bork, E. . . . . . . . . . . . . . . . . . . . . . . . . . 363
Raimon, A. . . . . . . . . . . . . . . . . . . . . . . . . . . . 25 van Santen, H. .. . . . . . . . . . . . . . . . . . . . . 364
Raja, S. .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 300 van Santen, L. . . . . . . . . . . . . . . . . . . . . . . 364
Ramírez Barat, B. .. . . . 153, 332, 351 Varallo, F. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 359
Rathore, S. . . . . . . . . . . . . . . . . . . . . . . . . . . . 300 Vassura, I. . . . . . . . . . . . . . . . . . . . . . . . . . . . 327
ICOM-CC | METAL 2022 | HELSINKI, FINLAND 367
Keyword index
A chalcocite.. . . . . . . . . . . . . . . . . . . . . . . . . . . 194 F
adsorptive inhibitors.. . . . . . . . . . . . 123 characterization. . . . . . . . . . . . . . . . . . . 203 feature maps. . . . . . . . . . . . . . . . . . . . . . . . 160
aerobic. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71 chitosan. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109 Fourier transform infrared
aerospace. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56 chloride. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 240 (FTIR) spectroscopy. . . . . . . . . . . . 50
aircraft heritage.. . . . . . . . . . . . . . . . . . . . . 78 freeware analytical tool. . . . . . . . . . . 87
coating. . . . . . . . . . . . . . . . . . . . . . . . 103, 109
air quality monitoring. . . . . . . . . . . 176
color. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 153 G
akaganeite. . . . . . . . . . . . . . . . . . . . . . . . . . . 240
colorimetry. . . . . . . . . . . . . . . . . . . . . . . . . . . 94 gilded silver.. . . . . . . . . . . . . . . . . . . . . . . . 272
Al cladding. . . . . . . . . . . . . . . . . . . . . . . . . . . 78
condition assessment. . . . . . . . . . . . 160 gloss.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94
alloy identification. . . . . . . . . . . . . . . . . 87
conservation.. . . . . . . . . . . 71, 203, 308 graphitized. . . . . . . . . . . . . . . . . . . . . . . . . . 114
aluminium alloys. . . . . . . . . . . . . . . . . . . 78
conservation laboratory. . . . . . . . 281 green chemistry. . . . . . . . . . . . . . . . . . . . . 25
aluminum. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
anaerobic. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71 consolidation.. . . . . . . . . . . . . . . . . . . . . . 114 green conservation. . . . . . . . . . . . . . . . . 35
analysis.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 248 contemporary art. . . . . . . . . . . . . . . . . 176
conversion coating. . . . . . . . . . . . . . . . . 65 H
anilite. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 194
heterogeneous tarnish. . . . . . . . . . . 248
antioxidant.. . . . . . . . . . . . . . . . . . . . . . . . . . . 50 cooling system. . . . . . . . . . . . . . . . . . . . . . . 87
Hildesheim. . . . . . . . . . . . . . . . . . . . . . . . . 184
archaeological.. . . . . . . . . . . . . . . . . . . . . 240 copper alloys. . . . . . . . . . 132, 221, 253
historiographic metal
archaeological copper alloys.. . 231 copper phosphates.. . . . . . . . . . . . . . . 221
collections. . . . . . . . . . . . . . . . . . . . . . . . 308
archaeological iron. . . . . . . . . . 16, 317 copper sulfide. . . . . . . . . . . . . . . . . . . . . . 194
hydrogels. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
archaeology. . . . . . . . . . . . . . . . . . . . . . . . . . . 35 corrosion. . . . . 25, 56, 71, 103, 153,
armor. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 140 176, 221, 231, 240 I
artificial coupon.. . . . . . . . . . . . . . . . . . 248 corrosion inhibitors. . . . . . . . . . . . . . 132 image analysis.. . . . . . . . . . . . . . . . . . . . . 153
atmospheric corrosion. . . . . . . . . . . . 40 corrosion monitoring.. . . . . . . . . . . 203 imaging. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 168
automaton.. . . . . . . . . . . . . . . . . . . . . . . . . . 272 covellite. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 194 Indonesia. . . . . . . . . . . . . . . . . . . . . . . . . . . . 103
craftswomen.. . . . . . . . . . . . . . . . . . . . . . . 140 industrial archaeology.. . . . . . . . . . 281
B industrial heritage. . . . . . . . . . . . . . . . 123
baghtar. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 140 D information gathering. . . . . . . . . . . 146
bell frames. . . . . . . . . . . . . . . . . . . . . . . . . . 146 dagger. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103 in situ preservation. . . . . . . . . . . . . . . . . 40
bells. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 146 decision-making. . . . . . . . . . . . . . . . . . 203 in situ treatment.. . . . . . . . . . . . . . . . . . . . 16
benzotriazole (BTA). . . . . . . . . . . . . 132 interpretation. . . . . . . . . . . . . . . . . . . . . . 212
deferoxamine B.. . . . . . . . . . . . . . . . . . . . . 25
biocleaning. . . . . . . . . . . . . . . . . . . . . . . . . . . 25 inventory. . . . . . . . . . . . . . . . . . . . . . . . . . . . 146
dehumidifier. . . . . . . . . . . . . . . . . . . . . . . 261
biomineralization. . . . . . . . . . . . . . . . . . . 35 iron. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 240, 253
desalination.. . . . . . . . . . . . . . . . . . . 16, 317
black spots. . . . . . . . . . . . . . . . . . . . . . . . . . 194 iron artifact. . . . . . . . . . . . . . . . . . . . . . . . . . . 40
dewatering fluids. . . . . . . . . . . . . . . . . 123
bodhisattva. . . . . . . . . . . . . . . . . . . . . . . . . 290 iron corrosion.. . . . . . . . . . . . . . . . . . . . . . . 35
DiscoveryMat. . . . . . . . . . . . . . . . . . . . . . . . 87
bronze. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 290 iron objects. . . . . . . . . . . . . . . . . . . . . . . . . . . 25
dissolved oxygen. . . . . . . . . . . . . . . . . . 317
bronze artifact. . . . . . . . . . . . . . . . . . . . . . . 40 iron-reducing bacteria. . . . . . . . . . . . 16
bronze disease. . . . . . . 132, 203, 231 documentation. . . . . . . . . . . . . . . . . . . . 146
iron uptake.. . . . . . . . . . . . . . . . . . . . . . . . . . . 35
Buddha.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 290
E
buried environment.. . . . . . . . . . . . . . . 40 J
efflorescence. . . . . . . . . . . . . . . . . . . . . . . 194
Joachim Fries. . . . . . . . . . . . . . . . . . . . . . 272
C electrochemical impedance
Judaica. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 308
calcium sulfonate. . . . . . . . . . . . . . . . . 123 spectroscopy (EIS). . . . . . . . . . . . . . . 94
capacity building. . . . . . . . . . . . . . . . . . 300 elemental sulfur. . . . . . . . . . . . . . . . . . . 194 K
car. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87 émail champlevé. . . . . . . . . . . . . . . . . . 184 karigar.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 140
cast. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 290 epidemiology.. . . . . . . . . . . . . . . . . . . . . . 253 keris. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103
cast iron.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114 etching. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103 Khirbet Yajuz.. . . . . . . . . . . . . . . . . . . . . . 221
368 ICOM-CC | METAL 2022 | HELSINKI, FINLAND
kris. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103 pollution. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 261 W

preventive conservation. 153, 231 watchmaker. . . . . . . . . . . . . . . . . . . . . . . . 272
L prompt gamma-ray neutron wind-up mechanism. . . . . . . . . . . . . 272
L-cysteine. . . . . . . . . . . . . . . . . . . . . . . . . . . 132 activation imaging (PGAI).. 184 wrought iron. . . . . . . . . . . . . . . . . . . . . . . . . 94
lead phosphates. . . . . . . . . . . . . . . . . . . 221 protective coatings. . . . . . . . . . . . 78, 94
linear sweep voltammetry.. . . . . 248 public art. . . . . . . . . . . . . . . . . . . . . . . . . . . . 109 X
long-term project. . . . . . . . . . . . . . . . . 281 pull-off testing. . . . . . . . . . . . . . . . . . . . . . . 94 X-ray fluorescence (XRF).. 176, 300
lubricant. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
R Z
M radiography.. . . . . . . . . . . . . . . . . . . . . . . . 290 zinc phosphates. . . . . . . . . . . . . . . . . . . 221
macrostructure. . . . . . . . . . . . . . . . . . . . 212 reburial.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
magnesium. . . . . . . . . . . . . . . . . . . . . . . . . . . 56 reflectance transformation
magnesium corrosion. . . . . . . . . . . . . 65 imaging (RTI). . . . . . . . . . . . . . . . . . . 160
mail makers. . . . . . . . . . . . . . . . . . . . . . . . 140 reliquary. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 184
management.. . . . . . . . . . . . . . . . . . . . . . . 203 Renaissance.. . . . . . . . . . . . . . . . . . . . . . . . 272
marine archaeological reverse engineering. . . . . . . . . . . . . . 272
conservation. . . . . . . . . . . . . . . . . . . . . 281 riveted rings. . . . . . . . . . . . . . . . . . . . . . . . 140
marine archaeology. . . . . . . . . . . . . . 114
metal carboxylates. . . . . . . . . . . . . . . . 168 S
metal conservation. 123, 168, 300 sampling. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 212
metallography. . . . . . . . . . . . . . . . . . . . . 212 shipwrecks. . . . . . . . . . . . . . . . . . . . . 71, 281
metal soaps. . . . . . . . . . . . . . . . . . . . . . . . . 168 showcases. . . . . . . . . . . . . . . . . . . . . . . . . . . 261
Mewar. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 140 siderophores.. . . . . . . . . . . . . . . . . . . . . . . . . 25
microstructure.. . . . . . . . . . . . . 212, 221 silver alloys. . . . . . . . . . . . . . . . . . . . . . . . . 248
(micro-X-ray fluorescence) soil corrosion.. . . . . . . . . . . . . . . . . . . . . . . . 40
μ-XRF.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 184 spionkopite. . . . . . . . . . . . . . . . . . . . . . . . . 194
modern metal conservation. . . . . 65 Sri Lanka. . . . . . . . . . . . . . . . . . . . . . . . . . . . 290
monitoring.. . . . . . . . . . . . . . . . . . . . . . . . . 153 stainless steel. . . . . . . . . . . . . . . . . . . . . . . 109
monitoring cleaning. . . . . . . . . . . . . 160 storage.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 261
museum preservation. . . . . . . . . . . 308 sunflower oil. . . . . . . . . . . . . . . . . . . . . . . . . 50
surface documentation.. . . . . . . . . 160
N surveys. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 253
natural conservation. . . . . . . . . . . . . . . 16 sustainability. . . . . . . . . . . . . . . . . . . . . . . 109
neutron imaging. . . . . . . . . . . . . . . . . . 184 Sweden. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 176
nickel silver alloys. . . . . . . . . . . . . . . . . . 71
non-invasive analysis. . . . . . . . . . . . 212 T
Tara. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 290
O thickener. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
Owatrol oil. . . . . . . . . . . . . . . . . . . . . . . . . . 114 Thiriyaya. . . . . . . . . . . . . . . . . . . . . . . . . . . . 290
oxidation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50 traditional coatings.. . . . . . . . . . . . . . 103
oxygen depletion. . . . . . . . . . . . . . . . . . 253 traditional craftsmen. . . . . . . . . . . . 140
training programme. . . . . . . . . . . . . 300
P
Trinkspiel. . . . . . . . . . . . . . . . . . . . . . . . . . . 272
painted metals. . . . . . . . . . . . . . . . . . . . . 168
tumulus. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
partial vacuum.. . . . . . . . . . . . . . . . . . . . 317
past treatments. . . . . . . . . . . . . . . . . . . . . . 56 U
phylacterion. . . . . . . . . . . . . . . . . . . . . . . . 184 ultraviolet (UV). . . . . . . . . . . . . . . . . . . . . 50
Pleco®. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 248
pollutants.. . . . . . . . . . . . . . . . . . . . . . . . . . . 176
The Metals Working Group is part of the
International Council of Museums – Committee
for Conservation (ICOM–CC).
ISBN: 978-2-491997-61-8

Icom Metal2022

Uploaded by

Copyright:

Available Formats

Icom Metal2022

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Icom Metal2022

Uploaded by

Copyright:

Available Formats

METAL2022

SEPTEMBER 5–9, 2022

PROCEEDINGS OF THE INTERIM MEETING OF THE ICOM-CC METALS WORKING GROUP

EDITED BY PAUL MARDIKIAN, LIISA NÄSÄNEN, AND AKI ARPONEN

SEPTEMBER 5–9, 2022

EDITED BY PAUL MARDIKIAN, LIISA NÄSÄNEN, AND AKI ARPONEN

Coatings, Consolidants & Corrosion Inhibitors

Keris Blade Protection – A Comparison of Traditional and Contemporary Coatings (YCP)

Atmospheric Particulate Matter: How to Include it in Artificial Ageing?

Testing Gap Fillers for Archaeological Metal

Elina Anttila Eero Ehanti

Liisa Näsänen, Aki Arponen, and Eero Ehanti

The ICOM-CC Coordination Team for 2020–2023 Ahmad N. Abu-Baker

Paul Mardikian We are grateful to Nicola Emmerson, João Cura D’Ars de

Cátia Viegas Wesołowska

Cheers, mate. We miss you.

“Tell me, I forget. Show me, I remember. Involve me, I understand.”

Henning Matthiesen* Lars Brock Andersen Philippe Dillmann Trine Wiinblad

Total Retrieved Retrieved Retrieved

cleaned due to its fragility. Cleaning was carried out

Thermodynamic modelling of the stability of different

Reburied Peat, lab NaOH

Figure 4. Backscatter SEM images of zones 1–5 from an untreated

average Cl− content was quantified in the DPL within

untreated reference samples and three treated samples)

NaOH treatment caused significant changes to the

(28–29 cm3/mol Fe) is a modest change in volume, so

as Cl− from artefacts will accumulate in the peat, and it is References

Table 1. Overview of the different gel formulations tested

Xanthan gum Agar Gellan gum

Table 2. Overview of the artefacts treated

Material Description Picture

1 – atmospheric corrosion layer

Candle Holder Cast iron

1 – dirt and sediments

1 – external elements (pebbles)

Figure 2. Iron samples treated for 10 min with

Extracted [Fe] mg/L

6·10−2 M - DFO-amended gel; 17.76 ± 3.43 19.26 ± 0.78

6·10−2 M -DFO-amended gel; 21.22 ± 0.68 37.86 ± 6.68

6·10−2 M EDTA-amended gel; 16.15 ± 2.97 41.41 ± 2.57

Figure 4. (a) Different concentrations of DFO-amended

Post-treatment Raman analyses confirmed this finding

Figure 6. Raman spectra (633 nm, 0.1 mW,

of green chemistry for the preservation of artistic and

Luana Cuvillier and Arianna Passaretti are PhD

Elodie Guilminot is a senior conservation scientist at

Aymeric Raimon and Virginie Dupuy are respectively

Microbial Stabilization of Archaeological Iron Objects

Sarah James Edith Joseph*

Figure 1. Formation of iron

Aerated a Acidic b Chlorides c SRB d

Table 2. Microbial mechanisms and their effect on iron-containing heritage materials

Microbial mechanisms Microorganisms Substrate References

Geobacter sulfurreducens; Volkland et al. 2001; Cote et al. 2015

bacteria: Pseudomonas sp.

Materials and methods

Thickett, D. 2012. Post excavation changes and preventive

Mathilde Monachon is currently a postdoctoral

Sathiyanarayanan Ganesan is currently a senior