Luppi F, 245426, PHD Final Thesis May 19

Download as pdf or txt
Download as pdf or txt
You are on page 1of 255

CRANFIELD UNIVERSITY

FEDERICO LUPPI

INERT AND NITRATED CROSS-LINKED β-CYCLODEXTRIN


BINDERS FOR ENERGETIC APPLICATIONS

CRANFIELD DEFENCE AND SECURITY


PhD in Defence and Security

PhD
Academic Year: 2015 - 2019

Supervisor: Eleftheria Dossi


Associate Supervisor: Mike Williams
March 2019
CRANFIELD UNIVERSITY

CRANFIELD DEFENCE AND SECURITY


PhD in Defence and Security

PhD

Academic Year 2015 - 2019

FEDERICO LUPPI

Inert and Nitrated Cross-linked β-Cyclodextrin Binders for Energetic


Applications

Supervisor: Eleftheria Dossi


Associate Supervisor: Mike Williams
March 2019

This thesis is submitted in partial fulfilment of the requirements for


the degree of PhD
(NB. This section can be removed if the award of the degree is
based solely on examination of the thesis)

© Cranfield University 2019. All rights reserved. No part of this


publication may be reproduced without the written permission of the
copyright owner.
Abstract
This PhD project focuses on the synthesis of new inert and nitrated cross-linked
cyclodextrin systems as binders for energetic formulations. Three diglycidyl
ethers with polyethylene glycol segments differing in length were used to cross-
link β-cyclodextrin. The physicochemical properties of the compounds were
investigated by proton nuclear magnetic resonance spectroscopy, differential
scanning calorimetry, thermogravimetric analysis and dynamic mechanical
analysis. The polyethylene glycol chains linked to the rigid β-cyclodextrin
conferred low glass transition temperatures (≥ –20 °C) on the inert binders and
also on their nitrated derivatives (≥ –14 °C). This is the first time that nitrated β-
cyclodextrin derivatives have shown viscoelastic behaviour at temperatures
below 0 °C. The viscoelasticity of both the inert and nitrated compounds
increased with the amount of polyethylene glycol chains in the system. Inert
binders with higher polyethylene glycol:β-cyclodextrin units ratios were softer
and exhibited self-healing behaviour. The thermo-mechanical characterisation
of these binders revealed that the system was exposed to mechanical stress
below the glass transition temperature, and the stress was directly related to the
proportion of the soft polyethylene glycol segments. The nitrated cross-linked
derivatives were characterised by decomposition temperatures of ~200 °C and
thermal degradation energies of 1400–2100 J g-1 strictly dependent on the
degree of cross-linking and nitration. Self-healing properties were confirmed in
nitrated products with a high polyethylene glycol segments content. Nitrated
samples with polyethylene glycol segments:β-cyclodextrin units ratios >3.8:1
were safer to handle in the laboratory as determined by small-scale hazard and
compatibility tests with various energetic fillers. Additionally, preliminary
Energetic Materials Testing Assessment Policy (EMTAP) tests confirmed the
samples were not sensitive to electrostatic discharge up to 4.5 J but were
sensitive to impact, with a figure of insensitiveness of 29. The nitrated samples
were unstable at temperatures >80 °C. The materials developed during this
PhD project could facilitate the manufacturing and storage of new binders and
may offer a suitable replacement for nitrocellulose and other binders in

i
energetic formulations. The stabilisation of the nitrated cross-linked binders
should be prioritised in future work.

Keywords: β-cyclodextrin, polyethylene glycol, cross-linking, thermal stability

This PhD was funded through the Weapons Science and Technology Centre
(WSTC) by the UK Defence Science and Technology Laboratory (DSTL)

ii
Acknowledgements
To the Centre for Defence Chemistry and my family. Many Thanks!

I am grateful to my father and mother, who have provided me thorough


emotional support and a moral compass in my life. I am also grateful to my
childhood friends who have supported me along the way.

Special gratitude goes out to the staff of the Centre for Defence Chemistry. A
special mention goes to my supervisor Eleftheria Dossi, who guided and
advised me over the past four years, helping me to overcome all the obstacles I
have met. Thank you to my adoptive mentor Guillaume Kister for having guided
me as if I was one of his own PhD students. I am also grateful to the other CfDC
university staff: Jeff Pons, Ian Wilson, Dan McAteer, Matt Parker, Lisa
Humphreys, Matt Weaver, Peter Wilkinson, Nicola Darcy, Susan Waring, Mark
Carpenter, Nathalie Mai, Philp Gill, Roger Cox, Mike Williams, Sally Gaulter,
Chris Stennet, Matt Goldsmith, Scott Clews, Susan Hardy and Patricia Pye for
their unfailing support and assistance.

And finally, last but by no means least, also to my friends, it was great sharing
the burden of the many challenges we and I have faced. To Hamish, Alejandro,
Erick, Andrea, Antonio and Christine:

Ubi amici, ibidem opes. (Plauto)

iii
TABLE OF CONTENTS
Abstract ............................................................................................................... i
Acknowledgements ............................................................................................ iii
LIST OF FIGURES ............................................................................................ vii
LIST OF TABLES ............................................................................................. xiv
LIST OF EQUATIONS ...................................................................................... xvi
LIST OF ABBREVIATIONS ................................................................................ 1
1 Overview ......................................................................................................... 4
1.1 Objectives ................................................................................................. 5
1.2 Academic contributions ............................................................................. 7
1.3 Thesis structure ........................................................................................ 7
2 Introduction...................................................................................................... 9
2.1 Insensitive munitions................................................................................. 9
2.2 Polymers and their properties ................................................................. 10
2.3 Polymers as binders in energetic formulations ....................................... 12
2.4 Cyclodextrins .......................................................................................... 19
2.4.1 Properties of CDs ............................................................................. 21
2.4.2 Synthesis of cyclodextrins ................................................................ 30
2.4.3 Cross-linking of cyclodextrins ........................................................... 30
2.4.4 Energetic applications of unlinked/cross-linked cyclodextrins .......... 34
3 Experimental ................................................................................................. 41
3.1 Materials ................................................................................................. 41
3.2 Methods .................................................................................................. 41
3.2.1 NMR spectroscopy ........................................................................... 41
3.2.2 FTIR spectroscopy ........................................................................... 41
3.2.3 Gel permeation chromatography ...................................................... 41
3.2.4 Differential scanning calorimetry and thermal gravimetric analysis .. 42
3.2.5 Dynamic mechanical analysis .......................................................... 42
3.2.6 Optical microscopy ........................................................................... 43
3.2.7 Scanning electron microscopy ......................................................... 44
3.3 Procedures ............................................................................................. 44
3.3.1 Triethylene glycol diglycidyl ether [120] .............................................. 44
3.3.2 Hexaethylene glycol diglycidyl ether [120] .......................................... 44
3.3.3 Cross-linked cyclodextrins (βCXCDs) .............................................. 45
3.3.4 Nitrated cross-linked cyclodextrins (βNCXCDs) ............................... 45
3.3.5 Hazard testing .................................................................................. 46
3.3.6 Compatibility tests ............................................................................ 47
3.3.7 Stability analysis ............................................................................... 48
4 Synthesis and characterisation of inert βCXCDs ........................................... 49
4.1 Synthesis and chemical characterisation ................................................ 49
4.1.1 Effect of the NaOH concentration ..................................................... 53

v
4.1.2 Effect of the reaction time................................................................. 54
4.1.3 Effect of the temperature .................................................................. 55
4.1.4 Effect of βCD alkoxide formation ...................................................... 56
4.1.5 Effect of the volume of NaOH solution ............................................. 56
4.1.6 Effect of cross-linker:βCD ratio......................................................... 57
4.1.7 Effect of the duration of cross-linker addition ................................... 59
4.2 The synthesis of βCXCDs ....................................................................... 60
4.3 Thermal and thermo-mechanical characterisation of βCXCDs ............... 78
4.3.1 Thermo-mechanical characterisation of the starting materials ......... 79
4.3.2 Thermal and thermo-mechanical characterisation of inert
βCXCDs .................................................................................................... 93
4.4 Conclusions .......................................................................................... 112
5 Synthesis and characterisation of energetic cross-linked βNCXCDs .......... 115
5.1 Attempts to nitrate the insoluble βCXCD precursors ............................. 117
5.2 Attempts to nitrate the soluble βCXCD precursors ............................... 119
5.3 Synthesis and chemical characterisation of βNCXCDs ........................ 121
5.4 Thermal and thermo-mechanical characterisation of energetic
βNCXCDs ................................................................................................... 133
5.4.1 Thermal characterisation of βNCXCDs prepared from insoluble
precursors ............................................................................................... 134
5.4.2 Thermal characterisation of βNCXCDs prepared from soluble
precursors ............................................................................................... 137
5.4.3 Thermo-mechanical characterisation of βNCXCDs ........................ 144
5.4.4 Qualitative analysis of the self-healing properties of βNCPCD34... 145
5.5 Conclusions .......................................................................................... 145
6 Compatibility, hazard and stability studies ................................................... 147
6.1 Compatibility tests – βCXCDs ............................................................... 147
6.2 Compatibility tests – βNCXCDs ............................................................ 155
6.3 Hazard tests – βNCXCDs ..................................................................... 164
6.3.1 EMTAP tests .................................................................................. 169
6.4 Stability of βNCXCDs determined by heat flow calorimetry .................. 171
6.5 Conclusion ............................................................................................ 173
7 Conclusion and recommendations .............................................................. 177
REFERENCES ............................................................................................... 183

vi
LIST OF FIGURES
Figure 2.1 Chemical structure of triethylene glycol diglycidyl ether .................. 19
Figure 2.2 Chemical structure of α, β and γ CDs [40]. ...................................... 20
Figure 2.3 Distribution of the CD publications by application (1996–2013)
according to CD News [38]. ....................................................................... 21
Figure 2.4 Structures of (a) α-D-glucopyranoside, the repeating unit of CD, and
(b) the toroid representation of CDs, with primary and secondary hydroxyl
groups in red and blue, respectively. H1, H2 and H4 are external to the
toroid, whereas H3 and H5 are internal ..................................................... 22
Figure 2.6 Schematic representation of a polyrotaxane formed by several CDs
threaded on a polymer chain with ending caps. ......................................... 24
Figure 2.7 The 1H-NMR spectra of CD (a) in D2O and (b) in DMSO-d6 [63]. .... 28
Figure 2.8 The infrared spectrum of βCD. ........................................................ 29
Figure 4.1 Dialysis apparatus for the purification of βCXCDs. .......................... 59
Figure 4.2 Cross-linking apparatus for the synthesis of βCXCDs. .................... 61
Figure 4.3 Physical characteristics of βCPCD. (a) Malleable IP13. (b) Powdery
IP17. .......................................................................................................... 64
Figure 4.4 Proposed chemical structure of the βCXCD products based on IP13
incorporating PEG segments, underlining the co-existence of PEG units as
spacers and as entanglements. ................................................................. 65
Figure 4.5 Comparative 1H NMR spectra of (a) sample IP13 (PEG:βCD ratio =
3.8:1), (b) PEGDGE, and (c) βCD in DMSO-d6 with assignments of signals.
The DMSO reference peak is at 2.5 ppm. R= H, cross-linker Ca-OR = H or
βCD. .......................................................................................................... 66
Figure 4.6 Comparative 1H NMR spectra of (a) IP13 (PEG:βCD ratio = 3.8:1),
(b) mixture of degraded/non degraded PEGDGE, and (c) PEGDGE in
DMSO-d6. .................................................................................................. 68
Figure 4.7 (a) The 1H-NMR spectrum of IP13 (PEG:βCD ratio = 3.8:1) in D2O
and (b) the proposed chemical structure of βCXCDs highlighting the two
different situations for the anomeric H-1, when in unit 1 or unit 2 of
βCXCDs. ................................................................................................... 69
Figure 4.8 The 1H-NMR spectrum of IP13 (PEG:βCD ratio = 3.8:1) in D2O. Int1 =
anomeric H-1, Int3 = CH2 of PEG. .............................................................. 71
Figure 4.9 Comparative 1H-NMR spectra of βCPCD samples (IP13, IP15–IP17)
with different PEG:βCD ratios. Lower ratios enhance the visibility of the
βCD proton (marked with the arrow). ......................................................... 72

vii
Figure 4.10 Comparative FTIR spectra of βCD (black curve), βCPCD sample
IP13 (red curve), and PEGDGE (blue curve). ............................................ 73
Figure 4.11 GPC analysis of βCD compared to polyethylene glycol/polyethylene
oxide standards. ........................................................................................ 74
Figure 4.12 GPC analysis of PEGDGE compared to polyethylene
glycol/polyethylene oxidestandards. .......................................................... 74
Figure 4.13 Analysis of βCXCDs by GPC compared to polyethylene
glycol/polyethylene oxide standards. (a) Sample IP22 (PEG:βCD ratio =
2.9). (b) Sample IH2 (HEG:βCD ratio = 3.2). (c) Sample IT2 (TEG:βCD ratio
= 3.1). ........................................................................................................ 76
Figure 4.14 Analysis of βCTCDs by GPC compared to polyethylene
glycol/polyethylene oxide standards. (a) Sample IT1 (TEG:βCD ratio = 3.6).
(b) Sample IT2 (TEG:βCD ratio = 3.1). (c) Sample IT3 (TEG:βCD ratio =
2.4). (d) Sample IT4 (TEG:βCD ratio = 1.9). .............................................. 77
Figure 4.15 Combined TGA and DSC analysis of βCD (10 °C min-1, 25–500 °C,
aluminium crucible). ................................................................................... 79
Figure 4.16 The decomposition of βCD observed by melting point analysis. At
317 °C, melting and caramelisation are dominant, but above 328 °C
charring prevails. ....................................................................................... 80
Figure 4.17 DSC heating curves of βCD (10 °C min-1) from –100 to 100 °C in
the first cycle (black), from –100 to 120 °C in the second cycle (blue), and
from –100 to 140 °C in the third cycle (red). .............................................. 81
Figure 4.18 DMA showing variation in the storage modulus (E′, solid lines) and
tanδ (dashed lines) of βCD (10 °C min-1, 1 Hz, third temperature cycle, –
100 to 140 °C, aluminium pocket). ............................................................. 82
Figure 4.19 Combined DSC/TGA thermograms of the cross-linkers TEGDGE
(black), HEGDGE (blue) and PEGDGE (red), from 30 to 500 °C (10 °C min -
1
, third temperature cycle, aluminium crucible). ......................................... 83
Figure 4.20 DSC thermogram of the cross-linkers: heating (solid) and cooling
(dashed) of TEGDGE (black), HEGDGE (blue) and PEGDGE (red) from –
100 to 120 °C (10 °C min-1, third temperature cycle, aluminium crucible). 85
Figure 4.21 DMA showing variation in the storage modulus (E′, solid lines) and
tanδ (dashed lines) of PEGDGE (10 °C min-1, 1 Hz, third temperature cycle
from –100 to 100 °C, aluminium pocket). .................................................. 87
Figure 4.22 DMA showing variation in the storage modulus (E′) of PEGDGE
with frequency (10 °C min-1, 1 and 10 Hz, third temperature cycle from –100
to 100 °C, aluminium pocket)..................................................................... 89
Figure 4.23 DMA showing variation in the storage modulus (E′, solid lines) and
tanδ (dashed lines) of HEGDGE (10 °C min-1, 1 Hz, third temperature cycle
from –100 to 100 °C, aluminium pocket). .................................................. 90

viii
Figure 4.24 DMA showing variation in the storage modulus (E′, solid lines) of
HEGDGE with the frequency (10 °C min-1, 1 and 10 Hz, third temperature
cycle from –100 to 100 °C, aluminium pocket). ......................................... 91
Figure 4.25 DMA showing variation in the storage modulus (E′, solid lines) and
tanδ (dashed lines) of TEGDGE (10 °C min-1, 1 Hz, third temperature cycle
from –100 to 100 °C, aluminium pocket). .................................................. 92
Figure 4.26 DMA showing variation in the storage modulus (E′, solid lines) and
tanδ (dashed lines) of TEGDGE in an extended temperature scan (10 °C
min-1, 1 Hz, third temperature cycle from –140 to 100 °C, aluminium
pocket). ...................................................................................................... 93
Figure 4.27 Combined DSC analysis and TGA of IP18 (PEG:βCD ratio = 4.1:1,
red line) and its precursors βCD (blue) and PEGDGE (green) over the
temperature range 30–500 °C (10 °C min-1, aluminium crucible). ............. 97
Figure 4.28 DSC analysis showing the plasticising effect of water in IP18 from –
100 to 120 °C (10 °C min-1, third temperature cycle, aluminium crucible). 98
Figure 4.29 DSC analysis of βCXCD products with high XEG:βCD ratios,
namely IT1 (3.6:1, black line), IH1 (4.0:1, blue line) and IP13 (3.8:1, red
line) from –100 to 120 °C (10 °C min-1, third temperature cycle, aluminium
crucible). .................................................................................................... 99
Figure 4.30 Photographs of βCPCD samples supported by (a) an aluminium
pocket, (b) a stainless-steel mesh, and (c) an aluminium pocket wrapped in
PTFE tape. .............................................................................................. 100
Figure 4.31 DMA showing variation in the storage modulus (E′, solid lines) and
tanδ (dashed lines) of sample IP9 (10 °C min-1, 1 Hz, third temperature
cycle from –100 to 140 °C, aluminium pocket). ....................................... 102
Figure 4.32 DMA showing variation in the storage modulus (E′) and tanδ of
sample IP9 (PEG:βCD ratio = 3.5, solid and dotted red lines) and IP17
(PEG:βCD ratio = 1.6, solid and dotted blue lines) (10 °C min -1, 1 Hz, third
temperature cycle from –100 to 140 °C, aluminium pocket). ................... 103
Figure 4.33 DMA showing variation in the storage modulus (E′, solid lines) and
tanδ (dashed lines) of IP9 (10 °C min-1, 1 Hz, third temperature cycle, –100
to 140 °C, stainless-steel mesh). ............................................................. 105
Figure 4.34 DMA showing variation in the storage modulus (E′, solid lines) and
tanδ (dashed lines) of IP9 (10 °C min-1, 1 Hz, third temperature cycle, –100
to 140 °C, aluminium pocket and PTFE tape).......................................... 106
Figure 4.35 Optical microscope images of IP9 (PEG:βCD ratio = 3.5) during the
first temperature cycle from –100 to 100 °C. The top row (a-d) shows the
cooling phase captured at 25 °C, –55 °C, –78 °C and –80 °C and the
bottom row (e-h) shows the heating phase captured at 2 °C, 43 °C, 67 °C
and 87 °C ................................................................................................ 107

ix
Figure 4.36 Optical microscope images of IP9 (PEG:βCD ratio = 3.5) during the
second temperature cycle from –100 to 100°C. The top row (a-d) shows the
cooling phase captured at 100 °C, 25 °C, –61 °C and –88 °C and the
bottom row (e-h) shows the heating phase captured at 5 °C, 43 °C, 68 °C
and 84 °C................................................................................................. 107
Figure 4.37 Self-healing of IP13 (PEG:βCD ratio = 3.8). (a) Sample after solvent
evaporation. (b) Sample after cutting. (c) The parts are placed in contact.
(d) The sample is heated to 70 °C for 30 min and pulled by the extremities.
................................................................................................................ 108
Figure 4.38 Scanning electron micrograph of βCD crystals. .......................... 109
Figure 4.39 Scanning electron micrographs of the βCTCDs (a) IT1 (b) IT2 (c)
IT3 and (d) IT4. ........................................................................................ 110
Figure 4.40 Scanning electron micrographs of the βCHCDs (a) IH1 and (b) IH4,
and the βCPCDs (c) IP19 and (d) IP17. .................................................. 111
Figure 4.41 DMA showing variation in the storage modulus (E′, solid lines) and
tanδ (dashed lines) of sample IP17 (10% w/w) formulated with 90% w/w
melamine (10 °C min-1, 1 Hz, third temperature cycle from –100 to 140 °C,
aluminium pocket). .................................................................................. 112
Figure 5.1 Physical appearance of the insoluble IP26 and IT5–IT7 precursors.
................................................................................................................ 118
Figure 5.2 Nitrated product degraded by the use of NaOH during purification.
................................................................................................................ 121
Figure 5.3 The 1H NMR spectra of (a) βNCD [116] and (b) NP1 in DMSO-d6. 123
Figure 5.4 Comparative 1H-NMR spectra of (a) inert precursor IP18 (PEG:βCD
ratio = 2.5:1) and (b) its nitrated product NP1, both in DMSO-d6. ............ 124
Figure 5.5 The 1H NMR spectra of products (a) NP1, (b) NH1 and (c) NT1 in
DMSO-d6. The XEG:βCD ratios were previously determined by NMR
spectroscopy. .......................................................................................... 125
Figure 5.6 The 1H NMR spectra of products (a) NP1, (b) NP2, (c) NP3 and (d)
NP4, which show a gradually decreasing PEG:βCD ratio. ...................... 126
Figure 5.7 Comparative FTIR spectra of sample NP1 (solid line) and its inert
precursor IP18 (dashed line). .................................................................. 129
Figure 5.8 Comparative FTIR spectra of sample NP1 (solid line) and βNCD
(dashed line). ........................................................................................... 130
Figure 5.9 Comparative GPC analysis of βNXCDs: (a) NP1 (PEG:βCD ratio =
3.8), (b) NH1 (HEG:βCD ratio = 4.0), and (c) NT1 (TEG:βCD ratio = 3.8).
................................................................................................................ 132

x
Figure 5.10 Comparative GPC chromatograms of βNPCDs with decreasing
PEG:βCD ratios: (a) NP1 (PEG:βCD ratio = 3.8), (b) NP2 (PEG:βCD ratio =
2.6), (c) NP3 (PEG:βCD ratio = 2.1), and (d) NP4 (PEG:βCD ratio = 1.6).
................................................................................................................ 133
Figure 5.11 DSC thermogram of NP9, from 30 to 300 °C (10 °C min-1,
aluminium crucible). ................................................................................. 135
Figure 5.12 DSC thermogram of NT7, from 30 to 500 °C (10 °C min -1,
aluminium crucible). ................................................................................. 135
Figure 5.13 DSC thermogram of NT9 from –100 to 100 °C (10 °C min-1, third
temperature cycle, aluminium crucible). .................................................. 137
Figure 5.14 DSC thermogram of NP1 from 25 to 300 °C (10 °C min-1, aluminium
crucible). .................................................................................................. 140
Figure 5.15 DSC thermogram of NP1 from –100 to 100 °C showing the
pronounced glass transition of the highly cross-linked products (10 °C min-1,
aluminium crucible). ................................................................................. 141
Figure 5.16 Plot of Tg against XEG:βCD ratio for βCHCDs (blue line) and
βCPCDs (orange line). ............................................................................ 142
Figure 5.17 DMA showing variation in the storage modulus (E′, solid lines) and
tanδ (dashed lines) of NP1 (10 °C min-1, 1 Hz, third temperature cycle from
–100 to 140 °C, aluminium pocket). ......................................................... 144
Figure 5.18 Self-healing ability of NP1 (PEG:βCD ratio = 4:1). ...................... 145
Figure 6.1 DSC thermogram of HMX (blue line), IP19 (red line), and a 50/50
(w/w) mixture (black line) from 30 to 300 °C (1 mg, 2 °C min -1, aluminium
crucible). .................................................................................................. 149
Figure 6.2 DSC thermogram of RDX (blue line), IP19 (red line), and a 50/50
(w/w) mixture (black line) from 30 to 300 °C (1 mg, 2 °C min-1, aluminium
crucible). .................................................................................................. 150
Figure 6.3 DSC thermogram of PETN (blue line), IP19 (red line), and a 50/50
(w/w) mixture (black line) from 30 to 300 °C (1 mg, 2 °C min-1, aluminium
crucible). .................................................................................................. 150
Figure 6.4 DSC thermogram of ADN (blue line), IP19 (red line), and a 50/50
(w/w) mixture (black line) from 30 to 300 °C (1 mg, 2 °C min -1, aluminium
crucible). .................................................................................................. 151
Figure 6.5 DSC thermogram of KClO3 (blue line), IP19 (red line), and a 50/50
(w/w) mixture (black line) from 30 to 400 °C (1 mg, 2 °C min -1, aluminium
crucible). .................................................................................................. 152
Figure 6.6 DSC thermogram of KNO3 (blue line), IP19 (red line), and a 50/50
(w/w) mixture (black line) from 30 to 300 °C (1 mg, 2 °C min -1, aluminium
crucible). .................................................................................................. 152

xi
Figure 6.7 DSC thermogram of NH4ClO4 (blue line), IP19 (red line), and a 50/50
(w/w) mixture (black line) from 30 to 500 °C (1 mg, 2 °C min -1, aluminium
crucible). .................................................................................................. 153
Figure 6.8 DSC thermogram of NH4NO3 (blue line), IP19 (red line), and a 50/50
(w/w) mixture (black line) from 30 to 300 °C (1 mg, 2 °C min -1, aluminium
crucible). .................................................................................................. 154
Figure 6.9 DSC thermogram of red phosphorus (blue line), IP19 (red line), and
a 50/50 (w/w) mixture (black line) from 30 to 550 °C (1 mg, 2 °C min -1,
aluminium crucible). ................................................................................. 154
Figure 6.10 DSC thermograms of HMX (blue line), NP13 (red line), and a 50/50
(w/w) mixture (black line) from 30 to 300 °C (1 mg, 2 °C min-1, aluminium
crucible). .................................................................................................. 156
Figure 6.11 DSC thermograms of RDX (blue line), NP13 (red line), and a 50/50
(w/w) mixture (black line) from 30 to 300 °C (1 mg, 2 °C min -1, aluminium
crucible). .................................................................................................. 157
Figure 6.12 DSC thermograms of PETN (blue line), NP13 (red line), and a
50/50 (w/w) mixture (black line) from 30 to 300 °C (1 mg, 2 °C min -1,
aluminium crucible). ................................................................................. 158
Figure 6.13 DSC thermograms of ADN (blue line), NP13 (red line), and a 50/50
(w/w) mixture (black line) from 30 to 300 °C (1 mg, 2 °C min -1, aluminium
crucible). .................................................................................................. 159
Figure 6.14 DSC thermograms of KClO3 (blue line), NP13 (red line), and a
50/50 (w/w) mixture (black line) from 30 to 400 °C (1 mg, 2 °C min -1,
aluminium crucible). ................................................................................. 159
Figure 6.15 DSC thermograms of KNO3 (blue line), NP13 (red line), and a 50/50
(w/w) mixture (black line) from 30 to 400 °C (1 mg, 2 °C min -1, aluminium
crucible). .................................................................................................. 160
Figure 6.16 DSC thermograms of NH4ClO4 (blue line), NP13 (red line), and a
50/50 (w/w) mixture (black line) from 30 to 500 °C (1 mg, 2 °C min -1,
aluminium crucible). ................................................................................. 160
Figure 6.17 DSC thermograms of NH4NO3 (blue line), NP13 (red line), and a
50/50 (w/w) mixture (black line) from 30 to 300 °C (1 mg, 2 °C min-1,
aluminium crucible). ................................................................................. 161
Figure 6.18 DSC thermograms of RedP (blue line), NP13 (red line), and a 50/50
(w/w) mixture (black line) from 30 to 550 °C (1 mg, 2 °C min-1, aluminium
crucible). .................................................................................................. 162
Figure 6.19 DSC thermograms of DPA (blue line), NP1 (red line), and a 50/50
(w/w) mixture (black line) from 30 to 300 °C (1 mg, 2 °C min -1, aluminium
crucible). .................................................................................................. 163

xii
Figure 6.20 DSC thermograms of 2,4-NDPA (blue line), NP1 (red line), and a
50/50 (w/w) mixture (black line) from 30 to 300 °C (1 mg, 2 °C min-1,
aluminium crucible). ................................................................................. 163
Figure 6.21 DSC thermograms of Centralite (blue line), NP1 (red line), and a
50/50 (w/w) mixture (black line) from 30 to 550 °C (1 mg, 2 °C min-1,
aluminium crucible). ................................................................................. 164
Figure 6.22 Thermal stability of pure NP1 and the same compound in the
presence of three different stabilisers (1% w/w). ..................................... 172

xiii
LIST OF TABLES
Table 2.1 List of binders in energetic formulations [23]. ................................... 13
Table 2.2 Example of polymers used in energetic applications [20]. ................ 15
Table 2.3 Examples of energetic polymers used as binders [20]. .................... 17
Table 2.4 The physical properties of α, β and γ CDs [38]. ................................ 22
Table 2.5 Thermal properties of α, β and γ CDs. .............................................. 23
Table 2.6 The 1H-NMR chemical shifts of CD protons in D2O and DMSO-d6 [63].
.................................................................................................................. 29
Table 2.7 The 13C NMR chemical shifts of CDs in D2O [63]. ............................ 29
Table 2.8 Applications of cross-linked CDs ...................................................... 31
Table 2.9 Applications of CDs functionalised with energetic molecules. .......... 34
Table 2.10 Summary of βCD nitration conditions [114]. ................................... 36
Table 2.11 Performance of TEMTN:βNCD propellant formulations [114]. ........ 37
Table 2.12 Sensitivity of nitrated cross-linked βCCD and RDX complexes [109]
.................................................................................................................. 38
Table 4.1 Effect of NaOH concentration on the yield of βCPCDs at a
PEGDGE:βCD ratio of 9:1. ........................................................................ 53
Table 4.2 Effect of the reaction time (Step 2) on the yield of βCPCDs. ............ 54
Table 4.3 Effect of the temperature on the yield of βCPCDs, when the
PEGDGE:βCD ratio is 5:1. ........................................................................ 55
Table 4.4 Effect of the time allowed for βCD alkoxide formation (Step 1) on the
yield of βCPCDs. ....................................................................................... 56
Table 4.5 Effect of PEGDGE:βCD ratio on the yield of βCPCDs. ..................... 57
Table 4.6 Effect of cross-linker addition time. ................................................... 60
Table 4.7 Synthesis of βCXCDs: summary of the most successful water-soluble
products. .................................................................................................... 63
Table 4.8 Combined DSC/TGA data for the cross-linkers. ............................... 84
Table 4.9 Summary of the thermal properties of the βCXCDs. ........................ 95
Table 4.10 Comparison of DMA and DSC results for βCXCD samples and their
precursors................................................................................................ 101
Table 5.1 Nitrogen content of the βNCXCDs. ................................................. 128

xiv
Table 5.2 Thermal properties of the energetic derivatives of gel-like insoluble
βCPCDs and insoluble βCTCD powders. ................................................ 136
Table 5.3 Thermal properties of βNCXCDs prepared from water-soluble
βCXCDs. ................................................................................................. 138
Table 6.1 Compatibility of inert βCPCD (sample IP19, PEG:βCD ratio = 4.0:1)
with various energetics. ........................................................................... 148
Table 6.2 Compatibility of nitrated βNCPCD (samples NP13 and NP1,
PEG:βCD ratio = 3.8:1) with various energetics and stabilisers. ............. 155
Table 6.3 Preliminary small-scale compatibility tests performed on βNCTCDs
and βNCTCDs containing TEG and PEG units, respectively. .................. 166
Table 6.4 EMTAP test data for sample NP17 based on Rotter impact and ESD
values. ..................................................................................................... 170
Table 6.5 DSC analysis of the decomposition temperatures of sample NP1
before and after HFC tests. ..................................................................... 173

xv
LIST OF EQUATIONS
Equation 1 Number of βCD anomeric protons determined by 1H-NMR ........... 70
Equation 2 Number of all XEG:βCD protons in βCXCDs ................................. 70
Equation 3 Correction factor: Determination of nitrogen content .................... 127
Equation 4 Nitrogen content (percentage) ..................................................... 127
Equation 5 Oxygen balance of explosives (percentage) ................................ 143

xvi
LIST OF ABBREVIATIONS

2,4-NDPA 2,4-Dinitrodiphenylamine
AP Ammonium perchlorate
CD Cyclodextrin
CEF Trichloroethyl phosphate
CL-20 Hexanitrohexaazaisowurtzitane
CTPB Carboxyl-terminated polybutadiene
D2O Deuterated water
DMA Dynamic mechanical analysis
DMSO-d6 Deuterated dimethylsulfoxide
DNAN 2,4-Dinitroanisole
DPA Diphenylamine
DS Degree of substitution
DSC Differential scanning calorimetry
E′ Storage modulus
EMTAP Energetic Materials Testing Assessment Policy
ESD Electrostatic discharge
FoI Figure of insensitiveness
FTIR Fourier transformed infrared
GPC Gel permeation chromatography
HEG Hexaethylene glycol unit
HEGDGE Hexaethylene glycol diglycidyl ether
HFC Heat flow calorimetry
HTPB Hydroxyl-terminated polybutadiene
IH Inert cross-linked βCD containing HEG units
IM Insensitive munition
IP Inert cross-linked βCD containing PEG units
IR Infrared
IT Inert cross-linked βCD containing TEG units
Kel-F 800 Poly(chlorotrifluoroethylene-co-vinylidene fluoride)
MIL-STD Military standard
MWCO Molecular weight cut-off

1
NC Nitrocellulose
NCD Nitrated cyclodextrin
NH Nitrated cross-linked βCD containing HEG units
NMR Nuclear magnetic resonance
NP Nitrated cross-linked βCD containing PEG segments
NT Nitrated cross-linked βCD containing TEG segments
PBAA Polybutadiene-acrylic acid copolymer
PBAN Polybutadiene-co-acrylonitrile
PEG Polyethylene glycol unit
PEGDGE Polyethylene glycol diglycidyl ether (500 Da)
PTFE Polytetrafluoroethylene
RT Room temperature
SEM Scanning electron microscopy
STANAG Standardisation agreement
tanδ Damping factor
TBAB Tetrabutyl ammonium bromide
TEG Triethylene glycol unit
TEGDGE Triethylene glycol diglycidyl ether
Tg Glass transition temperature
TGA Thermal gravimetric analysis
THF Tetrahydrofuran
TIVS Thermal initiated venting system
TMETN Trimethylolethane trinitrate
TNAZ 1,3,3-Trinitroazetidine
TNT Trinitrotoluene
TPE Thermoplastic elastomer
XEG Generic polyethylene glycol unit, used as a collective term for
TEG, HEG and PEG units
βCD β-Cyclodextrin
βCHCD Cross-linked βCD containing HEG units
βCPCD Cross-linked βCD containing PEG units
βCTCD Cross-linked βCD containing TEG units
βCXCD Generic cross-linked βCD system containing X segments, where

2
X=TEG, HEG or PEG
βNCD Nitrated βCD
βNCHCD Nitrated cross-linked βCD containing HEG units
βNCPCD Nitrated cross-linked βCD containing PEG units
βNCTCD Nitrated cross-linked βCD containing TEG units
βNCXCD Generic nitrated cross-linked βCD system containing X units,
where X=TEG, HEG or PEG
Ω Oxygen balance

3
1 Overview
The development of new binders is fundamental to the design of more
advanced insensitive munitions (IMs). The function of the binder is to support
the constituents of a formulation and create a cohesive mass. Binders also
improve the properties of formulations, increasing their mechanical strength
and/or the processability of the formulated energetics.

Binders used in the defence industry must enable the production of safe
formulations with low vulnerability to stimuli such as impact, heat and
electrostatic discharge (ESD) without compromising the energetic performance
of the formulations.

Most traditional binders are unsustainably derived from crude oil and do not
make an energetic contribution to the formulation. One exception is
nitrocellulose (NC), which is sustainably derived from cotton cellulose and has
been used extensively through history as a filler and energetic binder. However,
NC has inconsistent batch-to-batch physicochemical properties due to structural
differences caused by the origin of the cotton and the conditions of cultivation,
harvesting and storage. These differences affect the properties of the biomass
(lignocellulose) and ultimately those of the nitrated derivatives and the resulting
energetic formulations. The selection and certification of cotton batches suitable
for NC production is therefore a long and expensive process.

This PhD project focused on the development of reliable, sustainable binders


that contribute to the energetic content of a formulation. With insight from the
pharmaceutical industry, β-cyclodextrin (βCD) was considered as one of the
most promising building blocks for future binders. This compound can be
synthesised inexpensively from a renewable resource (starch) using a robust,
enzymatic process. Natural βCD can be used entrap small molecules such as
drugs in pharmaceutical formulations.

The functionalisation of the hydroxyl groups in βCD is a simple and convenient


method to produce new oligomeric and polymeric derivatives. The multiple
hydroxyl groups in βCD macrocycles allow the creation of three-dimensional

4
(3D) macrostructures with enhanced inclusion ability. The main constituents of
energetic formulations such as explosives, oxidisers and pyrotechnic fillers
would be less susceptible to external stimuli when embedded in a βCD-based
binder, making the formulation safer and easier to handle.

1.1 Objectives
The overall aim of this PhD project was to synthesise a series of energetic
cross-linked βCD derivatives and characterise them to determine their suitability
as novel binders. Pure βCD and its nitrated derivatives are unsuitable for the
processing of energetics due to the high crystallinity of these molecules.
Flexible polyethylene glycol segments of various lengths (XEG) were therefore
chosen to cross-link βCD and improve its processability.

The synthesis route selected for the production of energetic cross-linked βCDs
(βNCXCDs) comprises two steps, as summarised in Scheme 1.1:

1. Cross-linking of βCD using polyethylene glycol diglycidyl ethers of


various lengths to form βCXCDs.
2. Nitration of the βCXCDs.

The solubility of βCXCDs in common polar solvents such as water, methanol or


ethanol would facilitate formulation development. Therefore the studies focused
on finding optimal reaction parameters to obtain water-soluble compounds for
easy purification and processability. The properties of the prepared compounds
were investigated using the following analytical techniques:

1. Nuclear magnetic resonance (NMR) and Fourier transform infrared


(FTIR) spectroscopy for the chemical characterisation of the cross-linked
and nitrated products.
2. Iron sulfate titration to measure the nitrogen content of the nitrated cross-
linked products
3. Differential scanning calorimetry (DSC), thermal gravimetric analysis
(TGA) and dynamic mechanical analysis (DMA) for the thermo-
mechanical characterisation of the cross-linked and nitrated cross-linked
products to determine their compatibility with energetic fillers.

5
4. Hazard tests to assess the vulnerability of the compounds.

Scheme 1.1 Schematic representation of the synthesis of βCXCD and βNCXCD


binders.

6
1.2 Academic contributions
The work described in this thesis has thus far resulted in two peer-reviewed
publications (one in Propellants, Explosives, and Pyrotechnics, and another in
Polymer Testing) as well as two posters presented at one domestic conference
(Defence and Security Doctoral Symposium 2015–2016, Cranfield University,
UK) and one international conference (Macro 2018, 47th IUPAC World Polymer
Congress, Cairns, Australia). The work also contributed to the Cranfield
WSTC0037 Progress Internal Reports for the “Binders by Design Project” and
was presented in progress meetings with the PhD sponsor.

1.3 Thesis structure


Following this overview, Chapter 2 (Introduction) guides the reader through the
concept of binders and their use in energetic formulations, followed by a
summary of current knowledge about the chemical properties of cyclodextrins,
particularly their use in the energetics industry. Chapter 3 (Experimental)
describes the materials, reactions and methods used to synthesise and
characterise the compounds, as well as test conditions for hazard assessment.

Chapter 4 (Synthesis and characterisation of inert βCXCDs) explains the


rationale behind the optimisation of the first synthesis reaction, and also
describes the physicochemical and thermomechanical characterisation of the
inert compounds. Chapter 5 (Synthesis and chemical characterisation of
energetic cross-linked βNCXCDs) mirrors the structure of Chapter 4 but
describes the production and characterisation of the energetic cross-linked
systems. Chapter 6 (Compatibility, hazard and stability studies) provides the
data needed for the reader to understand the contribution and limitations of the
novel inert βCXCDs and energetic βNCXCDs in the development of new
energetic formulations.

Finally, Chapter 7 (Conclusions and recommendations) provides a summary of


the findings reported in the thesis and suggests additional work required to
develop this cross-linked system into a successful binder for energetic
applications.

7
The appendices contain additional experimental details that will help the reader
to understand the research, as well as the published articles based on the work
described in the thesis:

 Luppi F, Cavaye H, Dossi E (2018) Nitrated cross-linked β-cyclodextrin


binders exhibiting low glass transition temperatures. Prop Expl Pyrotech 43
(10) 1023–1031.
 Luppi F, Kister G, Carpenter M, Dossi E (2019) Thermomechanical
characterisation of cross-linked β-cyclodextrin polyether binders. Polym
Test 73, 338–345.

8
2 Introduction
2.1 Insensitive munitions
Energetic formulations used in the defence industry must withstand unintended
stimuli to prevent unwanted violent events, and manufacturing is therefore
moving towards improved safety during phases of life. The development of
insensitive munitions (IMs) is a direct response to the several accidents that
caused the involuntary initiation of explosives [1]. A definition developed in the
American Navy stands [1]:

“Insensitive munitions are those that reliably fulfil their performance, readiness,
and operational requirements on demand, but are designed to minimize the
violence of a reaction and subsequent collateral damage when subjected to
unplanned heat, shock, fragment or bullet impact, electromagnetic pulse (EMP),
or other unplanned stimuli.”

The development of IMs is guided by the handbooks STANAG 4439 and the
American standard MIL-STD-2105 “Hazard Assessment Tests for Non-Nuclear
Munitions”, which can be summarised in the following three strategies [2–4]:

1. The development of external packaging that improves the protection of


the munitions.
2. The development of mitigation devices for uncontrolled events in the
munitions.
3. The development of energetic materials with an intrinsic low vulnerability.

The first strategy relies on the design of new transport tools such as containers
better insulated from thermal and kinetic shock, or coated with intumescent
paint.

The second strategy is the development of passive or active devices to mitigate


the effects of uncontrolled events. Passive devices do not provoke a response
in the energetic materials, and examples include vent plugs, thread adaptors,
mitigation sleeves and intumescent paint [5–7]. Not as widely used, active
devices such as the thermal initiated venting system (TIVS) cause a controlled

9
thermal response to external stimuli in the IM, and thus reduce the severity of
unplanned events [8].

Energetic materials with low vulnerability intrinsically prevent undesired


consequences. The two global conflicts in the first half of the twentieth century
increased the development of novel energetic materials. Polymers were
introduced in energetic formulations as a medium to bind the explosive
ingredients, thus increasing the compactness of the munitions and reducing
their vulnerability [9]. In this role, polymers were used and defined as binders.

Polymers have a wide range of physicochemical properties that define their


efficiency in energetic applications [10].

2.2 Polymers and their properties


Polymers are macromolecules comprising repeat units of one or more
monomers, and they offer unique benefits when used as binders, including [11]:

 Wide variety of molecular architectures, e.g. linear, branched,


homopolymer, copolymer, random, alternating, star, block copolymer,
cross-linked;
 Dimensional stability (linear dimensional change in response to
temperature changes);
 Durability (lifetime, ageing, degradation, stabilisation);
 Easy processing (flow, forming, solidification by moulding, extrusion).

The chemical identity of a polymer defines its morphology and the physical
properties of the system. The synthesis route determines the molecular
structure and is designed to avoid side-reactions, thus favouring the recovery of
pure products. By controlling the reaction mechanism, the system can be
defined in terms of the regularity of the backbone (stereochemistry, sequencing
in copolymers), the molecular weight and polydispersity, and the polymer
architecture. Most of these molecular structures offer numerous opportunities
for functionalisation, process control and thus new applications.

10
Polymer morphology describes the arrangement and macroscale ordering of
polymer chains in space, which affects properties such as strength, toughness
and flexibility [12]. Crystallinity is often used to describe the morphology of a
polymer compound. Crystalline polymers are rigid and melt at higher
temperatures than amorphous polymers. Crystallinity increases mechanical
strength at the cost of low impact resistance. Amorphous polymers are softer,
with lower melting points than crystalline polymers, and there is a transition
point at which the polymer reversibly changes from glassy to rubbery behaviour
[13]. This is known as the glass transition temperature (T g) [12]. The glassy
state is hard, rigid and brittle like a crystalline solid, but retains the molecular
disorder of a liquid. When the material is heated the polymer will reach a
temperature at which segments of the entangled chains can move and the
behaviour will change to rubbery [11].

There are two general classes of polymers based on their behaviour when
exposed to heat: thermoplastics and thermosets [14]. Thermoplastic polymers
have relatively weak forces of attraction between the macrochains (electrostatic
forces, few cross-links), which are overcome when the material is heated [14].
Thermoplastics are permanently fusible and can always be softened, melted
down and recycled. When frozen, a thermoplastic becomes glassy and more
sensitive to fracture [14]. Thermoplastics have a distinctive stress–strain curve
(elongation to break). Thermosets comprise molecules that are cross-linked by
strong bonds during curing due to chemical reactions or the effects of heat or
radiation [14]. Effectively, the thermoset is one large molecule, with no
crystalline structure. Cured thermoset polymers do not soften and will char and
break down at high temperatures. Thermosets are generally harder than the
thermoplastics, more rigid and more brittle, and their mechanical properties are
not heat sensitive. They are also less soluble in organic solvents [15].

Thermoplastic elastomer (TPE) materials combine the functional performance


and properties of thermoset rubbers with the processability of thermoplastics.
TPEs are generally low-modulus, flexible materials that can be stretched
repeatedly to at least twice their original length at room temperature with an

11
ability to return to their approximate original length when stress is released [16].
TPEs are widely used to modify the properties of rigid thermoplastics, improving
impact strength. This is quite common for sheet goods and general moulding
TPEs [16]. Their soft texture, ease of manufacture and ability to absorb shocks
are preferred for energetic applications. Several classes of polymers can be
cross-linked in the defence industry to produce TPEs, including copolyesters,
styrenes, polyurethanes, polyamides and polyolefin blends [17]

2.3 Polymers as binders in energetic formulations


The binder is an important component of munition fillings because it holds the
explosive crystals together, allows the formulation to be safely casted or
machined into complex shapes, and reduces its vulnerability to accidental
stimuli during storage. Table 2.1 lists some examples of energetic formulations.

The properties of the binder strongly influence the physical properties of the
formulation and its applications. The plastic behaviour of certain polymers
allows the formulation of plastic explosives, which are malleable and can be
hand-moulded [18]. In contrast, viscoelastic behaviour reduces vulnerability to
mechanical stress by facilitating energy dissipation in large-calibre munitions
and nuclear warheads [19].

The chemical properties of the binder influence the stability of the formulations.
For example, the low chemical reactivity of fluoropolymers can protect the
formulations from harsh chemicals. For this reason, fluoropolymers are
compatible with many energetic ingredients and reduce the likelihood of
unpredicted chemical reactions [20]. The nitro-ester groups of NC undergo
autocatalytic decomposition, so stabilisers are required to ensure a sufficient
shelf-life [21]. Furthermore, epoxy-amines are incompatible with nitramines
such as 1,3,5-trinitro-1,3,5-triazinane (RDX) and 1,3,5,7-tetranitro-1,3,5,7-
tetrazocane (HMX) [22].

12
Table 2.1 List of binders in energetic formulations [23].

Binder1 Energetic Energetic


Application
Name % Name % formulation
Plastic
Polyisobutene 2.1 RDX 91 C4
explosives
RDX 63
Wax 1 Gun shells Comp. B
TNT 36
Nitrocellulose 8 Engineer
PETN 63 Detasheet C
elastomer 29 charges
Hydroxyl-terminated
polybutadiene Nuclear
5 HMX 29 EDC-29
warheads
(HTPB)
2,4-Dinitroanisole NTO 19.7
43.5 Gun shells IMX-101
(DNAN) NQ 36.8
NTO 53
DNAN 31.7 Mortars IMX-104
RDX 15.3
Nuclear
Viton-A 10 HMX 90 LX-07-2
warheads
< Nuclear
Polyurethane rubber HMX 95.5 LX-14-0
4.5 warheads
Nuclear
Kel-F 800 7.5 TATB 92.5 LX-17-0
warheads
Nitrocellulose
3 Nuclear
chloroethylphosphate HMX 94 PBX 9404
3 warheads
(CEF)
Nuclear
Kel-F 8003 5 TATB 95 PBX 9502
warheads
Missiles
Nylon n.a. RDX 85 PBXN-3
Warheads
Fluoro-elastomer 5 HMX 95 Gun shells PBXN-5
HYTEMP 44542 2 HMX 92 Various PBXN-9
Plastic
HTPB 11 RDX 88 PE7
explosives
NC,NQ, Single,
Solid rocket
NC n.a. / Double or
NG propellant
Triple base
Solid rocket AP/HTPB
HTPB n.a. AP /
propellant propellant
Styrene butadiene PETN 40.9 Plastic
9 Semtex H
rubber RDX 41.2 explosives
1 The percentage of binder varies according to the application.
2 Trade name of polyacrylate binder.

13
3 Trade name of fluoro-elastomers.

The constituents of energetic formulations must be compatible to avoid side


reactions. The introduction of polymers used for other applications in the
development of new energetic formulations can be slow, hence the rather small
number of polymers used successfully in energetic applications (Table 2.2).

NC was the first compound to be used in energetic formulations following the


invention of black powder. It was first synthesised in the nineteenth century, as
recorded by Christian Friedrich Schonbein in Basel and Rudolf Christian Bottger
in Frankfurt-am-Main [20]. A significant milestone in the defence applications of
NC was the production of the smokeless propellant Poudre B, by Paul Vieille in
1884 [24], where it was used as binder. A thick jelly was produced by kneading
NC in a mixture of ether and alcohol to produce thin films. The NC gave solidity
to the processed propellant and improved the control of its burning velocity.
However the compound shows limited elongation below room temperature [25].

The moderate cost of NC makes it difficult to replace in defence applications


even though the geographic and seasonal variations of raw cellulose mean that
expensive selection processes are needed every few years to comply with
military standards such as MIL-DTL-244 [26]. Special considerations are
required to ensure batch-to-batch consistency in the polymer architecture [27].
In addition, it is difficult to prevent the degradation of this compound [28].

The escalation of European conflicts in the second half of the twentieth century
stimulated the development of new binders. Polysulfides (PS) were the first
synthetic polymers used as binders for propellants by the Thiokol Chemical
Company in 1942 [29], although they are incompatible with the strong oxidisers
used in many propellants and pyrotechnics [30].

Polybutadienes were developed in the 1950s to replace PS, and are more
compatible with oxidisers in propellants. However, polybutadiene-acrylic acid
copolymer (PBAA) showed low reproducibility during propellant manufacturing
due to the synthetic nature of the radical polymerisation used to produce the
binder [30]. Polybutadiene was copolymerised with acrylonitrile and acrylic acid

14
(PBAN) to achieve better pulse from the rockets, but a high curing temperature
was needed for these binders [31].
Table 2.2 Example of polymers used in energetic applications [20].

Polymeric binder Chemical structure Applications


Propellants,
high explosives,
1888
Nitrocellulose

Propellants,
Polysulfide
1945

Polybutadienes Propellants

Poly butadiene-
acrylic acid
copolymer 1954
(PBAA)

Poly butadiene-co-
acrylonitrile 1954
(PBAN)

Carboxyl-
terminated
polybutadiene Late 1950s
(CTPB)

Propellants,
Polyurethane
mid 1950s

Hydroxyl- Propellants and


terminated high explosives,
polybutadiene 1972
(HTPB)

15
High
explosives,
Fluoropolymers
pyrotechnics,
propellants
Early 1960s

Kel-F 3700

Early 1960s
Teflon

Early 1960s

Viton

Thermoplastic
Propellants
elastomers (TPEs)

Estane 5703 [32] Early 1980s

Kraton Early 1980s

The development of new binders has been slow but continuous, although many
end up in non-defence applications.

Cross-linked binders are manufactured by reacting bifunctional or


multifunctional molecules such as isocyanates or polyols with these polymers to
create a toughened three-dimensional (3D) network, a process defined as
curing.

16
Hydroxyl-terminated polybutadiene (HTPB) is currently the most widely-used
polymer for the preparation of polyurethane elastomer binders with unique
properties such as low-temperature flexibility, hydrophobicity and resistance to
hydrolysis [20,33,34]. However, with the exception of NC, the polymers listed in
Table 2.2 do not contribute any energy to the formulation and negatively affect
its energetic output. Energetic polymers, in which the binder contributes to the
energy of the formulation, were developed to mitigate this major drawback
(Table 2.3). The energetic contribution is provided by functional groups that
decompose exothermically such as nitro, azido or nitro esters [35]

Table 2.3 Examples of energetic polymers used as binders [20].


Polymeric binder Chemical structure Applications

Polyglycidyl
Propellants, 1953
nitrate (PGN)

Glycidyl azide
polymer Propellants, 1976
(GAP)

Propellants, high
Polyoxetanes
explosives, 1984

3,3-
Bis(azidomethyl)
oxetane polymer
Poly(BAMO)

3-Azidomethyl-3-
methyloxetane
polymer
Poly(AMMO)

17
Polymeric binder Chemical structure Applications

Poly(3‐
nitratomethyl‐
3‐methyloxetane)
Poly(NIMMO)

Nitrated hydroxyl-
terminated
polybutadiene Propellants, 1980
(NHTPB)

Polyphosphazenes
Propellants, 2005
(PPZ)

Energetic
thermoplastic
elastomers
(ETPEs)

BAMO-GAP

Propellants, 1984

BAMO-AMMO

BAMO-NIMMO

18
Polymeric binder Chemical structure Applications

Nitrated High explosives,


cyclodextrin propellants,
polymers early 1990s

Most recent studies have focused on polyphosphazenes and energetic


cyclodextrin (CD) derivatives, both of which produce feasible energetic binders.
Polyphosphazenes are non-isocyanate curable energetic polymers which are
less toxic than current binders [36,37]. A successful synthetic curing process for
HTPB and polyphosphazenes using a non-toxic cross-linker was developed at
Cranfield University, based on the in-house diepoxide triethylene glycol
diglycidyl ether (TEGDGE), as shown in Figure 2.1 [12,18,28].

Figure 2.1 Chemical structure of triethylene glycol diglycidyl ether

The potential of CDs and their derivatives for energetic formulations is


discussed in the next section. The physicochemical properties and the ability of
CDs to envelop molecules is highlighted as means to reduce the vulnerability of
energetic formulations.

2.4 Cyclodextrins
CDs are cyclic organic molecules characterised by the ability to envelop smaller
molecules in their inner cavity. The three best-known CDs are α-cyclodextrin
(αCD), β-cyclodextrin (βCD), and γ-cyclodextrin (γCD), which are toroidal
macrocycles of six, seven and eight glucopyranose units, respectively (Figure

19
2.2) [38]. CDs with larger rings have been isolated but they are not used in the
defence industry [39] .

Figure 2.2 Chemical structure of α, β and γ CDs [40].

CDs were first synthesised in 1891 by Villiers and Schardinger. The work was
then picked up by Pringsheim and colleagues, who discovered that CDs had the
tendency to envelop organic molecules, leading to a burgeoning body of
research literature during the 1970s [38].

This early research showed that α, β and γ CDs were suitable for
pharmaceutical applications due to their enveloping properties [40]. However
misleading information about their toxicity was publicised, reducing interest in
these macrocycles. Modern toxicological tests show that CDs lack oral toxicity
because they are not absorbed in the intestine [41]. Most publications
concerning CDs focus on pharmaceutical applications and different
functionalised variants are commercially available, including randomly
methylated βCD (RAMEB), hydroxyalkylated derivatives (hydroxypropyl-βCD
and hydroxypropyl-γCD), acetylated-γCD, reactive derivatives such as
chlorotriazinyl-βCD, and branched versions such as glucosyl-βCD and maltosyl-
βCD [38,40,42,43]. Figure 2.3 shows in the distribution of CD publications by
application field over the last 20 years.

20
Figure 2.3 Distribution of the CD publications by application (1996–2013) according to
CD News [38].

Most publications have considered pharmaceutical applications, where modified


or unmodified CDs are used for drug delivery. The pharmaceuticals sector
consumes several tons of medical-grade CD per year, with high purity and high
development costs. There are smaller numbers of publications concerning
applications in the food, cosmetics and detergent industries, which require a
greater quantity of CDs but with less-stringent purity requirements and minimum
prices of 6–7 US$ per kg in the US and European markets [38], and 2–3 US$ in
the Chinese market. In Europe, Merck holds the bigger share of the CD market.

2.4.1 Properties of CDs


In the 1940s, the structures of CDs were defined by X-ray crystallography,
revealing their ability to trap small molecules and to form inclusion complexes.
The structural data obtained in these early studies showed how the primary
hydroxyl groups cluster at one end of the toroidal molecules, whereas the
secondary hydroxyl groups lie at the other end, as shown in Figure 2.4
(coloured in blue). The non-polar hydrogen atoms attached to C2 and C3 fill the
inner side of the cavity along with the ether group. This leads to hydrophilic
behaviour outside the toroid lipophilic cavity [40]. The physical properties of
CDs are summarised in Table 2.4.

21
Figure 2.4 Structures of (a) α-D-glucopyranoside, the repeating unit of CD, and (b) the
toroid representation of CDs, with primary and secondary hydroxyl groups in red and
blue, respectively. H1, H2 and H4 are external to the toroid, whereas H3 and H5 are
internal

Table 2.4 The physical properties of α, β and γ CDs [38].

CD α β γ
Number of glucose units 6 7 8
Molecular weight 972 1135 1297
Solubility in water, g 100 mL-1 at
14.5 1.85 23.2
room temperature
Cavity diameter, Å 4.7–5.3 6–6.5 7.5–8.3
[α]D 25 °C 150 ± 5 162 ± 0.5 177.4 ± 5
Height of torus, Å 7.9 ± 0.1 7.9 ± 0.1 7.9 ± 0.1
Diameter of outer periphery, Å 14.6 ± 0.4 15.4 ± 0.4 17.5 ± 0.4
Approx. volume of cavity, Å3 174 262 427
Approx. cavity volume in 1 mol CD
104 157 256
(ml)
Monoclinic
Hexagonal Quadratic
Crystal forms (from water) parallelogram
plates prisms
s
Crystal water (% w/w) 10.2 13.2–14.5 8.13–17.7
Diffusion constant at 40 °C 3.443 3.224 3.000
pK (by potentiometry) at 25 °C 12.332 12.202 12.081

22
The thermal properties of CDs, such as phase transitions or mass changes
during chemical reactions, have been studied by thermal analysis using
techniques such as differential scanning calorimetry (DSC) and thermal
gravimetric analysis (TGA). The literature reports slightly different thermal
behaviours among the three cyclodextrins [44]. The main phase transitions and
thermal events for the three CDs are summarised in Table 2.5.

Table 2.5 Thermal properties of α, β and γ CDs.

Water loss Phase transition Degradation


Cyclodextrin
(°C) (°C) (°C)

αCD 74–121 224 319–304


βCD 87–110 221 328–330
γCD 63–81 - 321–324

CDs are generally stable up to 300 °C, and degradation occurs at higher
temperatures. CDs are hygroscopic and can entrap water molecules in their
cavity, which is described as crystal water in Table 2.4. The high temperatures
cause phase transitions in the structures of α and β CDs at 224 and 221 °C,
respectively. The first part of the CD thermal profile is dominated by the loss of
hydration water and phase transitions in the crystal structure due to this loss
[45]. Among the three CDs, the water of hydration has the highest mobility in
γCD and is released earlier compared to the others [44], even though γCD is
the most soluble (Table 2.4).

The solubility of all CDs in water increases at higher temperatures, although


βCD is less soluble than the others [42]. The differences in solubility reflect the
different hydrogen bond interactions that occur between the hydroxyl groups of
each glucopyranose unit and the flexibility of the corresponding CD rings [46].
The inner cavity of αCDs cannot accommodate very large molecules due to its
small size (Table 2.4), but this is not the case for β and γ CDs, which are large
enough to incorporate various chemical compounds, including linear polymers
and aromatic rings [41]. In the less common and non-commercialised CDs (10,

23
14, 26, 32 repeat units) the spatial distortion causes cavity collapse, reducing
the inner volume to less than the value observed for γCDs, making them less
suitable for encapsulation [47].

The enveloping structures formed by CDs in solution are influenced by the


presence of solutes that could act as guest molecules. There is great interest in
these compounds, reflecting their ability to form different supramolecular
threads, allowing CDs to be used like polyrotaxanes (Figure 2.5) [48,49]. The
interest in developing these supramolecular structures has led to more specific
studies focusing on forces that control guest–host molecule interactions.

Figure 2.5 Schematic representation of a polyrotaxane formed by several CDs


threaded on a polymer chain with ending caps.

2.4.1.1 Cyclodextrin inclusion complexes

CD inclusion complexes are affected by two main parameters:

 Steric hindrance;
 Replacement of solvent molecules with guest molecules.

Steric hindrance depends on the relative sizes of the guest molecule and the
cavity, whereas the likelihood of replacement is based on thermodynamics. The
inclusion of guest molecules by CDs involves an enthalpy-dependent driving
force, i.e. water molecules inside the cavity have fewer interactions with the
hydroxyl groups than water molecules in the bulk solution [50]. The replacement
of water molecules with an organic molecule can therefore increase the stability,
resulting in the dissolution of a sparingly soluble or insoluble organic molecule

24
in water. The organic molecule/CD complex is often less soluble than the CD
alone [38].

Encapsulation tends to reduces the volatility, diffusion and reactivity of guest


molecules [38,42,51,52]. The thermal properties of guest compounds are also
affected, leading to increases in the glass transition and/or melting
temperatures when embedded in CD [53]. The encapsulation ability of CDs has
also been used to study their interactions with energetic molecules, as
discussed later in this section.

Complexes may form in the solid state or in solution, following processes such
as kneading, co-precipitation, dry mixing, sealing, slurry complexation,
neutralisation, spray drying, freeze-drying, and solvent evaporation This usually
leads to an inclusion ratio of 1:1 for host and guest molecules [42,54,55]. The
temperature affects the solubility of the complex as well as the stability of the
interaction [52]. The solubility of the complex is often inferior to the pure CD,
causing a fraction of the complex to precipitate when formed [56–60].

Most of these processes have only been tested at the laboratory scale, and they
have been claimed in various patents covering the manufacture of CD polymers
or derivatives. Many of the patents covering the manufacture of natural CDs
and modified variants such as hydroxypropyl-βCD have now expired, although
existing and novel CD polymer derivatives are still manufactured [41,61,62].

2.4.1.2 Chemical stability and characterisation of CDs

CDs are chemically stable compounds, hence their use as drug carriers. They
are relatively resistant to oxidation, with only strong oxidising environments
leading to ring-breaking reactions. CDs are more acid resistant than linear or
branched polysaccharides, but treatment with strong acids causes ring opening
and the formation of oligosaccharides. The magnitude of the ring opening rate is
influenced by the size of the CD molecule, so the process is slower for αCDs
than γCDs [40].

25
In alkaline environments, the CD ether bond is stable towards nucleophilic or
acid-base reactions and requires very strong conditions to react [40,42,59]. This
low reactivity is also conferred to guest molecules within the CD ring.

Nuclear magnetic resonance (NMR) spectroscopy is widely used to


characterise CDs because it determines their chemical structure in a complex.
Figure 2.6 shows the proton assignments typical in deuterated water (D2O) and
the assignment in deuterated dimethylsulfoxide (DMSO-d6).

26
Table 2.6 and Table 2.7 summarise the chemical shifts at 400 MHz reported for
1
H and 13C NMR, respectively [63]. The signals attributed to the protons are split
into two main regions. The region from 3.8 to 3.2 ppm includes most of the
signals due to the chemical shifts of the aliphatic protons of the CD H-2 to H-
6a,b, whereas the region from 5.8 to 4.4 ppm includes the chemical shifts of the
anomeric proton H-1 (Figure 2.6a). When DMSO-d6 is used, the chemical shifts
of the hydroxyl groups of the CD are also visible (Figure 2.6b). This region can
be integrated and compared with the integration of H-1 to determine the degree
of substitution (DS) that has occurred with the hydroxyl groups. The DS can
also be determined by titration of the hydroxyl groups following standard
methods [64–66].

Infrared (IR) spectroscopy is also widely used to characterise CDs [57,59,67–


70]. The IR spectrum and the main adsorption bands for βCD are shown in
Figure 2.7. A strong adsorption band at 3400–3300 cm–1 is due to the stretching
of hydroxyl groups. The peak at 2926 cm–1 is assigned to the stretching of the
C-H bond in CH and CH2 groups. The peaks at 1200–1400 cm–1 are assigned
to the bending of C-H in the primary and secondary hydroxyl groups of βCD
(1411, 1368, 1335, 1301 and 1246 cm –1) [71]. The strong adsorption at 1030–
1200 cm–1 is attributed to the stretching of the C-O-C bonds in the ether and
hydroxyl groups. Finally, the absorption bands at 700–950 cm–1 show the
bending vibrations of the C-H bonds and the stretching vibrations in the
glucopyranose ring [72]. The peak at 1626 cm–1 is actually due the water
absorption included in the CD molecule as reported for the spectrum of water
elsewhere [57,73].

27
Figure 2.6 The 1H-NMR spectra of CD (a) in D2O and (b) in DMSO-d6 [63].

28
Table 2.6 The 1H-NMR chemical shifts of CD protons in D2O and DMSO-d6 [63].

Cyclodextrin δH-1 δH-2 δH-3 δH-4 δH-5 δH-6a,b


D2O
αCD 4.60 3.19 3.57 3.08 3.39 3.44
βCD 4.68 3.26 3.58 3.19 3.47 3.49
γCD 4.53 3.08 3.35 3.00 3.26 3.30
DMSO-d6
αCD 4.79 3.29 3.78 3.40 3.59 3.65
βCD 4.82 3.29 3.64 3.34 3.59 3.64
γCD 4.89 3.32 3.65 3.36 3.56 3.65

Table 2.7 The 13C NMR chemical shifts of CDs in D2O [63].

Cyclodextrin δ C-1 δ C-2 δ C-3 δ C-4 δ C-5 δ C-6


αCD 102.19 72.61 74.21 82.07 72.91 61.37
βCD 102.58 72.67 73.89 81.94 72.89 61.17
γCD 102.42 73.19 73.82 81.33 72.69 61.21

Figure 2.7 The infrared spectrum of βCD.

29
2.4.2 Synthesis of cyclodextrins
The enzyme CD glucosyltransferase (CGTase) converts starch into CDs. It is
expressed by bacteria such as Paenibacillus macerans, Klebsiella oxytoca,
Bacillus circulans, and alkalophilic Bacillus species [38,42,74]. CD synthesis
begins with the liquefaction of starch by heating and hydrolysis to reduce the
viscosity of the raw material. The enzyme is then mixed with the starch and
different solvents are used as complexants to obtain the three CDs: toluene is
used for βCD, 1-decanol for αCD, and cyclohexadecenol for γCD. The CDs are
separated from the mixture by filtration, washed with water, and the organic
solvent is removed by distillation or extraction in water. The resulting aqueous
solution is treated with active carbon to remove trace impurities and filtered. The
cost of production depends on how easy it is to remove the CDs from the
organic solvent [38].

2.4.3 Cross-linking of cyclodextrins


CDs can be grafted to polymer chains, can form complexes with the branches
of an existing polymer, or can be entrapped in a polymer lattice such as the
polyrotaxane configuration (Figure 2.5). Polymeric CDs are formed primarily
due to a reaction between the primary and/or secondary hydroxyl groups with
functional groups of other molecules, or due to guest–host interactions. CD is
often cross-linked using a variety of cross-linkers to form 3D structures that are
usually insoluble in common solvents (Table 2.8).

The first cross-linked CDs were formed by reacting monomers with


epichlorohydrin under basic conditions. Changing the reaction parameters (pH,
solvent, temperature and time) led to different degrees of cross-linking [38]. The
solubility of the resulting products varies depending on the cross-linker and
reaction parameters [59,75,76]. The ability of polymers to swell depends on
whether the polymer is linear [77] or branched [71]. The synthesis of different
3D networks leads to different degrees of swelling and chiral properties [75].
Many cross-linked CDs are insoluble due to their high molecular weight (Mw),
so characterisation is restricted to methods such as solid-state NMR or Fourier
transform infrared (FTIR) spectroscopy [71,78]. Other synthesis routes include

30
the living polymerisation of CDs to form the core of thermosensitive polymers
[79].

Table 2.8 Applications of cross-linked CDs

Studies Cross-linker used Ref.


1,2,3,4-butanetetracarboxylic
Removal of bisphenol A from water [71]
dianhydride

4,4′-
Removal of pollutants such as
methylenebis(cyclohexyl
2-methylisoborneol and chlorinated [80]
isocyanate), toluene
disinfection by-products from water
diisocyanate
Absorption of phenol, methylene
Citric acid [68,70]
blue and aniline
Absorption of methylene blue and
Citric acid [81]
phenylalanine
Inclusion of aromatic compounds Epichlorohydrin [59]

Removal of urea from water Epichlorohydrin [51]


Manufacture of chromatographic
Epichlorohydrin [75] [77]
stationary phase
Ethylene glycol diglycidyl
Removal of bisphenol A from water [78]
ether
Solubility studies of the synthesised
Poly(carboxylic acids) [76]
polymers
Poly(ethylene glycol
Removal of bisphenol A from water [78]
diglycidyl) ether
Poly(N-isopropylacrylamide)
Synthesis of thermosensitive star
and poly(N,N- [79]
copolymers
dimethylacrylamide)
Manufacturing of antimicrobial
Poly(vinyl alcohol) citric acid [82]
packaging

Increasing the mechanical


resistance of microneedles for drug Poly(vinylpyrrolidone) [83]
injection
Control of drug release Polyethyelene glycol [84–88]

31
Cross-linked CD polymers have been extensively used to remove chemicals
from solution as reported in Table 2.8. The advent of green chemistry led to the
production of different water-insoluble CD polymers using citric or
polycarboxylic acids as environmentally friendly cross-linkers for wastewater
treatment [68,70,76,81]. The same cross-linking technique has been used to
manufacture antimicrobial packaging [82]. The thermal characterisation of these
compounds indicated that the presence of CDs increases the Tg of any
polymers to which they are grafted. The glass transition is defined by IUPAC as
the “Process in which a polymer melt changes on cooling to a polymer glass or
a polymer glass changes on heating to a polymer melt” [89]. Long-chained
cross-linkers are necessary to reduce the Tg of cross-linked CDs [82].

The synthesis of non-toxic materials is necessary when cross-linked CDs are


used for drug delivery [40]. Many reactants produce non-toxic compounds when
combined with CDs, expanding research into the synthesis of hydrogels for drug
delivery. The solubility, pH response and membrane permeability of these
systems can be tuned by adjusting the reaction conditions, making them ideal
carriers to bind specific drug molecules in pharmaceutical applications [90].
Hydrogels are insoluble matrices of polymers that can adsorb large amounts of
water, which cause them to swell [90]. This specific ability is valuable in other
fields such as wastewater treatment, to trap pollutants [91].

One of the most widely-used hydrogels is based on CDs cross-linked with


polyethylene glycol based cross-linkers [68,78,84–88,92]. The mechanical
properties of hydrogels are also beneficial in pharmaceutical applications. The
hydrogel must maintain its integrity to deliver the drug cargo to a specific target
in the human body. Several hydrogels based on CDs can self-heal, which
means they can at least completely or partially regenerate their mechanical
strength after damage [93–95], and many of these are CD/PEG systems
[92,96,97].

Self-healing is a characteristic possessed by many polymeric systems, in which


a material can restore its structural integrity in a damaged area [98]. Different

32
polymers rely on different self-healing mechanisms that involve three different
processes:

 Irreversible covalent bonds;


 Reversible covalent bonds;
 Reversible intermolecular interactions.

The first two processes require the presence of an encapsulated monomer


reservoir in the polymer bulk. When damage occurs, the capsules break and
release the healing agent into the fracture, followed by cross-linking to repair
the damage. This relies on the formation of reversible or irreversible covalent
bonds at sites where stress has caused the cleavage of bonds in the polymer
network. Examples of irreversible mechanisms include ring-opening metathesis
polymerisation (ROMP) and click chemistry [99], whereas reversible
mechanisms include cross-linked poly(dimethylsiloxane), retro Diels-Alder
reactions, and deformable/reformable disulfide and alkoxyamine bonds in
response to stimuli such as temperature or irradiation [100].

The healing behaviour of CD/PEG systems is based on reversible


supramolecular interactions, where various intermolecular forces can drive the
self-healing process:

 Inclusion complexes form between the cross-linker and CD [95,96,101];


 Electrostatic interactions and intermolecular hydrogen bonding
[50,94,102];
 Mobility of the polymeric chains in the bulk (reptation) [92,103,104];
 Mobility of the polymeric chains in the water of the hydrogel [105,106].

The combined effect of these driving forces allows the CD/PEG systems to
heal, which is useful in applications that subject the materials to mechanical
stress. In addition, the self-repairing ability of CD/PEG systems is enhanced
when the compounds are exposed to heat due to the increased mobility of the
polymer chains in water. Compounds that heal in response to external stimuli
such as temperature or mechanical pressure are defined as healable or
amendable [98,106].

33
2.4.4 Energetic applications of unlinked/cross-linked cyclodextrins
The ability of CDs to trap energetic molecules has been investigated in the past
for the treatment of military waste streams [67,107]. A list of studies involving
the energetic applications of CD is provided in Table 2.9.

Table 2.9 Applications of CDs functionalised with energetic molecules.

Application CD type Energetic Ref


Trinitrotoluene

Removal of (TNT)
βCD [67]
TNT from soil

Removal of
βCD
RDX from soil
Cyclotrimethyl
ene-
Inclusion [108,109]
βCD
studies1 trinitramine
(RDX)

High explosive βCD,


manufacturing γCD

1,3,3-
Trinitroazetidin
αCD e
Inclusion
βCD, (TNAZ) [110,111]
studies2
γCD

Inclusion βCD, Cyclotetrameth [110]


studies1 γCD ylene-
tetranitramine
(HMX)

34
Application CD type Energetic Ref
Inclusion γCD Hexanitrohexa [112,113]
studies3 azaiso-
wurtzitane
(CL-20)

βCD Trimethyloleth [114]


Manufacturing ane trinitrate
of gun
(TMETN)
propellants
with nitrated
βCDs
(βNCDs)

1
1 H-NMR solution studies.
2 Characterisation by IR spectroscopy and DSC.
3 Characterisation by IR spectroscopy.

CDs can be used as molecular sieves to remove nitroaromatic compounds such


as TNT and RDX from soil. Dilute aqueous solutions of hydroxypropyl-βCD and
methylated-βCD were used to absorb TNT from soil, achieving the removal of at
least 25% of the pollutant. These studies directly demonstrated that βCDs can
interact with and envelop aromatic explosive compounds [67,108].

The encapsulation of TNAZ, RDX and HMX has been attempted using α, β and
γ CDs. Interactions were characterised by 1H-NMR spectroscopy and showed
that the degree of interaction was relative to the size of the macrocycle and its
cavity, and the size of the molecule to be encapsulated. HMX only interacted
with γCD (having the larger cavity), whereas RDX interacted with both βCD and
γCD. In contrast, αCD did not interact with any of the energetics due to its
relatively smaller cavity size. TNAZ formed a solid complex with βCD, that was
also characterised by DSC, solid-state NMR, and Raman spectroscopy [111].
The physicochemical properties of this complex differed from those of the
physical mixture: the volatility of TNAZ was reduced and the CD thermal
decomposition profile was less exothermic, but the thermal stability of TNAZ
was not affected by the complexation [111]. Other attempts to encapsulate

35
larger molecules such as CL-20 have been successful with the larger γCD, but
not with βCD [112]. A more recent study showed that complexation also occurs
between CL-20 and the nitrated derivative of γCD [113]. The inclusion studies
offered insight into the ability of CDs to trap energetic molecules.

In addition, βNCD and βNCD cross-linked with epichlorohydrin and isocyanate


have been used to develop energetic formulations that are less vulnerable to
accidental stimuli [109,114]. These patents led to the registration of a warhead
featuring nitrated cross-linked CDs as part of the main charge [115]. The first
patent [109,114] claimed the use of βNCD to encapsulate trimethylolethane
trinitrate (TMETN). Different nitration levels are possible because not all of the
21 βCD hydroxyl groups need to be nitrated. The level of nitration affects the
energetic performance and physicochemical properties of the βNCD. When
more than three hydroxyl groups are nitrated to form nitro groups, the βNCD
becomes insoluble in water but remains soluble in organic solvents such as
acetone, methylene chloride or tetrahydrofuran (THF) [116]. Consequently the
βNCDs can be processed as acetone lacquers. The inventors tested different
reaction conditions to obtain βNCDs that are listed in Table 2.10.

Table 2.10 Summary of βCD nitration conditions [114].

CD1 90% HNO3 98% H2SO4 Time Extent of


(g) Mass (g) (g) (min) nitration (%)
200 1296 - 15 -2
4803
50 291 -2 -2

750 2650 550 120-180 12.96


1 Reaction temperature < 30 °C.
2 Values not reported.
3 Oleum.

The βNCD/TEMTN formulations were then used in hazard tests to determine


their sensitivity towards accidental stimuli. Hazard tests are systematic studies
to quantify the accidental initiation of energetic compounds using standardised
methods [117]. Such methods were developed independently by several
countries in the nineteenth century and the results differ in the way the

36
magnitude of the sensitivity is reported, which makes it difficult to compare
studies carried out in different countries [117]. The formulated products were
compared to pure TMETN and βNCD (Table 2.11), but no information on the
methodology was reported, although the compounds were likely to have been
tested under American standards.

The tests demonstrated that the addition of βNCD to TMETN reduced the
vulnerability of almost all the all formulations to shock, with only the 6:1 ratio
TMETN/βNCD formulation becoming more susceptible to impact than the single
ingredients. The addition of βNCD as a binder to TMETN did not improve the
thermal stability of the mixtures. The temperature of decomposition (T dec) of the
single ingredients was not reported, but the Tdec of the formulations was 4 °C
lower than the Tdec of TMETN and 20 °C lower than the Tdec of βNCD as
reported elsewhere. The sensitivity of βNCD to electrostatic discharge (ESD)
was mitigated by the formulation with TMETN up to the safe value of 12.5 J.
The TMETN/βNCD 2:1 propellant cured with R45M isocyanate was described
as a flexible gum, but its characterisation was not reported in the patent [114].

The nitration of βCDs featuring different cross-linkers (such as epichlorohydrin,


urethane and amines) was attempted to obtain softer binders less susceptible to
impact and ESD, but still able to contribute to the energy of the system [109]. A
cross-linked CD sample, obtained by cross-linking βCD with a large excess of
epichlorohydrin (1:12 ratio) was nitrated with 98% HNO3 and then formulated
with RDX in different ratios. The sensitivity towards accidental stimuli was then
determined in impact and ESD tests (Table 2.12).

Table 2.11 Performance of TEMTN:βNCD propellant formulations [114].

Hazard tests
DTA1 (Onset°C/ Card
Sample 5 kg Friction ESD Exotherm°C) gap
Impact
(mm) (lb) (J)

TMETN ≥ 600 ≥ 980 ≥ 12.5 1822 15–20


βNCD ≥ 600 ≥ 980 0.0125 200–2083 0
TMETN:βNCD ≥ 600 ≥ 980 ≥ 12.5 140/178 0

37
Hazard tests
DTA1 (Onset°C/ Card
Sample 5 kg Friction ESD Exotherm°C) gap
Impact
(mm) (lb) (J)
2:1
TMETN/βNCD
≥ 600 ≥ 980 ≥ 12.5 140/178 0
4:1
TMETN/βNCD
≥ 600 ≥ 980 ≥ 12.5 140/178 0
5:1
TMETN/βNCD
225 ≥ 980 ≥ 12.5 110/172 0
6:1
1 Values obtained from data thermal analysis (DTA): onset temperature of the decomposition
and exotherm peak temperature.
2 Exotherm peak value from [118].
3 This work.

The nitrated cross-linked βCD (βNCCD) reduced the sensitiveness of RDX


against impact depending on the amount of βNCCD used as the binder. The
presence of the binder slightly reduced the susceptibility of the formulation to
friction compared to pure RDX. Although the formulations also remained
vulnerable to ESD, the energy input required for ignition increased to that of the
binder (0.1288 J), which is much higher than the value for pure RDX (0.0585 J)
and significantly lower than the 0.02 J that can potentially be accumulated by an
operator [119] (Table 2.12).

Table 2.12 Sensitivity of nitrated cross-linked βCCD and RDX complexes [109]

Hazard Tests
RDX : βNCCD Impact
Sample Friction ESD
ratio Bruceton
(kg) (joule)
(cm)
RDX - 19 9.6 0.0595

βNCCD - 47 28.8 0.1288

RDX:βNCCD 1:1 42 12.8 0.1288


RDX:βNCCD 5:1 27 10.8 0.1288
RDX:βNCCD 10:1 30 10.8 0.1288

38
The CD derivatives used thus far still need design improvements before they
are considered suitable as binders for energetic applications. There is a need to
develop new systems that are more resistant to impact and ESD stimuli. To the
best of our knowledge, βCD systems cross-linked with XEG segments and
nitrated derivatives thereof have not yet been synthesised and characterised as
potential binders for energetic applications. The research described in Chapters
4–6 therefore focuses on the development and testing of these novel
formulations.

39
3 Experimental
3.1 Materials
The βCD used in this project (97%, Sigma-Aldrich) was used from stock with a
11–13% water content, based on TGA data. Polyethylene glycol diglycidyl ether
(PEGDGE 500 Mw, Sigma-Aldrich), sodium hydroxide (Fisher Chemicals),
benzylated dialysis membranes with a molecular weight cut-off (MWCO) of
2000 (Sigma-Aldrich), epichlorohydrin (Sigma-Aldrich), tetrabutyl ammonium
bromide (Sigma-Aldrich), hexaethylene glycol (Sigma-Aldrich) and triethylene
glycol (Sigma-Aldrich) were used without further purification. Triethylene glycol
diglycidyl ether (TEGDGE, 262 g mol-1) and hexaethylene glycol diglycidyl ether
(HEGDGE, 394 g mol-1) were synthesised as described in Section 3.3.

3.2 Methods

3.2.1 NMR spectroscopy


The 1H NMR spectra were recorded on a Bruker Ascend (400 MHz) with a
broad band fluorine observation (BBFO) probe in deuterated dimethyl sulfoxide
(DMSO-d6) and deuterium oxide (D2O) solutions. Signals representing the
solvents served as internal standards. The solvent peaks were referenced to 2.5
ppm (DMSO-d6) and 4.7 ppm (HDO, D2O). Peak multiplicities were described as
follows: singlet (s), multiplet (m), and broad (br).

3.2.2 FTIR spectroscopy


FTIR spectra were collected using a Bruker Alpha in attenuated total reflection
(ATR) mode, allowing the immediate characterisation of undiluted samples. The
spectra were collected in the region between 400 and 4000 cm -1.

3.2.3 Gel permeation chromatography


Gel permeation chromatography (GPC) was used to determine the Mw of the
products on a Waters size-exclusion chromatography system equipped with a
2410 refractive index detector. The samples in water (1–0 mg mL-1) were
passed through two columns (PL aquagel-OH MIXED-M 1000–500,000 Da, 8

41
µm, 300 x 7.5 mm). The buffer was a solution of 0.01 M LiNO3 and 0.5% (w/w)
NaN3 in distilled water. Nitrated samples in anhydrous THF (1.5 mg mL-1) were
passed through the column (10 µm PL-gel, Polymer Laboratories) at a flow rate
of 1.0 mL min-1. Calibration curves were created using polyethylene
oxide/polyethylene glycol standards for aqueous buffers, and polystyrene
standards for anhydrous THF.

3.2.4 Differential scanning calorimetry and thermal gravimetric


analysis
Thermal analysis of the βCXCD samples and their precursors was carried out
using a Mettler Toledo DSC3+ instrument for low-temperature profile studies
and a combination of Mettler Toledo TGA/DSC3+ and DSC30 instruments for
decomposition and compatibility studies. The temperature was cycled three
times for the low-temperature regime: the first heating run was set from –100 to
130 °C to eliminate the water present in the samples, whereas the second and
third runs were set from –100 to 120 °C, a higher temperature considered
sufficient to dry the sample. The degradation of the samples was characterised
in the temperature range 30–500 °C. A mass of 10 mg was placed in a 40-µL
aluminium pan with a pierced lid to analyse the degradation of the inert
materials, and low-temperature profiles of energetic and inert samples were
also assessed with 10-mg samples. The decomposition of the energetic
products was characterised using 1-mg samples. The DSC chamber was
continuously purged with N2 at a flow rate of 50 mL min-1. The variation of the
heat flow in the samples was recorded as a function of temperature and time.

3.2.5 Dynamic mechanical analysis


The thermo-mechanical properties of the βCXCDs were determined by dynamic
mechanical analysis (DMA) using a Perkin Elmer DMA8000 device. The
samples were subjected to a controlled sinusoidal displacement of 0.05 mm at
frequencies of 1, 5 and 10 Hz in the single cantilever clamped bending
configuration. The storage modulus (E′), loss modulus (E′′) and damping factor
(tanδ) were monitored as a function of temperature and time. The free sample
length between the vibrating and fixed cantilever clamps was ~15 mm. The

42
testing temperature was cycled three times between –100 and 140 °C at a rate
of either 2 or 10 °C min-1. Each material was tested in triplicate under each
condition. The samples were tested in aluminium pockets, a stainless steel
mesh or aluminium pockets with polytetrafluoroethylene (PTFE) tape because
the cross-linked CD samples were not capable of self-support. The aluminium
pockets consisted of rectangular shims (30 x 14 mm) cut from a 0.1 mm thick
aluminium strip (supplied by RS) and then folded lengthwise to form the
pockets. About 25 mg of the cross-linked CD sample was placed in the centre
of the pockets. This sample support was recommended by the DMA
manufacturer when testing powders, gels and liquids. PTFE tape (30 x 15 mm)
was used to fold the sample in the pocket and assess the bonding interaction
between the sample and the metallic support during cooling. Rectangular strips
of mesh (30 x 15 mm) were cut from a 0.65 mm thick sheet (supplied by RS).
The wire forming the mesh had a diameter of 0.4 mm and the dimensions of the
holes were 1 x 1 mm. About 30 mg of the crosslinked CD material was spread
around and in the centre of the mesh strips.

3.2.6 Optical microscopy


The dynamic physical properties of the βCXCD materials under the influence of
temperature were investigated by optical microscopy using a Leica DM
microscope fitted with a temperature-controlled stage (Linkam THMS 600). The
temperature was changed using a T95 controller and an automated LNP95
liquid N2 pump (both from Linkam). The material was placed on a 0.5 mm thick
quartz cover slip. The slide was placed in a carrier on the stage to allow visual
scanning. The stage was cooled to –100 °C and then heated to 100 °C at either
2 or 10 °C min-1. To prevent condensation forming on the windows of the cold
stage, the interior was purged with dry N2 gas prior to cooling. A digital Qicam
Fast 1394 CCD camera (QImaging) was used to continuously record any
changes in the samples during the temperature cycle. The sample was
illuminated by a white light source set in transmission mode.

43
3.2.7 Scanning electron microscopy
Scanning electron microscopy (SEM) was conducted using a Hitachi SU3500
instrument which is a tungsten filament variable pressure device with an
accelerating voltage of 20 kV at 80 Pa. Samples (5 mg) were pressed onto
conductive carbon tabs and supported with specimen stubs.

3.3 Procedures

3.3.1 Triethylene glycol diglycidyl ether [120]


Sodium hydroxide (40.00 g, 1.00 mol) in water (50 mL), tetrabutylammonium
bromide (1.19 g, 7.40 mmol) and epichlorohydrin (92.70 g, 1.00 mol) were
placed in a three necked round bottomed flask. The reaction mixture was then
stirred vigorously for 1 h at room temperature. Triethylene glycol (25.60 g, 0.17
mol) was added slowly over 3 h at room temperature with vigorous mechanical
stirring. The reaction mixture was stirred at 40 °C for 1 h, allowed to cool and
then filtered. The liquid phase was collected, dried overnight on sodium sulfate,
and the excess epichlorohydrin was evaporated under high vacuum to leave a
yellow-orange viscous liquid: 1H-NMR (400 MHz, CDCl3) : δ=4.00–3.40 (m,
16H, CH2-O), 3.15 (m, 2H, CH-CH2), 2.80 and 2.62 (2 m, 4H, CH2-CH);
13
C-NMR (CDCl3): δ=72.0 (CH2-O), 70.9, 70.7 and 70.6 (CH-CH2-O), 50.8
(CH-CH2) and 44.4 ppm (CH2-CH); DSC (10 °C min-1, N2) 203 °C (Tdec onset);
DSC (10 °C min-1, N2) Tg = 80 °C; yield 78%.

3.3.2 Hexaethylene glycol diglycidyl ether [120]


Sodium hydroxide (23.60 g, 0.59 mol) in water (29.5 mL), tetrabutylammonium
bromide (0.70 g, 4.35 mmol) and epichlorohydrin (54.70 g, 0.59 mol) were
placed in a three necked round bottomed flask. The reaction mixture was stirred
vigorously for 1 h at room temperature. Hexaethyelene glycol (28.50 g, 0.10
mol) was added slowly over 3 h at room temperature with vigorous mechanical
stirring. The reaction mixture was stirred at 40 °C for 1 h, allowed to cool and
then filtered. The liquid phase was collected and dissolved in dichloromethane
(50 mL). The solution was washed with half-saturated brine three times and
dried overnight on sodium sulfate. The excess epichlorohydrin was evaporated

44
under high vacuum to leave a yellow-orange viscous liquid: 1H-NMR (400 MHz,
CDCl3) : δ=4.00–3.40 (m, 26H, CH2-O), 3.15 (m, 2H, CH-CH2), 2.80 and 2.62 (2
13
m, 4H, CH2-CH); C-NMR (CDCl3): δ=72.0 (CH2-O), 70.9, 70.7 and 70.6 (CH-
CH2-O), 50.8 (CH-CH2) and 44.4 ppm (CH2-CH); DSC (10 °C min-1,N2) 273 °C
(Tdec onset); DSC (10 °C min-1, N2) Tg = 80 °C; yield 70%.

3.3.3 Cross-linked cyclodextrins (βCXCDs)


The reported procedure refers to the conditions optimised during this project.
For the conditions tested, see the appendices (Table A 1, Table A 2 Table A 3).
The βCD (5.00 g, 4.40 mmol) was dissolved in 5.6% w/w NaOH (21.0 mL) and
stirred mechanically for 16 h. Diglycidyl ether (TEGDGE, HEGDGE or
PEGDGE; 17.4–5.8 mL, 13.2–36.9 mmol) was then added dropwise over a
period of 20 min with vigorous stirring. The reaction mixture was heated to
30 °C for 6 h with vigorous stirring. After cooling to room temperature for
20 min, the mixture was neutralised with HCl (6 M). The volume of solvent was
reduced and the reaction mixture was precipitated in acetone. The solid was
dissolved in distilled water and dialysed against water using a benzylated
cellulose membrane (2000 MWCO) for 5 days. The dialysis water was replaced
every day. The dialysed white solid was collected and the water was
evaporated under reduced pressure. The final product was characterised by
1
H NMR and DSC. 1H NMR (400 MHz, DMSO-d6, ppm): δ5.9–5.6 (brm, C2-
OH, C3-OH), 5.1–4.8 (m, H-C1), 4.7–4.5 (br m, OHA, OHB), 4.4 (br m, C6-OH),
4.0–3.2 (brm, βCDOCH2(CHOH)CH2OCH2CH2). DSC 10°C min-1, N2: 240–
250°C (Tdec); yield 45-72%.

3.3.4 Nitrated cross-linked cyclodextrins (βNCXCDs)


Nitric acid (100%, 2.1 mL) was poured into a round-bottomed flask and cooled
below 10 °C in ice water. The bath was removed and βCXCD (200 mg) was
added in small fractions over 5 min, ensuring that the temperature remained
below 10 °C. The crude slurry or solution was then left stirring at room
temperature for 1 h. The reaction mixture was crushed into ice/water (10 mL).
The solid was decanted and rinsed with distilled water. The solid was dissolved
in acetone (5 mL) and re-precipitated in a solution of half saturated brine and

45
Na2CO3 (5% w/w, 100 mL) three times. The crude product was filtered and
dissolved in 10 mL acetone. The solution was filtered through Na2CO3 and dried
under vacuum. Small portions of products were characterised by 1H NMR and
DSC and stored under dichloromethane. Then 1H NMR spectroscopy was
performed in DMSO-d6 and used for comparison with the precursors. DSC (10
°C min-1, N2): 197–210 °C (Tdec); yield 80%.

3.3.5 Hazard testing


A set of small-scale hazard tests were performed according to Cranfield internal
procedures and following the Energetic Materials Testing Assessment Policy
(EMTAP) Manual of Tests.

3.3.5.1 Small scale hazard tests

Direct impact: steel hammer on steel anvil

A small amount of the synthesised compound (20 mg) was placed on the steel
anvil and struck 10 times with the steel hammer. Signs of decomposition, such
as smell, colour change and material consumption, were evaluated after each
blow.

Glancing blow: steel hammer on steel anvil

A small amount of synthesised compound (20 mg) was placed on the steel anvil
and struck with a glancing blow using the curved edge of the steel hammer.
Signs of decomposition were evaluated after each blow as above.

High temperature test

A small amount of compound (20 mg) was placed on a steel plate at 100 °C for
30 min. Signs of decomposition were evaluated as above.

Room temperature test

A small amount of compound (20 mg) was placed on a steel plate at 30 °C for
24 h. Signs of decomposition were evaluated as above.

46
Flame test

A small amount of compound (20 mg) was placed on a steel spoon and ignited
using a blow torch. The nature of the combustion process was described by an
expert member of the Cranfield University “Synthesis and Formulation” group.

3.3.5.2 Hazard tests based on the Energetic Materials Testing Assessment


Policy Manual of Tests

Electrostatic discharge test

Nylon spark test strips were filled with the material and sealed with copper tape.
Fifty samples were tested using a certified ESD testing apparatus and a spark
of 4.5 J was discharged through the composition. The samples were inspected
for perforation or signs of decomposition.

Rotter direct impact test

Samples (30–40 mg) were placed in a concavity at the centre of the supporting
frame of the Rotter tester apparatus. A free-fall weight (5 kg) and striker were
suspended above. The tests followed the Bruceton “up and down” testing
technique with 50 replicates and the results were based on the height at which
the compound was initiated 50% of the time, with the mean height reported as
the figure of insensitiveness (FoI) [119]. Initiation was determined by the
observation of parameters such as sound, smoke, flash and volume of gas
released immediately after impact, and is therefore affected by operator
judgment.

3.3.6 Compatibility tests


The tests were based on Test 4 of STANAG 4147 [121]. Approximately 0.5–1.0
mg of each material was analysed by DSC at a heating rate of 2 °C min-1 under
N2 and then a mixture of the two materials was measured in the same manner.
Any alteration in the shape, onset, or peak position of any measured thermal
event can be indicative of incompatibility.

Under “Applicability” STANAG 4147 Test 4 states: “This test is applicable to


explosives likely to come into contact with plasticizers, fuels, additives,

47
polymeric materials and other explosives.” It also states: “This test is not
concerned with compatibility between ingredients in explosive compositions and
the consequent stability of such compositions.” Even so, the method allows a
large number of mixtures to be investigated rapidly. As such, the results from
this tests are useful but not conclusive, and further experiments are required,
such as vacuum stability testing [122].

Edition 2 of STANAG 4147 Test 4 states that shifts in thermal events that vary
by less than 4°C indicate compatibility. For shifts between 4 and 20 °C, the test
is inconclusive and, as above, further experiments are required [122]. Shifts
greater than 20 °C are considered conclusive evidence that materials are
incompatible.

3.3.7 Stability analysis


The stability of the nitrated products was determined by heat flow calorimetry
(HFC). The samples (1.0 g) were placed in sealed vials (10 mL) with glass
beads to fill the head space. The samples were characterised by isothermal
calorimetry at 80 °C on a TAM IV device with a dedicated software package.

48
4 Synthesis and characterisation of inert βCXCDs
This chapter describes the cross-linking of βCD with the diglycidyl ethers
TEGDGE, HEGDGE or PEGDGE as a first step towards the synthesis of a new,
inert binder for energetic applications. The cross-linking reaction conditions
discussed below were investigated and optimised to obtain products that
exhibited the following properties, in alignment with the statement of
requirements developed by the PhD sponsor organisation, the Weapons
Science and Technology Centre (WSTC):

• Low Tg (possibly below 0 °C) to reduce vulnerability to shocks, ensuring the


cross-linked binder remains soft in a wide operative temperature range,
down to the Tg.
• Soluble in organic solvents and/or water, allowing the formulations to be
processed by preparing a lacquer with the binder. The lacquer and the solid
ingredients mixed with it can be extruded as a paste and granulated in
pellets [2]. Low-molecular-weight cross-linked products with a low cross-
linking ratio would guarantee the preparation of an extrudable paste.
• Easily converted to energetic derivatives by nitration. The new requirement
for binders in energetic formulations is to contribute to the energetic output
of the main filler. A simple and safe nitration procedure would produce a
suitable replacement for NC.
• Good chemical and thermal stability to avoid uncontrolled events during the
shelf life of the formulation.
• Good compatibility with energetic fillers such as high explosives, oxidisers,
and pyrotechnics to avoid undesirable reactions and instability when the
binder is mixed with the other ingredients of an energetic formulation.

4.1 Synthesis and chemical characterisation


The effect of the reaction conditions discussed in sections 4.1.1–4.1.7 and the
1
chemical characterisation of some of the βCXCDs by H-NMR and FTIR
spectroscopy have been published in a peer-reviewed article: Luppi F, Cavaye
H, Dossi E (2018) Nitrated cross-linked β-cyclodextrin binders exhibiting low

49
glass transition temperatures. Prop Expl Pyrotech 43 (10) 1023–1031. The
paper is attached at the end of the annex to this thesis.

The synthesis of βCXCDs was based on the cross-linking of βCD with


polyethylene glycol diglycidyl ethers containing ethylene glycol units, according
to Scheme 4.1. Triethylene glycol diglycidyl ether (TEGDGE) contains three
ethylene glycol units and was a synthetic sample. Hexaethylene glycol diglycidyl
ether (HEGDGE) contains six ethylene glycol units and was also a synthetic
sample. Polyethylene glycol diglycidyl ether (PEGDGE) was obtained from a
commercial source and was a polydisperse 500 Da preparation with an average
of nine ethylene glycol units. The abbreviation XEG is used to collectively refer
to the ethylene glycol segments in the cross-linkers and the βCXCD products.

To note that, TEG:βCD ratio, HEG:βCD ratio, PEG:βCD ratio is used refer to
the ratio of XEG units to βCD units in the reaction products βCXCDs

TEGDGE was prepared according to a published procedure [120] by reacting


triethylene glycol with an excess of epichlorohydrin under basic conditions to
produce the diglycidyl ether derivative. The same experimental conditions were
then used for the synthesis of HEGDGE. The synthesis of HEGDGE from
hexaethylene glycol has been described in the past following a similar route, but
gave a lower yield (55%) [123]. The yield using the method followed in this PhD
project was 70%.

As shown in Scheme 4.1, βCD (1) is dissolved in NaOH (5.6–50%) to give the
corresponding alkoxylated βCD (2). The equilibrium is strongly shifted towards
(1) (Step 1, Scheme 4.1) due to the pKa values of the acid dissociation of the
βCD hydroxyl groups. The pKa values for the βCD secondary alcohols in the
literature vary from 12.1 to 13.5 [124]. There is no information about the pKa
values of the primary hydroxyl groups because this is difficult to measure
directly [124]. The alcohol-alkoxide equilibrium (Step 1, Scheme 4.1) therefore
cannot be evaluated accurately. When a solution of 1.5 M NaOH (5.6% w/w) is
used, the amount of (1) is six orders of magnitude higher than the alkoxylated
derivative (2). It can be inferred that the alkoxylation of 1 is the rate-limiting step
in the cross-linking process.

50
Scheme 4.1 The synthesis of cross-linked βCXCDs.

51
The reaction can be accelerated by modifying certain conditions, such as the
temperature or concentration of NaOH or cross-linker, but a secondary reaction
competes with the main reaction and its outcome is the loss of cross-linker. The
competitive reaction is the opening of the diglycidyl rings of the cross-linker
under basic conditions yielding the corresponding tetrahydroxyl derivative
(Scheme 4.2).

Scheme 4.2 Competitive hydrolysis of diglycidyl rings under basic reaction conditions.

Both the cross-linking and competitive hydrolysis reaction kinetics are strongly
affected by:

 NaOH concentration
 Reaction time
 Temperature
 Effect of the βCD alkoxide formation
 Volume of NaOH aqueous solution
 Cross-linker:βCD ratio
 Addition rate of the cross-linker

Because all of these parameters affect the efficiency of cross-linking, it was


necessary to optimise them in order to avoid the competitive reaction as far as
possible and increase the yield of the desired product. PEGDGE is a
commercially available chemical and was used for most of the cross-linking
experiments, but a few tests were carried out using the TEGDGE produced in-
house.

52
4.1.1 Effect of the NaOH concentration
The effect of the concentration of NaOH was tested to determine which
conditions favoured the primary cross-linking reaction rather than the side
reaction in which the cross-linker is degraded. The effect of the NaOH
concentration was investigated at a high temperature (70 °C) for short reaction
times (Step 2, 1 h) to enable rapid screening. All the NaOH concentrations that
were tested resulted in a strongly alkaline environment, which can deprotonate
the βCD hydroxyl groups. Four βCXCD samples (IP1–IP4) were prepared using
50%, 40%, 36% and 5.6% w/w NaOH solutions and a PEGDGE:βCD ratio of
9:1 (Table 4.1) [78]. Higher concentrations of NaOH were used for the synthesis
of epichlorohydrin/βCD cross-linked systems [59,69,75], so these conditions
were tested to determine their effect on the balance between cross-linking and
PEGDGE degradation and also their effect on the solubility of the compounds.
The results are summarised in Table 4.1.

Table 4.1 Effect of NaOH concentration on the yield of βCPCDs at a PEGDGE:βCD


ratio of 9:1.

Reaction
2
Sample NaOH Time Yield3 Water
ID1 (% w/w) (Step 2) (%) solubility
(h)
IP1 50 1 <1 Y
IP2 40 1 <1 Y
IP3 36 1 <1 Y
IP4 5.6 0.66 88% N
1
PEGDGE:βCD ratio = 9:1.
2
Reaction temperature = 70 °C.
3
Yield measured as mass of products/mass of both reactants.

At 70 °C, the most dilute NaOH solution (5.6% w/w) promoted cross-linking
whereas all the stronger NaOH solutions (36%, 40% and 50% w/w) favoured
the rapid degradation of the cross-linker compared to the formation of the βCD
alkoxide. When the NaOH concentration was low, the reaction reached the
gelation point, thus causing a consistent change in viscosity of the reaction
mixture accompanied by the appearance of an insoluble hydrogel. The speed of

53
gelation indicated that cross-linking with 5.6% NaOH at 70 °C is extremely fast
and would make it difficult to recover soluble products. Accordingly, the studies
proceeded by investigating the effect of the reaction time in the presence of
40% NaOH. However the optimisation of the temperature and the amount of
cross-linker discussed in the following sections proved that this assumption was
incorrect, and that 5.6% NaOH was a more effective choice at lower
temperatures.

4.1.2 Effect of the reaction time


The slow step in the cross-linking mechanism is the formation of the alkoxide
(Step 1, Scheme 4.1), only small amounts of which are available to react with
the cross-linker. In order to increase the yield of βCPCDs, it appeared important
to increase the reaction time to reform the alkoxides that reacted with the cross-
linker and make more of these groups available for the cross-linking reaction.
Two samples were synthesised, maintaining the earlier reaction conditions
(40% NaOH, 70 °C) but extending the reaction time (Step 2) from 1 h (sample
IP2, Table 4.2) to 5 h (sample IP5, Table 4.2).

Table 4.2 Effect of the reaction time (Step 2) on the yield of βCPCDs.

Reaction
2
Sample NaOH Time Yield3 Water
ID1 (% w/w) (Step 2) (%) solubility
(h)
IP2 40 1 <1 Y
IP5 40 5 12 Y
1 PEGDGE:βCD ratio = 9:1.
2 Reaction temperature = 70 °C.
3 Yield measured as mass of products/mass of reactants.

The product yield increased from negligible amounts to 12% when the reaction
time was extended to 5 h (Step 2). This suggested that longer times are
necessary to replenish the consumed alkoxide, allowing the cross-linking
reaction to proceed. Although 12% was a significant improvement, it was not
considered ideal and the reaction conditions were investigated in further detail.
Accordingly, extending the time from 5 to 6 h improved the yield even further.

54
4.1.3 Effect of the temperature
Higher temperatures often accelerate endothermic reactions [125]. However,
both the cross-linking reaction and the degradation of the cross-linker are likely
to be accelerated by temperature. A trial was therefore carried out using the
following conditions:

 The temperature was reduced from 70 °C to 50 °C


 A low concentration of NaOH (5.6% w/w)
 A time of 5 h for the cross-linking reaction (Step 2, Scheme 4.1)
 The amount of cross-linker was reduced to a 5:1 ratio, to avoid the gelation
observed when the ratio was 9:1

The cross-linker appeared more stable at 50 °C when the reaction proceeded


for 5 h in the presence of 5.6% NaOH (sample IP6, Table 4.3). The lower yield
at 70 °C (sample IP7, Table 4.3) reflected the loss of cross-linker in the
competing degradation reaction, which is favoured by the higher temperature,
whereas lower temperatures favour cross-linking. Water-soluble polymers were
produced in the tests, confirming that a lower PEGDGE:βCD ratio was
necessary to generate soluble cross-linked βCPCDs.

Table 4.3 Effect of the temperature on the yield of βCPCDs, when the PEGDGE:βCD
ratio is 5:1.

Sample T Yield2
Water solubility
ID1
(°C) (%)

IP6 70 1 Y

IP7 50 13 Y
1 PEGDGE:βCD ratio = 5:1, 5 h reaction time (Step 2), aq. NaOH 5.6% w/w.
2 Yield measured as mass of products/mass of reactants.

55
4.1.4 Effect of βCD alkoxide formation
The time allowed for the system to reach alcohol/alkoxide equilibrium (Step 1,
Scheme 4.1) was extended by leaving the βCD stirring in 5.6% NaOH for 16 h,
thus increasing the amount of alkoxide available to react with the cross-linker.

Table 4.4 Effect of the time allowed for βCD alkoxide formation (Step 1) on the yield of
βCPCDs.

Sample Alcohol/alkoxide T Yield2 Water


ID1 time (Step 1) o
( C) (%) solubility

IP6 70 1 Y
0h
IP7 50 13 Y

IP8 50 30 Y
16 h
IP9 30 33 Y
1 PEGDGE:βCD ratio = 5:1, 5 h reaction time (Step 2), aq. NaOH 5.6% w/w.
2 Yield measured as mass of products/mass of reactants.

The difference between samples IP6 and IP7, where no time was allowed for
the system to reach alcohol/alkoxide equilibrium (Step 1, Scheme 4.1), and
samples IP8 and IP9 (where Step 1 in Scheme 4.1 involved stirring for 16 h),
revealed that the cross-linking process requires more time to increase the
availability of alkoxide groups for the reaction (Step 2, Scheme 4.1). Samples
IP8 and IP9 were obtained at higher yields (30% and 33%, respectively), with
the best results achieved at 30 °C. The malleability of the products prompted
thermal characterisation studies, revealing that the Tg of compounds IP8 and
IP9 were –18 and –22 °C, respectively. This confirmed that the processability of
crystalline βCD could be improved by the incorporation of soft polyethylene
glycol segments.

4.1.5 Effect of the volume of NaOH solution


When the concentration of βCD was higher than 0.23 gβCD mL-1aq.NaOH in the
reaction mixture, the solubility of the cross-linked βCPCDs was reduced,

56
causing their gelation (sample IP9, Table 4.4). In all subsequent experiments,
the safe value of 0.21 gβCD mL-1aq.NaOH was used.

4.1.6 Effect of cross-linker:βCD ratio


The next set of trials (Table 4.5) tested the effect of the PEGDGE:βCD ratio on
the yield and physical properties of the products, with the other parameters
maintained at the optimal values established above:

 5.6% w/w NaOH


 30 °C
 16 h for the formation of the alkoxide (Step 1, Scheme 4.1)
 5 h for the completed cross-linking reaction (Step 2, Scheme 4.1)
 PEGDGE as the cross-linker

The PEGDGE:βCD ratio influenced the final thermal properties of the βCPCD
products. As expected, increasing the amount of cross-linker from 2 to 5 molar
equivalents increased the cross-linking ratio in the product (Table 4.5). The
cross-linker consumed in the competitive degradation reaction negatively
affected the mass economy of the reaction, with the difference attributed to the
degradation of the PEGDGE. Thermal analysis of the cross-linked products
confirmed that lower amounts of cross-linker generated βCPCD products with
greater crystallinity and higher Tg values. The synthesis of sample IP11
achieved a higher yield than the other compounds, and led to a further
investigation of the purification method, which generally involved dialysis
against water.

Table 4.5 Effect of PEGDGE:βCD ratio on the yield of βCPCDs.

Sample PEGDGE:βCD Yield2 Water


ID1 feed ratio (%) solubility
IP9 5:1 343 Y
IP10 4:1 363 Y
IP11 3:1 68 Y
IP12 2:1 443 Y
1 5.6% (w/w) NaOH, 30 °C, 16 h for alkoxide formation (Step 1), 5 h reaction time (Step 2).
2 Yield measured as mass of products/mass of reactants.

57
3 Estimated value due to the leaking of the product from the purification equipment.

4.1.6.1 Purification of βCXCDs

A typical work up for the cross-linked compounds started with the recovery of
the crude product from the reaction mixture after neutralising the excess NaOH
with 6 M HCl. Most of the water was removed under vacuum and the product
was reprecipitated in acetone to eliminate unreacted and degraded cross-
linkers. The compound was dissolved again in a small amount of water and
dialysed against water to remove:

 NaCl formed during the neutralisation step


 Unreacted starting compounds (βCD and cross-linker)
 Any residual degraded cross-linker
 Low-molecular-weight oligomers

Dialysis was carried out using a cellulose membrane with a MWCO of 2000 Da
for 5 days at room temperature (Figure 4.1). The dialysate was changed daily.
At the end of the process, the retentate containing the desired cross-linked
product was NaCl-free, as assessed by AgNO3 qualitative analysis. When
characterised by 1H-NMR in DMSO-d6, the compounds were found to be free of
impurities. The NMR characterisation of the product is discussed in detail in
Section 4.2.

58
Figure 4.1 Dialysis apparatus for the purification of βCXCDs.

The dialysate residues were dried and weighed, revealing that 80–70% w/w of
the expected impurities on the first day were dialysate, reaching 85–90% w/w
on the second day. No residual NaCl was detected after replacing the water
four times, leaving only the degraded cross-linker and oligomeric products with
a Mw < 2000 Da. These were absent after 5 days of dialysis.

4.1.7 Effect of the duration of cross-linker addition


The duration of cross-linker addition to the reaction mixture containing the βCD
alkoxide was tested, initially with prolonged exposure times of up to 7 h to
determine whether generally lower amounts of cross-linker would react
preferentially with the βCD alkoxide rather than undergoing degradation. The
earlier results suggested that βCPCDs obtained with a PEGDGE:βCD ratio of
5:1 reflected the maximum ratio that still produced soluble compounds (Table
4.6). Considering a potential increase in the efficiency of the cross-linking
reaction, the first test involved 16 h for the alkoxide formation (Step 1, Scheme
4.1) and 7 h for the addition of the cross-linker (Step 2, Scheme 4.1). The
product obtained using the new reaction parameters (sample IP14) had a higher
PEG:βCD ratio than a similar setup with a brief exposure time (sample IP13)
reflecting an 8% higher cross-linking efficiency. No significant change in the

59
yield was observed, possibly due to losses of product during the purification
dialysis (Table 4.6).

Table 4.6 Effect of cross-linker addition time.

Sample PEGDGE:βCD PEG:βCD 1 Addition Yield


1 Cross-linker
Water
ratio by H yield
ID ratio feed time (h) (%)
NMR efficiency2 solubility
IP13 5:1 3.8 : 1 20 min 68 75 Y
IP14 3.8 : 1 3.2 : 1 7h 71 83 Y
1 Yield measured as mass of products/mass of PEGDGE and βCD.
2 Cross-linker yield is measured as the molar ratio between the feed of cross-linker and the
amount of polyethylene glycol units present in the product.

The longer cross-linker addition time improved the mass economy of the
reaction, but it seemed to have a negligible effect on the reaction yield and the
product solubility. A cross-linker addition time of 20 min was therefore used for
subsequent tests.

4.2 The synthesis of βCXCDs


The screening tests discussed in Sections 4.1.1–4.1.7 revealed the optimal
conditions for the cross-linking of βCD with diglycidyl ethers. The reaction time
was duly increased to 6 h to ensure that more unreacted cross-linker was
consumed. The optimised parameters are shown below:

 Low NaOH concentration (5.6% w/w) to reduce the competitive side reaction
(degradation of the cross-linker)
 Concentration of βCD in NaOH = 0.21 gβCD mL-1aq.NaOH to avoid gelation
 A 16 h (overnight) reaction for βCD alkoxide formation (Step 1, Scheme 4.1)
 Cross-linker:βCD feed ratio < 5:1 to avoid gelation
 Addition of the cross-linker in 20 min
 Low temperature (30 °C) to favour cross-linking over the degradation of the
cross-linker
 A 6-h reaction time (Step 2, Scheme 4.1)

60
Three sets of βCXCDs were synthesised using TEGDGE, HEGDGE and
PEGDGE cross-linkers, varying the feed cross-linker:βCD ratio from 2:1 to 5:1.
The water content of βCD can reach (13–14% w/w) [38]. The water content was
determined by TGA prior to each cross-linking reaction to maintain the βCD
concentration at 0.21 gβCD mL-1aq.NaOH, as discussed above. In a typical
synthesis reaction, 5 g of wet βCD was dissolved in 20.5 mL 5.6% NaOHaq w/w
and left to react at 30 °C overnight to form the βCD alkoxide. The cross-linker
was then added in 20 min and the reaction mixture was left for 6 h at 30 °C as
shown in Figure 4.2.

Figure 4.2 Cross-linking apparatus for the synthesis of βCXCDs.

The reaction mixture was then neutralised with 6 M HCl to remove excess
NaOH, reduced in volume and reprecipitated in acetone to eliminate organic
impurities such as degraded and unreacted cross-linker (Scheme 4.2). The
products were then purified by dialysis against water using a 2000 Da MWCO
benzoylated cellulose membrane. The retentate was collected and dried under
high vacuum for 48 h. The cross-linking of βCD with PEGDGE was replicated to
produce enough sample mass for the tests discussed in the following chapters.
No significant differences in the yield or properties of the synthesised products
were observed when comparing the replicates (Table A 1).

61
The desired βCXCD products were obtained with yields of up to 68% when
HEGDGE and PEGDGE cross-linkers were used, whereas slightly lower yields
(63%) were achieved when TEGDGE was used. This may reflect the outcome
of the dialysis step, with smaller βCTCD molecules not retained by the
membrane. Table 4.7 summarises the results of representative cross-linking
reactions with the best results in terms of yield and βCXCD properties. The
remaining synthesis reactions are reported in Table A 1.

62
Table 4.7 Synthesis of βCXCDs: summary of the most successful water-soluble products.

Cross-linker Physical
Sample XEG:βCD ratio Yield3
Cross-linker :βCD ratio appearance
ID1,2 by 1H NMR (%)
feed
IT1 5 :1 3.6 :1 63 Friable solid

IT2 4:1 3.1 :1 52 Brittle solid


TEGDGE
IT3 3:1 2.4 :1 56 Brittle solid

IT4 2:1 1.9 :1 45 Brittle solid

IH1 5:1 4.0 :1 68 Soft gum

IH2 4:1 3.2 :1 72 Soft gum


HEGDGE
IH3 3:1 2.6 :1 67 Brittle solid

IH4 2:1 1.8 :1 65 Brittle solid

IP13 5:1 3.8 :1 68 Soft gum

IP15 4:1 3.0 :1 65 Soft gum


PEGDGE
IP16 3:1 2.3 :1 68 Brittle solid

IP17 2 .1 1.6 :1 68 Brittle solid


1 5.6% (w/w) NaOH, 30°C, 16 h for the formation of βCD alkoxide, 6 h reaction time, 20 min for cross-linker addition.
2 Sample ID: I = inert (not nitrated), T/H/P = cross-link segment, numeral = unique sample reference.
3 Yield measured as mass of products/mass of cross-linker and βCD.

63
The off-white βCXCDs were soft and malleable at cross-linker ratios of 4:1 and
5:1, but powdery and fragile at cross-linker ratios of 3:1 and 2:1 (Figure 4.3).

Figure 4.3 Physical characteristics of βCPCD. (a) Malleable IP13. (b) Powdery IP17.

Attempts to cross-link βCD using a greater ratio of cross-linker:βCD than 5:1


generated highly cross-linked βCXCD products that were insoluble in water.
The βCXCDs produced with lower cross-link ratios were more soluble, and
among the three sets of products those synthesised with TEGDGE were easier
to dissolve than those synthesised with HEGDGE or PEGDGE. This is likely to
reflect the lower Mw, generating a less intricate cross-linked network.

All the products were characterised by 1H NMR:

 The NMR spectra were recorded in both DMSO-d6 and D2O solvents without
a reference because βCD can encapsulate these reference compounds
 DMSO-d6 was used as a solvent to investigate whether the hydroxyl groups
of the βCDs at 4.4–5.5 ppm reacted with the cross-linker: a decrease in
signal intensity would confirm this.
 D2O was used as a solvent to determine the cross-linking ratio (this was not
possible in DMSO-d6 due to the overlapping water and product signals.

64
1
H-NMR analysis in DMSO-d6

1
The H-NMR analysis in DMSO-d6 of the βCPCD sample IP13 and its
precursors PEGDGE and βCD led to the proposed βCXCD general chemical
structure shown in Figure 4.4. The corresponding 1H-NMR spectra are shown in
Figure 4.5.

Figure 4.4 Proposed chemical structure of the βCXCD products based on IP13
incorporating PEG segments, underlining the co-existence of PEG units as spacers
and as entanglements.

65
Figure 4.5 Comparative 1H NMR spectra of (a) sample IP13 (PEG:βCD ratio = 3.8:1), (b) PEGDGE, and (c) βCD in DMSO-d6 with
assignments of signals. The DMSO reference peak is at 2.5 ppm. R= H, cross-linker Ca-OR = H or βCD.

66
The NMR spectra of βCD and PEGDGE match the values reported in the
literature and discussed in Section 2.4.1 [63]. The spectrum of sample IP13 in
DMSO-d6 shows broadened peaks due to the macromolecular structure of the
cross-linked product, given the complex chemical environment of protons in
larger cross-linked molecules compared to the smaller βCD and cross-linkers.
This is a general observation valid for all the synthesised βCXCD systems,
confirming their polymeric nature. The integration of the NMR signals gave
reliable indications for the chemical structure of the products, including the
reduction of the integrals from OH groups at 5.9 ppm. The consumption of the
OH-6 groups cannot be confirmed because they overlap with new signals
between 5.0 and 4.6 ppm, which were attributed to the OHb on the PEG spacer
units after the opening of the diglycidyl ring of the PEGDGE (Scheme 4.2). The
anomeric proton H-1 was found at 4.7 ppm broadened by the macromolecular
structure and by its overlap with signals from OHa and OHb. The remaining
proton signals from βCD and methylene protons from the PEG units were found
between 3.9 and 3.2 ppm. Many signals overlap and peak attribution is therefore
difficult. The absence of the characteristic diglycidyl group signals between 3.2
and 2.6 ppm excludes the presence of the cross-linker in the purified product
(Figure 4.6a-c).

It was difficult to confirm the absence of degraded cross-linker (Figure 4.5b) in


the final cross-linked products. In order to help with the assignment of the
peaks, a sample of PEGDGE was left to react under the cross-linking reaction
conditions, resulting in partial degradation and the formation of a blend of
unreacted and degraded PEGDGE (Figure 4.6). During the synthesis of all
βCXCDs, the by-products were removed during purification by reprecipitation in
acetone followed by dialysis.

In addition, when only one of the glycidyl groups of PEGDGE reacts with βCD, a
new pendant PEG unit is formed which contains two new hydroxyl groups (OHb
and OHa) at the end of each entanglement. The NMR signals of the new OHs
appear as a broad peak between 4.9 and 4.55 ppm (Figure 4.6).

67
Figure 4.6 Comparative 1H NMR spectra of (a) IP13 (PEG:βCD ratio = 3.8:1), (b) mixture of degraded/non degraded PEGDGE, and
(c) PEGDGE in DMSO-d6.

68
1
H-NMR analysis in D2O

The 1H-NMR spectra of the βCXCDs were then recorded in D2O to determine
the degree of cross-linking. The spectrum of the representative sample IP13 in
Figure 4.7 is divided in two distinct regions: 5.2–4.6 ppm attributed to the
chemical shift of the H-1 anomeric protons in two different situations as seen in
Fig 4.7b and 4.1–3.4 ppm assigned to the chemical shift of the CH and CH2
protons of both βCD and PEG units in the βCXCDs.

Figure 4.7 (a) The 1H-NMR spectrum of IP13 (PEG:βCD ratio = 3.8:1) in D2O and (b)
the proposed chemical structure of βCXCDs highlighting the two different situations for
the anomeric H-1, when in unit 1 or unit 2 of βCXCDs.

The broad signal centred at 5.00 ppm is assigned to the anomeric H-1 protons
of the glucosidic unit 1 of βCD and represents the chemical environment of H-1

69
βCD when the protons of the OH-2 groups are not involved in the cross-linking.
In contrast, when the protons of the OH-2 groups were substituted with PEG
units (Figure 4.7, unit 2) the anomeric proton signal was shifted downfield at
5.18 ppm. The ratio between the two integrals in Figure 4.7 (Int1 and Int2) was
used to determine the degree of cross-linking in βCXCDs (XEG:βCD molar
ratio) [86]. Int1 refers to the seven anomeric protons of βCD units, whereas Int2
refers to all remaining protons in the βCXCD monomeric unit, specifically:

 42 protons of βCD units


 Protons of the polyethylene glycol unit (X), namely
 22 protons when X = TEG
 34 when X = HEG
 44 when X = PEG

The following system of two equations was used

𝐼𝑛𝑡1 = 7 H‐1 (1)

𝐼𝑛𝑡2 = 𝐻CD units + 𝑛𝐻𝑃𝐸𝐺 𝑜𝑟 𝐻𝐸𝐺 𝑜𝑟 𝑇𝐸𝐺 (2)

From the 1H-NMR spectra, the calculated XEG:βCD ratios in the βCXCDs were
25% lower than the initial feed ratio of the reaction. Using sample IP13 as an
example, the PEGDGE:βCD feed ratio was 5:1 but the PEG:βCD ratio in the
product was 3.8:1 (Table 4.7) reflecting the degradation of the cross-linker as
discussed in section 4.1.1. The PEG:βCD ratio was also confirmed by the
integrals ratio of the H-1 signals in unit 2 and unit 1 (Figure 4.6b).

The calculation of the XEG:βCD ratio in the βCXCDs was also attempted by
comparing Int1 and Int3, where Int3 between 3.7 and 3.6 ppm is the integral of
the four ethylene glycol CH2 protons present in the XEG repeat units in βCXCD
(Figure 4.8).

However, the cross-linking ratios obtained when comparing Int1 and Int3 were
5–10% higher in value in relation to the cited method (Equations 1 and 2) which
uses Int1 and Int2 [86], reflecting the partial overlap of the βCD protons and the

70
methylene groups of the XEG chains. The values obtained via the cited method
were therefore used for subsequent evaluations.

Figure 4.8 The 1H-NMR spectrum of IP13 (PEG:βCD ratio = 3.8:1) in D2O. Int1 =
anomeric H-1, Int3 = CH2 of PEG.

The NMR spectra of the three sets of βCXCDs prepared using PEGDGE were
similar, as shown in Figure 4.9. However, the intensity of the signal between 5.2
and 4.9 ppm, attributed to unit 2 (the anomeric H next to cross-linking point),
increased at higher PEG:βCD ratios. Three sharp signals attributed to the H-3
proton of βCD units appeared in the region between 3.6 and 3.4 ppm when the
PEG:βCD ratio was lower than 2.6, due to the absence of the broad signals at
high level of cross-linking (black arrows, Figure 4.9).

71
Figure 4.9 Comparative 1H-NMR spectra of βCPCD samples (IP13, IP15–IP17) with
different PEG:βCD ratios. Lower ratios enhance the visibility of the βCD proton
(marked with the arrow).

All βCXCDs were characterised by FTIR spectroscopy and the spectrum of a


representative sample (IP13) was compared to those of the precursors βCD
and PEGDGE (Figure 4.10). The strong absorption at 1077 and 1023 cm-1
(Figure 4.10, red curve) was attributed to the stretching of C-O-C cross-linking
bonds in βCXCDs. The absorption intensity at 2940 cm-1 (symmetric stretching
of CH2 and CH), 2871 cm-1 (asymmetric stretching of CH2 and CH), 1453 cm-1
(scissoring of CH2) and 1349 cm-1 (bending of CH) in the βCXCDs was higher
than in the βCD precursor, confirming the presence of the cross-links. The
signals in all the spectra at 1642 cm-1 represented the secondary vibration of
adsorbed water [73].

72
Figure 4.10 Comparative FTIR spectra of βCD (black curve), βCPCD sample IP13 (red
curve), and PEGDGE (blue curve).

The Mw of water-solubleβCXCDs containing TEG, HEG or PEG units were


determined by GPC. The samples were dissolved in aqueous 0.1 M LiNO3
containing 0.05% w/w NaN3, with polyethylene glycol/polyethylene oxide
standards for nine-point calibration in the range 106 Da to 1 MDa. Therefore, all
Mw values reported are standards equivalents. These standards were the
nearest available to the chemical composition of the βCXCDs.

GPC analysis of the precursors βCD and PEGDGE was carried out under the
same experimental conditions. The βCD chromatogram showed a peak at
retention time 13.7 min (Mw=1134.98 g mol-1, 200 Da standards equivalents)
that fell outside the calibration curve of the standards (Figure 4.11). As
expected, the standards eluted at different times compared to pure βCD. Their
equivalents cannot be used to assess the Mw of the oligomers, but provide a
useful qualitative estimate.

73
Figure 4.11 GPC analysis of βCD compared to polyethylene glycol/polyethylene oxide
standards.

The chromatogram of PEGDGE (Figure 4.12) revealed the polydispersity of the


compound as stated by the manufacturer (Sigma-Aldrich, average Mw=500).
Two major fractions were detected, with retention times of 12.7 and 13.5 min,
respectively (equivalent = 200 Da for the first fraction, with the other outside the
calibration range). The TEGDGE and HEGDGE cross-linkers were not analysed
by GPC because 1H-NMR analysis confirmed their monodisperse nature (Mw =
262 and 394 Da, respectively; Figure A 2 and Figure A 3).

Figure 4.12 GPC analysis of PEGDGE compared to polyethylene glycol/polyethylene


oxidestandards.

The βCXCD samples analysed by GPC (IT2I, IH2 and IP22) were synthesised
using TEGDGE, HEGDE and PEGDGE, respectively. Their chromatograms
(Figure 4.13) were composed of two major elusion peaks between 12 and 16
min. Sample IH2 presented a broader distribution of higher Mw, represented by

74
the onset of elution at 7.5 min corresponding to fractions with higher Mw (Figure
4.13b). The Mw of all βCXCDs, calculated using the elution times of the
standards equivalents, were lower than expected. Again this reflected the
different elution behaviour of the βCXCDs compared to the standards, due to
the presence of βCD with a different hydrodynamic volume [125]. The peak at
13.7–14 min suggested the sample was contaminated with βCD and/or dimers
that were not eliminated during the purification process.

GPC analysis of the βCXCDs confirmed that products with higher XEG:βCD
ratios (3.6:1) eluted at lower retention times, corresponding to their higher Mw.
Figure 4.14 shows the chromatograms of βCTCDs containing a decreasing ratio
of TEG units.

75
Figure 4.13 Analysis of βCXCDs by GPC compared to polyethylene
glycol/polyethylene oxide standards. (a) Sample IP22 (PEG:βCD ratio = 2.9). (b)
Sample IH2 (HEG:βCD ratio = 3.2). (c) Sample IT2 (TEG:βCD ratio = 3.1).

76
Figure 4.14 Analysis of βCTCDs by GPC compared to polyethylene
glycol/polyethylene oxide standards. (a) Sample IT1 (TEG:βCD ratio = 3.6). (b) Sample
IT2 (TEG:βCD ratio = 3.1). (c) Sample IT3 (TEG:βCD ratio = 2.4). (d) Sample IT4
(TEG:βCD ratio = 1.9).

77
4.3 Thermal and thermo-mechanical characterisation of
βCXCDs
The thermo-mechanical characterisation of some of the βCPCDs synthesised
using the PEGDGE cross-linker has been published as a peer-reviewed article:
Luppi F, Kister G, Carpenter M, Dossi E (2019) Thermomechanical
characterisation of cross-linked β-cyclodextrin polyether binders. Polym Test 73,
338–345 (see Appendix).

The thermo-mechanical properties of the βCXCDs were investigated to


determine their suitability as binders for energetic formulations. Their thermal
properties, specifically the decomposition temperature (Tdec), the enthalpy of
decomposition and the glass transition temperature (Tg) were determined by
DSC and DMA.

DSC was carried out using two temperature regimes: a low temperature range
(–100 to 100 °C) to determine Tg and a high-temperature range (25–500 °C) to
determine the Tdec. DMA was carried out solely in the low-temperature range.

The first task was the characterisation of the starting materials (βCD and the
cross-linkers TEGDGE, HEGDGE, PEGDGE). The low-temperature heat flow
changes were initially examined from –100 to 100 °C, but the upper
temperature was later extended to 140 °C to detect thermal events attributed to
βCD. The thermo-mechanical properties of βCD/ethylene glycol hydrogel
systems have been investigated before, revealing their viscoelastic properties
[86,87,92]. The βCXCDs were characterised to determine whether their
viscoelasticity was sufficient to reduce their sensitivity to shocks and hence their
vulnerability when used as binders.

In this chapter, the published data are combined with data for the βCTCDs and
βCHCDs. This chapter also discusses the self-healing properties of the
βCXCDs.

78
4.3.1 Thermo-mechanical characterisation of the starting materials
4.3.1.1 Thermal characterisation of βCD

The degradation of βCD under N2 gas was observed in the temperature range
30–500 °C by the combined use of DSC and DSC + TGA techniques. The βCD
thermogram is shown in Figure 4.15.

Figure 4.15 Combined TGA and DSC analysis of βCD (10 °C min-1, 25–500 °C,
aluminium crucible).

An endothermic event occurring before the temperature reached 110 °C was


attributed to the presence of water in the βCD stock sample, confirming the
precursor is hygroscopic. The mass loss percentage of 13.6% was similar to
that reported in the literature [38,126]. This value varied by several percentage
points depending on the moisture adsorbed daily by the βCD (13.1–13.7%). The
major event indicated by DSC was an endothermic peak at 328 °C, which
represented the degradation of βCD (Figure 4.15). Additional melting point
analysis revealed that βCD degradation passes through a caramelisation
phase, where the white βCD powder quickly undergoes three sequential stages
of browning, melting and charring (Figure 4.16).

79
Figure 4.16 The decomposition of βCD observed by melting point analysis. At 317 °C,
melting and caramelisation are dominant, but above 328 °C charring prevails.

The low-temperature scan to determine the Tg was initially cycled three times
from –100 to 100 °C to eliminate effects caused by the presence of water in the
thermogram (Figure 4.17). A second-order phase transition at 85 °C persisted
during these temperature cycles. Earlier reports attributed this phenomenon to
the dissolution of βCD crystals in water already present in the sample [45].
Molecular dynamics simulations predicted a T g of 61 °C for βCD [127] whereas
others determined an empirical Tg of 216 °C [128].

The temperature extreme was gradually increased by 10 °C up to a maximum


of 140 °C (Figure 4.17). The phase transition stabilised at ~100 °C in all
subsequent runs. The shifting of the transition to higher temperatures reflected
the evaporation of water molecules within the βCD cavity, which requires more
energy. This experiment suggested that the transition is not due to the
dissolution of βCD in water, as previously suggested [45]. The second-order
transition at 61 °C may represent the Tg of βCD predicted by modelling [127].

80
Figure 4.17 DSC heating curves of βCD (10 °C min-1) from –100 to 100 °C in the first
cycle (black), from –100 to 120 °C in the second cycle (blue), and from –100 to 140 °C
in the third cycle (red).

4.3.1.2 Thermo-mechanical characterisation of βCD

DMA was used to determine the storage modulus (E′) and damping factor (tanδ)
of βCD over three heating/cooling cycles. One difference between DSC and
DMA is the higher temperature at which transition occurs due to the effect of the
sinusoidal stress frequency during mechanical analysis [129]. The temperature
range was then set from –100 to 140 °C at 10 °C min-1 in order to investigate
the second-order transition due to the presence of βCD in the cross-linked
compounds. There were no significant differences between the second and
third thermal cycles. The last thermal cycle is shown in Figure 4.18.

81
Figure 4.18 DMA showing variation in the storage modulus (E′, solid lines) and tanδ
(dashed lines) of βCD (10 °C min-1, 1 Hz, third temperature cycle, –100 to 140 °C,
aluminium pocket).

The E′ values of the pure βCD sample were inversely related to the
temperature, whereas the tanδ values remained relatively constant during each
temperature cycle, showing there was no phase transition. This is consistent
with earlier experiments that defined βCD as a crystalline compound [14]. A
minor hysteresis was observed between cooling and heating. The lower E′
value during heating reflects the higher degree of relaxation in the material at
high temperatures. Therefore, the higher E′ value during cooling is due to the
stress generated by the high cooling rate.

The variation in E′ and tanδ was also investigated as a function of the oscillation
frequency. The increase in frequency during the temperature cycle had no
significant influence on either value. The corresponding thermogram is shown in
Figure A 4.

82
4.3.1.3 Thermal characterisation of TEGDGE, HEGDGE, and PEGDGE

The cross-linkers were also characterised by DSC, and their degradation


profiles (30–500 °C) and low-range temperature profiles (–100 to 140 °C) are
compared below.

Figure 4.19 Combined DSC/TGA thermograms of the cross-linkers TEGDGE (black),


HEGDGE (blue) and PEGDGE (red), from 30 to 500 °C (10 °C min-1, third temperature
cycle, aluminium crucible).

Heating the cross-linkers highlighted a continuous loss of substrate mass


starting at 160 °C. Part of the mass of the cross-linkers was lost even before the
onset of degradation (Table 4.8). The decomposition of TEGDGE was more
endothermic than its longer analogues, but the degradation events cannot be
analysed in detail using the available thermogram data. The TGA results (Table
4.8 and Figure 4.19) indicate that longer cross-linkers are more thermally
stable, which reflects the degradation of PEGDGE at 312 °C compared to the
shorter TEGDGE at 214 °C. The thermal stability of the longer chain can be
attributed to the scission mechanism of the chain extremities that occurs before
its complete decomposition. The PEGDGE cross-linker showed a mass loss of
27% before the main degradation event at 312 °C, which was greater than the
18–19% observed for the other cross-linkers. This can be attributed to the
degradation of the shorter chains present in the blend, which degraded earlier

83
than the predominant long-chain fraction resulting in a higher initial loss
compared to the two monodisperse cross-linkers (Table 4.8).

Table 4.8 Combined DSC/TGA data for the cross-linkers.

Mass loss
Number of Onset of
Mw before onset
Cross-linker ethylene degradation
-1 of degradation
glycol units (g mol )
(°C)
(% w/w)

TEGDGE 3 262 18 214

HEGDGE 6 394 19 278

PEGDGE 9 500 27 312

The Tg was initially characterised by three temperature cycles from –100 to


100 °C. Subsequently, the range was extended to 120 °C to completely remove
any water in the samples. The increase in temperature did not change the T g of
the cross-linkers (Figure 4.20). There were no significant differences among the
three runs and the third heating cycle of –100 to 120 °C is shown as an
example in Figure 4.20.

84
Figure 4.20 DSC thermogram of the cross-linkers: heating (solid) and cooling (dashed) of TEGDGE (black), HEGDGE (blue) and

PEGDGE (red) from –100 to 120 °C (10 °C min-1, third temperature cycle, aluminium crucible).

85
DSC analysis indicated the presence of a single, well-defined Tg for TEGDGE at
–78°C due to the monodisperse and pure nature of the sample. For HEGDGE,
the thermogram highlighted the presence of a first-order transition attributed to
chain melting as well as the glass transition beginning at –13 °C and peaking at
4 °C. Indeed, the baseline of the thermogram changes when the thermal event
concludes. The cooling curve for HEGDGE also reflected the sum of the two
transitions, with two crystallisation peaks centred at –41 °C and a lower
baseline after the thermal transition.

During heating (–100 to 120 °C), the heat flow adsorption of PEGDGE
increased between –70 and 10 °C in a multi-step transition that represented the
polydispersity of the sample as described by the manufacturer, the combination
of longer polymer chains melting (centred at 1 °C) and the increased mobility of
the shorter and middle-length chains after the glass transitions at –70, –18°C
and –37 °C, respectively. The multiple changes in the baseline from –71 °C
reflected a series of glass transitions involving different middle-length and
shorter polymer chains that can also act as plasticisers for the longer chains, as
reported for polyethylene glycol based polymers [130].

During cooling, the longer chains of HEGDGE and PEGDGE crystallised within
the temperature range from –23 to –50 °C, whereas the Tg of TEGDGE is
reiterated at –80°C. The cooling curves support the hypothesis that exotherms
during the crystallisation of HEGDGE and PEGDGE mask a vitrification event
that would explain the change in the baseline of the heat flow.

4.3.1.4 Thermo-mechanical characterisation of TEGDGE, HEGDGE,


PEGDGE

DMA was also used to determine the thermo-mechanical properties of the


cross-linkers, with three heating/cooling cycles from –100 to 100 °C at 10 °C
min-1. When PEGDGE was analysed, there were no significant differences
among the thermal cycles so only the third cycle is shown in Figure 4.21. The E′
value fell during heating, with two main transitions at –70 and –22 °C
corresponding to the glass transitions of the PEGDGE sample. The broad and
laddered decline in E′ reflects the specific properties of this commercial product,

86
a blend of PEG chains with an average molecular weight of 500 Da and a
polydispersity index of 1.7.

Figure 4.21 DMA showing variation in the storage modulus (E′, solid lines) and tanδ
(dashed lines) of PEGDGE (10 °C min-1, 1 Hz, third temperature cycle from –100 to
100 °C, aluminium pocket).

Previous studies have shown that the T g of PEGDGE is inversely related to its
polydispersity because the shortest polymer chains act as plasticisers, reducing
the brittleness and tensile strength while increasing the overall impact strength
of the material [46,47]. During cooling, PEGDGE underwent a single glass
transition event, increasing its stiffness. However, the rapid increase in E′ from
about –30 °C was followed by a sudden drop at about –75 °C, the latter
corresponding to 28% of the maximum E′ value. This phenomenon also
occurred in tests conducted at the lower heating/cooling rate of 2 °C min -1.
DMA was used to test the aluminium support in order to eliminate artefacts
caused by the machine and/or support matrix (Figure A 5). The E′ value of the
support showed linear variation within the experimental temperature range,
suggesting that the observed phenomenon is caused by the viscoelastic
behaviour of the PEGDGE and its strong interaction with the supporting pocket.
Hydroxyl, epoxy and carboxyl groups in polymers allow the formation of

87
hydrogen bonds with metal and glass [48]. The strong adhesive interactions and
the change in the stiffness of the PEGDGE sample during the glass transition
promote internal stress which is suddenly released, initiating cracking and de-
bonding of the sample from the support, recorded as a sudden drop in the E′
value. The physical damage (such as cracking) that emerged during the cooling
cycle provided an interesting initial assessment of the adhesive strength
between the sample and different support matrices. The same thermal profile
during the second and third cycles confirmed that PEGDGE recovers its initial
mechanical properties when heated above the T g in each cycle, and the
process is therefore reversible.

The effect of the oscillation frequency on the mechanical properties of PEGDGE


was tested at 1, 5 and 10 Hz. The E′ curves at 1 and 10 Hz are shown in Figure
4.22. The cooling and heating curves at both frequencies diverge at the point of
glass transition. During heating, the glass transition of the compound is
complete at ~25 °C as confirmed by the small decrease in E′ and the
overlapping of the curves at different frequencies. The Tg is directly proportional
to the frequency during both heating and cooling, and thus increases at higher
frequencies [49,50]. The drop in E′ value caused by cracking is also influenced
by the oscillation frequency: the E′ peak at 1 Hz occurs at –78 °C whereas the
10 Hz peak occurs at –70 °C.

88
Figure 4.22 DMA showing variation in the storage modulus (E′) of PEGDGE with
frequency (10 °C min-1, 1 and 10 Hz, third temperature cycle from –100 to 100 °C,
aluminium pocket).

DMA revealed that HEGDGE behaved similarly to PEGDGE (Figure 4.22).


During cooling, HEGDGE underwent significant stiffening and the E′ and tanδ
values increased below –20 °C. The E′ values dropped when HEGDGE passed
the vitrification point, and it cracked due to adhesion with the support. When
HEGDGE was heated, the rigidity decreased above –78 °C in a two-step
transition that concluded at 20 °C. The tanδ value gradually increased when
the sample was heated, reaching its maximum at 1 °C. The cross-linker
undergoes a broad transition representing the sum of its glass transition and
melting as observed by DSC (Figure 4.20). In addition, the drastic fall in tanδat
1 °C revealed the point at which the compound becomes a liquid.

89
Figure 4.23 DMA showing variation in the storage modulus (E′, solid lines) and tanδ
(dashed lines) of HEGDGE (10 °C min-1, 1 Hz, third temperature cycle from –100 to
100 °C, aluminium pocket).

The transition was also affected by the frequency of oscillation: a higher


frequency increased the E′ value by 2% up to 20°C (Figure 4.24).

90
Figure 4.24 DMA showing variation in the storage modulus (E′, solid lines) of HEGDGE
with the frequency (10 °C min-1, 1 and 10 Hz, third temperature cycle from –100 to 100
°C, aluminium pocket).

The DMA data for TEGDGE (Figure 4.25) was compared with the DSC results
(Figure 4.15). The Tg of the cross-linker was indicated by the tanδ peak at –71
°C and the transition was neat compared to the other cross-linkers, due to its
monodispersity and the absence of crystallisation in this short-chained polymer.

91
Figure 4.25 DMA showing variation in the storage modulus (E′, solid lines) and tanδ
(dashed lines) of TEGDGE (10 °C min-1, 1 Hz, third temperature cycle from –100 to 100
°C, aluminium pocket).

The Tg was found to be dependent on the frequency, as was the case for the
other cross-linkers. Due to the extreme vicinity of the Tg to the lowest extreme,
the temperature range was extended to the lower temperature of –140 °C to
determine whether cracking occurred. Accordingly, TEGDGE began to crack at
–100 °C (Figure 4.26).

92
Figure 4.26 DMA showing variation in the storage modulus (E′, solid lines) and tanδ
(dashed lines) of TEGDGE in an extended temperature scan (10 °C min-1, 1 Hz, third
temperature cycle from –140 to 100 °C, aluminium pocket).

4.3.2 Thermal and thermo-mechanical characterisation of inert


βCXCDs
Thermal analysis using DSC, TGA and DMA as described above was also used
to characterise the βCXCDs, to determine how the cross-linking conditions
influenced the thermo-mechanical properties of each βCXCD system.

4.3.2.1 Thermal characterisation of inert βCXCDs

The thermal degradation of all βCXCDs revealed similar thermal properties.


Table 4.9 summarises the Tdec, Tg and mass loss data. The cross-linking
reaction increased the Tdec of βCXCDs compared to the single cross-linkers,
reflecting the presence of βCD in the compounds, thus increasing the
temperature at which thermolysis affected their stability. The effect was stronger
in the βCTCD products (containing TEG units) due to their higher βCD content.
The Tdec of the βCXCD products was linearly dependent on the length of the

93
cross-linker, as seen for the single cross-linkers (Section 4.3.1.3). Accordingly,
the βCHCDs (containing HEG units) were associated with higher Tdec values
than the βCTCDs (containing TEG units). However, the βCPCDs (containing
PEG units) did not follow this trend, with Tdec values lower than the βCTCDs.
This reflects the presence of short chains in the polydisperse PEGDGE reagent
and subsequently in the cross-linked products, which set up decomposition
hotspots in the blend earlier than expected for a pure compound. A pure
compound with longer chains would have a higher T dec than HEGDGE.

Replicate samples produced using the same reaction conditions differed slightly
in terms of their mean cross-linker ratio due to the uncontrolled nature of the
reaction and the uncontrolled selection of hydroxyl groups that react with the
cross-linkers. However, products with the same feed ratio showed similar
thermal profiles (Table 4.9).

The glass transition event attributed to βCD was difficult to assess in the
βCXCDs. The Tg was inconsistent in the replicates and was only visible in
βCTCDs, and in the βCPCDs and βCHCDs with cross-link ratios of 3.8:1 or
1.8:1. This trend suggests that the transition is registered by DSC only when the
amount of βCD exceeds 51 % w/w. The mass lost during decomposition did not
differ significantly among the βCXCDs indicating similar thermodynamic
decomposition behaviours.

94
Table 4.9 Summary of the thermal properties of the βCXCDs.

Cross-
linker:βCD XEG:βCD Tdec Tdec Tg Tg βCD Mass loss
Cross- ratio midpoint midpoint decomposition
Sample onset peak
linker feed 1
by H NMR (°C)1 (°C)1 (°C) (°C) (%)
ratio
βCD - - - 318 328 - 101 -
TEGDGE - - - 2031 - -79 - -
HEGDGE - - - 2731 - -132 - -
PEGDGE - - - 3391 - -713 - -
IT1 5 :1 3.6 :1 252 287 - 99 84
IT2 4 :1 3.1 :1 257 277 - 95 80
TEGDGE
IT3 3 :1 2.4 :1 268 289 - 94 83
IT4 2 :1 1.9 :1 259 277 - 94 82
IH1 5 :1 4.0 :1 278 308 -10 - 87
IH2 4 :1 3.2 :1 279 303 -6 - 84
HEGDGE
IH3 3 :1 2.6 :1 276 307 - - 80
IH4 2 :1 1.8 :1 274 301 - 101 74
IP18 5 :1 4.1 :1 248 257 -20 - 87
IP19 5 :1 4.0 :1 247 256 -18 - 88
IP13 5 :1 3.8 :1 252 261 -19 110 -

95
Cross-
linker:βCD XEG:βCD Tdec Tdec Tg Tg βCD Mass loss
Cross- ratio midpoint midpoint decomposition
Sample onset peak
linker feed 1
by H NMR (°C)1 (°C)1 (°C) (°C) (%)
ratio
IP20 5 :1 3.7 :1 250 256 -19 - 84
IP21 5 :1 3.6 :1 247 256 -13 104 84
PEGDGE
IP9 5 :1 3.5 :1 237 254 -14 106 81
IP14 4 :1 3.2 :1 242 252 -8 96 81
IP15 4 :1 3.0 :1 236 250 -7 90 -
IP22 4 :1 2.9 :1 239 250 -3 90 80
IP23 4 :1 2.6 :1 239 246 -6 108 88
IP24 3 :1 2.5 :1 228 248 +5 106 82
IP25 3 :1 2.1 :1 226 243 - 102 84
IP17 2 :1 1.6 :1 221 233 - 90 75
1 DSC onset was used instead of TGA to determine the mass loss during decomposition (Figure 4.19, p. 83).
2 Onset of the transition of HEGDGE (Figure 4.19, p. 83).
3 First glass transition of PEGDGE (Figure 4.19, p. 83).

96
All DSC experiments were conducted in duplicate and the values shown in
Table 4.9 are averages. However, the replicates gave consistent results
differing by only ±1 °C from the average value.

The thermogram of product IP18 (PEG:βCD ratio = 4.1:1) compared to its


precursors (PEGDGE and βCD) is shown as a representative of the other
βCXCDs in Figure 4.27.

Figure 4.27 Combined DSC analysis and TGA of IP18 (PEG:βCD ratio = 4.1:1, red
line) and its precursors βCD (blue) and PEGDGE (green) over the temperature range
30–500 °C (10 °C min-1, aluminium crucible).

The thermogram of the product (red line) in each experiment indicates that
degradation occurred at 257 °C, whereas βCD degraded at 328 °C (blue line)
and PEGDGE initiated at 180 °C with a major drop at 338 °C (green line). The
thermal stability of each βCXCD lies midway between that of its precursors
(Table 4.9). Specifically, βCD improves the thermal stability of the short-chained
cross-linked systems (βCTCDs and βCHCDs) but reduces the thermal stability
of systems with longer cross-linkers (βCPCDs). The degradation of the
compound resembled the βCD thermogram, with a major endothermic event at

97
257 °C suggesting that the cross-linked products undergo a
degradation/caramelisation process.

Figure 4.28 DSC analysis showing the plasticising effect of water in IP18 from –100 to

120 °C (10 °C min-1, third temperature cycle, aluminium crucible).

The water entrapped by βCD affected the Tg of βCD and βCXCD (IP18) in a
similar manner, hence the water was eliminated during a first heating cycle up
to 130 °C to achieve the Tg of the pure compounds without the plasticising
effect of the water. The iterative heating cycles eliminated the endotherm
present in the first heating curve caused by the evaporation of water. The
second and third heating runs were stopped at 120 °C to reduce the thermal
shock on the βCXCDs. The absence of water therefore caused the Tg to
increase from approximately –23 to –20 °C, confirming its plasticising effect in
the βCXCDs. The third heating of each compound was used to report the Tg of
the dry product (Table 4.9). Comparison of the three cross-linked products IT1,
IH1 and IP13, which feature high cross-link ratios of 3.6:1, 4.0:1 and 3.8:1,
respectively (Table 4.9), indicated that the cross-linker type affects the thermal
properties of the products. The Tg of the three different βCXCD products based
on a temperature scan from –100 to 120 °C is shown in Figure 4.29.

98
Figure 4.29 DSC analysis of βCXCD products with high XEG:βCD ratios, namely
IT1 (3.6:1, black line), IH1 (4.0:1, blue line) and IP13 (3.8:1, red line) from –100 to 120
°C (10 °C min-1, third temperature cycle, aluminium crucible).

The cross-linker chains dominated the behaviour of the cross-linked samples at


low temperatures. A glass transition occurred in at least part of the products
with a notable shift towards positive temperatures compared to the cross-linkers
alone (Table 4.9). However, softening was observed in all the tested products
even when no evident glass transition occurred, as represented by the black
curve in Figure 4.29 related to the βCTCDs. The Tg was absent from products
cross-linked with TEGDGE, and general softening occurred in the entire thermal
spectrum. The products synthesised with the longer cross-linkers retained the
glass transition if the HEG:βCD ratio exceeded 3:1 or the PEG:βCD ratio
exceede 2.5:1 (Table 4.9). The heating curve revealed that the glass transition
of the βCXCDs was broad and gradual, as seen in the PEGDGE cross-linkers.
The broadness of the transition depended on the polydispersity of the products,
as also highlighted by GPC experiments. The secondary transition attributed to
βCD occurred at lower temperatures in the products cross-linked with TEGDGE
and PEGDGE, with few exceptions (Table 4.9). The βCHCD curves showed no
evidence of a secondary transition other than IH4 (HEG:βCD ratio = 1.8), which
contains the least amount of cross-linker. This general trend suggests that the
small secondary transition becomes visible by DSC only when a significant
proportion of βCD is present in the system.

99
4.3.2.2 Thermo-mechanical characterisation of the inert βCXCDs

The cross-linked βCPCD samples were tested with three different supports to
determine whether the adhesion of the material to the supports influenced the
tendency to crack and/or affected the thermal transitions. The supports were
aluminium pockets (Figure 4.30a), a stainless steel mesh (Figure 4.30b), or
aluminium pockets with PTFE tape (Figure 4.30c).

Figure 4.30 Photographs of βCPCD samples supported by (a) an aluminium pocket,


(b) a stainless-steel mesh, and (c) an aluminium pocket wrapped in PTFE tape.

4.3.2.2.1 Analysis of the βCXCD product in aluminium pockets

DMA was used to determine if the type of cross-linker and amount (XEG:βCD
ratios ranging from 4.3:1 to 1.6:1) affected the thermo-mechanical properties of
βCXCD. Initially the tests were performed with the samples held in aluminium
pockets as recommended by the DMA manufacturer for materials incapable of
self-support. The temperature was extended from 100 to 140 °C at 10 °C min -1
in order to cover the shift in temperatures for the minor second-order transition
identified by DSC. The DMA and DSC data are compared in Table 4.10.

The change in heat flow of the Tg recorded by DMA reduced as the amount of
βCD present in the βCXCDs increased. Furthermore, the secondary transition
captured by DSC was not detected by DMA up to 140 °C. The temperature
range was not extended any further to avoid the degradation of the βCXCDs.
The thermogram of IP9 (PEG:βCD ratio = 3.5) is used as an example to discuss
the significant changes in E′ and tanδ values at Tg (Figure 4.31).

100
Table 4.10 Comparison of DMA and DSC results for βCXCD samples and their
precursors.

Tg (°C)
Sample Cross-link ratio
tanδ DMA1 STD2 DSC

IT1 3.6 - -

IT2 3.1 - -

IT3 2.4 - -

IT4 1.9 - -

IH1 4.0 31 6.4 -10

IH2 3.2 53 4.5 -6

IH3 2.6 72 4.0 Soft

IH4 1.8 104 7.0 Soft

IP18 4.1 16 1.2 -19

IP19 4.0 41 4.8 -18

IP13 3.8 44 3.6 -19

IP20 3.7 47 3.1 -19

IP9 3.5 27 3.9 -16

IP15 3.0 67 7.5 -7

IP22 2.9 48 3.6 -3

IP17 1.6 - -
1 Third cooling cycle at 1 Hz, Tg determined as the tanδ peak.
2 Standard deviation (STD) of three replicates.

101
Figure 4.31 DMA showing variation in the storage modulus (E′, solid lines) and tanδ
(dashed lines) of sample IP9 (10 °C min-1, 1 Hz, third temperature cycle from –100 to
140 °C, aluminium pocket).

Figure 4.31 shows the E′ and tanδ values for sample IP9 (containing 67% w/w
PEGDGE) during the third temperature cycle from –100 to 140 °C. There were
minor differences in each value between the first cycle and the second and third
cycles, suggesting that the cross-linked system retained water adsorbed from
the air and during the purification process.

During heating, DMA revealed a gradual stepwise decrease in E′ between the


temperature extremes with a major step at 55 °C coinciding with a tanδ peak.
This broad transition was frequency dependent, suggesting it represented a
glass transition event for sample IP9 highlighted at –10 °C by DSC. The
softening of the IP9 product was represented by a change in the slope of the E′
curve at –31 °C, matching the onset of the tanδ peak, and this relates to the
viscoelasticity of the PEG soft segments present in all the cross-linkers. The
difference in Tg determined by DSC and DMA was much larger than the typical
shift of 10–20 °C often reported in the literature at 10 Hz [131,132].

102
With an aluminium support in place, the DMA thermogram for IP9 showed a
similar profile to PEGDGE during the cooling phase, which can be attributed to
the cracking of the sample. This was shown by the sudden drop in E′ from its
maximum value to 31% at around –30 °C. The temperature at which cracks
begin to appear in the IP9 sample was not significantly dependent on the
oscillation frequency, whereas cracking of the PEG chains in PEGDGE shifted
from –78 °C at 1 Hz to –70 °C at 10 Hz. The cracking temperature was not
dependent on the material used, but was dependent on overall changes in the
sample preparation method and differences in stress dissipation within the bulk
compound. The lower mobility of the cross-linked samples reflects the existence
of a cross-linked matrix compared to the viscous PEGDGE liquid, which has a
higher adhesive surface area.

Figure 4.32 DMA showing variation in the storage modulus (E′) and tanδ of sample IP9
(PEG:βCD ratio = 3.5, solid and dotted red lines) and IP17 (PEG:βCD ratio = 1.6, solid
and dotted blue lines) (10 °C min-1, 1 Hz, third temperature cycle from –100 to 140 °C,
aluminium pocket).

The DMA curves of samples IP9 (PEG:βCD ratio = 3.5) and IP17 (PEG:βCD
ratio = 1.6) are compared in Figure 4.32. The lower PEG content of IP17
resulted in cracks appearing at a much higher temperature (35 °C) and the E′

103
value dropped to 10% of its maximum value. In comparison, the E′ value of IP9
with the higher PEG content fell to only 30% of its maximum. The relatively low
cross-linker content of sample IP17 therefore appears to cause earlier cracking
compared to sample IP9 and reduces its viscoelastic behaviour, also limiting its
binding strength given that the low E′ value is related to the strength of adhesion
to the metallic support. The glass transitions determined from the peak tanδ
values shift to higher temperatures when the PEG content is low. The absence
of first-order transitions in the βCPCD samples during cooling was confirmed by
DSC (Figure A 6) suggesting that βCPCD products are completely amorphous
at temperatures between –100 and 140 °C.

4.3.2.2.2 Analysis of the βCXCD product on a steel mesh support

The interaction between the βCXCDs and the support had a noticeable effect in
the cooling thermogram. Sample IP9 (PEG:βCD ratio = 3.5) was analysed with
a stainless-steel mesh in lieu of the aluminium pocket. The sample was spread
over the mesh prior to analysis (Figure 4.30). The glass transition of IP9 as
determined by DMA was interrupted by stress relaxation at 15 °C (Figure 4.33).
Cracking was still observed during cooling, but the phenomenon was less
severe compared to the sample held in an aluminium pocket (Figure 4.33). The
cracking manifested itself over a wider temperature range (–14 to –70 °C)
compared to the sudden drop observed with the aluminium pocket. The
characteristic drop in E′ began at about –17 °C and reached 3% of the
maximum value. The drop was visible during each cycle, confirming the stress
relaxation effect that occurs with each round of heating. The metallic mesh
changed the manner in which the stress introduced during each cooling cycle
was dissipated. The mesh had a greater surface area to which IP9 could bind
(362 mm2) compared to the aluminium pocket (180 mm2) but nevertheless
allowed effective stress dissipation within the bulk sample.

104
Figure 4.33 DMA showing variation in the storage modulus (E′, solid lines) and tanδ
(dashed lines) of IP9 (10 °C min-1, 1 Hz, third temperature cycle, –100 to 140 °C,
stainless-steel mesh).

4.3.2.2.3 Analysis of the βCXCD product in aluminium pockets with PTFE tape

In order to confirm that the drop in the E′ values during cooling was due to the
de-bonding of the sample from its support, PTFE tape was used to hold
βCPCD1 samples within an aluminium pocket, thus allowing the samples to
contract freely as the temperature declined. Preliminary DMA characterisation
of the PTFE tape (blank) showed a softening at the onset temperature of 18 °C
(Figure A 7). The absence of adhesion between sample IP9 and the support
when the sample was enfolded by PTFE tape was confirmed by the drop in E′
during cooling (Figure 4.34).

105
Figure 4.34 DMA showing variation in the storage modulus (E′, solid lines) and tanδ
(dashed lines) of IP9 (10 °C min-1, 1 Hz, third temperature cycle, –100 to 140 °C,
aluminium pocket and PTFE tape).

The hysteresis between the heating and cooling curves during the same
temperature cycle diminished with each cycle. Therefore, the PTFE tape almost
completely eliminated the de-bonding phenomenon and the thermal stress in
the sample during the cooling phase. The T g of IP9 under these conditions was
41°C, identical to the value recorded in the aluminium pocket without tape,
confirming that the experimental setup does not affect the T g.

4.3.2.3 Analysis of the βCXCDs by optical microscopy

Optical microscopy was used to characterise the cracking of the IP9 product
(PEG:βCD ratio = 3.5) in more detail during cooling (Figure 4.35). A glass
support was considered adequate as a transparent substitute for the metallic
support we used to investigate the adhesion of the compound. The sample was
cooled from 25 °C (Figure 4.35a) and began to crack at about –55 °C (Figure
4.35b), with the cracks propagating further as the temperatures was reduced to
–100 °C (Figure 4.35c,d). The same sample was then heated to 100 °C and the
cracks began to self-repair, starting at 0 °C (Figure 4.35e) until healing was
complete at ~80 °C (Figure 4.35h).

106
Figure 4.35 Optical microscope images of IP9 (PEG:βCD ratio = 3.5) during the first
temperature cycle from –100 to 100 °C. The top row (a-d) shows the cooling phase
captured at 25 °C, –55 °C, –78 °C and –80 °C and the bottom row (e-h) shows the
heating phase captured at 2 °C, 43 °C, 67 °C and 87 °C .

During the second thermal cycle in the same sample, cracking began at –58°C
and initiated at a different location (Figure 4.36c). This confirms the evidence
provided by the similar E′ values in each thermal cycle, i.e. the thermal history is
erased by heating as previously reported for a hydrogel CD/PEG system [33].
During the second cycle, sample healing occurred at ~80 °C as in the first cycle
(Figure 4.36h). Furthermore, IP9 placed on PTFE tape showed no evidence of
cracking during the cooling phase, confirming that stress relief by cracking was
due to the bonding of IP9 to the glass support (Figure A 8).

Figure 4.36 Optical microscope images of IP9 (PEG:βCD ratio = 3.5) during the
second temperature cycle from –100 to 100°C. The top row (a-d) shows the cooling
phase captured at 100 °C, 25 °C, –61 °C and –88 °C and the bottom row (e-h) shows
the heating phase captured at 5 °C, 43 °C, 68 °C and 84 °C.

107
This experiment confirmed the self-healing of the compound and explained why
the E′ curves are identical during multiple temperature cycles. The self-healing
behaviour is thought to reflect the reformation of hydrogen bonds and host–
guest interactions as seen in CD hydrogel systems due to the lower viscosity of
the cross-linked system when heated above the Tg [33,52–54]. The physical
transformation occurs at the onset of material flow behaviour, allowing the crack
to be filled in, and probably involves a rheological model involving the snake-like
displacement of the polymeric chains, described as reptation [55].

4.3.2.4 Qualitative assessment of the self-healing properties

Figure 4.37 Self-healing of IP13 (PEG:βCD ratio = 3.8). (a) Sample after solvent
evaporation. (b) Sample after cutting. (c) The parts are placed in contact. (d) The
sample is heated to 70 °C for 30 min and pulled by the extremities.

The rupture strength of sample IP13 was qualitatively assessed by intentionally


cutting the sample (Figure 4.37b). The parts were subsequently rejoined and
annealed in the oven at 70 °C for 30 min (Figure 4.37c). The joined parts were
found to be cohesive after this qualitative test (Figure 4.37d). The sample was
gripped at its extremities and pulled gently at room temperature. The sample
showed considerable elongation. Diffusion of the polymeric chains by reptation
and the reforming of hydrogen bonds in the fracture allowed the material to
qualitatively self-repair and self-heal.

4.3.2.5 Scanning electron microscopy

The surface morphology of the βCXCDs was explored by scanning electron


microscopy (SEM) and compared to the precursor βCD, which is a crystalline
material with roughly square particles of different sizes (Figure 4.38).

108
Figure 4.38 Scanning electron micrograph of βCD crystals.

In contrast to the precursor, the surface of the βCXCDs was more amorphous
due to the cross-linking reaction. The foamy appearance of the surface was due
to the evaporation of water during the drying process. The βCTCD samples with
various TEG:βCD ratios showed significant differences when analysed by SEM
(Figure 4.39). The surface of IT1 (TEG:βCD ratio = 3.6) was amorphous with
large aggregates, whereas the higher TEG content of samples IT2 (TEG:βCD
ratio = 3.2), IT3 (TEG:βCD ratio = 2.4) and IT4 (TEG:βCD ratio = 1.8) caused
the particles to appear progressively less amorphous and to resemble the
crystalline morphology of native βCD.

109
Figure 4.39 Scanning electron micrographs of the βCTCDs (a) IT1 (b) IT2 (c) IT3 and
(d) IT4.

The βCXCDs containing HEG and PEG units were not significantly affected by
the degree of cross-linking and were characterised by a homogeneous smooth
surface (Figure 4.40). Crystalline domains were observed in the βCPCDs with
fewer cross-links (Figure 4.40d).

The upper part of Figure 4.40 compares the surface morphology of sample IH1
(HEG:βCD ratio = 4.0) and IH4 (HEG:βCD ratio = 1.8), whereas the lower part
compares samples IP19 (PEG:βCD ratio = 4.0) and IP17 (PEG:βCD ratio =
1.6). The crystalline domains in the less cross-linked sample IP17 may have a
higher βCD content given their similarity to native βCD. The exposure of these
regions to the electron beam for a few minutes in the vacuum chamber
generated a honeycomb structure, due to the evaporation of water and/or the
degradation of these regions.

110
Figure 4.40 Scanning electron micrographs of the βCHCDs (a) IH1 and (b) IH4, and
the βCPCDs (c) IP19 and (d) IP17.

4.3.2.6 The βCXCD product in an early-stage formulation

A basic formulation was prepared to study the effect of the inert binder on the
formulation’s mechanical properties. Sample IP17 (10% w/w) was mixed with
melamine in water. The melamine was used as inert substitute for RDX given
the similarities in their particle sizes. The mixture was left stirring for 1 h at room
temperature before drying under vacuum. The dry mixture contained
aggregates rather than finely-dispersed melamine. DMA experiments were
conducted, resulting in the thermogram shown in Figure 4.41.

111
Figure 4.41 DMA showing variation in the storage modulus (E′, solid lines) and tanδ
(dashed lines) of sample IP17 (10% w/w) formulated with 90% w/w melamine (10 °C
min-1, 1 Hz, third temperature cycle from –100 to 140 °C, aluminium pocket).

The formulation inherited some of the mechanical properties of the binder given
the Tg highlighted by the tanδ peak at 60 °C and the drop in E′ at -45 °C,
matching the values of the binder alone (Figure 4.31). The cracking attributed to
IP17 was not abolished in the formulation and a small drop in E′ was observed
at –12°C. The drop in E′ was lower than the equivalent drop for the pure
cross-linker.

4.4 Conclusions
This chapter summarised the outcome of cross-linking βCD with polyethylene
glycol diglycidyl ethers containing three, six and nine repeat units, as a first step
towards the synthesis of a new inert semi-synthetic binder for energetic
applications. The investigation focused on the optimisation of the cross-linking
reaction, which was strongly affected by several factors including NaOH
concentration and volume, reaction time, temperature, βCD alkoxide formation,
cross-linker:βCD ratio, and the addition rate of the cross-linker in the βCD
alkoxide solution. The optimal conditions for the reaction were 5.6% NaOH,

112
0.21 gβCD mL-1aq.NaOH, 30 °C, 16 h to form the βCD alkoxide, 20 min for the
addition of the cross-linker, a cross-linker:βCD feed ratio < 5:1, and 6 h for the
cross-linking. A unique synthesis procedure was adopted for all βCXCDs.
Samples with different cross-linker:βCD feed ratios (5:1, 4:1, 3:1 and 2:1) were
synthesised using the three different cross-linkers. Up to 25% of the cross-linker
was consumed in a parallel degradation reaction. Characterisation by 1H-NMR
and FTIR spectroscopy confirmed the formation of βCXCDs containing units of
1
βCD and XEG chains. Characterisation by H-NMR in D2O allowed the
XEG:βCD ratio in the repeat unit of the macromolecule to be calculated.
Samples with XEG:βCD ratios between 4:1 and 3:1 were malleable whereas
compounds with lower ratios were friable.

The compounds synthesised using the shorter cross-linker TEGDGE were


friable and easy to dissolve in water regardless of the TEG:βCD ratio. The
βCXCDs synthesised using longer cross-linkers (HEGDGE and PEGDGE) took
longer to dissolve when the XEG:βCD ratio defined by 1H-NMR was > 3.4:1.

The thermal and mechanical properties of the precursor βCD were tuned by the
cross-linking, giving βCXCDs with low Tg values (from –20 to 5 °C). The
softness of the βCXCD samples was dependent on the quantity and length of
the polyethylene glycol units. In products with a XEG:βCD ratio > 3.7:1, the Tg
was as low as –20 °C, whereas the glass transition was absent in the βCXCDs
with the lowest number of cross-links. No glass transition was observed in
βCTCDs, which contained the shorter TEG units.

The βCXCD samples were found to crack when cooled during DMA due to the
formation of strong hydrogen bonds with the metallic support. The βCXCDs also
healed rapidly when heated, retaining some of the healing properties of
hydrogels synthesised with βCD and polyethylene glycol reported in
pharmaceutical research projects. The characterisation of βCHCDs and
βCPCDs by SEM confirmed their amorphous nature, but the βCTCDs
increasingly resembled the crystalline morphology of native βCD when the
TEG:βCD ratio was < 2.4:1.

113
Having established a robust synthesis route for βCXCDs and investigated the
properties of the water-soluble products, the project moved on to the synthesis
and characterisation of nitrated derivatives, as described in Chapter 5.

114
5 Synthesis and characterisation of energetic cross-
linked βNCXCDs
In order to produce new binders that provide energy to energetic formulations,
the cross-linked βCXCDs systems described in Chapter 4 were nitrated using
conventional protocols. The nitration procedure was initially developed at a
scale of 200 mg. The energetic properties of the nitrated products were
assessed by Cranfield Defence and Security experts before synthesis was
scaled up to 1 g.

The number of βCD hydroxyl groups that can be nitrated, i.e. converted to
nitrato groups (ONO2), can vary from 1–3 per glucosidic unit of βCD and 1–21
per βCD molecule (Scheme 5.1). The available hydroxyl groups could become
sterically hindered when the precursor molecules are cross-linked to form
βCXCDs. Therefore, to ensure a high degree of nitration, a large excess of the
nitrating mixture was used as discussed below.

Two conventional synthesis routes used for the preparation of other nitro-esters
were tested [21]. Route 1 used a mixture of concentrated sulfuric acid (98%)
and nitric acid (95%) whereas route 2 used fuming nitric acid (Scheme 5.1).

Scheme 5.1 Synthesis routes selected to achieve the nitration of βCXCDs.

Route 1 is typically used for the synthesis of NC [27]. The sulfuric acid catalyses
the formation of the nitronium ion, resulting in an electrophilic attack on the
organic substrate, and its desiccant power removes the water generated during
nitration as shown in Scheme 5.2. The removal of water is a highly exothermic

115
process and has the potential to degrade βCD if not properly controlled. During
the synthesis of NC, sulfuric acid can also promote the formation of sulfate
esters during the nitration process, which affect the yield and stability of the
nitrated product. The presence of sulfate esters, which are sensitive to moisture
and temperature, could reverse the nitration reaction thus degrading the
nitroesters. This negatively affects the physicochemical properties and stability
of the products, as reported for NC [27].

Scheme 5.2 Nitration of βCD units: (a) the sulfuric acid catalyses the formation of
nitronium ions, and (b) the formation of nitroesters.

In route 2, fuming nitric acid was used as the nitrating mixture. This procedure
has already been used for the synthesis of nitrated βCD and the experimental
data from this PhD project can therefore be compared with data reported in the
literature [109,114,116].

Two-phase nitration was also attempted for the nitration of βCXCDs using the
halogenated solvent dichloromethane, which can act as a heat sink, removing
the nitrated product from the nitrating medium and increasing the yield. This
approach is commonly used in the synthesis of polymeric nitroesters such as
polyNIMMO and polyGLYN [133]. Two-phase nitration is also used to extract
nitrated organic substrates such as nitroglycerine from the nitrating mixture
[133]. However, dichloromethane was found to be inefficient for the nitration of
βCXCDs because neither the cross-linked precursors nor the nitrated products
dissolved in this solvent. Therefore, two-phase nitration was considered
unsuitable as method to produce nitrated βCD derivatives.

116
The βCXCD systems containing TEG, HEG and PEG were nitrated using
fuming HNO3. The conditions and work-up were optimised primarily using the
water-soluble βCPCD products, given the commercial availability of PEGDGE.
However, the insoluble products generated by cross-linking with TEGDGE and
PEGDGE were initially used to assess the efficiency of nitration of insoluble
precursors, as described in Section 5.1.

5.1 Attempts to nitrate the insoluble βCXCD precursors


The first nitration attempts targeted the insoluble products synthesised during
the optimisation of the cross-linking reaction conditions. The nitration of the
cross-linked products was influenced by three parameters:

 The physical properties of the starting material;


 The strength of the nitrating medium;
 The duration of the nitration reaction.

Four insoluble βCXCDs were chosen, namely samples IP26 and IT5–IT7
containing PEG and TEG units, respectively (Appendix, Table A 4) and nitrated
using the sulfuric and nitric acid mixture (Route 1, Scheme 5.1) or fuming nitric
acid (Route 2, Scheme 5.1). The degree of cross-linking reflected on the
swelling power of the two precursors in water. The optimisation of the synthesis
parameters was guided by the data collected during the characterisation of the
thermal properties of the βNCXCD products, as discussed below.

117
Figure 5.1 Physical appearance of the insoluble IP26 and IT5–IT7 precursors.

The insoluble βCXCDs were mixed with a large excess of the nitrating mixture
or fuming nitric acid (~1 mL) relative to the mass of cross-linked substrate (200
mg). The samples were crushed into small pieces before nitration to increase
the available surface area for the nitration. The nitration mixtures were stirred at
200–900 rpm during the trials to grind the largest particles and guarantee good
contact with the acids. The starting materials physically disintegrated under
these conditions to give a cloudy suspension, indicating that the insoluble
compounds were not efficiently converted to soluble nitrated derivatives. The
nitration reaction was interrupted after 1 h at room temperature by quenching
the mixture in excess water and ice. The crude slurry was rinsed with large
amounts of water to dilute and eliminate the acids and the solids were partially
extracted with acetone. The insoluble solid products were rinsed with water and
then acetone. Small fractions were recovered when rinsing with acetone and
these were dried. Both fractions were characterised by 1H-NMR spectroscopy
and DSC.

Several replicates of the nitration of βCTCDs containing TEG units (samples


IT5–IT7) led only to insoluble βNCTCDs, whereas the βNCPCD mixtures
generated small amounts of acetone-soluble material comprising highly-nitrated
derivatives. The nitrated samples were not soluble in acetone (Appendix, Table
A 5) and they showed good thermal stability with a decomposition peak at ~200
°C. The energy content of the samples was relatively low for energetic binders,

118
at 450–800 J g-1 (Table 5.2), and seemed to be independent of the nitration
conditions. Generally, the yields of the nitration reactions of βCTCDs were high
(450–1400 J g-1, Table 5.2). In contrast, the synthesis of βCPCDs from
insoluble precursors was more difficult given the physical properties of the
starting material, which was less gummy and friable. Nitration thus produced a
variety of nitrated insoluble compounds with different thermal properties, as well
as some acetone-soluble fractions.

In conclusion, the properties of the nitrated products were strongly dependent


on the method used to disperse the insoluble particles of the inert βCXCD
precursors in the nitration medium. The βCTCD precursors were well-dispersed
fine powders whereas the βCPCD precursors were hydrogels that were difficult
to penetrate and wet by the acids. The traces of βNCPCD soluble in acetone
consisted of highly nitrated oligomers with crystalline properties, similar to the
nitrated βCD (Table 5.1) and were not pursued further.

5.2 Attempts to nitrate the soluble βCXCD precursors


Initial tests revealed that the solubility of the βCXCD precursors in the nitrating
media was inversely proportional to the degree of cross-linking. The dissolution
of βCXCD substrates took up to 1 h at room temperature for derivatives with a
XEG:βCD ratio of 3:1, and less than 20 min when the XEG:βCD ratio was < 3:1.
(As stated in Chapter 4, the abbreviation XEG is used here to refer collectively
to the TEG, HEG and PEG segments in the cross-linked products.) The volume
of nitric acid used per 200 mg of βCXCD was therefore increased to 2.1 mL,
and most precursors took less than 40 min to fully dissolve. Subsequently, all
reaction mixtures were left to stir at room temperature for 1 h before quenching
in excess ice-water. The liquid was decanted and the solid residue was washed
several times with water, dissolved in a minimum amount of acetone, and
re-precipitated in water (450 mL). The product was dispersed in the water,
making it difficult to recover, so a half-saturated solution of brine was used
instead of distilled water in order to change the zeta potential of the solution and
promote the flocculation of product particles. The precipitate was subsequently
re-dissolved in acetone, the inorganic components were filtered off and the

119
βNCXCDs were dried under high vacuum. Small portions were then used for
chemical and thermal characterisation. The βNCXCD samples obtained from
soluble βCXCD precursors showed a high thermal stability (Tdec ~200°C, Edec
~1700 J g-1; Table 5.3). These values were also consistent when compared to
the nitrated products of the insoluble precursors, as discussed above.

The compounds emitted a smell consistent with traces of acid (NOx) when
briefly stored (hours) in a vial, indicating the purification method was inefficient.
Subsequent tests were therefore carried out to assess the efficiency of three
variations of the purification method for βNCXCDs:

 Multiple re-precipitations from acetone in water;


 Re-precipitation from acetone in boiling water;
 Re-precipitation from acetone in an aqueous CaCO3 suspension (100 or 25
°C).

When the compounds were purified by multiple re-precipitations in boiling water,


they retained the characteristic smell of NOx albeit less pungent compared to
the original purification method. The use of a boiling water/CaCO3 mixture was
adopted from the NC manufacturing process and was considered an
unnecessary extreme choice at this initial stage of research, where the stability
of the βNCXCDs was not fully understood. The crude nitrated products from the
initial ice-water wash were rinsed several times with water, re-dissolved in small
amounts of acetone and re-precipitated in an excess of water containing 1%
(w/w) CaCO3. The resulting products lacked the NOx smell suggesting that the
latter purification method was the most efficient.

A side experiment using a strong base (NaOH) at room temperature as a


neutralising agent resulted in the degradation of the product, with the release of
NOx from the nitrated product and absorption onto its surface (Figure 5.2).

120
Figure 5.2 Nitrated product degraded by the use of NaOH during purification.

A number of replicates confirmed that re-precipitation from acetone in 0.5% w/w


CaCO3 in water at room temperature gave consistent results with clean, acid-
free βNCXCDs, and this purification method was adopted for their preparation.

5.3 Synthesis and chemical characterisation of βNCXCDs


Based on the pilot studies described above, several βCXCD samples were
nitrated and extracted in acetone using the following general procedure. Nitric
acid (2.1 mL, 100%) was cooled below 10 °C in an ice-water bath, and 200 mg
of βCXCD was added in small portions over 5 min, ensuring that the
temperature remained below 10 °C. The crude slurry was then stirred at room
temperature for 1 h and the nitration mixture changed from the slurry to a clear
solution, indicating complete dissolution of the βNCXCD products. To quench
the reaction, the nitrating mixture was poured into excess ice-water and the
solid compounds were collected and washed with distilled water to eliminate
acid residues. The products were then re-dissolved in a small amount of
acetone and precipitated in an excess of half-saturated brine. Subsequent
re-precipitation from acetone in a 0.5% w/w aqueous CaCO3 suspension at
room temperature gave a crude precipitate which was dissolved in a small
amount of acetone and filtered to remove the inorganic salts. The βNCXCDs

121
were then chemically characterised by 1H-NMR spectroscopy. The nitrated
products were soluble in acetone-d6 and DMSO-d6 allowing the nitrogen content
to be determined (Table 5.1). The NMR spectra of the βNCXCDs were
compared with those of the βCXCD precursors, revealing similar profiles
although lower chemical shifts for the nitrated derivatives due to the presence of
the nitro groups.

An attempt was made to attribute the signals to specific protons or groups of


protons in the βNCXCD molecules. The literature provides a number of
assignments for the parent nitrated βCD molecule (βNCD), and the
corresponding 1H NMR and FTIR spectra were used as references. However,
the chemical characterisation of similar cross-linked systems has not been
published [116,134]. In order to facilitate the assignment of peaks, a sample of
βCD precursor was nitrated as an internal comparison using the method
described above for the nitration of βCXCDs. Figure 5.3 compares the 1H NMR
spectra of βNCD and sample NP1, prepared from the inert precursor IP18 with
a PEG:βCD ratio of 4.0:1. The similarities between the spectra confirmed the
formation of βNCPCDs.

122
Figure 5.3 The 1H NMR spectra of (a) βNCD [116] and (b) NP1 in DMSO-d6.

Figure 5.4 compares the 1H-NMR spectra of the inert precursor IP18 and a
representative nitrated sample NP1. The detailed properties of both samples
are summarised in Appendix A (Table A 1 and Table A 6).

123
Figure 5.4 Comparative 1H-NMR spectra of (a) inert precursor IP18 (PEG:βCD ratio =
2.5:1) and (b) its nitrated product NP1, both in DMSO-d6.

The assignment of 1H-NMR signals in the βNCXCD spectra was complex. The
spectrum of the nitrated sample is characterised by three sets of broad signals
at 5.70–5.20 ppm, 5.10–4.00 ppm and 3.80-3.40 ppm. The first set of peaks at
5.70–5.20 ppm indicated different chemical environments in the nitrated
macromolecule due to the presence of the stronger electro-withdrawing nitrato
groups. The anomeric proton H-1 of sample IP18 with a chemical shift centred
at 4.8 ppm was shifted at 5.60 ppm in the NP1 spectrum. The shoulder at 5.60
ppm is possibly the H-1 signal when the cross-link was added at position OH-2
of unit 2, as described in Scheme 5.1. The broad signals centred at 5.45 and
5.35 ppm can be assigned to protons H-2 and H-3 of the βCD units and H-b,
next to a chiral centre in the PEG chain. The region 5.10–4.00 ppm represents
the modified chemical shift of all the βCD and PEG units protons situated close
to the nitrato groups. The two broad signals between 5.15 and 4.65 ppm can be
assigned to Ha and H6 protons. The set of small, broad signals at 4.60–3.80
ppm can be assigned to all CH2 protons belonging to βCD and PEG units that

124
are not adjacent to a nitrato group, but are close enough to be influenced by
them. The complex environment between 3.80 and 3.40 ppm can be assigned
to the remaining βCD protons which are situated far enough away from the
nitrato groups to maintain the chemical shifts of their precursors.

No significant differences were observed among the spectra for βCTCDs,


βCHCDs and βCPCDs, containing TEG, HEG and PEG units, respectively, and
with similar XEG:βCD ratios (Figure 5.5).

Figure 5.5 The 1H NMR spectra of products (a) NP1, (b) NH1 and (c) NT1 in DMSO-
d6. The XEG:βCD ratios were previously determined by NMR spectroscopy.

The compounds with fewer cross-links featured a sharper H-1 anomeric signal
at 5.7–5.5 ppm (Figure 5.6). Fewer cross-links promote the comparable
conversion of all three available hydroxyl groups in the βCD unit into ONO2
groups (Figure 5.6d), whereas the conversion of OH-2 was more frequent in the
highly cross-linked products as highlighted by the presence of the shoulder at
5.7–5-6 ppm (Figure 5.6a).

125
Figure 5.6 The 1H NMR spectra of products (a) NP1, (b) NP2, (c) NP3 and (d) NP4,
which show a gradually decreasing PEG:βCD ratio.

The degree of substitution (DS) of the hydroxyl groups in βNCD was determined
by NMR spectroscopy as described in the literature [116], revealing that 2.6 of
the 3 hydroxyl groups available per βCD repeat unit were nitrated under the
adopted reaction conditions. The technique has an error of ± 2.5% in terms of
the nitrogen content [116] and the values were assessed also by iron sulfate
titration (Table 5.1) [135]. The nitrogen content of βNCD was also determined
by titration to confirm the consistency of the results.

The iron sulfate titration method is based on the hydrolysis of nitrato groups in
excess 98% sulfuric acid while stirring for 1 h at room temperature. The nitric
acid generated in this reaction forms NOx complexes with the iron titrant and
turns the solution pink, allowing quantification by colour change [135]. KNO3 is
used as a standard to determine the correction factor (F) from ideal to real
titration conditions using (3:

126
𝐴 𝑥 13.833 (3)
𝐹=
𝐵

where 13.833 is the nitrogen content of KNO3 (g mol-1), A is the weight of NT or


NH or NP (g), and B is the amount of titrant used to reach the endpoint (mL).

SeveralβNCXCDs were subsequently titrated and the nitrogen content then


determined using (44:

𝑉𝑋𝐹
𝑁% = (4)
mNC

where N% is the nitrogen content, mNC is the weight of the NT, NH or NP


samples (g), V is the amount of titrant required to reach equivalency (mL), and
F is the correction factor from Equation 3.

The nitrogen content was inversely dependent on two factors:

 The XEG:βCD ratio (degree of cross-linking);


 The length of the polyethylene glycol repeat unit in the βNCXCDs.

The TEG, HEG and PEG units increase the overall mass of carbon atoms in
the βNCXCDs and therefore reduce the percentage of nitrogen in the total
mass of the macromolecule. The only exception was NT4, which diverged from
the trend in all three triplicates and further investigation is therefore needed.
The DS in the βNCXCDs was generally greater than 2.1/3. The lower DS in the
βNCPCDs can be attributed to the lower efficiency of nitration in the
less-soluble βNCXCDs, which take longer to dissolve in the nitrating medium.

127
Table 5.1 Nitrogen content of the βNCXCDs.

Inert precursor
Nitrogen content DS per βCD
Sample unit
ID
Name XEG:βCD ratio Error (ideal max = 3)
(% w/w)1
(%)

βNCD - - 11.6 11 2.1

NT1 IT1 3.6 8.2 2 2.4


NT2 IT2 3.1 9.4 8 2.7
NT3 IT3 2.4 9.5 8 2.4
NT4 IT4 1.9 6.3 15 1.3

NH1 IH1 4.0 5.6 20 3.1


NH2 IH2 3.2 7.2 5 2.2
NH3 IH3 2.6 9.0 4 2.8
NH4 IH4 1.8 9.4 13 2.5

NP1 IP18 4.0 6.3 3 2.4


NP5 IP13/20 3.7 6.2 2 2.4
NP6 IP13/20 3.7 6.8 25 2.5
NP2 IP23 2.6 7.0 5 2.2
NP3 IP33 2.5 6.9 12 2.2
NP4 IP32 1.6 8.8 28 2.4
1 Values are means of three replicate experiments with a variance of ±0.2%.

128
The nitrated products were then characterised by FTIR spectroscopy in
attenuated total reflectance (ATR) mode for the rapid analysis of solid-phase
samples. A representative sample NP1 is compared with its precursor in Figure
5.7.

Figure 5.7 Comparative FTIR spectra of sample NP1 (solid line) and its inert precursor
IP18 (dashed line).

The absorption peak at 3440 cm-1 was assigned to the O-H stretching vibration
of the remaining hydroxyl groups in the cross-linked system, and the peak at
2900 cm-1 was assigned to the C-H stretching vibration. The absorbance was
significantly reduced at 3440 cm-1 and strong absorption peaks, attributed to the
introduction of nitrato groups, appeared at 1646, 1274 and 831 cm-1. Sample
NP8 was also compared with βNCD to identify any differences in the absorption
spectrum (Figure 5.8).

129
Figure 5.8 Comparative FTIR spectra of sample NP1 (solid line) and βNCD (dashed
line).

The NP1 spectrum revealed stronger absorption peaks at 2900 and 2800 cm-1
due to the vibration of the C-H bond of the methylene groups (symmetric and
asymmetric stretching) and weaker absorption by the OH groups (bending) at
3400 cm-1. This last difference in the OH bond vibration can be attributed to the
Mw of the two materials: free OH groups in βCD have a greater mass than the
same groups in the macromolecule.

The peaks at 1460 cm-1 and 1100 cm-1 were more intense in NP1 than βNCD
and were assigned to the bending vibrations of the C-H bond of the methylene
groups in the PEG units of NP1. There were no significant differences between
the nitrated systems with different XEG:βCD ratios (Appendix, Figure A 9).

The molecular weights of the βNCXCD derivatives were determined by GPC in


the solvent THF. Polystyrene standards were used for a nine-point calibration

130
(600 Da to 1.7 MDa) and all Mw values are reported are polystyrene
equivalents.

The broad chromatograms of the βNCXCDs were similar to their precursors and
confirmed the polymeric structure of the nitrated products (Figure 5.9). The Mw
distribution in the three βNCXCD systems is compared in Figure 5.9. The
systems have similar XEG:βCD ratios (~3.8) but contain different repeat units,
namely PEG (Figure 5.9a), HEG (Figure 5.9b) or TEG (Figure 5.9c). All
samples had a broad Mw distribution with the elution of high-Mw fractions
beginning at 12 min. Sample NP1 displayed a broader profile, whereas NH1
and NT1 differed slightly due to the presence of two major elution peaks as
observed following the GPC analysis of βCXCD.

131
Figure 5.9 Comparative GPC analysis of βNXCDs: (a) NP1 (PEG:βCD ratio = 3.8),
(b) NH1 (HEG:βCD ratio = 4.0), and (c) NT1 (TEG:βCD ratio = 3.8).

As expected, both the Mw distribution and polydispersity index of the samples


were dependent on the XEG:βCD ratio. As observed for the inert βCXCDs, the
GPC analysis of βNCXCDs confirmed that products with higher XEG:βCD ratios
(3.8:1) eluted at lower retention times corresponding to their higher Mw (Figure
5.10).

132
Figure 5.10 Comparative GPC chromatograms of βNPCDs with decreasing PEG:βCD
ratios: (a) NP1 (PEG:βCD ratio = 3.8), (b) NP2 (PEG:βCD ratio = 2.6), (c) NP3
(PEG:βCD ratio = 2.1), and (d) NP4 (PEG:βCD ratio = 1.6).

5.4 Thermal and thermo-mechanical characterisation of


energetic βNCXCDs
As discussed above, two sets of βNCXCDs were synthesised from inert
precursors having differing in terms of their solubility in water. The thermo-

133
mechanical characterisation of the inert βCXCDs by DSC, TGA and DMA
(Chapter 4) was therefore extended to the nitrated derivatives, the βNCXCDs.

5.4.1 Thermal characterisation of βNCXCDs prepared from insoluble


precursors
The thermal behaviour of βNCXCDs obtained by nitrating the insoluble βCXCDs
was variable. The thermal stability of the nitrated products was affected by a
combination of four different factors:

 Strength of the nitrating medium;


 Cross-link ratio of the inert precursor;
 Nitration reaction time;
 Physicochemical properties of the inert precursors.

The conditions used for the synthesis of the βNCPCDs (samples NP7–NP11)
and βNCTCDs (samples NT5–NT15) from inert precursors containing PEG and
TEG units, respectively, are summarised in Appendix A (Table A 4 and Table A
5, respectively). The nitration of the insoluble βCTCDs released similar amounts
of energy in every trial, but this was variable and random in the βNCPCDs.

The thermal profiles of samples NP7–NP11 synthesised from the same


precursor (IP26) were found to be inconsistent during decomposition due to the
insolubility of the precursor in the nitrating medium. However, longer nitration
times promoted higher degrees of nitration and higher decomposition enthalpies
compared to analogues NT1–NT10 containing TEG units (Appendix A, Table A
5). The reaction conditions were changed in each trial to increase the efficiency
of the reaction, but the nitration was strongly affected by the gel-like physical
properties of the βCPCD precursors. Increasing the nitration time to 2 h (NP7–
NP12) and the stirring speed (900 rpm), facilitated the dispersion of the
hydrogel. The products demonstrated greater thermal stability up to Tdec (peak =
200 °C) which were attributed to the more efficient purification. However, the
thermal properties of the compounds still appeared to be random, with NP8 and
NP9 exceeding 1000 J g-1 (Figure 5.11), but NP7 and NP10–NP12 falling within
the range 400–600 J g-1 (Table 5.2).

134
Figure 5.11 DSC thermogram of NP9, from 30 to 300 °C (10 °C min-1, aluminium
crucible).

All the βNCTCDs containing TEG segments showed similar thermal stabilities,
with a decomposition peak centred at ~200 °C (Figure 5.12).

Figure 5.12 DSC thermogram of NT7, from 30 to 500 °C (10 °C min-1, aluminium
crucible).

The energy associated with the decomposition of NT5–NT14 was always in the
range 430–810 J g-1 regardless of the nitration conditions (Table 5.2) and the
thermal profiles were quite similar when characterised by DSC. The
decomposition energy was low compared to the values for NC (~2000 J g-1

135
when measured under the same conditions [28]), indicating a low degree of
nitration in the samples. The residue began to char at 370 °C.

Table 5.2 Thermal properties of the energetic derivatives of gel-like insoluble βCPCDs
and insoluble βCTCD powders.

Inert
Product
Precursor
Sample ID Cross-linker: Tdec (°C) ΔHdec Tg
ID βCD feed
ratio1 Onset Peak (J g-1) (°C)

NP7 IP26 9:1 100 108 690 -


NP8 IP26 9:1 129 135, 1392 1400 -
NP9 IP26 9:1 174 203 1132 -8
NP10 IP26 9:1 181 208 460 -8
NP11 IP26 9:1 109 208 460 -28
NT5 IT5 9:1 183 210 530 -24

NT6 IT5 9:1 183 208 650 -19

NT73 IT5 9:1 179 208 760 -19

NT83 IT5 9:1 179 208 810 -18

NT93 IT5 9:1 179 208 760 -22

NT103 IT6 3:1 182 209 430 -22

NT113 IT7 6:1 175 206 570 -22

NT123 IT5 9:1 175 208 590 -21

NT13 IT6 3:1 181 208 570 -22

NT14 IT7 6:1 176 207 510 -21

NT15 IT5 9:1 183 210 530 -24


1 Double spiked decomposition.
2 Compounds are insoluble. The cross-linker:βCD feed ratio is reported.
3 Nitration time = 2 h.

The low-temperature characterisation of the compounds was achieved by


cycling the temperature from –100 to 100 °C three times. The βNCXCDs
derived from insoluble precursors appeared to be hygroscopic, retaining water
in their structures. This affected the Tg of the product which fell to –26 °C

136
(Figure 5.13). By heating the sample, the water evaporated allowing the real Tg
to be determined in the second heating cycle, revealing a value of –22 °C. The
transition at 80 °C attributed to the βCD units in βNCXCDs was also recorded in
several nitrated samples (Figure 5.10).

Figure 5.13 DSC thermogram of NT9 from –100 to 100 °C (10 °C min-1, third
temperature cycle, aluminium crucible).

In conclusion, the nitration of insoluble precursors generated sparsely nitrated


products. It was not possible to control the degree of nitration and thus the
thermal properties of the βNCXCDs so these products were not considered in
the subsequent experiments.

5.4.2 Thermal characterisation of βNCXCDs prepared from soluble


precursors

The βNCXCDs prepared from water-soluble βCXCDs were soluble in organic


solvents such as acetone and THF. The thermal properties were duly
investigated by DSC and TGA, and the nitrogen content was determined by
titration (Table 5.3).

137
Table 5.3 Thermal properties of βNCXCDs prepared from water-soluble βCXCDs.

Sample Inert precursor Nitrated product


Tdec (°C)
ID XEG:βCD ratio Tg (°C) ΔHdec (J.g-1) Nitrogen content % Ω (%) Tg (°C)
Onset Peak
βNCD - - 193 207 1990 11.6 -40 -
NT1 3.6 -1 180 196 2080 8.2 -81 -1
NT2 3.2 -1 180 203 1430 9.4 -71 -1
NT3 2.5 -1 179 200 1690 9.3 -69 -1
NT4 1.6 - 176 198 1850 6.2 -87 -
NH1 4 -10 179 191 1750 8.2 -87 +6
NH2 3.2 -6 174 198 1730 7.1 -93 +26
NH3 2.6 -1 178 199 1720 9.0 -76 +47
NH4 1.8 -1 174 196 1720 9.4 -70 +65
NP1 4.0 -20 176 198 1750 6.2 -109 -14
NP5 3.7 -19 178 190 1800 6.2 -101 +7
NP6 3.7 -19 175 190 1540 6.6 -105 +7
NP2 2.8 -6 172 196 1840 6.8 -97 +31
NP3 2.5 +5 174 197 1630 7.0 -94 +23
NP4 1.6 - 174 193 1730 8.7 -76 +45
1 Change in heat flow due to the softening of the compound.

138
The offsets for βNCXCD decomposition fell within the range 172–193 °C (Table
5.3) whereas the degradation peaks were in the range 190–203 °C and were
comparable to those of the corresponding products obtained from insoluble
βCXCD precursors in water and also to NC (201–205 °C) [28]. The ΔHdec
values were 1360–2080 J g-1, lower than the 2200–1800 J g-1 released by the
less fuel-rich NC [28]. Some general observations were made for the three sets
of βNCXCDs and are reported below.

For samples NT1–NT4 containing TEG units, the decomposition temperatures


were very close to 200 °C regardless of the TEG content whereas the
decomposition enthalpies fell within the range 1430–2080 J g-1. No glass
transition was observed, confirming that the length of TEG units was insufficient
to make the βNCTCD products soft and rubbery.

For samples NH1–NH4 containing HEG units, the decomposition temperatures


were variable (191–199 °C) whereas the decomposition enthalpies were stable
at 1700 J g-1. A glass transition occurred between 6 and 65 °C confirming that
the length of the HEG units was sufficient to influence the thermo-mechanical
properties of the βNCHCD products.

For samples NP1–NP6 containing PEG units, the decomposition temperatures


were again variable (190–197 °C) and the decomposition enthalpies varied in
the range 1360–1840 J g-1. A glass transition occurred between 7 and 45 °C
confirming that the length of PEG units was also sufficient to influence the
thermo-mechanical properties of the βNCPCD products. Exceptionally, NP1
was found to undergo a glass transition at –14°C.

Finally, the Tg of the βNCXCDs appeared to be linearly dependant on the


HEG/PEG content, as shown in Figure 5.16.

139
Figure 5.14 DSC thermogram of NP1 from 25 to 300 °C (10 °C min-1, aluminium
crucible).

140
The Tg of the βNCXCDs was 20–30 °C higher than that of the precursors (Table
5.3). The presence of hydrogen bonds increases the Tg [136]. The replacement
of hydroxyl groups with nitrato groups in the βNCXCDs should reduce the
number of hydrogen bonds but the Tg nevertheless increased. One potential
explanation is that the nitrato groups expanded the polar surface area of the
compounds, therefore increasing their cohesive energy and limiting chain
mobility to an even greater extent than hydrogen bonds [137]. The cohesive
energy is the increase in internal energy when all intermolecular forces are
eliminated [137]. The trend in Tg values could be evaluated by Hildebrand or
Hansen solubility parameter modelling to assess the effect of hydroxyl and
nitrato groups on the cohesive energy of the βNCXCDs [138].

Figure 5.15 DSC thermogram of NP1 from –100 to 100 °C showing the pronounced
glass transition of the highly cross-linked products (10 °C min-1, aluminium crucible).

141
Figure 5.16 Plot of Tg against XEG:βCD ratio for βCHCDs (blue line) and βCPCDs
(orange line).

The nitrogen content of the βNCXCDs was determined by titration as discussed


above and was found to be inversely proportional to the length of the PEG
segment. The βNCPCDs contained less nitrogen (6.5%) than the βNCTCDs
and βNCHCDs (8–9%). The longer polyethylene glycol chains reduced the
nitrogen content because the spacer increases the proportion of C, H and O
relative to N. Furthermore, the βCPCDs are less soluble than the analogous
molecules synthesised with the HEGDGE and TEGDGE cross-linkers, reducing
the efficiency of the nitration reaction.

The nitrogen content of the βNCXCDs was needed to determine the chemical
formula of each compound and thus its contribution to the combustion of
energetic formulations. The difference between inert and energetic binders is
the contribution to the overall energetic output of the formulation. The power of
an explosive is proportional to the amount of molecular oxygen it contains,
which determines the gaseous products produced during detonation. This is
expressed as the oxygen balance (Ω), as shown in Equation 5:

142
𝑏
(𝑑−2𝑎− )𝑥 1600 (5)
2
Ω= %
𝑀𝑤

where d is the number of oxygen atoms in the compound, a is the number of


carbon atoms, b is the number of hydrogen atoms and Mw is the molecular
weight of the compound.

Equation 5 indicates the ability of the compound to produce gaseous products


and release energy as heat [2]. Positive Ω values correspond to the complete
oxidation of the compounds in the formulation and the production of large
quantities of gases such as CO2 and hence more energy, whereas negative Ω
values indicate the compound is deficient in oxygen, does not achieve complete
oxidation, and that there is a higher yield of gases such as CO and solid
residues rich in carbon, and hence less energy.

The Ω values for all βNCXCDs were negative due to the high proportion of
carbon compared to oxygen. Furthermore, the Ω value of βNCD is –40%, which
is identical to NC with a nitrogen content of 11.6%. The XEG units in the
βNCXCDs reduced Ω even further, yielding values of –69% to –109%.

143
5.4.3 Thermo-mechanical characterisation of βNCXCDs
Due to the energetic properties of βNCXCDs, only one sample containing PEG
units (NP1) was characterised by DMA. Preliminary small-scale hazard tests
were completed before the characterisation and sample NP1 was declared safe
for testing under the induced stresses generated by the DMA measurement
conditions. NP1 was not susceptible to ignition during the impact and friction
mallet tests as described in Chapter 6. The Tg of NP1 was similar to that of its
precursor IP18, with an onset at 15 °C and a tanδ peak at 25 °C (Figure 5.17).
The nitrated derivative interacted strongly with the support, just like its
precursors (Section 4.3.2.2) and accordingly, cracking was observed at –51 °C
during the cooling cycle at the onset of E′.

Figure 5.17 DMA showing variation in the storage modulus (E′, solid lines) and tanδ
(dashed lines) of NP1 (10 °C min-1, 1 Hz, third temperature cycle from –100 to 140 °C,
aluminium pocket).

144
5.4.4 Qualitative analysis of the self-healing properties of βNCPCD34
The self-healing ability of NP1 was qualitatively visible, as for the inert βCPCDs
(Chapter 4). NP1 was assessed by first cutting the sample (Figure 5.18a,b) and
then re-joining the parts by leaving it at rest for 4 h (Figure 5.18c). The sample
was then pulled gently by its extremities, showing a high degree of elongation
(Figure 5.18d).

Figure 5.18 Self-healing ability of NP1 (PEG:βCD ratio = 4:1).

These experiments showed that the loss of some hydrogen bonds due to the
replacement of hydroxyl groups with nitrato groups did not inhibit the mobility of
the polymer chains in NP1 and did not eliminate its thermal healability. The
healing of NP1 was driven by supramolecular interactions, such as hydrogen
bonds and Van der Walls forces, as was the case for its precursor. The higher
mobility of the dangling PEG chains accelerated the self-healing of these
compounds.

5.5 Conclusions
The results presented in this chapter provide key data required for the nitration
of βCXCDs, which are produced from βCD and diglycidyl ethers such as
TEGDGE, HEGDGE and PEGDGE, and thus for the preparation of βNCXCDs
as new energetic binders. The nitration reaction with fuming nitric acid led to
two sets of βNCXCDs differing in terms of their solubility in water and other
organic solvents, and their properties were strongly dependent on the
physicochemical characteristics of the βCXCD precursors.

Insoluble βCXCD systems yielded cross-linked derivatives with various degrees


of nitration that were insoluble in acetone, whereas water-soluble βCXCD
precursors produced βNCXCDs that were soluble in acetone and easy to

145
characterise and process. The degree of nitration in these products was not
only dependent on the properties of the precursor, but also the duration of the
nitration reaction and the XEG:βCD ratio of the products. Better control of the
properties of the βNCXCD products was achieved when soluble βCXCD
precursors were used, and the workup was simpler compared to the insoluble
precursors. A purification method based on the NC manufacturing process
removed acid traces from the nitration reaction more efficiently, stabilising the
compound during storage.

All of the synthesised products were characterised by 1H-NMR and FTIR


spectroscopy to confirm their effective nitration. No significant differences in the
degree of nitration were observed for systems containing variable quantities of
the linkers TEG, HEG and PEG. The nitrogen content of the βNCXCDs was
determined by the titration of free hydroxyl groups in the molecules and was
dependent on the degree of cross-linking and the length of the polyethylene
glycol chains in the βNCXCDs. A lower degree of nitration was observed in
derivatives containing PEG segments compared to those containing TEG or
HEG segments due to the lower solubility of the βCPCD precursors in the
nitration media.

The thermal stability of the βNCXCDs (Tdec = 190–200 °C) was independent of
the degree of cross-linking and type of linker segment in the macromolecule,
whereas ΔHdec was influenced by the type of linker segment and ranged from
1300 to 2100 J g-1. The βNCXCDs retained much of the softness of their
βCXCD precursors, with Tg values in the range 6–65 °C. The analysis of one
βNCPCD sample showed that it retained the self-healing characteristics of its
precursor after 4 h at room temperature.

146
6 Compatibility, hazard and stability studies
It is expected that the main constituents of munitions, such as explosives,
oxidisers and pyrotechnic fillers, would be less susceptible to external stimuli
when embedded in a βCD binder, making the formulation safer and easier to
handle. Compatibility, hazard and stability testing help to ensure the safety of an
energetic formulation during manufacturing and storage. Compatibility tests
determine whether materials such as binders, plasticisers and other additives in
energetic formulations have an adverse effect on stability when in contact with
each other. The tests check for chemical compatibility between the constituents
of a formulation and with the tools used during manufacturing and storage. The
effect of ageing on the stability of the materials is determined by accelerated
ageing at high temperatures. Hazard tests offer a comprehensive understanding
of the sensitiveness of formulations and their components against thermal,
mechanical and electrostatic stimuli, and help with the development of
guidelines for safe manufacturing, storage, transport and use.

This chapter discusses the compatibility, hazard and stability testing of the inert
βCXCDs and nitrated βNCPXCDs developed in Chapters 4 and 5, according to
NATO STANAG 4147 Test 4 S [121] and EMTAP guidelines [119]. Any
changes in the shape, onset, or peak position of any thermal event may be
indicative of incompatibility.

6.1 Compatibility tests – βCXCDs


A set of preliminary compatibility tests was carried out by mixing sample IP19
(an inert βCPCD with a PEG:βCD ratio of 4:1) with a set of inert, oxidising and
energetic materials following the procedure described in Chapter 3. The
decomposition temperatures of the pure compounds and 50:50 w/w mixtures
were determined by DSC, and the results are summarised in Table 6.1. All DSC
tests for compatibility were carried out with a heating rate of 2 °C min-1 rather
than the 10 °C min-1 used to record the thermal properties in Chapters 4 and 5.
The differences in thermal behaviour between the pure compounds and
mixtures were taken as indications of compatibility/incompatibility as per

147
STANAG 4147 (Ed.2) Test 4 (see Section 3.3.6). Specifically, shifts of less than
4 °C generally indicate compatibility, shifts of more than 20 °C generally indicate
incompatibility, and values between are inconclusive and warrant further testing.

Table 6.1 Compatibility of inert βCPCD (sample IP19, PEG:βCD ratio = 4.0:1) with
various energetics.

Tdec
(°C) ΔT
Formulation Shape change
(°C)
Energetic Mixture
IP19/HMX 279 279 0 Significant
IP19/RDX 226 212 –14 Significant
IP19/PETN 185 184 –1 Minor
IP19/ADN 172 149 –23 Significant
IP19/KClO3 -1 -1 0 None
IP19/KNO3 -1 -1 0 None
IP19/NH4ClO4 264 260 –4 Minor
IP19/NH4NO3 -1 188 0 Significant
IP19/RedP 400/492 404/471 +4/–11 Significant
1 No decomposition

Pure HMX decomposed at 279 °C (Figure 6.1) and there was no shift in the
position of the decomposition peak when mixed with IP19. However, the shape
and onset of the decomposition curve changed slightly in the mixture, with a
broader degradation ramp starting at 10 °C, which is earlier than pure HMX.
Further evaluation of the compatibility is therefore required according to NATO
STANAG recommendations.

148
Figure 6.1 DSC thermogram of HMX (blue line), IP19 (red line), and a 50/50 (w/w)
mixture (black line) from 30 to 300 °C (1 mg, 2 °C min-1, aluminium crucible).

Pure RDX melted at 204 °C before decomposing at 226 °C (Figure 6.2, blue
curve) whereas the mixture with IP19 decomposed directly at 212 °C with a
lower onset of decomposition at 30 °C (Figure 6.1, black curve). The 14 °C shift
in the decomposition temperature between pure RDX and the RDX/IP19
mixture suggests a degree of incompatibility and further investigation is needed.
There was a significant difference in compatibility between the IP19
formulations with RDX and HMX, which may reflect the relative size of the two
nitramines. RDX is the smallest (4–5 Å) and can be partially encapsulated by
the βCD cavity (6–6.5 Å), which translates to the small hump in the degradation
curve. In contrast, HMX remains free due to its larger size (5–8 Å). The βCD-
based binder was therefore anticipated to encapsulate energetic fillers but no
direct chemical reaction was expected.

149
Figure 6.2 DSC thermogram of RDX (blue line), IP19 (red line), and a 50/50 (w/w)
mixture (black line) from 30 to 300 °C (1 mg, 2 °C min-1, aluminium crucible).

Pentaerythritol tetranitrate (PETN) melted at 142 °C and decomposed at 185 °C


(Figure 6.3, blue curve) and these properties were not affected by mixing with
IP19 (Figure 6.3, black curve). The energy of the mixture increased, suggesting
that the binder had a synergistic effect on the energy output of the system
during decomposition.

Figure 6.3 DSC thermogram of PETN (blue line), IP19 (red line), and a 50/50 (w/w)
mixture (black line) from 30 to 300 °C (1 mg, 2 °C min-1, aluminium crucible).

150
Neat ammonium dinitramide oxidiser (ADN) melted at 93 °C (Figure 6.4, blue
curve), and this did not change in the mixture with IP19 (Figure 6.4, black
curve). ADN decomposed at 172 °C, but the mixture showed a biphasic
decomposition curve with major peaks at 149 and 172 °C. This indicated a
degree of incompatibility between the two compounds, perhaps reflecting the
encapsulation of some of the ADN molecules within the βCD cavity, as
proposed for RDX.

Figure 6.4 DSC thermogram of ADN (blue line), IP19 (red line), and a 50/50 (w/w)
mixture (black line) from 30 to 300 °C (1 mg, 2 °C min-1, aluminium crucible).

KClO3 showed no exothermic peaks below 400 °C with a small endotherm


visible at 303 °C (Figure 6.5, blue curve), possibly due to the compound
melting. The mixture with IP19 exhibited the same endotherm and a tiny
additional endotherm at 214 °C, suggesting good compatibility between these
components (Figure 6.5, black curve).

151
Figure 6.5 DSC thermogram of KClO3 (blue line), IP19 (red line), and a 50/50 (w/w)
mixture (black line) from 30 to 400 °C (1 mg, 2 °C min-1, aluminium crucible).

KNO3 melted at 335 °C (Figure 6.6, blue curve) and no changes were observed
in the mixture with IP19 (Figure 6.6, black curve). These components were
therefore considered to be compatible.

Figure 6.6 DSC thermogram of KNO3 (blue line), IP19 (red line), and a 50/50 (w/w)
mixture (black line) from 30 to 300 °C (1 mg, 2 °C min-1, aluminium crucible).

152
The thermal profile of pure NH4ClO4 showed a melting point at 242 °C and
decomposition at 264 °C (Figure 6.7, blue curve). These values were
unchanged in the mixture with IP19 (Figure 6.7, black curve). Decomposition
involved a principal event followed by two minor exothermic peaks. These
results indicated that the two components were compatible.

Figure 6.7 DSC thermogram of NH4ClO4 (blue line), IP19 (red line), and a 50/50 (w/w)
mixture (black line) from 30 to 500 °C (1 mg, 2 °C min-1, aluminium crucible).

Pure NH4NO3 showed a broad endotherm above ~200 °C caused by the volatile
compound evaporating after melting (Figure 6.8, blue curve). When mixed with
IP19, one endotherm became an exotherm with a peak at 188 °C (Figure 6.8,
black curve), possibly indicating chemical reactivity between molten ammonium
nitrate and the binder. This indicates the two ingredients are likely to be
incompatible.

153
Figure 6.8 DSC thermogram of NH4NO3 (blue line), IP19 (red line), and a 50/50 (w/w)
mixture (black line) from 30 to 300 °C (1 mg, 2 °C min-1, aluminium crucible).

Two decompositions were observed for pure red phosphorous (RedP) at


400 and 492 °C (Figure 6.9, blue curve) but the mixture with IP19 showed a
broader decomposition profile with maxima at 409 and 471 °C (Figure 6.9, black
curve). Much more energy is released when RedP is mixed with the binder. The
two ingredients could therefore be described as compatible, with reservations.

Figure 6.9 DSC thermogram of red phosphorus (blue line), IP19 (red line), and a 50/50
(w/w) mixture (black line) from 30 to 550 °C (1 mg, 2 °C min-1, aluminium crucible).

154
6.2 Compatibility tests – βNCXCDs
In the next set of compatibility tests, sample NP13 (a nitrated βNCPCD with a
PEG:βCD ratio of 3.8:1) was mixed with a set of inert, oxidising and energetic
materials and analysed by DSC to determine their compatibility in accordance
with STANAG 4147 guidelines as described above for the inert precursor IP10.
Like NC, the βNCXCDs are nitroesters and need to be mixed with stabilisers to
ensure an adequate shelf life [139,140]. Therefore, the widely-used stabilisers
diphenylamine (DPA) and N,N′-diethyl-N,N′-diphenylurea (Centralite) were also
tested in the compatibility tests, as well as 2,4-nitrodiphenylamine (2,4-NDPA),
which is an intermediate product of DPA. Due to the small-scale synthesis (200
mg) of the βNCXCD compounds, the compatibility tests with the stabilisers were
conducted using a different sample (NP1), although this has the same chemical
composition as NP13. The decomposition temperatures of the pure compounds
and mixtures were determined by DSC, and the results are summarised in
Table 6.2.

Table 6.2 Compatibility of nitrated βNCPCD (samples NP13 and NP1, PEG:βCD ratio
= 3.8:1) with various energetics and stabilisers.

Formulation Tdec2
ΔT
(°C) Shape change
(°C)
Energetic Mixture
NP13/NP11 186 - - -
HMX 279 278 –1 Significant
RDX 226 210 –13 Significant
PETN 185 187 +2 Minor
ADN 172 161 –11 significant
KClO3 -3 -3 0 none
KNO3 -3 1853 –13 minor
NH4ClO3 264 265 +1 minor
NH4NO3 1893 +34 Significant
RedP 400/492 404/481 +4/–11 minor

155
NP1 184 - - -
DPA 1404 –44
4
2,4-NDPA 190 +6
4
Centralite 185 +1
1 PEG:βCD ratio = 3.8:1
-1
2 At 2 °C min , N2 atmosphere.
3 No decomposition.
4 Decomposition is related to the solely energetic binder.

The decomposition peaks of pure HMX at 279 °C (Figure 6.10, blue line) and
the energetic binder NP13 at 186 °C (Figure 6.10, red line) were not affected
when the two components were mixed (Figure 6.10, black line). As discussed
for the inert precursor IP19, the onset of degradation remained the same, but a
portion of the HMX started to degrade earlier, causing an initial step at 272 °C.
This revealed a degree of incompatibility between NP13 and HMX that warrants
further investigation.

Figure 6.10 DSC thermograms of HMX (blue line), NP13 (red line), and a 50/50 (w/w)
mixture (black line) from 30 to 300 °C (1 mg, 2 °C min-1, aluminium crucible).

Pure RDX melted at 204 °C before decomposition with a major peak at 226 °C
(Figure 6.11, blue line). When mixed, RDX and NP13 decomposed in the same
temperature range between 140 and 230 °C. The RDX decomposition moved
from 226 to 210 °C, with the shift of –16°C suggesting a degree of
incompatibility which needs investigation. In contrast, the NP13 was 5 °C more

156
stable when mixed with RDX, with its Tdec shifted from 186 to 191 °C. As for the
inert binder, there was a noticeable difference between NP13 formulations
containing RDX and HMX, which likewise can be attributed to the different sizes
of the nitramines and resulting differences in their degree of encapsulation by
the βCD units of the energetic binder.

Figure 6.11 DSC thermograms of RDX (blue line), NP13 (red line), and a 50/50 (w/w)
mixture (black line) from 30 to 300 °C (1 mg, 2 °C min-1, aluminium crucible).

PETN and NP13 are both nitroesters containing nitrato groups and this is
reflected in their very similar decomposition profiles (Figure 6.12, blue and red
curves respectively). The melting of PETN at 141 °C was unchanged in the
mixture. Furthermore, the mixture produced more energy than either of the
components alone (Figure 6.12, black curve). The two components were
therefore declared compatible.

157
Figure 6.12 DSC thermograms of PETN (blue line), NP13 (red line), and a 50/50 (w/w)
mixture (black line) from 30 to 300 °C (1 mg, 2 °C min-1, aluminium crucible).

ADN melted at 93°C, a feature retained in the mixture with NP13 (Figure 6.13,
blue and black curves). The ADN profile featured an exothermic decomposition
with an onset at 137 °C and a peak at 172 °C, but in the mixture the onset
shifted to 142 °C with a major peak at 161 °C, and a secondary peak at 157 °C
matching the decomposition temperature of NP13 (Figure 6.13, red and black
curves). The shape of the mixture’s decomposition diverged from the profile of
the pure compounds and became more indented, with a shift of +9 °C
compared to ADN. This indicated a degree of incompatibility between the two
components, and also suggested there was a ~20% increase in the amount of
energy released by the mixture.

158
Figure 6.13 DSC thermograms of ADN (blue line), NP13 (red line), and a 50/50 (w/w)
mixture (black line) from 30 to 300 °C (1 mg, 2 °C min-1, aluminium crucible).

KClO3 consistently melted at 303 °C (Figure 6.13, blue curve). The mixture
displayed the same endotherm at 305 °C, as well as a shoulder in the very
sharp exotherm (Figure 6.13, black curve). The data suggested these two
compounds are highly compatible, as seen for the analogous experiments with
IP19.

Figure 6.14 DSC thermograms of KClO3 (blue line), NP13 (red line), and a 50/50 (w/w)
mixture (black line) from 30 to 400 °C (1 mg, 2 °C min-1, aluminium crucible).

KNO3 showed no evidence of decomposition below 550 °C but melted at 335°C


(Figure 6.15, blue curve). When the mixture was tested, there was no change in

159
the thermal profile (Figure 6.15, black curve). This indicated that the two
components were compatible.

Figure 6.15 DSC thermograms of KNO3 (blue line), NP13 (red line), and a 50/50 (w/w)
mixture (black line) from 30 to 400 °C (1 mg, 2 °C min-1, aluminium crucible).

NH4ClO4 melted at 242 °C and this property did not change in the mixture with
NP13 (Figure 6.16, blue and black curves). The subsequent decomposition
consisted of a main decomposition event at 264 °C followed by small
exothermic peaks. The thermograms of the individual components and the
mixture were comparable, suggesting the components were compatible.

Figure 6.16 DSC thermograms of NH4ClO4 (blue line), NP13 (red line), and a 50/50
(w/w) mixture (black line) from 30 to 500 °C (1 mg, 2 °C min-1, aluminium crucible).

160
NH4NO3 displayed an endothermic peak at 67 °C (Figure 6.15, blue curve) but
this disappeared in the mixture (Figure 6.17, black curve). Furthermore, the
endotherm at 265 °C (caused by the evaporation of NH4NO3) led to a strong
exothermic event at 208 °C. This may have been caused by a reaction between
the NH4NO3 vapour and the binder NP13, indicating that the two components
are likely to be incompatible.

Figure 6.17 DSC thermograms of NH4NO3 (blue line), NP13 (red line), and a 50/50
(w/w) mixture (black line) from 30 to 300 °C (1 mg, 2 °C min-1, aluminium crucible).

The decomposition of the mixture of RedP and NP13 differed very slightly from
the individual components due to the drop in the baseline (Figure 6.18). The
first RedP decomposition event shifted from 400 to 404 °C, whereas the second
shifted from 492 to 481 °C. Despite the changes in the mixture, it is unlikely that
any in-service munition containing these two components would reach such
temperatures. Therefore, the components can be considered as compatible.

161
Figure 6.18 DSC thermograms of RedP (blue line), NP13 (red line), and a 50/50 (w/w)
mixture (black line) from 30 to 550 °C (1 mg, 2 °C min-1, aluminium crucible).

The pure DPA stabiliser melted at 54 °C (Figure 6.19, blue curve) and this did
not change in the mixture (Figure 6.19, black curve). Evaporation of the DPA
from the pin hole in the crucible commenced as soon as the sample was heated
above 100 °C due to the flow of nitrogen in the experiment. The development of
vapours caused the early decomposition of NP1, from 157 to 130 °C. The DPA
was not gassed away from the nitrogen purge when the crucible was sealed.
The mixture was tested with sealed and unsealed vessels and the thermograms
were identical, confirming that the evaporation of DPA causes NP1 to
decompose at a lower temperature. This means that the DPA vapour does not
stabilise NP1 but actually accelerates its decomposition.

162
Figure 6.19 DSC thermograms of DPA (blue line), NP1 (red line), and a 50/50 (w/w)
mixture (black line) from 30 to 300 °C (1 mg, 2 °C min-1, aluminium crucible).

The 2,4-NDPA melted at 157 °C (Figure 6.20, blue curve) and the same melting
peak was also present in the mixture (Figure 6.20, black curve). Unlike DPA,
the 2,4-NDPA was found to stabilise NP1, shifting the degradation of the binder
from 184 to 190 °C.

Figure 6.20 DSC thermograms of 2,4-NDPA (blue line), NP1 (red line), and a 50/50
(w/w) mixture (black line) from 30 to 300 °C (1 mg, 2 °C min-1, aluminium crucible).

Centralite melted at 73 °C (Figure 6.21, blue curve) and this did not change
when it was mixed with NP1 (Figure 6.21, black curve). Centralite also
evaporated in the temperature range 130–154 °C, as observed above for DPA.

163
A second test with sealed crucibles showed no change in the thermogram of the
mixture. Therefore, the degradation of the binder was not affected by the
presence of Centralite, making it a suitable stabiliser.

Figure 6.21 DSC thermograms of Centralite (blue line), NP1 (red line), and a 50/50
(w/w) mixture (black line) from 30 to 550 °C (1 mg, 2 °C min-1, aluminium crucible).

6.3 Hazard tests – βNCXCDs


The new energetic βNCXCDs were characterised using internal hazard tests at
Cranfield Defence and Security School to determine the conditions under which
the energetic binders are likely to ignite. Two sets of small-scale (20 mg) hazard
tests were performed using two different types of βNCXCDs.

The first set of tests was carried out on insoluble nitrated products synthesised
from insoluble βCDTCD and βCDPCD precursors, containing TEG and PEG
units respectively. These experiments are described in the appendix: samples
NT1–NT5 (Table A 5) and samples NP13–NP15 (Table A 4).

The second set of tests was carried out on acetone-soluble βNCPCD samples
NP1, NP5, NP6 and NP16–NP17 (Table 6.3). These were synthesised from
βCPCDs with PEG:βCD ratios > 3.5:1, and their softness makes them
promising in terms of processability. Sample NP16, with a low PEG:βCD ratio of
2.6:1 was characterised as well as βNCD to determine any increase in hazards

164
caused by the crystalline structure (Table A6). The following tests were
performed:

 Impact and friction sensitivity test;


 Colour change tests to assess thermal stability;
 Temperature ignition test to assess combustion properties.

Materials which clearly failed these tests were considered too hazardous for
manual handling and were not considered further. These tests were not
standardised methods to characterise the energetic properties of a material, but
were carried out to provide a degree of confidence in handling before full
Energetic Materials Testing and Assessment Policy (EMTAP) tests on a larger
scale. The test results are summarised in Table 6.3.

165
Table 6.3 Preliminary small-scale compatibility tests performed on βNCTCDs and βNCTCDs containing TEG and PEG units, respectively.

Polyethylene Colour
Colour
glycol:βCD change Impact
Sample change Friction Ignition4 Evaluated5
feed ratio1 100 °C 30 (GO)
RT 24 h
min
βNCD - No change Browning 7/10 No response HVB Hazardous
NP13 6:1 No change Browning 2/10 No response HVB Hazardous
NP15 9:1 No change Browning 0/10 (smell) No response VB4 Not hazardous
NP16 3:1 No change Browning 3/10 No response HVB2 Hazardous
NP5 5:1 No change Browning2 0/10 No response VB Not hazardous
NP6 5:1 No change Browning2 0/10 No response VB Not hazardous
NP17 5:1 No change Browning2 0/10 No response VB Not hazardous
2
NP1 5:1 No change Browning 0/10 No response VB Not hazardous
NT1 9:1 No change Browning 0/10 No response VB Not hazardous
NT2 9:1 No change Browning 0/10 No response VB Not hazardous
NT3 9:1 No change Browning 0/10 No response VB Not hazardous
NT4 9:1 No change Browning 0/10 No response VB Not hazardous
NT5 9:1 No change Browning 0/10 No response VB Not hazardous
1. Feed ratio of cross-linker TEGDGE or PEGDGE, as appropriate to βCD.
2. The compound softened when heated.
3. HVB - Highly vigorous burning, similar to the ignition of a match.
4. VB - Vigorous burning, like the flame of a candle.

166
5. Assessed by CDS experts.

167
Colour change test
A colour change generally indicates that a decomposition reaction occurs
(Table 6.3). The effect of temperature on the colour of each compound was
assessed using two methods. In the first method, 20 mg of each sample was
heated to 100 °C and maintained at this temperature for 30 min. In the second
method, the samples were left at room temperature for 24 h. Samples NP1,
NP5, NP6 and NP13–17 displayed a change in viscosity when heated to 100
°C, as anticipated from their thermal profiles. In contrast, the viscosity of the
insoluble gels NT5–NT8 did not change as the temperature increased. All the
samples became slightly brown during the test reflecting the decomposition of
the nitroesters to form NOx, which was adsorbed by the samples. The inert
precursor IP19 (PEG:βCD ratio = 4.0) was also heated to exclude the possibility
that browning reflected the caramelisation of the βCD or the ageing of the PEG
segments in the cross-linked system. IP19 showed no signs of browning after
30 min, but browning was observed when this inert sample was kept at 100 °C
for 24 h.

Impact and friction test

The impact and friction sensitivity of insoluble gels NT5–NT10 and NP15, and
acetone-soluble samples NP1, NP5, NP6 and NP13–NP17 was assessed using
the steel-on-steel mallet test (Table 6.3). The detection of smell, flash or sound
was left to the judgment of an operator trained for the test. This determined
whether the compound was recorded as go (decomposition detected) or no-go
(no decomposition detected). The βNCD control was quite susceptible to impact
(7/10 go) whereas almost all of the synthesised compounds showed no
evidence of decomposition, the exceptions being samples NP13, NP15 and
NP16. This was attributed to the purification method, which left traces of acid,
and the low degree of cross-linking of these products. However, the three
sensitive samples showed evidence of a localised initiation that did not
consume the entire sample mass. The soft samples (NP1, NP5, NP6 and
NP17) were not sensitive to impact but underwent deformation due to the

168
impact itself of the heat thus generated, confirming the softness of these
compounds.

Ignition test

The samples were burnt over an open flame and the energy of the flame was
qualitatively described to evaluate the susceptibility of the compound to burning.
Fast burning refers to a type of combustion that is often compared to the ignition
of a safety match, and vigorous burning is like the flame produced by wood
burning. As expected from the earlier hazard tests, samples NP13, NP16 and
βNCD burnt more vigorously than the others, with a sparking flame. The
intensity of the ignition was directly dependent on the Ω value of the compound,
with the less cross-linked compounds (less rich in carbon chains) burning more
vigorously.

6.3.1 EMTAP tests


The EMTAP tests described below followed the certificated methods presented
in the EMTAP Manual of Tests, which sets the British Standards relevant to the
qualification of energetic materials and associated technical assessments [119].
Each test is based on the repetition of several forms of analysis to increase
confidence in the result. The Rotter impact test (Test No. 1A) and electrostatic
discharge (ESD) test (Test No. 4) were performed on sample NP17 (Table 6.4).
Due to the large quantity of sample required and the safety restrictions applied
at Cranfield University, the NP17 sample was obtained by dissolving and
blending in acetone the products of four nitration reactions at the 2.5-g scale
(Table A 6) using a precursor with a PEG:βCD ratio of 3.8:1. The resulting blend
passed the small-scale tests described above and was considered safe for
EMTAP testing. Given the properties of the sample, the material was soft and
malleable, making it a promising candidate for future processing. Further work
is needed to collect more data for the scaled-up manufacturing of this
compound and to increase confidence in the assessment of its sensitivity.

169
Table 6.4 EMTAP test data for sample NP17 based on Rotter impact and ESD values.

Energetic
Impact by drop-weight input (cm)1 ESD2 4.5 J
sample

NP17 29.3 0/50


1 Impact value is the mean of 50 tests.
2 ESD value is the mean of 50 tests at 4.5 J.

The Rotter impact test indicated a figure of insensitiveness (FoI) of 29 for NP17
with a release of 1 mL gas per test. The initiated samples were not completely
consumed, indicating a localised initiation as seen in the small-scale tests
described above.

The FoI value indicated that the compound was highly sensitive to impact, given
that sensitive compounds have values in the range 100 ≥ FoI ≥ 30, and initiators
have FoI values < 30. Dry βNCD has an FoI value of 60 [114] but this value is
not comparable to the FoI of NP17 described here because it was based on the
results of an impact test conducted under American standards that differ from
the British system. Furthermore, the FoI value reported for a similar
epichlorohydrin:βCD cross-linked nitrated product was 47 [109]. The values
determined in the Rotter impact test therefore provide an indication of the
potential sensitivity of NP17 but should not be considered conclusive. The test
is affected by several factors, including the homogeneity of the blend and the
presence of impurities. Given that this was the first attempt to synthesise a
significant quantity of the nitrated product, more tests are required to achieve a
confident result.

The ESD test (spark sensitivity test) determines the sensitivity of a sample to
initiation by electrostatic discharges. NP17 was not susceptible to ESD up to an
energy input of 4.5 J, which is a significant improvement compared to βNCD,
which initiates at 0.0125 J [114], and the epichlorohydrin:βCD cross-linked
nitrated product discussed above, which initiates at 0.1288 J [109].

170
6.4 Stability of βNCXCDs determined by heat flow calorimetry
The single-temperature stability test is a standardised technique in STANAG
4582 that uses heat flow calorimetry (HFC) to determine the shelf life of
energetic formulations [139]. The test was repurposed in this project as a
method to obtain indicative data about the lifetime thermal stability of
βNCXCDs.

Four lots (~1 g) of sample NP1 were prepared, with a PEG:βCD ratio of 4:1.
One lot was kept pure, and the others were mixed with one of three stabilisers
(DPA, 2,4-NDPA or Centralite) as a 1% w/w formulation. A sample of pure (non-
stabilised) NC with a nitrogen content of 12.6% was used as benchmark.

The heat flow from of the samples was measured at a constant temperature of
80 °C. At this temperature, the STANAG criterion is that the sample should
have a heat flow less than a tabulated value of 114 µW g-1 (Figure 6.22, dashed
line) for a time of 10.6 days, in order to be defined as stable. The heat flow is
recorded after the release of the first 5 J of energy. Therefore, the thermograms
do not present the initial stabilisation of the temperature when the samples are
introduced in the machine. The results are summarised in Figure 6.22.

171
Figure 6.22 Thermal stability of pure NP1 and the same compound in the presence of
three different stabilisers (1% w/w).

The experiment was automatically halted after 70 h by the safety control of the
HFC instrument due to the high heat flow values that were registered. All
samples containing the energetic binder NP1 were unstable throughout the
recording (Figure 6.22). The heat flow from NC was lower than the limit value of
114 µW g-1 (light blue curve). The viscous consistency of the sample is thought
to influence its stability because gaseous and liquid materials decompose more
quickly than crystalline solids [141,142]. The air and moisture trapped in the
spongy matrix of the samples, as well as that in the head space of the vials,
probably accelerated the decomposition, as reported for other nitroesters [140].

The highest heat flow was recorded for the pure NP1 sample. The presence of
a stabiliser reduced the heat flow during decomposition, indicating that NP1 can
be stabilised. Interestingly, DPA reduced the heat flow particularly during the
first 4 h, but this was followed by an increase at ~4.5 h probably reflecting the
consumption of the stabiliser. Despite the consumption of the DPA, a residual
stabilising effect was observed, probably conferred by DPA nitro-derivatives,

172
which include 2,4-NDPA in the latter stages of decomposition [143]. This
explains why DPA has a greater stabilising effect than 2,4 NDPA alone.

The addition of Centralite did not influence the stability of the binder, as shown
by the similar heat flow curves over time with and without this additive. It would
be interesting to determine the quantity of stabiliser remaining throughout the
experiment, but the specialised chromatography technique required for this type
of experiment was not available [139]. However, the samples were analysed
before and after testing by 1H-NMR spectroscopy, which showed that the
stabilisers decomposed during the experiment (Figure A 10, Figure A 11 and
Figure A 12). The aged samples were also thermally characterised by DSC.
The Tdec was not significantly changed following exposure to a constant
temperature of 80 °C for 3 days under HFC testing conditions (Table 6.6). The
presence of stabilisers slightly improved the stability of NP1, increasing the Tdec
marginally compared to the pure sample, but the variations were small and
within the range of experimental error.

Table 6.5 DSC analysis of the decomposition temperatures of sample NP1 before and
after HFC tests.

Sample Tdec before HFC tests Tdec ( after 3 days at 80 °C)


(°C) (°C)
NP1 195 195
NP1/DPA 198 197
NP1/2,4-NDPA 198 196
NP1/Centralite 198 195

6.5 Conclusion
EnergeticβNCXCDs were characterised by compatibility, hazard and stability
tests to determine their suitability for energetic applications. Most of the studies
were performed on the βNCPCD derivatives containing PEG segments because
these had superior mechanical properties making them potentially suitable for

173
industrial exploitation. The compatibility studies determined whether the binder
was likely to initiate unwanted reactions among the other constituents.

The βCPCDs and βNCPCDs behaved similarly in the tests. They were
compatible with several pyrotechnic ingredients such as KClO3, KNO3 and
NH4ClO4, as well as explosives such as HMX and PETN. A degree of
incompatibility of both inert and nitrated compounds was observed with RDX,
possibly due to the size of the nitramines, which were small enough to be
encapsulated within the βCD macrocycles. Centralite was also compatible with
the energetic βNCPCDs, but DPA reduced the Tdec of the binder by 44 °C
(specifically when the DPA evaporated at 143 °C) suggesting that DPA vapours
are incompatible with the binder.

Small-scale hazard tests were performed on a set of the new energetic binders
differing in terms of their cross-linker ratio. The most malleable βNCPCD
compounds (PEG:βCD ratio > 3.0) were insensitive to impact and friction with a
mallet. In contrast, samples with PEG:βCD ratios < 3.0 were susceptible to
impact due to the low cross-linker content and thus greater crystallinity. All the
compounds underwent a colour change when left at 100 °C for 30 min. At this
temperature, the nitroesters undergo rapid thermal cleavage to form NOx
molecules that are adsorbed into the matrix of the compound.

A highly cross-linked βNCPCD product (PEG:βCD ratio = 3.8:1) was subjected


to EMTAP tests for impact and ESD resistance. The recorded FoI value of 29
indicated that the nitrated products can be sensitive to impact, but the same
sample was insensitive to ESD stimuli up to 4.5 J, which is a considerable
improvement.

HFC-based stability tests were performed on the soft nitrated compound NP1,
either in pristine form or mixed with 1% (w/w) of one of three different stabilisers
(DPA, 2,4-NDPA or Centralite). The results indicated that the softer compounds
were unstable when incubated at 80 °C for 3 days even in the presence of
stabiliser, with heat flow exceeding 200 µW g-1. The inherent viscosity of these
materials is thought to accelerate the degradation of the nitroester groups. The

174
Tdec of the samples did not change significantly after the HFC tests, although
DPA showed a stronger stabilising effect on the βNCXCDs.

175
7 Conclusion and recommendations
Conclusions

This PhD project has contributed to the development of new inert and energetic
binders for energetic applications. The function of the binder is to hold together
the constituents of an energetic formulation, improving its response to external
stimuli and, when possible, contributing to the energy of the system. The project
focussed on the use of βCD, a cyclic polysaccharide that has been investigated
as a drug delivery vehicle due to its ability to envelop guest molecules. The
design and synthesis of a set of inert and nitrated cross-linked βCD systems
was achieved by the cross-linking of βCD with diglycidyl ethers containing a
variety of polyethylene glycol units followed by a nitration reaction.

Three polyethylene glycol diglycidyl ethers (TEGDGE, HEGDGE and PEGDGE,


containing three, six and an average of nine ethylene glycol units, respectively)
were used to generate βCXCD products containing TEG, HEG or PEG
segments. The abbreviation XEG is used to describe these segments
collectively.

The cross-linkers were used to tune the properties of βCD and generate
malleable cross-linked βCXCD compounds suitable for processing. The
investigation focused on the optimisation of the cross-linking reaction, which
was strongly affected by several factors including NaOH concentration and
volume, reaction time, temperature, βCD alkoxide formation, cross-linker:βCD
feed ratio, and the addition rate of the cross-linker in the βCD alkoxide solution.

The optimal experimental conditions for the cross-linking of βCD were


investigated extensively using PEGDGE as the cross-linker due to its low cost
and commercial availability. Cross-linked βCPCD products that were insoluble
in water and other common solvents were produced using PEGDGE:βCD feed
ratios higher than 5:1. The optimal conditions were a PEGDGE:βCD ratio of up
to 5:1, 5.6% NaOH aqueous solution, 16 h to form the βCD alkoxide, 20 min for
the addition of the cross-linker, and 6 h reaction time at 30 °C. Under these
conditions, several βCPCDs containing different quantities of PEG units were

177
synthesised. The same procedure was then adopted for the synthesis of the
other βCXCD systems. The products were characterised by 1H-NMR and FTIR
spectroscopy, confirming the polymeric nature of the βCXCDs. In all trials, up to
25% of the cross-linker was consumed in a parallel degradation reaction
resulting in a lower XEG:βCD ratio in the products compared to the feed ratio.
The XEG:βCD ratio in the βCXCDs was determined by 1H-NMR in D2O. The
length of the XEG segments and the degree of cross-linking influenced the
physical appearance of the βCXCDs. Samples with PEG:βCD ratios between
4:1 and 3:1 were malleable whereas those with PEG:βCD ratios below 3:1 were
friable. Samples containing TEG units were friable and easy to dissolve
regardless of the TEG:βCD ratio. Also, βCXCDs containing HEG and PEG units
took longer to dissolve if the XEG:βCD ratio was higher than 3.4:1, as
determined by 1H-NMR spectroscopy.

As expected from the design of the materials, the thermal and thermo-
mechanical properties of βCD precursors were affected by the cross-linking
reaction, yielding βCXCDs with low Tg values. The softness of the βCXCD
samples was directly dependent on the length of the XEG unit and the
XEG:βCD ratio. Therefore, βCXCDs containing HEG and PEG segments with a
XEG:βCD ratio exceeding 3.7:1 were characterised by Tg values as low as –20
°C. No glass transition was observed for the βCXCDs with few cross-links or for
those containing TEG segments. 

Cracking was observed in βCXCD samples cooled below their Tg during DMA
experiments due to strong interactions with the metallic supports via hydrogen
bonds. This was also confirmed by optical microscopy using glass supports.
The βCXCDs demonstrated accelerated healing when heated above room
temperature, much like the healing properties of hydrogels containing PEGand
βCD segments, which are used in pharmaceutical applications.

The research project then focused on the synthesis and characterisation of


nitrated derivatives of βCXCDs, aiming to improve the binder’s contribution to
the energy of the formulation. The nitration reaction with fuming nitric acid
yielded two sets of βNCXCDs differing in terms of their solubility in water and

178
other organic solvents, and properties correlated to the physicochemical
characteristics of the βCXCD precursors. Insoluble βCXCD systems yielded
cross-linked derivatives with various degrees of nitration that were insoluble in
acetone, whereas water-soluble βCXCD precursors produced βNCXCDs that
were soluble in acetone and easy to characterise and process. A purification
method adopted from the NC manufacturing process removed acid traces from
the nitration reaction more efficiently, stabilising the compound during storage.

The βNCXCDs were characterised by 1H-NMR and FTIR spectroscopy to


determine the degree of nitration, which was found to be dependent on the
nitration reaction time, the length of the cross-linker and the cross-linkerβCD
feed ratio. This was confirmed by measuring the nitrogen content of the
products. The more soluble products generally featured a lower degree of
nitration.

The thermal stability of βNCXCDs (Tdec = 190–200 °C) was independent of the
length of the XEG unit and the XEG:βCD ratio whereas the ΔHdec of each
system (1300–2100 J g-1) was dependent on the length of the XEG segment.

The βNCXCDs retained a certain degree of softness from their βCXCD


precursors with Tg values in the range –14 to +65 °C. The higher Tg values
probably reflected the higher polarity of the system, which increases the
cohesive energy of the compounds and inhibits the mobility of the XEG chains.
One βNCXCD product containing PEG units retained the self-healing ability of
its precursor after 4 h at room temperature.

The suitability of βNCXCDs as binders in energetic formulations was


determined by conducting compatibility, hazard and stability tests using NATO
STANAG and EMTAP protocols. Most of these studies were performed on the
βNCPCD derivatives containing PEG segments. The compatibility studies
determined the likelihood that the binder would initiate undesirable reactions in
the other constituents of the energetic formulations.

Both βCPCD and βNCPCD behaved similarly when tested and were found to
be compatible with several pyrotechnic ingredients (including KClO3, KNO3 and

179
NH4ClO4) as well as explosives such as HMX and PETN. The degree of
incompatibility determined for RDX, needs further investigation. The different
behaviour of mixtures containing HMX and RDX was attributed to the difference
in the size of these energetic nitramines, promoting different interactions with
the cavities of βCPCDs and βNCPCDs. Only Centralite was compatible with the
energetic βNCPCD binders among three common stabilisers used with
nitroesters

Small-scale laboratory hazard tests performed on a set of βNCPCD binders with


different degrees of cross-linking showed that the most malleable compounds
were insensitive to impact and friction with a mallet, but reducing the degree of
cross-linking increased their sensitivity to impact. All the compounds changed
colour when left for 30 min at 100 °C suggesting they produced NOx which was
subsequently absorbed.

A highly cross-linked βNCPCD sample was found to be sensitive to impact (FoI


= 29) but insensitive to ESD stimuli up to 4.5 J. The ESD test result is very
significant because nitrated βCD is highly sensitive to ESD (0.0125 J) and the
low sensitivity of βNCPCDs indicates they could be promising as a new
generation of energetic binders.

Stability tests indicated that softer βNCPCD compounds are unstable when
placed at 80 °C for 3 days even in the presence of a stabiliser. This may be due
to the inherent viscosity of these materials, which is thought to speed up the
degradation of the nitroester groups.

In conclusion, the new inert and nitrated βCD-based binders designed and
developed during this PhD project offer a promising alternative to current
binders for energetic formulations. The greatest strengths of the project are the
use of and environmentally sustainable βCD precursor, the absence of
flammable organic solvents in the synthesis method, and the easy synthesis
and processability of the inert and energetic products. The highlight of the
nitrated derivatives is their insensitivity to ESD at values up to 4.5 J compared
to the low value of 0.12555 J for similar compounds reported in literature.

180
Notably, an inert βCPCD derivative with a PEG:βCD ratio of 3.7:1 is now being
exploited by BAE Systems in Glascoed as a potential alternative to NC in
propellant formulations.

Recommendations

The scope for future work in the area covered by this PhD project is
considerable. Areas worthy of further investigation include, but are not limited
to:

1. Investigating the inclusion properties of βCXCDs, which are widely exploited


in pharmaceutical research because they can change the physical
properties of encapsulated materials. These tests could help to optimise the
manufacturing process for energetics containing βCXCDs and their nitrated
derivatives. The characterisation of Mw during this project was not ideal due
to the limited resources available for this task. Studying the Mw of the
products should be extended to determine the effect of the reaction
conditions and the consistency of the properties in each trial.
GPC/MALDI-TOF (matrix-assisted laser desorption/ionisation time of flight)
mass spectrometry could also be applied to confirm the results of the
GPC/refractive index studies with the βCXCDs.
2. The phenomenon of thermal healability looks interesting for manufacturing
purposes and further investigations of the healing process are needed to
determine the degree of healing inherited by the formulations.
3. Curing is another step towards the manufacturing of energetic devices.
Appendix A1 describes the first attempted curing experiments in the project
to obtain polymer-bonded explosive (PBX) formulations from the remaining
free hydroxyl groups in the βCXCDs, although the results are still
preliminary.
4. Scaling up the formulations even further (500 g mixes) and investigating the
effects of plasticisers, filler particle size distributions and automated mixing

181
processes on the mechanical properties and sensitivity of pressed/cast PBX
pellets.
5. Further compatibility testing is necessary to determine the differences in
compatibility between the βCXCDs and βNCXCDs when the XEG:βCD ratio
varies. In addition, compatibility with nitroglycerine and NC should be
assessed.
6. All forms of analysis should be extended to the level of triplicates in order to
increase confidence in the results.
7. Scaling up the synthesis of the energetic derivatives for EMTAP tests
should be extended to all compounds in order to correlate the mechanical
characteristics and hazard properties of the βNCXCDs. In addition, hazard
tests should be extended to real formulations to determine the suitability of
the binders for real-world applications.
8. Stabilisation studies were performed on only one βNCXCD compound.
Stability tests should be extended to obtain data for all the compounds
synthesised in the project, to determine how the rheology of the products
affects their stability.

The starting material was βCD cyclodextrin because it is less expensive than
γCD. However, the studies performed with βCD should be replicated with γCD
so that the complexation of energetic molecules with this larger macrocycle can
be exploited.

182
REFERENCES
1. Raymond L Beauregard. The History of Insensitive Munitions.
http://www.insensitivemunitions.org/history/preface/

2. Akhavan J. Chemistry of Explosives. Chemistry of Explosives. 2004. pp. P001–P004.


DOI:10.1039/9781847552020

3. NATO - STANAG 4439 - POLICY FOR INTRODUCTION AND ASSESSMENT OF


INSENSITIVE MUNITIONS (IM). 2018.

4. DEPARTMENT OF DEFENSE., TEST METHOD STANDARD. MIL-STD-2105 C


HAZARD ASSESSMENT TESTS NON-NUCLEAR MUNITIONS. 2003.

5. Smith P. Coating Technologies for Insensitive Munitions. 2006.

6. Madsen T., DeFisher S., Baker EL., Suarez D., Al-Shehab N., Wilson A., et al. Explosive
venting techonology for cook-off response mitigation. 2010.

7. Peterson NR., Sweitzer JC. Composite Material Particle Impact Mitigation Sleeve
Testing. Procedia Engineering. Elsevier; 1 January 2015; 103: 475–481.
DOI:10.1016/J.PROENG.2015.04.062

8. Sharp M., Sloan M. Active Passive Mitigation Devices S3 Assessment.

9. G. G. Ang; Pisharath Sreekumar., Ang HG., Pisharath S. Energetic Polymers - Binders


and Plasticizers for Enhancing Performance. Energetic Polymers - Binders and
Plasticizers for Enhancing Performance. Wiley-VCH; 2012. 510 p.
DOI:10.1002/prep.201280003

10. Fakirov S. Fundamentals of Polymer Science for Engineers. Weinheim, Germany: Wiley-
VCH Verlag GmbH & Co. KGaA; 2017. DOI:10.1002/9783527802180

11. Ebewele RO. Polymer science and technology. CRC Press; 2000. 483 p.

12. Guo Q. Polymer morphology : principles, characterization, and processing.

13. Kavesh S., Schultz JM. Meaning and measurement of crystallinity in polymers: A
Review. Polymer Engineering and Science. John Wiley & Sons, Ltd; 1 November 1969;
9(6): 452–460. DOI:10.1002/pen.760090612

14. Peters EN. Thermoplastics, Thermosets, and Elastomers-Descriptions and Properties.


Mechanical Engineers’ Handbook. Hoboken, NJ, USA: John Wiley & Sons, Inc.; 2015.
pp. 1–48. DOI:10.1002/9781118985960.meh109

15. Tadmor Z., Gogos CG. Principles of polymer processing. Wiley-Interscience; 2006. 961
p.

16. McKeen LW. The effect of creep and other time related factors on plastics and
elastomers. W. Andrew; 2009. 406 p.

17. Sanghavi RR., Asthana SN., Karir JS., Singh H. Studies on thermoplastic elastomers
based RDX-propellant compositions. Journal of Energetic Materials. Taylor & Francis
Group ; January 2001; 19(1): 79–95. DOI:10.1080/07370650108219393

18. Cooper PW. Explosives engineering. VCH; 1996. 460 p.

183
19. Junisbekov TM., Kestelman VN., Malinin NI. Stress Relaxation in Viscoelastic Materials.
Science Publishers; 2003. 1-82 p.

20. G. G. Ang; Pisharath Sreekumar. Polymers as Binders and Plasticizers - Historical


Perspective. Energetic Polymers - Binders and Plasticizers for Enhancing Performance.
2012. 510 p. DOI:10.1002/prep.201280003

21. Bohn M a. NC-based energetic materials - stability , decomposition , and ageing.


Nitrocellulose - Supply, Ageing and Characterisation. 2007;

22. Agrawal J. High energy materials. 2010. 495 p. DOI:10.1002/9783527628803

23. Laboratory & RCRA Group Los Alamos. Waste Explosives Treated by Open Detonation
at Los Alamos.

24. Persson P-A., Holmberg R., Lee J. Rock blasting and explosives engineering. CRC
Press; 1994. 540 p.

25. Ang HG., Pisharath S. Energetic polymers : binders and plasticizers for enhancing
performance. Wiley-VCH; 2012. 218 p.

26. NPFC - MIL-DTL-244 - NITROCELLULOSE | E. 2014.

27. Saunders CW., Taylor LT. A review of the synthesis, chemistry and analysis of
nitrocellulose. Journal of Energetic Materials. Taylor & Francis Group ; September 1990;
8(3): 149–203. DOI:10.1080/07370659008012572

28. Pourmortazavi SM., Hosseini SG., Rahimi-Nasrabadi M., Hajimirsadeghi SS., Momenian
H. Effect of nitrate content on thermal decomposition of nitrocellulose. Journal of
Hazardous Materials. 15 March 2009; 162(2–3): 1141–1144.
DOI:10.1016/j.jhazmat.2008.05.161

29. Sutton E. From polymers to propellants to rockets - A history of Thiokol. 35th Joint
Propulsion Conference and Exhibit. Reston, Virigina: American Institute of Aeronautics
and Astronautics; 1999. DOI:10.2514/6.1999-2929

30. Hendel., J. F. Review of solid propellants for space exploration. 1 October 1965;

31. Cesaroni AJ. US 6,740,180 B1: Thermoplastic polymer propellant compositions. US:
United States Patent and Trademark Office; 2004.

32. Salazar MR., Kress JD., Lightfoot JM., Russell BG., Rodin WA., Woods L. Low-
temperature oxidative degradation of PBX 9501 and its components determined via
molecular weight analysis of the Poly[ester urethane] binder. Polymer Degradation and
Stability. December 2009; 94(12): 2231–2240.
DOI:10.1016/j.polymdegradstab.2009.08.011

33. Chaturvedi S., Dave PN. Solid propellants: AP/HTPB composite propellants. Arabian
Journal of Chemistry. Elsevier; 8 January 2015; DOI:10.1016/j.arabjc.2014.12.033

34. Wingborg N. Increasing the tensile strength of HTPB with different isocyanates and
chain extenders. Polymer Testing. 2002; 21(3): 283–287. DOI:10.1016/S0142-
9418(01)00083-6

35. Keshavarz MH., Klapötke TM. The Properties of Energetic Materials. Berlin, Boston: De
Gruyter; 2017. DOI:10.1515/9783110521887

184
36. GOLDING P., BELLAMY AJ., CONTINI AE., DOSSI E. POLYPHOSPHAZENES. 28
December 2013;

37. Badgujar DM., Talawar MB., Zarko VE., Mahulikar PP. New directions in the area of
modern energetic polymers: An overview. Combustion, Explosion, and Shock Waves.
Pleiades Publishing; 31 July 2017; 53(4): 371–387. DOI:10.1134/S0010508217040013

38. Szejtli J. Introduction and General Overview of Cyclodextrin Chemistry. Chemical


Reviews. 1998; 98(5): 1743–1754. DOI:10.1021/cr970022c

39. Zheng M., Endo T., Zimmermann W. Enzymatic Synthesis and Analysis of Large-Ring
Cyclodextrins. Australian Journal of Chemistry. CSIRO PUBLISHING; 2002; 55(2): 39.
DOI:10.1071/CH01189

40. Del Valle EMM. Cyclodextrins and their uses: A review. Process Biochemistry. 2004;
39(9): 1033–1046. DOI:10.1016/S0032-9592(03)00258-9

41. Challa R., Ahuja A., Ali J., Khar RK. Cyclodextrins in drug delivery: An updated review.
AAPS PharmSciTech. 2005; 6(2): E329–E357. DOI:10.1208/pt060243

42. Das SK., Rajabalaya R., David S., Gani N., Khanam J., Nanda A. Cyclodextrins-the
molecular container. Research Journal of Pharmaceutical, Biological and Chemical
Sciences. 2013; 4(2): 1694–1720.

43. Redenti E., Szente L., Szejtli J. Drug/cyclodextrin/hydroxy acid multicomponent systems.
Properties and pharmaceutical applications. Journal of Pharmaceutical Sciences. 2000;
89(1): 1–8. DOI:10.1002/(SICI)1520-6017(200001)89:1<1::AID-JPS1>3.0.CO;2-W

44. Hădărugă NG., Bandur GN., David I., Hădărugă DI. A review on thermal analyses of
cyclodextrins and cyclodextrin complexes. Environmental Chemistry Letters. 7
September 2018; DOI:10.1007/s10311-018-0806-8

45. Specogna E., Li KW., Djabourov M., Carn F., Bouchemal K. Dehydration, dissolution,
and melting of cyclodextrin crystals. Journal of Physical Chemistry B. American
Chemical Society; 29 January 2015; 119(4): 1433–1442. DOI:10.1021/jp511631e

46. Sabadini E., Cosgrove T., Egídio F do C. Solubility of cyclomaltooligosaccharides


(cyclodextrins) in H2O and D2O: a comparative study. Carbohydrate Research.
February 2006; 341(2): 270–274. DOI:10.1016/j.carres.2005.11.004

47. Larsen KL. Large Cyclodextrins. Journal of Inclusion Phenomena and Macrocyclic
Chemistry. Kluwer Academic Publishers; 2002; 43(1/2): 1–13.
DOI:10.1023/A:1020494503684

48. Gattuso G., Gargiulli C., Parisi MF. Threading cyclodextrins in chloroform: A
[2]pseudorotaxane. International Journal of Molecular Sciences. 2007; 8(10): 1052–
1063. DOI:10.3390/i8101052

49. Gattuso G., Nepogodiev SA., Stoddart JF. Synthetic cyclic oligosaccharides. Chemical
Reviews. 1998; 98(5): 1919–1958. DOI:10.1021/cr960133w

50. Liu L., Guo QX. The driving forces in the inclusion complexation of cyclodextrins. Journal
of Inclusion Phenomena. 2002; 42(1–2): 1–14. DOI:10.1023/A:1014520830813

51. Bing-Lin He (Ping-Lum Ho)., Zhao X Bin. Study of the oxidation of crosslinked β-
cyclodextrin polymer and its use in the removal of urea. I. Reactive Polymers. 1992;
18(3): 229–235. DOI:10.1016/0923-1137(92)90653-J

185
52. Mocanu G., Vizitiu D., Carpov A. Cyclodextrin polymers. Journal of Bioactive and
Compatible Polymers. 2001; 16(4): 315–342. DOI:10.1106/JJUV-8F2K-JGYF-HNGF

53. Oliveira T., Botelho G., Alves NM., Mano JF. Inclusion complexes of α-cyclodextrins with
poly(d,l-lactic acid): Structural, characterization, and glass transition dynamics. Colloid
and Polymer Science. 2014; 292(4): 863–871. DOI:10.1007/s00396-013-3127-2

54. Jana M., Bandyopadhyay S. Molecular Dynamics Study of β-Cyclodextrin–Phenylalanine


(1:1) Inclusion Complex in Aqueous Medium. The Journal of Physical Chemistry B.
American Chemical Society; 8 August 2013; 117(31): 9280–9287.
DOI:10.1021/jp404348u

55. Shanmugam M., Ramesh D., Nagalakshmi V., Kavitha R., Rajamohan R., Stalin T.
Host–guest interaction of l-tyrosine with β-cyclodextrin. Spectrochimica Acta Part A:
Molecular and Biomolecular Spectroscopy. Elsevier; 1 November 2008; 71(1): 125–132.
DOI:10.1016/J.SAA.2007.10.054

56. Francisco R., Biagi G. Characterization and Synthesis of Cyclodextrin Inclusion


Complexes and their Applications as Fluorescent Probes for Sensing
Biomacromolecules by Characterization and Synthesis of Cyclodextrin Inclusion. 2012;

57. Rachmawati H., Edityaningrum CA., Mauludin R. Molecular Inclusion Complex of


Curcumin–β-Cyclodextrin Nanoparticle to Enhance Curcumin Skin Permeability from
Hydrophilic Matrix Gel. AAPS PharmSciTech. 29 December 2013; 14(4): 1303–1312.
DOI:10.1208/s12249-013-0023-5

58. Giordano F., Novak C., Moyano JR. Thermal analysis of cyclodextrins and their inclusion
compounds. Thermochimica Acta. December 2001; 380(2): 123–151.
DOI:10.1016/S0040-6031(01)00665-7

59. Akira H., Masaoki F., Shun-ichi N. Inclusion of aromatic compounds by a b-cyclodextrin-
epichlorohydrin polymer. Polymer Journal. 1981. pp. 777–781.
DOI:10.1295/polymj.13.777

60. Giglio V., Sgarlata C., Vecchio G. Novel amino-cyclodextrin cross-linked oligomer as
efficient carrier for anionic drugs: A spectroscopic and nanocalorimetric investigation.
RSC Advances. Royal Society of Chemistry; 2015; 5(22): 16664–16671.
DOI:10.1039/c4ra16064a

61. Szejtli J., Fenyvesi E., Zoltan S., Zsadon B., Tudos F. Cyclodextrin-polyvinyl alcohol
polymers and a process for the preparation thereof in a pearl, foil, fiber or block form. 23
June 1981;

62. Cyclodextrin-epichlorohydrin polymers for separating and adsorbing substances. 26


December 1991;

63. Schneider H-J., Hacket F., Rüdiger V., Ikeda H. NMR Studies of Cyclodextrins and
Cyclodextrin Complexes. Chemical Reviews. 1998; 98(5): 1755–1786.
DOI:10.1021/cr970019t

64. ISO 6796:1981 - Polyglycols for industrial use -- Determination of hydroxyl number --
Phthalic anhydride esterification method. https://www.iso.org/standard/13294.html

65. ASTM E222-17, Standard Test Methods for Hydroxyl Groups Using Acetic Anhydride
Acetylation, ASTM International, West Conshohocken, PA, 2017.

66. ASTM E1899-16, Standard Test Method for Hydroxyl Groups Using Reaction with p-

186
Toluenesulfonyl Isocyanate (TSI) and Potentiometric Titration with Tetrabutylammonium
Hydroxide, ASTM International, West Conshohocken, PA, 2016.

67. Sheremata TW., Hawari J. Cyclodextrins for desorption and solubilization of 2,4,6-
trinitrotoluene and its metabolites from soil. Environmental Science and Technology.
2000; 34(16): 3462–3468. DOI:10.1021/es9910659

68. Zhao D., Zhao L., Zhu C., Tian Z., Shen X. Synthesis and properties of water-insoluble
β-cyclodextrin polymer crosslinked by citric acid with PEG-400 as modifier. Carbohydrate
Polymers. Elsevier Ltd; 2009; 78(1): 125–130. DOI:10.1016/j.carbpol.2009.04.022

69. Jiang H., Yang Z., Zhou X., Fang Y., Ji H. Immobilization of β-cyclodextrin as insoluble
β-cyclodextrin polymer and its catalytic performance. Chinese Journal of Chemical
Engineering. 2012; 20(4): 784–792. DOI:10.1016/S1004-9541(11)60249-8

70. Zhao D., Zhao L., Zhu CS., Huang WQ., Hu JL. Water-insoluble β-cyclodextrin polymer
crosslinked by citric acid: Synthesis and adsorption properties toward phenol and
methylene blue. Journal of Inclusion Phenomena and Macrocyclic Chemistry. 2009;
63(3–4): 195–201. DOI:10.1007/s10847-008-9507-4

71. Kono H., Nakamura T. Polymerization of β-cyclodextrin with 1,2,3,4-


butanetetracarboxylic dianhydride: Synthesis, structural characterization, and bisphenol
A adsorption capacity. Reactive and Functional Polymers. Elsevier Ltd; 2013; 73(8):
1096–1102. DOI:10.1016/j.reactfunctpolym.2013.04.006

72. Roik N V., Belyakova LA. Thermodynamic, IR spectral and X-ray diffraction studies of
the “β-cyclodextrin-para-aminobenzoic acid” inclusion complex. Journal of Inclusion
Phenomena and Macrocyclic Chemistry. 27 April 2011; 69(3–4): 315–319.
DOI:10.1007/s10847-010-9737-0

73. Maréchal Y., Yves. The molecular structure of liquid water delivered by absorption
spectroscopy in the whole IR region completed with thermodynamics data. Journal of
Molecular Structure. October 2011; 1004(1–3): 146–155.
DOI:10.1016/j.molstruc.2011.07.054

74. Krause, Rui: Mamba BB. 9. Cyclodextrin polymers : Synthesis and Application in Water
Treatment. Cyclodextrins: Chemistry and Physics. 2010;

75. Yi JM., Tang KW. Preparation of insoluble β-cd polymer and its application to
enantiomers and isomers separation. Journal of Central South University of Technology
(English Edition). 2001; 8(1): 57–59. DOI:10.1007/s11771-001-0026-3

76. Martel B., Ruffin D., Weltrowski M., Lekchiri Y., Morcellet M. Water-soluble polymers and
gels from the polycondensation between cyclodextrins and poly(carboxylic acid)s: A
study of the preparation parameters. Journal of Applied Polymer Science. 2005; 97(2):
433–442. DOI:10.1002/app.21391

77. Davis ME., Gonzalez H., Hwang SSJ. Linear Cyclodextrin Copolymers. Patent
US/2006/7091192B1. Int. Appl. C07H 3/00. 2006;

78. Kono H., Nakamura T., Hashimoto H., Shimizu Y. Characterization, molecular dynamics,
and encapsulation ability of β-cyclodextrin polymers crosslinked by polyethylene glycol.
Carbohydrate Polymers. Elsevier Ltd.; 5 September 2015; 128: 11–23.
DOI:10.1016/j.carbpol.2015.04.009

79. Wycisk A., Döring A., Schneider M., Schönhoff M., Kuckling D. Synthesis of ß-
cyclodextrin-based star block copolymers with thermo-responsive behavior. Polymers.

187
2015; 7(5): 921–938. DOI:10.3390/polym7050921

80. Mhlanga SD., Mamba BB., Krause RW., Malefetse TJ. Removal of organic contaminants
from water using nanosponge cyclodextrin polyurethanes. Journal of Chemical
Technology & Biotechnology. Wiley-Blackwell; April 2007; 82(4): 382–388.
DOI:10.1002/jctb.1681

81. Kayaci F., Aytac Z., Uyar T. Surface modification of electrospun polyester nanofibers
with cyclodextrin polymer for the removal of phenanthrene from aqueous solution.
Journal of Hazardous Materials. Elsevier B.V.; 2013; 261: 286–294.
DOI:10.1016/j.jhazmat.2013.07.041

82. Birck C., Degoutin S., Tabary N., Miri V., Bacquet M. New crosslinked cast films based
on poly(vinyl alcohol): Preparation and physico-chemical properties. Express Polymer
Letters. 2014; 8(12): 941–952. DOI:10.3144/expresspolymlett.2014.95

83. Chen W., Wang C., Yan L., Huang L., Zhu X., Chen B., et al. Improved
polyvinylpyrrolidone microneedle arrays with non-stoichiometric cyclodextrin. Journal of
Materials Chemistry B. 2014; 2(12): 1699–1705. DOI:10.1039/c3tb21698e

84. Radia O., Rogalska E., Moulay-Hassane G. Preparation of meloxicamβ-


cyclodextrinpolyethylene glycol 6000 ternary system: Characterization, in vitro and in
vivo bioavailability. Pharmaceutical Development and Technology. Taylor & Francis; 23
October 2012; 17(5): 632–637. DOI:10.3109/10837450.2011.565347

85. Salmaso S., Semenzato A., Bersani S., Matricardi P., Rossi F., Caliceti P.
Cyclodextrin/PEG based hydrogels for multi-drug delivery. International Journal of
Pharmaceutics. 10 December 2007; 345(1–2): 42–50.
DOI:10.1016/j.ijpharm.2007.05.035

86. Nielsen TT., Wintgens V., Larsen KL., Amiel C. Synthesis and characterization of
poly(ethylene glycol) based β-cyclodextrin polymers. Journal of Inclusion Phenomena
and Macrocyclic Chemistry. Springer Netherlands; 30 December 2009; 65(3): 341–348.
DOI:10.1007/s10847-009-9591-0

87. Cesteros LC., Ramírez CA., Peciña A., Katime I. Poly(ethylene glycol-ß-cyclodextrin)
gels: Synthesis and properties. Journal of Applied Polymer Science. 15 October 2006;
102(2): 1162–1166. DOI:10.1002/app.24390

88. Klaewklod A., Tantishaiyakul V., Hirun N., Sangfai T., Li L. Characterization of
supramolecular gels based on β-cyclodextrin and polyethyleneglycol and their potential
use for topical drug delivery. Materials Science and Engineering C. May 2015; 50: 242–
250. DOI:10.1016/j.msec.2015.02.018

89. Meille S V., Allegra G., Geil PH., He J., Hess M., Jin J-I., et al. Definitions of terms
relating to crystalline polymers (IUPAC Recommendations 2011). Pure and Applied
Chemistry. De Gruyter; 3 August 2011; 83(10): 1831–1871. DOI:10.1351/PAC-REC-10-
11-13

90. Peppas N. Hydrogels in pharmaceutical formulations. European Journal of


Pharmaceutics and Biopharmaceutics. 3 July 2000; 50(1): 27–46. DOI:10.1016/S0939-
6411(00)00090-4

91. Crini G. Recent developments in polysaccharide-based materials used as adsorbents in


wastewater treatment. Progress in Polymer Science (Oxford). 2005; 30(1): 38–70.
DOI:10.1016/j.progpolymsci.2004.11.002

188
92. Van De Manakker F., Vermonden T., El Morabit N., Van Nostrum CF., Hennink WE.
Rheological behavior of self-assembling PEG-β-cyclodextrin/ PEG-cholesterol hydrogels.
Langmuir. 4 November 2008; 24(21): 12559–12567. DOI:10.1021/la8023748

93. Jia Y-GG., Zhu XX. Self-healing supramolecular hydrogel made of polymers bearing
cholic acid and β-cyclodextrin pendants. Chemistry of Materials. American Chemical
Society; 13 January 2015; 27(1): 387–393. DOI:10.1021/cm5041584

94. Jeong D., Joo S-W., Shinde VV., Jung S. Triple-crosslinkedβ-cyclodextrin oligomer self-
healing hydrogel showing high mechanical strength, enhanced stability and pH
responsiveness. Carbohydrate Polymers. Elsevier; 15 October 2018; 198: 563–574.
DOI:10.1016/J.CARBPOL.2018.06.117

95. Guo K., Lin M-S., Feng J-F., Pan M., Ding L-S., Li B-J., et al. The Deeply Understanding
of the Self-Healing Mechanism for Self-Healing Behavior of Supramolecular Materials
Based on Cyclodextrin-Guest Interactions. Macromolecular Chemistry and Physics. John
Wiley & Sons, Ltd; 1 May 2017; 218(10): 1600593. DOI:10.1002/macp.201600593

96. Miao T., Fenn SL., Charron PN., Oldinski RA. Self-Healing and Thermoresponsive Dual-
Cross-Linked Alginate Hydrogels Based on Supramolecular Inclusion Complexes.
Biomacromolecules. NIH Public Access; 14 December 2015; 16(12): 3740–3750.
DOI:10.1021/acs.biomac.5b00940

97. Poudel AJ., He F., Huang L., Xiao L., Yang G. Supramolecular hydrogels based on poly
(ethylene glycol)-poly (lactic acid) block copolymer micelles and α-cyclodextrin for
potential injectable drug delivery system. Carbohydrate Polymers. 15 August 2018; 194:
69–79. DOI:10.1016/j.carbpol.2018.04.035

98. Burattini S., Greenland BW., Chappell D., Colquhoun HM., Hayes W. Healable polymeric
materials: a tutorial review. Chemical Society Reviews. 2010; 39(6): 1973.
DOI:10.1039/b904502n

99. Binder W (Wolfgang H. Self-healing polymers : from principles to applications. Wiley-


VCH; 2013. 425 p.

100. Garcia SJ., Fischer HR. Self-healing polymer systems: properties, synthesis and
applications. Smart Polymers and their Applications. Woodhead Publishing; 1 January
2014; : 271–298. DOI:10.1533/9780857097026.1.271

101. Kakuta T., Takashima Y., Harada A. Highly elastic supramolecular hydrogels using host-
guest inclusion complexes with cyclodextrins. Macromolecules. American Chemical
Society; 11 June 2013; 46(11): 4575–4579. DOI:10.1021/ma400695p

102. Jeong YK., Kwon TW., Lee I., Kim TS., Coskun A., Choi JW. Hyperbranched β-
cyclodextrin polymer as an effective multidimensional binder for silicon anodes in lithium
rechargeable batteries. Nano Letters. American Chemical Society; 12 February 2014;
14(2): 864–870. DOI:10.1021/nl404237j

103. Klein J. Evidence for reptation in an entangled polymer melt. Nature. Nature Publishing
Group; 12 January 1978; 271(5641): 143–145. DOI:10.1038/271143a0

104. Muthukumar M., Baumgärtner A. Diffusion of a Polymer Chain in Random Media.


Macromolecules. American Chemical Society; July 1989; 22(4): 1941–1946.
DOI:10.1021/ma00194a071

105. Guo K., Lin M-SS., Feng J-FF., Pan M., Ding L-SS., Li B-JJ., et al. The Deeply
Understanding of the Self-Healing Mechanism for Self-Healing Behavior of

189
Supramolecular Materials Based on Cyclodextrin–Guest Interactions. Macromolecular
Chemistry and Physics. Wiley-Blackwell; 1 May 2017; 218(10): 1600593.
DOI:10.1002/macp.201600593

106. Hart LR., Harries JL., Greenland BW., Colquhoun HM., Hayes W. Healable
supramolecular polymers. Polymer Chemistry. 2013; 4(18): 4860.
DOI:10.1039/c3py00081h

107. Tavengwa NT., Hintsho N., Durbach S., Weiersbye I., Cukrowska E., Chimuka L.
Extraction of explosive compounds from aqueous solutions by solid phase extraction
using β-cyclodextrin functionalized carbon nanofibers as sorbents. Journal of
Environmental Chemical Engineering. June 2016; 4(2): 2450–2457.
DOI:10.1016/j.jece.2016.04.024

108. Hawari J., Paquet L., Zhou E., Halasz A., Zilber B. Enhanced Recovery of the Explosive
Hexahydro-1,3,5-Trinitro-1,3,5-Triazine (RDX) From Soil: cyclodextrin versus anionic
surfactants. Chemosphere. Pergamon; 1 May 1996; 32(10): 1929–1936.
DOI:10.1016/0045-6535(96)00102-6

109. Ruebner A., Statton GL., Consaga JP. Polymeric cyclodextrin nitrate esters. 2003. p. 10.
DOI:10.1074/JBC.274.42.30033.(51)

110. Cahill S., Bulusu S. Molecular complexes of explosives with cyclodextrins I.


Characterization of complexes with the nitramines RDX, HMX and TNAZ in solution
by1H NMR spin-lattice relaxation time measurements. Magnetic Resonance in
Chemistry. August 1993; 31(8): 731–735. DOI:10.1002/mrc.1260310808

111. Cahill S., Owens FJ. Molecular Complexes of Explosives with Cyclodextrins. 2.
Preparation and Characterization of a Solid Complex of B-Cyclodextrin with the
Nitramine 1,3,3-Trinitroazetidine (TNAZ). The Journal of Physical Chemistry. 1994;
98(28): 7095–7100.

112. Maksimowski P., Rumianowski T. Properties of the Gamma-Cyclodextrin /CL-20 System.


Central European Journal of Energetic Materials. 2016; 13(1): 217–229.
DOI:10.22211/cejem/64973

113. Maksimowski P., Grzegorczyk A., Cieślak K., Gołofit T., Chmielarek M., Tomaszewski
W., et al. γ-Cyclodextrin Nitrate/CL-20 Complex: Preparation and Properties.
Propellants, Explosives, Pyrotechnics. John Wiley & Sons, Ltd; 15 November 2018;
DOI:10.1002/prep.201800301

114. Consaga JP., Collignon SL. Energetic composites of cyclodextrin nitrate esters and
nitrate. Patent US/1992/5114506. Int. Appl. C06B 25/00. 1992;

115. Consaga JP., Plata L., Ents USPD. Chemically reactive fragmentation warhead. 2001; 1.

116. Romanova LB., Barinova LS., Lagodzinskaya G V., Kazakov AI., Mikhailov YM.
Preparation and NMR analysis of β-cyclodextrin nitrates. Russian Journal of Applied
Chemistry. 2014; 87(12): 1884–1889. DOI:10.1134/S1070427214120155

117. Zukas JA., Walters WP (William P. Explosive effects and applications. 431 p.

118. Copperhead Chemical Database. http://www.copperheadchemical.com/

119. Energetic Materials Testing Assessment Policy Manual of Tests (EMTAP) Manual of
Tests. 2007.

190
120. Dossi E., Akhavan J., Gaulter SE., Williams RG., Doe WJ. Cross-linking of Hydroxyl-
terminated Polyols with Triethyleneglycol Diglycidyl Ether: An Alternative to Toxic
Isocyanates. Propellants, Explosives, Pyrotechnics. Wiley-Blackwell; 1 March 2018;
43(3): 241–250. DOI:10.1002/prep.201700220

121. STANAG 4147 (Ed. 2), 2001, “Chemical compatibility of ammunition components with
explosives (non-nuclear applications)”.

122. Mazzeu MAC., Mattos E da C., Iha K. Studies on compatibility of energetic materials by
thermal methods. Journal of Aerospace Technology and Management. 2010; 2(1): 53–
58. DOI:10.5028/jatm.2010.02015358

123. Weiss F., Finkelmann H. Lyotropic liquid crystalline epoxide-amine addition polymers.
Colloid & Polymer Science. 1 October 2003; 281(11): 1007–1014. DOI:10.1007/s00396-
003-0951-9

124. Gaidamauskas E., Norkus E., Butkus E., Crans DC., Grinciene G. Deprotonation of β-
cyclodextrin in alkaline solutions. Carbohydrate Research. 2009; 344(2): 250–254.
DOI:10.1016/j.carres.2008.10.025

125. Atkins PW (Peter W., De Paula J. Atkins’ Physical chemistry. Oxford University Press;
2010. 972 p.

126. Hădărugă NG., Hădărugă DI. Water content of natural cyclodextrins and their essential
oil complexes: A comparative study between Karl Fischer titration and thermal methods.
Food Chemistry. Elsevier; 15 June 2012; 132(4): 1741–1748.
DOI:10.1016/J.FOODCHEM.2011.11.003

127. Zhou G., Zhao T., Wan J., Liu C., Liu W., Wang R. Predict the glass transition
temperature and plasticization of β-cyclodextrin/water binary system by molecular
dynamics simulation. Carbohydrate Research. Elsevier Ltd; 2015; 401: 89–95.
DOI:10.1016/j.carres.2014.10.026

128. Rodríguez-Tenreiro C., Alvarez-Lorenzo C., Concheiro Á., Torres-Labandeira JJ.


Characterization of cyclodextrincarbopol interactions by DSC and FTIR. Journal of
Thermal Analysis and Calorimetry. Kluwer Academic Publishers; 2004; 77(2): 403–411.
DOI:10.1023/B:JTAN.0000038981.30494.f4

129. Abiad MG., Campanella OH., Carvajal MT. Assessment of Thermal Transitions by
Dynamic Mechanical Analysis (DMA) Using a Novel Disposable Powder Holder.
Pharmaceutics. Multidisciplinary Digital Publishing Institute (MDPI); 24 March 2010;
2(2): 78–90. DOI:10.3390/pharmaceutics2020078

130. Tabary N., Garcia-Fernandez MJ., Danède F., Descamps M., Martel B., Willart J-FF.
Determination of the glass transition temperature of cyclodextrin polymers. Carbohydrate
Polymers. Elsevier; 5 September 2016; 148: 172–180.
DOI:10.1016/j.carbpol.2016.04.032

131. van Ekeren PJ., Carton EP. Polyurethanes for potential use in transparent armour
investigated using DSC and DMA. Journal of Thermal Analysis and Calorimetry. 16
August 2011; 105(2): 591–598. DOI:10.1007/s10973-011-1665-8

132. Sepe MP., Rapra Technology Limited. Thermal analysis of polymers. Rapra Technology
Ltd; 1997. 120 p.

133. Contini AE., Flood N., McAteer D., Mai N., Akhavan J. Low hazard small-scale synthesis
and chemical analysis of high purity nitroglycerine (NG). RSC Advances. The Royal

191
Society of Chemistry; 13 October 2015; 5(106): 87228–87232.
DOI:10.1039/C5RA17951C

134. Rodin MD., Romanova LB., Darovskih A V., Gorbunova MA., Tarasov AE. Ir-
Spectroscopic Method for Determining the Degree of Nitration of β-Cyclodextrin. Journal
of Applied Spectroscopy. Springer US; 20 September 2018; 85(4): 691–696.
DOI:10.1007/s10812-018-0706-5

135. Kolthoff IM., Sandell EB., Moskovitz B. The Volumetric Determination of Nitrates with
Ferrous Sulfate as Reducing Agent. Journal of the American Chemical Society. April
1933; 55(4): 1454–1457. DOI:10.1021/ja01331a020

136. Kwei TK. The effect of hydrogen bonding on the glass transition temperatures of polymer
mixtures. Journal of Polymer Science: Polymer Letters Edition. June 1984; 22(6): 307–
313. DOI:10.1002/pol.1984.130220603

137. Park SJ., Seo MK. Solid-Liquid Interface. Interface Science and Technology. Elsevier; 1
January 2011; 18: 147–252. DOI:10.1016/B978-0-12-375049-5.00003-7

138. Hansen CM. Hansen solubility parameters : a user’s handbook. CRC Press; 2007. 519
p.

139. Vogelsanger B. Chemical Stability, Compatibility and Shelf Life of Explosives. CHIMIA
International Journal for Chemistry. 1 June 2004; 58(6): 401–408.
DOI:10.2533/000942904777677740

140. Trache D., Tarchoun AF. Stabilizers for nitrate ester-based energetic materials and their
mechanism of action: a state-of-the-art review. Journal of Materials Science. Springer
US; 23 January 2018; 53(1): 100–123. DOI:10.1007/s10853-017-1474-y

141. Manelis GB. Thermal decomposition and combustion of explosives and propellants.
Taylor & Francis; 2003. 362 p.

142. Burov YM. Thermal Decomposition of Solid Energetic Materials.


http://www.abitura.com/modern_physics/x1.html

143. Apatoff JB., Norwitz G. Role of Diphenylamine as a Stabilizer in Propellants; Analytical


Chemistry of Diphenylamine in Propellants (A Survey Report). 1973.

192
Appendix A
Table A 1 List of βCPCDs synthesised using PEGDGE as the cross-linker.

Cross-
Sample Logbook Feed βCD Water PEGDGE NaOH NaOH Time T linker Time Yield
1
addition Solubility
ID ref. ratio (g) (%) (mL) (% w/w) (mL) alkox. (h) (°C) reaction (%)
time.
IP3 βCPCD1 9:1 5.00 N.d 17.4 36 21 0 70 ~20min 0.5 1 Water
0.5+4.5
IP24 βCPCD2 9:1 5.00 N.d 17.4 50 21 0 70 ~20min 1 Water
RT

IP27 βCPCD3 9:1 5.00 N.d 17.4 5.6 21 0 70 ~20min 0.66 88 Insoluble

IP4 βCPCD4 3:1 5.00 N.d 5.8 5.6 21 0 70 ~20min 1 5 Water

IP26 βCPCD5 6:1 5.00 N.d 11.6 5.6 21 0 70 ~20min 0.66 91 Insoluble

IP28 βCPCD6 9:1 5.00 N.d 17.4 5.6 21 0 70 ~20min 1 1 Water

IP1 βCPCD7 9:1 5.00 N.d 17.4 50 21 0 70 ~20min 0.5 1 Water

IP2 βCPCD8 9:1 5.00 N.d 17.4 40 21 1 70 ~20min 1 1 Water

IP5 βCPCD9 9:1 5.00 N.d 17.4 40 21 1 70 ~20min 5 12 Water

IP29 βCPCD10 1:1 5.00 N.d 1.9 40 21 3 70 ~20min 1 1 Water

IP6 βCPCD11 5:1 5.00 N.d 9.7 5.6 21 1 70 ~20min 5 1 Water

193
Cross-
Sample Logbook Feed βCD Water PEGDGE NaOH NaOH Time T linker Time Yield
1
addition Solubility
ID ref. ratio (g) (%) (mL) (% w/w) (mL) alkox. (h) (°C) reaction (%)
time.

IP7 βCPCD12 5:1 5.00 N.d 9.7 5.6 21 0 50 ~20min 5 13 Water

IP30 βCPCD13 5:1 5.00 N.d 9.7 5.6 21 0 50 ~20min 2 9 Water

IP8 βCPCD14 5:1 5.00 N.d 9.7 5.6 21 16 50 ~20min 5 30 Water

IP31 βCPCD15 5:1 5.00 N.d 9.7 5.6 21 16 30 ~20min 7.5 23.6 Insoluble

IP32 βCPCD16 5:1 5.00 N.d 9.7 5.6 21 16 30 5 hours 1 23.9 Water

IP9 βCPCD17 5:1 5.00 N.d 9.7 5.6 21 16 30 ~20min 6 33.6 Water

IP33 βCPCD18 5:1 5.00 N.d 9.7 5.6 21 16 30 ~20min 7 33.2 Water

IP34 βCPCD19 5:1 2.5 N.d 4.8 20 10.5 16 70 ~20min 6 15 Water

Not
IP35 βCPCD20 5:1 2.5 N.d 4.8 5.6 KOH 10.5 16 30 5 hours 1.5 Insoluble
recovered
not
IP36 βCPCD21 1:1 2.5 N.d 9.7 5.6 100 16 30 ~20min 7.5 Water
recovered

IP37 βCPCD22 5:1 2.5 N.d 4.8 5.6 10+42.5 16 30 ~20min 24 48 Water

IP11 βCPCD23 3:1 2.5 N.d 2.9 5.6 10.5 16 30 ~20min 24 68 Water

IP38 βCPCD24 4:1 5.00 N.d 7.7 5.6 21 16 30* ~20min 24 39 Water

194
Cross-
Sample Logbook Feed βCD Water PEGDGE NaOH NaOH Time T linker Time Yield
1
addition Solubility
ID ref. ratio (g) (%) (mL) (% w/w) (mL) alkox. (h) (°C) reaction (%)
time.

IP10 βCPCD25 4:1 5.00 N.d 7.7 5.6 21 16 30 ~20min 24 36 Water

Not
IP39 βCPCD26 4:1 5.00 N.d 7.7 5.6 21 16 30 7 hours 17 Insoluble
recovered

IP40 βCPCD27 3:1 5.00 N.d 5.8 5.6 21 16 30 7 hours 17 48 Water

IP12 βCPCD28 2:1 5.00 N.d 3.9 5.6 21 16 30 7 hours 17 44 Water

not
IP41 βCPCD29 3.5:1 5.67 0.12 9.7 5.6 21 16 30 7 hours 17 Insoluble
recovered

IP42 βCPCD30 5:1 5.63 0.12 9.7 5.6 21 16 30 ~20min 6 61 Insoluble

not
IP43 βCPCD31 5:1 5.63 0.12 9.7 5.6 21 16 30 ~20min 5 Insoluble
recovered
not
IP44 βCPCD32 5:1 5.64 0.12 9.7 5.6 21 16 30 ~20min 5 Insoluble
recovered

IP13 βCPCD33 5:1 4.35 0.13 8.4 5.6 20.7 16 30 ~20min 6 68 Water

IP20 βCPCD34 5:1 4.35 0.13 8.4 5.6 20.7 16 30 ~20min 6 64 water

IP22 βCPCD35 4:1 4.40 0.12 6.8 5.6 20.8 16 30 ~20min 6 63 water

IP15 βCPCD36 4:1 4.40 0.12 6.8 5.6 20.8 16 30 ~20min 6 65 water

IP45 βCPCD37 2:1 4.15 0.17 3.2 5.6 18.7 16 30 ~20min 6 68 water

195
Cross-
Sample Logbook Feed βCD Water PEGDGE NaOH NaOH Time T linker Time Yield
1
addition Solubility
ID ref. ratio (g) (%) (mL) (% w/w) (mL) alkox. (h) (°C) reaction (%)
time.

IP17 βCPCD38 2:1 4.15 0.17 3.2 5.6 18.7 16 30 ~20min 6 68 water

IP46 βCPCD39 6:1 4.34 0.13 8.4 5.6 19.8 16 30 ~20min 6 gel Insoluble

IP47 βCPCD40 6:1 4.34 0.13 8.4 5.6 20.7 16 30 ~20min 6 37%+gel Insoluble

IP16 βCPCD41 3:1 4.15 0.17 4.8 5.6 20.7 16 30 ~20min 6 73 water

IP48 βCPCD42 3:1 4.15 0.17 4.8 5.6 19.8 16 30 ~20min 6 68 water

IP14 βCPCD43 3.8:1 4.34 0.13 8.4 5.6 19.8 16 30 7 hours 6 71 water

IP21 βCPCD44 5:1 4.34 0.13 8.4 5.6 19.8 16 30 ~20min 6 68 water

IP25 βCPCD45 3:1 4.34 0.13 5 5.6 19.8 16 30 ~20min 6 67 water

IP23 βCPCD46 4:1 4.34 0.13 6.7 5.6 19.8 16 30 ~20min 6 69 water

IP19 βCPCD47 5:1 4.34 0.13 8.4 5.6 19.8 16 30 ~20min 6 67 water

IP18 βCPCD48 5:1 4.34 0.13 8.4 5.6 19.8 16 30 ~20min 6 68 water

1 Dry mass of βCD determined by TGA after sample βCDPCD28.

196
Table A 2 List of βCHCDs using HEGDGE as the cross-linker.

Sample Logbook Feed 1 Water NaOH NaOH T Cross-linker Time Yield


βCD (g) HEGDGE (g) Time alkox. (h) Sol.
ID ref. ratio (%) (% w/w) (mL) (°C) addition time reaction (%)

IH1 βCHCD1 5:1 4.00 0.12 6.9 5.6 19 16 30 ~20min 6 68 water

IH2 βCHCD2 4:1 4.00 0.12 5.6 5.6 19 16 30 ~20min 6 72 water

IH3 βCHCD3 2:1 4.00 0.13 2.8 5.6 14.3 16 30 ~20min 6 65 water

IH4 βCHCD4 3:1 4.00 0.13 4.2 5.6 14.3 16 30 ~20min 6 67 water

1 Dry mass of βCD determined by TGA.

Table A 3 List of βCTCDs synthesised using TEGDGE as the cross-linker.

Cross-linker
Sample Logbook Feed βCD Water NaOH NaOH Time alkox. T Time Yield
TEGDGE (g) addition Sol.
ID ref. ratio (g) (%) (% w/w) (mL) (h) (°C) reaction (%)
time

IT5 βCTCD1 9:1 5 N.d 9.35 40 21 1 70 ~20min 5 35 Insoluble

IT6 βCTCD2 3:1 5 N.d 3.1 40 21 1 70 ~20min 5 40 Insoluble

IT7 βCTCD3 6:1 5 N.d 6.2 40 21 1 70 ~20min 5 33 Insoluble

IT8 βCTCD4 5:1 5 N.d 5.76 5.6 21 1 70 ~20min 5 10 water

IT9 βCTCD5 5:1 5 N.d 5.76 5.6 21 16 30 ~20min 5 38 water

197
Cross-linker
Sample Logbook Feed βCD Water NaOH NaOH Time alkox. T Time Yield
TEGDGE (g) addition Sol.
ID ref. ratio (g) (%) (% w/w) (mL) (h) (°C) reaction (%)
time

IT1 βCTCD6 5:1 2.00 0.12 2.3 5.6 9.5 16 30 ~20min 6 63 water

IT2 βCTCD7 4:1 2.00 0.12 1.84 5.6 9.5 16 30 ~20min 6 52 water

IT3 βCTCD8 3:1 2.00 0.12 1.39 5.6 9.5 16 30 ~20min 6 56 water

IT4 βCTCD9 2:1 2.00 0.12 0.93 5.6 9.5 16 30 ~20min 6 45 water

1 Dry mass of βCD determined by TGA.

198
Table A 4 Nitration of insoluble βCPCD precursor, sample IP26.

Precursor HNO3 % Volume HNO3 (mL) H2SO4 Volume Time DCM (mL) Stirr Yield
Sample Logbook (h)
% H2SO4 rpm (%)
ID ref. Name Solubility
(mL)
1,2
βNCPCD3 IP26 N 100 0.5 100 0.5 1.0 0.0 200 32
NP9
1,2
βNCPCD4 N 100 0.5 100 0.5 2.0 0.0 400 30
NP10 IP26
1,2
βNCPCD5 N 100 0.5 100 0.5 1.5 1.8 900 44
NP11 IP26
1,2
βNCPCD6 N 100 0.5 100 0.5 1.5 0.0 900 37
NP12 IP26

βNCPCD7 N 100 0.5 100 0.5 1.5 1.8 900 90


NP13 IP26

βNCPCD8 N 100 0.5 100 0.5 2.0 1.8 900 87


NP14 IP26

1 Low degree of nitration, gum-like βCPCD precursors.


2 Purification process affected the yield of the reaction.

Table A 5 Nitration of insoluble βCTCD precursors.

Precursor Volume
Sample Logbook H2SO4 Stirr Yield
HNO3 % Volume HNO3 (mL) H2SO4 Time (h) DCM (mL)
ID ref. Name Solubility % rpm (%)
(mL)

βNCTCD1 IT5 N 95 1.0 0 0.0 1.0 0.0 200 87


NT5

199
βNCTCD2 N 100 1.0 0 0.0 1.0 0.0 200 80
NT6 IT5

βNCTCD3 N 100 0.5 100 0.5 1.0 0.0 200 35


NT7 IT5

βNCTCD4 N 100 0.5 100 0.5 0.5 1.8 400 85


NT8 IT5

βNCTCD5 N 100 0.5 100 0.5 1.0 0.0 900 88


NT9 IT5

βNCTCD6 IT7 N 100 0.5 100 0.5 1.0 0.0 900 88


NT10

βNCTCD7 IT6 N 100 0.5 100 0.5 1.0 0.0 900 84


NT11

βNCTCD8 IT5 N 100 0.5 100 0.5 1.0 1.8 900 86


NT12

βNCTCD9 IT6 N 100 0.5 100 0.5 1.0 1.8 900 91


NT13

βNCTCD10 IT7 N 100 0.5 100 0.5 1.0 1.8 900 92


NT14

Table A 6 Nitration of βCPCD precursors.

1
H2SO4 Stirr Yield
HNO3 Mass of Volume
Sample Logbook Precursor Time DCM
precursor HNO3 Mass Sol
ID ref. name % (mL) (h) (mL) rpm %
(mg) (mL)
Sol. % mg

NP14 βNCPCD1 IP5 Y 100 200 0.5 100 0.5 0.5 1.8 200 163

NP15 βNCPCD2 IP5 Y 95 200 1 0 0 1 0 200 199

200
1
H2SO4 Stirr Yield
HNO3 Mass of Volume
Sample Logbook Precursor Time DCM
precursor HNO3 Mass Sol
ID ref. name % (mL) (h) (mL) rpm %
(mg) (mL)
Sol. % mg
purification
is difficult,
NP7 βNCPCD3 IP26 N 100 200 0.5 100 0.5 1 0 200
insoluble
precursor
purification
is difficult,
NP8 βNCPCD4 IP26 N 100 200 0.5 100 0.5 2 0 400
insoluble
precursor
purification
is difficult,
NP9 βNCPCD5 IP26 N 100 200 0.5 100 0.5 1.5 1.8 900 70
insoluble
precursor
purification
is difficult,
NP10 βNCPCD6 IP26 N 100 3000 0.5 100 0.5 1.5 0 900 300
insoluble
precursor
purification
is difficult,
NP11 βNCPCD7 IP26 N 100 200 0.5 100 0.5 1.5 1.8 900 103
insoluble
precursor
purification
is difficult,
NP12 βNCPCD8 IP26 N 100 200 0.5 100 0.5 2 1.8 900 169
insoluble
precursor

NP18 βNCPCD9 IP8 Y 100 200 1 1 700 180

201
1
H2SO4 Stirr Yield
HNO3 Mass of Volume
Sample Logbook Precursor Time DCM
precursor HNO3 Mass Sol
ID ref. name % (mL) (h) (mL) rpm %
(mg) (mL)
Sol. % mg

NP19 βNCPCD10 IP8 Y 95 200 1 1 700 187

almost
nothing
recovered
NP20 βNCPCD11 IP8 N 100 200 1 1 700
(filter
paper
problem)

NP21 βNCPCD12 IP11 Y 100 200 1 1 700 190

1+(1
NP22 βNCPCD13 IP10 Y 100 1000 5 5min 700 1070
dis)

NP16 βNCPCD14 βCPCD27 Y 100 1000 5 1 700 1230

NP23 βNCPCD15 IP10 Y 100 200 1 1 700 103

compound
is not easy
NP24 βNCPCD16 IP10 Y 100 200 1 1 700 117 to purify
cloud
solution
IP13/20
NP25 βNCPCD17 Y 100 1000 5 1 700 451
blend

IP13/20
NP13 βNCPCD18 Y 100 500 2.5 1 700 476
blend

202
1
H2SO4 Stirr Yield
HNO3 Mass of Volume
Sample Logbook Precursor Time DCM
precursor HNO3 Mass Sol
ID ref. name % (mL) (h) (mL) rpm %
(mg) (mL)
Sol. % mg

IP13/20
NP26 βNCPCD19 Y 100 2000 10 1 700 1637
blend

IP13/20
NP27 βNCPCD20 Y 100 1000 10 1 700 656
blend

IP13/20
NP28 βNCPCD21 Y 100 1000 10 1 700 395
blend

samples
IP13/20
damaged
NP29 βNCPCD22 blend Y 100 500 5 1 700 NA trying to
use
Na2CO3

IP13/20
NP30 βNCPCD23 Y 100 500 5 1 700 NA
blend

IP13/20
NP31 βNCPCD24 Y 100 300 2.1 1 700 150
blend

IP13/20
NP32 βNCPCD25 Y 100 300 2.1 1 700 98
blend

NP33 βNCPCD26 IP13/20 Y 100 300 2.1 1 700 72

203
1
H2SO4 Stirr Yield
HNO3 Mass of Volume
Sample Logbook Precursor Time DCM
precursor HNO3 Mass Sol
ID ref. name % (mL) (h) (mL) rpm %
(mg) (mL)
Sol. % mg
blend

IP13/20
NP5 βNCPCD27 Y 100 2000 1 700 1301 N% 6.2
blend

IP13/20
NP6 βNCPCD28 Y 100 2000 14 1 700 1755 N% 6.6
blend

NP34 βNCPCD29 IP21 Y 100 2000 14 1 700 1562

IP18/19
NP17 βNCPCD30 Y 100 2500X4 17.5X4 1 700 9000
blend

NP2 βNCPCD31 IP23 Y 100 2500 17.5 1 700 2760 N% 6.8

NP4 βNCPCD32 IP17 Y 100 2100 14.7 1 700 1556 N% 7

NP3 βNCPCD33 IP25 Y 100 2200 15 1 700 2294 N% 8.7

NP1 βNCPCD34 IP18 Y 100 6000 28 1 700 4764 62

204
1 Where yield is not reported, masses are less than required to determine the nitrogen content, thus were not calculated.

Table A 7 Nitration of βCHCD precursors.

Stirr Yield
Sample Logbook Mass of Volume 100% HNO3 N
Precursor name Solubility precursor Time (h)
ID ref. (mg) (mL) Mass (%)
rpm (%)
mg

NH1 βNCHCD1 IH1 Y 2000 14 1 700 1683 60 8.2


NH2 βNCHCD2 IH2 Y 2060 14 1 700 1897 71 7.1
NH3 βNCHCD3 IH3 Y 2000 14 1 700 1798 62 9.0
NH4 βNCHCD4 IH4 Y 2000 14 1 700 1939 66 7.4

Table A 8 Nitration of soluble βCTCD precursors.

1
H2SO4 Stirr Yield N
Mass of Volume
Sample Logbook Precursor HNO3 Volume Time DCM (%)
precursor H2SO4 Mass
ID ref. name HNO3 (mL) % (h) (mL) rpm (%)
(mg)
Sol. % (mL) mg

NT5 βNCTCD1 NT5 N 95 200 1 0 0 1 0 200 170

NT6 βNCTCD2 NT5 N 100 200 1 0 0 1 0 200 112

NT7 βNCTCD3 NT5 N 100 200 0.5 100 0.5 1 0 200 146

NT8 βNCTCD4 NT5 N 100 200 0.5 100 0.5 0.5 1.8 400 148

NT9 βNCTCD5 NT5 N 100 200 0.5 100 0.5 1 0 900 103

205
NT10 βNCTCD6 NT7 N 100 200 0.5 100 0.5 1 0 900 157
NT11 βNCTCD7 NT6 N 100 200 0.5 100 0.5 1 0 900
NT12 βNCTCD8 NT5 N 100 200 0.5 100 0.5 1 1.8 900 126
NT13 βNCTCD9 NT6 N 100 200 0.5 100 0.5 1 1.8 900 110

NT14 βNCTCD10 NT7 N 100 200 0.5 100 0.5 1 1.8 900 196

NT1 βNCTCD11 NT10 Y 100 1000 7 0 0 1 0 700 1038 74 8.2

NT2 βNCTCD12 NT11 Y 100 1000 7 0 0 1 0 700 886 60 9.4

NT3 βNCTCD13 NT12 Y 100 1000 7 0 0 1 0 700 884 60 9.3

NT4 βNCTCD14 NT13 Y 100 1000 7 0 0 1 0 700 870 68 6.2

1Where yield is not reported, masses are less than required to determine the nitrogen content, thus were not calculated.

206
Figure A 1 1H-NMR spectrum of PEGDGE in DMSO-d6.

207
Figure A 2 1H-NMR spectrum of HEGDGE in DMSO-d6.

208
Figure A 3 1H-NMR spectrum of TEGDGE in DMSO-d6.

209
Figure A 4 DMA showing variation in the storage modulus (E′) and tanδ of βCD at 1, 5
and 10 Hz (10 °C min-1, third temperature cycle from –100 to 140 °C, aluminium
pocket).

210
Figure A 5 DMA showing variation in the storage modulus (E′) of the aluminium
support (10 °C min-1, 1 Hz, third temperature cycle from –100 to 140 °C, aluminium
pocket).

Figure A 6 DSC showing cooling of IP13 from –100 to 100 °C (10 °C min-1, third temperature
cycle, aluminium crucible).

211
Figure A 7 DMA showing variation in the storage modulus (E′) of the aluminium
support and PTFE (10 °C min-1, 1 Hz, third temperature cycle from –100 to 140 °C,
aluminium pocket).

Figure A 8 Optical microscope images of IP9 (PEG:βCD ratio = 3.5) placed on PTFE
tape. The sample was annealed at 140 °C for 5 min, top row (a-d), to remove any
physical defects (a-b, black shades). The arrow indicates the disappearance of the
black shades. The bottom row (e-h) shows the cooling phase captured at –3, –55, –69
and –90 °C and confirms the absence of cracking. The dark shapes on the bottom left
of the picture represent condensed water in the cooling chamber.

212
Figure A 9 FTIR spectra of the βNCPCD samples NP1–NP4 containing PEG spacer
units.

213
Figure A 10 1H NMR spectra of (a) NP1 + 1% w/w DPA at time 0, and (b) NP1 + 1%
w/w DPA at time 3 d at 80 °C in DMSO-d6.

214
Figure A 11 1H NMR spectra of (a) NP1 + 1% w/w 2,4-NDPA at time 0, and (b) NP1 +
1% w/w 2,4-NDPA at time 3 d at 80 °C in DMSO-d6.

215
Figure A 12 1H NMR spectra of (a) NP1 + 1% w/w Centralite at time 0 and (b) NP1 +
1% w/w Centralite a time 3 d at 80 °C in DMSO-d6.

A.1 Preliminary curing tests of inert βCXCDs and energetic


βNCXCDs
Initial tests were performed on inert βCPCD and energetic samples βNCPCD to
investigate further cross-linking in the presence of TEGDGE. The curing tests at
70 °C were designed to check the possibility that cured products could be used
during the formulation process.

Preliminary curing tests using inert IP18 and nitrated NP1 samples (PEG:βCD
ratio = 4.1) were performed with TEGDGE as the curing agent at a
βCPCD:TEGDGE ratio of 1:0.2 (Figure A 13). The curing mixtures were
prepared by dissolving IP18 or NP1 and the cross-linker in acetone. The
samples were stirred for 10 min, the acetone was removed under high vacuum
at 40 °C and the samples were left to cure in sealed flasks in the oven at 70 °C
for 20 days. The analysis of the sample involved checking the solubility of a

216
small fraction of the curing samples and analysing the consumption of the
cross-linker by 1H-NMR spectroscopy.

Scheme A 1 Schematic representation of the curing of βCPCD and βNCPCD


binders using TEGDGE as the cross-linker.

A visible change in the colour of the mixture was observed after 2 days at 70 °C
(Figure A.13b) which was attributed to the ageing of the compounds in the
oven. The IP18 sample became partially insoluble in water after 3 days.
Characterisation by 1H-NMR revealed that the cross-linker was consumed
and/or degraded under curing conditions, suggesting that cross-linking had
occurred.

Figure A 13 (a) IP18 before curing and (b) after curing for 2 days at 70 °C.

A colour change was also observed after curing NP1 for 1 day at 70 °C (Figure
A.14b,c). This was attributed primarily to the degradation of the nitrate esters, in

217
addition to the ageing of the compound in the oven. The NP1 sample was still
soluble in acetone after 3 days. Characterisation by 1H-NMR revealed that the
cross-linker was consumed and/or degraded under curing conditions,
suggesting that no cross-linking occurred.

Additional curing tests for the βCXCDs and βNCXCDs are necessary,
potentially including the use of curing accelerators.

Figure A 14 Comparison of (a) NP1 and the NP1/TEGDGE curing mixture at (b) the
mixing time, and (c) after 20 days at 70 °C.

218
Full Paper
1 DOI: 10.1002/prep.201800137
2
3
4
Nitrated Cross-linked b-Cyclodextrin Binders Exhibiting
5
6
Low Glass Transition Temperatures
7
Federico Luppi+,[a] Hamish Cavaye+,[a] and Eleftheria Dossi+*[a]
8
9
10
11 Abstract: Polymeric binders such as b-cyclodextrins (bCDs) trated systems (bNCXCDs) were therefore synthesised using
12 are used to bind with other constituents of energetic for- a 1 : 1 (v/v) ratio of 98 % sulfuric acid/100 % nitric acid or
13 mulations and to prevent accidental ignition. One of the 100 % fuming nitric acid, increasing their solubility in ace-
14 advantages of bCDs is the ability to tune their properties by tone and tetrahydrofuran. The nitrated derivatives were
15 chemical modification. Here, we synthesised nitrated cross- characterised by decomposition temperatures (200 8C) and
16 linked bCDs (bNCXCDs) to produce new binders for en- energies (up to 1750 J g 1) comparable to nitrocellulose.
17 ergetic formulations. The cross-linking of bCD with non-tox- Moreover, the glass transition of the inert bCXCDs at low
18 ic triethylene glycol diglycidyl ether (TEGDGE, X=T) and temperatures (< 0 8C) was conserved in the corresponding
19 poly(ethylene glycol) diglycidyl ethers (PEGDGE, X=P) yield- nitrated bNCXCDs, ensuring the desensitisation of energetic
20 ed soft, water soluble oligomeric compounds (bCXCDs) compositions even at low temperatures. This is the first
21 which can improve the processability of energetic for- time that nitrated derivatives of bCD with glass transition
22 mulations and contribute to their desensitisation. When the temperatures below 0 8C have been reported, suggesting
23 PEGDGE cross-linker was used, lower glass transition tem- such derivatives could make suitable replacements for ni-
24 peratures were achieved, which extended the operative trocellulose and other binders in energetic formulations.
25 range of the bCPCD binder to 20 8C. The analogous ni-
26
Keywords: cross-linked b-cyclodextrin · diglycidyl ethers · energetic binder · low Tg
27
28
29
30 1 Introduction The inclusion properties of bCD and gCD binders can
31 lead to stabilisation of energetic molecules because these
32 Synthetic and semi-synthetic polymeric binders are im- molecules have larger cavities than aCDs [3–5]. Fur-
33 portant components of most energetic formulations [1]. The thermore, nitrated cyclodextrins (NCDs) have been devel-
34 role of the binder is to improve the overall mechanical oped as energetic components, although fully-nitrated cy-
35 properties of the formulation and to coat the energetic clodextrins are sensitive to electrostatic discharge (ESD),
36 molecules, thus shielding them from accidental stimuli and e. g. bCD nitrated to 85 % is sensitive to ESD ignition at
37 environmental degradation. Nitrocellulose, produced by the 0.0125 J [6]. To reduce the ESD hazard, pre-functionalisation
38 nitration of natural cellulose, is the most frequently used before nitration can be achieved using inert molecules as
39 semi-synthetic polymer in energetic formulations [1]. cross-linkers [7, 8]. NCDs initially cross-linked with epichlor-
40 During the past three decades, cyclodextrins (CDs) have ohydrin in NaOH, polyallylamine in KOH or isocyanate in
41 been introduced as energetic binders in a small number of DMSO, are less sensitive to ESD ignition at 0.1288 J [8],
42 studies because they are natural molecules with a composi- which is a 10-fold reduction in sensitivity. In contrast, longer
43 tion similar to cellulose and they have useful molecular in- inert cross-linkers such as poly(ethylene glycol) (PEG) have
44 clusion properties [1, 2]. Typical CDs are cyclic compounds been used in CD systems for pharmaceutical applications,
45 containing six, seven or eight sugar rings (a, b and gCDs, resulting in the formation of insoluble gels [9, 10]. Water-
46 respectively) linked together via a-glycosidic bonds. They
47 are particularly useful in the food, cosmetic and pharma-
48 ceutical industries because they form complexes within [a] F. Luppi,+ H. Cavaye,+ E. Dossi+
49 their toroidal cavities with a wide range of molecules [2]. Centre for Defence Chemistry, Cranfield University, Defence Acad-
50 The physical properties of the CDs can be tuned to satisfy emy of United Kingdom, Shrivenham, SN6 8LA, UK
*e-mail: [email protected]
51 specific applications such as shielding other molecules from
[email protected]
52 external stimuli [2, 3], which is important in explosive appli- [email protected]
53 cations [4–5]. The availability of hydroxyl functional groups [+] The manuscript was written with contributions from all authors. All
54 allows the functionalisation of the macrocycles to further authors have approved the final version of the manuscript.
55 tune their physicochemical properties [2, 3]. Supporting information for this article is available on the WWW
56 under https://doi.org/10.1002/prep.201800137

Propellants Explos. Pyrotech. 2018, 43, 1023–1031 © 2018 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim 1023
Full Paper F. Luppi, H. Cavaye, E. Dossi

1 soluble CD polymers were obtained when acyl chloride PEG 2 Experimental Section
2 derivatives were used [11] but to the best of our knowledge
3 these derivatives have never been converted to nitrated de- 2.1 Materials
4 rivatives.
5 One of the most important properties of binders in en- The bCD ( 97 %, Sigma-Aldrich) was used from stock (TGA,
6 ergetic formulations is the glass transition temperature (Tg), 13 % water content). Polyethylene glycol diglycidyl ether
7 at which the rubbery binder becomes glassy and brittle. (PEGDGE 500 Mw, Sigma-Aldrich), sodium hydroxide (Fisher
8 PEG-based cross-linkers are favoured for their low glass Chemicals), acetylated dialysis membrane (2000 MWCO, Sig-
9 transition temperatures because this increases the operative ma-Aldrich), tetra-n-butylammonium bromide (TBAB, Sig-
10 temperature range within which the cross-linked materials ma-Aldrich), and triethylene glycol (TEG, Sigma-Aldrich)
11 can successfully contribute as a binder [12] and desensitise were obtained from commercial sources and used without
12 the energetic formulation. The replacement of highly toxic further purification. Triethylene glycol diglycidyl ether
13 cross-linkers with non-toxic materials is one of the require- (TEGDGE, 262.0 g/mol) was synthesised as previously de-
14 ments for new energetic formulations. The REACH regu- scribed [14]. The bCXCD cross-linking ratio was determined
15 lation of the European Union, has introduced restrictions on by 1H NMR.
16 the usage of certain isocyanates to improve the protection
17 of human health and the environment from the risks that
18 can be posed by chemicals [13, 14]. In this context, the use 2.2 Characterisation
19 of CDs and non-toxic ethylene glycol diglycidyl ethers en-
20 sures relatively mild and environmentally sustainable cross- NMR spectra were recorded on a Bruker Ascend (400 MHz)
21 linking conditions: CDs are obtained from natural and sus- with a BBFO probe in deuterated dimethyl sulfoxide
22 tainable sources, water is used as the reaction solvent, and (DMSO-d6) solution using tetramethylsilane (TMS) as an in-
23 the reaction temperatures are kept low. The formation of ternal reference. Spectra were also recorded in deuterated
24 stable ether linkages (C O C) between CDs and the digly- water (D2O) with 3-(trimethylsilyl)-1-propane-sulfonic acid
25 cidyl ethers also supports further functionalisation such as sodium salt as an internal reference. Peak multiplicities were
26 the nitration of the cross-linked cyclodextrins (CXCDs). Sev- described as follows: singlet (s), doublet (d), triplet (t), mul-
27 eral nitration methods for pure or cross-linked CDs have tiplet (m), doublet of doublet (dd), and broad (br). Thermal
28 been reported, ranging from pure nitric acid or nitric/sulfu- properties were determined by DSC using a Mettler Toledo
29 ric acid mixtures to stronger and more advanced systems DSC822 or DSC30 with heating and cooling rates of
30 such as nitration in liquid CO2 with nitrogen pentoxide 10 8C min 1 and a flow of dry nitrogen. The reported values
31 [8, 15]. The latest methods achieve full nitration whereas were the measurements performed on samples with a
32 earlier methods resulted in up to 90 % nitration of the avail- mean weight of 10 mg for inert materials and low temper-
33 able hydroxyl groups [15]. The complexation of explosive ature, whereas 1 mg was used to assess the decomposition
34 molecules in the CD cavities also reduces the sensitivity of temperature of energetic materials. Yields were measured
35 the explosive to external stimuli [8]. as mass of products over mass of reactants. Gel Permeation
36 Here we report the synthesis of a bCXCD system and its Chromatography (GPC) measurements were performed in
37 nitrated derivative bNCXCD using TEGDGE and PEGDGE tetrahydrofuran (THF) at 35 8C, using Agilent PLgel 10 mm
38 cross-linkers. The two cross-linkers were chosen as soft seg- mixed B columns and Agilent polystyrene calibration kit (Mw
39 ments for their low glass transition temperatures of 80 500–6.9 3 106).
40 and 68 8C, respectively [16, 17]. We selected bCD for the
41 initial tests because it is less expensive than gCD. However,
42 the gCD cavity would encapsulate larger energetic mole- 2.3 Synthesis of bCXCDs
43 cules and more tests using gCD will be published in the fu-
44 ture. All bCXCDs were subsequently nitrated to determine The bCD (5–5.6 g, 3.8–4.4 mmol) was dissolved in 5.6–50 %
45 the impact of the bNCXCD systems on degradation en- w/w NaOH (21.0 mL) and stirred mechanically for 0–16 h.
46 ergies and the glass transition temperature. We also carried Diepoxide (TEGDGE n = 3 or PEGDGE n = 9) (17.4–5.8 mL,
47 out preliminary studies to determine the degree of compat- 13.2–36.9 mmol) was then added dropwise in 20 min with
48 ibility between the inert and nitrated products and en- vigorous stirring. The reaction mixture was heated to 30–
49 ergetic fillers. 70 8C for 30 min with vigorous stirring. After cooling for
50 20 min, the mixture was neutralised with 6 M HCl. The puri-
51 fication method depended on the solubility of the product
52 in water. The volume of solvent for soluble products was re-
53 duced and the crude solute was precipitated in acetone
54 three times. The solid was then collected and dissolved in
55 distilled water and dialysed against water using a cellulose
56 membrane (2000 MWCO) for 5 days. The insoluble products

1024 www.pep.wiley-vch.de © 2018 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim Propellants Explos. Pyrotech. 2018, 43, 1023–1031
Nitrated Cross-linked b-Cyclodextrin Binders Exhibiting Low Glass Transition Temperatures

1 were filtered from the crude reaction mixture and washed react at room temperature with the epoxides. Heating the
2 with acetone. The insoluble solid was then collected and reaction mixture (up to 70 8C) under variable basic con-
3 suspended in distilled water and dialysed against water us- ditions (5.6 % w/w, 40 % w/w and 50 % w/w) led to the par-
4 ing a cellulose membrane (2000 MWCO) for 5 days. The di- tial substitution of all three hydroxyl groups attached to C-
5 alysis water was replaced every day. The dialysed solid was 2, C-3 and C-6 atoms of bCD units in bCXCDs as from 1H-
6 collected and the water was evaporated under reduced NMR characterisation in dimethylsulfoxide (DMSO); this is
7 pressure. The final product was characterised by 1H NMR discussed later in this paper. As a general rule for the re-
8 and DSC. 1H NMR (400 MHz, DMSO-d6, ppm): d = 5.9–5.6 (br activity of bCD hydroxyl groups, the primary hydroxyl
9 m, OH C2, OH C3), 5.1–4.8 (m, H-1), 4.7-4.5 (br m, OHA, OHB groups attached to C-6 atom (Figure 1), are considered the
10 and OHC), 4.4 (br m, OH C6), 4.0-3.2 (br m, most reactive [18–20]. The difference of reactivity of the pri-
11 bCD OCH2 (CH OH) CH2 O CH2 CH2). DSC (10 8C min 1, mary and secondary hydroxyl groups of bCD cannot be as-
12 N2) 240–250 8C (dec) (See Supporting Information). sessed by the authors in these earlier studies due to the
13 overlap of the hydroxyl proton signals in the NMR analyses.
14 The number of unreacted hydroxyl groups attached to C-2
15 2.4 Synthesis of bNCXCDs and C-3 atoms was evaluated by 1H-NMR spectroscopy in
16 dimethylsulfoxide (DMSO) and was found to depend on the
17 Nitric acid (95–100 % 0.5–1 mL) or 100 % nitric acid/100 % concentration of NaOH. The least concentrated NaOH sol-
18 sulfuric acid mixture (50 : 50 v/v, 1 mL) was poured into a ution (5.6 % w/w) promoted more cross-linkages whereas
19 round-bottomed flask and cooled to below 10 8C in ice wa- stronger NaOH solutions (40–50 % w/w) inhibited the for-
20 ter. Dichloromethane (DCM) (1.8 mL) was added when route mation of the cross-linked product (Table 1) because of i)
21 2 reported later in the paper (Figure 2) was followed. The the higher amount of degraded cross-linker by the reaction
22 bath was removed and bCXCD (X = P and T, 200.0 mg) was conditions discussed in the following paragraph and ii) the
23 added in small fractions over 5 min, ensuring that the tem- supressed formation of bCD-alkoxide in the very viscous re-
24 perature remained below 10 8C. The crude slurry or solution action mixtures.
25 was then left stirring at room temperature for 1 h. The re- Under the basic cross-linking conditions we used, a
26 action mixture was poured into ice/water (10 mL) and the competitive side reaction occurs and the degradation of the
27 solid was dissolved in acetone (5 mL) and precipitated in diepoxide cross-linker produces a tetra-hydroxyl by-prod-
28 water (100 mL). The clean product was collected, dissolved
29 in acetone and dried under vacuum. Small portions neces-
30 sary for characterisation were taken and the rest was stored
31 under DCM. The 1H NMR analysis was performed in ace-
32 tone-d6 and DMSO-d6. DSC (10 8C min 1, N2) 197–210 8C
33 (dec) (See Supporting Information).
34
35
36 3 Results and Discussion
37
38 3.1 Synthesis of bCXCDs
39
40 Several bCXCDs were synthesised at 70 8C to determine the
41 effect of different parameters on the yield of the reaction
42 and the physicochemical properties of the products. The
43 cross-linked compounds were recovered after cooling the Figure 1. Proposed chemical structure of bCD, bCXCD and bNCXCD
(X=T,P) with numbered H and C atoms.
44 reaction mixture followed by neutralisation with 6 M HCl.
45 The final products were obtained by precipitation from wa-
46 ter in acetone followed by dialysis against water using a cel- Table 1. Effect of NaOH concentration on the properties of
47 lulose membrane with a molecular weight cut-off (MWCO) bCPCDs.
48 of 2000 for 5 days.
49 The cross-linking reaction involved the alkoxidation of Sample1 NaOH2 Reaction Yield3 Tg Water
50 bCD combined with diglycidyl ethers of two different (% w/w) time (%) (8C) solubility
(h)
51 lengths: PEGDGE with n = 9 repeating ethylene glycol units
52 and TEGDGE with n = 3 ethylene glycol units. The hydroxide bCPCD1 50.0 1.00 <1 – Y
53 (OH ) ions from the basic medium (NaOH) reacted first with bCPCD2 40.0 1.00 <1 – Y
54 the hydroxyl groups of the bCDs to form alkoxides, and the bCPCD3 5.6 0.66 88 30 N
55 latter then reacted with the epoxide ring of the diglycidyl 1
PEGDGE:bCD ratio = 9 : 1, Reaction temperature = 70 8C, 3 Yield
2

56 ether molecule. The hydroxyl groups of the bCDs would not measured as mass of products/mass of reactants.

Propellants Explos. Pyrotech. 2018, 43, 1023–1031 © 2018 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim www.pep.wiley-vch.de 1025
Full Paper F. Luppi, H. Cavaye, E. Dossi

1 uct, thus affecting the yield of the reaction and physical 70 8C reduce the yield due to the loss of cross-linker in the
2 properties of the final product. The effect of the NaOH con- competitive degradation reaction.
3 centration was investigated at 70 8C for short reaction times The effect of the cross-linker:bCD ratio was investigated
4 ( 1 h) (Table 1) and this had a significant influence on the under the optimal conditions determined thus far, i. e. 5.6 %
5 reaction. When the NaOH concentration was low, the re- w/w NaOH, 30 8C, 16 h for the formation of the alkoxide and
6 action become more efficient and reached the gelation 5 h for the completed cross-linking reaction (Table 3). The
7 point accompanied by the appearance of a solid product thermal properties of the cross-linked products at low tem-
8 and the increasing of the viscosity of the reaction mixture, peratures were affected, i. e. lower amounts of cross-linker
9 whereas strong NaOH solutions delayed the formation of resulted in products with higher degree of crystallinity and
10 gels and affected the yield of the reaction due to faster higher Tg. The ratio of cross-linker:bCD in the cross-linked
11 cross-linker degradation. product is lower than in the initial feed of the reaction. The
12 We also investigated the effect of the number of ethyl- difference is caused by the degradation of the cross-linker
13 ene glycol units in the cross-linker when the cross-link- as mentioned above and the NMR analysis of the com-
14 er:bCD ratio was maintained at 9 : 1, and 40 % w/w NaOH pounds is discussed in detail later, in the section dealing
15 was used for all the reactions. The trend of the reactions with characterisation.
16 suggested that TEGDGE produces gels more quickly than
17 PEGDGE. Given this result and the insoluble nature of the
Table 3. Effect of cross-linker : bCD ratio on the properties of
18 gels produced with TEGDGE, we decided to focus on the
bCPCDs.
19 longer cross-linker (PEGDGE), which was expected to confer
20 better mechanical properties upon the cross-linked prod- Sample1 Cross-link- Cross-linker : bCD Yield Tg Water
21 ucts. er : bCD ratio ratio by NMR (%) (8C) solubility
22 Another parameter that affects the cross-linking of bCD feed
23 is the time needed to prepare its alkoxide. Initial trials sug- bCPCD7 5:1 3.75 : 1 34 22 Y
24 gested that leaving the bCD stirring in basic conditions for bCPCD8 4:1 3:1 36 13 Y
25 16 h leads to better yields of soluble products (Table 2). It is bCPCD9 3:1 2.25 : 1 68 3 Y
26 therefore necessary to allow the cyclodextrin to form the al- bCPCD10 2:1 1.5 : 1 44 +8 Y
27 koxidic groups prior to the addition of the cross-linker as 1
5.6 % (w/w) NaOH, 30 8C, 16 h alkoxide formation, 5 h reaction
28 discussed below. Indeed, the alkoxide promoted successful time.
29 ether linkage formation between cyclodextrin and PEGDGE
30 (see the Supporting Information).
31 A final investigation demonstrated that the duration of
32 the addition of the cross-linker to the reaction mixture also
Table 2. Role of the temperature on the yield of bCPCDs.
33 influenced the reaction yield. If prolonged addition times
34 Sample T Yield Tg Water were used, a small proportion of cross-linker was lost in the
35 (8C) (%) (8C) solubility competing degradation reaction. Further studies are need-
36 ed to determine the optimal conditions to improve the
bCPCD41 70 1 – Y
37 bCPCD51 50 13 +6 Y yield of water-soluble samples ( 68 % at present). Higher
38 bCPCD62 50 30 18 Y yields were achieved only for insoluble gels. The cross-link-
39 bCPCD72 30 34 22 Y ing of bCPCDs therefore requires further optimisation. The
40 1 samples obtained in these early experiments were used for
Cross-linker : bCD = 5 : 1, 0 h alkoxide formation, 5 h reaction time,
41 2
Cross-linker : bCD = 5 : 1, 16 h alkoxide formation, 5 h reaction time. the nitration reactions described in the following section.
42 The images of bCPCD7 and its precursor bCD were com-
43 pared by scanning electron microscopy (SEM) and the cor-
44 Higher temperatures affect the kinetics of cross-linking responding images are provided in the Supporting In-
45 by increasing the rate of cross-linking and also by accelerat- formation. The images show the homogeneity of the
46 ing the degradation of the cross-linker. Two sets of trials surface of the bCPCDs, whereas the bCDs have a crystalline
47 were carried out using the following conditions: a low con- appearance.
48 centration of NaOH (5.6 % w/w), a cross-linker:bCD ratio of
49 5 : 1, 0 or 16 h for the formation of the alkoxide and 5 h for
50 the cross-linking reaction itself. The longer time for alkoxide 3.2 Synthesis of bNCXCDs
51 formation was set at 16 h based on the results discussed
52 above. The cross-linker:bCD ratio was set to 5 : 1 to prevent The bCXCDs were functionalised with nitro groups to form
53 gelation, which occurred at the higher ratio of 9 : 1. The data energetic derivatives that can contribute energy to ex-
54 summarised in Table 2 suggest that the cross-linker is more plosive formulations. The nitration of a set of water-soluble
55 stable at 30 8C, but higher temperatures such as 50 8C and bCPCDs and insoluble bCTCD precursors with variable phys-
56

1026 www.pep.wiley-vch.de © 2018 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim Propellants Explos. Pyrotech. 2018, 43, 1023–1031
Nitrated Cross-linked b-Cyclodextrin Binders Exhibiting Low Glass Transition Temperatures

1 ical properties was carried out using the reaction conditions raphy (GPC) in THF and compared with poly(styrene) stand-
2 shown in Figure 2. ards. The results Mn = 7350 Da and Mw = 15140 Da, con-
3 firmed the polymeric nature of the sample (see Supporting
4 Information).
5 Like the bCPCDs precursors, the Tg of the nitrated prod-
6 ucts was dependent on the number of soft ethylene glycol
7 units in the cross-linker. The Tg increased when less PEGDGE
8 was used (Table 4) and we speculate from very preliminary
9 modelling [23] investigation on nitrated CDs that the polar
10 nitro groups could inhibit the mobility of the polymeric
11 Figure 2. Overview of the synthesis of bNCXCDs. chains. On the other hand, the large number of OH to ONO2
12 (nitrato) transformations contributed to the overall energy
13 of the system. The decomposition temperature (Tdec) stabi-
14 In initial tests we evaluated the following nitration lised at 196 8C, which is similar to nitrocellulose samples
15 methods: (i) using 1 : 1 98 % sulfuric acid/100 % nitric acid; with a 12.5 % nitrogen content [24]. As expected, the en-
16 (ii) two-phase nitration, using 1 : 1 98 % sulfuric acid/100 % ergy of the system was also affected, with higher decom-
17 nitric acid followed by dichloromethane (DCM); and (iii) us- position energies (DHdec) such as 1750 J g 1 for sample
18 ing 100 % fuming nitric acid as the nitrating phase. DCM did bNCPCD3. The thermal properties of the bNCPCDs can be
19 not improve the control and/or efficiency of the nitration tuned to give the desired product for specific applications.
20 reaction because the bNCXCDs were insoluble in DCM and A very small, second-order transition was observed in all
21 were not extracted from the acidic water phase during the thermograms at ~ 80 8C, which was attributed to the glass
22 reaction. Therefore, DCM was not included in subsequent transition of the rigid bCD units in the cross-linked systems.
23 experiments. Attempts to nitrate insoluble gel compounds When TEGDGE was used as the cross-linker, both the
24 also achieved no control of the nitration reaction, and thus bCTCD precursors and the bNCTCDs derivatives were in-
25 the physicochemical and hazard properties of the products soluble gels. The bNCTCDs were characterised by low Tg val-
26 were unpredictable, so these precursors were also aban- ues, thermal stability and relatively low decomposition en-
27 doned. ergies compared to the analogous bNCPCDs (Table 4).
28 The bCXCDs were difficult to dissolve in the mixed acids Although the cross-linker:bCD ratio was 9 : 1 to guarantee
29 and their conversion to analogous nitrated products was low transition temperatures, the degree of nitration was
30 compromised. The sulfate esters [21, 22] of bCXCDs formed lower in these materials due to the inability of the acids to
31 in these nitration conditions are expected to affect their sol- penetrate the polymer structure and efficiently nitrate the
32 ubility and consequently the efficiency of the nitration re- hydroxyl groups of the bCD units. The nitration of the same
33 action. In contrast, the bCXCDs dissolved efficiently in the bCTCD resulted in the synthesis of bNCTCDs with variable
34 fuming nitric acid within the 1 h duration of the nitration properties. This highlights the lack of control over the nitra-
35 reaction. Table 4 summarises the data obtained when 100 % tion of insoluble products. Also, it was difficult to purify the
36 fuming nitric acid was used as the nitrating agent. All insoluble bNCTCDs from the acid traces. The stability of
37 bNCPCDs were soluble in organic solvents such as acetone these products and their hazard properties is strongly af-
38 and THF, suggesting good processability, which is important fected by the efficiency of the purification process.
39 in energetic formulations because the different components
40 must be mixed thoroughly. The bNCPCDs were purified by
41 re-precipitation from acetone in water and their thermal 3.3 Chemical Characterisation
42 stability was satisfactory, as shown by their high decom-
43 position energy and negative Tg values. A nitrated sample The cross-linked products we synthesised were charac-
44 bNCPCD1 was examined by gel permeation chromatog- terised by 1H NMR spectroscopy in deuterated dimethylsulf-
45
46
47 Table 4. Thermal properties of bNCXCD oligomers.
48 Nitrated Precursor Reaction Tg Tdec DHdec
49 sample1 Name cross-linker : bCD time (8C) (8C) (J g 1)
50 (h)
51 bNCPCD1 bCPCD7 5:1 1.0 20 202 1500
52 bNCPCD2 bCPCD8 4:1 1.0 142 197 1640
53 bNCPCD3 bCPCD9 3:1 1.0 19 199 1750
54 bNCTCD1 bCTCD1 9:1 1.0 21 210 530
bNCTCD2 bCTCD1 9:1 1.0 20 208 650
55
1
56 Using 200 mg of precursor and 1 mL 100 % HNO3, 2 Very broad transition; more investigation needed.

Propellants Explos. Pyrotech. 2018, 43, 1023–1031 © 2018 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim www.pep.wiley-vch.de 1027
Full Paper F. Luppi, H. Cavaye, E. Dossi

1 oxide (DMSO-d6) and deuterated water (D2O) as solvents. * Integral 2 at 3.2–4.0 ppm = 6nbCD H-2 to H-6 protons +
2 DMSO-d6 was used to investigate the reactivity of the hy- 44ncross-linker protons
3 droxyl groups of the bCDs at 4.4–5.5 ppm (Figure 3 and 4). The results show that the cross-linker is not consumed
4 in the competing degradation reaction at 30 8C, allowing
5 the reactants to react in a quasi-stoichiometric manner
6 (25 % lower than the anticipated theoretical value, Table 3).
7 The bNCTCDs and bNCPCDs were characterised by 1H
8 NMR in acetone-d6 and DMSO-d6. The spectra recorded in
9 DMSO-d6 were compared with those of the starting oligom-
10 ers and with a sample of nitrated bCD (bNCD) synthesised
11 under the same conditions [15]. Figure 5 shows the 1H NMR
12 spectra of bNCD and bNCXCD in DMSO-d6.
13
14
15
16 Figure 3. 1H NMR spectrum of bCD and bCPCD8 in DMSO-d6, sug-
gesting the assignment of the proton of the OR groups when no
17
substitution occurs. The OHC assignment is expected when R1 = H.
18
19
20
21
22
23
24
25
26 Figure 5. 1H NMR spectra of bNCD (top) and bNCPCD1 (bottom) in
27 DMSO-d6. In the bNCPCD1 spectrum (bottom), the signals near the
water peak are assigned to protons that are not near the nitrato
28
groups.
29
30
31
32 The bNCPCD1 signals were broader due to the poly-
33 meric nature of this product. The broadness of the peaks
34 and the down field shift of the signals compared to bCPCD
35 reflect the complex chemical environment present in this
36 energetic polymer. The different degrees of nitration on the
37 cyclodextrin units result in overlapping signals representing
38 Figure 4. 1H-NMR spectra of bCD (top) and bCPCD8 (bottom) in protons close to the nitrato groups. The reference peak of
39 D2O. proton H-1 close to the glycoside bond is present at
40 5.60 ppm. The broad signals at 5.55–5.10 ppm can be as-
41 signed to protons H-2 and H-3 attached to asymmetric car-
42 The spectrum of the bCPCDs shows broadened peaks bon atoms 2 and 3, whereas the signals at 5.10–4.60 ppm
43 due to the larger cross-linked molecules, when compared to can be assigned to CH2, which is adjacent to nitrato groups
44 the spectrum of pure bCD [2]. New peaks appear in the in both bCD and the cross-linker units. Finally, the signals at
45 spectrum (Figure 4) and are attributed to the three hydroxyl 4.6–4.0 ppm can be assigned to CH2 protons belonging to
46 groups (OHA, OHB and OHC when RC = H) on the cross-link- both bCD and the cross-linker units that are not adjacent to
47 ing units. Due to the similar environments in the com- a nitrato group, but close enough to be influenced by
48 pound, many signals overlap. The water peak which over- them. The signals between 3.8 ppm and 3.2 ppm can be as-
49 laps the signals of the product at 3.3–3.4 ppm in DMSO-d6 signed to the methylene of the ethylene glycol units which
50 moves in the hydrogen-deuterium oxide (HDO) when D2O is are farthest from and least affected by the nitrato groups.
51 used, allowing us to determine the cross-linking ratio as re- The degree of nitration in bNCD was determined using a
52 ported below. previously described analytical method [15] and was nearly
53 The cross-linker:bCD ratio in bCPCDs was calculated [11] 90 %. The degree of nitration in the bNCPCDs will be as-
54 from the 1H NMR in D2O (Figure 4). sessed in future studies based on iron sulfate titration and
* Integral 1 at 4.8–5.1 ppm = n
55 bCD H-1 protons comparison with ion chromatography data.
56

1028 www.pep.wiley-vch.de © 2018 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim Propellants Explos. Pyrotech. 2018, 43, 1023–1031
Nitrated Cross-linked b-Cyclodextrin Binders Exhibiting Low Glass Transition Temperatures

1 3.4 Thermal Characterisation


2
3 The Tg of the bCXCDs and bNCXCDs was determined by dif-
4 ferential scanning calorimetry (DSC) within the temperature
5 range 100 8C to 100 8C at 10 8C min 1 and the results are
6 reported in Tables 1–4. The Tg of the bCXCDs ranged be-
7 tween 22 8C and + 8 8C and increased in line with the pro-
8 portion of bCD because the cross-linker introduces more
9 mobility into the system. The glass transitions reported for a
10 similar cross-linked system based on a carboxylic acid PEG
11 linker ranged from 20 8C to 16 8C [11]. These values are Figure 7. DSC thermogram of the decomposition of bNCPCD1.
12 consistent with the properties gained by the cross-linked
13 system studied herein.
14 The low Tg of the bCXCD precursors was transferred to The attempted nitration of insoluble bCXCD products
15 the bNCXCD derivatives and ranged from 20 8C to + 19 8C also yielded compounds with a decomposition event at
16 (Table 4). This transition is particularly wide compared to the ~ 200 8C, but the decomposition energy was 500–800 J g 1.
17 non-nitrated precursor (Figure 6) and the material begins to This reduction in energy probably reflects the presence of
18 soften at very low temperatures. The Tg midpoint is difficult fewer nitrato groups after nitration given the less accessi-
19 to determine in the nitrated product due to the broadness bility to the hydroxyl groups of bCXCDs and a higher pro-
20 of the transition, which may reflect the sum of different ar- portion of cross-linker in the system.
21 rangements of the entangled cross-linker chains in the com-
22 pound. Dynamic mechanical analysis (DMA) of bCXCD and
23 bNCXCD samples is currently underway and the results will 3.5 Compatibility Assessments
24 be published in a separate article.
25 Initial compatibility tests based on DSC were carried out to
26 determine whether contact between the bNCPCDs and en-
27 ergetic ingredients in a formulation could lead to un-
28 desirable or unexpected hazards. Sample bNCPCD1 was
29 mixed with energetic fillers such as oxidisers, pyrotechnics
30 and high explosives, and preliminary DSC compatibility
31 tests were carried out according to STANAG 4147 Test 4
32 [25]. Any chemical interaction between the ingredients
33 should lead to a change in the decomposition profile of the
34 formulation. The thermal decomposition of the single in-
35 gredients and their mixture was compared (Table 5). The
36 Figure 6. DSC thermogram of the glass transition of bCPCD8 (solid compounds were mixed in a 1 : 1 w/w ratio (total amount
37 line) and bNCPCD1 (dashed line) between 100 8C and 100 8C. 1.0 mg) and heated from 30 to 500 8C at a rate of 2 8C min 1.
38 Sample bNCPCD1 was tested with oxidisers such as am-
39 monium dinitramide (ADN), potassium chlorate (KClO3), am-
40 All bNCPCDs derived from water-soluble bCPCD pre- monium perchlorate (NH4ClO4) and ammonium nitrate
41 cursors showed similar thermal stabilities, with decom-
42 position occurring at ~ 200 8C. This is comparable with the
43 decomposition of nitrocellulose [24]. The decomposition Table 5. Summary of compatibility tests using bNCPCD1.
44 energy fell within the range 1500–1750 J g 1 (Figure 7).
Energetic Mixture
45 The energy released by the bNCXCDs was similar to that Name Tdeg Tdeg DT Change in shape
46 released by nitrocellulose samples with a nitrogen content (8C) (8C) (8C)
47 of 12.5 %, measured using the same method [24]. As ex-
ADN 174 161 13 Significant
48 pected, cross-linking affected the maximum nitrogen con-
KNO31 – – – Minor
49 tent of the bNCXCDs, which declined with the increasing NH4ClO4 300 242 58 Significant
50 content of inert ethylene glycol chains. DSC revealed that NH4NO31 – – – Significant
51 the bNCXCD systems displayed lower decomposition en- KClO31 – – – –
52 ergies than the bNCD sample with 90 % nitration Red Phosphorous 400 404 +4 Minor
53 (1880 J g 1). Micro-calorimetry measurements will be per- PETN 185 187 2 None
54 formed in the future to measure the decomposition energy HMX 279 279 0 Minor
RDX 225 210 15 Minor
55 of the bNCXCDs and the effective release of energy from
1
56 these systems. No degradation observed.

Propellants Explos. Pyrotech. 2018, 43, 1023–1031 © 2018 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim www.pep.wiley-vch.de 1029
Full Paper F. Luppi, H. Cavaye, E. Dossi

1 (NH4NO3). No changes in either the decomposition temper- 4 Conclusions


2 ature (Tdec) or the curve shape were observed for mixture
3 bNCPCD1/KClO3 compared to the thermogram of the pure The synthesis of bCXCD systems from bCD and diglycidyl
4 oxidiser (Table 5). We found that bNCPCD1 is not compat- ethers yielded cross-linked insoluble or water-soluble de-
5 ible with NH4ClO4 or NH4NO3 and that ADN shifted the Tdec rivatives with low Tg values down to 30 8C, the first such
6 by 13 8C and changed the shape of the curve. Further vac- observation for this type of product. The physicochemical
7 uum stability tests are needed to confirm compatibility with properties of the products were affected by several reaction
8 ADN. Sample bNCPCD1 showed good compatibility with parameters, including the temperature, cross-linker:bCD ra-
9 pyrotechnics such as red phosphorous. All thermograms are tio, concentration of NaOH, time allowed for the formation
10 provided in the Supporting Information. of bCD alkoxide, and the duration of reaction with the
11 Sample bNCPCD1 also showed good compatibility with cross-linker. The cross-linking reaction parameters were
12 symmetric nitro-esters such as pentaerythritol tetranitrate tuned to obtain soft, soluble precursors to make the sub-
13 (Figure 8) suggesting that bNCPCDs could potentially be sequent nitration reaction safer and more consistent. The
14 combined with nitroglycerine in double-based and triple- polymers were water soluble if prepared with a PEGDG-
15 based propellants [26]. E:bCD ratio of up to 5 : 1 at 30 8C. The Tg of the cross-linked
16 materials was primarily affected by the quantity and length
17 of PEG spacer present in the system.
18 The bNCPCDs retained some of the mechanical proper-
19 ties of the precursor systems. The Tg was maintained below
20 0 8C after the nitration of bCPCD derivatives made of high
21 bCD:cross-linker ratio. The presence of polar nitrato groups
22 on the cross-linked molecules increased the packing density
23 of the molecules and thus reduced their freedom, increas-
24 ing the Tg. Thermal analysis revealed that the nitrated prod-
25 ucts soften from 60 8C in a linear manner. The thermal sta-
26 bility and energy released by the bNCPCDs is similar to that
27 Figure 8. Thermograms of pentaerythritol tetranitrate (dotted line), observed for nitrocellulose samples with a nitrogen content
28 bNCPCD1 (dashed line) and the mixture (solid line) at 2 8C min 1. of 12.5–13.5 % making them promising nitrocellulose sub-
29 stitutes in energetic compositions. The further processing of
30 bNCXCDs using the same purification methods applied to
31 Nitramines such as RDX and HMX were also tested for nitrocellulose would improve the thermal stability even
32 compatibility with bNCPCD1. Although the Tdeg of RDX was more.
33 15 8C lower when mixed with bNCPCD1, the degradation of Initial compatibility tests indicated that bNCPCDs may
34 HMX was not affected (Figure 9). be suitable binders in formulation with selected energetics.
35 The reaction will need to be scaled up for proper compati-
36 bility and Energetic Materials Testing and Assessment Policy
37 (EMTAP) tests to determine the sensitivity of the new ni-
38 trated cross-linked compounds to ESD.
39
40
41 5 Recommendations/Future work
42
43 The assessment of the properties of the new nitrated cross-
44 linked b-cyclodextrins developed in this work is in progress.
45 Investigation of the change of properties with time and
46 Figure 9. Thermograms of HMX (dotted line), bNCPCD1 (dashed overall ageing of the nitrated binder as well as their com-
47 line) and the mixture (solid line) at 2 8C min 1. patibility with stabilisers/plasticisers is recommended. The
48 cross-linking reaction of bCD with TEGDGE and other ethyl-
49 ene glycol diglycidyl ethers is currently investigated.
50 The small amount of HMX used in the test underwent
51 degassing within the degradation temperature range of
52 bNCPCD1. The phenomenon was detected by DSC as an en- Supporting Information
53 dothermic peak at 210 8C. The measurement was repeated
54 several times and the endothermic peak shifted at different The following files are available free of charge.
55 temperatures every time, supporting the degassing hypoth-
56 esis.

1030 www.pep.wiley-vch.de © 2018 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim Propellants Explos. Pyrotech. 2018, 43, 1023–1031
Nitrated Cross-linked b-Cyclodextrin Binders Exhibiting Low Glass Transition Temperatures

1 Acknowledgements tion, Evaluation, Authorisation and Restriction of Chemicals


2 (REACH), establishing a European Chemicals Agency, 2006.
3 This project was funded through the Weapons Science and Tech- [14] E. Dossi, J. Akhavan, S. E. Gaulter, R. G. Williams, W. J. Doe,
nology Centre (WSTC) by the Defence Science and Technology Lab- Cross-linking of Hydroxyl-terminated Polyols with Triethylene-
4
oratory (Dstl). The authors would like to thank Roxel UK for kindly glycol Diglycidyl Ether: An Alternative to Toxic Isocyanates, Pro-
5 providing the ADN sample and BAES Land for assistance in the ex- pellants Explos. Pyrotech. 2018, 43, 241–250.
6 ploitation of this work. [15] L. B. Romanova, L. S. Barinova, G. V. Lagodzinskaya, A. I. Kaza-
7 kov, Y. M. Mikhailov, Preparation and NMR Analysis of b-Cyclo-
8 dextrin Nitrates, Russ. J. Appl. Chem. 2014, 87, 1884–1889.
9 References [16] W. J. Doe, New PBX formulations free of toxic isocyanates, MSc
10 Thesis, Cranfield University, Shrivenham, UK, 2012.
[17] G. Tripodo, C. Wischke, A. Lendlein, Highly Flexible Poly(ethyl-
11 [1] H. G. Ang, S. Pisharath, Energetic Polymers, Wiley-VCH., 2012.
2-cyanoacrylate) Based Materials Obtained by Incorporation of
12 [2] J. Szejtli, Introduction and General Overview of Cyclodextrin
Oligo(ethylene glycol)diglycidylether, Macromol. Symp. 2011,
13 Chemistry, Chem. Rev. 1998, 98, 1743–1753.
309/310, 49–58.
[3] H. J. Schneider, F. Hacket, V. Rdiger, NMR Studies of Cyclo-
14 [18] R. Khan, P. Forgo, K. J. Stine, V. T. D’Souza, Methods for Se-
dextrins and Cyclodextrin Complexes, Chem. Rev. 1998,
15 lective Modifications of Cyclodextrins, Chem. Rev. 1998, 98,
98,1755-1786.
16 1977–1996.
[4] S. Cahill, S. Bulusu, Molecular Complexes of Explosives with Cy-
[19] E. Gaidamuskas, B. Norkus, E. Botkus, D. C. Crans, G. Grinciene,
17 clodextrins, Magn. Reson. Chem. 1993, 31, 731–735.
Deprotonation of b-cyclodextrin in alkaline solutions, Carbo-
18 [5] P. Maksimowski, T. Rumianowski, Properties of the Gamma-Cy-
hyd. Res. 2009, 344, 250–254.
19 clodextrin/CL-20 System, Cent. Eur. J. Energ. Mater. 2016, 13,
[20] P. Mischnick, D. Momcilovic, Chemical Structure Analysis of
20 217–229.
Starch and Cellulose Derivatives, Adv. Carbohydr. Chem. Bio-
[6] J. P. Consaga, Composites of Cyclodextrin Nitrate Ester Plasti-
21 chem. 2010, 64, 117–210.
cizers, United States Patent, 5,114,506, United States Dept. of
22 [21] G. Norwitz, Spectrophotometric Determination of sulfate in
the Navy, USA 1992.
23 propellants and nitrocellulose, Department of the Army Frank-
[7] J. P. Consaga, Chemically Reactive Fragmentation Warhead,
ford Arsenal, AMS code 4931.OM.6350, Philadelphia, 1970,
24 United States Patent, 6,293,201 B1, United States of America,
http://www.dtic.mil/dtic/tr/fulltext/u2/711884.pdf (cited Oc-
25 Secretary of the Navy, USA 2001.
tober 2017).
26 [8] A. Ruebner, G. L. Statton, J. P. Consaga, Polymeric Cyclodextrin
[22] A. G. Bricknell, L. W. Trevoy, J. W. T. Spinks, A study of the sul-
27 Nitrate Esters, United States Patent, 6,527,887, Mach I, Inc., USA
fate in nitrocellulose using S35O4 – as tracer, Can. J. Chem.
1991.
28 1957, 35, 704–14.
[9] G. Mocanu, D. Vizitiu, A. Carpov, Cyclodextrin Polymers, J. Bio-
29 [23] Porter, D. Group Interaction Modelling of Polymer Properties;
act. Compat. Polym. 2001, 16, 315–342.
30 Marcel Dekker: New York, 1995.
[10] C. Rodriguez-Tenreiro, C. Alvarez-Lorenzo, A. Rodriguez-Perez,
[24] S. M. Pourmortazavi, S. G. Hosseini, M. Rahimi-Nasrabadi, S. S.
31 A. Concheiro, J. J. Torres-Labandeira, New Cyclodextrin Hydro-
Hajimirsadeghi, H. Momenian, Effect of nitrate content on ther-
32 gels Cross-linked with Diglycidylethers with a High Drug Load-
mal decomposition of nitrocellulose, J. Hazard. Mater. 2009,
33 ing and Controlled Release Ability, Pharm. Res. 2006, 23, 121–
162, 1141–1144.
34 130.
[25] NATO Standardization Agreement STANAG 4147, Chemical
[11] T. T. Nielsen, V. Wintgens, K. L. Larsen, C. Amiel, Synthesis and
35 Compatibility of Ammunition Components with Explosives and
characterization of poly(ethylene glycol) based b-cyclodextrin
36 Propellants (Non-Nuclear Application), 1992.
polymers, J. Inclusion Phenom. Macrocyclic Chem. 2009, 65,
37 [26] J. Akhavan, The Chemistry of Explosives, The Royal Society of
341–348.
Chemistry, Cambridge, 1998.
38 [12] P. Golding, A. J. Bellamy, A. E. Contini, E. Dossi, Poly-
39 phosphazenes, Patent WO, WO/2013/190260 A3, Int. Appl.
40 PCT/GB2013/000276, The Secretary of State for Defence, UK;
Cranfield University 2013. Received: May 1, 2018
41
[13] Regulation (EC) No 1907/2006 of the European Parliament and Revised: July 18, 2018
42
of the Council of 18 December 2006 concerning the Registra- Published online: September 4, 2018
43
44
45
46
47
48
49
50
51
52
53
54
55
56

Propellants Explos. Pyrotech. 2018, 43, 1023–1031 © 2018 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim www.pep.wiley-vch.de 1031
Polymer Testing 73 (2019) 338–345

Contents lists available at ScienceDirect

Polymer Testing
journal homepage: www.elsevier.com/locate/polytest

Material Properties

Thermomechanical characterisation of cross-linked β-cyclodextrin polyether T


binders
Federico Luppi, Guillaume Kister, Mark Carpenter, Eleftheria Dossi*
Centre for Defence Chemistry, Cranfield University, Defence Academy of the United Kingdom, Shrivenham, SN6 8LA, UK

A R T I C LE I N FO A B S T R A C T

Keywords: Cyclodextrins are promising building blocks for the synthesis of industrial binders. A new binder was prepared
Differential scanning calorimetry by cross-linking β-cyclodextrin with variable amounts of polyethylene glycol diglycidyl ether (40–60% w/w) to
Dynamic mechanical analysis produce a soft polyether network that was soluble in water and alcohol, and the thermomechanical properties of
Polyethylene glycol the binder were determined. Increasing the amount of cross-linker reduced the glass transition temperature of
Processability
the binder, as determined by differential scanning calorimetry and dynamic mechanical analysis. Cooling ex-
Self-healing
periments revealed sudden stress relief below the glass transition temperature, reflecting the de-bonding of the
Adhesiveness
polymer from the metallic supports. This was prevented by contact with polytetrafluoroethylene tape. Optical
microscopy confirmed the stress relief in the form of cracking, and revealed self-healing by reptation, promoted
by a higher cross-linker content and temperature. The information gained on the influence of the support
medium on the thermomechanical properties of the cross-linked β-cyclodextrins can be used by industry for
optimising manufacture and storage methods for new binders.

1. Introduction carried by CDs are easily derivatised to yield binders that can be used to
manufacture electrodes [18–20] or drug products [21–24]. Polymeric
Binders are typically large molecules that are used to bond the derivatives of all three types of CDs can respond to external stimuli such
particulate components in formulations and improve the physico- as pH or temperature, and safely deliver their cargoes of drugs to target
chemical properties of the mixture. The paint manufacturing industry is organs [25–27]. In order to process the encapsulated drugs, crystalline
historically one of the major investors in the development of new CDs have been chemically modified to allow the incorporation of drugs
polymeric binders in order to improve the cohesion of pigment particles during extrusion [26–28]. CDs were cross-linked with either multi-
and adhesion to the coated substrate following evaporation of the sol- functional isocyanate or ethylene glycol diepoxides to form insoluble
vent [1–3]. The electronics industry uses binders to improve the con- hydrogels, which can easily be formulated into pills or capsules
ductivity and mechanical strength of electrodes in batteries [4–6]. [15,29–32]. These hydrogels are thermoresponsive and some of them
Binders are also used in rocket manufacturing to improve the safety and self-heal following physical damage [31,33]. The self-healing me-
ballistic performance of propellants [7,8]. Finally, medical researchers chanism relies on the supramolecular interactions in the hydrogel, such
seek biocompatible binders that facilitate controlled drug release [9]. as hydrogen bonding promoted by the hydroxyl groups in the polymer
Most current binders are derived from petrochemical resources and [34] and interactions between CD macrocycles and the polyether chains
there is a strong demand for more sustainable alternatives [4,10–12]. [35–37]. Furthermore, the soft hydrogel matrix allows the polymer
Cyclodextrins (CDs) are polysaccharides derived from starch that chains to diffuse and re-bond in a damaged area [38–40]. The self-re-
provide a promising set of sustainable building blocks for the synthesis pair of damaged synthetic materials typically involves physical flow of
of new binders, particularly in the pharmaceutical industry [14]. As the material at or near the damaged area followed by reversible cova-
shown in Fig. 1, these toroidal macrocycles comprise six, seven or eight lent bonding or supramolecular chemistry, which also includes metal-
glucopyranose units to form α, β and γ CDs, respectively [14]. CDs have ligand coordination, π-π stacking and ionic interactions [41].
the ability to complex small molecules and alter their physicochemical Most self-repairing CD systems show a high degree of cross-linking
properties, allowing the controlled release of drugs [15,16] or the en- and are insoluble in water and common organic solvents. This limits
trapment and inactivation of toxins [17]. The many hydroxyl groups their use as binders because post-synthesis reactions cannot be achieved

*
Corresponding author.
E-mail address: e.dossi@cranfield.ac.uk (E. Dossi).

https://doi.org/10.1016/j.polymertesting.2018.11.034
Received 16 October 2018; Accepted 23 November 2018
Available online 26 November 2018
0142-9418/ © 2018 Elsevier Ltd. All rights reserved.
F. Luppi et al. Polymer Testing 73 (2019) 338–345

Fig. 1. a) Chemical structure of α (n = 6), β (n = 7) and γ (n = 8) cyclodex-


trins. b) Toroid structure of a cyclodextrin molecule [13].

without modifying the chemical structure, and the hydrogel becomes


Fig. 2. Physical characteristics of βCPCD: a) malleable βCPCD1, and b) pow-
brittle because it cannot dissipate mechanical stress when swollen [42]. dery βCPCD3.
To address these drawbacks, we recently synthesised water and alcohol
soluble cross-linked βCDs using polyethylene glycol diglycidyl ether
(βCPCDs) [13]. Here, the thermomechanical properties of the cross- configuration. The storage modulus (E′), loss modulus (E″) and
linked βCPCD system are characterized and the thermosensitive, self- damping factor (tanδ) were monitored as a function of temperature and
healing and adhesive behaviour of the synthetic materials investigated time. The free sample length between the vibrating and fixed cantilever
by differential scanning calorimetry (DSC) and dynamic mechanical clamps was ~15 mm. The test temperature was cycled three times be-
analysis (DMA). These experiments provide more information about tween −100 and 140 °C at a rate of either 2 or 10 °C min−1. Each
these new binders and will facilitate the development of appropriate material was tested in triplicate under each condition.
manufacturing and storage methods. The cross-linked βCPCD samples were incapable of self-support and
were, therefore, tested in aluminium pockets (Fig. 3a), a stainless steel
mesh (Fig. 3b), or aluminium pockets with polytetrafluoroethylene
2. Materials
(PTFE) tape (Fig. 3c) with the aim of assessing the debonding effect of
the polymer on the support. The aluminium pockets recommended by
The βCPCD samples were synthesised from βCD (≥97% purity,
Perkin Elmer consisted of rectangular shims (30 × 14 mm) cut from a
Sigma-Aldrich) prepared as a stock (13% water content, determined by
0.1 mm thick aluminium strip (supplied by RS) and then folded
thermogravimetric analysis) and polyethylene glycol diglycidyl ether
lengthwise to form the pockets. Approximately 25 mg of the cross-
(PEGDGE, Sigma-Aldrich, Mw = 500 Da, polydispersity index = 1.7),
linked βCPCD sample was placed in the centre of each pocket. PTFE
as previously described [13]. The chemical structure and purity of all
tape (30 × 15 mm) was used to fold the sample in the pocket and assess
precursors and products were assessed by proton nuclear magnetic re-
the bonding interaction between the sample and the metallic support
sonance (1H NMR) spectroscopy and DSC. The PEGDGE/βCD ratios
during cooling. Rectangular strips of type 18 mesh (30 × 15 mm) were
used are reported in Table 1, and the PEG/βCD ratio in the cross-linked
cut from a 0.65 mm thick sheet (supplied by RS). Approximately 30 mg
products was determined by 1H NMR, as previously described [35]. The
of the crosslinked βCPCD material was spread around and in the centre
physical appearance of the cross-linked products was influenced by the
of the mesh strips.
proportion of PEGDGE: cross-linked malleable products were generated
when the PEGDGE content exceeded 50% w/w (Fig. 2a, sample
βCPCD1), whereas powdery products were generated when less
3.2. Differential scanning calorimetry
PEGDGE was present (Fig. 2b, sample βCPCD3).
Thermal analysis of the βCPCD samples and their precursors was
3. Experimental carried out using a Mettler Toledo DSC3+ device. 10 mg of the material
was placed in a 40 μl aluminium pan with a pierced lid. The DSC
3.1. Dynamic mechanical analysis chamber was continuously purged with N2 gas at a flow rate of
50 ml min−1. The test temperature was cycled three times between
The thermomechanical properties of the βCPCD products were de- −100 and 140 °C. The variation of the heat flow in the samples was
termined by DMA using a Perkin Elmer DMA8000 device. The samples recorded as a function of temperature and time.
underwent a controlled sinusoidal displacement of 0.05 mm at fre-
quencies of 1, 5 and 10 Hz in the single cantilever clamping bending
3.3. Optical microscopy
Table 1
Dynamic mechanical analysis and differential scanning calorimetry data for The dynamic physical properties of the βCPCD materials under the
βCPCD samples and their precursors. influence of temperature were investigated by optical microscopy using
Sample PEGDGE:βCD Cracking E′ Tg (°C) a Leica DM microscope fitted with a temperature-controlled stage
(% w/w) temperaturea drop (Linkam THMS 600). The temperature was changed using a T95 con-
(°C) (%) DMAb DSCc
troller and an automated LNP95 liquid N2 pump (both from Linkam).
βCD – – – – 83 The material was placed on a 0.5 mm thick quartz microscopy cover
PEGDGE 100/0 −70 23 −60/-25 −73/-41/-20d slip. The slide was placed in a carrier within the stage to allow visual
βCPCD1 60–40 −42 33 41 −17 scanning. The stage was cooled to −100 °C and then heated to 100 °C at
βCPCD2 55–45 −31 21 75 −8 either 2 or 10 °C min−1. To prevent condensation forming on the
βCPCD3 40–60 +42 11 93 +96
windows of the cold stage, the interior was purged with dry N2 gas prior
a
The third cooling cycle at 1 Hz. to cooling. A digital Qicam Fast 1394 CCD camera (QImaging) was used
b
The tanδ peak. to continuously record any changes in the samples during the tem-
c
Midpoint. perature cycle. The sample was illuminated by a white light source set
d
The tanδ peaks of the three major transitions of PEGDGE. in transmission mode.

339
F. Luppi et al. Polymer Testing 73 (2019) 338–345

Fig. 3. Photographs of βCPCD samples supported by a) an aluminium pocket, b) a stainless-steel mesh, and c) an aluminium pocket wrapped in PTFE tape.

Fig. 4. Dynamic mechanical analysis showing the variation of the storage Fig. 5. Differential scanning calorimetry thermogram of βCD (10 °C min−1,
modulus (E′) (solid lines) and tanδ (dashed lines) of βCD (10 °C min−1, 1 Hz, third temperature cycle from −100 °C to 100 °C).
third temperature cycle from −100 °C to 140 °C, aluminium pocket).

4. Results and discussion

4.1. Precursor analysis – βCD

DMA was used to determine the storage modulus (E′) and damping
factor (tanδ) of βCD over three heating/cooling cycles from −100 °C to
140 °C at 10 °C min−1. The last cycle is shown in Fig. 4. The E′ value
was slightly higher during the first thermal cycle compared to the
second and third cycles, reflecting the presence of synthesis-dependent
stresses. There were no significant differences between the second and
third thermal cycles, which are shown in the Supporting Information
(SI).
The E′ values of the pure βCD sample were inversely related to the
temperature, whereas the tanδ values remained relatively constant
during each temperature cycle, showing there was no phase transition. Fig. 6. Dynamic mechanical analysis showing the variation of the storage
This is consistent with earlier experiments that defined βCD as a crys- modulus (E′) (solid lines) and tanδ (dashed lines) of PEGDGE (10 °C min−1,
talline compound [14]. Minor hysteresis was observed between cooling 1 Hz, third temperature cycle from −100 °C to 100 °C, aluminium pocket).
and heating. The lower E′ value during heating reflects the higher de-
gree of relaxation in the material at high temperatures. Therefore, the of the PEGDGE cross-linker, again with three heating/cooling cycles
higher E′ value during cooling is due to the stress created by the high from −100 °C to 140 °C at 10 °C min−1. There was no significant dif-
cooling rate. ference between the thermal cycles so only the third cycle is shown in
The variation in E′ and tanδ was also investigated as a function of Fig. 6. The E′ value fell during heating, with two main transitions at
the oscillation frequency. The increase in frequency during the tem- −70 °C and −22 °C, corresponding to the glass transitions of the
perature cycle had no significant influence on either value. The corre- PEGDGE sample. The broad and laddered decline in E′ reflects the
sponding thermogram is shown in the SI. A phase transition indicated specific properties of this commercial product, a blend of polyethylene
by DSC at 83 °C (Fig. 5) persisted at temperature cycles up to 140 °C. glycol (PEG) chains with an average molecular weight of 500 Da and a
Earlier reports attributed this phenomenon to the dissolution of βCD polydispersity index of 1.7.
crystals in the water present in the sample [43]. Molecular dynamics Previous studies have shown that the Tg of PEGDGE is inversely
simulations predicted a glass transition temperature (Tg) for βCD of related to its polydispersity because the shortest polymer chains act as a
61 °C [44] whereas others reported a measured Tg value of 216 °C [45]. plasticiser, reducing the brittleness, lowering the tensile strength and
increasing impact strength of the material overall [46,47]. During
cooling, PEGDGE underwent a single glass transition event, increasing
4.2. Precursor analysis – PEGDGE cross-linker
its stiffness. However, the rapid increase in E′ from about −30 °C was
followed by a sudden drop at about −75 °C, the latter corresponding to
DMA was also used to determine the thermomechanical properties

340
F. Luppi et al. Polymer Testing 73 (2019) 338–345

28% of the maximum E′ value. This phenomenon also occurred in tests measured by DMA occurred at a higher temperature compared to that
conducted at the lower heating/cooling rate of 2 °C min−1. DMA was recorded by DSC (Table 1), probably reflecting the oscillation frequency
used to test the aluminium support in order to eliminate artefacts applied to each sample and the different sample masses used in the two
caused by the machine and/or support matrix (SI). The E′ value of the analytical techniques.
support showed linear variation within the experimental temperature
range, suggesting that the observed phenomenon is caused by the vis- 4.3. Analysis of the βCPCD product in aluminium pockets
coelastic behaviour of the PEGDGE and its strong interaction with the
supporting pocket. Hydroxyl, epoxy and carboxyl functionalities in a The characteristics of the precursors allowed selection of suitable
polymer allow the formation of hydrogen bonds with metal and glass methods for the analysis of βCPCD product samples. First, DMA and
[48]. The strong adhesive interactions and the change in the stiffness of DSC were used to determine how the amount of PEGDGE cross-linker
the PEGDGE sample during the glass transition promote internal stress (ranging from 40% to 60%) affected the thermomechanical properties
which is suddenly released, initiating cracking and de-bonding of the of βCPCD, initially with the samples held in aluminium pockets as re-
sample from the support, recorded as a sudden drop in the E′ value. The commended for materials incapable of self-support. The DMA and DSC
physical damage (such as cracking) that emerged during the cooling data for βCD, PEGDGE and βCPCD are compared in Table 1.
cycle provided an interesting initial assessment of the adhesive strength The E′ and tanδ values for βCPCD1 containing 60% PEG units
between the sample and different support matrices. The same thermal during the third temperature cycle from −100 to 140 °C are shown in
profile during the second and third cycles confirmed that PEGDGE re- Fig. 9. There were no significant differences in either value when
covers its initial mechanical properties when heated above the Tg in comparing the three temperature cycles, suggesting that the cross-
each cycle, and the process is, therefore, reversible. linked system is not affected by the thermal history of the sample, as
The effect of the oscillation frequency on the mechanical properties observed for PEGDGE (SI). The PEG chains dominated the behaviour of
of PEGDGE was tested at 1, 5 and 10 Hz. The E′ curves at 1 and 10 Hz the cross-linked samples, with a notable shift in the transitions towards
are shown in Fig. 7. The cooling and heating curves at both frequencies positive temperature values. During heating, DMA revealed a gradual
diverge at the point of glass transition. During heating, the glass tran- stepwise decrease in E′ between the temperature extremes, with a major
sition of the compound is complete at ~25 °C, as confirmed by the small step at 55 °C coinciding with a tanδ peak. This broad transition was
decrease in E′ and the overlapping of the curves at different frequencies. frequency dependent, suggesting it represented a glass transition event
The Tg is directly proportional to the frequency during both heating and for βCPCD1 (SI). The softening of the βCPCD1 product was represented
cooling, and thus increases at higher frequencies [49,50]. The drop in by a change in the slope of the E′ curve at −10 °C, matching with the
the E′ values caused by cracking is also influenced by the oscillation onset of the tanδ peak, and this relates to the viscoelasticity of the PEG
frequency: the E′ peak at 1 and 5 Hz occurs at −78 °C whereas the soft segments.
10 Hz peak occurs at −70 °C. With an aluminium support in place, the DMA thermogram for
The cracking of the PEGDGE samples was analysed by DSC with a βCPCD1 showed a similar profile to PEGDGE during the cooling phase,
temperature cycle between −100 and 100 °C. The heat flow variation which can be attributed to the cracking of the sample. A sudden drop in
in the sample during the third temperature cycle is shown in Fig. 8. DSC E′ from its maximum value to 31% occurs at around −42 °C. The
revealed a thermal transition during the cooling interval from −23 to temperature at which cracks begin to appear in the βCPCD1 sample was
−50 °C reflecting both the crystallisation and vitrification of the blend. not significantly dependent on the oscillation frequency, whereas
Therefore, the mechanical phenomenon observed by DMA is likely to cracking of the PEG chains shifted from −78 °C at 1 Hz in PEGDGE to
occur when the PEGDGE becomes fragile following crystallisation of the −70 °C at 10 Hz. The lower mobility of the βCPCD1 samples reflects the
longer chains (−34 °C) and vitrification of the shorter chains (−42 °C). existence of a cross-linked matrix compared to the viscous PEGDGE
The lower baseline heat flow adsorption after the crystallisation peak liquid, which has a higher adhesive surface area.
from −22 °C possibly indicates the presence of a glass transition due to The DMA curves of samples βCPCD1 (60% w/w PEGDGE) and
the middle-length and shorter polymer chains that act as a plasticiser, βCPCD3 (40% w/w PEGDGE) are compared in Fig. 10. When the PEG
as reported for PEG [51]. During heating, the heat flow adsorption of content was lower, the βCPCD cracked at a much higher temperature
the PEGDGE increases between −70 and 10 °C in a multi-step transition (35 °C) and the E′ value dropped to 10% of its maximum, far below the
that represents the combination of longer polymer chains melting 30% of maximum observed for the sample with a higher PEG content. A
(~1 °C) and the increased mobility of the middle length and shorter low PEG content appears to increase the brittleness of the sample and
chains after the glass transition. Notably, the phase transformation reduces its viscoelastic behaviour, also reducing its binding strength
given that the drop in E′ is related to the strength of adhesion to the
metallic support. The glass transitions determined from the peak tanδ
values shift to higher temperatures when the PEG content is low. The
absence of first-order transitions in the βCPCD samples during cooling
was confirmed by DSC (Fig. 11 and SI) suggesting that βCPCD products
are completely amorphous at temperatures between −100 and 100 °C.

4.4. Analysis of the βCPCD product on a steel mesh support

The way in which βCPCD1 (60% w/w PEGDGE) and βCPCD3 (40%
w/w PEGDGE) interacted with a stainless-steel mesh in lieu of the
aluminium pocket was also investigated. The samples were spread over
the mesh prior to analysis (Fig. 1b). The glass transition of βCPCD1 as
determined by DMA was interrupted by stress relaxation at 15 °C
(Fig. 12). The relaxation declined between the thermal cycles but was
still present during the third cycle (SI). Cracking was still observed
Fig. 7. Dynamic mechanical analysis showing the variation of the storage during cooling, but the phenomenon was less severe when compared to
modulus (E′) of PEGDGE with frequency (1 and 10 Hz) during cooling (blue and the sample held in an aluminium pocket (Fig. 12). The cracking also
green lines) and heating (black and red lines) (10 °C min−1, third temperature manifested itself over a wider temperature range (−14 °C to −70 °C)
cycle from −100 °C to 100 °C, aluminium pocket). compared to the sudden drop observed with the aluminium pocket. The

341
F. Luppi et al. Polymer Testing 73 (2019) 338–345

Fig. 8. Differential scanning calorimetry thermogram of PEGDGE from −100° to 100 °C (10 °C min−1, third temperature cycle, aluminium crucible).

Fig. 9. Dynamic mechanical analysis showing the variation of the storage


modulus (E′) (solid lines) and tanδ (dashed lines) of βCPCD1 (10 °C min−1, Fig. 11. Differential scanning calorimetry thermogram of βCPCD1 from −100°
1 Hz, third temperature cycle from −100 °C to 140 °C, aluminium pocket). to 100 °C (10 °C min−1, third temperature cycle, aluminium crucible).

Fig. 10. Dynamic mechanical analysis showing the variation of the storage Fig. 12. Dynamic mechanical analysis showing the variation of the storage
modulus (E′) and tanδ of βCPCD1 (60% w/w PEGDGE, solid and dotted red modulus (E′) (solid lines) and tanδ (dashed lines) of βCPCD1 (10 °C min−1,
lines) and βCPCD3 (40% w/w PEGDGE, solid and dotted blue lines) (10 °C 1 Hz, third temperature cycle from −100 °C to 140 °C, stainless-steel mesh).
min−1, 1 Hz, third temperature cycle from −100 °C to 140 °C, aluminium
pocket). characteristic drop in E′ began at about −17 °C and reached 3% of the
maximum value. The drop was visible during each cycle, confirming the
stress relaxation effect that occurs with each round of heating. The
metallic mesh changes the manner in which the stress introduced

342
F. Luppi et al. Polymer Testing 73 (2019) 338–345

the metallic support used to investigate the adhesion of the compound.


The sample was cooled from 25 °C (Fig. 14a) and began to crack at
about −55 °C (Fig. 14b), with the cracks propagating further as the
temperature was reduced to −100 °C (Fig. 14c and d). The same sample
was then heated to 100 °C and the cracks began to self-repair, starting at
0 °C (Fig. 14e) until complete healing was observed at about 80 °C
(Fig. 14h).
During a second thermal cycle on the same sample, cracking began
at −58 °C and initiated at a different location (Fig. 15c). This confirms
the evidence provided by the similar E′ value profiles in each thermal
cycle, i.e. the thermal history of the sample is erased by heating, as
previously reported for a hydrogel CD/PEG system [33]. In the second
cycle, sample healing occurred at about 80 °C as in the first run
(Fig. 15h). Furthermore, βCPCD1 placed on PTFE tape showed no
evidence of cracking during the cooling phase, confirming that stress
Fig. 13. Dynamic mechanical analysis showing the variation of the storage relief by cracking was due to the bonding of βCPCD1 to the glass
modulus (E′) (solid lines) and tanδ (dashed lines) of βCPCD1 (10 °C min−1, support (SI).
1 Hz, third temperature cycle from −100 °C to 140 °C, aluminium pocket and This experiment confirmed the self-healing of the compound and
PTFE tape).
explained why the E′ curves are identical during multiple temperature
cycles. The self-healing behaviour is thought to reflect the reformation
during each cooling cycle is dissipated. The mesh has a greater surface of hydrogen bonds and host–guest interactions, as seen in CD hydrogel
area to which the βCPCD1 can bind (362 mm2) compared to the alu- systems due to the reduction of the viscosity of the cross-linked system
minium pocket (180 mm2) but, nevertheless, allows stress to be dis- when heated above Tg [33,52–54]. The physical transformation occurs
sipated within the bulk sample much more effectively. at the onset of material flow behaviour, allowing the crack to be filled
in, and probably involves a rheological model involving the snake-like
4.5. Analysis of the βCPCD product in aluminium pockets with PTFE tape displacement of the polymeric chains, described as reptation [55].
The rupture strength of sample βCPCD1 was qualitatively assessed
Finally, to confirm that the drop in the E′ values during cooling was by intentionally cutting the sample (Fig. 16b). Both parts were subse-
due to the de-bonding of the sample from its support, PTFE tape was quently re-joined and annealed in the oven at 70 °C for 30 min
used to hold βCPCD1 samples within an aluminium pocket, thus al- (Fig. 16c). The joined parts were found to be cohesive after this qua-
lowing the samples to contract freely with the decreasing temperature. litative test (Fig. 16d). The sample was gripped at its extremities and
Preliminary DMA characterisation of the PTFE tape (blank) showed a pulled gently. The sample showed considerable elongation. Diffusion of
softening at the onset temperature of 18 °C (SI). The absence of adhe- the polymeric chains by reptation and the reforming of hydrogen bonds
sion between βCPCD1 and the support when enclosed by PTFE tape was in the fracture allowed the material to qualitatively self-repair and self-
confirmed by the drop in E′ during cooling as shown in Fig. 13. The heal. In future work, the self-healing capabilities of this compound will
hysteresis between the heating and cooling curves during the same be assessed in more detail.
temperature cycle diminished with each cycle. Therefore, the PTFE tape
almost completely eliminated the de-bonding phenomenon and the 5. Conclusions
formation of thermal stress in the sample during the cooling phase. The
Tg of βCPCD1 under these conditions was 41 °C, identical to the value The thermomechanical properties of a new cross-linked binder
recorded in the aluminium pocket without tape, confirming that the based on βCD and PEG segments were assessed by DSC and DMA. The
experimental setup does not affect the Tg. cross-linking improved the processability of the material compared to
crystalline βCD, which has unsuitable physical properties for a binder.
4.6. Analysis of the βCPCD product by optical microscopy All cross-linked βCPCD products showed viscoelastic responses similar
to those observed for the pure cross-linker. The viscoelasticity was di-
Optical microscopy was used to characterise in more detail the rectly related to the proportion of cross-linker in the βCPCD system,
cracking that occurred in the βCPCD1 product during cooling (Fig. 14). with Tg values as low as −17 °C (determined by DSC) when cross-linker
A glass support was considered adequate as a transparent substitute for content was > 50% w/w. DMA revealed a sudden drop in E′ during the

Fig. 14. Optical microscope images of βCPCD1 (60% w/w PEGDGE) during the first temperature cycle from −100 to 100 °C. The top row (a–d) shows the cooling
phase captured at 25, –55, −78 and −80 °C, and the bottom row (e–h) shows the heating phase captured at 2, 43, 67 and 87 °C.

343
F. Luppi et al. Polymer Testing 73 (2019) 338–345

Fig. 15. Optical microscope images of βCPCD1 (60% w/w PEGDGE) during the second temperature cycle from −100 to 100 °C. The top row (a–d) shows the cooling
phase captured at 100, 25, −61 and −88 °C, and the bottom row (e–h) shows the heating phase captured at 5, 43, 68 and 84 °C.

Fig. 16. Self-healing of βCPCD1 (60% w/w PEGDGE). a) Sample after solvent evaporation. b) Sample after cutting. c) The parts are placed in contact. d) The sample
is heated to 70 °C for 30 min and pulled by the extremities.

cooling sequence, and the amplitude of the drop was directly propor- 210–218, https://doi.org/10.1016/j.porgcoat.2017.07.023.
tional to the amount of cross-linker present. This phenomenon is [4] B. Andres, C. Dahlström, N. Blomquist, M. Norgren, H. Olin, Cellulose binders for
electric double-layer capacitor electrodes: the influence of cellulose quality on
thought to reflect the de-bonding of βCPCD from the metallic support electrical properties, Mater. Des. 141 (2018) 342–349, https://doi.org/10.1016/j.
when the sample was cooled below Tg. The adhesiveness of the material matdes.2017.12.041.
was drastically reduced when folded in PTFE film, indicating that the [5] D. Mazouzi, Z. Karkar, C.R. Hernandez, P.J. Manero, D. Guyomard, L. Roué,
B. Lestriez, Critical roles of binders and formulation at multiscales of silicon-based
adhesive strength is due to the ability to form hydrogen bonds with the composite electrodes, J. Power Sources 280 (2015) 533–549, https://doi.org/10.
metallic support. Optical microscopy confirmed that the adhesiveness 1016/j.jpowsour.2015.01.140.
between the samples and the support, combined with the changes in [6] J.E.S. Technol, N. Choi, S. Ha, Y. Lee, J.Y. Jang, M. Jeong, W.C. Shin, M. Ue, Recent
progress on polymeric binders for silicon anodes in lithium-ion batteries, J.
viscoelasticity of the PEG and βCPCD segments at temperatures below Electrochem. Sci. Technol. 6 (2015) 35–49, https://doi.org/10.5229/JECST.2015.
Tg, produced mechanical stress that resulted in the propagation of 6.2.35.
cracks in the cross-linked material. Interestingly, the damaged βCPCD [7] G.G. Ang, Pisharath Sreekumar, Polymers as binders and plasticizers - historical
perspective, Energ. Polym. - Bind. Plast. Enhancing Performance. 37 (2012) 510,
structure was able to restore its integrity by self-healing, which began at
https://doi.org/10.1002/prep.201280003.
0 °C and achieved complete self-repair at about 80 °C. The results show [8] S. Chaturvedi, P.N. Dave, Solid propellants: AP/HTPB composite propellants, Arab.
that the thermomechanical properties of the βCPCD system are superior J. Chem. (2015), https://doi.org/10.1016/j.arabjc.2014.12.033.
to those of pure βCD. The influence of the support matrix on the [9] N. Bertrand, P. Colin, M. Ranger, J. Leblond, Designing Polymeric Binders for
Pharmaceutical Applications, in: 2013: pp. 483–517. doi:10.1039/9781849737821-
thermomechanical properties of βCPCD could facilitate the develop- 00483.
ment of optimal manufacturing and storage methods for new binders. [10] P. Tourneroche, J.C. Gelin, M. Sahli, T. Barrière, Development and thermo-physical
characterization of polymers/metallic powder mixtures for MIM application,
Procedia Eng 81 (2014) 2530–2536, https://doi.org/10.1016/j.proeng.2014.10.
Funding 362.
[11] A. Royer, T. Barriére, J.C. Gelin, Development of bio-sourced binder to metal in-
This project was funded through the Weapons Science and jection moulding, AIP Conf. Proc, AIP Publishing LLC, 2016, p. 020009, , https://
doi.org/10.1063/1.4963413.
Technology Centre (WSTC) by the UK Defence Science and Technology [12] S. Vasantrao Patil, S. Laxman Ghatage, S. Shankar Navale, N. Kadar Mujawar,
Laboratory (DSTL). Natural binders in tablet formulation, Int. J. PharmTech Res. 6 (2014) 1070–1073
https://www.researchgate.net/publication/265644099_Natural_Binders_in_Tablet_
Formulation.
Appendix A. Supplementary data
[13] F. Luppi, H. Cavaye, E. Dossi, Nitrated cross-linked β-cyclodextrin binders ex-
hibiting low glass transition temperatures, Propellants, Explos. Pyrotech. (2018),
Supplementary data to this article can be found online at https:// https://doi.org/10.1002/prep.201800137.
[14] J. Szejtli, Introduction and general overview of cyclodextrin chemistry, Chem. Rev.
doi.org/10.1016/j.polymertesting.2018.11.034.
98 (1998) 1743–1754, https://doi.org/10.1021/cr970022c.
[15] R. Challa, A. Ahuja, J. Ali, R.K. Khar, Cyclodextrins in drug delivery: an updated
References review, AAPS PharmSciTech 6 (2005) E329–E357, https://doi.org/10.1208/
pt060243.
[16] E.M.M. Del Valle, Cyclodextrins and their uses: a review, Process Biochem. 39
[1] K.K. Sutna, S. Jacob, R. Joseph, Paint formulation using water based binder and (2004) 1033–1046, https://doi.org/10.1016/S0032-9592(03)00258-9.
property studies, Macromol. Symp. 277 (2009) 144–151, https://doi.org/10.1002/ [17] R. Solaro, E. Dossi, E. Chiellini, G. Mazzanti, New multifunctional polymeric ma-
masy.200950318. terials for the treatment of chronic uremia, J. Bioact. Compat Polym. 12 (1997)
[2] F.F. Abdel-Mohsen, H.S. Emira, A study of the effects of different binders and fillers 27–46, https://doi.org/10.1177/088391159701200103.
on the properties of flame retardant paints, Pigment Resin Technol. 36 (2007) [18] Y.K. Jeong, T.W. Kwon, I. Lee, T.S. Kim, A. Coskun, J.W. Choi, Hyperbranched β-
67–73, https://doi.org/10.1108/03699420710733493. cyclodextrin polymer as an effective multidimensional binder for silicon anodes in
[3] V. Alvarez, M. Paulis, Effect of acrylic binder type and calcium carbonate filler lithium rechargeable batteries, Nano Lett. 14 (2014) 864–870, https://doi.org/10.
amount on the properties of paint-like blends, Prog. Org. Coating 112 (2017) 1021/nl404237j.

344
F. Luppi et al. Polymer Testing 73 (2019) 338–345

[19] J. Wang, Z. Yao, C.W. Monroe, J. Yang, Y. Nuli, Carbonyl- β -cyclodextrin as a novel 1021/cr970019t.
binder for sulfur composite cathodes in rechargeable lithium batteries, Adv. Funct. [37] T. Miao, S.L. Fenn, P.N. Charron, R.A. Oldinski, Self-healing and thermoresponsive
Mater. 23 (2013) 1194–1201, https://doi.org/10.1002/adfm.201201847. dual-cross-linked alginate hydrogels based on supramolecular inclusion complexes,
[20] F. Zeng, W. Wang, A. Wang, K. Yuan, Z. Jin, Y.S. Yang, Multidimensional polycation Biomacromolecules 16 (2015) 3740–3750, https://doi.org/10.1021/acs.biomac.
β-cyclodextrin polymer as an effective aqueous binder for high sulfur loading 5b00940.
cathode in lithium-sulfur batteries, ACS Appl. Mater. Interfaces 7 (2015) [38] K. Guo, M.S. Lin, J.F. Feng, M. Pan, L.S. Ding, B.J. Li, S. Zhang, The deeply un-
26257–26265, https://doi.org/10.1021/acsami.5b08537. derstanding of the self-healing mechanism for self-healing behavior of supramole-
[21] W. Chen, C. Wang, L. Yan, L. Huang, X. Zhu, B. Chen, H.J. Sant, X. Niu, G. Zhu, cular materials based on cyclodextrin–guest interactions, Macromol. Chem. Phys.
K.N. Yu, V.A.L. Roy, B.K. Gale, X. Chen, Improved polyvinylpyrrolidone micro- 218 (2017) 1600593, https://doi.org/10.1002/macp.201600593.
needle arrays with non-stoichiometric cyclodextrin, J. Mater. Chem. B. 2 (2014) [39] H. Fujita, Y. Einaga, Self diffusion and viscoelasticity in entangled systems I. Self-
1699–1705, https://doi.org/10.1039/c3tb21698e. diffusion coefficients, Polym. J. 17 (1985) 1131–1139, https://doi.org/10.1295/
[22] V. Giglio, C. Sgarlata, G. Vecchio, Novel amino-cyclodextrin cross-linked oligomer polymj.17.1131.
as efficient Carrier for anionic drugs: a spectroscopic and nanocalorimetric in- [40] M. Muthukumar, A. Baumgärtner, Diffusion of a polymer chain in random media,
vestigation, RSC Adv. 5 (2015) 16664–16671, https://doi.org/10.1039/ Macromolecules 22 (1989) 1941–1946, https://doi.org/10.1021/ma00194a071.
c4ra16064a. [41] Y. Yang, X. Ding, M.W. Urban, Chemical and physical aspects of self-healing ma-
[23] R. Solaro, S. D'Antone, L. Bemporad, E. Chiellini, New polyfunctional derivatives of terials, Prog. Polym. Sci. 49–50 (2015) 34–59, https://doi.org/10.1016/j.
β-cyclodextrin suited for the formulation of drug release systems, J. Bioact. Compat progpolymsci.2015.06.001.
Polym. 8 (1993) 236–250, https://doi.org/10.1177/088391159300800303. [42] X. Zhao, Multi-scale multi-mechanism design of tough hydrogels: building dis-
[24] O. Radia, E. Rogalska, G. Moulay-Hassane, Preparation of meloxicamβ-cyclodex- sipation into stretchy networks, Soft Matter 10 (2014) 672–687, https://doi.org/10.
trinpolyethylene glycol 6000 ternary system: characterization, in vitro and in vivo 1039/C3SM52272E.
bioavailability, Pharmaceut. Dev. Technol. 17 (2012) 632–637, https://doi.org/10. [43] E. Specogna, K.W. Li, M. Djabourov, F. Carn, K. Bouchemal, Dehydration, dissolu-
3109/10837450.2011.565347. tion, and melting of cyclodextrin crystals, J. Phys. Chem. B 119 (2015) 1433–1442,
[25] Y.Y. Liu, X.D. Fan, Synthesis and characterization of pH- and temperature-sensitive https://doi.org/10.1021/jp511631e.
hydrogel of N-isopropylacrylamide/cyclodextrin based copolymer, Polymer [44] G. Zhou, T. Zhao, J. Wan, C. Liu, W. Liu, R. Wang, Predict the glass transition
(Guildf) 43 (2002) 4997–5003, https://doi.org/10.1016/S0032-3861(02)00350-6. temperature and plasticization of β-cyclodextrin/water binary system by molecular
[26] J.-T. Zhang, S.-W. Huang, J. Liu, R.-X. Zhuo, Temperature sensitive poly[N-iso- dynamics simulation, Carbohydr. Res. 401 (2015) 89–95, https://doi.org/10.1016/
propylacrylamide-co-(acryloyl?-cyclodextrin)] for improved drug release, j.carres.2014.10.026.
Macromol. Biosci. 5 (2005) 192–196, https://doi.org/10.1002/mabi.200400167. [45] C. Rodríguez-Tenreiro, C. Alvarez-Lorenzo, Á. Concheiro, J.J. Torres-Labandeira,
[27] V. Bennevault, C. Huin, P. Guégan, K. Evgeniya, X.P. Qiu, F.M. Winnik, Characterization of cyclodextrincarbopol interactions by DSC and FTIR, J. Therm.
Temperature sensitive supramolecular self assembly of per-6-PEO-β-cyclodextrin Anal. Calorim. 77 (2004) 403–411, https://doi.org/10.1023/B:JTAN.0000038981.
and α,ω-di-(adamantylethyl)poly(N-isopropylacrylamide) in water, Soft Matter 11 30494.f4.
(2015) 6432–6443, https://doi.org/10.1039/c5sm01293g. [46] S.K. Behera, D. Saha, P. Gadige, R. Bandyopadhyay, Effects of polydispersity on the
[28] J. Thiry, F. Krier, S. Ratwatte, J.M. Thomassin, C. Jerome, B. Evrard, Hot-melt glass transition dynamics of aqueous suspensions of soft spherical colloidal parti-
extrusion as a continuous manufacturing process to form ternary cyclodextrin in- cles, Phys. Rev. Mater. 1 (2017) 055603, , https://doi.org/10.1103/
clusion complexes, Eur. J. Pharmaceut. Sci. 96 (2017) 590–597, https://doi.org/10. PhysRevMaterials.1.055603.
1016/j.ejps.2016.09.032. [47] S.J. Li, S.J. Xie, Y.C. Li, H.J. Qian, Z.Y. Lu, Influence of molecular-weight poly-
[29] H. Kono, T. Nakamura, H. Hashimoto, Y. Shimizu, Characterization, molecular dispersity on the glass transition of polymers, Phys. Rev. E. 93 (2016) 012613, ,
dynamics, and encapsulation ability of β-cyclodextrin polymers crosslinked by https://doi.org/10.1103/PhysRevE.93.012613.
polyethylene glycol, Carbohydr. Polym. 128 (2015) 11–23, https://doi.org/10. [48] B. ESCAIG, Binding metals to polymers. A short review of basic physical mechan-
1016/j.carbpol.2015.04.009. isms, Le J. Phys. IV. 03 (1993), https://doi.org/10.1051/jp4:19937120 C7-753-C7-
[30] S.K. Osman, G.M. Soliman, M. Amin, A. Zaky, Self-assembling hydrogels based on B- 761.
Cyclodextrin polymer and poly (ethylene glycol) bearing hydrophobic moieties for [49] Y. Bai, L. Jin, Characterization of frequency-dependent glass transition temperature
protein delivery, Int. J. Pharm. Pharmaceut. Sci. 6 (2014) 591–597. by Vogel-Fulcher relationship, J. Phys. D Appl. Phys. 41 (2008) 152008, https://
[31] F. Van De Manakker, T. Vermonden, N. El Morabit, C.F. Van Nostrum, doi.org/10.1088/0022-3727/41/15/152008.
W.E. Hennink, Rheological behavior of self-assembling PEG-β-cyclodextrin/PEG- [50] G. Kister, E. Dossi, Cure monitoring of CFRP composites by dynamic mechanical
cholesterol hydrogels, Langmuir 24 (2008) 12559–12567, https://doi.org/10. analyser, Polym. Test. 47 (2015) 71–78, https://doi.org/10.1016/j.polymertesting.
1021/la8023748. 2015.08.009.
[32] S. Salmaso, A. Semenzato, S. Bersani, P. Matricardi, F. Rossi, P. Caliceti, [51] N. Tabary, M.J. Garcia-Fernandez, F. Danède, M. Descamps, B. Martel, J.F. Willart,
Cyclodextrin/PEG based hydrogels for multi-drug delivery, Int. J. Pharm. 345 Determination of the glass transition temperature of cyclodextrin polymers,
(2007) 42–50, https://doi.org/10.1016/j.ijpharm.2007.05.035. Carbohydr. Polym. 148 (2016) 172–180, https://doi.org/10.1016/j.carbpol.2016.
[33] Y.G. Jia, X.X. Zhu, Self-healing supramolecular hydrogel made of polymers bearing 04.032.
cholic acid and β-cyclodextrin pendants, Chem. Mater. 27 (2015) 387–393, https:// [52] M.D. Hager, P. Greil, C. Leyens, S. Van Der Zwaag, U.S. Schubert, Self-healing
doi.org/10.1021/cm5041584. materials, Adv. Mater. 22 (2010) 5424–5430, https://doi.org/10.1002/adma.
[34] F. Herbst, D. Döhler, P. Michael, W.H. Binder, Self-healing polymers via supramo- 201003036.
lecular forces, Macromol. Rapid Commun, Wiley-VCH Verlag GmbH & Co. KGaA, [53] S.R. White, B.J. Blaiszik, S.L.B. Kramer, S.C. Olugebefola, J.S. Moore, N.R. Sottos,
Weinheim, Germany, 2013, pp. 203–220, , https://doi.org/10.1002/marc. Self-healing polymers and composites, Am. Sci. 99 (2011) 392–399, https://doi.
201200675. org/10.1179/095066010X12646898728408.
[35] T.T. Nielsen, V. Wintgens, K.L. Larsen, C. Amiel, Synthesis and characterization of [54] A. Klaewklod, V. Tantishaiyakul, N. Hirun, T. Sangfai, L. Li, Characterization of
poly(ethylene glycol) based β-cyclodextrin polymers, J. Inclusion Phenom. supramolecular gels based on β-cyclodextrin and polyethyleneglycol and their po-
Macrocycl. Chem. 65 (2009) 341–348, https://doi.org/10.1007/s10847-009- tential use for topical drug delivery, Mater. Sci. Eng. C 50 (2015) 242–250, https://
9591-0. doi.org/10.1016/j.msec.2015.02.018.
[36] H.-J. Schneider, F. Hacket, V. Rüdiger, H. Ikeda, NMR studies of cyclodextrins and [55] J. Klein, Evidence for reptation in an entangled polymer melt, Nature 271 (1978)
cyclodextrin complexes, Chem. Rev. 98 (1998) 1755–1786, https://doi.org/10. 143–145, https://doi.org/10.1038/271143a0.

345

You might also like