Modeling solubility of carbohydrates in ionic liquids using PC-SAFT

Aristides P. Carneiro 1, Christoph Held 2, Oscar Rodrguez 1, Eugnia A. Macedo 1, Gabriele Sadowski 2

1LSRE

Laboratory of Separation and Reaction Engineering Associate Laboratory LSRE/LCM, Faculdade de Engenharia, Universidade
do Porto. Rua Dr. Roberto Frias s/n 4200 465 Porto, Portugal
2Department for Biochemical and Chemical Engineering, TU Dortmund, Dortmund, Germany
Technische Universitt Dortmund, Emil-Figge-Str. 70, 44227 Dortmund, Germany

Abstract
Ionic liquids (ILs) have proved their ability as suitable solvents for biorefining[1]. The phase equilibrium between biomass-derived carbohydrates and ionic liquids is critical for process
design and optimization in biorefineries.
Thermodynamic models to describe accurately these mixtures are of key importance for process design and optimization. PC-SAFT equation of state [2], accounting for chain formation,
dispersive forces and association, represents a powerful tool for the modeling of complex fluid mixtures, including ILs[3].
This work presents the modeling results of solubility of four carbohydrate-derived solutes in three imidazolium-based ILs using PC-SAFT. The pure component parameters were obtained
using binary aqueous data for carbohydrates and for ILs pure density was considered for parameter estimation.
The obtained parameters allow for good solubility predictions (with ARDs lower than 30 % in most of the cases). Generally, very low kij parameters were fitted to solubility data, giving even
more accuracy to the quantitative description of the phase equilibria (ARDs lower than 10 % in most of the cases).

Methods and materials

Modeling
PC-SAFT equation of state:

Carbohydrates : monosaccharides and sugar alcohols

Parameters :
donor site

glucose

sorbitol

xylitol

fructose

miseg

acceptor site

AiBi/ kB and kAiBi

ui /kB

Parameter estimation - carbohydrates

Water content (wt%)

Ionic Liquids :

N of association sites (Nassoc. = N + N) : OH groups

Four Adjustable parameters:

Data to fit parameters:

[emim][EtSO4]

0.10

miseg

Aqueous solution densities at 298 K

Aqueous solution osmotic coefficients at 298 K

[emim][N(CN)2]

ui /kB

Association parameters: AiBi/ kB = 5000

KAiBi = 0.1

0.18

kij water- carbohydrate

Parameter estimation Ionic Liquids

Association parameters[3]:

[emim][CF3CO2]

AiBi/

0.12

KAiBi

Three adjustable
parameters:

kB = 5000
= 0.1

miseg

ui /kB

kij IL 1-propanol = 0 [4]

Data to fit parameters:

Experimental Methodology:
IL
+
carbohydrate

To the best overall

predictions

Pure IL densities ( 278 - 343 K )

Determination of the association scheme:
Select a
scheme

HPLC
quantification

Fit parameters

Predict
solubility of
sugars in IL

The chosen association

scheme

EoS

Phase separation

Stirring time ~ 48h

Results
Solubility modeling predictions (kij =0)

water solubility predictions PC-SAFT

Parameters carbohydrates
C. Held, A. Carneiro, O. Rodriguez, E.A. Macedo, G. Sadowski,
Modeling thermodynamic properties of aqueous single-solute and
multi-solute sugar solutions with PC-SAFT, 2013, (submitted)

xylitol
fructose

sorbitol
glucose

Parameters ionic liquids

%ARD

Ionic liquid
[emim][EtSO4]

0.14

[emim][N(CN)2]

0.22

[emim][CF3CO2]

0.06

Pure IL densities : PC-SAFT modeling

Solubility modeling - correlation

fructose

xylitol

sorbitol

glucose

[emim][CF3CO2]

[emim][EtSO4]

[emim][N(CN)2]

kij carbohydrate/IL
[emim][EtSO4]
[emim][N(CN)2]
[emim][CF3CO2]

Remarkable low kijs

Solubility of
glucose in the ILs
glucose

sorbitol
fructose

[emim][CF3CO2]
[emim][N(CN)2]
[emim][EtSO4]

good solubility predictions

xylitol

[emim][EtSO4]

[emim][N(CN)2]

[emim][CF3CO2]

Conclusions
Solubility of carbohydrates in ILs was sucessfully modeled using the PC-SAFT model. Very satisfactory results were obtained using a strategy which considers the number of association
sites according to the OH groups in carbohydrates and, fitting IL parameters to pure IL density.
Good solubility predictions were obtained with this strategy with only three adjustable IL parameters. The number association sites of each ionic liquid were determined through the
optimization of solubility predictions.
An accurate quantitative description of solubility data was possible by fitting kij parameters to the experimental solubility data. The low obtained values of kij (in magnitude) also indicates
that PC-SAFT in its predictive form models the data quite satisfactorily.

Aknowledgements:

Email address: carneiro.aristides@gmail.com

References:

[1]
[2]
[3]
[4]

S.S.Y. Tan, D.R. MacFarlane, Top. Curr. Chem. 290, 2010, 311339.
J. Gross, G. Sadowski, Ind. Eng. Chem. Res., 40, 2001, 12441260
Xiaoyan J, C. Held, G. Sadowski, Fluid Phase Equilib. 335, 2012, 64-73
A. Nann, J. Mndges, C. Held, S.P. Verevkin, G. Sadowski, The Journal of Physical Chemistry B,
2013, 117, 3173.

Departamento de Engenharia Qumica, Faculdade de Engenharia da Universidade do Porto

Rua Dr. Roberto Frias, s/n, 4200-465 Porto, Portugal