Solubility of sugars and sugar alcohols in ionic liquids: Measurement and PC-SAFT modeling

Article abstract of DOI:10.1021/jp404864c

Biorefining processes using ionic liquids (ILs) require proper solubility data of biomass-based compounds in ILs, as well as an appropriate thermodynamic approach for the modeling of such data. Carbohydrates and their derivatives such as sugar alcohols represent a class of compounds that could play an important role in biorefining. Thus, in this work, the pure IL density and solubility of xylitol and sorbitol in five different ILs were measured between 288 and 339 K. The ILs under consideration were 1-ethyl-3-methylimidazolium dicyanamide, 1-butyl-3-methylimidazolium dicyanamide ([bmim][DCA]), Aliquat dicyanamide, trihexyltetradecylphosphonium dicyanamide, and 1-ethyl-3-methylimidazolium trifluoroacetate. Comparison with the literature data was performed, showing good agreement. With the exception of [bmim][DCA], the solubility of these sugar alcohols in the other ILs is presented for the first time. The measured data as well as previously published solubility data of glucose and fructose in these ILs were modeled by means of PC-SAFT using a molecular-based associative approach for ILs. PC-SAFT was used in this work as it has shown to be applicable to model the solubility of xylitol and sorbitol in ILs (Paduszynì?ski;et al. J. Phys. Chem. B 2013, 117, 7034-7046). For this purpose, three pure IL parameters were fitted to pure IL densities, activity coefficients of 1-propanol at infinite dilution in ILs, and/or xylitol solubility in ILs. This approach allows accurate modeling of the pure IL data and the mixture data with only one binary interaction parameter k_ij between sugar and the IL or sugar alcohol and the IL. In cases where only the pure IL density and activity coefficients of 1-propanol at infinite dilution in ILs were used for the IL parameter estimation, the solubility of the sugars and sugar alcohols in the ILs could be predicted (k_ij = 0 between sugar and the IL or sugar alcohol and the IL) with reasonable accuracy. ? 2013 American Chemical Society.

Full text of DOI:10.1021/jp404864c

Subscriber access provided by Saint Joseph's University | Francis A. Drexel Library
Article
Solubility of Sugars and Sugar Alcohols in Ionic
Liquids: Measurement and PC-SAFT Modeling
Aristides Pinto Carneiro, Christoph Held, Oscar Rodriguez,
Gabriele Sadowski, and Maria Eugénia Rebello Almeida Macedo
J. Phys. Chem. B, Just Accepted Manuscript • DOI: 10.1021/jp404864c • Publication Date (Web): 30 Jul 2013
Downloaded from http://pubs.acs.org on August 5, 2013
Just Accepted
“Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted
online prior to technical editing, formatting for publication and author proofing. The American Chemical
Society provides “Just Accepted” as a free service to the research community to expedite the
dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts
appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been
fully peer reviewed, but should not be considered the official version of record. They are accessible to all
readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered
to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published
in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just
Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor
changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers
and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors
or consequences arising from the use of information contained in these “Just Accepted” manuscripts.
The Journal of Physical Chemistry B is published by the American Chemical Society.
1155 Sixteenth Street N.W., Washington, DC 20036
Published by American Chemical Society. Copyright © American Chemical Society.
However, no copyright claim is made to original U.S. Government works, or works
produced by employees of any Commonwealth realm Crown government in the course
of their duties.

Page 1 of 56
The Journal of Physical Chemistry
1
2
3
4
5
Solubility of sugars and sugar alcohols in ionic liquids: measurement and PC-SAFT
modeling
6
7
8
9
Aristides P. Carneiro a,Christoph Held *,b, Oscar Rodríguez a,†, Gabriele Sadowski b,
Eugénia A. Macedo a
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
a LSRE – Laboratory of Separation and Reaction Engineering – Associate Laboratory LSRE/LCM,
Faculdade de Engenharia, Universidade do Porto, Rua Dr. Roberto Frias, 4200-465 Porto,
Portugal
b Laboratory of Thermodynamics, Department of Biochemical and Chemical Engineering,
Technische Universität Dortmund, Emil-Figge-Str. 70, 44227 Dortmund, Germany
† Present address: Department of Chemical Engineering, University of Santiago de Compostela, Spain.
ACS Paragon Plus Environment

The Journal of Physical Chemistry
Page 2 of 56
1
2
3
4
Abstract
5
6
7
8
9
Biorefining processes using ionic liquids (ILs) require proper solubility data of biomass-
based compounds in ILs, as well as an appropriate thermodynamic approach for the
modeling of such data. Carbohydrates and their derivatives such as sugar alcohols
represent a class of compounds that could play an important role in biorefining. Thus,
in this work pure-IL density and solubility of xylitol and sorbitol in five different ILs
were measured between 288 and 339 K. The ILs under consideration were: 1-ethyl-3-
methylimidazolium dicyanamide, 1-butyl-3-methylimidazolium dicyanamide, Aliquat®
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
dicyanamide,
trihexyltetradecylphosphonium
dicyanamide,
and
1-ethyl-3-
methylimidazolium trifluoroacetate. Comparison with the literature data was
performed, showing good agreement. With the exception of [bmim][DCA], the
solubility of these sugar alcohols in the other ILs is presented for the first time. The
measured data as well as previously published solubility data of glucose and fructose in
these ILs were modeled by means of PC-SAFT using a molecular-based associative
approach for ILs. PC-SAFT was used in this work as it has shown to be applicable to
model the solubility of xylitol and sorbitol in ILs (Paduszyński et al., J. Phys.Chem. B
117, 7034-7046, 2013). For this purpose, three pure-IL parameters were fitted to
pure-IL density, activity coefficients of 1-propanol at infinite dilution in ILs, and/or
xylitol solubility in ILs. This approach allows accurate modeling of the pure-IL data and
the mixture data with only one binary interaction parameter kij between sugar and IL
or sugar alcohol and IL. In cases where only pure-IL density and activity coefficients of
1-propanol at infinite dilution in ILs were used for the IL-parameter estimation, the
solubility of sugars and sugar alcohols in ILs could be predicted (kij=0 between sugar
and IL or sugar alcohol and IL) with reasonable accuracy.
Keywords: sugars, sugar alcohol, ionic liquids, solubility, PC-SAFT, modeling, prediction
* corresponding author: Christoph.Held@bci.tu-dortmund.de
ACS Paragon Plus Environment

Page 3 of 56
The Journal of Physical Chemistry
1
2
3
4
1. Introduction
5
6
7
8
It is common sense that future sustainability depends on the exploitation of alternative
sources for energy and chemicals. Biomass-derived sugars and sugar alcohols (both
referred to as “sugars” in the following) are potential precursors of commodity
chemicals and added-value compounds1. Processing biomass in a biorefinery requires
suitable solvents, which are able to dissolve biomass, catalyze the hydrolysis of the
main polysaccharides (cellulose and hemicellulose) to sugars, and promote the
transformation of sugars into intermediate chemicals or final products. Ionic liquids
(ILs), a still recent class of "molten salts", have been extensively applied in academia as
suitable solvents for many applications in a broad spectrum of fields (e.g. 2-4). Their
unique properties 5-7, mainly the so-called "non-volatility" and high thermal stability,
made them attractive tools to face many operatory challenges. Their versatility allows
the properties to be "tailored" by choosing an appropriate anion-cation pair. Thus, an
IL can be designed according to the physical or chemical demands for a given
application. In biomass processing, ILs have been successfully applied as solvent media,
extraction agents, as well as catalysts for several reactions 8,9. The main advantage of
their use in biorefining is the fact that ILs can either dissolve the whole biomass10,11 or
dissolve the main polysaccharides (cellulose, hemicellulose and lignin) selectively12. In
addition, dissolution and hydrolysis reactions can be performed in relatively mild
conditions13,14 when compared to standard processes for biomass pretreatment and
hydrolysis (alkali pretreatment, acidic hydrolysis, steam explosion, pyrolysis, etc.).
Furthermore, ILs can improve the conversion and the selectivity of several sugar
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
reactions 15,16
.
Phase-equilibrium data as well as their modeling are of key relevance for
designing biorefining processes involving ILs as solvents. Especially solubility data are
needed to design separation and reaction units. Different experimental procedures
have been applied to measure solubility of sugars or sugar alcohols in ILs 17-23. Most of
them use an approach in which the solubility is obtained at constant temperature
(isothermal approach) 17-22,24. To ensure that equilibrium is reached, usually
incremental amounts of the sugar component are added to an IL, until saturation is
reached17,20,21. Another method consists in adding an excess of sugar to the IL 22,25 and
ACS Paragon Plus Environment

The Journal of Physical Chemistry
Page 4 of 56
1
2
3
4
5
6
7
8
9
measure the sugar concentration of the IL solution over the time until saturation (no
variation in sugar concentration). Another possible approach is to measure the
equilibrium temperature at constant composition (polythermal approach ). Sugar
solubility data in ILs are still scarce in literature, and in most of the studies the data are
available only at a single temperature. Moreover, the available literature data cover
essentially only the most well-known sugars: glucose, fructose, and sucrose as well as
polysaccharides such as cellulose. Our previous works in this field 22,25,26 aimed to close
this gap of data, providing accurate solubilities in ILs within a broad range of
temperature. Regarding experimental solubility of sugar alcohols in ILs, Payne and
Kerton27 were the pioneers, reporting solubility of xylitol in two ILs at 373 K. In 2011,
Conceição et al.23 measured the solubilities of mannitol and xylitol in 10 ILs in a wide
range of temperature. Later on, our group published solubility of xylitol and sorbitol in
three ILs in the temperature range 288-343 K25. More recently, Paduszyński et al.24
reported solubility of xylitol and sorbitol in dicyanamide-based ILs. Special care has to
be given to the fact that ILs are highly-hygroscopic compounds. In addition, the
solubility of sugar-like compounds is strongly affected by the presence of water in
ILs21,28. This requires keeping the water content in ILs as low as possible and further
measuring and reporting the amount of water in ILs prior and after their use.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Thermodynamic models used to describe and predict properties and phase
equilibria of mixtures containing sugars and ILs have to account for several specific
interactions. Correlation of those data has already been performed22,25,26 by means of
local composition models such as NRTL29 and UNIQUAC30. SAFT-based equations of
31
state (EoS) were recently applied to model sugar/water systems. Feng et al. used
32
original SAFT to represent water activity in sugar solutions and solubility of glucose,
fructose, and sucrose in water. To obtain the sugar pure-component parameters, the
authors used the critical point and normal boiling temperature of the sugar, which
they calculated using group contribution methods. In a second step, they modeled the
sugar solubility in water by fitting quadratic temperature-dependent kij values. This
shortcoming was probably caused by inappropriate pure-component parameters.
Later, Ji et al.33 also used SAFT and the same parameter-estimation procedure to
predict the density of aqueous sugar solutions. In a following work, the same research
ACS Paragon Plus Environment

Page 5 of 56
The Journal of Physical Chemistry
1
2
3
group published the use of SAFT for modeling water activities, mixture densities, and
4
5
34
oxygen solubilities in sugar alcohol solutions . They applied four association sites for
6
7
all sugars and three association sites for water.
8
9
In our previous work35 we demonstrated that PC-SAFT is suitable to model
sugars in aqueous and non-aqueous solutions. The association scheme for each sugar
was chosen according to its molecular structure (i.e. number of OH groups). Using only
four adjustable parameters, which were fitted to aqueous-solution density and
osmotic-coefficient data, PC-SAFT was able to predict satisfactorily sugar and sugar
alcohol solubility in single and mixed solvents and even in multi-solute sugar systems.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Besides these works on sugar modeling, ILs have also been modeled extensively
36,37
with SAFT-based EoS. Two excellent and complete reviews
containing the main
research on modeling IL systems, including cubic and molecular-based equations of
state as well as molecular dynamics studies, also report the use of SAFT-based
approaches. Kroon et al.38 were the pioneers applying SAFT to systems containing ILs.
They used the tPC-PSAFT EoS to model solubilities of CO2 in ILs. Many other versions of
SAFT were also applied to systems with ILs, e.g. PC-SAFT39-41, Soft-SAFT42, SAFT-
gamma43, SAFT-VR44, and hetero-SAFT45. They differ among each other in the
reference-fluid system, in the combining rules for fluid mixtures, and in the kind of
molecular interactions which are taken into account. It is also noteworthy to mention
that in SAFT-based models, the ILs can be treated as homo-segmented or hetero-
segmented molecules, or as fully dissociated species (ePC-SAFT). The original version
of PC-SAFT has been used for modeling many mixtures with ILs, e.g. the solubility of
CO2 in imidazolium-based ILs39-41. In a recent work41 ILs were treated as homo-
segmented non-dissociated species and only three parameters (segment number,
diameter and dispersion energy) were fitted to experimental pure-IL density. The
association parameters were kept constant (equal to those of 1-alkanols). Two
association sites of type B and one site of type A were considered for each IL and only
AB interactions were allowed. Temperature-independent kij values were fitted to
binary data to correlate CO2 solubilities with good accuracy. In a similar way, Rahmati-
44
Rostami et al.
used PC-SAFT to model the solubility of hydrogen sulfide in
imidazolium-based ILs containing the anions [BF4]-, [PF6]-, and [NTf2]-. Another
ACS Paragon Plus Environment

The Journal of Physical Chemistry
Page 6 of 56
1
2
3
4
5
6
7
8
9
approach for the description of CO2 and CH4 solubility in ILs was proposed by Ji et al. 46
who used the PC-SAFT model assuming either complete dissociation of the IL into ions
(ion based approach , ePC-SAFT) or non-dissociation (molecular based approach).
Therewith, quantitative gas solubility predictions in ILs were possible over a wide range
of pressure and temperature. Recently, Domańska et al. 40 applied PC-SAFT using a 10-
site association scheme for ILs to model binary mixtures containing isoquinolinium-
based ILs and an organic solvents. The pure-IL parameters were fitted to IL density and
Hildebrand solubility parameters. Paduszyński and Domańska 47 modelled the solubility
of aliphatic hydrocarbons in piperidinium ILs. They proposed the use of infinite-dilution
activity coefficients for fitting temperature-dependent binary interaction parameters,
while the pure-component parameters were fitted to pure-IL density data and IL-
solubility data that were determined from Hildebrand parameters at different
temperatures. Nann et al.39 showed that pure-IL densities and activity-coefficient data
(in this case 1-butanol) are appropriate data for the estimation of pure-IL PC-SAFT
parameters. Very recently, the solubility of sugar-like molecules in ILs was modeled for
the very first time using PC-SAFT by Paduszyński et al.24 Therein ILs were modeled with
a symmetric 6-site association scheme. The pure-component parameters for
[bmim][DCA] were fitted to experimental atmospheric liquid-density data and vapor-
pressure data of pure IL. The other two dicyanamide-based ILs in that work were
modeled with the same association scheme and just the segment number, miseg, and
segment diameter, σi, were fitted to density of those ILs, being the remaining
parameters (ui/kB, εAiBi/kB and κAiBi) transferred from [bmim][DCA]. For the sugar
alcohols a 2B scheme for each OH group was assigned and pure-component
parameters were fitted to liquid-density data and vapor-pressure data of pure sugar
alcohols at elevated temperatures. With this strategy, the authors obtained a good
solubility modeling applying kij values for dispersion and association energy.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
In this work, experimental solubility data of xylitol and sorbitol in five ILs are
presented. Four dicyanamide ILs and one trifluoroacetate IL were used in this work: 1-
ethyl-3-methylimidazolium dicyanamide [emim][DCA], 1-butyl-3-methylimidazolium
dicyanamide [bmim][DCA], Aliquat®[DCA], trihexyltetradecylphosphonium dicyanamide
[P6,6,6,14][DCA], and 1-ethyl-3-methylimidazolium trifluoroacetate [emim][TFA]. The
ACS Paragon Plus Environment

Page 7 of 56
The Journal of Physical Chemistry
1
2
3
pure-component densities of these ILs were measured and compared with literature
4
5
48
values. Original PC-SAFT was applied to model solubility data of sugars (glucose and
6
7
8
9
fructose) and sugar alcohols (xylitol and sorbitol) in these ILs. An additional IL,
[emim][EtSO4], was also considered herein for the modeling results. Two different
approaches for the PC-SAFT IL-parameter estimation are presented. In a first approach,
pure-IL density and activity coefficients at infinite dilution of 1-propanol in the ILs were
used. A different parameter-estimation procedure was applied to ILs for which
experimental 1-propanol activity-coefficient data at infinite dilution were not available:
parameters were fitted to pure-IL density and solubility of xylitol in these ILs.
Experimental solubility data and modeling results obtained for systems with
[bmim][DCA] were compared to those published by Paduszyński et al.24.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
2. Experiments
Materials
The ILs 1-ethyl-3-methylimidazolium dicyanamide, [emim][DCA] (> 99 wt%), 1-
butyl-3-methylimidazolium dicyanamide, [bmim][DCA] (> 98 wt%) and
trihexyltetradecylphosphonium dicyanamide, [P6,6,6,14][DCA] (> 95 wt%) were supplied
by Iolitec GmbH (Germany) and used without further purification. Xylitol (99 wt%, Alfa
Aesar) and D (-) sorbitol (99.9 wt%, Merck KGaA, Germany) were the sugar alcohols
used in this work. They were dried under vacuum prior to the measurements. The
chemicals needed for the synthesis of 1-ethyl-3-methylimidazolium trifluoroacetate
were ethyl trifluoroacetate (>99.9 wt%, Merck KGaA, Germany) and 1-
methylimidazole (>99 wt%, Merck KGaA, Germany). Sodium dicyanamide (>97 wt%,
Acros Organics) and Aliquat®336a with 3.86 wt% of water (measured by Karl-Fisher
titration in this work) and with an average molecular weight of 442 g∙mol-1 were
supplied by Acros Organics and used to synthesize Aliquat®[DCA]. Other chemicals
such as toluene (>99.99 wt%, Fisher Scientific), acetone (>99.9 wt%, Labsolve), n-
heptane (>99 wt%, BDH Prolabo), and dichloromethane (99.99 wt% Fisher Scientific)
were needed for the synthesis. Silver nitrate, AgNO3 (Ph. Eur, Merck KGaA, Germany)
was used for the quantification of chloride content in Aliquat®[DCA].
ACS Paragon Plus Environment

The Journal of Physical Chemistry
Page 8 of 56
1
2
3
a
Aliquat®336 is a commercial name for a chloride ionic liquid with a mixture of the cations:
4
5
6
7
trioctylmethylammonium and tridecylmethylammonium, predominating the former (0.55 of mole
fraction).
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
The chemical structures of the ion species of the ILs considered in this work are
illustrated in Figure 1.
Synthesis of [emim][TFA]
A solvent-free reaction was carried out to produce [emim][TFA] according to the
literature49. Roughly 9 g of 1-methylimidazole and 16 g of ethyl trifluoroacetate were
mixed in a closed stainless steel reactor at 373 K. The mixture was stirred for 24 h, and
the unconverted reactants were removed under vacuum. The obtained [emim][TFA]
contained 1200 ppm of water (measured by Karl-Fisher titration) and the chemical
1
structure of [emim][TFA] was checked by H NMR. Detailed information about this
synthesis is available in a previous publication26.
Synthesis of Aliquat®[DCA]
The metathesis reaction between Aliquat®336 and sodium dicyanamide for anion
exchange was performed according to a literature procedure50. A mixture of the two
reactants (with 25% molar excess of sodium dicyanamide) in dichloromethane was
stirred for 48h. Then, the suspension was filtered and the obtained liquid was distilled
to remove the excess of dichloromethane. Vacuum was applied to purify the IL, and a
final water content of 500 ppm was determined by Karl-Fisher titration. The final
chloride content was found to be lower than 1300 ppm (measured by Mohr’s method).
1
The cation structure was verified through H NMR. Further detailed information about
this synthesis is available in a previous publication26.
Solubility measurement of the sugar alcohols in the ILs
An isothermal technique was applied to determine the sugar alcohol solubilities in ILs.
In this procedure, 25 mL-jacketed equilibrium cells were used, in which roughly 10 g of
the IL were placed together with a given amount of sugar alcohol. Water from a
ACS Paragon Plus Environment

Page 9 of 56
The Journal of Physical Chemistry
1
2
3
JULABO F12-ED recirculation bath (Julabo, USA) was pumped through the cell jacket in
4
5
order to keep a constant temperature in the inner cell with a precision of ± 0.1 K. The
6
7
8
9
binary mixture was magnetically stirred until equilibrium was reached. The system was
supervised to ensure that a solid phase was present during equilibration. In order to
evaluate whether the system reached the equilibrium state or not, samples were
drawn at different dissolution times, and the concentration of the sugar alcohol was
measured. At the point where no variation in sugar alcohol concentration was
observed, the saturation equilibrium was assumed to be reached. Hence, this
measured concentration corresponded to the solubility concentration. Before each
sampling, the stirring was stopped and phase separation took place until the whole
solid phase was settled down in the cell. Sampling was made in triplicate and the sugar
alcohol concentrations were quantified by HPLC (LaChrom Elite®, HITAKI, USA). A 250×4
mm HPLC column with Purospher®STAR RP-18e (5μm) as stationary phase was used.
An isocratic method with water as eluent and a flow of 0.7 mL∙min- 1 was chosen. Sugar
alcohol detection was carried out with continuous refractive index measurement at
313 K.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Density of the pure ILs
Densities of the ILs [emim][DCA], [bmim][DCA], [emim][TFA], Aliquat®[DCA],
and [P6,6,6,14][DCA] were measured between 278 - 343 K. The apparatus used was a
vibrating-tube densimeter DMA 4500 M (Anton Paar, Germany) with a precision of ±
0.00001 g∙cm-3 for density and ±0.01K for temperature.
3. Modeling
3.1. PC-SAFT equation of state
The PC-SAFT model was developed based on a perturbation theory using the hard-
chain fluid as the reference fluid system. The model assumes molecules as being chains
of spherical segments. Specific interactions such as hydrogen bonding between
molecules are described by considering association sites on the segments. PC-SAFT
ACS Paragon Plus Environment

The Journal of Physical Chemistry
Page 10 of 56
1
2
3
calculates the reduced residual Helmholtz energy of a system (ares) as a sum of the
4
5
reference-chain contribution (ahc), dispersive interactions (adisp), and association (aassoc
)
6
7
effects:
8
9
ꢀꢁꢂꢃ ꢄ ꢀꢅꢆ ꢇ ꢀꢈꢉꢃꢊ ꢇ ꢀꢋꢃꢃꢌꢆ
(1)
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
The hard-chain term, ahc, is the reference contribution in which only repulsive
interactions are taken into account. The non-spherical shape of the molecules is also
considered in this term, by portraying the molecules as chains of spherical segments -
hard-chain fluid. Then a dispersive term, adisp, resulting from the application of a
second order perturbation theory to the hard-chain reference system is added to
account for attractive interactions between the chains through dispersive forces.
Specific interactions such as association (i.e. hydrogen bonding) are considered in the
associative term, aassoc, in which the magnitude of these interactions is measured by
the association strength between donor and acceptor sites which are assigned to each
molecule. Other contributions such as polar or Coulomb interactions can be also taken
into account but were not considered herein. In this work, the ILs and the sugars were
described as non-dissociated molecules. Self-association was considered for both
sugars and ILs as it is well-known that these molecules interact strongly through
hydrogen bonding. In concentrated solutions, ILs are often considered to be present as
non-dissociated molecules using SAFT-based modeling approaches47,51. As this work
deals with those solutions, IL dissociation and Coulomb forces due to charges were
neglected. In addition, there are experimental and theoretical evidences that ILs and
carbohydrates interact essentially through hydrogen bonding52,53. Thus, a polar term
was not considered. Such term would also increase the complexity of the model,
requiring the knowledge of dipolar moments for ILs or the fitting of additional
parameters. As for ion pairing, this is accounted within the association term of PC-
SAFT. The expressions to calculate the reduced residual Helmholtz energy
contributions of Eq.(1) are those used in the original PC-SAFT48. In general, five pure-
component parameters are required to describe an associating compound: the
number of segments miseg, the diameter of the segments σi, the van der Waals-
interaction (dispersion) energy parameter between segments of two different
molecules ui/kB, and two additional parameters to account for association between
ACS Paragon Plus Environment

Page 11 of 56
The Journal of Physical Chemistry
1
2
3
4
5
6
7
8
9
acceptor sites Ai and donor sites Bi, namely the association-energy parameter εAiBi/kB
and the association-volume parameter κAiBi. Moreover, the number of association sites
in each compound must be set prior to the parameter estimation. To handle fluid
mixtures, the conventional Berthelot-Lorentz combining rules were used to calculate
the mixture parameters:
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
ꢍꢉꢎ ꢄ ꢏꢐ ꢑꢍꢉ ꢇ ꢍ ꢒ
(2)
(3)
ꢎ
ꢓ ꢄ ꢑ1 ꢔ ꢕ ꢒ ꢓ ꢓ
ꢉ ꢎ
ꢖ
ꢉꢎ
ꢉꢎ
The binary interaction parameter kij in Eq.(3) was introduced to correct the dispersive
energy between two components. This parameter is usually obtained by fitting to
binary experimental data such as activity coefficients or solubilities. Keeping kij=0, the
predictive form (no binary data used) of the model is conserved. To account for cross
association between a sugar and an IL, the combining rules proposed by Wolbach and
Sandler54 were used in this work:
ꢏ ꢜ
ꢗꢘ ꢚ ꢄ ꢐ
ꢝ
ꢘ ꢚ
ꢛ ꢛ
ꢘ ꢚ
ꢙ
ꢛ
ꢙ
ꢙ
ꢗ
ꢇ ꢗ
(4)
(5)
ꢤ
ꢠ ꢠ
ꢖ
ꢙ
ꢛ
ꢘ ꢚ
ꢛ ꢛ
ꢞꢘ ꢚ
ꢄ
ꢞ
ꢟꢏꢡ ꢑꢠ ꢢꢠ ꢒꢣ
ꢘ ꢚ
ꢙ ꢙ
ꢙ
ꢛ
ꢖ
ꢞ
ꢙ ꢛ
ꢐ
3.2 Calculation of thermodynamic properties with PC-SAFT
Various physical and thermodynamic properties can be obtained from the residual
Helmholtz energy. First, the compressibility factor, Z is calculated from PC-SAFT as a
derivative of ares with respect to density, ρ :
ꢥ ꢦ ꢄ 1 ꢇ ꢦ ꢧꢨꢋꢩꢪꢫꢭ
(6)
ꢜ ꢝ
ꢨꢬ
ꢮ,ꢯ
where x stands for the composition of the mixture and T for the temperature.
For density calculations, Eq.(7) has to be solved:
ꢜ ꢝ
ꢰ ꢔ ꢦꢱꢲꢥ ꢦ ꢄ 0
(7)
ACS Paragon Plus Environment

The Journal of Physical Chemistry
Page 12 of 56
1
2
3
where R is the ideal gas constant (8.314 J∙mol∙K-1) and p the system pressure. Herein,
4
5
the density roots are obtained iteratively at given temperature and pressure.
6
7
8
The fugacity coefficient ϕi of component i can be obtained as follows:
9
ꢩꢪꢫ
ꢩꢪꢫ
ꢨꢋ
ꢨꢋ
ln ꢳꢉ ꢄ ꢥ ꢔ 1 ꢔ lnꢥ ꢇ ꢀꢁꢂꢃ ꢇ
(8)
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
ꢜ
ꢝ
∑
ꢔ ꢉ ꢴꢉ
ꢨꢯ
ꢙ
ꢨꢯ
ꢙ
Activity coefficients ꢵꢏꢶof component 1 at infinite dilution in component 2 can be
calculated from the fugacity coefficients:
ꢹ
ꢷ
ꢜꢮ,ꢊꢝ
ꢜꢮ,ꢊꢝ
ꢸ
ꢵꢏꢶ ꢄ ꢷ
(9)
ꢺꢸ
where 0 and ∞ stands for pure-compound state and infinite dilution, respectively.
Considering a binary mixture of a volatile (1) and a non-volatile (2) component, the
osmotic coefficient ꢻ of this mixture can be obtained from the knowledge of fugacity
coefficients of component 1:
ꢿ
ꢸ
ꢻ ꢄ ꢔ ꢏꢼꢼꢼ ꢽꢾꢧꢯ
(10)
ꢭ
ꢸ
ꢿ
ꢺꢸ
ꣀ ꣂ
ꣁ
ꢸ
with m2 and M1 being the molality of a non-volatile component 2 in (moles of 2 per kg
of pure 1) and the molecular weight of the volatile component 1 in (g/mol),
respectively
In order to calculate the solubility xi (in mole fraction) of a component i in a liquid, the
following solid-liquid equilibrium condition 55 was applied in this work:
꣆꣇꣄
ꢺꢙ
꣆꣇꣄
꣋,ꢙ
꣆꣇꣄
ꢺꢙ
꣆꣇꣄
ꢺꢙ
ln ꢧꢴꢉ ꣃ ꢷ ꣄ꢭ ꢄ ∆ꣅ
ꢟ1 ꢔ ꢮ ꢣ ꢔ ꣉꣊ ꣌ꢟ1 ꢔ ꢮ ꢣ ꢇ ln ꢟꢮ꣆꣇꣄ꢣ꣍
(11)
ꢷ
ꢙ
ꢺꢙ
꣆꣇꣄
꣈ꢮ
ꢮ
꣈
ꢮ
ꢮ
ꢺꢙ
ꢺꢙ
S-L
In Eq.(11), the subscript i stands for a sugar component in this work and ΔH0iS-L, T0i
,
S-L
and Δcp,i are the melting enthalpy, melting temperature, and heat-capacity
difference of the pure sugar between its solid and its (hypothetical) liquid state,
L
respectively. ϕ0i and ϕi stand for the sugar fugacity coefficient as pure subcooled liquid
(hypothetical state) and in the liquid mixture, respectively. Solubility data and
modeling results are presented in this work in molality, i.e. moles sugar per kg pure IL.
Mole fractions of sugars x2 were converted to molality m2 using the molar mass of IL
M1 according to
ACS Paragon Plus Environment

Page 13 of 56
The Journal of Physical Chemistry
1
2
3
4
꣎ꢐ ꢄ ꢏꢼꢼꢼꢯ
(12)
ꣁ
ꣂ ꢯ
ꢸ
ꢸ
5
6
7
3.3. Parameter estimation
8
9
Modeling a sugar+IL mixture requires PC-SAFT parameters for the pure IL, the
pure sugar, and binary interaction parameters kij. Due to their non-volatile character,
ionic liquids and sugars can hardly be parameterized using pure liquid densities and
pure vapor pressures. Moreover, sugars do not exist as pure liquids. Therefore,
alternative parameter estimation strategies are needed to model systems containing
these compounds.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
i) Sugar parameters
The sugar parameters were taken from a previous work35. They were obtained
by adjusting four parameters: miseg, σi, ui/kB, and a binary kij between sugar and water
to experimental aqueous-solution densities and osmotic coefficients at 298 K. The
association-energy and association volume parameters were set to constant values.
The association scheme was set according to the number of hydroxyl groups present in
the solutes. These parameters allowed for satisfactory solubility predictions in water as
well as in solvent mixtures (e.g. water/alcohol). Therefore, they were also applied in
this work although the solvent is an IL instead of water. Table 1 presents these pure-
component parameters for the sugars and sugar alcohols considered herein as well as
their melting properties which were taken from literature.
ii) IL parameters
To obtain the PC-SAFT parameters for ILs pure-IL density data are appropriate.
However, fitting to volumetric data alone is often not sufficient for estimating energy
parameters such as ui/kB. Therefore, additional experimental data are necessary to
provide a set of parameters that represents well both the volumetric and the energetic
properties of the ILs. In general, vapor-pressure data of pure ILs would be appropriate
for the pure-IL parameter estimation. However, such data are difficult to measure with
high precision and are available only for very limited set of ILs. Due to this fact, the use
of binary experimental data is a valid option, e.g. osmotic coefficients of water/IL or
alcohol/IL solutions63-65. However, the limited maximum IL concentration (e.g. < 5
ACS Paragon Plus Environment

The Journal of Physical Chemistry
Page 14 of 56
1
2
3
4
5
6
7
8
9
mol∙kg-1 in vapor-pressure osmometry experiments) may not be appropriate to model
systems with very high IL contents. Another possibility is the use of activity coefficients
of alcohol at infinite dilution in ILs 39,47. These data are more suitable to describe
systems at high (nearly pure) IL concentrations.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
In this work, two strategies for the estimation of pure-IL parameters were applied. In
the first strategy, pure-IL densities and infinite-dilution activity coefficients of 1-
propanol in ILs were used. A second strategy was applied to systems for which these
data were not available; in this second strategy IL parameters were fitted to pure-IL
densities and xylitol solubilities in ILs. This approach is expected to provide good
correlating results for sugar solubility, but may have limitations when using the IL
parameters for prediction of other (solubility-independent) thermodynamic data.
Independent of the strategy, the following objective function, OF, was minimized in
the parameter estimation:
ꣁ
꣓꣔꣇꣆꣕꣖꣗
꣒
꣚
ꢙ꣞ꢸ
꣛꣜꣝꣜
ꢙ
∑
꣐ꢏ꣑
꣙
ꢪ꣘꣋.
ꢙ
꣒
꣏
꣠
(13)
OF ꢄ logꢏꢼ
꣟
꣛꣜꣝꣜
where Ndata is the total number of experimental points used and Y refers to a generic
property, being pure-IL density, alcohol activity coefficient at infinite dilution, or sugar
solubility in IL in this work.
IL parameter estimation: strategy 1
The ILs [emim][EtSO4], [emim][DCA], and [emim][TFA] were parameterized using pure-
IL density and activity-coefficient data at infinite dilution ꢵꢏꢶ꣑ꢊ of 1-propanol
(component 1) in these ILs (component 2). Reasons for choosing these data was their
availability in the literature and the fact that sugars and alcohols are similar with
respect to the interactions they can exhibit: both are able to self-associate and cross-
associate and further have apolar parts. In this work, preliminary studies were carried
out using experimental pure-IL densities together with experimental alcohol activity
coefficients at infinite dilution for the IL-parameter estimation. Comparing the use of
ethanol, 1-propanol, or 1-butanol activity coefficients at infinite dilution these studies
showed that using the so-determined pure-IL parameters yield a similar performance
for sugar solubility predictions, independent of the alcohol chosen for the ꢵꢏꢶ data.
ACS Paragon Plus Environment

Page 15 of 56
The Journal of Physical Chemistry
1
2
3
4
5
6
Due to the fact that experimental 1-propanol activity coefficients at infinite dilution
were available for the largest number of ILs investigated in this work, 1-propanol was
the alcohol chosen for ꢵꢏꢶ data to be used in the parameter estimation.
7
The temperature range for the chosen values of ꢵꢏꢶ꣑ꢊ was set according to the
temperature range (288 - 339 K) of solubility data. Besides ꢵꢏꢶ꣑ꢊ data also experimental
pure-IL densities stemming from this work and from literature were used for the
parameter estimation. According to our previous work46, the association parameters
for ILs were set to εAiBi/kB = 5000 K and κAiBI = 0.1 as these values had been found to be
appropriate to model densities of imidazolium-based ILs. The kij values between ILs
and 1-propanol were set to zero, according to a previous research conducted by us39.
Thus, only three adjustable IL parameters were determined using this strategy, namely
miseg, σi , and ui/kB.
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Different association schemes for each IL were tested in order to determine the
scheme which yields the highest overall accuracy with respect to IL-density, activity
coefficient ꢵꢏꢶ꣑ꢊ, and sugar solubility (in terms of ARD, see Eq.(14)). Symmetric
association schemes (number of donor sites equal to the number of acceptor sites)
with the total number of association sites ranging between 2 (1+1) and 10 (5+5) were
applied in these preliminary analyses. Table 2 summarizes the obtained parameters for
the ILs with the best association scheme, and the average relative deviations (ARDs)
for density and ꢵꢏꢶ꣑ꢊ. These ARDs were calculated by:
ꢪ꣘꣋.
ꢙ
꣓꣔꣇꣆꣕꣖꣗
꣣꣤
꣑꣤
꣣
ꢏꢼꢼ
꣟
ꢉ꣥ꢏ
ꢙ
꣛꣜꣝꣜
∑
꣡ꢱ꣢ꢜ%ꢝ ꢄ ꣟
(14)
ꢪ꣘꣋.
꣤
꣛꣜꣝꣜
ꢙ
where Y refers to pure-IL density, ꢵꢏꢶ꣑ꢊ , or sugar solubility in the IL .
IL parameter estimation: strategy 2
For the other three ILs investigated in this work ([bmim][DCA], Aliquat[DCA], and
[P6,6,6,14][DCA]), data for alcohol activity coefficients at infinite dilution are not available
in the literature. Therefore, pure-IL density and solubility data of a reference sugar in
the ILs were used for the parameter estimation. The reference sugar chosen was
xylitol, which presents roughly an average value for the experimental solubility among
the four solutes considered (see Tables 6 and 7 or Figure 8 in section 4 of this
manuscript). Preliminary tests using this fitting strategy showed that the association
ACS Paragon Plus Environment

The Journal of Physical Chemistry
Page 16 of 56
1
2
3
4
5
6
7
8
9
scheme for ILs should be the same as for the reference solute. Thus, the association
scheme 5+5 was chosen for the ILs parameterized with this approach. As done for
strategy 1, the association parameters of the ILs were set to the same constant values
(εAiBi/kB = 5000 K and κAiBi = 0.1). In strategy 2, a kij between the ILs and xylitol was
considered as an additional adjustable parameter; however assumed to be
temperature independent values. As a result, four adjustable parameters were fitted
to pure-IL density and xylitol solubilities: miseg, σi, ui/kB, and a kij between xylitol and IL.
Table 3 lists the pure-IL and kij parameters obtained with this strategy 2.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Afterwards, and independently on the fitting strategy used, a kij for each binary
system (sugar+IL) was applied to accurately describe the experimental solubility in ILs.
They are listed in Table 4 for each parameter-estimation strategy. It can be observed
that the absolute kij values are very low for strategy 1 and distinctly higher (one order
of magnitude) for strategy 2. This already shows that it is highly recommended to use
non-solubility related data for the IL-parameter estimation in order to obtain kij values
that are as low (absolute) as possible.
4. Results and discussion
4.1. Experimental pure-IL densities
The densities of the pure ILs measured in this work are presented in Table 5 as a
function of temperature. Figures 2 and 3 compare the data measured in this work with
other data from open literature.
It can be observed from Figures 2 and 3 that our measured density data are in
excellent agreement with literature data. An exception was found for Aliquat®[DCA]
for which lower densities were obtained compared with literature data. This is
probably due to the different amount of water in the IL. The Aliquat®[DCA] used in this
work contained 0.05 wt% of water, which is lower than that present in the literature
work79 (0.2 wt%) and might explain the density discrepancy shown in Figure 3.
The large alkyl chains present in the hydrophobic ILs (Aliquat®[DCA] and [P6,6,6,14][DCA])
cause high system volumes and low densities, even lower than 1000 kg/m3 in the
whole temperature range (see Figure 3). Densities are slightly different between these
ACS Paragon Plus Environment

Page 17 of 56
The Journal of Physical Chemistry
1
2
3
4
5
6
7
8
9
two hydrophobic ILs although the alkyl-chain volume is similar. [P6,6,6,14][DCA]
possesses a higher number of alkyl groups (larger alkyl chains), which at a first glance
would let one assigning to it a lower density than for Aliquat®[DCA], due to a larger
volume occupancy of [P6,6,6,14][DCA]. However, the molar mass of the ions also
influences the IL density 82: heavy ions cause high mass-density ILs. Thus, as the
phosphorous central atom is heavier than nitrogen, [P6,6,6,14][DCA] is “heavier” than
Aliquat®[DCA]. Among the considered hydrophilic imidazolium-based ILs, [bmim][DCA]
has a lower mass density than [emim][DCA] due to the higher number of alkyl groups
in the cation which causes higher volumes related to a higher flexibility/mobility of the
[bmim]+. [emim][TFA] presents higher mass density than [emim][DCA] caused by the
much higher molar mass of its anion (roughly the double). Besides the anion molar
mass effect, the anion-cation interaction (both coulombic and hydrogen bonding) plays
also a role in yielding more or less densier IL-structures. Thus, the acetate-based anion
[TFA]- due to its higher lewis basicity (ability to accept hydrogen bonds) compared with
[DCA]- can interact more strongly with the [emim]+ cation resulting in an IL with higher
density.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
4.2. Experimental sugar alcohol solubility in ILs
Tables 6 and 7 show the experimental data of sugar alcohol solubility in the five ILs
considered in this work. The solubility is presented in molality units, i.e. moles of sugar
alcohol per kg of pure IL. The standard deviation is related to the triplicate sampling,
and the relative uncertainty is calculated according to the theory of uncertainty
propagation.
The presence of water in these systems might be an important factor that
influences solubility values. Thus, in Table 8 the weight-percent values of water in the
ILs before and after all the solubility measurements are presented. It was found out
within this work that there is no significant impact on solubility data for water wt%
differences lower than 0.2 wt%. As it can be seen from Table 8, water wt% differences
for the ILs was always below this value and therefore reliable solubility data was
obtained. For some ILs, the natural water uptake during sampling procedures was
ACS Paragon Plus Environment

The Journal of Physical Chemistry
Page 18 of 56
1
2
3
4
5
higher than for others (e.g. for Aliquat®[DCA] in Table 8), but the critical difference of
0.2 wt% was never exceeded.
6
7
8
9
The analysis of the measured solubility data (see Tables 6 and 7) provides
several conclusions. First, the ability of these ILs to dissolve xylitol and sorbitol
depends on both the cation and the anion nature. Due to their electron-donor
properties18,20 both anions considered in this work ([TFA]- and [DCA]-) show high
affinity for sugars and sugar alcohols (see also Figure 1). The high free-electron density
allows for strong hydrogen bonding between the IL anions and the hydroxyl groups of
the sugar alcohols. This leads to high xylitol and sorbitol solubilities in ILs containing
these anions. Comparing both anions, [TFA]- might have a higher hydrophilicity as the
sugar solubility is generally higher in [emim][TFA] than in [emim][DCA]. This is
illustrated in Figure 4. In addition, the free electrons in [TFA]- are more accessible due
to steric effects. Thus, a higher hydrophilicity, i.e. stronger interactions with sugar OH
groups and steric advantages, yields higher solubilities in [TFA]-based ILs compared to
[DCA]-based ILs.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Besides the anion influence, the hydrophilicity of the ILs considered in this work
depends strongly on the cation character. The hydrophilic ILs under investigation
contain imidazolium-based cations ([emim][DCA], [bmim][DCA], and [emim][TFA]),
whereas the hydrophobic ILs used in this work are Aliquat®[DCA] and [P6,6,6,14][DCA].
An evident conclusion is that the hydrophilic ILs dissolve much more xylitol and
sorbitol than the hydrophobic ones (see Tables 6 and 7). This might be due to the fact
that the large alkyl chains of the [Aliquat]+ and the [P6,6,6,14]+cations interact preferably
with each other. This causes strong IL-IL interactions and only weak solute-IL
interactions.
The effect of the alkyl chain length of the ILs on sugar alcohol solubilities can be
analyzed in Figure 5. Investigating the solubility of xylitol and sorbitol in [emim]+ and
[bmim]+-based ILs with the same anion (dicyanamide), it was observed that the
presence of the two additional CH2 groups in [bmim][DCA] reduces the solubility of
xylitol and sorbitol by more than 30 % (related to molality). Among the two
hydrophobic ILs (Aliquat®[DCA] and [P6,6,6,14][DCA]), the higher solubility values were
found in the Aliquat IL. As the same effect was observed in a previous work for the
ACS Paragon Plus Environment

Page 19 of 56
The Journal of Physical Chemistry
1
2
3
4
5
6
7
8
9
solubility of glucose and of fructose26, it seems that ILs with ammonium-based cations
are generally more suitable as solvents for sugar-like molecules than ILs with
phosphonium-based cations. However, it is important to note that the [P6,6,6,14]+ cation
has more alkyl (apolar) groups than the Aliquat+ cation, which also affects solubility (as
shown for the [emim]+ and [bmim]+ cations in Figure 5).
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
As xylitol and sorbitol have very similar chemical structures, their solubility
behavior in the selected ILs could be expected to be identical. On the one hand,
sorbitol has an additional OH group compared to xylitol, which causes more hydrogen
bonds with the ILs. On the other hand, sorbitol has also one CH group more (less
polarity) than xylitol which causes the opposite effect. These contrary effects could
compensate each other and as consequence, one could expect the xylitol/IL and the
sorbitol/IL interactions to be very similar. Nevertheless, experimental data indicate
that xylitol is slightly more soluble than sorbitol. This can probably be attributed to the
different melting properties (Table 1), which are more favorable for xylitol solubility.
However, sorbitol is more soluble than xylitol in [emim][DCA] and [bmim][DCA] until
ca. 320 K (Figure 5). This might be due to higher hydrogen-bond strength at lower
temperatures, causing the sugar with more OH groups (sorbitol) to be slightly more
soluble. In addition, the melting-properties effect is not that significant at these lower
temperatures, and a small increase in hydrogen bonding strength can reverse the
solubility ranking.
Comparison between solubility data obtained in this work with those published
by Paduszyński et al.24 was performed. It should be noted that the experimental
techniques for the determination of the solubilities applied in both works were clearly
distinct. Nevertheless, Figure 5 shows that there is a good agreement between the
data measured in this work and those reported by Paduszyński et al.24
4.3. Pure-IL PC-SAFT parameters
All the PC-SAFT parameter sets for the ILs listed in Tables 2 and 3 allow for accurate
pure-IL density modeling (ARD < 1 %) in the temperature range from 278 to 343 K
(Figures 2 and 3). Considering Tables 2 and 3, it can be observed that pure-IL density
calculations are in better agreement with experimental data using strategy 1
(parameters fitted to IL density and 1-propanol activity coefficients at infinite dilution)
ACS Paragon Plus Environment

The Journal of Physical Chemistry
Page 20 of 56
1
2
3
4
5
6
7
8
9
instead of strategy 2 (parameters fitted to IL density and xylitol solubility). This is
probably due to the fact that the parameters in strategy 1 were estimated using data
only at xIL ≈ 1 (pure-IL densities and 1-propanol activity coefficients at infinite dilution)
whereas data at much lower IL concentrations (xIL<<1) were used in strategy 2.
Nevertheless, the density ARD value in density calculated for [bmim][DCA] (0.16 %) is
close to that obtained by Paduszyński et al.24 (ARD=0.10 %), who fitted the parameters
to density data and vapor-pressure data of the pure IL.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
The physical meaning of the parameter sets was assessed by calculating the
vapor pressure of pure ILs. The scarcely reported experimental vapor pressure83 of ILs
indicate that vapor pressure of ILs should be lower than 3 Pa at temperatures until 500
K (note that such high temperature is close to the decomposition temperature of some
ILs). Applying the parameters determined in this work (Tables 2 and 3) all the IL
parameter sets presented in this work yield vapor pressures below 0.1 Pa up to 600 K,
which is sounding with the "non-volatility" of the ILs. It is worth to note that this was
achieved without applying any direct constraint. The selection of high values for
association parameters and the number of association sites played a crucial role
herein, and acted as an indirect constraint to vapor pressure.
In order to further test the suitability of the pure-IL parameters to describe the
physical properties of ILs, the parameters were used to predict osmotic-coefficient
data in IL+water and IL+(1-butanol) mixtures. Figure 6 compares experimental and PC-
SAFT modeled osmotic coefficients in these solutions. It can be seen that the
parameters provide a satisfactory qualitative prediction of the osmotic-coefficient
data. This result is independent of the strategy applied for the parameter-estimation
procedure: Figure 6a presents a system with pure-IL parameters estimated with
strategy 1, whereas Figure 6b presents a system with pure-IL parameters estimated by
strategy 2. A good quantitative modeling was obtained by fitting a kij to experimental
data with better accuracy for the system where IL parameters were fitted using
strategy 1. Figure 6 shows that the pure-IL PC-SAFT parameters determined in this
work allow calculating thermodynamic properties that were not used in the
parameter-estimation procedure.
ACS Paragon Plus Environment

Page 21 of 56
The Journal of Physical Chemistry
1
2
3
4
5
6
7
8
9
As the PC-SAFT parameters for the ILs are applicable to different systems, different
conditions, and various thermodynamic properties, they are proven to be physically
meaningful and allow for a qualitative analysis. Since the association parameters were
kept constant for all ILs used, the IL dispersion-energy parameter might be related with
the solubility of the sugars in the ILs. In Tables 2 and 3 it can be observed that the
higher the uIL/kB values, the higher the xylitol and sorbitol solubilities in the respective
IL (data in Tables 6 and 7). This dependence holds for both, strategy 1 and 2.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
3
-1
A linear relation between (mILseg∙σIL ∙MIL ) and the specific volume at 298 K was
identified for the IL parameters determined by both strategies, which is illustrated by
the circles in Figure 7a. A more complete IL database would allow for a better linear
regression, which could be useful to decrease the number of adjustable parameters for
other ILs with less data available. However, ILs with parameters obtained with strategy
2 (triangles in Figure 7a) produced some dispersion in the linear trend, which might be
due to the fact that this strategy was applied without using data which accounts for IL-
3
interactions at xIL≈1. Nevertheless, strategy 2 IL parameters (mILseg∙σIL and uIL/kB)
depend linearly on the IL molecular weight (see triangles in Figure 7b and 7c), which
indicates that these parameters are also physically reasonable as similar relations were
observed in many PC-SAFT modeling studies. These trends were not observed for
strategy 1 (circles in Figure 7b and 7c) due to the very narrow range of molecular
weights of the ILs used in that approach.
4.4. Modeling sugar and sugar alcohol solubility in ILs
The IL parameters obtained with strategy 1 (using pure-IL density and ꢵꢏꢶ꣑ꢊ data) were
applied to directly predict the solubility of glucose, fructose, xylitol, and sorbitol in
these ILs. That is, kij values between sugar and IL were set to zero. Figure 8 illustrates
that solubilities are predicted reasonably well, with ARDs for glucose varying between
3.4 - 14.4, for xylitol between 15.5 - 34.9 %, and for sorbitol between 8.3 - 13.8 %. The
ranking of solubility of the four sugars in all the ILs above 300 K is predicted correctly.
Noteworthy, these predictions were performed using only three adjustable pure-IL
parameters. The worst results were obtained for fructose, for which PC-SAFT
ACS Paragon Plus Environment

The Journal of Physical Chemistry
Page 22 of 56
1
2
3
4
5
6
7
8
9
overpredicts the solubility in the three ILs considered leading to ARDs of 31.3, 33.7,
and 112.2% for [emim][EtSO4]22 [emim][DCA], and [emim][TFA], respectively. For
xylitol in [emim][TFA], the solubility at 300 K is modeled in accordance to experimental
data whereas the temperature dependence of the experimental solubility could not be
S-L
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
correctly predicted. This might be due to the fact that temperature-independent ꢀcp
values were used for the modeling with Eq.(11), which might require applying
temperature-dependent kij values.
Nevertheless, these predictions are quite satisfactory considering the fact that usually
non-zero kij values are expected for the modeling of complex binary mixtures with PC-
SAFT.
For the quantitative modeling of sugar solubility in ILs, kij values between a
sugar or sugar alcohol and an IL were fitted to solubility data. The remarkably low kij
values in these systems (see Table 4) also confirm that the solubility predictions with
PC-SAFT are already in a reasonable agreement with the experimental data. Despite
the fact that fructose solubility predictions were not as good as for the other solutes,
quite satisfactory modeling results were obtained when kij values were implemented
(Figure 9). Thus, Introducing kij parameters improves the solubility descriptions
(maximum ARDs of 3.8, 5.3, and 10.9% for glucose, fructose, and sorbitol solubility,
respectively) (see Table 9). For xylitol the maximum ARDs was 21.7%, obtained for
solubility in [emim][TFA], due to the reasons explained above. However, for
[emim][EtSO4] and [emim][DCA] the ARDs were 7.0% and 10.8%, respectively.
The xylitol solubility fittings with IL-parameters obtained using strategy 2 are
presented in Figure 10 (solid lines) and compared to the experimental data (squares in
Figure 10). The experimental solubility data of xylitol shown in Figure 10 strongly
depends on the kind of IL. Whereas the solubility of xylitol is relatively high in
[bmim][DCA], it is extremely low in Aliquat®[DCA] and [P6,6,6,14][DCA] even at elevated
temperatures. With the Fitting of small kij values between xylitol and each IL (see Table
3), PC-SAFT is able to describe these solubility differences with good accuracy. The
parameters yield very satisfactorily ARD values of 1.6, 6.9, and 11.2 % for [bmim][DCA],
Aliquat®[DCA], and [P6,6,6,14][DCA], respectively.
ACS Paragon Plus Environment

Page 23 of 56
The Journal of Physical Chemistry
1
2
3
4
5
6
7
8
9
In Figure 10, the modeling results of glucose, fructose, and sorbitol solubility in
ILs (using IL parameters obtained with strategy 2) are presented. Overall, the results
are quite satisfactory, yielding ARDs (see also Table 10) in the range of 4.9 - 10.6% for
glucose, 5.3 - 25.8% for fructose, and 5.4 - 17.3% for sorbitol. However, the steep
increase of the fructose solubility in [P6,6,6,14][DCA] in the temperature range from 318
to 328 K could not be quantitatively reproduced with PC-SAFT. This increase differs
from the temperature dependence of the solubility behavior in all other systems. As
the IL parameters were fitted to xylitol solubility where this unusual steep increase
does not occur it might be assumed that the parameters are not appropriate to
describe the extreme increase in fructose solubility at T > 318 K accurately.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Among the ILs parameterized with strategy 2, [bmim][DCA] has already been
modeled by Paduszyński et al.24 who used PC-SAFT to model the solubility of sugar
alcohols over a wide temperature range (up to the melting temperature). However, in
this work the solubility data were measured and modeled in a narrower temperature
range. It is worth to assess how this model results can be extrapolated to higher
temperatures. Thus, in Figure 11 the typical patterns of SLE diagrams calculated in this
work are presented.
Figure 11 illustrates the application of PC-SAFT to the modeling of xylitol in
[bmim][DCA] over a broad temperature range. The kij value between xylitol and
[bmim][DCA] was adjusted in this work to solubility data in the temperature range
between 290 and 330 K. Nevertheless, this kij can be applied also to extrapolate the
solubility of xylitol up to the melting temperature of xylitol satisfactorily. Using PC-
SAFT between 330 and 360 K yields only slightly higher xylitol solubilities than the
experimental data of Paduszyński et al.24. In that work, the authors observed
instabilities in the calculated SLE curve at high temperatures using the predictive form
of PC-SAFT(kij=0). In this work the predictions (kij=0) do not present any instability
region when calculating solubilities close to melting temperature for systems with
[bmim][DCA]. This behavior was similar in the systems with the ILs [emim][EtSO4],
[emim][DCA] and [emim][TFA], independent of the considered sugar or sugar alcohol.
For all these ILs, PC-SAFT either predict well the solubilities or overpredict them (in the
case of [bmim][DCA]). However, for the two ILs [P6,6,6,14][DCA] and Aliquat®[DCA]
instabilities were observed, similar to those reported by Paduszyński et al.24.
ACS Paragon Plus Environment

The Journal of Physical Chemistry
Page 24 of 56
1
2
3
4
5
6
7
8
9
Furthermore, liquid split phase was calculated in these two systems: [P6,6,6,14][DCA] +
xylitol and Aliquat®[DCA] + xylitol. Comparing the pure-component parameters of the
ILs (see Table 3) and also those from Paduszyński et al. such instabilities might be due
to too low values of ui/kB.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
5. Conclusions
In this work, pure-IL density and solubility of sugars and sugar alcohols in ILs were
investigated experimentally as well as modeled with PC-SAFT. Two sugars (glucose and
fructose), two sugar alcohols (xylitol and sorbitol) and six ILs ([emim][DCA],
[bmim][DCA], Aliquat[DCA], [P6,6,6,14][DCA], [emim][ETSO4], and [emim][TFA]) were
considered.
Measured pure-IL densities are in very good agreement with available literature
data. Considering ILs with the same cation, the pure-IL densities were found to be
generally higher for bigger anions. For ILs with the same anion, such a general trend
could not be observed: On the one hand, pure densities of hydrophilic ILs
([emim][DCA] and [bmim][DCA]) were high for an IL with a short cation alkyl chain
length. On the other hand, in hydrophobic ILs (Aliquat®[DCA] and [P6,6,6,14][DCA]) high
densities were found for the IL with a long cation alkyl chain length.
The measured solubility data of xylitol and sorbitol is in good agreement with
literature data for [bmim][DCA]. The solubility data of these sugar alcohols in the other
five ILs were presented for the first time and followed the trends found in our previous
works for solubility of glucose and fructose in the same ILs. ILs with the [TFA]- anion
have higher ability to dissolve the sugar alcohols than [DCA]--based ILs. The effect of
the IL alkyl chain length on sugar alcohol solubility in these ILs revealed that increasing
the alkyl chain length causes a strong decrease of solubilities. Among the hydrophobic
ILs, the ammonium-based Aliquat®[DCA] dissolves higher amounts of xylitol and
sorbitol than the phosphonium-based IL [P6,6,6,14][DCA]. In general, xylitol was found to
be more soluble than sorbitol, which might be explained by the more favorable
ACS Paragon Plus Environment

Page 25 of 56
The Journal of Physical Chemistry
1
2
3
4
5
melting properties of xylitol. However, an exception was found in hydrophilic [DCA]-
based ILs at low temperatures, in which sorbitol is more soluble than xylitol.
6
7
8
9
PC-SAFT pure-IL parameters were fitted for six ILs using two different
strategies. In both strategies the pure-IL density was used to fit the parameters. In
strategy 1, additionally 1-propanol infinite dilution activity coefficients, ꢵꢏꢶ꣑ꢊ (taken
from literature) were used. In strategy 2 instead, xylitol solubility data in ILs were used
for the parameter estimation together with the pure-density data. Comparing both
strategies, slightly better pure IL-density descriptions were obtained with strategy 1.
Further, sugar solubility in ILs could be predicted with a reasonable accuracy (ARDs <
30% in most cases) with strategy 1, i.e. without any knowledge of solubility data and kij
= 0 between sugar and IL. The modeling results for systems with [bmim][DCA] were
compared with those presented by Paduszyński et al.24 (who first modeled SLE of sugar
alcohols in ILs using PC-SAFT) showed overall good agreement and satisfactory results.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
In sum, this work shows that PC-SAFT allows for quantitative modeling of the
solubility of sugars in ILs, with only three pure-IL parameters. Further, this work has
shown that accounting for association is crucial to model ILs. However, the association
parameters do not need to be treated as adjustable parameters. Considering the
obtained results, this work recommends fitting three pure-IL parameters to IL densities
and infinite-dilution activity coefficients whenever available (strategy 1). In case of
non-available experimental ꢵꢏꢶ data, the parameters can also be adjusted to solubility
data of solutes in ILs (strategy 2 in this work). This requires three pure-IL parameters
and a kij and allows only for the correlation of solubility data.
6. Acknowledgements
This work is partially supported by project PEst-C/EQB/LA0020/2011, financed by
FEDER through COMPETE - Programa Operacional Factores de Competitividade and
by FCT - Fundação para a Ciência e a Tecnologia. A.C. acknowledges financial
support (Grant SFRH/BD/62105/2009) from Fundação para a Ciência e a Tecnologia
(FCT, Portugal).
ACS Paragon Plus Environment

The Journal of Physical Chemistry
Page 26 of 56
1
2
3
4
7. Nomenclature
5
6
7
Roman symbols
8
9
a
kB
kij
miseg
m2
J∙mol-1
J∙K-1
molar Helmholtz free energy
Boltzmann constant, 1.38065∙10-23 J∙K-1
binary interaction parameter
segment number
molality (moles of component 2 per kg of pure
component 1)
molar mass
number of association sites
pressure
ideal gas constant, 8.314 J∙mol-1∙K-1
solubility of a sugar in IL (moles of sugar per kg of pure IL)
temperature
melting temperature of pure component i
dispersion-energy parameter
uncertainty
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
mol ∙kg-1
g∙mol-1
M
Nsites
p
R
m
T
T0i
ui/kB
u
wt%
xi
bar
J∙mol-1∙K-1
mol∙kg-1
K
K
K
SꢀL
weight fraction
mole fraction
compressibility factor
Z
Greek symbols
∞
γ1
activity coefficient at infinite dilution of compound 1 in
component 2
SꢀL
ꢁcp,i
ꢁH0i
J∙mol-1∙K-1
Heat capacity difference between solid and liquid sugar
melting enthalpy of pure component i
association-energy parameter
association-volume parameter
density
segment diameter
fugacity coefficient
osmotic coefficient
J∙mol-1
SꢀL
εAiBi/kB
K
κAiB
ρ
kg∙m-3
Å
σi
φi
φ
Subscripts
i, j
S-L
0
1
2
component indexes
solid-liquid
pure component
volatile component
non-volatile component
1-propanol
1-p
∞
infinite dilution
ACS Paragon Plus Environment

Page 27 of 56
The Journal of Physical Chemistry
1
2
3
4
5
Superscripts
6
7
8
9
assoc
disp
hc
association
dispersion
hard chain
residual
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
res
Abbreviations
ARD
EoS
ILs
average relative deviation
equation of state
ionic liquids
OF
objective function
PC-SAFT Perturbed-chain statistical-associating fluid theory
[BF4]-
tetrafluoroborate anion
[emim]+ 1-ethyl-3-methylimidazolium cation
[bmim]+ 1-butyl-3-methylimidazolium cation
[DCA]-
[NTf2]-
[P6,6,6,14
[PF6]-
dicyanamide anion
bis(trifluoromethylsulfonyl)imide anion
trihexyltetradecylphosphonium cation
hexafluorophosphate anion
trifluoroacetate
+
]
[TFA]-
ACS Paragon Plus Environment

The Journal of Physical Chemistry
Page 28 of 56
1
2
3
4
8. References
5
6
7
8
9
(1)
Werpy, T. A.; Petersen, G. Top Value Added Chemicals From Biomass: I. Results
of Screening for Potential Candidates from Sugars and Synthesis Gas, Pacific Northwest
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
National Laboratory, 2004.
(2)
Plechkova, N. V.; Seddon, K. R. Applications of ionic liquids in the chemical
industry Chem. Soc. Rev. 2008, 37, 123-150.
(3)
Zhao, H.; Xia, S. Q.; Ma, P. S. Use of ionic liquids as 'green' solvents for
extractions J. Chem. Technol. Biotechnol. 2005, 80, 1089-1096.
(4)
Zhao, D. B.; Wu, M.; Kou, Y.; Min, E. Ionic liquids: applications in catalysis Catal.
Today 2002, 74, 157-189.
(5)
Welton, T. Room-temperature ionic liquids. Solvents for synthesis and catalysis
Chem. Rev. 1999, 99, 2071-2083.
(6)
Brennecke, J. F.; Maginn, E. J. Ionic liquids: Innovative fluids for chemical
processing AIChE J. 2001, 47, 2384-2389.
(7)
Huddleston, J. G.; Willauer, H. D.; Swatloski, R. P.; Visser, A. E.; Rogers, R. D.
Room temperature ionic liquids as novel media for 'clean' liquid-liquid extraction Chem.
Commun. 1998, 1765-1766.
(8)
Tan, S. S. Y.; MacFarlane, D. R. Ionic Liquids in Biomass Processing Top. Curr.
Chem. 2009, 290, 311-339.
(9)
da Costa Lopes, A.; João, K.; Morais, A.; Bogel-Łukasik, E.; Bogel-Łukasik, R.
Ionic liquids as a tool for lignocellulosic biomass fractionation sustain chem process 2013, 1, 1-
31.
(10)
Sun, N.; Rodriguez, H.; Rahman, M.; Rogers, R. D. Where are ionic liquid
strategies most suited in the pursuit of chemicals and energy from lignocellulosic biomass?
Chem. Commun. 2011, 47, 1405-1421.
(11)
da Costa Lopes, A. M.; João, K. G.; Rubik, D. F.; Bogel-Łukasik, E.; Duarte, L. C.;
Andreaus, J.; Bogel-Łukasik, R. Pre-treatment of lignocellulosic biomass using ionic liquids:
Wheat straw fractionation Bioresour. Technol. 2013, 142, 198-208.
(12)
sugarcane bagasse in the presence of acidic ionic liquids Catal. Today 2013, 200, 99-105.
(13) Shill, K.; Padmanabhan, S.; Xin, Q.; Prausnitz, J. M.; Clark, D. S.; Blanch, H. W.
Long, J.; Li, X.; Guo, B.; Wang, L.; Zhang, N. Catalytic delignification of
Ionic Liquid Pretreatment of Cellulosic Biomass: Enzymatic Hydrolysis and Ionic Liquid Recycle
Biotechnol. Bioeng. 2011, 108, 511-520.
(14)
Iguchi, M.; Aida, T. M.; Watanabe, M.; Smith, R. L. Dissolution and recovery of
cellulose from 1-butyl-3-methylimidazolium chloride in presence of water Carbohydr. Polym.
2013, 92, 651-658.
(15)
Moreau, C.; Finiels, A.; Vanoye, L. Dehydration of fructose and sucrose into 5-
hydroxymethylfurfural in the presence of 1-H-3-methyl imidazolium chloride acting both as
solvent and catalyst J Mol Catal a-Chem 2006, 253, 165-169.
(16)
Forsyth, S. A.; MacFarlane, D. R.; Thomson, R. J.; von Itzstein, M. Rapid, clean,
and mild O-acetylation of alcohols and carbohydrates in an ionic liquid Chem. Commun. 2002,
714-715.
(17)
Forsyth, S. A.; MacFarlane, D. R. 1-alkyl-3-methylbenzotriazolium salts: ionic
solvents and electrolytes J. Mater. Chem. 2003, 13, 2451-2456.
(18)
Liu, Q.; Janssen, M. H. A.; van Rantwijk, F.; Sheldon, R. A. Room-temperature
ionic liquids that dissolve carbohydrates in high concentrations Green Chem. 2005, 7, 39.
ACS Paragon Plus Environment

Page 29 of 56
The Journal of Physical Chemistry
1
2
3
(19)
Lee, S. H.; Dang, D. T.; Ha, S. H.; Chang, W.-J.; Koo, Y.-M. Lipase-catalyzed
synthesis of fatty acid sugar ester using extremely supersaturated sugar solution in ionic liquids
4
5
Biotechnol. Bioeng. 2008, 99, 1-8.
6
7
8
9
(20)
enzyme-compatible ionic liquids that can dissolve carbohydrates Green Chem. 2008, 10, 696.
(21) Rosatella, A. A.; Branco, L. C.; Afonso, C. A. M. Studies on dissolution of
Zhao, H.; Baker, G. A.; Song, Z.; Olubajo, O.; Crittle, T.; Peters, D. Designing
carbohydrates in ionic liquids and extraction from aqueous phase Green Chem. 2009, 11, 1406-
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
1413.
(22)
ionic liquids - Experimental data and modeling Fluid Phase Equilib. 2012, 314, 22-28.
(23) Conceiçao, L. J. A.; Bogel-Lukasik, E.; Bogel-Lukasik, R. A new outlook on
solubility of carbohydrates and sugar alcohols in ionic liquids Rsc Adv 2012, 2, 1846-1855.
(24) Paduszyński, K.; Okuniewski, M.; Domańska, U. Renewable Feedstocks in Green
Carneiro, A. P.; Rodriguez, O.; Macedo, E. A. Solubility of monosaccharides in
Solvents: Thermodynamic Study on Phase Diagrams of d-Sorbitol and Xylitol with Dicyanamide
Based Ionic Liquids The Journal of Physical Chemistry B 2013, 117, 7034-7046.
(25)
ionic liquids - Experimental data and modeling J. Chem. Thermodyn. 2012, 55, 184-192.
(26) Carneiro, A. P.; Rodríguez, O.; Macedo, E. A. Fructose and Glucose Dissolution
in Ionic Liquids: Solubility and Thermodynamic Modeling Ind. Eng. Chem. Res. 2013.
(27) Payne, S. M.; Kerton, F. M. Solubility of bio-sourced feedstocks in 'green'
solvents Green Chem. 2010, 12, 1648-1653.
Carneiro, A. P.; Rodriguez, O.; Macedo, E. A. Solubility of xylitol and sorbitol in
(28)
Zakrzewska, M. E.; Bogel-Lukasik, E.; Bogel-Lukasik, R. Solubility of
Carbohydrates in Ionic Liquids Energ. Fuel. 2010, 24, 737-745.
(29)
Renon, H.; Prausnitz, J. M. Local compositions in thermodynamic excess
functions for liquid mixtures AIChE J. 1968, 14, 135-144.
(30)
Abrams, D. S.; Prausnitz, J. M. Statistical Thermodynamics of Liquid-Mixtures -
New Expression for Excess Gibbs Energy of Partly or Completely Miscible Systems AIChE J.
1975, 21, 116-128.
(31)
of state to biomass fast pyrolysis liquid Chem. Eng. Sci. 2005, 60, 617-624.
(32) Huang, S. H.; Radosz, M. Equation of State for Small, Large, Polydisperse, and
Associating Molecules Ind. Eng. Chem. Res. 1990, 29, 2284-2294.
(33) Ji, P. J.; Feng, W.; Tan, T. W. Density calculation of sugar solutions with the
SAFT model J. Chem. Eng. Data 2007, 52, 135-140.
Feng, W.; Vanderkooi, H.; Deswaanarons, J. Application of the SAFT equation
(34)
Ji, P. J.; Feng, W.; Tan, T. W.; Zheng, D. X. Modeling of water activity, oxygen
solubility and density of sugar and sugar alcohol solutions Food Chem. 2007, 104, 551-558.
(35)
Held, C.; Carneiro, A. P.; Rodríguez, O.; Sadowski, G.; Macedo, E. A. Modeling
thermodynamic properties of aqueous single-solute and multi-solute sugar solutions with PC-
SAFT AIChE J. 2013. DOI: 10.1002/aic.14212
(36)
Maia, F. M.; Tsivintzelis, I.; Rodriguez, O.; Macedo, E. A.; Kontogeorgis, G. M.
Equation of state modelling of systems with ionic liquids: Literature review and application
with the Cubic Plus Association (CPA) model Fluid Phase Equilib. 2012, 332, 128-143.
(37)
Vega, L. F.; Vilaseca, O.; Llovell, F.; Andreu, J. S. Modeling ionic liquids and the
solubility of gases in them: Recent advances and perspectives Fluid Phase Equilib. 2010, 294,
15-30.
(38)
Kroon, M. C.; Karakatsani, E. K.; Economou, I. G.; Witkamp, G. J.; Peters, C. J.
Modeling of the carbon dioxide solubility in imidazolium-based ionic liquids with the tPC-PSAFT
equation of state J. Phys. Chem. B 2006, 110, 9262-9269.
(39)
Nann, A.; Mündges, J.; Held, C.; Verevkin, S. P.; Sadowski, G. Molecular
Interactions in 1-Butanol + IL Solutions by Measuring and Modeling Activity Coefficients The
Journal of Physical Chemistry B 2013, 117, 3173-3185.
ACS Paragon Plus Environment