Process and Equipment Development for

Textile Dyeing in Supercritical Carbon Dioxide

Martijn VAN DER KRAAN

 Process and Equipment Development for

Textile Dyeing in Supercritical Carbon Dioxide

Proefschrift

ter verkrijging van de graad van doctor

aan de Technische Universiteit Delft,
op gezag van de Rector Magnificus prof. dr. ir. J.T. Fokkema,
voorzitter van het College van Promoties,
in het openbaar te verdedigen op maandag 12 september 2005 om 15:30 uur

door

Martijn VAN DER KRAAN

scheikundig ingenieur
geboren te Nieuwkoop.
Dit proefschrift is goedgekeurd door de promotor:
Prof. dr. G.J. Witkamp

Samenstelling promotiecommissie:

Rector Magnificus Voorzitter

Prof. dr. G.J. Witkamp Technische Universiteit Delft, promotor
Prof. dr. ir. J.T.F. Keurentjes Technische Universiteit Eindhoven
Prof. dr. ir. A.B. de Haan Universiteit Twente
Prof. dr. F. Recasens Polytechnical University of Barcelona
Prof. ir. G.J. Harmsen Technische Universiteit Delft
Prof. dr. J.A. Moulijn Technische Universiteit Delft
Dr. ir. G.F. Woerlee FeyeCon D&I B.V.

Dr. ir. G.F. Woerlee heeft als begeleider in belangrijke mate aan de totstandkoming
van dit proefschrift bijgedragen.

Dit werk is financieel ondersteund door de ministeries van EZ, OCenW en VROM
via het EET programma van SenterNovem, project EET K99101.

ISBN-10: 9090199128
ISBN-13: 9789090199122
Printed by Febodruk B.V., Enschede

Copyright © 2005 by M. van der Kraan

All rights reserved. No part of the material protected by this copyright notice may be reproduced or utilised in any
form or by any means, electronic or mechanical, including photo copy, recording or by any information storage
and retrieval system, without prior permission from the publisher.
Voor mijn ouders
en voor Uschi
Summary

Process and Equipment Development for Textile

Dyeing in Supercritical Carbon Dioxide

The large-scale water pollution by the textile dyeing industry is a global

environmental problem. The ever more stringent regulations on wastewater also
make it an economical problem. In the last two decades therefore, research has
been done on an environmentally benign technology, using supercritical carbon
dioxide (scCO2) as a dye solvent, rather than water.

The applicability of the technology is limited at this moment by two factors.

Firstly, there is not enough knowledge on reactive and non-reactive dyeing.
Secondly, because the process operates at conditions of typically 120ºC and 300
bar, high-pressure equipment is needed which results in high investment costs that
stand in the way of industrial implementation. In this work, the supercritical dyeing
process was investigated experimentally regarding both reactive and non-reactive
dyeing and also new equipment was designed to minimize the equipment and
process costs.

Silk, wool, nylon and polyester were dyed successfully in scCO2 with two non-
polar dyes containing vinylsulphone or dichlorotriazine reactive groups. It was
found that the presence of a small amount of water increased the coloration of wool
and nylon significantly and that silk only reacted with the dyes when water was
added. This was the first time that silk was dyed successfully in scCO2. Maximum
color depths were observed when both the scCO2 and the textiles were saturated
with water.

To predict how much water is to be added to a supercritical dyeing vessel in order

to obtain optimal coloration with reactive dyes, a thermodynamic model was
developed. This model describes the equilibrium distribution of water over the
supercritical and the textile phase, as a function of pressure, temperature, vessel
volume and textile mass. One part of the model is the calculation of the solubility
of water in a supercritical fluid, which is done with a new equation, derived from
statistical thermodynamics.

The non-reactive dyeing of polyester was investigated experimentally in a new 40-

liter pressure vessel. The saturation coloration increased and the distribution
coefficient decreased with temperature and scCO2-density. The adsorption of the
dye on the polyester was exothermic and followed Nernst adsorption. Since the
values of the saturation colorations, the temperature dependence of the distribution
coefficient and also the adsorption mechanism were the same as in aqueous dyeing,
it was concluded that the disperse dyeing in scCO2 behaved thermodynamically the
same as in water.

Because no generally accepted procedure exists for the design of heat exchangers
for supercritical fluids, this was investigated with computational fluid dynamics
simulations. Heat transfer coefficients were determined for different temperatures,
pressures and Reynolds numbers, for CO2 flowing up or down through a vertical
pipe.

The impairment and enhancement of heat transfer caused by the temperature-

induced variation of physical properties was investigated, as well as the effect of
buoyancy. It was determined when (i.e. for which process conditions) and how (i.e.
with which Nusselt relation) these effects are to be taken into account in the
calculation of heat transfer coefficients for the design of heat exchangers working
at supercritical conditions.

A technical-scale, 100-litre dyeing machine was designed and built to test polyester
beam dyeing in scCO2 at 300 bar and 120ºC. A new type of pressure vessel was
used, consisting of a steel liner with carbon fibers wound around to take up the
radial forces and a yoke construction for the axial forces. This configuration lowers
the investment cost but also the operating cost, because the amount of steam
required to heat the vessel is lower than for a completely steel vessel. Furthermore,
because the carbon fiber vessel requires less heating due to the low heat capacity of
the carbon fibers, the process time is shortened. To circulate the CO2 with the
dissolved dye through the textile, a low-pressure centrifugal pump was designed
for service in scCO2 and placed inside the dyeing vessel.
Also a commercial-scale 1000-liter supercritical dyeing machine was designed, for
treating 300 kg polyester while recycling all dye and 96% of the CO2. An
economical analysis showed that, although the purchase cost for a supercritical
machine is higher (500 k€) than for an aqueous machine (100 k€), the operating
cost is lower (0.35 instead of 0.99 € per kg polyester). This is caused by the higher
rate of dyeing and by the simpler dye formulations that can be used in scCO2. The
overall result is a 50% lower process cost for the supercritical process.

This thesis contributes to the development of supercritical fluid dyeing technology

in two ways. Firstly, knowledge was generated on reactive and non-reactive
dyeing, two subjects that were insufficiently explored up to now. Secondly, new
equipment was designed, lowering the investment and operating costs of the
process. It was shown that the supercritical process is not only environmentally
superior to aqueous dyeing, but also economically. The overall result of this thesis
is a better understanding of textile dyeing in supercritical carbon dioxide and a
lower threshold for the implementation in industry.

Martijn van der Kraan

Samenvatting

Proces- en apparaatontwikkeling voor textiel verven

in superkritisch kooldioxide

De grootschalige watervervuiling door de textielindustrie is een wereldwijd

milieuprobleem. De steeds strengere regelgeving betreffende afvalwater maken het
tevens een economisch probleem. Daarom is in de laatste twee decennia door
diverse wetenschappers onderzoek gedaan naar een milieuvriendelijke technologie
waarbij superkritisch kooldioxide (scCO2) wordt gebruikt als oplosmiddel voor
kleurstof, in plaats van water.

De toepasbaarheid van de technologie wordt op dit moment beperkt door twee

factoren. Ten eerste is er niet genoeg kennis van reactief en niet-reactief verven in
scCO2. Ten tweede, omdat het proces plaatsvindt bij condities van typisch 120ºC
en 300 bar, zijn hoge-druk machines nodig, hetgeen resulteert in hoge
investeringskosten die in de weg staan van industriële implementatie. In dit werk is
het superkritische verfproces experimenteel onderzocht voor wat betreft reactief en
niet-reactief verven en er is tevens nieuwe apparatuur ontwikkeld om de
investerings- en proceskosten te minimaliseren.

Zijde, wol, nylon en polyester zijn met goed resultaat in scCO2 geverfd met twee
apolaire kleurstoffen met de reactieve groep vinylsulfon respectievelijk
dichloortriazine. Er werd vastgesteld dat de aanwezigheid van een kleine
hoeveelheid water de aankleuring van wol en nylon significant deed toenemen en
dat zijde alleen reageerde met de kleurstof wanneer water werd toegevoegd. Dit
was de eerste keer dat zijde in scCO2 is geverfd met een goed resultaat. Maximale
kleurdiepten werden gevonden wanneer zowel het scCO2 als het textiel waren
verzadigd met water.

Om te voorspellen hoeveel water moet worden toegevoegd aan een superkritisch

verfvat om optimale aankleuring te verkrijgen, werd een thermodynamisch model
ontwikkeld. Dit model beschrijft de evenwichtsverdeling van water over de
superkritische fase en de textielfase, als functie van druk, temperatuur, vatvolume
en textielmassa. Een onderdeel van het model is de berekening van de
oplosbaarheid van water in een superkritisch fluïdum, hetgeen werd uitgevoerd met
een nieuwe vergelijking, afgeleid uit de statistische thermodynamica.

Het niet-reactief verven van polyester werd experimenteel onderzocht in een nieuw
40-liter drukvat. De verzadigingsaankleuring nam toe en de verdelingscoëfficiënt
nam af met de temperatuur en dichtheid van de scCO2. De adsorptie van kleurstof
aan het polyester was exotherm en volgde Nernst adsorptie. Aangezien de waarden
van de verzadigingsaankleuringen, de temperatuurafhankelijkheid van de
verdelingscoëfficiënt en ook het adsorptiemechanisme hetzelfde waren als in
waterverven, werd geconcludeerd dat dispers verven in scCO2 hetzelfde
thermodynamische gedrag vertoont als in water.

Omdat geen algemeen geaccepteerde procedure bestond voor het ontwerp van
warmtewisselaars voor superkritische fluïda, werd dit onderzocht met
“computational fluid dynamics” simulaties. Warmteoverdrachtscoëfficiënten
werden bepaald voor verschillende temperaturen, drukken en Reynoldsgetallen,
voor omhoog- en omlaagstromende CO2 in een verticale pijp.

De verslechtering of verbetering van de warmteoverdracht, veroorzaakt door de

temperatuurafhankelijke variatie van de fysische eigenschappen, werd onderzocht,
evenals het effect van de zwaartekracht. Er werd vastgesteld wanneer (voor welke
procescondities) en hoe (met welke Nusseltrelatie) deze effecten in rekening
moeten worden gebracht bij de calculatie van warmteoverdrachtscoëfficiënten voor
het ontwerp van warmtewisselaars werkend onder superkritische condities.

Een 100-liter verfmachine op technische schaal werd ontworpen en gebouwd om

polyester boomverven te testen in scCO2 bij 300 bar en 120ºC. Een nieuw type
drukvat werd gebruikt, bestaande uit een stalen binnenvat omwikkeld met
koolstofvezels om de radiale krachten op te vangen en met een jukconstructie voor
de axiale krachten. Deze configuratie verlaagt de investeringskosten maar ook de
operationele kosten, doordat de hoeveelheid energie om het vat te verwarmen lager
is dan bij een volledig stalen vat. Om de CO2 met de opgeloste kleurstof door het
textiel te circuleren, is een lage-druk centrifugaalpomp ontworpen voor gebruik in
scCO2, deze is binnen in het drukvat geplaatst.

Er is eveneens een 1000-liter superkritische verfmachine ontworpen, voor het

verven van 300 kg polyester en het hergebruiken van alle kleurstof en 96% van de
CO2. Een economische analyse toonde aan dat, hoewel dat de aanschafkosten van
een superkritisch machine hoger zijn (500 k€) dan van een waterverfmachine (100
k€), de operationele kosten lager zijn (0.35 in plaats van 0.99 € per kg polyester).
Dit wordt veroorzaakt door de hogere snelheid van het verfproces en door de
eenvoudigere kleurstofformuleringen die in scCO2 gebruikt kunnen worden. Het
overall resultaat is een 50% lagere proceskost voor het superkritische proces.

Dit proefschrift draagt bij tot de ontwikkeling van de superkritische

verftechnologie op twee manieren. Ten eerste is er kennis gegenereerd op het
gebied van reactief en niet-reactief verven, twee onderwerpen die tot nog toe
onvoldoende waren onderzocht. Ten tweede is er nieuwe apparatuur ontworpen die
de investerings- en operationele kosten hebben verlaagd. Er is aangetoond dat het
superkritische proces niet alleen milieutechnisch superieur is aan waterverven,
maar ook economisch. Het overall resultaat van dit proefschrift is een beter begrip
van textiel verven in superkritisch kooldioxide en een lagere drempel voor de
industriële implementatie.

Martijn van der Kraan

Contents

1 Introduction 1

2 Dyeing of Natural and Synthetic Textiles in Supercritical

Carbon Dioxide with Disperse Reactive Dyes 11

3 Equilibrium Distribution of Water in the Two-Phase System

Supercritical Carbon Dioxide – Textile 27

4 Equilibrium Study on the Disperse Dyeing of Polyester

Textile in Supercritical Carbon Dioxide 49

5 The Influence of Variable Physical Properties and Buoyancy

on Heat Exchanger Design for Near- and Supercritical
Conditions 71

6 Process and Equipment Design for a 100-Liter

Polyester Beam Dyeing Machine 89

7 Economical Evaluation of Polyester Beam Dyeing in

Supercritical Carbon Dioxide 115

Epilogue 131

Dankwoord 133

Curriculum Vitae 135

Publications 137
Chapter 1

Introduction

1.1. Environmental implications of textile dyeing

The importance of the textile processing industry is clearly illustrated by the world
production in 2003, being 52 Mton. The two main fiber types are polyester (22
Mton) and cotton (20 Mton), the remainder is accounted for by wool, silk, nylon,
viscose and some specialty fibers [1].

Virtually all of this textile is dyed, using worldwide 0.7 Mton of dye powder, with
a total value of 19 billion € [2]. Approximately 5 to 10 % of the non-reactive dyes
are lost with the wastewater, for reactive dyes this is even 50 %. The resulting
pollution lies between 40 and 300 liter per kg textile [3], giving a total amount of 2
to 20 billion cubic meters of wastewater annually, which indicates the seriousness
of this environmental problem.

Although other chemicals, such as salts and dispersing agents, are used in the
dyeing process, the most problematic pollutant is the dye itself. Inherent to their
purpose, dye molecules are designed to be resistant to degradation by light, water
and many chemicals [4]. Therefore, treatment in municipal water purification
facilities can not decolorize dye house effluents. Instead, the wastewater has to be
treated at the site.

1.2. Solutions to the wastewater problem

Dye molecules can be decomposed in water by a range of chemical, physical and

biological treatments [5]. The most widely used technique is the oxidative process,
where hydrogen peroxide is added to the water and activated by ultra violet light to
oxidize the dye molecules. An important draw-back of this technique is the

1
production of toxic sludge, which has to be disposed of or incinerated. All other
techniques to purify the water are characterized by either a high process cost or a
production of toxic waste. Since the treated water has to satisfy ever more stringent
environmental regulations, the end-of-pipe solutions become increasingly
expensive for the textile industry. It is therefore that a solution at the source of the
problem, i.e. replacement of water as a dyeing solvent, is preferable.

In the sixties and seventies, this consideration has led to some research on the use
of chlorinated hydrocarbons, mainly perchloroethylene, as dyeing media [6]. This
so-called solvent dyeing has, however, a major disadvantage in the toxicity of the
solvents. The technique has not been applied in industry and all research on the
subject has been stopped.

A more promising alternative to water is the use of supercritical fluids. Such

solvents have not yet been implemented in textile industry, but their use as dyeing
medium has received considerable attention from researchers in the past two
decades, as is illustrated by the 143 articles mentioned in the review by Bach et al.
[7].

1.3. Supercritical fluids

When a solid is heated, the thermal motion of the molecules increases, the solid
melts and a liquid and a vapor phase are formed. In figure 1.1 these three states of
matter are graphically presented. When a vapor below its critical temperature is
compressed, it condenses when the vapor-liquid equilibrium line is crossed. Above
the critical temperature however, the thermal energy of the vapor molecules is so
high, that condensation is no longer possible, no matter how much the pressure is
increased. The vapor-liquid equilibrium line ends at the critical temperature. When
a fluid is above its critical temperature and the corresponding critical pressure, it
can not be regarded as a vapor or a liquid and it is referred to as a supercritical
fluid.

The most widely used supercritical fluid is carbon dioxide, because it combines a
relatively mild critical point with non-flammability, non-toxicity and a low price.
Because of its green and safe character, it is the best supercritical solvent for textile

2
 p (bar)
supercritical
fluid
solid liquid
74 bar
critical point

vapor
triple point

31ºC T (ºC)

Figure 1.1. Phase diagram of carbon dioxide, phase behavior as a function of

pressure (p) and temperature (T)

dyeing. The CO2 is a waste product of combustion, fermentation and ammonia

synthesis, so that no CO2 has to be produced especially for dyeing.

To allow solubilization of low vapor pressure compounds such as the solid dyes,
both the density and the temperature of CO2 need to be sufficiently high, in the
order of 600 kg/m3 and 120ºC respectively. To obtain this set of conditions, the
pressure needs to be around 300 bar, so that both the temperature and pressure are
far above the critical point (31ºC, 74 bar).

The main advantage of using scCO2 instead of water in a dyeing process, is the
easy separation of the CO2 and the unused dye that remains after the dyeing
process. Depressurization leads to precipitation of the excess dye and gives clean,
gaseous CO2, so that both compounds can be recycled and no waste is generated.
Furthermore, after the dyeing process, the textile does not need an energy-intensive
drying step as it does in aqueous dyeing.

The above advantages are consequences of the high vapor pressure of CO2, but also
the physical properties that are characteristic of all supercritical fluids facilitate the
dyeing process: The intense thermal motion of the molecules and the low viscosity

3
result in a high diffusivity in the fluid, facilitating the mass transport of the dye
towards the fibers.

Additional advantages of scCO2 exist for the case of polyester dyeing. Because
scCO2 is a non polar solvent, no dispersing agents are needed when non polar dyes
are used. This means that simpler dye formulations can be used than in aqueous
polyester dyeing where dispersing agent makes up around 50% of the dye powder.
Another advantage specifically for polyester is that under supercritical conditions
the CO2-molecules penetrate and swell the polymer. This plasticizes the textile
fibers and increases diffusion coefficients of dyes inside the polyester with one
order of magnitude [9], relative to aqueous dyeing.

1.4. Textile dyeing in scCO2 – state of the art

1.4.1. Process

In supercritical textile dyeing, the choice of dye depends on the dyeing mechanism
which, in its turn, depends on the type of fiber that is to be dyed. For non polar
textiles, such as polyester, non polar, non reactive dyes are dissolved into the
supercritical phase, transported to the fiber and adsorbed to the surface. Finally, the
dye molecules diffuse into the CO2-swollen polymer matrix, where they are bound
to the polyester molecules by physical attraction, mainly dispersion forces. Upon
depressurization, the CO2-molecules exit the shrinking fibers and the dye
molecules are retained. This dyeing mechanism was suggested by Saus et al. [10]
and subsequently confirmed by Tabata et al. [11]. Because these non polar dyes are
also used in aqueous dyeing processes, where they are dispersed in the water, they
are generally called disperse dyes.

The dyeing mechanism and therefore the type of dye is different for natural
textiles. Since these are polar due to the presence of hydroxyl groups (cotton) or
amino groups (wool, silk), they have no physical attraction towards disperse dyes.
To bind a dye molecule to such a fiber, a chemical bond is needed to realize
sufficient fastness, for which reason reactive dyes have to be used.

4
As is the case for polyester, for natural textiles the currently available dyes that are
used in water can be used in scCO2. However, these reactive dyes are polar and
therefore not soluble in scCO2. It has been shown by Jun et al. [12] that it is
possible to dye natural fibers with these polar dyes in scCO2 by using reverse
micelles. This method is not suited for dyeing cotton/polyester blends since the
latter fiber is not capable of forming chemical or physical bonds with the polar
dyes. Another disadvantage is that the use of extra chemicals like dispersing agents
complicates the process and therefore it is preferable to use new, non polar reactive
dyes for natural fibers.

Non polar reactive dyes were used on cotton by Schmidt et al. [13] but the results
were not satisfactory, and an unpractical long dyeing time of 4 hours was needed
and during the process, corrosive byproducts were formed. Other researchers have
tried to dye cotton in scCO2 but did not succeed [7]. The successful development of
a process for the dyeing of cotton in scCO2 was carried out by Fernandez [14].
Since this falls beyond the scope of the present work, it will not be discussed
further. Wool was dyed well in scCO2 with non polar reactive dyes by Schmidt et
al. [15], but silk has never been dyed to an acceptable depth [7].

Despite the research on disperse dyeing of non-polar fibers and the reactive dyeing
of polar fibers, there is by far not enough knowledge in both fields, which currently
limits the further development of supercritical fluid dyeing.

1.4.2. Equipment

It is generally accepted [16, 17] that a supercritical dyeing process should be

operated as is shown in figure 1.2. During the dyeing, the CO2 is circulated through
a heat exchanger, a vessel where the dye is dissolved and through a vessel where
the dye is delivered to the textile. When the desired coloration is attained, dye is
still left in the CO2, which is removed by passing the CO2 through a pressure
reducing valve into a separator vessel. In the separator, the CO2 is gasified, so that
the dye precipitates and the clean CO2 can be recycled by pumping it back to the
dyeing vessel.

Pilot plants such as drawn in figure 1.2 have been constructed [16, 17] but, up to
now, no commercial-size machine has been built. This is caused by the high cost

5
associated with large-scale equipment operating around 300 bar. To facilitate
commercialization, the price of large-scale pressure equipment has to be lowered.
To do this, there are some important engineering issues to be faced.

circulation reducing valve

pump separator

dye
dyeing holder
vessel
DYEING SOLVENT
CYCLE RECYCLE CO2
storage
vessel

heater pressurization pump

Figure 1.2. Simplified flow diagram of a process to dye textiles in supercritical

CO2.

The first issue concerns the pressure vessel containing the textile. Since industrial
dye baths have volumes in the order of 1000 liters, the thick-walled supercritical
dyeing vessels will consist of a large amount of steel compared to aqueous dyeing
vessels. This does not only increase the investment costs, but also the operating
costs because the vessel has to be heated to the dyeing temperature in each batch.
Because more energy is needed for the heating, a larger heat exchanger or a longer
process time is required, which increases either the investment or the operating cost
even more.

The second engineering issue is the circulation of scCO2 through the dyeing vessel.
Centrifugal pumps for static pressures of around 300 bar are expensive and weigh
heavily on the equipment cost [16].

A further complication encountered when designing a supercritical dyeing

machine, concerns heat transfer. Literature gives no clear indications on how to

6
calculate heat transfer coefficients for the design of a heat exchanger for high-
pressure CO2.

The above three problems have not been solved up to now, which hinders the
implementation of supercritical fluid dyeing in industry.

1.5. Aim and contents of this thesis

This work treats the above discussed limiting aspects on both the process (non-
reactive and reactive dyeing) and on the equipment (pressure vessel, circulation
pump and heat transfer).

The aim of this thesis is:

To develop and apply the knowledge required to solve the process and equipment
technology issues that are currently limiting the further development and
implementation of the supercritical textile dyeing technology.

As was discussed above, the dyeing of natural fibers in scCO2 is a research area
where only a start has been made. Therefore, in chapter 2 experiments are
discussed with new, reactive, non polar dyes on several textile types.

Since it was found in chapter 2 that the presence of a small amount of water in the
scCO2 increases the coloration in reactive dyeing, a model was developed which is
described in chapter 3, to predict how much water is to be added to a supercritical
dyeing vessel to obtain optimum coloration.

In chapter 4, the dyeing of polyester is discussed, experiments in a 40-liter pressure

vessel are described and the influence of temperature and CO2-density on the
coloration and distribution coefficient in scCO2 is investigated quantitatively.

Chapter 5 describes computational fluid dynamics simulations of heat transfer to

near- and supercritical CO2, to clarify how a supercritical fluid heat exchanger can
be designed.

7
In chapter 6, the process and equipment design of a 100-liter beam dyeing machine
for supercritical polyester dyeing is treated. A new type of pressure vessel and
centrifugal pump are described, both equipment items are meant to decrease the
process cost significantly.

A short description of a commercial-scale 1000-liter beam dyeing machine is given

in chapter 7, together with an economical comparison of the supercritical and the
conventional aqueous process.

In the epilogue, recommendations are made for further research and the future of
the technology of textile dyeing in supercritical carbon dioxide is discussed.

References

1. E.M. Aizenshtein, World Chemical Fiber and Thread Production in

2003, Fibre Chem. 36 (6), 467 (2004).
2. H.S. Rai, M.S. Bhattacharyya, J. Singh, T.K. Bansal, P. Vats, and U.C.
Banerjee, Removal of Dyes from the Effluent of Textile and Dyestuff
Manufacturing Industry: A Review of Emerging Techniques with
Reference to Biological Treatment, Crit. Rev. Env. Sci. Tec. 35, 219
(2005).
3. U. Rott, Multiple Use of Water in Industry – The Textile Industry
Case, J. Env. Sci. Health, part A – Tox. A38 (8), 1629 (2003).
4. H. Zollinger, Color Chemistry, Synthesis, Properties of Organic Dyes
and Pigments, 2nd ed., VCH Publishers, New York, 92 (1987).
5. T. Robinson, G. McMullan, R. Marchant, and P. Nigam, Remediation
of Dyes in Textile Effluent: A Critical Review on Current Treatment
Technologies with a Proposed Alternative, Bior. Technol. 77, 247
(2001).
6. R.B. Love, Rev. Prog. Color. 6, 18 (1975).
7. E. Bach, E Cleve, and E. Schollmeyer, Past, Present and Future of
Supercritical Fluid Dyeing Technology – An Overview, Rev. Prog.
Color. 32, 88 (2002).
8. G. Brunner, Topics in Physical Chemistry 4: Gas Extraction,
Steinkopf, Darmstadt (1994).

8
9. S. Sicardi, L. Manna and M. Banchero, Comparison of Dye Diffusion
in Poly(ethylene terephtalate) Films in the Presence of a Supercritical
or Aqueous Solvent, Ind. Eng. Chem. Res. 39, 4707 (2000).
10. W. Saus, D. Knittel and E. Schollmeyer, Dyeing of Textiles in
Supercritical Carbon Dioxide, Text. Res. J. 63, 135 (1993).
11. I. Tabata, J. Lyu, S. Cho, T. Tominaga, and T. Hori, Relationship
Between the Solubility of Disperse Dyes and the Equilibrium Dye
Adsorption in Supercritical Fluid Dyeing, Color. Technol. 117, 346
(2001).
12. J.H. Jun, K. Sawada and M. Ueda, Application of Perfluoropolyether
Reverse Micelles in Supercritical CO2 on Dyeing Process, Dyes and
Pigments 61, 17 (2004).
13. A. Schmidt, E. Bach and E. Schollmeyer, in: E. Reverchon (Ed.), Proc.
6th Conf. Supercrit. Fluids Appl., Maiori, Italy, 557 (2001).
14. M.V. Fernandez Cid, Cotton Dyeing in Supercritical Carbon Dioxide,
Delft University of Technology, Dissertation, Delft (2005).
15. A. Schmidt, E. Bach and E. Schollmeyer, The Dyeing of Natural
Fibers with Reactive Disperse Dyes in Supercritical Carbon dioxide,
Dyes and Pigments 56 (1), 27 (2003).
16. W.A. Hendrix, Progress in Supercritical CO2 Dyeing, J. Ind. Text. 31
(1), 43 (2001).
17. E. Bach, E. Cleve, E. Schollmeyer, M. Bork P. and Koerner,
Experience with the Uhde CO2-Dyeing Plant on Technical Scale. Part
1: Optimization Steps of the Pilot Plant and First Dyeing Results,
Melliand Int. 3, 192 (1998).

9
10
Chapter 2

Dyeing of Natural and Synthetic Textiles in

Supercritical Carbon Dioxide with Disperse Reactive
Dyes

Abstract

Polyester, nylon, silk and wool were dyed with disperse reactive dyes in
supercritical carbon dioxide (scCO2). The dyes were substituted with either
vinylsulphone or dichlorotriazine reactive groups. Since earlier research showed
that water, distributed over the scCO2 and the textile, increased the coloration,
experiments were done with the vinylsulphone dye with varying amounts of water
in the dyeing vessel, to investigate if there is an optimum water concentration. The
amounts were such, that no liquid water was present. The maximum coloration was
obtained when both the scCO2 and the textiles were saturated with water. At the
saturation point, deep colors were obtained with the vinylsulphone dye for
polyester, nylon, silk and wool, with fixation percentages between 70 and 92%
when the dyeing time was 2 hours. The positive effect of water was due to its
ability to swell fibers or due to an effect of water on the reactivity of the dye-fiber
system. Also the dichlorotriazine dye showed more coloration when the scCO2 was
moist. With this dye, experiments were conducted in water-saturated scCO2,
varying the pressure from 225 to 278 bar and the temperature from 100 to 116ºC.
The coloration of polyester increased with pressure, the results for silk and wool
were not sensitive to pressure. Increasing the temperature had no influence on the
dyeing of polyester, silk and wool. The fixations on polyester, silk and wool, being
between 71 and 97%, were also independent of pressure and temperature.

Contents submitted to the Journal of Supercritical Fluids

11
2.1. Introduction

In current industrial textile dyeing processes, large amounts of wastewater are

produced. This is an environmental burden and, due to the ever more stringent
regulations on water pollution, also an economical problem. The use of scCO2 as
dyeing medium solves this problem: the CO2 and the residual dye remaining in the
dye bath after the process can easily be separated so both can be recycled.
Additional advantages are the high diffusivity and low viscosity of scCO2, which
make the dyeing process faster than in water. The low surface tension allows the
scCO2 to penetrate small pores easier than water. Furthermore, after dyeing no
energy-demanding drying step is needed. These considerations have led to a
considerable research effort on scCO2 dyeing in the last two decades, as was
reviewed by Bach et al. [1].

Textiles can be classified into synthetics (e.g. polyester and nylon) and natural
textiles. The latter category can be divided into proteins (e.g. silk and wool) and
cellulosics (e.g. cotton). In this work, polyester, nylon, silk and wool are
investigated. Cotton is also under investigation in our laboratory but will not be
discussed in this work.

When dyeing in scCO2, non-polar dyes are used to enable dissolution. Polyester
(polyethylene terephtalate or PET) is also non-polar and during the dyeing process,
the dye molecules can diffuse into the polymer matrix, where they are physically
bonded. Because of its non-polarity, polyester can be dyed in scCO2 with non-
reactive, so-called disperse dyes. Nylon is non-polar as well, although slightly
more polar than PET, and it can also be dyed with disperse dyes. Since the nylon
molecules have amino end-groups, it is also possible to use dyes that are able to
react with these nucleophiles, forming a covalent bond, as was shown by Liao et al.
[4]. Silk and wool are polar and therefore have no affinity for the non-polar dye
molecules. It is only possible to dye these textiles in scCO2 when the dyes are
reactive towards the amino groups in the protein fibers of silk and wool. These
non-polar reactive dyes are generally called disperse reactive.

In the disperse dyeing of synthetic textiles, the CO2 penetrates and swells the
fibers, thereby facilitating the diffusion of dye molecules through the polymer.
Upon depressurization, the CO2 molecules exit the shrinking fiber and the dye is

12
retained in the textile. The mechanism of PET dyeing in scCO2 has been discussed
extensively, e.g. by Saus et al. [2] and Tabata et al. [3].

In reactive dyeing, four steps can be distinguished in the process, each of which is
potentially rate determining:
1. Transport of dye to the fibers. The dye powder is dissolved in the scCO2
and transported towards the textile. The solubility of the dye determines
the transport rate of dye towards the textile.
2. Diffusion of dye into the fibers. Natural fibers are porous and their
accessibility needs to be sufficient to allow diffusion of dye molecules into
the pores.
3. Adsorption of dye to the textile. Before a dye molecule can react with a
fiber, it needs to be adsorbed to its surface. The affinity or substantivity of
the dye for the textile determines whether or not this step takes place.
4. Reaction of the dye with the textile. The dye has to form a covalent bond
with the amino groups of the proteins.

Two different dyes are used in this work:

1. Vinylsulphone dye. This dye is chosen because the vinylsulphone group is
able to react with protein fibers in an aqueous dyeing process, according to
Heyna et al. [5]. Fig. 2.1.A gives the reaction of the dye with a textile
amino group.
2. Dichlorotriazine dye. Although the dichlorotriazine group was originally
developed and patented by Rattee et al. [6] for cotton dyeing in alkaline
water by reaction of the dye with the hydroxyl groups of cotton, it is
investigated here if it is also able to react with amino groups of silk and
wool in scCO2. Fig. 2.1.B shows the reaction of a dichlorotriazine dye with
a textile amino group.

It is known from Schmidt et al. [7] that a vinylsulphone dye is able to react with the
functional amino groups of nylon and wool in scCO2 but they found no coloration
of silk. Schmidt et al. [8] also found that a dichlorotriazine dye was unable to dye
silk; they did not report on dyeing of wool with a dichlorotriazine dye. However,
these publications describe dyeing in dry CO2, while the patent by Veugelers et al.
[9] claims that the addition of water enhances coloration when natural fibers are
dyed with reactive dyes in scCO2.

13
 SO2 HC CH2
Et2N N N

textile NH2

SO2 CH2 CH2 NH textile

Et2N N N

A. Vinylsulphone reaction

Cl N Cl textile HN N Cl

N N N N

HN textile NH2 HN
O2 N N N NEt2 O2N N N NEt2
- HCl

B. Dichlorotriazine reaction
Fig. 2.1. Reaction of vinylsulphone (A) and dichlorotriazine (B) dyes with textile
amino sites

It is not yet known why water has this positive effect on the process. One possible
explanation is that it acts as a modifier, enhancing the solubility of the dye. It could
also be that water works as a swelling agent for the polar, natural textiles. As
explained by Kraessig [10], the macromolecules of a natural fiber are kept together
by intermolecular hydrogen bonds. Water breaks up these interactions and allows
the chain molecules to increase their distance, thereby swelling the textile fiber.
Dye molecules can then penetrate more easily into the fibrous structure. It is
expected that nylon, silk and wool are susceptible to this effect of water and
polyester is not. The third possible effect of water on the supercritical dyeing
process is that it somehow participates in the reaction between the dye and the
fiber. There are, however, no reports on this last possibility.

The aim of the present work is to investigate whether a vinylsulphone and a

dichlorotriazine dye are suitable for protein textiles in moist scCO2. For the
vinylsulphone dye, it is determined how much water should be added to the process

14
to obtain maximum coloration. Experiments are done with different dyeing times,
to get an indication of the rate of the process. With the dichlorotriazine dye, the
influence of water addition is also investigated. Furthermore, the effects of pressure
and temperature on the coloration of different textiles in moist scCO2 are
investigated with the dichlorotriazine dye. Finally, by dyeing polyester and nylon
in the same batch as silk and wool it is checked if also blends of synthetics and
natural textiles can be dyed in scCO2 with both dyes.

2.2. Experimental section

2.2.1 Materials

The orange vinylsulphone dye and the purple dichlorotriazine dye, shown in
Fig.1A and 1B were designed for use in scCO2 and synthesized by a well-known
dye manufacturer. Both powders were used as received. The textiles were all dyed
as pieces of cloth. The polyester (polyethylene terephtalate; 120 g/m2) and the
nylon (polyamide 6.6; 150 g/m2) were knitted and free of spinning oils. The wool
(62 g/m2) and the silk (120 g/m2) were woven. Polyester was received from Ames
Europe, the other textiles were from the Center for Test Materials (Netherlands).
The carbon dioxide (99.97 %) was purchased from HoekLoos. The acetone that
was used in the extractions was technical grade from Chemproha. Demineralised
water was used, in the extractions and in the dyeing vessel.

2.2.2. Equipment and procedure

A 4-litre stainless steel autoclave from Uhde Hochdrucktechnik (Germany) was

used as pressure vessel (Fig. 2.2). The temperature of the dyeing process was
determined by setting the temperature of an oil heater and pumping the oil through
the jacket of the dyeing vessel for 2 hours prior to each experiment. After this
preheating step, dye powder (0.2 ± 0.01 g) was placed at the bottom of the vessel,
between two stainless steel filter plates with a pore size of 10 µm, to prevent
entrainment of undissolved dye particles. A piece of cotton (20 ± 0.2 g) was folded
around small pieces of polyester, nylon, silk and wool (each 0.2 ± 0.02 g). The
textile was placed in the dyeing vessel in such a way, that the CO2 was forced to
flow through the layers of textile.

15
 textile
oil heater
autoclave with
heating jacket

CO2 supply
pump oil pump

circulation pump
dye powder

Fig. 2.2. Experimental set-up for textile dyeing in scCO2

In the experiments in moist scCO2, the cotton was wetted with the desired amount
of water prior to placing it in the vessel. The system was then pressurized with an
air-driven plunger pump, from Williams Instrument Company. The CO2 was
pumped around for 2, 4 or 6 hours, at a flow rate of 0.10 ± 0.02 m3/hr, with a
centrifugal pump with magnetic coupling from Autoclave Engineers. The flow
direction was upwards through the vessel: first through the dye and then through
the textiles. Temperature and pressure increased slowly in the first hour but were
constant afterwards (± 1ºC and ± 2 bar). The stable values in the second hour are
the reaction conditions that are mentioned in the presentation of the results below.
In the experiments of 2 and 4 hours, there was dye powder left between the filter
plates. The amount was different for each experiment and varied from 0.3 to 0.1 g.
In the 6 hour experiments, no remaining dye was found.

2.2.4 Color analysis

The dyed pieces of polyester, nylon, silk and wool were analyzed by measuring the
reflectance curve between 350 and 750 nm with a portable spectrophotometer from
Avantes. The minimum of the curve (Rmin) was used to determine the ratio of light
absorption (K) and scatter (S) via the Kubelka-Munk function [10]:

(1 − R min ) 2
(K/S )dyed = (1)
2R min

16
The ratio K/S is used as a measure of coloration in textile industry and also in this
paper. After this analysis, each sample was stripped of unfixed dye by Soxhlet
extraction with a 50 weight% solution of acetone in water for 30 min. The K/S-
value of the extracted textile (K/S)extr was determined and used to calculate the
percentage of dye molecules that was fixed to the textile (F):

F = (K / S )extr (K / S )dyed * 100 % (2)

2.3. Results and discussion

2.3.1 Vinylsulphone dye

The influence of water addition was investigated in a series of 5 experiments.

Different amounts of water were added and the equilibrium distribution of water
was calculated with a model that is presented elsewhere (Van der Kraan et al. [11]).
This calculation was done neglecting the water-uptake of silk, wool, polyester and
nylon, since the mass of cotton in the vessel is 97% of the total mass of textiles.
The model allows calculation of the relative humidity RH of the scCO2 from
pressure, temperature, vessel volume, textile mass and amount of water added to
the vessel. For the 5 experiments done in the vinylsulphone series, the relative
humidities were: 4, 25, 51, 73 and 97 %. Earlier experiments at our laboratory have
shown that the presence of liquid water in the dyeing vessel led to staining of the
textile. Therefore, in this work, RH < 100% in all experiments. The amounts of
water that were absorbed by the textile from the air, prior to the experiment, were
measured gravimetrically and taken into account in the calculation of RH.
Therefore, when no water was added to the dyeing vessel, a relative humidity of
4% was obtained. Although the results (Fig. 2.3) are given as a function of the
relative humidity RH of the scCO2, it could just as well be the relative humidity of
the textile that plays a role in the dyeing process. However, since a rise in CO2
humidity also means a rise in the humidity of the textile, it effectively makes no
difference which humidity is treated as the independent variable.

17
 100 100
80 fixation 80

60 60

F (%)
K/S
A. Polyester
40 40
(K/S)dyed
20 20
(K/S)
0 0
0 20 40 60 80 100
RH (%)

100 100
80 fixation 80
(K/S)
60 60

F (%)
B. Nylon K/S
40 (K/S)extr 40
20 20
0 0
0 20 40 60 80 100
RH (%)

100 100
80 80
fixation
60 60

F (%)
K/S

C. Silk
40 (K/S) 40
20 (K/S) 20
0 0
0 20 40 60 80 100
RH (%)

100 100
fixation
80 80
60 60
F (%)
K/S

D. Wool
(K/S)dyed
40 40
20 (K/S) 20
0 0
0 20 40 60 80 100
RH (%)

Fig. 2.3. Influence of relative CO2-humidity (RH) on the coloration (K/S) and
fixation (F) of textiles dyed in scCO2 with a vinulsulphone dye during 2 hours at
230 bar and 112ºC, for polyester (A), nylon (B), silk (C) and wool (D)

18
The 5 experiments in this series were all done with the vinylsulphone dye, in 2
hours dyeing at a pressure of 230 bar and a temperature of 112ºC. The coloration
and fixation of polyester were independent of humidity. A dark orange color was
obtained before ((K/S)dyed = 25.5 ± 1.5) and after Soxhlet extraction ((K/S)extr =
19.5 ± 1.5), the fixation percentage was 75 ± 5 %. Nylon was also dyed well. Fig.
2.3.B shows an increase of coloration with humidity, but also in dry CO2 nylon
could be colored deeply (K/S = 20). The fixation percentages showed a slight
increase with humidity. The results for silk show that no coloration was obtained
without water addition, as was found by Schmidt et al. [7]. However, Fig. 2.3.C
shows that, as the humidity increased, so did the coloration and the fixation of silk.
Wool gave light colors in dry scCO2 but, as can be seen in Fig. 2.3.D, a higher
coloration and fixation were obtained when water was added.

For these amino containing textiles, the experiments gave maximum colorations
when the textile and the scCO2 were nearly saturated (RH = 97%). To investigate
the influence of dyeing time, the experiments were repeated at the same pressure
and temperature but with 4 and 6 hours dyeing time, with relative scCO2 humidity
RH = 97%. It was found that the polyester coloration before and after extraction
increased between 2 and 4 hours from (K/S)extr = 19.5 to 49 (Fig. 2.4.A). Between 4
and 6 hours, the coloration did not increase any further for polyester. This indicates
that the polyester was saturated with dye molecules after 4 hours. Also the fixation
(84 ± 1.5) does not change significantly between 4 and 6 hours.

Although silk is colored well by both dyes in moist scCO2, it remains white when
dyeing in dry scCO2, despite the fact that dye is dissolved in the scCO2, as is
indicated by the dark coloration of polyester in dry scCO2. It can be stated from
this that the positive effect of water on the coloration of the amino containing
textiles is not due to an action of water as a solubility enhancer. Because water is a
polar component, it can also not enhance the substantivity between the dye and the
fiber. An increase in fixation with water content, as observed with the amino
containing textiles, indicates that water plays a role in the reaction between the dye
and the fiber. Whether or not the accessibility of the fibers is enhanced by the

19
 100
80
60

K/S
(K/S)dyed
40
(K/S)extr
20
0
2 4 6
dyeing time (hours)
100
80
60
K/S

40
20
0
2 4 6
dyeing time (hours)
100
80
60
K/S

40
20
0
2 4 6
dyeing time (hours)

100
80
60
K/S

40
20
0
2 4 6
dyeing time (hours)

Fig. 2.4. Coloration of textiles with a vinylsulphone dye in scCO2 at 230 bar,
112ºC and different CO2-humidities (RH) as a function of time, for polyester (A),
nylon (B), silk (C) and wool (D)

20
swelling effect of water, does not follow from the results, but can also not be ruled
out.

For the amino containing textiles, the results given in Fig. 4B-D show that
coloration after extraction increased with time up to 6 hours; after 4 hours nylon,
silk and wool were not yet saturated with dye. Fig. 2.4.B-D also reveal that, as the
dyeing time of the amino containing textiles progressed, the fixation increased.
This reflects the dyeing mechanism described above: First the dye adsorbs to the
textile surface and, after that, it reacts with the amino sites.

2.3.2 Dichlorotriazine dye

Firstly, two experiments were done with the dichlorotriazine dye: one in dry scCO2
(RH = 4%) and one in moist, almost saturated scCO2 (RH = 97%). As can be seen
in table 2.1, the results correspond qualitatively with those of the vinylsulphone
experiments for all textiles: The coloration before and after extraction and the
fixation of polyester were independent of water addition. The coloration after
extraction of silk was negligible in dry scCO2 but increased with relative scCO2-
humidity, the fixation on silk increased with water addition. Wool could be colored
lightly in dry scCO2 and darker in moist scCO2, the fixation on wool increased
when water was added. As was the case for the vinylsulphone dye, for the
dichlorotriazine dye it follows that the reactivity is increased by water addition.
Whether or not the accessibility of the fibers is also increased does not follow from
the results.

Table 2.1. Influence of relative scCO2-humidity RH on the coloration before

((K/S)dyed) and after ((K/S)extr) Soxhlet extraction and the calculated fixations F of
textiles dyed with dichlorotriazine dye in scCO2 at 250 bar and 100ºC for 2 hours
Textile (K/S)dyed (K/S)extr F (%)
RH = 4% RH = 97% RH = 4% RH = 97% RH = 4% RH = 97%
Polyester 10.1 10.3 10.1 10.1 100 98
Silk 0.44 4.24 0.25 3.22 57 76
Wool 4.26 5.79 2.47 4.05 58 70

With the dichlorotriazine dye, the influence of pressure and temperature on the
coloration and fixation of polyester, silk and wool was investigated. In all of these

21
experiments, the relative humidity of the scCO2 was 97%. For the calculation of the
amounts of water in the textile and the scCO2, i.e. to calculate how much water had
to be added, the same procedure as above was followed, taking only the mass of
the cotton into account. Because the distribution of water between cotton and
scCO2 depends on temperature and pressure, different amounts of water were
added in each experiment, varying from 22 g at (225 bar, 100ºC) to 30 g water at
(250 bar, 116ºC), to reach the same CO2-humidity of 97%.

In table 2.2 the results are given for varying pressures (from 225 to 278 bar) at
100ºC and 2 hours dyeing time. The colorations (K/S)dyed and (K/S)extr of polyester
were good and increased with pressure at constant temperature, as was reported by
Chang et al. [12] for disperse non-reactive dyes. The increase of dye uptake with
pressure is caused by enhanced dye solubility and by increased swelling of the
polyester. Although less than polyester, silk and wool were dyed well and the
(K/S)dyed and (K/S)extr varied within the experimental error range, i.e. the
colorations were independent of pressure. This is clear from Fig. 2.5, which gives a
graphical representation of the influence of pressure on the coloration after
extraction. Table 2.2 shows that the fixations on polyester, silk and wool are
independent of pressure.

Table 2.2. Influence of pressure on the coloration before ((K/S)dyed) and after
((K/S)extr) Soxhlet extraction and the calculated fixations F of textiles dyed with
dichlorotriazine dye in scCO2 at 100ºC for 2 hours

P polyester silk wool

bar (K/S)dyed (K/S)extr F (K/S)dyed (K/S)extr F (K/S)dyed (K/S)extr F

% % %
225 6.83 5.95 87 3.91 3.57 91 3.29 2.65 81
235 8.83 8.73 99 4.60 4.32 94 4.71 3.52 75
250 10.4 10.1 97 4.73 4.05 86 4.40 3.22 73
260 10.7 9.82 92 4.66 4.31 92 4.05 3.20 79
265 11.5 11.2 97 4.02 3.60 90 4.68 4.00 85
275 11.5 10.9 95 4.60 4.05 88 3.73 2.94 79

22
Table 2.3 shows that, when temperature was varied from 100 to 116ºC at constant
pressure (250 bar), the colorations of polyester, silk and wool before and after
extraction showed no significant changes. This is also clear from Fig. 2.6, where
the colorations after extraction are presented graphically. The fixation percentages
of polyester, silk and wool showed no dependency on temperature. Although the
K/S-values for polyester were much higher in all the dichlorotriazine experiments,
the K/S-values for silk and wool, being between 3 and 4, also corresponded with a
dark purple color.

12
10
8 polyester
(K/S)extr

silk
6
wool
4
2
0
225 235 245 255 265 275
p (bar)

Fig. 2.5. Influence of pressure on the coloration (K/S)extr of polyester, silk and wool
dyed for 2 hours at 100ºC with dichlorotriazine dye in scCO2 with relative humidity
RH = 97%.

Table 2.3. Influence of temperature on the coloration before ((K/S)dyed) and after
((K/S)extr) Soxhlet extraction and the calculated fixations F of textiles dyed with
dichlorotriazine dye in scCO2 at 250 bar for 2 hours

T polyester silk wool

ºC (K/S)dyed (K/S)extr F (K/S)dyed (K/S)extr F (K/S)dyed (K/S)extr F
% % %
100 10.4 10.1 97 4.66 4.05 87 4.19 3.22 89
105 11.2 11.0 98 4.29 3.98 93 4.00 3.53 88
111 10.9 10.9 100 4.26 3.85 90 4.40 3.64 83
116 11.8 11.5 97 5.31 4.80 90 4.46 4.05 91

23
Pressure and temperature are the two parameters that determine the solubility of
dye in scCO2. Therefore, the negligible influence of both parameters on the
coloration of the amino containing textiles suggests that the solubility is not the
rate determining factor in the process. However, since it is not known how strong
the solubility varies with pressure and temperature, this cannot be stated with
certainty. Research is needed on the influence of pressure and temperature on the
solubility of this particular compound in scCO2.

14
12
10 Polyester
(K/S)extr

8 Silk
6 Wool

4
2
0
100 105 110 115
temperature (ºC)
Fig. 2.6. Influence of temperature on the coloration (K/S)extr of polyester, silk and
wool dyed for 2 hours at 250 bar with dichlorotriazine dye in scCO2 with a relative
humidity of 97%

The observation that temperature has negligible influence on coloration also

suggests that the reaction rate is not rate determining. However, since no
information is available on the kinetics of this reaction in scCO2, this cannot be
positively concluded. Additional research is needed on the kinetics of the dye-fiber
reaction.

2.4. Conclusions

Disperse dyes containing a reactive vinylsulphone or a dichlorotriazine group, are

suitable for dyeing textiles containing polyester, nylon, silk, wool or blends of
these fibers in supercritical carbon dioxide. The dye uptake by polyester is

24
independent of water addition. For the amino-containing textiles, the coloration
increases with the concentration of water in the scCO2 and the textiles.
Experiments with the vinylsulphone dye show that maximum coloration of nylon,
silk and wool is obtained when both the scCO2 and the textiles are saturated with
water. The fixation percentage of vinylsulphone dye on polyester was 75% and
independent of pressure. Fixations on nylon, silk and wool increased with relative
humidity of the scCO2. In saturated scCO2, values are reached in 2 hours of 94, 85
and 90%, respectively. When dyeing with the dichlorotriazine dye in scCO2
saturated with water, the fixations of polyester, silk and wool are 93, 88 and 79%.

When the dyeing time is varied for the vinylsulphone dye, the coloration of the
amino containing textiles increases as expected. Also the fixation grows in time,
indicating that more and more of the dye molecules present in the textile are
covalently bonded to the amino sites.

The positive effect of water on the dyeing process is caused either by water
facilitating the chemical reaction between the dye and the fiber, or by water acting
as a swelling agent for the textiles. Whatever the reason for this phenomenon, it is
concluded that water should be added in supercritical dyeing of nylon, silk and
wool with disperse dyes containing vinylsulphone or dichlorotriazine reactive
groups.

The experimental results on the dichlorotriazine dye suggest that the solubility and
the reactivity can be ruled out as the rate-determining step of the dyeing process.
However, for this conclusion to be made with certainty, more research is needed on
solubility and reaction kinetics.

References

1. E. Bach, E. Cleve and E. Schollmeyer, Past, Present and Future of

Supercritical Fluid Dyeing Technology – an Overview, Rev. Prog. Color.
32, 88 (2002).
2. W. Saus, D. Knittel and E. Schollmeyer, Dyeing of Textiles in
Supercritical Carbon Dioxide, Text. Res. J. 63 (3), 135 (1993).

25
 3. I. Tabata, J. Lyu, S. Cho, T. Tominaga and T. Hori, Relationship between
the Solubility of Disperse Dyes and the Equilibrium Dye Adsorption in
Supercritical Fluid Dyeing, Color. Technol. 117, 346 (2001).
4. S.K. Liao, Y.C. Ho, P.S. Chang, Dyeing of Nylon 66 with a Disperse-
Reactive Dye using Supercritical Carbon Dioxide as the Dyeing Medium,
J. Soc. Dyers and Color. 116, 403 (2000).
5. J. Heyna, W. Schumacher and Hoe, German P 925,902 (1949).
6. I.D. Rattee, W.E. Stephen and ICI, BP 772,030 774,925 871,930 (1954).
7. A. Schmidt, E. Bach and E. Schollmeyer, Use of Fiber Reactive Groups in
Supercritical Carbon Dioxide, Melliand Textilberichte 83 (9), 648 (2002).
8. A. Schmidt, E. Bach and E. Schollmeyer, The Dyeing of Natural Fibers
with Reactive Disperse Dyes in Supercritical Carbon Dioxide, Dyes and
Pigments 56 (1), 27 (2003).
9. W.J.T. Veugelers, H. Gooijer, J.W. Gerritsen and G.F. Woerlee, European
Patent EP 1 126 072 A2, 2001.
10. R. McDonald (Ed.), Colour Physics for Industry, 2nd ed., Society of Dyers
and Colourists, Bradford (1995).
11. M. van der Kraan, M.V. Fernandez Cid, G.F. Woerlee, W.J.T. Veugelers,
C.J. Peters and G.J. Witkamp, Equilibrium Distribution of Water in the
Two-Phase System Supercritical Carbon Dioxide-Textile, J. Supercrit.
Fluids, submitted for publication.
12. K.-H. Chang, H.-K. Bae and J.-J. Shim, Dyeing of PET Textile Fibers and
Films in Supercritical Carbon Dioxide, Korean J. Chem. Eng. 13 (3), 310
(1996).

26
Chapter 3

Equilibrium Distribution of Water in the Two-phase

System Supercritical Carbon Dioxide - Textile

Abstract

When natural fibers are dyed in supercritical carbon dioxide, the addition of a small
amount of water increases coloration. For a process design it is important to know
how much water has to be added to obtain a desired humidity of both textile and
carbon dioxide. In this work a thermodynamic model is proposed to calculate the
distribution of water over the textile phase and the supercritical phase as a function
of pressure and temperature. The phase equilibrium is described with Raoult’s law
for non-ideal fluids. The absorbed water in the textile is a condensed phase and is
modeled here as a non-ideal liquid, using the NRTL-equation. The non-ideality of
the supercritical phase is described by a solubility enhancement factor, a new
equation derived from statistical thermodynamics. Although the model is applied to
cotton, viscose, silk and wool, it can be used for all water absorbing textiles.

Contents submitted to the Journal of Supercritical Fluids

27
Symbols

A Antoine constant [MPa]

b, b1, b2, b12 Van der Waals volume [m3/mole]
B Antoine constant [K]
c12, c21 NRTL-parameters [J/mole]
C Antoine constant [K]
g NRTL-parameter [-]
G12, G21 NRTL-parameters [-]
h Planck’s constant [Js]
k Boltzmann’s constant [J/K]
L relative scCO2-humidity [-]
m particle mass [kg]
mCO2 mass of scCO2 [kg]
m CO 2
H 2O mass of water in scCO2 [kg]
m Htext2 O mass of water in textile [kg]
mH2O total mass of water in dyeing vessel [kg]
mtext mass of textile [kg]
MH2O, MCO2 molar mass of water and carbon dioxide [g/mole]
n number of particles [-]
Ni α molar concentration of component i in phase α [mole/m3]
NA Avogadro’s number [mole-1]
p pressure [MPa]
p sat
H 2O water vapor pressure [MPa]
qr,v rotational and vibrational part of partition function [-]
r distance [m]
R universal gas constant [J/(moleK)]
T temperature [K]
U potential energy [J]
V volume [m3]
Vm molar volume [m3/mole]
Vex excluded volume [m3]
x regain [kg water/kg dry textile]
xmax regain in saturated textile [kg water/kg dry textile]
y mass fraction of water in the supercritical phase [-]

28
Y mole fraction of water in scCO2 [-]
ysat water mass fraction in saturated scCO2 [-]
Z partition function [-]

αmax constant [-]

β = 1/(kT) [J-1 ]
γH2O activity coefficient of water in textile [-]
Γ enhancement factor in terms of density [-]
ε Lennard-Jones parameter [J]
θH2O water content of textile, relative to saturation [-]
θtext textile fraction in wet textile [-]
κ constant [Km3/kg]
λ constant [K]
µ chemical potential [J/mole]
ρ density [kg/m3]
ρCO2 density of scCO2 [kg/m3]
ρtext density of textile [kg/m3]
ρ CO
max
2 maximum density of CO2 [kg/m3]
ρ sat
H 2O water vapor density at vapor pressure [kg/m3]
σ, σ1, σ2, σ12 Lennard-Jones parameter [m]
τ12, τ21 NRTL-parameters [-]
Φ non-ideality factor in terms of pressure [-]

3.1. Introduction

Using supercritical carbon dioxide (scCO2) as a solvent for dyeing textiles has been
the subject of several researchers in the last few years, as was reviewed by Bach et
al. [1]. Typical dyeing conditions are 373 K and 30 MPa. Veugelers et al. [2] and
Van der Kraan [3] reported that in reactive dyeing of natural textiles the addition of
a small amount of water to the dyeing vessel increased the uptake of dye.

The positive effect of water can be attributed to one or more of the following three
causes. Firstly, water can act as a modifier, increasing the solubility of dye in

29
scCO2. Secondly, water can participate in the reaction of dye with the textile
reactive sites. This option has not yet been investigated. Thirdly, water can have an
influence on the structure of the textile. Natural textiles consist of proteins (e.g. silk
and wool) or cellulose (e.g. cotton). These molecules contain amine or hydroxyl
groups that are capable of forming intermolecular hydrogen bonds that keep the
protein or cellulose chains and, therefore, the whole textile structure together.
When water is added to the textile, it breaks up the hydrogen bonds between the
chains, forms its own hydrogen bonds with the amine or hydroxyl groups and takes
position between the chains. This means that the chains are driven apart and that
the textile volume increases, i.e. water acts as a swelling agent (Nevell [4]). The
swollen structure of the textile allows dye molecules to penetrate the textile and,
therefore, has a positive effect on a dyeing process.

For a process design, it is important to be able to calculate the desired amount of

water in a specific dyeing batch. When the cause of coloration improvement by
water addition is the enhancement of dye solubility, the water content of the scCO2
is the factor to be calculated. When coloration is improved by participation of
water in the reaction or by the swelling effect, the humidity of the textile is the
factor of interest. A model is needed to calculate the water content of both the
textile and the scCO2.

In a dyeing vessel, the water is distributed over the textile and the supercritical
phase. In this work a thermodynamic equilibrium distribution is modeled, taking
into account the non-ideal behavior of water in textile and in scCO2. To quantify
the interactions water-textile and water-scCO2, equations are derived that are fitted
to experimental data from literature. The resulting thermodynamic model enables
calculation of how much water has to be added to a supercritical dyeing process to
obtain a desired water content in the textile and in the scCO2. An example is given
showing how to apply the model in the case that the humidity of the textile is the
factor to be calculated. The case that the humidity of the scCO2 is to be calculated
is an analogous procedure. The model is developed for cotton, silk, wool and
viscose but can also be used for other water absorbing textiles.

30
3.2. Thermodynamic model

3.2.1. Equilibrium water distribution

A dyeing vessel is considered containing textile, scCO2 and water. All water is
dissolved in either the textile or the scCO2; no liquid water is present. The textile is
modeled here as a homogeneous phase with ntot sites available for water adsorption,
of which nH2O are occupied by adsorbed water molecules. Since the water
molecules that are adsorbed to the textile sites are in fact a condensed phase, they
are modeled here as a non-ideal liquid. When the mole fraction of water in the
textile is defined as θH2O = nH2O/ntot, the equilibrium distribution of water over the
textile and the scCO2 can be described by Raoult’s law for non-ideal fluids (Smith
and Van Ness [5]):

H 2 O = YΦp
θ H 2 O γ H 2 O p sat (1)

where γH2O is the activity coefficient of water in the textile, p sat

H 2 O is the vapor

pressure of pure water, Y is the mole fraction of water in the scCO2, p is the
pressure inside the dyeing vessel and Φ is the factor describing the non-ideality of
the supercritical phase. Φ is a function of the fugacity coefficient of water in scCO2
and of the Poynting factor and therefore a complex function of pressure and
temperature [5].

In this study, it is convenient to write Eq. (1) in terms of mass and mass density
instead of moles and pressure:

H 2 O = yρ
θ H 2 O γ H 2 O Γρ sat (2)

where ρ sat
H 2 O is the mass density of pure water at the vapor pressure, ρ is the mass

density of the supercritical phase, y is the mass fraction of water in the supercritical
phase and Γ is the factor describing the non-ideality of the supercritical phase. The
advantage of using mass densities instead of pressures is that Γ is a simple function
of density and temperature (see Eq. (19)). Eq. (2) is the basis upon which the
thermodynamic model is constructed.

31
The first term in Eq. (2), θH2O, is defined in Eq. (1) as a mole fraction but is equal
to the following mass fraction:

x
θ H 2O = (3)
x max

where x is the regain, a term used in textile industry to indicate textile humidity. Its
maximum value xmax is the regain in saturated scCO2. The regain is defined as the
mass of water in the wet textile m Htext2 O relative to mass of dry textile m text :

m Htext2 O
x= (4)
m text

Inspection of Eq. (3) and Eq. (4) reveals that θH2O is the same in terms of mass as in
terms of moles.

If the total mass of water distributed over the textile and the scCO2 is mH2O and the
mass of water in the scCO2 is m CO 2
H 2 O , then the water mass balance is:

m H 2 O = m Htext2 O + m CO 2
H 2O (5)

with the amount of water dissolved in the scCO2 given by:

m CO 2 CO 2
(
H 2 O = y m H 2 O + m CO 2 ) (6)

where y is the mass fraction of water in the supercritical phase and mCO2 denotes
the mass of scCO2 in the dyeing vessel, calculated from the volume of the vessel,
the density’s of water and textile (ρCO2 and ρtext) and the mass of (dry) textile mtext:

m text
m CO 2 = ρ CO 2 (V − ) (7)
ρ text

32
The values for the textile density ρtext, taken from Morton and Hearle [6], are given
in table 3.1. The density ρCO2 needed in Eq. (7) is taken from the IUPAC tables
(Angus et al. [7]). The density ρCO2 is also used for ρ in Eq. (2) because the
solubility of water in scCO2 is small and therefore it is assumed that it has
negligible influence on the CO2-density: according to Evelein et al. [8], the value of
y at typical dyeing conditions of 30 MPa and 373 K is y sat = 10-2.

Table 3.1. Density ρtext of dry cotton, silk, wool and viscose rayon from Morton and
Hearle [6]
Textile material ρtext (kg/m3)
cotton 1550
silk 1340
wool 1300
viscose rayon 1520

The terms in Eq. (2) and Eq. (3) that remain to be discussed represent the
thermodynamics of the water-textile mixture (activity coefficient γH2O and
maximum regain xmax) and the thermodynamics of the water-scCO2 mixture (non-
ideality factor Γ).

3.2.2. Thermodynamics of the system water-textile

maximum regain xmax

The maximum regain xmax is defined as the humidity of a textile that is in
equilibrium with air that is saturated with water. In such a situation the chemical
potential of the adsorbed water is equal to the chemical potential of water in
saturated air which, in its turn, is equal to the chemical potential of liquid water.
When not air but high-pressure CO2 is regarded, the situation remains the same: the
chemical potential of the adsorbed water is the same as in liquid water. Since the
latter is independent of pressure and the surrounding medium, it can be stated that
the chemical potential of adsorbed water is the same in air as in scCO2, at the same
temperature. Because the chemical potential depends on the water content of the
textile, also the maximum regain xmax is the same in air as in scCO2. In this work,
experimental data measured in air are used to fit an equation for xmax that is to be
used in supercritical dyeing.

33
Wiegerink [9] observed that in humid air at atmospheric pressure the logarithm of
the regain x in silk, wool, cotton and viscose is inversely proportional to
temperature for a given relative air humidity. In the case of saturated air the water
regain is at its maximum xmax :

λ
x max = α max exp (8)
T

where αmax and λ are constants and T is temperature in K. Below the saturation
point of the textile, Eq. (8) is valid for calculating x, with a different pre-
exponential factor (α) but with the same λ [9]. Therefore, the ratio x/xmax (= θH2O)
is independent of temperature.

Eq. (8) can be used for supercritical dyeing with parameters determined by fitting
the equation to data from Wiegerink who measured the regains of cotton, silk, wool
and viscose as a function of relative air humidity for several temperatures. Taking
from Wiegerink [9] data points (T, xmax) allows calculation of the parameters αmax
and λ for a textile material (table 3.2). For a given temperature, the value of xmax
can now be calculated from Eq. (8).

Table 2: Empirical parameters of Eq. (8), calculated from Wiegerink [9]

Textile material αmax/10-2 λ (K)
mercerized cotton 0.97 1031
degummed silk 2.27 711
clothing wool 3.57 625
viscose rayon 4.33 594

activity coefficient
The activity coefficient γH2O in Eq. (2) follows from the NRTL equation (Prausnitz
et al., [10]):

 G 21 G 12 
ln γ H 2 O = θ 2text  τ 21 ( ) 2 + τ12  (9)
 θ H 2 O + θ text G 21 (θ text + θ H 2 O G 12 ) 2 

34
where θtext ≡ 1-θH2O. The four parameters τ12, τ21, G12 and G21 are determined by
three independent parameters c12, c21 and g:

c12 c 21
τ12 = τ 21 = (10)
RT RT

G 12 = exp(−gτ12 ) G 21 = exp(−gτ 21 ) (11)

It has been shown earlier that the chemical potential of water in textile is the same
in air as in scCO2. Therefore, also the activity coefficient is independent of the
surrounding medium. The NRTL-equation can be used for scCO2 when it is fitted
to Wiegerink’s data [9] measured in air. This is done by defining a relative scCO2-
humidity L as:

yρ CO 2
L= (12)
Γρ sat
H 2O

and combining Eq. (2), (3) and (12) to obtain:

x
θ H 2O γ H 2O = γ H 2O = L (13)
x max

Taking points (L,x) from Wiegerink [9] for a certain temperature, calculating xmax
from Eq. (8) and using Eq. (13), allows calculation of points (θH2O,γH2O ) that are
used to fit the NRTL-equation, i.e. to determine the parameters G12, G21, τ12 and
τ21. Fig. 3.1 shows that temperature has a negligible influence on the activity
coefficient of water in cotton, silk, wool and viscose, from 347 to 377 K.

The observed independence of the activity coefficient on temperature is not seen in

the formal NRTL-equation (see Prausnitz et al. [10]), in which temperature-
dependent mass- or mole fractions are used instead of normalized mass fractions
θH2O and θtext that are temperature-independent. The temperature-dependence
introduced by Eq. (10) is concluded to be negligible between 347 and 377. In the
practice of textile dyeing the temperature-independence has the advantage that for

35
each textile type one set of NRTL-parameters can be used for all dyeing
temperatures between 347 and 377 K.

mercerised cotton degummed silk

γH2O γH2O
1 1

G12 = 1.28 τ12 = 20.23 G12 = 1.36 τ12 = 14.78

G21 = 0.72 τ21 = -17.68 G21 = 0.65 τ21 = -21.04
0 0
0,0 0,5 1,0 0,0 0,5 1,0
0.0 0.5 θH2O 1.0 0.0 0.5 θH2O 1.0

clothing wool viscose rayon

γH2O γH2O

1 1

G12 = 1.26 τ12 = 12.73 G12 = 1.25 τ12 = 24.45

G21 = 0.73 τ21 = -17.31 G21 = 0.75 τ21 =-31.59
0 0
0,0 0,5 1,0 0,0 0,5 1,0
0.0 0.5 1.0 0.0 0.5 1.0
θH2O θH2O

Fig. 3.1. Activity coefficient of water in textile (γH2O ) as a function of relative

textile humidity (θH2O), for 377 K (♦) and 347 K ( ϭ ) and the NRTL-equation ()
with the fitted parameters G12, G21, τ12 and τ21

3.2.3. Thermodynamics of the system water-supercritical CO2

For calculation of the water mass concentration ρ sat

H 2 O in Eq. (2), the Antoine

equation is the starting point:

 B 
H 2 O = A exp
p sat  (14)
T −C

36
with temperature T expressed in K and the constants taken from Poling et al. [11]:
A = 1.32⋅104 MPa, B = -3884 K and C = 43.0 K. Eq. (14) is combined with the
ideal gas law:

ρ sat
H 2O =
p sat H 2O
RT (15)
1000M H 2 O

where MH2O is the molar mass of water, R is the universal gas constant and the
factor 1000 is introduced to match the units. This allows elimination of vapor
pressure to obtain ρ sat
H 2 O , the density of pure gaseous water at the vapor pressure:

1000AM H 2 O  B 
H 2O =
ρ sat exp  (16)
RT T −C

Eqs. (14) to (16) describe the behavior of ideal gases. For a supercritical phase, the
factor Γ in Eq. (2) corrects for the non-ideality in the fluid. Eq. (2) can be rewritten
as:

yρ CO 2
Γ= (17)
θ H 2 O γ H 2 O ρ sat
H 2O

The behavior of the supercritical phase is independent of the properties of the

condensed phase, or: Γ is independent of θ2O and γH2O. Furthermore, Γ is
independent of y because, as was shown above, the solubility of water in scCO2 is
low. From these considerations it follows that Γ is the same below and at
saturation; for Γ one can substitute in Eq. (2) the value calculated with:

y sat ρ CO 2
Γ= (18)
ρ sat
H 2O

Inspection of Eq. (18) reveals that Γ is the water concentration in scCO2 (= yρCO2)
relative to the water concentration in the absence of scCO2 (= ρ sat
H 2 O ) , i.e. Γ

describes the extra solubility of water in the gas (or supercritical) phase that is

37
caused by non ideal behavior, or by the interaction of water molecules with CO2-
molecules. Therefore, Γ is called here the solubility enhancement factor.

The following expression, derived in appendix A, is used to calculate the

enhancement factor:

y sat ρ CO 2  ρ CO 2   κρ CO 2 b1 ρ CO
max
2 ρ CO 2

Γ= = 
 1 − 
max 
exp
 −  (19)
ρ H 2O
sat
 ρ CO 2   T M CO 2 ρ CO 2 − ρ CO 2 
max

where ρ CO
max 3
2 = 1560 kg/m is the maximum (solid state) density of CO2 (Perry and

Green [12]), MCO2 = 44.0 g/mole is the molar mass of CO2. As is shown in
appendix A, b1 represents the molar excluded volume and can be taken as the Van
der Waals volume b1 = 0.0305 liter/mole (Lide [13]). In Eq. (19), κ is the only
parameter that is to be determined by fitting the equation to literature data. Evelein
et al. [8] give water solubilities in scCO2 as a function of pressure for several
temperatures. These data are used to determine κ = 2.08 K⋅m3/kg and to construct
the fitted curves shown in Fig. 3.2. The solubilities predicted by Eq. (19) are in
good agreement with the data from literature, especially at typical dyeing
conditions of 373 K and 30 MPa. This proves that the approximations introduced in
the derivation of Eq. (19), in appendix A, are justified, at least for the system H2O-
CO2.

38
 100

323 K, fit
323 K, data
348 K, fit

enhancement factor
348 K, data
373 K, fit
10 373 K, data
423K, fit
423K, data

1
0 100 200 300 400 500 600 700 800 900
carbon dioxide density (kg/m3)

Fig. 3.2. Fitted curves of the enhancement factor ( y sat ρ CO 2 / ρ sat

H 2 O ) as a function of

pressure for several temperatures

3.2.4. Application of the model

The aim of the model is to predict how much water has to be added to a
supercritical dyeing process (mH2O) to obtain a desired textile humidity (x), at given
process conditions (p, T, vessel volume, mtext and textile type). The procedure is as
follows:
1. From temperature, ρ sat
H 2 O and xmax are calculated with Eq. (16) and (8)

respectively.
2. From p, T and vessel volume V follow: ρCO2 (IUPAC tables) and Γ
(Eq. (19)).
3. From ρCO2, mtext, ρtext and V, the carbon dioxide mass mCO2 is
calculated (Eq. (7)).
4. The model is now reduced to a system of 6 equations: Eq. (2) to Eq.
(6) and Eq. (9). From experience it has to be known at which value of
x the coloration of a certain type of textile is maximal. Substituting this
x in Eq. (3) and Eq. (4) leaves 6 unknown variables: θH2O, γH2O, y,

39
 m Htext2 O , mH2O and m CO 2
H 2 O . The system can be solved and the output

mH2O gives the amount of water that has to be added to a dyeing vessel
to obtain maximum textile coloration.

To illustrate the application of the model an example is given. 100 kg Mercerized

cotton is dyed in scCO2 (30 MPa, 373 K) in a 500-liter vessel. If it is known from
experience that the coloration has an optimum at a water content of 0.126 g water/g
dry textile, the procedure is:
1. For 373 K, Eq. (16) delivers ρ sat 3
H 2 O = 0.59 kg water/m . Eq. (8)

gives xmax = 0.15 kg water/kg dry textile.

2. From the IUPAC tables [7] it follows that for p = 30 MPa and T =
373 K the density of scCO2 is ρCO2 = 662 kg/m3. The enhancement
factor calculated with Eq. (19) is Γ = 10.4.
3. From Eq. (7) it follows that mCO2 = 288 kg.
4. The desired textile humidity is x = 0.126 kg/kg. Solving Eq. (2) to
Eq. (6) and Eq. (9) gives the six unknown variables:
θH2O = 0.84 γH2O = 1.1 y = 0.0084
m CO 2
H 2 O = 2.3 kg m Htext2 O = 13 kg m H 2 O = 15 kg
The model predicts that when mH2O = 15 kg of water are added to the dyeing
vessel, optimal coloration is obtained. It should be noted that the model assumes
that the textile does not contain any water when it is placed in the dyeing vessel.
This is only true when the textile is in equilibrium with dry air. When the textile is
stored in humid air prior to dyeing, a correction should be made. This is illustrated
with the example treated above: if the cotton is stored at 293 K and with a relative
air humidity of 65%, according to Bobeth et al. [13], the moisture content is 0.11
kg water / kg dry cotton. This means that for the 100 kg of cotton not 15 kg of
water, as was predicted by the model, but only 4 kg of water have to be added to
obtain the desired water content of 0.126 g water/g dry cotton during the dyeing
process. For other storage temperatures and air humidities, textile humidities are
readily available in the tables and graphs from Bobeth et al. [13] and Morton and
Hearle [6], for cotton, viscose, silk and wool.

40
3.3. Conclusions

A model was proposed to calculate the optimal amount of water in a supercritical

dyeing process from dyeing temperature, pressure, vessel volume and textile mass.
The model can be used for cotton, viscose, silk and wool in supercritical carbon
dioxide.

The solubility and the activity coefficient of water in textile were modeled. The
resulting equations were fitted between 347 and 377 K with experimental data for
air but are also valid under supercritical dyeing conditions.

The solubility of water in scCO2 as a function of density and temperature was

modeled with a solubility enhancement factor derived from statistical
thermodynamics. At typical textile dyeing conditions (373 K, 30 MPa) the
solubilities predicted by this factor are in agreement with experimental data from
literature.

Appendix A

From statistical thermodynamics (Hill [15]) it follows that the chemical

potential µ i of a system of n i particles i at temperature T can be calculated from
the canonical ensemble partition function Z:

1 ∂ ln Z
µi = − (a1)
β ∂n i

where β = 1/kT and k is the Boltzmann constant.

According to Vera and Prausnitz [16] and Reif [17] the partition function for non-
ideal fluids can be approximated by the generalized Van der Waals form:

3
n
1  2πm  2
Z =  2  [(V − Vex ) exp(−βU)] q nr , v
n
(a2)
n!  h β 

41
where m is particle mass, h is Planck’s constant, Vex is the volume excluded by
the particles and U is the average attractive potential energy between one particle
and all others in the fluid. The final term, q nr , v , accounts for rotational and
vibrational energies of the particles and according to Gasser [18] it is a function of
temperature only for small molecules like water or carbon dioxide. To allow the
derivation of an analytical solution like Eq. (18), a simplification is needed here: It
is assumed that the rotational and vibrational term q nr , v is negligible compared to
the other contributions in Eq.(a2). The validity of this assumption is checked in a
later stage by fitting the final result of the derivation Eq. (18) to experimental data
from literature.

The total potential U is the sum of ½ n pair potentials (u) where the factor ½ is
introduced because each potential is shared by 2 particles. For u an average value
can be estimated from the attractive term of the Lennard-Jones potential:

∞ 6 ∞ 4
1 1 n σ 6 n 1 8πσ 3 εn
2 V ∫σ V ∫σ  r 
U = nu = − 4 ε   4 πr 2
dr = −8 πσ ε   dr = − (a3)
2 r 3V

where ε and σ are the Lennard-Jones parameters and r is the distance between a
pair of particles.

Substituting the right-hand term of Eq. (a3) in Eq. (a2) and taking the natural
logarithm gives:
3  2 πm  8βπσ 3 εn 2
ln Z = −n ln n + n + n ln 2  + n ln(V − Vex ) + (a4)
2 h β 3V

because lnn! = nlnn-n for large values of n.

Combining Eq. (a4) with Eq. (a1) yields the pure-compound chemical potential:

42
 1 ∂  3  2 πm  8βπσ 3 εn 2 
µ=−  − n ln n + n + n ln 2  + n ln(V − Vex ) + 
β ∂n  h β 3V 
2  
1 3  2 πm  Vex 16βπσ 3 εn 
= −  − ln n + ln 2  + ln(V − Vex ) − +  (a5)
β 2 h β V − Vex 3V 

where the term Vex/(V-Vex) is introduced because of the dependence of Vex on n.

Eq. (a5) can be applied for gases, liquids and supercritical fluids. When Eq. (a5) is
substituted in the liquid (L) - gas (G) equilibrium condition µL =µG, the term
3/2ln(2πm/h2β) cancels because it is the same for a gas and a liquid. The
equilibrium condition becomes:

VexL 16βπσ 3 εn L
− ln n L + ln(V L − VexL ) − + =
V L − VexL 3V L
(a6)
VG 16βπσ 3 εn G
− ln n + ln(V − V ) − G ex G +
G G G

V − Vex
ex
3V G

or, taking one mole in each phase:

b 16βπσ 3 εN A
− ln N A + ln(VmL − b) − + =
VmL − b 3VmL
(a7)
b 16βπσ 3 εN A
− ln N A + ln(V − b) − G
G
+
Vm − b
m
3VmG

if NA is Avogadro’s number, Vm is the molar volume and b is the molar excluded

volume. The two-phase system described by Eq. (a7) is at the vapor pressure of the
pure compound. Here, water is regarded at temperatures up to 377 K, so that it can
be stated that the pressure of the system is so low that the term b (VmG − b) can be
neglected. The parameter σ is eliminated by setting the molar excluded volume
equal to the Van der Waals volume: b = 2/3 πNAσ3 :

43
 b 8βbε 8βbε
ln(VmL − b) − + = ln( V G
− b ) + (a8)
VmL − b VmL
m
VmG

VmL − b  1 1 b  8βbε b
= exp 8βbε ( G − L )( L )  = exp(− L + L ) (a9)
Vm − b
G
 Vm Vm Vm − b  Vm Vm − b

because the reciprocal of the molar gas volume is negligible compared to that of
the liquid. As discussed above, the pressure is low, so that VmG − b is
approximately VmG and Eq. (a9) becomes:

1 1  8βbε b 
= L exp − L + L  (a10)
Vm Vm − b
ig
 Vm Vm − b 

where ig denotes “ideal gas”. This equation is used in combination with Eq. (a18)
that will now be derived.

When a solution of water (component 1) in supercritical carbon dioxide (2) is

regarded, the total partition function is the product of the 2 individual partition
functions or Z = Z1Z2:
n
  σ + σ
3
  1
  
3
n  8βπ 1 2
 ε 12 n 2  
1  2πm1  2 1   8βπσ13 ε 11n 1  2   
Z=   (V − Vex ) exp + q1,r , v  .....
n 1!n 2 !  h 2β    3 V 3 V 
 
   
   
n
   σ1 + σ 2 
3
  2
3
n   8βπ  ε 21 n 1  
 2 πm 2  2 2   8βπσ 32 ε 22 n 2  2   
.... 2  (V − Vex ) exp + q 2 , r , v 
(a1)
β 
 h    3 V 3 V  
 
   

where ε12 = ε21 represents the binary interaction energy between a water and a
carbon dioxide molecule. The corresponding Lennard-Jones distance is [(σ1
+σ2)/2]3. Substitution of Z in Eq. (a1) and carrying out the differentiation to the

44
particles 1 gives the chemical potential of water µ1 dissolved in supercritical carbon
dioxide. In the differentiation , it is taken into account that the excluded volume is
occupied by both components: Vex = Vex,1 + Vex,2, or, if vi is the volume excluded
by a single particle of component i: Vex = n1v1 + n2v2.

 3  2 πm  nv 16βπσ13 ε 11 n 1 
 − ln n 1 + ln 2 1  + ln(V − Vex ) − 1 1 + +
 2  h β  V − Vex 3V 
1  (a12)
µ1 = − 
β 16βπ σ1 + σ 2  ε 12 n 2
3

 
  2  n v
− 2 1 + ln q 1,r , v 
 3V V − Vex 

Using the molar excluded volume b1 = 2/3 πNAσ13, introducing the combined
excluded volume b12 = 2/3 πNA [(σ1+σ2)/2]3 and setting µ 1L = µ 1G :

n 1L v1 8βb1ε 11 n 1L 8βb12 ε 12 n L2 n L2 v1
− ln n + ln(V − V ) − L
L L L
+ + − L =
V − VexL V − VexL
1 ex
VLNA VLNA
(a13)
n 1G v1 8βb1ε 11 n 1G 8βb12 ε 12 n G2 n G2 v1
− ln n + ln(V − V ) − G
G G G
+ + − G
V − VexG V − VexG
1 ex
VG NA VG NA

Or, taking one mole of water and neglecting the solubility of CO2 in the liquid
phase:
b1 8βb1ε 11 n 1L
− ln N A + ln(VmL − b1 ) − + =
VmL − b1 VLNA
(a14)
nGv 8βb1ε 11 n 1G 8βb12 ε 12 n G2 n G2 v1
− ln n + ln(V − V ) − G 1 1 G +
G G G
+ −
V − Vex V G − VexG
1 ex
VG NA VG NA

8βb1ε 11 n 1L 8βb1ε 11 n 1G n G2 v1 n 1G v1
Furthermore: >> and >> because the
VLNA VG NA V G − VexG V G − VexG
solubility of water in the gas phase is low. Neglecting both small terms and
substituting b1/NA for v1:

45
 b1 8βb1ε 11 n 1L
− ln N A + ln(VmL − b1 ) − + =
VmL − b1 VLNA
(a15)
8βb12 ε 12 n G2 n G2 v1
− ln n + ln(V − V ) +
G G G
−
V G − VexG
1 ex
VG NA
n iα
Introducing the molar concentration of component i in phase α as N iα =
Vα NA
gives:
b1 8βb1ε 11 n 1L
ln(VmL − b1 ) − + =
VmL − b1 VLNA
(a16)
1 VexG N A n G2 N A V G b1
ln( G − ⋅ G ) + 8βb12 ε 12 N G2 −
N1 n1 G
n2 NAV G
1 / N 2 − 1 / N 2,max
G

b1 8βb1ε 11 N A
ln(VmL − b1 ) − + =
V − b1
L
m VLNA
==> (a17)
 1 N G2  b NG N
ln G −  + 8βb12 ε 12 N G2 − 1 2 2,max
N G  N 2,max − N G2
 1 N 2,max N 1 

N 1G
=
1 − N G2 / N 2,max
==> (a18)
1  b1 8βb1ε 11 b1 ρ CO
max
2 ρ CO 2

exp  − + 8β b ε N G
− 
Vm − b1  M CO 2 ρ CO 2 − ρ CO 2 
 Vm − b1
L L L 12 12 2 max
Vm

Combining Eq. (a10) and Eq. (a18) :

N 1G 1  8b12 ε 12 N G2 b1 ρ CO
max
2 ρ CO 2

= exp
 −  (a19)
1 − N 2 / N 2,max Vm
G ig
 kT M CO 2 ρ CO 2 − ρ CO 2 
max

Substituting N1ig for 1/Vmig, 1000 ρCO2/MCO2 for N2G :

N 1G  N G2   8b ε 1000ρ CO 2 b1 ρ CO
max
2 ρ CO 2

= 1 −  exp 12 12 −  (a20)
N 1ig  N 2,max   kTM M ρ max
− ρ 
  CO 2 CO 2 CO 2 CO 2 

46
Setting N1G/N1ig = yρCO2/ρH2Osat = Γ and N2G/N2,max = ρCO2/ ρ CO
max
2 , where ρ CO 2
max

denotes the maximum density of carbon dioxide and introducing the constant κ:

y sat ρ CO 2  ρ CO 2   κρ CO 2 b1 ρ CO
max
2 ρ CO 2

= 
 1 − 
max 
exp
 −  (a21)
ρ H 2O
sat
 ρ CO 2   T M CO 2 ρ CO 2 − ρ CO 2 
max

References

1. E. Bach, E. Cleve and E. Schollmeyer, Past, Present and Future of

Supercritical Fluid Dyeing Technology - An Overview, Rev. Prog. Color,
32, 88 (2002).
2. W.J.T. Veugelers, H. Gooijer, J.W. Gerritsen and G.F. Woerlee, European
Patent EP 1 126 072 A2, 2001.
3. M. Van der Kraan, M.V. Fernandez Cid, G.F. Woerlee, W.J.T. Veugelers,
G.J. Witkamp, Dyeing Natural and Synthetic Textiles in Supercritical
Carbon Dioxide with Reactive Dyes, in: A. Bertucco (Ed.), Proceedings of
the Fourth International Symposium on High Pressure Technology and
Chemical Engineering, vol.1, 2002, p.199.
4. T.P. Nevell, Cellulose: structure, properties and behaviour in the dyeing
process, in: J. Shore (Ed.), Cellulosics Dyeing, Society of Dyers and
Colourists, Bradford, 1995, 1-80.
5. J.M. Smith and H.C. Van Ness, Introduction to Chemical Engineering
Thermodynamics, 4th ed., McGraw-Hill, New York, 1987, 346.
6. W.E. Morton and J.W.S. Hearle, Physical Properties of Textile Fibres, 3rd
ed., The Textile Institute, Manchester, 1993, 174.
7. S. Angus, B. Armstrong and K.M. de Reuck, International Thermodynamic
Tables of the Fluid State, vol.3, Pergamon Press, Oxford, 1976, table 3.
8. K.A. Evelein, R.G. Moore and R.A. Heidemann, Correlation of the Phase
Behavior in the Systems Hydrogen Sulfide–Water and Carbon Dioxide–
Water, Ind. Eng. Chem., Process Des. Dev. 15 (3), 423 (1976).
9. J.G. Wiegerink, The Moisture Relations of Textile Fibres at Elevated
Temperatures, Text. Res. 10, 357 (1940).

47
10. J.M. Prausnitz, R.N. Lichtenthaler and E. Gomes de Azevedo, Molecular
Thermodynamics of Fluid-Phase Equilibria, 3rd ed., Prentice Hall, New
Jersey, 1999, 262.
11. B.E. Poling, J.M. Prausnitz and J.P. O’Connel, The Properties of Gases
and Liquids, 5th ed., McGraw-Hill, New York, 2001, A-59.
12. R.H. Perry and D.W. Green, Perry’s Chemical Engineers’ Handbook, 7th
ed., McGraw-Hill, New York, 1997, 2-12.
13. D.R. Lide, Handbook of Chemistry and Physics, 75th ed., CRC Press, Boca
Raton, 1994, 6-48.
14. W. Bobeth, W. Berger, H. Faulstich, P. Fischer, A. Heger, H.-J. Jacobasch,
A Mally and I. Mikut, Textile Faserstoffe, Springer-Verlag, Berlin, 1993,
237.
15. T.L. Hill, An Introduction to Statistical Thermodynamics, Addison-Wesley
Publishing Company, Massachusetts, 1960, 19.
16. J.H. Vera and J.M. Prausnitz, Generalized van der Waals Theory for Dense
Fluids, Chem. Eng. J. 3, 1 (1972).
17. F. Reif, Fundamentals of Statistical and Thermal Physics, McGraw-Hill,
New York, 1965, 426.
18. R.P.H. Gasser and W.G. Richards, An Introduction to Statistical
Thermodynamics, World Scientific Publishing Co., Singapore, 1995, 40.

48
Chapter 4

Equilibrium Study on the Disperse Dyeing of

Polyester Textile in Supercritical Carbon Dioxide

Abstract

The dyeing of polyester textile in supercritical carbon dioxide (scCO2) was

investigated experimentally. The influence of temperature and density of the scCO2
on the process was studied in the ranges (85-125)ºC and (400-550) kg/m3. The dye
saturation concentration in the polyester increased and the distribution coefficient
decreased with temperature, the latter showing a logarithmic dependence on the
reciprocal of temperature. Increasing the fluid density gave an increasing saturation
concentration and a decreasing distribution coefficient. When the right temperature
and solvent density were chosen for the supercritical process, the same dye
concentration could be attained as in aqueous dyeing. It was found that the
adsorption of the dye on the polyester followed Nernst adsorption, as is the case in
aqueous dyeing. The experiments showed that the dyeing was exothermic, with a
negative change of entropy accompanying the transfer of dye from the dissolved to
the adsorbed state. The thermodynamic characteristics of supercritical and aqueous
dyeing were concluded to be roughly the same, with similar saturation
concentrations, thermodynamic affinities and heats and entropies of dyeing. The
observed shrinking of the polyester during the dyeing process was also similar to
what is normal in aqueous dyeing. The exposure to scCO2 had no effect on the
strength of the textile.

Contents submitted to the Textile Research Journal

49
4.1. Introduction

The large-scale water pollution in current industrial dyeing processes poses an

environmental problem and an economical burden for the textile industry. In the
last few decades, researchers have therefore investigated supercritical fluids
(SCF’s) as an alternative to water as a dyeing medium.

The best suited SCF for this application is carbon dioxide, because it is
inexpensive, non-toxic and non-flammable. Due to its green and safe character,
supercritical carbon dioxide (scCO2) is already in use on an industrial scale in the
extraction of caffeine from coffee and tea. Typical conditions for textile dyeing in
scCO2 are 30 MPa and 120ºC, resulting in a liquid-like CO2-density of 585 kg/m3.

The most important advantage of dyeing in scCO2 instead of water is the easier
separation of solvent and residual dye: depressurization after the dyeing leads to
precipitation of the dye and delivers clean, gaseous CO2, so that both compounds
can be recycled.

An additional advantage of dyeing in scCO2 is caused by the physical properties of

the supercritical state: The density is liquid-like, so that low-vapor pressure
compounds, such as dyes, can be dissolved. The viscosity is lower and the
diffusion coefficient in a SCF is higher than in a liquid, facilitating mass transport.
When dyeing non-polar textiles such as polyester, the scCO2 acts as a swelling
agent, plasticizing the polymer and increasing the rate of dye diffusion inside the
fiber [1]. Since CO2 is a non-polar solvent, disperse dyes can be dissolved without
the need for dispersing agents, i.e. simpler dye formulations can be used than in
aqueous dyeing.

The review by Bach et al. [2] mentions that natural and synthetic fibers have been
dyed successfully in scCO2. Schmidt et al. [3] have shown that scCO2-dyeing does
not damage the fibers, as long as the temperature is kept below 160ºC.

Most publications in the field of supercritical dyeing treat the measurement and
modeling of dye solubility in scCO2 [4-21]. The present chapter aims to contribute
to a less researched [22-26] subject in polyester dyeing: it describes a quantitative

50
equilibrium study of saturation colorations, adsorption isotherms and distribution
coefficients.

An equilibrium study or, in other words, an investigation of the dyeing

thermodynamics, is important because, as will be clear from the above
considerations, a supercritical solvent forms a quite different dye bath than a liquid
solvent. Furthermore, as is discussed by Kikic e.a. [27], scCO2 dissolves in
polyester, changing its properties and possibly also its thermodynamic behavior in
a dyeing process. Up to now, it has been reported, e.g. by Chang e.a. [22], that the
coloration of polyester in scCO2 increases with pressure and temperature, but a
more systematic study on this subject is desirable.

The aim of this work is to investigate the influence of the density and temperature
of the scCO2 on the polyester saturation coloration and the distribution coefficient.
Also the thermodynamic parameters governing the supercritical dyeing process are
determined for two dyes. The coloration and the thermodynamic parameters are
compared to those reported for aqueous dyeing. Finally, the adsorption isotherms
of two dyes on polyester are determined to assess if Nernst adsorption is the dyeing
mechanism, as is the case for dyeing in water.

4.2. Theory of Dyeing

4.2.1. Dye solubility in scCO2

To describe the thermodynamics of the supercritical dyeing system, it is necessary

to know how much dye is present in the dyeing vessel. The solubility of
compounds in scCO2 can be described by several semi- or purely empirical
expressions, fitted to experimental data. In the present work, the purely empirical
equation 1 is used, because Jouyban et a. [28] showed that this expression is
accurate for 106 low vapor pressure compounds, including dyes:

lny = M0 + M1p + M2p2 + M3 pT + M4T/p + M5 ln ρ (1)

51
where y is the solubility in mole fraction, ρ is the density of pure scCO2 (kg/m3) at
the dyeing pressure p (MPa) and dyeing temperature T (K) and M0 - M5 are fit
parameters. The value of CCO2 (g/g) is now:

M dye (2)
C CO 2 = y
M CO 2

where Mdye and MCO2 are the molar masses (g/mol) of dye and carbon dioxide.

4.2.2. Distribution coefficient

The thermodynamic affinity (–∆µ0, J/mole) of dye for textile is defined as [29]:

a PET (3)
− ∆µ 0 = RT ln
a CO 2

where R is the universal gas constant (J/mole/K), T is the dyeing temperature (K)
and aPET and aCO2 are the activities of the dye in the polyester and the scCO2 phase.
The activities are defined as the ratio’s of the dye fugacities in the corresponding
phase (fPET and fCO2) relative to the standard-state fugacity f0 , the standard state
being solid dye at the dyeing temperature and 1 bar:

f PET f CO 2 (4)
a PET = a PET =
f0 f0

The fugacities in the PET and the scCO2 follow from the definitions of the activity
coefficients γPET and γCO2:

f PET = C PET γ PET f f CO 2 = C CO 2 γ CO 2 f (5)

where CPET and CCO2 are the dye concentrations (g/g) in both phases and f is the
fugacity of pure solid dye at the dyeing temperature and at the dyeing pressure.
Eliminating fPET and fCO2:

52
 C PET γ PET f C CO 2 γ CO 2 f (6)
a PET = a CO 2 =
f0 f0

In aqueous dyeing it is common practice [30] to put f = f0, i.e. to assume that the
fugacity of the dye is independent of pressure. This is also assumed for the
supercritical process. The system regarded here is saturated with dye, so that γPET =
γCO2 = 1. The activities are now equal to the concentrations and equation 3 can be
simplified to:

C PET (7)
− ∆µ 0 = RT ln
C CO 2

The ratio of CPET (g/g) and CCO2 (g/g) is the distribution coefficient K:

C PET (8)
K ≡ CO 2
C

In this work, the values for K are determined at several temperatures and scCO2-
densities, using equation 8.

The thermodynamic affinity is equal to the standard free energy change of dyeing
∆G0 (J/mole), and therefore:

∆µ0 = ∆G0 = ∆H0 - T∆S0 (9)

where ∆H0 and ∆S0 are the changes in standard enthalpy (J/mole) and entropy
(J/mole/K) during the dyeing process. Combining equations 7, 8 and 9 gives:

 ∆H 0  ∆S 0 (10)
ln K = −  +
 RT  R

for a saturated dye bath. According to equation 10, constructing a Van ‘t Hoff plot,
i.e. plotting ln K as a function of 1/T leads to a straight line and gives the values of
the standard enthalpy and entropy of dyeing. Equation 10 is known to hold for

53
aqueous dyeing processes [29] and it is investigated here if it is also valid for
supercritical dyeing.

As discussed by Burkinshaw [30], disperse aqueous polyester dyeing is an

exothermic process (∆H0<0) and the accompanying change in entropy is negative
(∆S0<0). The enthalpy and entropy changes are those associated with the transfer
of dye from the dissolved to the adsorbed state. In the present work, this is
investigated for the case of scCO2, for two disperse dyes. Also the standard
affinities of the dyes are determined.

The distribution coefficients are determined here at maximum coloration, i.e. for
polyester saturated with dye. The values for CPET are measured. The values of CCO2
that are to be substituted in equation 8, are the dye solubilities in scCO2, from
equation 2.

4.2.3. Adsorption isotherm

In disperse aqueous dyeing, when the equilibrium concentration of dye in polyester

is plotted against the concentration in the dye bath, a straight line is obtained,
indicating that the dyeing obeys a Nernst or partition mechanism. The linearity is
observed below and at the saturation point, i.e. the distribution coefficient is
independent of concentration in the fiber or in the water. Up to now, only two
publications have appeared on dye adsorption isotherms in scCO2, one reporting a
linear adsorption isotherm [25] and the other a Langmuir isotherm [24]. In this
work the dye isotherms in scCO2 are investigated further, for two dyes.

4.2.4.Effect of supercritical dyeing on polyester textile

In the aqueous dyeing process, polyester is heated with steam of 150 to 200ºC,
prior to dyeing. This is done to relax the polymer molecules in the fiber, to remove
the intermolecular stresses that are formed in the fiber spinning process. Without
this so-called heat-setting, excessive shrinking of the cloth occurs whenever it is
exposed to high temperatures, e.g. in a dye bath. It is investigated in this work what
the shrinking behavior of heat-set and non-heat-set polyester is when it is dyed in
scCO2. Also the tensile strength before and after dyeing are determined, to check
for possible fiber damage.

54
4.3. Experimental

4.3.1. Materials

The polyester (poly(ethylene terephtalate)) was piqué-knitted, washed to remove

the spinning oils and heat-set at 195ºC by the supplier: Ames Europe, Enschede
(The Netherlands). The yarn was 167 dtex, the specific weight of the cloth was 120
g/m2. The CO2 was purchased from HoekLoos (The Netherlands) and had a purity
of 99.97 %. The acetone was technical grade from Chemproha (The Netherlands).
The N,N-dimethylformamide was 99.8% from Merck Chemicals (The
Netherlands).

The properties of the dyes used in this study are given in table 4.1. Their structures
are given in figure 4.1. All dyes were purchased from Sigma Aldrich (The
Netherlands) and used without further purification.

Table 4.1. Properties of the dyes used in this study, with molecular weight (M),
wavelength of maximum absorption (λmax) and the purity according to the supplier
Dye M (g/mole) λmax (nm) purity (%)
Disperse Orange 13 352 427 90
Disperse Orange 3 242 443 90
Disperse Red 1 314 502 95

CH2 CH3

O2N N N NH2 O 2N N N N
CH2 CH2 OH

Disperse Orange 3 Disperse Red 1

N N N N OH

Disperse Orange 13
Figure 4.1. Molecular structures of the dyes used in this study

55
4.3.2. Dyeing equipment and procedure

The experiments were conducted in a 40-liter stainless steel pressure vessel with a
perforated drum inside that contained 100 g textile cloth. The drum rotated at 60
rpm. The dye was put in a 50 ml porous stainless steel cylinder (pore size 10µm)
that was fixed inside the drum. A simplified scheme of the equipment is given in
figure 4.2. The system was pressurized in 10 minutes, with a air-driven
reciprocating pump (Resato International, The Netherlands). The dyeing vessel was
heated with a steam jacket. During the heating, the scCO2 reached the desired
pressure and temperature in 30 minutes. After that, pressure and temperature
remained constant. During the dyeing, the scCO2 was circulated through the vessel
by a centrifugal pump (Autoclave Engineers, USA) at a flow rate of 30 liter/min, to
intensify mixing inside the vessel.

steam
rotating drum, with
textile and dye

motor 40-liter
pressure vessel

CO2 supply condensate

pump

CO2 circulation pump

Figure 4.2. Experimental 40-liter dyeing machine

In the experiments in scCO2 where the influence of temperature and density on the
saturation coloration was investigated, the PET was dyed for 6 and 8 hours. No
difference in coloration was measured. Because of this observation and because,
after these experiments, dye powder was left in the porous steel cylinder, it was
concluded that the PET had reached saturation. The distribution coefficient was
calculated with equation 8, taking CPET from equation 12 (below) and CCO2 from
equation 2.

56
Also the experiments in scCO2 for the determination of the sorption isotherms were
done for 6 and 8 hours. Again, no color differences were observed but now all dye
had disappeared from the steel cylinder, leading to the conclusion that after these
dyeing times, the dye distribution between the scCO2 and the PET was at
equilibrium but below saturation. The amount of dye in the PET-cloth (CPET) was
taken from equation 12, the amount of dye remaining in the solvent (CCO2) was
calculated from the dye mass balance:

m TOTAL = m PET C PET + m CO 2 C CO 2 (11)

with mTOTAL denoting the mass of pure dye (g) introduced into the vessel prior to
the experiment, mPET the mass of dyed polyester (100g) and mCO2 the mass (g) of
scCO2 in the dyeing vessel. The latter was calculated as the product of density at
120ºC and 300 bar (585 kg/m3) and vessel volume (40 liter): mCO2 = 23000 g.

To check the shrinkage of the polyester during the dyeing, the weight per unit area
cloth was measured before and after the dyeing. This was done for both heat-set
and non-heat-set polyester.

4.3.3. Analyses

After each dyeing experiment, the dyed PET-cloth was washed with acetone at
room temperature to remove any unfixed dye. The PET was then subjected to a
Soxhlet-extraction with boiling N,N-dimethylformamide (DMF) in an oil bath of
180ºC. After all dye had been removed, the dye concentration in the DMF was
determined with a UV/VIS spectrophotometer, at the wavelength of maximum
absorption given in table 1. The dye concentration in the PET was then calculated
with:

C DMF m DMF (12)

C PET
=
m PET

where CPET is the concentration (g/g) of dye in the PET, CDMF is the dye
concentration (g/g) in the DMF, mDMF is the mass of DMF after extraction (g) and
mPET is the mass of PET (g).

57
4.4. Results and Discussion

4.4.1. Dye solubility in scCO2

The solubility fits are constructed for three dyes: Disperse Orange 3, Disperse
Orange 13 and Disperse Red 1. The fit parameters of equation 1 are given in table
4.2. The fits are presented graphically in figures 4.3, 4.4 and 4.5. It can be seen that
the solubilities of Disperse Orange 3 and Disperse Red 1 are described well, that of
Disperse Orange 13 less accurately, but acceptable for our purpose. For Disperse
Orange 3 and Disperse Red1, the fits cover almost the entire dyeing temperature
range, for Disperse Orange 13, the fit range does not cover the higher dyeing
temperatures.

Table 4.2. Fit parameters for equation 1, describing the solubility of Disperse
Orange 3, Disperse Orange 13 and Disperse Red 1 in supercritical carbon dioxide
Dye Disperse Orange 3 Disperse Orange 13 Disperse Red 1
Reference [6] [8] [14]
M0 -0.006745 -48.69 0.07694
M1 -0.2452 0.7175 -0.6607
M2 -0.003408 -0.01720 -5.833·10-4
M3 1.102·10-3 9.821·10-4 1.776·10-3
M4 -0.2199 0.3863 -0.2913
M5 0.6311 4.747 1.448

58
 20

y/10 mol/mol
15

10

-6
5

0
300 400 500 600 700
density / (kg/m3)

Figure 4.3. Solubility (y) of Disperse Orange 3 as a function of carbon dioxide

density, with drawn lines representing the fitted curves of equation 1 and points
representing literature data [6]:
353 K,  373 K,  393 K

30
25
y /10 mol/mol

20
15
-6

10
5
0
300 400 500 600 700
density / (kg/m3)
Figure 4.4. Solubility (y) of Disperse Orange 13 as a function of carbon dioxide
density, with drawn lines representing the fitted curves of equation 1 and points
representing literature data [8]:
330 K,  370 K

59
 350
300

y / 10 mol/mol
250
200

-6
150
100
50
0
100 300 500 700 900
3
density / (kg/m )

Figure 4.5. Solubility (y) of Disperse Red 1, with drawn lines as the fitted curves
and points representing literature data [14]:
323 K,  353 K,  383 K

4.4.2. Distribution coefficient

For two dyes, Disperse Orange 3 and 13, the saturation coloration and the
distribution coefficient was investigated. The measured values of the saturation
coloration CPET are shown for Disperse Orange 3 in table 4.3, for two scCO2-

Table 4.3. Saturation concentrations of Disperse Orange 3 in polyester (CPET) and

in carbon dioxide (CCO2) and the distribution coefficients K, as a function of
temperature T, for two CO2-densities
density T Pressure CPET CCO2 K
3 -2 -6
kg/m (K) MPa / 10 g/g / 10 g/g
400 368 165 1.61 9.88 1630
378 178 1.79 16.6 1080
388 192 1.93 27.3 707
393 199 2.04 34.8 586
550 358 195 1.50 26.0 577
368 218 1.95 46.6 418
378 241 2.19 79.8 274
388 264 2.77 134 207
393 276 2.94 170 173
398 287 3.18 217 147

60
Table 4.4. Saturation concentrations of Disperse Orange 13 in polyester (CPET) and
in carbon dioxide (CCO2) and the distribution coefficients K
density T pressure CPET CCO2 K
kg/m3 (K) MPa / 10-2 g/g / 10-6 g/g
400 368 165 1.50 9.88 4230
378 178 1.69 16.6 3190
388 192 1.85 27.3 2280
393 199 1.90 34.8 1880
550 368 218 1.83 46.6 998
378 241 2.08 79.8 730
388 264 2.22 134 532

densities, together with the calculated values for CCO2 and the resulting distribution
coefficients K. The corresponding values for Disperse Orange 13 are given in table
4.4. To get a clear impression for the dyeing conditions, also the dyeing pressures
are given in the tables. The saturation coloration of Disperse Orange 13 in scCO2 at
378 K (105ºC) and 550 kg/m3 is 2.1 mass percent, close to the value reported in
literature for aqueous dyeing at 100ºC (2.0 mass percent) [31].

It can be seen from the tables that coloration increases with the temperature and the
density (or pressure) of the scCO2. This was reported earlier for other dyes in
scCO2, e.g. by Chang e.a. [22].

The Van ‘t Hoff plots of the distribution coefficient are given in figure 4.6 for
Disperse Orange 3. The R-squared values for 400 and 550 kg/m3 are 0.976 and
0.998, indicating good agreement of the experimental results with equation 10.
Also figure 4.7, showing the plots for Disperse Orange 13, shows agreement with
equation 10, the R-squared values for 400 and 550 kg/m3 being 0.994 and 0.999
respectively.

61
 8.5

7.5

ln K
6.5

5.5

4.5
0.0025 0.0026 0.0027 0.0028
1/T(K)
Figure 4.6. Distribution coefficient K of Disperse Orange 3 as a function of dyeing
temperature T, for different scCO2-densities:  400 kg/m3,
550 kg/m3

8.5

7.5
ln K

6.5

5.5

4.5
0.0025 0.0026 0.0027 0.0028
1/T(K)
Figure 4.7. Distribution coefficient K of Disperse Orange 13 as a function of
dyeing temperature T, for different scCO2-densities:  400 kg/m3,
550 kg/m3.

The fact that equation 10 is valid here, confirms the assumption made between
equations 6 and 7, about the negligible influence of pressure on the fugacity f of the
pure dye was allowed, even though the pressures in scCO2 are far higher than in
aqueous dyeing.

It can be seen in figures 4.6 and 4.7, that the distribution coefficient decreases with
temperature. Regarding equation 10 it can then be stated that the enthalpies of

62
dyeing are negative, i.e. the process is exothermic. This is an apparent
contradiction with the observation made earlier in tables 3 and 4, that the coloration
CPET increases with temperature. The explanation is that also CCO2 increases with
temperature and relatively more than CPET, so that the ratio of CPET and CCO2, the
distribution coefficient, decreases. In other words, the transfer processes of dye
from the standard (solid) state to either the adsorbed state in the PET or to the
dissolved state in the scCO2 are both endothermic processes. The transfer of dye
from the dissolved to the adsorbed state, the actual dyeing step, is exothermic.

There is also a negative entropy of dyeing, caused by the dye molecules taking on a
state of higher order when going from the supercritical to the adsorbed state on the
polyester molecules. As was discussed above, negative enthalpy and entropy
changes are also reported for aqueous dyeing.

The thermodynamic affinities of dyeing are calculated with equation 9, taking a

temperature of 100ºC as example. The affinity becomes less negative as density
increases, reflecting that, going from 400 to 550 kg/m3, the distribution coefficient
decreases. This shows that the relative increase of CPET with density is less than the
relative increase in CCO2.

Table 4.5 shows values of ∆µ0 that are in or near the range that is characteristic of
physisorption, which is -20 to 0 kJ/mol [32]. The affinity of Disperse Orange 13 for
polyester in water at 100ºC is ∆µ0 = -13 kJ/mol.[31], not equal to but in the same
order as is calculated here for scCO2. This value was determined for polyester that
was heat-set at 195ºC, just as the polyester used in this work. The enthalpies,
entropies and affinities mentioned in the table are all similar to those measured in
water for other disperse dyes [33].

Table 4.5. Standard enthalpy (∆H0), entropy (∆S0) and affinity (∆µ0) of dyeing in
scCO2 at 400 and 550 kg/m3, for two disperse dyes
Dye Density ∆H0 ∆S0 ∆µ0 (100ºC)
kg/m kJ/mole Jmole/K kJ/mol
Disperse 400 -43 -57 -22
Orange 3 550 -41 -61 -18
Disperse 400 -39 -36 -25
Orange 13 550 -37 -44 -21

63
4.4.3. Adsorption isotherm

In these experiments, Disperse Orange 3 and Disperse red 1 were used to dye
polyester. To investigate whether or not the adsorption follows a linear Nernst
isotherm, dyeing experiments were done with both the scCO2 and the polyester
below saturation. The results are given for two dyes in figure 4.8. The R-squared
coefficients for Disperse Orange 3 and Disperse red 1 are both 0.996, showing a
linear dependence of CPET on CCO2.

2.5
Disperse Orange 3 2.5
Disperse Red 1
2 2

CPET / 10 -2 g/g
CPET / 10-2 g/g

1.5 1.5
1 1
0.5 0.5
0 0
0 5 10 15 0 5 10 15 20
CCO2 / 10-5 g/g CCO2 / 10-05 g/g
Figure 4.8. Adsorption isotherms for two disperse dyes, in scCO2 at 550 kg/m3 and
115ºC

4.4.4.Effect of supercritical dyeing on polyester

To investigate the effect of the dyeing process on the polyester cloth, the shrinkage
and the tensile strength were determined, at the laboratory of Ames Europe. The
results are given in table 4.6. The increase of the mass per unit cloth area of the
non-heat-set polyester increases by 70% during supercritical dyeing. For the heat-
set polyester, this is only 20%, similar to the percentages normally observed in
aqueous dyeing [34]. This indicates that heat-set polyester has to be used for
dyeing in scCO2, to avoid excessive shrinkage. In table 4.6 it can be seen that the
tensile strength of the polyester cloth is not significantly affected by the exposure
to the supercritical dye bath. In the dyeing experiment on which table 4.6 is based,
Disperse Red 1 was used as dye, the temperature was 115ºC and the scCO2-density
was 550 kg/m3.

64
Table 4.6. Specific mass and tensile strength of heat-set and non-heat-set polyester,
before and after dyeing in scCO2 of 115ºC and 550 kg/m3
specific mass tensile strength
(g/m2) (N)
length width
heat-set before dyeing 123 483 457
after dyeing 147 494 476
non-heat-set before dyeing 123 335 479
after dyeing 206 634 529

4.5. Conclusions

When polyester is dyed with disperse dyes in supercritical carbon dioxide, the
saturation coloration increases with temperature and density of the scCO2. The
same coloration can be achieved in scCO2 as in water, as long as the density is
sufficiently high.

The distribution coefficient K decreases with density and temperature, showing a

logarithmic dependence on the reciprocal of temperature. This temperature-
dependence is similar to that observed in aqueous dyeing. The polyester swelling
by the carbon dioxide does not affect the temperature-dependence of the
distribution coefficient.

The transfer of dye from the dissolved state to the adsorbed state, i.e. the actual
dyeing step of the process, is an exothermic process, although the saturation dye
concentration in the polymer increases with temperature. This apparent
contradiction is caused by the increase of dye concentration in the CO2 with
temperature being relatively larger than the increase of the dye concentration in the
polyester. The dyeing step is accompanied by a negative entropy change. The
values of the enthalpies, entropies and thermodynamic affinities in supercritical
dyeing are in the same order of magnitude as in aqueous dyeing.

65
The linearity of the adsorption isotherms that is observed in water, is also found in
supercritical carbon dioxide, indicating a Nernst adsorption mechanism.

Dyeing polyester in scCO2 leads to similar shrinkage of the cloth as in aqueous

dyeing, provided that heat-set polyester is used. The textile fibers are not damaged
by the supercritical dyeing process.

It can be concluded that the disperse dyeing process of polyester behaves

thermodynamically much the same in supercritical carbon dioxide as it does in
water and, what is more important, the same color depths can be achieved in both
dyeing media.

References

1. S. Sicardi, L. Manna and M. Banchero, Comparison of Dye Diffusion in

Poly(ethylene terephtalate) Films in the Presence of a Supercritical or
Aqueous Solvent, Ind. Eng. Chem. Res. 39, 4707 (2000).
2. E. Bach, E. Cleve and E. Schollmeyer, Past, Present and Future of
Supercritical Fluid Dyeing Technology – an Overview, Rev. Prog. Color.
32, 88 (2002).
3. A. Schmidt, E. Bach and E. Schollmeyer, Damage to Natural and Synthetic
Fibers Treated in Supercritical Carbon Dioxide at 300 bar and
Temperatures up to 160 Degrees C, Text. Res. J. 72 (11), 1023 (2002).
4. S.L. Draper, G.A. Montero, B. Smith and K. Beck, Solubility
Relationships for Disperse Dyes in Supercritical Carbon Dioxide, Dyes and
Pigments 45, 177 (2000).
5. J. Fasihi, Y. Yamini, F. Nourmohammadian and N. Bahramifar,
Investigations on the Solubilities of Some Disperse Azo Dyes in
Supercritical Carbon Dioxide, Dyes and Pigments 63, 161 (2004).
6. A. Ferri, M. Banchero, L. Manna and S. Sicardi, An Experimental
Technique for Measuring High Solubilities of Dyes in Supercritical Carbon
Dioxide, J. Supercrit. Fluids 30 (1), 41 (2003).
7. B. Guzel and A. Akgerman, Solubility of Disperse and Mordant Dyes in
Supercritical CO2, J. Chem. Eng. Data 44, 83 (1999).

66
8. U. Haarhaus, P. Swidersky and G.M. Schneider, High-Pressure
Investigations of the Solubility of Dispersion Dyestuffs in Supercritical
Gases by VIS/NIR Spectroscopy. Part I – 1,4-Bis(octadecylamino)-9,10-
anthraquinone and Disperse Orange 13 in CO2 and N2O up to 180 MPa, J.
Supercrit. Fluids 8, 100 (1995).
9. S.N. Joung and K.-P. Yoo, Solubility of Disperse Anthraquinone and Azo
Dyes in Supercritical Carbon Dioxide at 313.15 to 393.15 K and from 10
to 25 MPa, J. Chem. Eng. Data 43, 9 (1998).
10. S.N. Joung, H.Y. Shin, Y.H. Park and K.-P. Yoo, Measurement and
Correlation of Solubility of Disperse Anthraquinone and Azo Dyes in
Supercritical Carbon Dioxide, Korean J. Chem. Eng. 15 (1), 78 (1998).
11. T. Kraska, K.O. Leonhard, D. Tuma and G.M. Schneider, Correlation of
the Solubility of Low-volatile Organic Compounds in Near- and
Supercritical Fluids Part II. Applications to Disperse Red 60 and Two
Disubstituted Anthraquinones, Fluid Phase Equilibria 194-197, 469
(2002).
12. J.W. Lee, J.M. Min and H.K. Bae, Solubility Measurement of Disperse
Dyes in Supercritical Carbon Dioxide, J. Chem. Eng. Data 44, 684 (1999).
13. H.-M. Lin, C.-Y. Liu, C.-H. Cheng, Y.-T. Chen and M.-J. Lee, Solubilities
of Disperse Dyes of Blue 79, Red 153 and Yellow 119 in Supercritical
Carbon Dioxide, J. Supercrit. Fluids 21 (1), 1 (2001).
14. T. Shinoda and K. Tamura, Solubilities of C.I. Disperse Red 1 and C.I.
Disperse Red 13 in Supercritical Carbon Dioxide, Fluid Phase Equilibria
213, 115 (2003).
15. B. Wagner, C.B. Kautz and G.M. Schneider, Investigations on the
Solubility of Anthraquinone Dyes in Supercritical Carbon Dioxide by a
Flow Method, Fluid Phase Equilibria 158-160, 707 (1999).
16. D. Tuma, B. Wagner and G.M. Schneider, Comparative Solubility
Investigation of Anthraquinone Disperse Dyes in Near- and Supercritical
Fluids, Fluid Phase Equilibria 182, 133 (2001).
17. T. Shinoda and K. Tamura, Solubilities of C.I. Disperse Orange 25 and C.I.
Disperse Blue 354 in Supercritical Carbon Dioxide, J. Chem. Eng. Data
48, 869 (2003).
18. K. Mishima, K. Matsuyama, H. Ishikawa, K.-I. Hayahi and S. Maeda,
Measurement and Correlation of Solubilities of Azo Dyes and

67
 Anthraquinone in Supercritical Carbon Dioxide, Fluid Phase Equilibria
194-197, 895 (2002).
19. A.S. Ozcan, A.A. Clifford and K.D. Bartle, Solubility of Disperse Dyes in
Supercritical Carbon Dioxide, J. Chem. Eng. Data 42, 590 (1997).
20. D. Tuma and G.M. Schneider, Determination of the Solubilities of
Dyestuffs in Near- and Supercritical Fluids by a Static Method up to 180
MPa, Fluid Phase Equilibria 158-160, 743 (1999).
21. H.-D. Sung and J.-J. Shim, Solubility of C.I. Disperse Red 60 and C.I.
Disperse Blue 60 in Supercritical Carbon Dioxide, J. Chem. Eng. Data 44,
985 (1999).
22. K.-H. Chang, H.-K. Bae and J.-J. Shim, Dyeing of PET Textile Fibers and
Films in Supercritical Fluids, Korean J. Chem. Eng. 13 (3), 310 (1996).
23. M.R. De Giorgi, E. Cadoni, D. Maricca and A. Piras, Dyeing Polyester
Fibres with Disperse Dyes in Supercritical CO2, Dyes and Pigments 45, 75
(2000).
24. A.S. Özcan and A. Özcan, Adsorption Behavior of a Disperse Dye on
Polyester in Supercritical Carbon Dioxide, J. Supercrit. Fluids, in press.
25. I. Tabata, J. Lyu, S. Cho, T. Tominaga and T. Hori, Relationship between
the Solubility of Disperse Dyes and the Equilibrium Dye Adsorption in
Supercritical Fluid Dyeing, Color. Technol. 117, 346 (2001).
26. M.-W. Park and H.-K. Bae, Dye Distribution in Supercritical Dyeing with
Carbon Dioxide, J. Supercrit. Fluids 22, 65 (2002).
27. I. Kikic, F. Vecchione, P. Alessii, A. Cortesi, F. Eva and N. Elvassore,
Polymer Plasticization Using Supercritical Carbon Dioxide: Experiment
and Modeling, Ind. Eng. Chem. Res. 42, 3033 (2003).
28. A. Jouyban, H.-K. Chan and N.R. Foster, Mathematical Representation of
Solute Solubility in Supercritical Carbon Dioxide Using Empirical
Expressions, J. Supercrit. Fluids 24, 19 (2002).
29. H.H. Sumner, in: A. Johnson (Ed.), The Theory of Coloration of Textiles,
2nd ed., Society of Dyers and Colourists, Bradford (U.K.), 255 (1989).
30. S.M. Burkinshaw, Chemical Principles of Synthetic Fiber Dyeing,
Chapman and Hall, London (1995).
31. E. Merian, J. Carbonell, U. Lerch and V. Sanahuja, The Saturation Values,
Rates of Dyeing, Rates of Diffusion and Migration of Disperse Dyes on
Heat-Set Polyester Fibres, Am. Dyestuff Rep. 53(26), 11 (1964).

68
32. Y. Yu, Y.-Y. Zhuang and Z.-H. Wang, Adsorption of Water-soluble Dye
onto Activated Functionalized Resin, J. Colloid Interf. Sci. 242, 288
(2001).
33. R.H. Peters, Textile Chemistry, Volume III, Elsevier, Amsterdam, 105
(1975).
34. S. Fransman, Ames Europe B.V., Personal communication.

69
70
Chapter 5

The Influence of Variable Physical Properties and

Buoyancy on Heat Exchanger Design for Near- and
Supercritical Conditions

Abstract

Computational fluid dynamics simulations were done on heated supercritical

carbon dioxide flowing up or down in a vertical pipe. The impairment or
enhancement of heat transfer caused by the temperature-induced variation of
physical properties was investigated, as well as the effect of buoyancy. The
simulations show, for non-buoyant flow, that for pressures above 120 bar, the
effect of variation in physical properties is small and a constant-property Nusselt
relation can be used for a heat exchanger design. For pressures below 120 bar, the
variation in physical properties has to be taken into account for a correct heat
exchanger design. For non-buoyancy conditions the Krasnoshchekov - Protopopov
equation can be used to calculate heat transfer coefficients. It was observed that
buoyancy can enhance heat transfer coefficients up to a factor 3. When buoyancy is
active, the highest heat transfer coefficients are realized when the fluid flows
downward. The Jackson and Hall correction factor for the calculation of heat
transfer coefficients under buoyancy was confirmed by the simulations.

Contents published in J. Supercrit. Fluids 34, 99 (2005)

71
Symbols

a constant -
b constant -
c constant -
cp specific heat at constant pressure Jkg-1K-1
d pipe diameter m
e constant -
g gravitational acceleration ms-2
Gr Grashof number, Gr = gd3ρ(ρb-ρw) /µ2 -
H enthalpy Jkg-1
h heat transfer coefficient Wm-2K-1
k thermal conductivity Wm-1K-1
Nu Nusselt number, Nu = hd/k -
p pressure bar
Pr Prandtl number, Pr = µcp/k -
q heat flow Js-1
Re Reynolds number, Re = ωd/µ -
T temperature K and °C
Uo overall heat transfer coefficient Wm-2K-1

Greek
ρ density kgm-3
µ dynamic viscosity Pa⋅s
ω mass flux kgm-2s-1
Φm mass flow kgs-1

subscripts
b at bulk conditions
cp at constant properties
i inside of pipe
o outside of pipe
pc at pseudo-critical temperature
vp at variable properties
w at wall conditions
c critical

72
5. 1. Introduction

In designing heat exchangers for supercritical fluids a complication arises that is

not encountered when dealing with liquids or gases. The strong variation of the
fluid physical properties with temperature around the critical point influences heat
transfer strongly.

This problem has led to a great deal of research and many publications are to be
found in which Nusselt relations for the calculation of heat transfer coefficients
have been constructed and fitted to experimental work, as is reviewed by Pitla et al.
[1]. The present work is an investigation into the phenomena around the critical
point and into the buoyancy phenomenon, i.e. free convection caused by
temperature-induced density differences. Computational fluid dynamics
simulations of heated supercritical carbon dioxide flows through vertical pipes are
carried out.

The objective is to study when and how the variable properties and the buoyancy
must be taken into account in the Nusselt-relation for heating. For buoyancy
conditions in a heated flow, it is investigated what the position of a pipe and the
direction of the flow should be.

The importance of having information on tube-side heat transfer coefficients in

supercritical fluid heat transfer can be illustrated by regarding the individual heat
transfer resistances of shell-side, wall and tube-side. The overall heat transfer
coefficient for a steel pipe is calculated with:

1 1 d o ln(d o d i ) d o
= + + (1)
Uo ho 2k steel dihi

where Uo is the overall heat transfer coefficient corresponding with the outside area
of the tube, ho and hi are the outside and inside film coefficients, do and di are the
outside and inside tube diameters and ksteel is the thermal conductivity of the steel
tube wall. As an example is taken here supercritical carbon dioxide (scCO2) being
heated with steam in a pipe with do=16 mm, di=13 mm and ksteel=16 Wm-1K-1 so
that 2ksteel/(doln(do/di))=10000 Wm-2K-1. The heat transfer coefficient for

73
condensing steam is ho=8000 Wm-2K-1. It can be seen in several publications [1-7]
that, without the influence of variable properties or buoyancy, the tube-side heat
transfer coefficient for scCO2 is in the order of hi=6000 Wm-2K-1. All three
contributions of Eq. (1) have the same order of magnitude, so it is important to take
into account enhancement or impairment of tube-side heat transfer caused by
variations in physical properties or buoyancy.

5.2. Nusselt-relations

For fluids with physical properties that can be regarded as temperature-

independent, the generally applied Nusselt-relation for the calculation of the tube-
side heat transfer coefficient is, according to Jackson and Hall [2,3]:

h cp d i
Nu cp = = 0.0183 Re 0b.82 Prb0.5 (2)
kb
or:
h cp = 0.0183 ⋅ d i−0.17 ω 0.82 k 0b.5 µ −b0.33 c 0p.,5b (3)

where Nucp is the Nusselt number in case of constant properties, evaluated at bulk
conditions, hcp is the corresponding tube-side heat transfer coefficient, kb is the
thermal conductivity of the fluid at bulk conditions, Reb and Prb are the Reynolds
and Prandtl numbers at bulk conditions, ω is the mass flux and µb and cp,b are the
dynamic viscosity and the specific heat at constant pressure at bulk conditions. In
Eq. (2), the physical properties are assumed constant in radial direction. In the
longitudinal direction of a pipe, the bulk temperature and therefore the physical
properties do change. Eq. (2) delivers the local heat transfer coefficient.

Around the pseudo-critical point, thermal conductivity, viscosity and specific heat
vary strongly. It should be noted here that this temperature is higher than the
critical temperature (31°C) when the pressure is above its critical value (74 bar). As
is shown in Fig. 5.1 for 80 bar (pseudo-critical temperature Tpc = 34°C) , the
variation in the factor c 0p.,5b is much larger than the change in the factor k 0b.5 µ −b0.33 ,
so that it can be expected that the heat transfer coefficient from Eq. (3) will follow

74
the trend of the specific heat, i.e. showing peaks at the pseudo-critical temperature.
Indeed, these peaks were observed in the experiments of Swenson et al. [4] for
water.

Eq. (2) can be used to predict the heat transfer coefficient at supercritical
conditions only at low temperature differences between wall and bulk because then
the physical properties can be regarded as constant in radial direction.

10

6 0 .5
k b Cpµb −0.33
[( W /( mK )) 0.5 (Pa.s) −0.33 )]
wortel

4 0.5
c p,b [(kJ /(kgK))
product
0.5
]

0
25 30 35 40 45
fluid temperature (°C)

Fig. 5.1. Influence of the variation of carbon dioxide physical properties around
the pseudo-critical temperature at 80 bar on the constant-property heat transfer
coefficient of Eq. (3)

At higher wall-to-bulk temperature differences, specific heat, viscosity, thermal

conductivity and density will show radial gradients and Eq. (2) is no longer valid.
Instead, the variable properties have to be taken into account. This is mostly done
by introducing correction factors in the form of ratios of wall to bulk properties
raised to an empirical power:

a
 cp   ρw 
b
 µw 
c
 kw 
e

Nu vp = Nu cp         (4)
 c p,b   ρb   µb   kb 
 

75
where Nuvp is the variable-property Nusselt number, a, b, c and e are constants, ρ is
the fluid density and c p is the average specific heat that can be calculated from the
specific enthalpy H and temperature T according to Pitla et al. [1]:

Hw − Hb
cp = (5)
Tw − Tb

In practice, not all corrections in Eq. (4) are necessary. Krasnoshchekov and
Protopopov [5] only corrected for specific heat and density and still obtained a
relation that could be fitted to experimental data for a heated carbon dioxide flow
in a vertical pipe:

a
h vp d i  cp   ρw 
b

Nu vp = = Nu cp     (6)
kb  c p ,b   ρb 
 
with a=0.4 at Tb<Tw<Tpc and 1.2Tpc<Tb<Tw,
a=0.4+0.2[(Tw/Tpc)-1] at Tb<Tpc<Tw,
a=0.4+0.2[(Tw/Tpc)-1][1-5{(Tb/Tpc)-1}] at Tpc<Tb<1.2Tpc and Tb<Tw,
and b=0.35-0.05p/pc.

where hvp is the variable-property tube-side heat transfer coefficient, Tb, Tw and Tpc
are temperatures at bulk, wall and pseudo-critical conditions and p and pc are
pressure and critical pressure respectively.

Other forced convection correlations for heating CO2 in a vertical pipe have been
proposed but according to Jackson and Hall [2], Eq. (6) is the most accurate and is
supported by the most experimental data. Although Krasnoshchekov and
Protopopov used a more complex constant-property Nusselt number, Jackson and
Hall [2] found that substituting Eq. (2) in Eq. (6) gave equally good results so this
will also be done in this work.

For a very high wall-to-bulk temperature difference an extra complication is

introduced. In this case, temperature-induced density variations cause free
convection or buoyancy that disturbs the flow profile and thereby influences the
heat transfer. When the fluid is heated while flowing down, buoyancy action leads

76
to a sharpened velocity profile relative to normal turbulent flow (Fig. 5.2a, b). This
leads to a steeper shear stress profile and therefore to more turbulence; the heat
transfer is enhanced.

shear stress profiles

velocity profiles

a) normal turbulent b) sharpened velocity

flow profiles profile and increased
shear stress

Fig. 5.2. Schematic drawing of buoyancy effect, in downward heated pipe flow,
leading to a sharpened velocity profile and a steeper shear stress profile, relative
to normal turbulent pipe flow

When the fluid is flowing upwards, buoyancy causes either enhancement or

impairment of heat transfer, depending on the magnitude of the buoyancy-effect:
Low buoyancy action (Fig. 5.3a) leads to a flattened velocity profile, less shear
stress and impairment of heat transfer whereas high buoyancy action (Fig. 5.3b)
gives an M-shaped velocity profile, increased shear stress (W-profile) and therefore
enhanced heat transfer.

77
 velocity profiles

shear stress profiles

a) flattened velocity b) M-shaped velocity profile

profile and less shear and increased shear stress
stress

Fig. 5.3. Schematic drawing of buoyancy effect, in upward pipe flow, leading to
either a flattened velocity profile and decreased shear stress or to an M-shaped
velocity profile and increased shear stress

It was observed by Walisch et al. [6] that heat transfer coefficients for heated
horizontal pipes are below those of vertical pipes with down flow. Since Walisch et
al. conducted their experiments both far away and close to the pseudo-critical
temperature and for different pressures, it is assumed here that the heat transfer
coefficients from the simulations for down flow will be larger than for horizontal
flow for the range of conditions treated here. Therefore, simulations for horizontal
flow are not carried out in this work.

Hall and Jackson [7] found that, in a vertical pipe with up- or down flow, buoyancy
has to be taken into consideration when the bulk Grashof and Reynolds numbers
are such that:

Grb Re −b2.7 > 5 ⋅ 10 −6 (7)

where Grb is the Grashof number evaluated at bulk conditions.

No correlation for buoyancy in upward flow exists but buoyancy in a downward

heated flow was described by Jackson and Hall [3] with a correction factor so that
the Nusselt number for mixed forced and free convection Numix becomes:

78
 a 0.91 1 / 3
 cp   ρw 
b
  Gr  
Nu mix = 0.0183 Re 0.82
Pr 
0.5    1 + 2750 −b2.7   (8)
b b
 c p,b   ρb    Re b  
 

The tube-side heat transfer coefficients predicted by Eq. (8) are used here to
compare with the results from computational fluid dynamics simulations.

5.3. Computational fluid dynamics simulations

Heated turbulent CO2-flows in a vertical pipe were simulated using the

computational fluid dynamics program FLUENT, version 5.0. Turbulence was
modeled with the k-ε model. The flow entering the pipe was hydrodynamically
fully developed. Variation of the physical properties of the CO2 (density, specific
heat, thermal conductivity and viscosity) with temperature was approximated by
piecewise linear functions (isobars). The deviations between the approximate linear
values and values from the NIST database [8] were less than 1% for all physical
properties.

In the simulations, the enthalpy Hin of the CO2 flowing into the simulated pipe and
the mass flow rate Φm were known. The program delivered the enthalpy Hout of the
CO2 at the end of the pipe. The heat q that was transported to the CO2 was
calculated with q = Φm(Hout-Hin). The tube-side heat transfer coefficient hi was then
calculated with q = hiA(Tw-Tb). The length of the simulated pipe was short so that
the calculated hi can be regarded as a local heat transfer coefficient. Only tube-side
coefficients were calculated.

5.4. Results and discussion

Simulations were carried out for constant bulk pressure and for constant bulk
density. In industrial practice these two cases are encountered in continuous and
batch processes respectively. For both cases, the wall temperature was set constant
in the simulations. In a continuous process this is realistic, in a batch heating
process it is not. In the latter case, therefore, the heat transfer coefficients resulting
from the simulations should be regarded as momentary values. It should be noted

79
here that the conditions “constant pressure” and “constant density” refer to the bulk
fluid. On a microscopic level, gradients in pressure and density do exist due to the
transport of thermal energy.

The constant bulk-pressure process was simulated for heated down flow in a pipe
of internal diameter 13 mm, mass flow 0.3 kg/s, bulk temperature Tb = 30 °C. The
tube-side heat transfer coefficients hi were calculated from the simulation results
and plotted as a function of wall temperature in Fig. 5.4. According to Eq. (7),
buoyancy was negligible so that only the variation of physical properties in radial
direction is expected to influence heat transfer.
heat transfer coefficient h i (kWm K )

16
-1
-2

14
12
10
8
6
P=80 bar
4
P=100 bar
2 P=120 bar
0
20 40 60 80 100 120
wall temperature Tw (°C)

Fig. 5.4. Heat transfer coefficient as a function of wall temperature and pressure,
for heated down flow at bulk temperature 30°C. Simulation results showing the
effect of the temperature-induced variation of specific heat

It can be seen in Fig. 5.4 that the heat transfer coefficient follows the same trend as
the specific heat does at the pseudo-critical temperature: showing peaks that
decrease in height and shift to higher temperatures when the pressure increases.
This can be understood by realizing that a peak in the specific heat in the thermal
boundary layer will give a peak in the heat transfer coefficient. However, this
would mean that, at a low wall-to-bulk temperature difference, the peak in transfer
coefficients should be observed when the wall has the pseudo-critical temperature.
From Liao and Zhao [9] it is known that at 80 and 100 bar, the pseudo-critical

80
temperatures are 35 and 45°C respectively. The peaks in Fig. 5.4 lie at higher
temperatures because even at low wall-to-bulk temperature difference, in reality
the wall will have to be somewhat above pseudo-critical to realize a pseudo-critical
temperature in the thermal boundary layer.

To illustrate the influence of the variable properties, the values of hi from Fig. 5.4
were compared to the constant-property heat transfer coefficient hcp calculated from
Eq. (2). In Fig. 5.5 the ratio hi/hcp is plotted as a function of wall temperature. The
horizontal dashed line at hi/hcp = 1 represents the constant-property situation. Going
from low to high wall temperature, it can be seen that first the simulations agree
with the constant-property relation Eq. (2) within 10%, i.e. the ratio hi/hcp is near
unity.

1.2

1.0

0.8
hi / hcp

0.6
P=80 bar
0.4
P=100 bar
0.2 P=120 bar

0.0
20 40 60 80 100 120
wall temperature Tw (°C)

Fig. 5.5. Normalized heat transfer coefficient as a function of wall temperature.

Simulation results showing the effect of specific heat and pressure, at bulk
temperature 30°C

This corresponds to a wall-to-bulk temperature difference so low, that no radial

gradient in specific heat exists, i.e. the condition for using Eq. (2) is met. Going
further up in wall temperature, the peaks of Fig. 5.5 indicate that Eq. (2)
underestimates the peaks in heat transfer for the case of 80 bar.

81
On the right-hand side of the peaks in Fig. 5.5, impairment of heat transfer occurs,
i.e. the ratio hi/hcp falls below unity, especially at low pressures. This cannot be
caused by buoyancy, according to Eq. (7). It is caused by the variation of density
and specific heat in radial direction.

density-effect
At high wall temperatures, the density at the wall is significantly lower than in the
bulk and a low-density, gas-like fluid transports heat less effectively. The density
variation in radial direction is less prominent at higher pressure and so is the
resulting impairment of heat transfer.

heat capacity effect

Due to the choice of bulk temperature (30°C), the specific heat in the bulk is higher
than at the wall, especially for high wall temperatures. This leads to an
overestimation of the heat transfer coefficient by Eq. (2). The specific heat effect is
less significant for higher pressures.

The impairment by density variation is a general phenomenon; whether or not the

effect of specific heat occurs depends on the choice of bulk temperature. From Fig.
5.5 it also follows that for 120 bar and higher, the constant-property relation Eq. (2)
can be used for design purposes for wall temperatures up to 120°C.

A comparison between heat transfer coefficients from the simulations and from the
variable-property relation Eq. (6) is made in Fig. 5.6 for the same conditions as in
the 80 bar case of Fig. 5.5. It can be seen that Eq. (6) has an accuracy of ± 30 %.

The above cases are examples of constant bulk-pressure heat transfer,

corresponding with industrial continuous processes. For heat transfer at constant
bulk-density, corresponding with industrial batch processes, the simulation results
are shown in Fig. 5.7. During the simulations, the fluid was heated from 80 bar and
33°C to 300 bar and 112 °C at a wall temperature of 120°C. Buoyancy-effects were
not important according to Eq. (7). It can be seen that the heat transfer coefficient
deviates significantly from the constant-property case at low pressure. This is the
same observation that was made earlier because the left-hand side of the heating
curve corresponds with a large temperature difference between wall and bulk, i.e.

82
an overestimation of heat transfer by Eq. (2) because of specific heat and density
variation in radial direction.

simulations
1.2
Eq. (6)
1.0

hi/hcp 0.8

0.6

0.4

0.2

0.0
20 40 60 80 100 120

wall temperature Tw (°C)

Fig. 5.6. Normalized heat transfer coefficient as a function of wall temperature, for
80 bar and a bulk temperature of 30°C. Comparison of simulation results with Eq.
(6)

1.0

0.8

0.6
hi/hcp

300 bar,
0.4 112°C
80 bar,
0.2 33°C

0.0
50 100 150 200 250 300
pressure (bar)

Fig. 5.7. Simulation results showing the change in normalized heat transfer
coefficient during isochoric heating of carbon dioxide of 615 kg/m3, flowing at 0.3
kg/s in a 13 mm pipe

83
It should be noted here that Eq. (2), containing the specific heat at constant
pressure cp, is used for a process occurring at constant bulk density, where the
heating leads to a pressure increase. This is allowed because Eq. (2) is not used to
calculate the whole heating process from 33 to 112 °C, in which case the specific
heat at constant volume would be expected, but only the momentary heat transfer
coefficient. For a short piece of pipe, such as is simulated here, the bulk-pressure
can be regarded as constant and therefore the use of cp and of Eq. (2) is permitted.

The importance of taking into account the variation of physical properties is

illustrated by Fig. 5.5 and Fig. 5.7. For conditions where hi/hcp = 0.5, for example,
Eq. (1) indicates that the overall heat transfer coefficient Uo is 30% lower than
predicted by Eq. (2).

Because in Fig. 5.4 to Fig. 5.7 no buoyancy is active, the influence of varying
physical properties could be investigated. Simulations were also carried out at 120
bar and large Grashof numbers so that not only forced but also free convection is
significant. It was observed earlier that the variation of physical properties has a
limited influence at 120 bar, so that in this case any significant deviation from the
constant-property situation will be caused by buoyancy. In the simulations wall and
bulk temperatures were 134 and 104°C respectively. Fig. 5.8 shows the resulting
heat transfer coefficients as a function of Reynolds number for a heated vertical
flow of CO2 in a 13 mm pipe, both for upward and downward flow.

It follows from Fig. 5.8 that, at large Reynolds number, there is no difference
between up- and down flow. At these conditions, buoyancy is negligible and the
constant-property equation is satisfactory for design purposes. At intermediate Re
values (from the dashed line down to Re = 40000), the heat transfer coefficient for
upward flow is lower than predicted by the constant-property relation Eq. (2) (hi/hcp
<1) while the down flow coefficient stays approximately constant. This is the effect
of the flattening of the velocity profile in the upward flow (Fig. 5.3a). Going from
Re = 40000 further down in Reynolds number, the heat transfer coefficients for
both up- and down flow become larger than predicted by Eq. (2), i.e. the ratio hi/hcp
is larger than unity. The enhancement is caused by the steeper shear stress profile
in the down flow (Fig. 5.2b) and by the W-shaped shear stress profile in the up
flow (Fig. 5.3b). In the buoyancy regime, to the left of the dashed line, the heat

84
transfer for the down flow is larger than for the up flow for the whole range of Re.
At very low flow rates (very low Re), as the regime of pure free convection is
approached, the difference between up- and down flow vanishes, which was also
reported by Jackson and Hall [3]. The limit for the onset of buoyancy from Eq. (7)
(GrRe-2.7 = 5⋅10-6) corresponds with the dashed line; the simulation results confirm
the validity of the buoyancy criterion.

4
heating, upward flow

3 heating, downward flow

hi / hcp

0
1E+03 1E+04 1E+05 1E+06 1E+07

Re

Fig. 5.8. Influence of buoyancy on the normalized heat transfer coefficient for
carbon dioxide flowing in a vertical pipe at 120 bar. Comparison of simulation
results for up- and downward flow. The dashed line indicates the onset of buoyancy
(Eq. (7)).

In Fig. 5.9, the above discussed simulation results for the heated down flow are
plotted together with the heat transfer coefficients that are predicted by Eq. (8). The
figure shows that the agreement is good, confirming the validity of Eq. (8).

The results from the mixed convection simulations (Figs. 5.8 and 5.9) show that
buoyancy can increase heat transfer significantly relative to the constant-property
situation. For example, if a value of hi/hcp = 2 can be realized, Eq. (1) then predicts
an increase in the overall heat transfer coefficient Uo of 25%.

All simulations and model calculations were carried out for CO2 in 13 mm pipes.
However, the results presented here are also applicable for other systems (other

85
fluids and pipe diameters), as long as the relevant dimensionless numbers (Re, Pr
and Gr) of a regarded system are the same as the ones in this work.

4
Eq. (8)
simulations
3
hi / hcp
2

0
1E+03 1E+04 1E+05 1E+06 1E+07
Re
Fig. 5.9. Influence of buoyancy on the normalized heat transfer coefficient for
carbon dioxide flowing down in a heated vertical pipe at 120 bar. Comparison of
simulations with Eq. (8)

5.5. Conclusions

For a heated vertical flow of CO2 without buoyancy, the constant-property Nusselt
relation is not suitable for pressures below 120 bar as it can introduce an
overestimation of the overall heat transfer coefficient by 30 %. For such conditions,
the variable-property relation of Krasnoshchekov and Protopopov is acceptable for
design purposes.

The criterion that Jackson and Hall developed, to determine when buoyancy is
important, was confirmed by the CFD-results.

In the mixed convection regime in downward heated flow, the CFD simulations
gave results that agreed with the Jackson and Hall buoyancy correction factor
combined with the Krasnoshchekov-Protopopov relation. The action of buoyancy
can lead to an increase in heat transfer coefficient as large as 25 %.

86
A heat exchanger for heating should be placed in a vertical position with the fluid
flowing downwards when buoyancy is active. A vertical up flow or a horizontal
flow will always result in a smaller heat transfer coefficient.

References

13. S.S. Pitla, D.M. Robinson, E.A. Groll, S. Ramadhyani, Heat Transfer from
Supercritical Carbon Dioxide in Tube Flow: A Critical Review, HVAC&R
Research 4 (3), 281 (1998).
14. J.D. Jackson, W.B. Hall, Forced Convection Heat Transfer, in: S. Kakaç,
D.B. Spalding (Eds.), Turbulent Forced Convection in Channels and
Bundles, vol.2, 1979, 563.
15. J.D. Jackson, W.B. Hall, Influences of Buoyancy on Heat Transfer to
Fluids Flowing in Vertical Tubes under Turbulent Conditions, in: S.
Kakaç, D.B. Spalding (Eds.), Turbulent Forced Convection in Channels
and Bundles, vol.2, 1979, 640.
16. H.S. Swenson, J.R. Carver, C.R. Kakarala, Heat Transfer to Supercritical
Water in Smooth-Bore Tubes, ASME Journal of Heat Transfer nov. 1965.
17. A. Krasnoshchekov, V.S. Protopopov, Heat Exchange in the Supercritical
Region During the Flow of Carbon Dioxide and Water, Teploenergetika 6
(12), 26130 (1972).
18. T. Walisch, W. Dörfler, Ch. Trepp, Heat Transfer to Supercritical Carbon
Dioxide in Tubes with Mixed Convection, HTD-Vol. 350, vol. 12, ASME
1997, National Heat Transfer Conference, 79.
19. W.B. Hall, J.D. Jackson, Laminarization of Turbulent Pipe Flow by
Buoyancy Forces, ASME Paper no. 69-HT-55 (1969).
20. National Institute for Standards and Technology (NIST) Chemistry
Webbook, http://webbook.nist.gov/chemistry.
21. S.M. Liao and T.S. Zhao, Measurements of Heat Transfer Coefficients
from Supercritical Carbon Dioxide Flowing in Horizontal Mini/Micro
Channels, J. Heat Transfer 124, 413(2002).

87
)

88
Chapter 6

Process and Equipment Design for a 100-Liter

Polyester Beam Dyeing Machine

Abstract

A technical-scale, 100-litre machine was designed and built to test polyester beam
dyeing in scCO2, with respect to the performance of the equipment and the
process.

The design aims at introducing 1 mass percent of dye into 13 kg polyester cloth in
a 2-hour process. Per batch, 95% of the CO2 is recovered and stored for reuse. The
dyeing conditions are 300 bar and 120ºC.

A new pressure vessel was designed, consisting of a steel liner with carbon fibers
wound around it to take up the radial forces. The axial pressure forces are taken up
by an external steel yoke keeping both lids of the vessel in place.

To circulate the CO2 with the dissolved dye through the textile roll, a low-pressure
centrifugal pump was placed inside the dyeing vessel.

Although the equipment designed in this chapter is smaller, the process is the same
as in the industrial-scale 1000-litre machine discussed in chapter 7. This gives the
opportunity to test and optimize the performance of the process. The two most
costly items of an industrial-scale machine, the dyeing vessel and the circulation
pump, were both especially designed for the technical-scale machine in a cheaper
form, thereby facilitating the scale-up of the supercritical dyeing process.

89
6.1. Introduction

In the previous chapters, different machine configurations were used in the

experiments. In the present chapter, a polyester beam-dyeing machine is discussed.
In such a machine, also used in current aqueous dyeing, the knitted or woven
polyester cloth is wound around a perforated pipe. The pipe is closed at one end
and the water or, in this case, the scCO2, is pumped into the other end of the pipe.
The flow is then forced through the PET-layers; the dye molecules simultaneously
diffuse into the fibers. Figure 6.1 gives a schematic impression of the dyeing
principle.
beam
(perforated pipe)
CO2 out
polyester
CO2 and dye in cloth

Figure 6.1. Schematic representation of a beam dyeing process

The dyeing machine developed in this chapter is large compared to the dyeing
machines described in literature [1] and in chapters 2 and 4 but is still an order of
magnitude smaller, in volume, than the industrial-size (1000-liter) machine
discussed in chapter 7. The aim of the currently discussed 100-liter dyeing machine
is testing the process with respect to:
1. Effect of heating and cooling rate. It is known from aqueous dyeing that
too fast a temperature rise or drop results in uneven dyeings of the PET. It
is to be tested if this phenomenon also puts a restriction on the heating and
cooling rate in scCO2.
2. Evenness of dyeing as a function of dyeing time, CO2-flow rate, beam
diameter and the diameter of the holes in the beam.
3. Dye dosing. An appropriate way of introducing and dissolving the dye
powder into the scCO2 is to be developed.
4. The leveling time. After all dye has diffused into the PET, the inside of the
textile roll is darker than the outside. When the CO2-flow is now
continued, the color will be equilibrated. It should be determined how long
this leveling takes.

90
 5. The required time and temperature of rinsing. After dyeing, the CO2
contains residual dye that is to be flushed out of the vessel to avoid
contamination of the machinery and the cloth. The optimal rinsing time
and temperature must be determined.
6. Problems with deposition of cyclic trimers. It has been observed by
Montero et al. [2], that PET-dyeing in scCO2 can lead to deposition of
ethylene terephtalate trimers inside the dyeing machinery. This problem is
to be investigated with the 100-liter machine.

It should be noted that the machinery was designed and built, but no experimental
results are available at this moment. Both the process and the equipment were
designed in close cooperation with FeyeCon D&I B.V. (The Netherlands).

6.2. Boundary conditions for the 100-litre beam-dyeing machine

The machine is designed for one set of boundary conditions (table 6.1). The aim of
the process is to introduce 1 mass% of dye into 13 kg of polyester. The textile, as
rolled up on the beam, has a density of 500 kg/m3. Since chapter 4 showed
sufficient coloration at 120ºC and 300 bar, these values are taken as the working
temperature and pressure in the design.

Table 6.1. Boundary conditions for the 100-liter beam-dyeing machine

Mass of PET roll (kg) 13
Length x diameter (m) 0.45 x 0.28
Beam diameter (m) 0.06
Dyeing temperature (ºC) 120
Design temperature (ºC) 150
Dyeing Pressure (bar) 300
Design pressure (bar) 350
Desired dye concentration (g/(gPET) 0.01

91
6.3. Process steps

The consecutive steps of the process are given in table 6.2, together with the times
for which the corresponding equipment items are designed. The aim is a total
process time of 2 hours. Figure 6.2 gives a simplified process flow diagram.

Table 6.2. Process steps and times

Process step Time (min)
Loading textile and dye 3
Pressurization 15
Dyeing at 120ºC and 300 bar 45
Leveling 15
Rinsing at 80ºC and 300 bar 20
Depressurization 20
Unloading textile 2
Total batch time 120

circulation reducing valve

pump separator

dye condenser
dyeing holder
vessel
DYEING SOLVENT
CYCLE RECYCLE CO2
storage
vessel

heater/cooler pressurisation chiller

pump

Figure 6.2. Simplified process flow diagram for the supercritical beam dyeing
process

92
After the textile and the dye powder are loaded, the dyeing vessel is pressurized
with CO2. In the last part of the pressurization step, the circulation pump and the
heat exchanger are started. The CO2 is heated at the same time as new CO2 is
pumped into the machine until the desired temperature (120ºC) and pressure (300
bar) are reached.

During the dyeing step, the CO2 is circulated from the dyeing vessel, over the dye
holder, the heat exchanger and back into the dyeing vessel. In this way, dye
dissolution and PET dyeing take place simultaneously.

After 45 minutes, the dye dosing stops. In the leveling time that follows, the flow
rate, temperature and pressure are kept constant. After this step, the dye is
distributed over the PET and the scCO2. Depressurization would lead to
precipitation of dye on both the textile and the inner parts of the dyeing machine
and therefore the vessel is rinsed with fresh CO2. During the rinsing, clean CO2
from the storage vessel is pumped into the machine while CO2 containing residual
dye is removed through the reducing valve. The circulation pump and the heater
are still running at this time.

The reducing valve keeps the pressure in the dyeing vessel at 300 bar during the
rinsing step. The CO2 exiting the valve enters the separator, where it is gasified to
precipitate the dye in order to obtain a flow of clean CO2 in the direction of the
condenser.

The gaseous CO2 is condensed and stored in liquid form. During the rinsing, the
pressure in the CO2-recycling part is kept constant by keeping the temperature in
the storage vessel at the saturation temperature at 60 bar (22ºC), using the coolant
flowing through the condenser.

The aim of the rinsing step is to lower the dye concentration in the vessel, so that
depressurization does not lead to precipitation of dye onto the cloth and the
machinery. In the rinsing step, dye will be extracted from the fibers. This negative
effect is minimized by lowering the temperature from 120 to 80ºC while the
pressure is kept constant by the pressurization pump. However, as is illustrated for
5 dyes in appendix 6.1, this leads to a decrease in dye solubility of maximally a
factor 2. When the coloration of the polyester is to be above one half of the

93
maximum coloration, the CO2 at the beginning of the rinsing step contains dye
concentrations above one half of the solubility, following from the linearity of the
absorption isotherms in chapter 4. Cooling then results in precipitation of dye in the
machine and on the cloth. To avoid this, the first part of the rinsing time can be
carried out at 120ºC and only then the temperature is lowered to 80ºC.

When the dyeing machine is regarded as ideally mixed, the time t (s) required to
decrease the concentration from c0 to c in the vessel volume V (0.10 m3) by rinsing
with volume flow rate ΦV (pump flow 1.2·10-4 m3/s) is given by integrating the
mass balance:
dc
V = −Φ V c (6.1)
dt
from t = 0; c = c0 to t = t; c = c, leading to:
 tΦ 
c = c 0 exp − V  (6.2)
 V 
From equation 6.2 it follows that the first 5 minutes of the rinsing have to be
carried out at 120ºC, the concentration is then c = 0.7 c0. The next 5 minutes,
rinsing and cooling occur simultaneously. In this way, the temperature of 80ºC is
reached at the moment that the dye concentration c = 1/2 c0, i.e. after 10 minutes of
rinsing. It is also calculated that a total of 20 minutes of rinsing leads to a decrease
in dye concentration of the scCO2 by 75%, i.e. c = 1/4 c0 . It is assumed that the
remaining 25% does not significantly affect the coloration of the next batch.

The process and equipment design allows changes in the times of heating, cooling,
dyeing, leveling and rinsing as well as changes in CO2 flow rate, beam dimensions
and in the dye dosing device, so that all six issues mentioned in section 6.1 can be
investigated.

6.4. Solvent cycle

The process is drawn in a temperature-entropy diagram in figure 6.3. Point A

represents the storage vessel conditions (saturated liquid at 22ºC, 60 bar). From
point A to B, the CO2 from the storage vessel is sub cooled isobaric to 20ºC, a
temperature with boiling pressure 57 bar, to prevent cavitation of the pressurization

94
pump. From B to C, the CO2 is pumped into the dyeing vessel. This is assumed to
be an isentropic process. Energy dissipation in the pump will bend the top of line
BC to the right. On the other hand, the CO2 in the dyeing vessel will lose part of its
heat to the steel and the polyester, bending the top of line BC to the left.

From point C on, the circulation and heating of the CO2 are started while the
pressurization pump still runs. The effect is a simultaneous pressure, temperature
and density increase, leading to point D that represents the dyeing conditions
(120ºC, 300 bar). From D to E, the CO2 is cooled isobaric from 120 to 80ºC; the
pressurization pump is used to simultaneously increase the density in the dyeing
vessel from 585 to 745 kg/m3.

During the rinsing period, the conditions inside the dyeing vessel remain those of
point E (80ºC, 300 bar). The CO2 that is used for the rinsing, follows the route
E
F
G
A
B
E (line BE is not drawn in the diagram). From E to F, the CO2

T (ºC)
D
supercritical
120
phase region
E
80
C critical point

31
liquid phase gas phase
22 G
region B A F region
gas-liquid two-phase region

saturated liquid line saturated gas line

entropy

Figure 6.3. Temperature-entropy diagram for CO2, showing the solvent cycle
during the supercritical dyeing process

passes through the regulating valve, undergoing an isenthalpic expansion into the
two-phase region, ending at the 60 bar isobar (line AG). The liquid mass fraction in
point F, being 0.4, represents the amount of CO2 that is to be evaporated in the
separator. During evaporation, the CO2 goes to the saturated gas line (point G). In

95
the condenser it crosses from G to A through the two-phase region. As liquid CO2
reenters the storage vessel, the cycle is closed.

After the rinsing, the pressure in the dyeing vessel is lowered through the
regulating valve until it is equal to the pressure in the storage vessel (60 bar). The
pressure in the dyeing vessel is then further reduced to 10 bar with the gas booster.
The final temperature in the dyeing vessel is estimated to be between –20 and
20ºC. This corresponds with a density of (21±2) kg/m3, giving a total amount of
95% of CO2 that is recycled per batch.

6.5. Equipment design

6.5.1. Dyeing vessel

The vessel was designed by Solico B.V. (The Netherlands). The internal length and
diameter are 1350 and 350 mm. The volume available in the vessel for the scCO2
is 100 liters. The working and design pressures are 300 and 350 bar, respectively,
the working and design temperatures are 120 and 150ºC.

Configuration of the vessel.

As discussed in chapter 7, a thin-walled steel liner with composite fibers wound
around it tangentially to take up the pressure forces in radial direction [3], is
economically favorable to use, instead of a completely steel dyeing vessel. Two
lids are used, not connected to the vessel, to distribute the axial forces over 6 steel
plates that are placed around the vessel as a yoke. The yoke is moved up and down
by two spindles, both driven by the same motor to ensure equal movement of both
sides of the yoke. During movement, the yoke is guided by 4 positioners, as is
shown in figure 6.4. The outer length, width and thickness of the plates are 2250,
800 and 30 mm.

96
 flanged CO2 exit connection

CO2 entrance connection

carbon fibers
lid

yoke plates

spindle

positioner

Figure 6.4. Photograph of the closed 100-litre dyeing vessel

To load and unload polyester rolls, the yoke is lowered so that one of the lids can
be moved away from the vessel by means of a pneumatic linear drive underneath
the vessel. Attached to the lid, a cart on wheels, supporting the beam with textile, is
then also pushed out. Inside the vessel, the cartwheels are guided by rails, welded
onto a stainless steel shell that covers the lower 115º of the liner. Figures 6.5 A, B
give a schematic impression of the opened vessel and a cross-section of the cart
loaded with textile, respectively.

Because of economical reasons, as discussed in chapter 7, the circulation pump is

placed inside the dyeing vessel. The lid on the pump-side of the vessel can be
pulled out manually, to perform maintenance on the pump. The CO2 entrance and
exit are on top of the lid, as is shown in figure 6.6. Figure 6.7 gives a complete
impression of the vessel and its contents, together with the flow path of the CO2.

97
 1 2 3 4 5 6 7
4
3

8
2

A B

1. vessel wall 5. yoke plates

2. cart bottom 6. lid
3. beam 7. support table
4. polyester roll 8. cart wheel

Figure 6.5. Schematic representation of the dyeing vessel in open position, with a
longitudinal section (A) and a cross-section of the loaded cart (B)

exit, to dye holder and heat

exchanger
entrance, into pump

Figure 6.6. Schematic presentation of the pump-side lid,

showing the CO2 entrance and exit.

When the yoke is lowered or raised, contact with the lids is to be avoided. A
distance of 3 mm is realized by pressing the lids against the vessel with:
• the pneumatic linear drive, on the beam-side and
• springs attached to the lid on the pump-side.
When the vessel is pressurized, the lids move 3 mm outwards until there is contact
with the yoke. On the lids, lip seals are placed, designed to slide along the internal
surface of the liner. Regarding the internals of the dyeing vessel, the movement of

98
 connection for
electrical wiring
of pump
pump casing beam carbon fibre layer
motor casing
seal bearings polyester liner

Figure 6.7. Detailed drawing of the 100-litre dyeing vessel

O-rings impeller rotor O-rings Teflon layer shell plate cart

magnetic spools

99
the lids is compensated by allowing the O-rings in figure 6.7 to slide 6 mm.
Because of the 3 mm movement of the pump-side lid, the CO2 entrance and exit of
the vessel are connected to the rest of the machine by flexible stainless steel hoses.

Materials of construction.
All parts of the vessel and the internals are made of stainless steel (AISI 316), with
the exception of:
• the lip seals, made from Rulon®, with PEEK back-up rings,
• the 4 O-rings in figure 6.7, made of a Viton® core with Teflon® jacket and
• the axes of the cartwheels are made of brass because this material is dry-
running. The use of steel is avoided here because the required grease on the
bearings would dissolve into the scCO2.

In the choice of the material for the liner and the composite, care is taken that the
strain of the liner is higher so that the pressure-induced deformation of the liner
will follow that of the composite. For the liner, duplex (stainless steel 1.4462) is
chosen since it has a large tensile strength, relative to other stainless steel types.
For the composite material, the standard carbon fiber type T 700 is taken because it
is cheaper and has a higher stiffness and tensile strength than other composite
materials, like aramid or glass fiber [4]. The matrix to keep the carbon fibers
together is a high-temperature epoxy resin, because it binds well with carbon fibers
and it can be applied at the design temperature of the dyeing vessel (150ºC).

The physical properties of the liner and the fiber material are given in table 6.3.
Calculations [4] show that at the working pressure of 300 bar, the stress in the liner
in tangential direction is 430 MPa, below the creep limit. At this pressure, the liner
is pressed outwards against the composite layer. At a pressure of 2070 bar inside
the vessel, the composite layer is at its tensile strength of 2450 MPa.

Cyclic fatigue tests on composite samples at 150ºC have shown a 50 % failure

probability for 2·106 cycles at a stress of 510 MPa, well above the stress of 430
MPa at the working pressure. For the liner this fatigue limit was calculated to be
565 MPa. This suffices since the vessel is designed for only 25000 dyeing batches.
The diameter increase at the dyeing pressure of 300 bar is 1.5 mm. The seals are
designed to compensate for this enlargement.

100
Table 6.3. Mechanical properties of the liner and the composite layer for the 100-
litre dyeing vessel [4]
Liner Material Duplex stainless steel
Thickness 10 mm
Creep limit 460 MPa
Fatigue limit (2·106 cycles) 565 MPa
Composite layer Material Carbon fiber in epoxy resin
Thickness 40 mm
Tensile strength 2450 MPa
Fatigue limit (2·106 cycles) 510 MPa

6.5.2. Circulation pump

Pump capacity
At 120ºC and 300 bar, dye solubilities lie between 10-4 and 10-5 g dye / g CO2 [5-
8]. It is assumed here that:
• the CO2 is saturated, containing 10-5 g dye / g CO2, when leaving the dye
holder and
• the CO2 contains a negligible amount of dye when exiting the textile roll.
To introduce 1 mass% of dye into 13 kg PET, a total amount of 13000 kg CO2
needs to be circulated in the 45 minutes mentioned in table 6.2. Since the density of
scCO2 at 120ºC and 300 bar is 585 kg/m3, this means that a pump capacity of 30
m3/hour is needed. The resulting CO2 velocity in the 2” SCH 160 pipes is 5.7 m/s.

Pressure drop
The required head of the circulation pump is determined by several pressure drop
contributions. The calculations are given in appendix 6.2. It is concluded that the
total pressure drop over the machine, to be delivered by the circulation pump, is 2
bar.
The combination of capacity and head have led to the choice of a 3000 rpm
centrifugal pump with a maximum capacity of 30 m3/hour at 3 bar head, designed
in cooperation with and supplied by Packo Inox N.V. (The Netherlands). The
electric motor for the pump has a capacity of 7.5 kW, is controlled with a
frequency drive and was delivered by Electromotorenfabriek Nijmegen (The
Netherlands).

101
Materials of construction
All parts of the motor and pump are made of stainless steel (AISI 316 and 304),
with the exception of:
• The electrical copper spools of the motor.
• The resin in which the spools are fixed is a product from
Electromotorenfabriek Nijmegen (The Netherlands). This product was
tested at the Delft University of Technology and found to be not soluble in
scCO2 and not damaged by repeated depressurizations.
• The bearings. Since grease dissolves in scCO2, dry-running bearing are
used. The balls are made of silicium nitride and run in steel cages with a
dry-lubrication coating. The bearings are a product from SKF Nederland.
• The electrical wiring to the motor is copper, coated with Teflon®.
• The rotor is made from soft iron.

6.5.3. Pressurization pump

Because an air-driven piston-cylinder pump was found to be suitable for liquid CO2
in the 40-litre machine discussed in chapter 4, such a device was also chosen here.
The pump has to fill the vessel in 15 minutes (table 6.2) and rinse the vessel
sufficiently in 20 minutes. The latter duty determines the required flow rate,
following from equation 6.2 as 1.2·10-4 m3/s at 300 bar. This leads to the choice of
a type P200-65-2 from Resato Int. B.V. (The Netherlands). The pump has two
heads; both suction pipes are chilled in 1.5 meter double-pipe exchangers by water
from the cooling machine.

6.5.4. Dye holder

To prevent dye powder from reaching the dyeing vessel, which would lead to
staining of the textile, the dye has to be molecularly dissolved in the scCO2. A
device is used consisting of a combination of a filter housing and a powder dosing
device, as shown in figure 6.8.

102
After the dye has been dosed, it is kept dye powder falling
from entrainment by a stainless steel into the dye holder
candle filter, with pore size 10 µm.
The scCO2 flows past the dissolving candle filter
particles and through the filter
material towards the dyeing vessel.
The volume of the filter housing is 3
liters. The dosing device has yet to be
developed; the possibility of dosing CO2
powder into a pressurized machine has entering
been proven by Eggers et al. [9].
CO2 and dissolved
dye exiting
6.5.5. Heat exchanger
Figure 6.8. Dye dissolution unit with
As discussed above, each dyeing cycle dosing and candle filter
involves heating and cooling of the
CO2, polyester and steel. To save on
equipment cost, this is both done in one heat exchanger. Because in a dyeing
process the ease of cleaning of the tubes is important, a single-pass exchanger is
used. Two configurations are possible for this equipment item. Firstly, the steam
and cooling water compartments on the shell-side can be separated, as shown in
figure 6.8. Secondly, it is possible to send both the steam and the cooling water
through the same compartment, as is shown in figure 6.9.

In the case of figure 6.9, care should be taken to avoid flow of steam into the
cooling water system and vice versa, because cooling water systems are designed
for lower pressures than steam systems and because the presence of water in steam
pipes leads to water hammer. Therefore, figure 6.9 contains six more valves than
figure 6.8. Four valves separate the cooling water from the steam system and two
valves are needed to squeeze water out of the shell and into the drain, using
pressurized air, when the duty from the exchanger switches from cooling to
heating. The advantage of the configuration in figure 6.8 is an easier control; it is
therefore used in current industrial aqueous dyeing processes. The advantage of the
configuration shown in figure 6.9 is a smaller, cheaper heat exchanger and it is
therefore chosen for the 100-litre dyeing machine.

103
 regulating valve
steam supply
partition
cooling water return

CO2 in

baffles
condensate return
regulating valve
cooling water supply
Figure 6.8. Possible configuration for alternate heating and cooling, with separate
compartments for steam and cooling water

pressurised air

cooling water return

regulating valve steam supply

condensate return

regulating valve cooling water supply

drain

Figure 6.9. Heat exchanger configuration for the 100-litre dyeing machine, with a
single shell-side compartment for both steam and cooling water

104
The temperature of the scCO2 in the dyeing vessel is measured and determines the
flow rate of the steam or cooling water into the heat exchanger via the
corresponding regulating valves. Since the heating and cooling times are yet to be
investigated, the heat exchanger is designed for relatively short, minimum heating
and cooling periods, in both cases 5 minutes.

Heating
The circulation pump, and thus the heating, starts when a pressure of 150-200 bar
is reached. It was concluded in chapter 5 that, for these conditions, the influence of
temperature-induced variations of physical properties in radial direction of a heat
exchanger pipe can be neglected, i.e. a constant-property relation for calculation of
the heat transfer can be used. Also buoyancy can be neglected according to chapter
5.

Walish et al. [10] measured that for Reynolds number larger than 105, upward,
downward and horizontal flows of scCO2 all have the same heat transfer
coefficients (±20%). Since the Reynolds number in the dyeing process is 6·105, a
horizontal heat exchanger for the 100-litre dyeing machine is designed, using the
constant-property equation 5.2. A heating time of 5 minutes is calculated when 23
pipes of 1/4” SCH 40 (external diameter 13.72 mm, wall thickness 2.24 mm, length
1950 mm) are used. A steam temperature of 150ºC is taken, the tube-side heat
transfer coefficient is 8000 Wm-2K-1, making the overall coefficient 2000 Wm-2K-1,
according to equation 5.1.

Cooling
The tube-side coefficient is 8000 Wm-2K-1. The water, from a cooling tower, has a
temperature of 20ºC. The shell-side coefficient is calculated with Kern’s method
[11a] to be 9000 Wm-2K-1. This value is an order of magnitude larger than what is
normal for a shell-side liquid. The reason is the small flow area between the pipes,
leading to large Reynolds numbers and fast heat transfer. The overall heat transfer
coefficient from equation 5.1 is now 2000 Wm-2K-1. The time needed to cool the
scCO2, polyester and steel from 120 to 80ºC is 5 minutes.

The resulting heat exchanger (figure 6.9) was designed in cooperation with and
supplied by GTI Siersema Procestechnologie (The Netherlands). It is built at an

105
incline of 1º to allow the condensate and cooling water to flow out. For the same
purpose, a space of 5 mm is left open between the baffles and the shell wall. To
allow thermal expansion of the exchanger, a bellow is implemented in the shell
wall. All parts of the machine are made of stainless steel (AISI 304), with the
exception of the Teflon® O-rings in the heads.

6.5.6. Regulating valve

In the regulating valve, the pressure of the CO2 changes from 300 bar to 60 bar.
Since this process is fast and the piping is insulated, the expansion is isenthalpic, so
that the temperature-entropy diagram of carbon dioxide gives the temperature
(22ºC) and the mass fraction of liquid (0.4) in the CO2 downstream of the reducing
valve, flowing into the separator vessel.

The valve selection is based on the required flow coefficient KV, generally defined
as:

K V = Q G ∆p (6.3)

where Q is the volume flow (m3/hour), G is the specific gravity or the CO2-density
relative to the density of water at 20ºC and ∆p is the pressure drop over the valve
(bar).

It is calculated that a flow coefficient KV = 0.03 is needed to realize the throughput

(0.4 m3/hour) needed to rinse the vessel sufficiently in 20 minutes, according to
equation 6.2. To depressurize the machine from 300 to 100 bar, the regulating
valve is used; the required flow coefficient for this step is KV = 0.11. The further
pressure drop from 100 to 60 bar is realized by opening a by-pass valve over the
regulating valve. The chosen valve, supplied by Samson Regeltechniek B.V. (The
Netherlands), has characteristics KV = 0.16 and regulating ratio 50:1.

106
6.5.7. Separator

The maximum amount of liquid that is to be boiled off by the steam jacket of the
separator is 0.05 kg/s. A conservative estimation
of the inside film coefficient of 800 W/m2/K [11b]
gives an overall heat transfer coefficient of 400
W/m2/K. With a steam temperature of 150ºC and
internal and external vessel diameters of 0.183 and
0.220 m, it follows that the height of the steam
jacket should be 0.2 m. To compensate for fouling
by dye powder, a height of 0.3 m is taken. The
content of the separator is 20 liters.

To minimize entrainment of precipitated dye

particles, the CO2 is introduced into the bottom of
the vessel tangentially. For the same reason, the
CO2 leaving the separator passes through a filter,
mounted inside the vessel, above the liquid level.
To avoid flooding of the separator, which would
lead to pollution of the storage vessel with dye, the
temperature of the CO2 is measured below and
above the liquid level. The valve regulating the Figure 6.10. Separator
CO2-flow into the separator shuts when the liquid vessel for the 100-litre
rises too much. dyeing machine

All parts of the separator are made of stainless

steel (AISI 316), with the exception of the screw lid (stainless steel 1.6582) and its
Teflon® O-ring. The working pressure being 60 bar, the design pressure of the
separator is taken as 100 bar.

6.5.8. Gas booster

An air-driven piston-cylinder gas booster is used from Maximator Gmbh

(Germany). It was calculated that a type DLE5-2-NN-C has sufficient capacity to
lower the pressure from 60 to 10 bar in the 20 minutes mentioned in table 6.2.

107
6.5.9. Cooling machine

As cooling medium, water with 10% ethylene glycol is used. Since the required
cooling capacity for the suction pipes of the pressurization pump is small, the
capacity of the cooling machine is determined by the heat flow through the
condenser, amounting 12 kW. A CH380 chiller from Piovan (Italy) is chosen, with
a capacity of 50 kW.

6.5.10. Condenser

The flow through the condenser during the rinsing is 0.084 kg/s, requiring a
condenser of minimally 12 kW. The outside film coefficient, following from the
Nusselt relation for condensation outside horizontal tubes [11c], is 2200 W/m2/K.
The overall coefficient becomes 1000 W/m2/K, so that the required pipe length is
20 m, when the cooling medium is 10ºC and flowing through a pipe of 16 mm
outside diameter and 2 mm wall thickness. Inside the condenser, two pipe spirals of
each 16 m are placed concentrically, as shown in figure 6.11.

6.5.11. Storage vessel

To save on equipment cost, the condenser and the storage vessel are combined into
one structure. The storage vessel has a volume of 250 liters, and working and
design pressures of 60 and 100 bar, respectively. The liquid level in the storage
vessel is detected with a differential pressure transmitter.

All parts of the condenser and the storage vessel are made of stainless steel (AISI
316), with the exception of the Teflon®/steel flange seal.

108
 CO2-gas entrance
2 exits for
cooling liquid
inner cooling spiral
outer cooling spiral

2 entrances for
cooling liquid
flange connection between
gas condenser and storage vessel

liquid

CO2-liquid exit

Figure 6.11. Integrated condenser-storage vessel unit for the 100-litre dyeing
machine

Appendix 6.1. Comparison of dye solubilities at 80 and 120º at 300 bar.

Table 6.4. Solubilities (g/m3) of several dyes in supercritical carbon dioxide at 300
bar, for different temperatures (T)
Dye T (ºC) Reference
80 120
Solvent Brown 1 62 73 [12]
Disperse Blue 79 260 425 [13]
Disperse Red 324 36 45 [13]
Disperse Red 153 21 21 [14]
Disperse Yellow 119 12 26 [14]

109
Appendix 6.2 Pressure drop calculation for the 100-litre beam-dyeing
machine.

All pressure drop contributions are calculated for rinsing conditions: 300 bar, 80ºC.

1. Pressure drop over the textile

The fiber diameter was determined to
be 200 µm by taking SEM pictures of
the polyester cloth (figure 6.12). The
Reynolds numbers at the inside and
outside of the textile roll follow from
the velocity (0.1 and 0.02 m/s
respectively), density during rinsing
(745 kg/m3), fiber diameter and
viscosity (6.4·10-5 Pa·s) and amount
230 and 45 respectively. The flow
through the textile roll is laminar and
the corresponding pressure loss can
Figure 6.12. Scanning Electron
thus be calculated with the Darcy
Microscope (SEM) picture of the
equation for radial flow through a
knitted polyester
porous cylinder [15]:

dp µΦ V
= (6.4)
dr 2παL

with p = pressure (Pa), r = radius (m), µ = viscosity, ФV = volume flow rate

(8.3·10-3 m3/s), L = cylinder length (0.45 m) and α is the medium-specific flow
resistance coefficient is assumed to be similar to that in [15]: 1.0·10-11 m-2.

Integrating equation 6.4 from the inside (radius r) to the outside (radius R) of the
textile roll gives the pressure drop ∆p:

µΦ V  R 
∆p = ln  (6.5)
2παL  r 

110
With r = 0.03 m and R = 0.14 m, it follows that the pressure drop contribution from
the textile roll is 0.29 bar.

2. Pressure drop over the piping

The machine has a total of L = 4 m 2” SCH 160 pipes, with internal diameter D =
43 mm. The pipe velocity is 5.7 m/s, Re = 3·106. The Fanning friction factor now
being f = 0.0025 [17a], the pressure drop is ∆p = 2fρv2L/D = 0.1 bar. The flexible
hoses, both 1.5 m in length, have an internal diameter of only 35 mm and therefore
give a pressure loss of 0.2 bar. There are ten 90º bends in the machine, each with a
friction factor KW = 0.8 [17b], giving a joint pressure drop of ∆p = 10 KW 0.5ρv2 =
0.8 bar. The total pressure drop contribution of the piping is thus 1.1 bar.

3. Pressure drop over the dye holder

As discussed below, the CO2 passes through a stainless steel filter candle in the
vessel into which the dye is dosed. The pressure drop follows from the equation
(6.6) which, together with the properties of the SIKA-FIL 10 filter material is given
by the manufacturer: GKN Sinter Metals (Germany) .

Φ V sµ
∆p = (6.6)
Aα

with s = filter layer thickness (0.0008 m), A = filter surface (0.04 m) and α is the
permeability coefficient of the filter material (9·10-12 m2). The contribution to the
pressure drop is relatively small: 0.04 bar. The joint pressure loss in the entrance
and exit of the dye holder (total KW = 0.5 [17b]) and in the sharp 90º bend inside
(KW = 1.5) is 0.2 bar. The total pressure drop over the dye holder is 0.24 bar.

4. Pressure drop over the heat exchanger

As explained below, the heat exchanger has 23 pipes of internal diameter 9.24 mm.
The resulting pipe velocity is 5.4 m/s, so that Re = 6·105, the Fanning friction
factor is 0.004 and the pressure drop 0.3 bar. The entrance from the 2” pipe into the
head and the exit of the 9.24 mm pipes into the other head both give negligible
pressure loss due to the large flow area in the heads. The entrance from the head
into the 9.24 mm pipes is characterized by KW = 1 [17b]. The entrance from the
head into the 2” pipe has a factor KW = 0.5. The resulting pressure drop from

111
entrance and exit effects is 0.05 bar, giving a total pressure drop over the heat
exchanger of 0.35 bar.

The total pressure drop in the 100-litre dyeing machine is concluded to be 2 bar.

References

1. E. Bach, E. Cleve, E. Schollmeyer, M. Bork, P. Koerner, Experience with

the Uhde CO2-Dyeing Plant on Technical Scale. Part 1: Optimization Steps
of the Pilot Plant and the First Dyeing Results, Melliand International 3,
192 (1998).
2. G. Montero, D. Hinks and J. Hooker, Reducing Problems with Cyclic
Trimer Deposition in Supercritical Carbon Dioxide Polyester Dyeing
Machinery, J. Supercrit. Fluids 26(1), 47 (2003).
3. L.J.M.M. Van Schepdael and Stork Prints B.V., Device and Method for the
Piece-wise or Batch-wise Refining of Pieces of Substrate, in Particular a
Textile Substrate, Under High Pressure, Patent WO 2004/009897A1, 2004.
4. L.J.M.M. Van Schepdael, Composietvat voor gebruik bij superkritisch
verven, Feasibility Study Report d.d. 22-03-02, Solico B.V. (The
Netherlands).
5. H.-D. Sung and J.-J. Shim, Solubility of C.I. Disperse Red 60 and C.I.
Disperse Blue 60 in Supercritical Carbon Dioxide, J. Chem. Eng. Data 44,
985 (1999).
6. T. Shinoda and K. Tamura, Solubilities of C.I. Disperse Red 1 and C.I.
Disperse Red 13 in Supercritical Carbon Dioxide, Fluid Phase Equilibria
213, 115 (2003).
7. J.W. Lee, J.M. Min and H.K. Bae, Solubility Measurements of Disperse
Dyes in Supercritical Carbon Dioxide, J. Chem. Eng. Data 44, 684 (1999).
8. S.L. Draper, G.A. Montero, B. Smith and K. Beck, Solubility
Relationships for Disperse Dyes in Supercritical Carbon Dioxide, Dyes and
Pigments 45,177 (2000).
9. R. Eggers, J. von Schnitzler and T. Kempe, Continuous dosing of solids in
high pressure systems, 9th Meeting on Supercritical Fluids, 7th Italian
Conference on Supercritical Fluids and Their Applications, June 2004,
Trieste, Italy.

112
10. T. Walish, W. Doerfler and Ch. Trepp, Heat Transfer to Supercritical
Carbon Dioxide in Tubes With Mixed Convection, HTD volume 350,
ASME National Heat Transfer Conference 12 (1997) 79.
11. J.M. Coulson, J.F. Richardson and R.K. Sinnot, Chemical Engineering
vol.6: Design, 2nd ed., Pergamon Press, Oxford, 1993, a600, b569, c636.
12. A. Ferri, M. Banchero, L. Manna and S. Sicardi, A new Correlation for the
Solubilities of Azoic and Anthraquinone Derivatives in Supercritical
Carbon Dioxide, J. Supercrit. Fluids 32(1-3), 27 (2004).
13. A. Ferri, M. Banchero, L. Manna and S. Sicardi, An Experimental
Technique for Measuring High Solubilities of Dyes in Supercritical Carbon
Dioxide, J. Supercrit. Fluids 30 (1), 41 (2003).
14. H.-M. Lin, C.-Y Liu, C.-H Cheng, Y.-T. Chen and M.-J. Lee, Solubilities
of Disperse Dyes of Blue 79, Red 153 and Yellow 119 in Supercritical
Carbon Dioxide, J. Supercrit. Fluids 21(1), 1 (2001).
15. W.A. Rapp, Y. Yang, G. O’Neal, H. Boyter and G. Hughey, Evaluating
Liquor Flow Profiles in Fabric Beam Dyeing, AATCC Review 2, 42 (2002).
16. J.M. Smith, E. Stammers and L.P.B.M. Janssen, Fysische
Transportverschijnselen I, Delftse Uitgeversmaatschappij, 6th ed., Delft,
1989, a42, b47.

113
114
Chapter 7

Economic Evaluation of Polyester Beam Dyeing in

Supercritical Carbon Dioxide

7.1. Introduction

In the previous chapters, it has been shown that textile dyeing in supercritical
carbon dioxide is technically possible. Whether or not it is also economically
attractive is investigated in this chapter by comparing the cost of a polyester beam
dyeing process in supercritical CO2 with the cost of the same process in water.

Dyeing methods other than beam dyeing are possible (e.g. jet-dyeing, drum dyeing)
but with beam dyeing, the ratio of textile volume and bath volume is the largest. A
small bath volume per unit textile volume means smaller equipment and less CO2
to recycle over the storage vessel. In other words, for a supercritical process to
have low investment- and operating-costs, a beam dyeing process is the best
option.

7.2. Boundary conditions

The raw material of a polyester finishing plant is usually PET fibers contaminated
with spinning oil, an additive that functions as a lubricant during spinning, weaving
and knitting. After the PET is woven or knitted, the spinning oil has to be removed
from the cloth by passing it through a washing machine. This step is the same for
aqueous and supercritical dyeing and therefore these washing costs are left out of
consideration in the comparison of both processes. Before supercritical or aqueous
dyeing, the polyester has to be heat-set to avoid excessive shrinking of the textile in
the dye bath, due to thermal effects. Also this step is identical for the aqueous and
the supercritical process and is left out of consideration.

115
It is assumed that pressurised air, steam and cooling water from a tower are present
at the site. The use of these utilities is incorporated in the evaluation, but the
required compressor, steam boiler and cooling tower are left out of consideration.

The comparison between both processes is done for dyeing batches of 300 kg PET,
an amount that is in the same order as in current industrial practice. The density of
solid polyester is 1380 kg/m3; rolled upon a beam in woven or knitted form it is
250-750 kg/m3; in this work it is taken to be 500 kg/m3 in the roll. The perforated
pipe (beam) onto which the PET is rolled is assumed to be 0.1 m in diameter for
the supercritical and 0.3 m for the aqueous process. Table 7.1 gives an overview of
the boundary conditions for both processes.

Table 7.1. Boundary conditions for the supercritical and aqueous dyeing process
scCO2-dyeing aqueous dyeing

Mass of PET roll (kg) 300 300

Length x diameter of textile roll (m) 2 x 0.63 2 x 0.69
Beam diameter (m) 0.1 0.3
Dyeing temperature (ºC) 130 130
Pressure (bar) 300 3
Desired dye concentration (g/(gPET) 0.02 0.02
Annual process time (hr/yr) 3500 3500

7.3. Process design for polyester beam dyeing in scCO2

For the industrial supercritical dyeing process, the same process steps and times are
used as in chapter 5, leading to a total batch time of 2 hours. Table 7.2 gives an
overview. The same process flow diagram applies as was used in chapter 6.

116
Table 7.2. Process steps and times for polyester dyeing in supercritical carbon
dioxide
Process step time (min)
Loading textile and dye 3
Pressurisation 15
Dyeing 45
Levelling 15
Rinsing 20
Depressurisation 20
Unloading textile 2
Total batch time 120

A dye is chosen with a solubility of 10-4 g/(g CO2). As in chapter 6, it is assumed

that the scCO2 can be saturated with dye in the dye holder and that the dye
concentration in the scCO2 exiting the PET roll is negligible. The density of the
scCO2 is 550 kg/m3. The transport of 6 kg dye (2 mass % of the textile) to the PET
in the dyeing time of 45 minutes requires a flow of 145 m3/hour. Assuming a flow
rate of 6 m/s, this requires standard 4” schedule 160 piping, with inner diameter 87
mm and wall thickness 13.5 mm.

The pressure drop calculations are done in the same manner as in chapter 6, leading
to a drop of maximally 0.3 bar over the PET roll and 0.6 bar over the piping. The
pressure drop over the heat exchanger is 0.2 bar and also over the dye holder 0.2
bar. The total pressure drop, to be overcome by the circulation pump, is 1.3 bar.

7.4. Equipment for supercritical dyeing

Since no industrial-scale supercritical dyeing machine exists at this moment, the

capital cost is calculated by designing such a machine and estimating the purchase
costs of all components. Below, the main plant items are discussed.

Dyeing vessel
A dyeing vessel is used with a stainless steel liner reinforced by carbon fibres
wound around the circumference, thereby taking up the radial pressure forces. The
two lids are kept in place by a steel yoke that takes up the axial pressure forces.

117
The length of the vessel is 3 m: 2 for the PET roll and 1 m for the pump and motor
discussed below. As the textile roll is 0.63 m, the inner diameter of the dyeing
vessel is taken as 0.7 m.

The advantage of using a carbon fibre vessel instead of an entirely steel vessel is
twofold. Firstly, a steel vessel of inner diameter 0.7 m and a design pressure of 350
bar would have a wall thickness of 0.11 m, making the vessel mass 7000 kg,
without yoke or closure. For heating and cooling the CO2, PET and steel, there is
now twice the amount of steam and cooling water per batch needed. The second
disadvantage of a steel pressure vessel is that it requires a heat exchange area twice
as large as the carbon fibre vessel, if the heating and cooling times are to be equal
to the carbon fibre case. The extra costs for cooling water and steam and for the
larger heat exchanger lead to an increase of 6 % in the process cost relative to the
carbon fibre case.

Circulation pump
The choice and price of the circulation pump is determined by the combination of
flow (145 m3/hour) and pressure drop (1.3 bar) and by the working pressure of 300
bar. The high flow means that a centrifugal pump is required. The high working
pressure demands that a magnetic coupling be used between the pump and its
motor. Such a pump would cost approximately 100 k€ [2] and weighs heavily on
the investment cost. An alternative is to use a standard low-pressure centrifugal
pump and place it inside the pressure vessel, together with the motor. The cost of
pump and motor is now reduced to 20 k€ [3]. It can be calculated from the tables
below that the total process cost is then reduced with 5%. Dry-running bearings are
used and a special resin for the electrical spools of the motor, as was discussed in
chapter 6.

Pressurisation pump unit

To pump liquid CO2 from 60 to 300 bar, either a membrane pump or an air-driven
piston-cylinder pump can be used. The latter type is chosen here, since it has a 3 to
4 times lower purchase cost. The pressurisation unit contains 6 pumps, with a total
flow of 70 liter/min.

118
Dye holder
The pressure vessel where the dye powder is dissolved in the scCO2 is a stainless
steel filter housing of 200 mm internal diameter and height 600 mm. The dye is put
into the vessel, upstream of the filter, so that the scCO2 flows past the dye particles
and then through the filter.

Heat exchanger
As was discussed in chapter 6, there are 2 possible configurations for the heat
exchanger: one in which the steam and cooling water flow through the same shell-
side space and one in which the steam- and cooling water parts are segregated. The
latter configuration requires less valves,is easier to use in an industrial environment
and is therefore chosen here.

The heat exchange area for heating and cooling the scCO2, steel and PET is
calculated in the same manner as in chapter 6. As is shown in chapter 5, the heat
transfer coefficients for heating and cooling can be calculated with constant-
property Nusselt relations. Both the steam- and the cooling water part of the shell
have a heat exchange area of 3.4 m2. Heating and cooling can be accomplished in
10 minutes if 50 stainless steel pipes of type ½” shedule 40 are used. Both the
heating and the cooling part should be 1 m in length, the shell diameter is 250 mm.

Separator
The precipitation and collection of residual dye is done in a separator vessel. By
means of a steam jacket, the liquid fraction of the CO2 that results from the
expansion from 300 to 60 bar is boiled off and flows to the storage vessel. The
separator is designed in the same way as in chapter 6, resulting in a vessel with an
internal diameter of 0.3 m and a height of 1 m.

Gas booster unit

The gas boosters are of the air-driven piston-cylinder type. The unit is used to
pump the gaseous CO2 out of the dyeing vessel, after the pressures in the dyeing
vessel and the storage vessel are equalised. The unit consists of 2 gas boosters, with
a total flow of 50 liters/minute and is able to lower the pressure in the dyeing vessel
from 60 to 10 bar in 20 minutes. The remaining CO2 (20 kg) is vented to the
atmosphere, which means that 96% of the CO2 is recycled per batch.

119
Condenser
The condenser is designed to liquify the flow of CO2 leaving the separator during
the rinsing step: 0.83 kg/s. The same condenser configuration is taken as in chapter
6: CO2 flowing on the shell-side and the cooling water on the tube-side. With a
cooling water temperature of 0ºC, the required area of heat exchange is 6.4 m2. The
calculation procedure is similar to the one in chapter 6.

Cooling machine
The cooling water needed for the condenser is brought to 0ºC with an electrical
cooling machine. The required capacity is estimated to be 140 kW.

Storage vessel
The CO2 is stored at 22ºC and 60 bar, in liquid form. The required pressure vessel
has a volume of 1200 liters.

7.5. Capital costs for supercritical and aqueous dyeing

The capital cost of the main plant items discussed above are listed in table 7.3.

Table 7.3. Investment cost for the main plant items of an supercritical dyeing
machine
Main plant item Capacity Purchase cost (k€)
Dyeing vessel 1150 liter 90a
3
Circulation pump 150 m /hr 20b,c
Pressurisation pump unit 70 liter/min 20d
Dye holder 20 liter 20e
Heat exchanger 2 x 3.4 m2 30e
Separator 70 liter 20e
Gas booster unit 50 liter/min 14f
Condenser 6.4 m2 20e
Cooling machine 140 kW 55g
Storage vessel 1200 liter 20h
Total main plant items: 309 k€

120
Price estimations from: aSolico B.V., bPacko Inox N.V., cDACE Price booklet,
d
Resato Int. B.V., eEstimations based on costs of the 100 liter pilot machine from
chapter 6, fMaximator GmbH, gPiovan S.p.A., hVan Steen Apparatenbouw.

The secondary capital costs are presented in table 7.4.

Table 7.4. Overview of the secondary equipment costs for a supercritical dyeing
machine
Cost (k€)
Instrumentation and control 25a
Piping 10b
Valves 20a
Frame and isolation 20a
Working hours 14c
Total secondary costs: 89 k€
a
Estimations based on costs of the 100 liter pilot machine from chapter 6, bNoxon Stainless B.V., c10 weeks à
€35/hr.

From tables 7.3 and 7.4 it follows that the cost of producing the supercritical
dyeing machine is 398 k€. If an overhead of 25 % is assumed, the purchase cost for
a textile dyer would be 500 k€ for a beam dyeing machine for 300 kg polyester
batches. This is to be compared with the purchase cost of a machine for aqueous
beam dyeing: 100 k€ [4].

For both the supercritical and the aqueous process, the annual capital charge a is
calculated with the annuity equation:

i
a= (7.1)
1 − (1 + i) − N

With an interest rate i of 6 % and a depreciation time N of 10 years, the annual

capital charge is 14 %. Table 7.5 gives an overview of the resulting capital charges
for both processes.

121
Table. 7.5. Capital costs for polyester beam dyeing in scCO2 and water
CO2 Water
Investment k€ 500 100
Annual capital charge k€ / year 70 14
Batch time hour 2 3.5a
Production capacity ton / year 525 300
Capital charge € / batch 40 14
€ / (kg PET) 0.13 0.047
a
Data from Ames Europe B.V. and Stork Prints B.V.

7.6 Operating costs for supercritical and aqueous dyeing

The prices of the different utilities and compounds that are needed to operate a
dyeing machine are listed in table 7.6.

Table 7.6. Prices of utilities, relevant compounds and operator labour

Unit Cost (€ / unit)
3
Pressurised air (7 bar) Nm 7·10-3 a
Steam ton 26b
Electricity kWh 0.10a
Cooling water (tower) m3 0.05a
Water m3 2.27c
CO2 kg 0.11d
Operator labour hour 28
a
DACE price booklet, bData from Stork Prints B.V., cPrice of purchasing, purification and disposal of water, data
from Stork Prints B.V., dHoekLoos B.V.

The use of utilities and compounds is calculated from the specifications of the
equipment and the demands of the process. The heat balance is made with the data
from table 7.7.

122
Table 7.7. Data for the heat balances of the dyeing processes
Material Specific heat Heat of Amount in dyeing machine (kg)
(J/kg/ºC) vaporisation CO2-process Aqueous
(kJ/kg) process
Steel 460 1700 1000b
PET 1300 300 300
CO2 2200a 140 560 0
c
Water 4200 2200 0 1000b
a
Average value, taken between 20 and 130ºC. bData from Stork Prints B.V. c At 150ºC

Supercritical process
In table 7.8 the calculated operational costs are shown for the supercritical process.

Table 7.8. Operational costs per batch for polyester beam dyeing in scCO2
Item Amount per batch Price (€)
3
Pressurised air 1110 Nm 7.55
Steam 137 kg 3.55
Electricity 105 kWh 10.50
3
Cooling water 16 m 0.80
CO2 20 kg 2.20
Dyes 6 kg 60
a
Maintenance 1.2
Labour costb 0.67 hour 18.67
Total operating cost: 104.47 € / batch
0.35 € / (kg PET)
a
Maintenance is 3 % of the capital cost. b 1 operator for 3 machines

The steam consumption in the table is the sum of the amount of steam needed to
heat all steel, PET and CO2 and the amount that is needed to vaporise the liquid
fraction of CO2 that enters the separator during the rinsing step of the process.

The electricity use of the process is calculated by adding the use of the motor of the
centrifugal pump and the use of the cooling machine.

The cost of the dye powder in table 7.8 is based on the current price of dye for
aqueous dyeing: 15 €/kg. This includes dispersing agents and since these are not

123
needed in supercritical dyeing, the dye will be cheaper. It is therefore roughly
estimated to be 10 €/kg. It follows from the distribution coefficients given in
chapter 4 that the amount of dye remaining in the scCO2 after the dyeing process is
small compared to the amount in the PET; it is therefore neglected.

Aqueous process
The dye for the aqueous process contains around 50 % actual dye molecules, the
rest is mainly dispersing agents. Therefore, to obtain 2 mass% of dye into the 300
kg of PET, 12 kg dye is needed. Also in the aqueous process, the dye concentration
in the machine after the process is neglected.

After the dyeing step, the PET is washed in 1000 liter water with surfactants at
90ºC and subsequently rinsed 3 times with 100 liter water at 60ºC. The time
required for dyeing, washing and rinsing together is 3.5 hours [4]. Finally, the PET
is dried, using steam. The resulting operating costs are given in table 7.9.

Table 7.9. Operating costs per batch for aqueous polyester beam dyeing
Process step Item Amount per batch Price (€)
Maintenance 3
Dyeing Steam 253 kg 6.58
Electricity 50 kWha 5
Water 1000 litera 2.27
Dyes 12 kg 180
b a
Chemicals 3 kg 9
Labour cost 32.67c
Washing and rinsing Steam 319 kg 8.30
a
Water 4000 liter 9.08
Chemicals 12.3a
a
Electricity 50 kWh 5
Drying Steam 910 kg 23.66
Total operating costs: 297 €/batch
0.99 € / (kg PET)
a
Information from Stork Prints B.V. bLevelling agents and pH modifiers. c Labour cost for dyeing, washing and
rinsing together, 1 operator for 3 machines.

124
7.7. Total process costs

The costs for supercritical and aqueous dyeing are summarised in table 7.10. It is
clear that the capital costs are higher and the operating costs are lower for the
supercritical process. Because the process costs are largely determined by the
operating costs, the process costs are lower when polyester is dyed in scCO2.

Table 7.10. Comparison of costs (€ / (kg PET) for supercritical and aqueous
polyester dyeing
CO2 Water
Capital cost 0.13 0.047
Operating cost 0.35 0.99
Process cost 0.48 1.04

7.8. Discussion on the proess cost

For the process time, 2 hours was taken in the above calculations. It is possible
that, in industrial practice, the time will be longer. It will, however, always be
shorter than for the aqueous process. To illustrate the influence of the time on the
costs of the supercritical process, the calculations were performed also for a time of
3 hours. The capital and operating costs now become 0.20 and 0.42 €/(kg PET)
respectively. The total process cost increases from 0.48 to 0.64 when the time
changes from 2 to 3 hours.

In the calculation of the operating costs for supercritical dyeing, the price of dye
was estimated roughly to be 10 €/kg, since there is no dye commercially available
for dyeing in scCO2. The value must lie between the price of the press cake, the
unpurified product of dye synthesis (4 €/kg) and the price of dye for aqueous
dyeing, containing dispersing agents (15 €/kg). This means that the process costs
for the supercritical process lie between 0.36 and 0.58 € per kg polyester.

In the supercritical process evaluated above, the dye remaining in the CO2 after
dyeing, is separated from the CO2 by vaporisation of the latter. This separation
technique requires continuous vaporisation, condensation and pressurisation of the
CO2 during the rinsing step. An energetically prefferable method is to remove the

125
dye from the CO2 by adsorption onto a medium like activated carbon. This
separation technique was investigated for dye and scCO2 on a laboratory scale by
Von Schnitzler [5], with bleaching earth as adsorptive material. It is unknown how
this would affect the process time and it is also unknown what the price of
regeneration or disposal of the adsorptive medium is. Therefore, at this moment,
the costs of a supercritical process with this alternative separation cannot be
estimated.

7.8. Environmental considerations

Closely related to the economical evaluation are the environmental implications of

the dyeing process. The dye solvent (water or scCO2) , the chemicals (mainly dyes
and dispersing agents) and the different types of energy consumed during the
dyeing process are all quantified in this chapter in terms of costs but their use also
has environmental consequences.

An quantitative environmental comparison between supercritical and aqueous

dyeing can only be made by making life cycle analyses of all compounds and
utilities involved. This is beyond the scope of this thesis, but the economical
comparison between the supercritical and aqueous processes can also be used as a
rough environmental comparison, since the prices of compounds and utilities are a
measure for the resources required to produce them. It is possible however, to get a
clearer picture of this issue by regarding the solvents, chemicals and energy types
qualitatively.

It will be clear that the main environmental advantage of supercritical dyeing is the
reuse of 95 % of the solvent. Because the CO2 is obtained as a waste product from
industrial processes like combustion or ammonia synthesis, the venting of the 5 %
CO2 is not a pollution on the account of textile dyeing. As discussed in chapter 1,
in polyester dyeing 5-10 % of the dye is not used and ends up in the wastewater. In
scCO2-dyeing, all dye can be reused, which forms a second environmental
advantage of the supercritical process. In short, the supercritical process is
environmentally superior to the aqueous process when the use of compounds is
regarded. Table 7.11 gives an overview of the use of compounds and energy in
both the supercritical and the aqueous process. The disperse dyes that are used in

126
supercritical dyeing are the same as in aqueous dyeing, so that the human and
ecological toxicity of these compounds does not play a role in the comparison of
the processes.

It can be seen in table 7.11 that different types of energy are used in both dyeing
processes. The use of cooling water is neglected here. For a comparison with
respect to energy consumption, it is calculated how much carbon dioxide is formed
in the production of the pressurized air, steam and electricity. The generation of 1
Nm3 pressurised air of 7 bar requires 0.125 kWh [6], the CO2-production for 1
kWh electricity is 0.61 kg [7] and the CO2-production for steam formation is 118
kg CO2 / ton steam [7]. The resulting amounts of carbon dioxide associated with
the energy use in the dyeing processes are given in table 7.12. Supercritical dyeing
requires less energy and is therefore associated with a 26% lower CO2-emission.

Table 7.11. Environmental comparison between the supercritical and the aqueous
polyester dyeing process, regarding the uses of compound and utilities
compound/ unit CO2 water
utility
water kg 0 5000a
CO2 kg 20 0
dye pure dye kg 6 6
dispering agent kg 0 6
chemicals kg 0 3
3
pressurized air Nm 1110 0
steam kg 137 1380c
electricity kWh 105 100
3
cooling water m 16 0
a b c
For dyeing, washing and rinsing. 5% of the amount of CO2 that is used per batch (560 kg). For dyeing,
washing, rinsing and drying.

127
Table 7.12. Carbon dioxide emission associated with the energy used in
supercritical and aqueous dyeing, per 300 kg polyester batch
energy equivalent CO2-emission (kg CO2)
supercritical dyeing aqueous dyeing
steam 16 163
electricity 64 61
pressurised air 85 0
total 165 224

7.9. Conclusions

From the process and equipment design for beam dyeing of 300 kg batches of
poyester in scCO2, it follows that the purchase cost for the machine is 500 k€. The
equivalent machine for dyeing in water costs 100 k€. The large difference in
investment cost is caused by the high pressure of the supercritical process and by
the fact that 96 % of the CO2 is recycled.

The operating costs for the supercritical process are lower than for the aqueous
process: 0.35 and 0.99 € per kg of polyester, respectively. This is mainly caused by
the higher rate of dyeing and the lower price of dye for the supercritical process.

The overall effect is that it is more than twice as cheap to dye polyester in scCO2
than it is in water. The process costs are 0.48 and 1.04 € per kg of polyester,
respectively.

The economic evaluation was based on a supercritical process where the residual
dye is separated from the CO2 by gasifying the latter. When, instead, the dye is
removed by adsorption, the supercritical process could be made cheaper. At this
moment, no estimation can be made as to how much cheaper.

As the supercritical process requires less compounds and energy, it is concluded to

be environmentally benign compared to the aqueous process.

The overall conclusion is that dyeing polyester in scCO2 instead of water, is

environmentally superior and reduces the costs with 50 %.

128
References

1. J. v. Zuidam, Ames Europe B.V., Enschede, The Netherlands, personal

communication, 2003-2004.
2. G.F. Woerlee, FeyeCon D&I B.V., internal report, april 1999.
3. DACE price booklet, Dutch Association of Cost Engineers, 22nd ed.,
Elsevier, 2002.
4. P. Leerkamp, Stork Prints B.V., internal report, february 2003.
5. J. Von Schnitzler, Der Stofftransport in Faerbeprozessen von Polymeren
mit Ueberkritischem CO2, Dissertation Technische Universitaet Hamburg-
Harburg, 2001.
6. H. van Heyster, GrassAir, personal communication.
7. Tufts Institute for the Environment, Method for Conducting a Greenhouse
Gas Emission Inventory for Colleges and Universities, april 2002.

129
130
Epilogue

Suggestions for further research

As was concluded in this thesis, one of the main reasons that the supercritical
process is economically superior to the aqueous process, is the higher rate of
dyeing. It is therefore recommendable to investigate the kinetics of both disperse
and reactive dyeing. Knowledge on the kinetics and the identification of the rate-
determining step in non-reactive and reactive dyeing open the possibility of further
reducing the process times.

For disperse polyester dyeing, a large number of dyes is commercially available.

With respect to reactive dyeing however, more dyes should be synthesized and
tested in scCO2, so that the whole color spectrum can be covered.

Regarding the equipment, both the rotating drum machine (chapter 4) and the beam
dyeing machine (chapter 6) are to be tested for their applicability in reactive
dyeing. The process time and the evenness of the beam have to be measured and
optimized in the 100-liter dyeing machine described in chapter 6, both for disperse
and reactive dyeing. Finally, an equipment item should be designed, built and
tested to introduce the dye powder into the pressurized dyeing machine.

Future of supercritical textile dyeing

It is clear that a large driving force exists for the development and implementation
of textile dyeing in scCO2. Water pollution is becoming an ever greater
environmental concern and it is likely that the regulations on the quality of treated
wastewater will become more stringent in the future. Therefore, the end-of-pipe
treatments will require more effort and cost, so that the economical advantage of
scCO2 will become even greater. Both the environmental and the economical
incentive will increase the demand for a clean dyeing process.

For the implementation of a new dyeing process in industry, the environmental and
economical implications, the product quality (color depth, fastness) and the

131
robustness of the process in practice have to be ascertained. Regarding polyester
dyeing, the environmental advantage is clear. The second two issues were
investigated satisfactorily in this thesis. The development of the 100-liter polyester
beam dyeing machine, described in chapter 6, creates the possibility of carrying out
large-scale experiments to test the robustness of the process. Only then can a full-
scale 1000-liter pilot machine be designed, built and tested in an industrial
environment.

Also for reactive dyeing, the environmental advantage is straightforward.

Insufficient information is available, e.g. on kinetics, to evaluate the process
economically. The product quality was found to be good in this work: dark colors
were obtained and a good fastness. To assess the robustness of the process of
reactive dyeing, there is not enough information regarding large-scale experiments.
Another issue in implementation concerns the dyes: disperse dyes are
commercially available but new reactive dyes have to be designed and synthesized
first on a small scale, their performance is to be tested in scCO2 on a small and then
on a large scale, investigated with respect to toxicity and finally synthesized on an
industrial scale.

In view of the above, disperse dyeing of polyeste will probably be implemented in

industry first and reactive dyeing will follow. This thesis treats several of the issues
that need to be solved before implementation in industry is possible. A machine is
designed and built with which the robustness of disperse dyeing can be tested and
the performance of reactive dyes in scCO2 is investigated. Therefore, this thesis is
an important step towards the implementation of disperse and reactive textile
dyeing in supercritical carbon dioxide.

132
Dankwoord

Tijdens de vier jaren van mijn promotieonderzoek op API heb ik met veel mensen
samengewerkt. De mede- “API-bewoners” zorden ervoor dat het niet alleen een
leerzame en productieve tijd was, maar ook een vrolijke.

Allereerst wil ik mijn promotor, Geert-Jan Witkamp, bedanken voor het

vertrouwen en de kans aan een interessant en veelzijdig onderzoek te werken. Ook
Geert Woerlee bedank ik hiervoor. En, Geert: Bedankt voor de hulp op momenten
dat ik vast kwam te zitten. Cor Peters wil ik hierbij bedanken voor de vriendelijke
hulp met hoofdstuk 3.

Met Wim Veugelers heb ik het genoegen gehad veel samen te werken. Een
genoegen vanwege je efficiënte manier van werken en je praktijkervaring, Wim,
maar ook omdat je een bijzonder prettige collega bent, altijd bereid tijd vrij te
maken om anderen te helpen.

Jan, Martijn, Carel, Stefan, André, Jos en Gerard: Hartelijk dank voor de inzet bij
het tekenen en construeren van de opstellingen en jullie praktische adviezen.
Bovendien wil ik jullie bedanken voor de vele keren dat er reparaties en
veranderingen aan de apparatuur moesten worden uitgevoerd.

Een grote hulp bij het onderzoek waren Marjan, Özcan, Frans, Elmehdi en Zinaida.
Hun werk, in de vorm van experimenten en CFD-simulaties, is overal in dit
proefschrift terug te vinden.

Bij de analyses van de geverfde lappen is meermalen de kennis van Michel en Paul
van grote waarde gebleken. Hen wil ik bedanken voor de hulp bij zaken als
chemicaliën en analytische apparatuur.

Tijdens mijn gehele promotie heb ik veel geleerd van de mensen die vanuit Stork
Prints B.V. op het project werkten. Vooral Jan-Willem Gerritsen en Peter
Leerkamp bedank ik hiervoor. Ook de medewerkers van Ames Europe en Triade
wil ik hierbij bedanken voor hun bijdragen. Ludo en Hans van Solico zijn we
dankbaar voor het prachtige ontwerp van het koolstofvezelversterkte verfvat.

133
Vanaf het begin van mijn API-tijd heb ik veel steun en ideën gekregen van de
medewerkers van FeyeCon D&I B.V. Behalve Geert en Wim zijn Ernst, Frank,
Gerard en Hubert met hun ervaring op superkritisch gebied behulpzaam geweest.
Bedankt!

Vanesa, bedankt voor de aangename samenwerking en de gezellige tijd, in de

kamer maar ook op de verschillende congressen en cursussen. Christof, jou wil ik
bedanken voor het feit dat je mij, als afstudeerder, op het superkritische pad hebt
gebracht. Daar heb ik tijdens mijn promotie geen spijt van gehad. Maaike, bedankt
voor de collegialiteit en de gezellige vrijdagavonden met Sietse en jou. Aris en
Marcel, gelukkig heb ik met jullie verschillende keren wat stoom kunnen afblazen
op momenten dat overdruk dreigde. Jaap, Yohana, Christiana, Frans, Perize,
Maaike, Helma, Andreanne, Paolo, Abilash, Stephen, Peter, Mark, Elif, Robert,
Brenda, Dima, Marcos, Daniela, Joop en de andere stafleden wil ik bedanken voor
de vriendschap en voor de goede sfeer op API. Jelan, bedankt voor je hulp met de
omslag.

Uschi, jouw steun is heel belangrijk geweest de afgelopen jaren. Ook voor je
geduld wil ik je bedanken, er ging nogal wat tijd en energie in deze klus zitten. Nu
hebben we gelukkig weer meer tijd voor elkaar.

Tot slot wil ik opmerken dat ik zonder de goede zorgen van mijn ouders nooit tot
een promotie gekomen zou zijn. Bedankt voor alles!

Martijn.

134
Curriculum Vitae

Martijn van der Kraan

Born: June 5th 1971, Nieuwkoop, The Netherlands

1983 - 1989 Atheneum, Rijksscholengemeenschap Bergen op Zoom

1989 – 2001 Chemical Engineering, Delft University of Technology

2001 – 2005 Ph.D. at the Laboratory for Process Equipment

Delft University of Technology

2005 Post-doc at the Laboratory for Process Equipment

Delft University of Technology
Project Manager, FeyeCon Development and Implementation B.V.

135
136
Publications

M. van der Kraan, M.M.W. Peeters, M.V. Fernandez Cid, G.F. Woerlee, W.J.T.
Veugelers and G.J. Witkamp The Influence of Variable Physical Properties and
Buoyancy on Heat Exchanger Design for Near- and Supercritical Fluids, J.
Supercrit. Fluids 34, 99 (2005).

M. van der Kraan, M.V. Fernandez Cid, G.F. Woerlee, W.J.T. Veugelers and G.J.
Witkamp, Equilibrium Study on the Disperse Dyeing of Polyester in Supercritical
Carbon dioxide, submitted to The Textile Research Journal.

M. van der Kraan, M.V. Fernandez Cid, G.F. Woerlee, W.J.T. Veugelers and G.J.
Witkamp, Dyeing of Natural and Synthetic Textiles in Supercritical Carbon
Dioxide with Disperse Reactive Dyes, submitted to The Journal of Supercritical
Fluids.

M. van der Kraan, M.V. Fernandez Cid, G.F. Woerlee, W.J.T. Veugelers, C.J.
Peters and G.J. Witkamp, Equilibrium Distribution of Water in the Two-phase
System Supercritical Carbon Dioxide – Textile, submitted to the Journal of
Supercritical Fluids.

M. van der Kraan, O. Bayrak, M.V. Fernandez Cid, G.F. Woerlee, W.J.T.
Veugelers and G.J. Witkamp, Textile Dyeing in Supercritical Carbon Dioxide,
Proceedings of the 6th International Symposium on Supercritical Fluids, Versailles,
France, 2003, 2119.

M. van der Kraan, M.M.W. Peeters, M.V. Fernandez Cid, G.F. Woerlee, W.J.T.
Veugelers and G.J. Witkamp, Heat Transfer in High-pressure Carbon Dioxide,
Proceedings of the 6th International Symposium on Supercritical Fluids, Versailles,
France, 2003, 517.

M. van der Kraan, M.V. Fernandez Cid, G.F. Woerlee, W.J.T. Veugelers and G.J.
Witkamp, Dyeing Natural and Synthetic Textiles in Supercritical Carbon Dioxide

137
with Reactive Dyes, Proceedings of the 4th International Symposium on High
Pressure Technology and Chemical engineering, Venice, Italy, 2002, 199.

M. van der Kraan, E. Abdaou, M.V. Fernandez Cid, W.J.T. Veugelers, G.F.
Woerlee, G.J. Witkamp, Maximum attainable Colouration of Polyester in
Supercritical Carbon Dioxide, Proceedings of the 7th Italian Conference on
Supercritical Fluids and their Applications, Trieste, Italy, 2004.

138