Accepted Manuscript
Content
Accepted Manuscript
Title: Solubility of monosaccharides in ionic liquids Experimental data and modeling Authors: Aristides P. Carneiro, Oscar Rodr´ ıguez, Eug´ enia A. Macedo PII: DOI: Reference: To appear in: Received date: Revised date: Accepted date: S0378-3812(11)00490-0 doi:10.1016/j.fluid.2011.10.011 FLUID 9009 Fluid Phase Equilibria 30-7-2011 11-10-2011 13-10-2011
Please cite this article as: A.P. Carneiro, O. Rodr´ ıguez, E.A. Macedo, Solubility of monosaccharides in ionic liquids - Experimental data and modeling, Fluid Phase Equilibria (2010), doi:10.1016/j.fluid.2011.10.011 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
*Highlights
Highlights Ionic Liquids as solvents for biomass carbohydrates Solubility of glucose, fructose, xylose and galactose in two ionic liquids Correlation of the experimental data with NRTL and UNIQUAC Determination of the apparent thermodynamic functions of dissolution
Ac
ce pt
ed
M
an
Page 1 of 22
us
cr
ip t
*Revised Manuscript
Solubility of monosaccharides in ionic liquids – Experimental data and modeling
Aristides P. Carneiro, Oscar Rodríguez, Eugénia A. Macedo*
Laboratory of Separation & Reaction Engineering, Associate Laboratory LSRE/LCM Faculdade de Engenharia da Universidade do Porto, Rua Dr. Roberto Frias, s/n, 4200-465 Porto, Portugal.
* To whom correspondence should be addressed. Tel.: +351 22 5081653. Fax: +351 22 508 1674. E-mail: [email protected].
Abstract
Keywords: Ionic liquids, Carbohydrates, Solubility, Thermodynamic models
Ac
Biomass represents nowadays one of the more sustainable alternatives as a source of fuels and chemicals. The capability of ionic liquids to act as selective solvents and catalysts for biomass processing has already been proven. Thus they are a serious alternative to conventional solvents, provided that phase equilibria with biomass derived compounds is studied. To overcome the lack of experimental data on phase equilibria of biomass carbohydrates in Ionic liquids, the solubilities of monosaccharides such as D-(+)-Glucose, D-(-)-Fructose, D-(+)-Xylose and D-(+)Galactose in two ILs were measured in a temperature range from 288 K to 328 K. The ionic liquids selected for this work were the 1-etyhl-3-methylimidazolium ethylsulfate, [emim][EtSO4], and the Aliquat®336. The experimental technique used a combination of an isothermal method to attain the solid-liquid equilibrium, and quantitative analysis by HPLC with refractive index detector. For the two ionic liquids, the ranking of solubility follows the series: D-(-)-Fructose > D-(+)-Xylose > D-(+)-Glucose > D-(+)-Galactose. The correlation of the solubility data was achieved through local composition thermodynamic models NRTL and UNIQUAC. For both models the average relative deviations are almost all below 4 %, which is a satisfactory result. The apparent standard thermodynamic functions of dissolution were also determined from the experimental data using an approach based on a modified Van’t Hoff equation. The results show good accuracy.
ce pt
ed
M
an
us
Page 2 of 22
cr
ip t
1. Introduction The human exploitation of nature has been somehow chaotic over the years. Feedstocks, mainly the fossil ones, are being extracted at increasing rates, probably faster than they can be regenerated. Direct consequences are problems raised in several fields of society such as politics, economics, social and environment. In order to attain equilibrium in our planet’s energy and mass balances, other sources for feedstocks must be explored. The output needs to be the same as from petroleum-based production, but using a greener and more sustainable path. Biomass is an abundant and renewable source of carbon, which can be used to produce energy, fuel and chemicals[1]. Lignocellulosic biomass feedstocks are composed mainly of cellulose, hemicellulose, starch and lignin. These carbohydrates can be easily converted into fermentable sugars or phenolic acids, which can be further used to produce fuel or as a source of added-value chemicals[2]. The depolymerisation of cellulose into fermentable sugars can be attained by gasification, pyrolysis, liquefaction and acid hydrolysis of biomass[3]. While sugars can be further processed through reactions of fermentation, dehydration, acetylation and hydrogenation producing compounds such as 5hydroxymethyl furfural, sorbitol or bioethanol (among others), which are potential "building blocks" for the production of a large number of chemicals. Connecting the production of these three fundamental basic resources (energy, fuel and chemicals) on an optimized perspective is the goal of newer integrated biorefineries[4]. On one hand, high volume and low value products can be obtained (energy and fuel); on the other hand, low volume and high value chemicals are produced to maintain the economic feasibility of the biorefinery[5]. Biorefining consists on a set of chemical and biological processes, including pretreatment of the biomass, conversion of biopolymers into other compounds (e.g., fermentable sugars), fermentation reactions, separations, etc[6-9]. All of these processes have a common feature: the need to use solvents. Whether acting as reaction media or as an extraction agent, solvents play a key role in this biorefining concept. Room temperature ionic liquids (ILs) are a novel class of green solvents: salts which are liquid near to room temperature[10-13]. As a kind of molten salts, they are composed only of ions, usually a combination of a larger organic cation and a smaller anion. Most of them exhibit interesting properties such as negligible vapour pressure, high thermal stability, high conductivity, high heat capacities, nonflammability, etc. Small modifications on the cation or the anion can often produce a large variation on physical properties such as density and viscosity, and also in their chemical affinity with other compounds. This behaviour makes them able to dissolve both polar and non-polar compounds, and even polymers. They are often labelled as “designer solvents” due to this ability to change their properties and chemistry with small modifications on the ions (properties can be tuned for each desired application). With all of these characteristics, ILs have the potential to act as solvents in many applications and replace common organic solvents avoiding losses by evaporation and atmospheric contamination. The use of ILs as solvents for biomass processing has been recently reported by several authors[3,14,15]. Their capacity to dissolve natural polymers present in biomass, like cellulose [16,17] and even wood[18] has also been proved. Such capacity is related mainly with the ionic liquid anion’s ability to establish hydrogen bonds
Ac
ce pt
ed
M
an
us
cr
ip t
Page 3 of 22
2. Materials and methods
2.1. Materials
Aliquat®336 containing 4.18 mass % of water (measured through KarlFisher titration) and with average molecular weight of 442, was purchased from Alfa Aesar. When drying this IL, it became solid at ~ 1 mass % of water in mass, so it was decided to work with it as received, without further purification. 1-Ethyl-3methylimidazolium ethylsulfate was prepared in our laboratory, following a procedure available in the literature[27]. 1-methylimidazole and diethylsulfate used in this synthesis were purchased from Merck with purities higher than 99% in both cases. Toluene was purchased from Fisher Scientific (99.99 %, HPLC grade). D-(+)-Glucose and D-(-)-Fructose were purchased from Merck. D-(+)Xylose and D-(+)-Galactose were purchased from Fluka. All these carbohydrates have purities greater than 99.0 % and were dried at 318 K for 24 h, before the experiments. A description of the compounds used in this work is presented in Table 1.
2.2. Preparation of 1-ethyl-3-methylimidazolium ethylsulfate - [emim][EtSO4]
Approximately 0.83 mol of 1-methylimidazole were diluted in 250 mL of toluene. Then, an equimolar amount of diethylsulfate was added dropwise at room temperature. The reaction was performed under nitrogen atmosphere and an ice bath was used to maintain temperatures below 313 K due to the highly exothermic reaction. The mixture was stirred for 24 h and after that, the two liquid phases were separated by decantation. The lower IL-rich phase was washed with toluene (3 x 200 mL). The excess toluene was removed by distillation under reduced pressures at about 323 K in a rotary evaporator (IKA RV 10 D S40). Finally the
Ac
ce pt
ed
M
with the hydroxyl groups of carbohydrates[19], while the cation’s structure also affects solvation properties, however with a smaller effect. Therefore, and despite the majority of them has still high purchase and production costs nowadays, which is a serious drawback to the scale-up of their applications, ILs can play an important role in the development of new biorefinery processes, either used as reaction media, catalysts, extraction solvents, or entrainers, among other functions. Solubility data of biomass-derived compounds such as carbohydrates in ionic liquids is an important issue in order to develop chemical or bio-processes involving these compounds. In the literature there are some data reported already[20-22], however it is far from being enough to have a good knowledge on phase equilibria of these systems. In most cases, these data are also unavailable in a sufficiently large temperature range, which turns their modeling a very difficult task. In this work, solubility of several monosaccharides, D-(+)-Glucose, D-(-)-Fructose, D-(+)-Xylose and D-(+)-Galactose, in two ILs has been measured in a temperature range from 288 K to 328 K using an isothermal method. The 1-ethyl-3methylimidazolium ethylsulfate, [emim][EtSO4] and the Aliquat®336 (a mixture of methyltrioctylammonium chloride and methyltridecylammonium chloride) were the two ionic liquids used as solvents. The apparent thermodynamic functions of dissolution, , and , have been also determined from a modified Van’t Hoff equation[23,24],and the data have been successfully correlated with local composition thermodynamic models: NRTL[25] and UNIQUAC[26].
an
us
cr
ip t
Page 4 of 22
resultant liquid was dried for 48 h under vacuum (P < 1×10-3 bar) at about 353 K, to remove the volatile compounds, including water. A pale yellow and slightly viscous liquid was obtained (yield greater than 90%), its water content (960 ppm) was measured by Karl Fisher titration in a Metrohm 870 KF Titrino plus titrator. The chemical structure was checked by 1H-NMR, with results very similar to those reported in the literature[27]. H (400 MHz, CDCl3) 1.239 (3H, t, NCH2CH3), 1.524 (3H, t, OCH2CH3), 3.981 (3H, s, NCH3), 4.047 (2H, q, OCH2), 4.284 (2H, q, NCH2), 7.504 (2H, s, C(4,5)H), 9.442(1H, s, C(2)H). The solubility measurements were performed using an isothermal technique based on the research group experience on solid-liquid equilibria[28]. First, to obtain saturated solutions, the carbohydrate was added, in a slight excess of the estimated solubility, to a specified amount of ionic liquid (10-15 g) in a 25 cm3 glass cell. Preliminary experiments were made to estimate the amounts to be added of each component. The equilibrium cells were equipped with a jacket where water from a thermostatic bath (JULABO F12-ED) was flowing to maintain the desired temperature, with a precision of ±0.1 K. The internal temperature of the cells was also measured with calibrated platinum probes, and its history was recorded by data acquisition software. Mixtures were magnetically stirred for at least 48 h, the time considered to be sufficient to reach the equilibrium, based on some preliminary tests and according to the experience of the team[28-30]. Once the solutions were saturated, the stirring was stopped and the excess solid gradually deposited in the bottom, resulting in a clear top liquid phase. Sampling was made in triplicate, using 2 mL syringes equipped with filter membranes (0.45 μm). Quantitative analysis was performed by reverse phase HPLC, using a LichroCART®250 mm x 4 mm column with Purospher®STAR RP-18e (5µm) as stationary phase. The eluent was water (100%) flowing at 0.7 mL∙min-1, and the detection was made by refractive index at 313 K using a HITACHI L-2490, with an injection volume of 20 µL of diluted sample. All samples were diluted conveniently and analysed three times each. Concentrations were calculated from calibration curves, previously prepared for each carbohydrate analyzed using the adopted chromatographic method. In the case of Aliquat®336, which is hydrophobic, the carbohydrate was re-extracted to an aqueous phase after precipitation with an antisolvent, dichloromethane[20]. This operation enables the quantification using the previous chromatographic method. The precision of the analytical method was estimated to be 0.88 %. 3. Results and discussion
3.1. Solubility data
Dissolving a solid in a liquid can be approximated, under some assumptions, to a melting process. Hence, solute melting properties are relevant to this study and they are presented in Table 2 for all the solutes used in this work. Another important issue is the water content of ionic liquids, which affects both the physical properties and their phase equilibria with other compounds. Thus, the water content of ILs used in this work was measured by Karl-Fisher titration, before and after the solubility measurements, as presented in Table 3.
Ac
ce pt
ed
M
an
us
cr
ip t
Page 5 of 22
2.3. Experimental determination of solubilities
In the case of 1-ethyl-3-methylimidazolium ethylsulfate, there is a small increase in the water content, which does not affect the properties neither the solubility since the ionic liquid is over 99.8% pure before and after measurements. On the other hand, Aliquat®336 shows a decrease in its water content; this is caused by the fact that the commercial product has 4.18 mass% of water and when submitted to temperatures around 328 K, there is a tendency to evaporate some water and its content decreases. However, these variations did not appear to be sufficiently large to produce effects on the solubility, as will be shown in results tendency (Table 4). As can be observed from Figures 1 and 2, the expected linear increase on solubility is verified in all cases, due to the relatively short range of temperatures studied. [emim][EtSO4] has shown a higher capacity to dissolve the carbohydrates than Aliquat®336. This can be explained, on one hand, with the higher ability of ethylsulfate anion to establish hydrogen bonds with the hydroxyl groups of carbohydrates. On the other hand, Aliquat®336 has a cation with large alkyl chains (hydrophobic), which reduces the affinity with polar substances such as monosaccharides. A solubility ranking could be drawn based on the melting properties of the solutes (independent of the ionic liquid), since all the solutes are chemically identical. Direct inspection of Tables 2 and 4 shows that D-(-)-Fructose is the most soluble carbohydrate, whereas D-(+)-Galactose is the less soluble. Their melting temperature and their melting enthalpy are related with this solubility ranking. The larger the melting temperature and enthalpy, the lower the solubility. Literature values for the studied binary systems are scarce. Solubility of glucose in Aliquat®336 at 308.15 K has been measured against the water content in the ionic liquid by Rosatella et al.[20] The data presented by the authors, for the same water content as the one used in this work, was also plotted in Figure 1 for comparison purposes. There is a significant deviation from the value obtained in this work, as observed in Figure 1. Actually, Aliquat®336 is a mixture of two ILs, the trioctylmethylammonium chloride and the tridecylmethylammonium chloride. Therefore, a possible reason for this deviation could be attributed to different ratios of C10 and C8 alkyl chains between ionic liquids used in the two works. A larger content of C10 chains could cause a decrease on glucose solubility due to an increase of the hydrophobic effect. Another possible reason could be related with the experimental techniques. However, the authors of the mentioned work did not present sufficient information, for example in respect to the time needed to reach the solid-liquid equilibrium, which is a very important issue in solubility experiments. Concerning the [emim][EtSO4], glucose solubility at 333 K has been published elsewhere[31]. The reported solubility can be considered in good agreement with the value obtained in this work, extrapolating to that temperature (see Figure 2). However, no information has been given about the water content in the ionic liquid, and this could be related with the deviation. A recent patent[32], also reported solubility values for glucose and fructose in [emim][EtSO4] at room temperature .Those values are far from those obtained in this work and [31], as shown in Figure 2, but a reliable comparison is not possible since the authors do not mention how the measurements were carried out neither the water content on the ionic liquid.
Ac
ce pt
ed
M
an
us
cr
ip t
Page 6 of 22
3.2. Thermodynamic functions of dissolution
Apparent thermodynamic functions of dissolution, , and can be obtained from the experimental solubility data. These thermodynamic functions are useful to provide information about the dissolution process to explain the molecular behaviour during the dissolution, and even to obtain energetic parameters for molecular-based equations of state to model the phase equilibria[33]. The approach proposed by Krug et al.[23], based on a modified Van´t Hoff equation, was used in this work,
us
where represents the mole fraction of the solute at saturated conditions (solubility), R is the ideal gas constant, T is the temperature in K, and is the harmonic average of mean of the experimental temperatures given by:
cr
where, Np is the number of experimental data points. The apparent standard enthalpy change of dissolution, , is taken from the slope of the linear plot To obtain the apparent standard change of Gibbs energy during the dissolution, , this approach uses the following expression:
ed
M
of
against
, Figure 3.
an
ce pt
where the constant k, is the yy axis intercept of the linear plot represented on Figure 3. Finally, through the definition of Gibbs energy, one can calculate the apparent standard entropy change of dissolution, . (4)
Figure 3 explains why not all the solubility data were used to calculate the thermodynamic functions for three systems. The data point neglected, the one which presented larger deviations to the linear behaviour, was in all cases at the lowest temperature of the range, 288 K. This may be related to the difficulty in efficient stirring due to the high viscosities of ionic liquids at this temperature. Table 5 presents the thermodynamic functions for each binary system studied. The obtained values for are positive in all cases, showing an endothermic behaviour in these dissolution processes. The lower this enthalpy, the easier it is for the system to overcome the energetic barrier to the dissolution. However, we cannot focus only in enthalpy to justify values of solubility. Entropy also plays an important role as an indicator if the configuration of solvent molecules after dissolution, is more or less favourable than the configuration in pure solvent, from an entropic point of view. Hence, positive values in all of the studied systems indicate more degrees of freedom to the ionic liquids when they
Ac
ip t
(1) (2) (3)
Page 7 of 22
are solvating molecules of carbohydrate than they have in a pure ionic liquid state. Finally, the positive values of show that, these dissolution processes are all non-spontaneous; requiring a continuous energy supply to occur.
3.3. Modeling solubilities through local composition models
M
In this work, two local composition models were used, to represent the solubility data, NRTL and UNIQUAC. Although these models were initially developed for non-electrolytes, they have been used with a good accuracy in many studies involving phase equilibria with ILs[24,34,35]. The solubility of the carbohydrates in the ILs, were determined from an expression based on the symmetric convention for the calculation of the activity coefficients, i.e., the pure liquid at the solution temperature as a standard state for the carbohydrate, together with their fusion entalpy, , melting temperature, , and heat capacity difference between pure liquid and solid states, . Table 2, shows these properties for the carbohydrates used in this work. For a general case, this equation can be derived through an idealized thermodynamic cycle[36] between the solid and liquid carbohydrate phase states, under some assumptions: (i) The solvent does not appear in the solid phase. (ii) The difference on the heat capacity between the pure liquid and the pure solid carbohydrate is considered temperature independent and equal to such difference at melting temperature, . Thus, the resulting expression to the calculation of the solubilities is:
an
us
cr
Where, is the mole fraction of solute (subscript 2) in the saturated solution, the activity coefficient of solute in the liquid phase, T, the temperature and R, the ideal gas constant. In order to compute solubility values, must be known. Its values are obtained from the models mentioned above, since they are explicit for activity coefficients. The binary parameters of NRTL and UNIQUAC models were obtained for each binary system fitting the model to the experimental data. Beyond the binary parameters, NRTL model has also a parameter, α, related with the nonrandomness of the mixture. When the number of available points for solubility data are reduced, this parameter has to be fixed and it can be made using some heuristics[37]. Thus, in this work a value of 0.3 was adopted for α. UNIQUAC model is characterized with structural parameters of volume and area, r and q, respectively, for each pure component. Available values in the literature were used in this work, except for Aliquat®336 which were estimated through an empiric correlation proposed by Domanska and Mazurowska, 2004[38] using the molar volume of the ionic liquid at 298.15 K (0.8865 g∙cm-3, measured experimentally on an Anton Paar DMA 4500M densimeter). Table 6, presents the structural parameters used for each compound and also the melting properties of the carbohydrates. Hence, the binary parameters have been obtained for both models and binary systems studied, using a Matlab™ framework, and the solubilities were calculated using these parameters to generate the activity coefficients in eq 5.
Ac
ce pt
ed
ip t
(5)
Page 8 of 22
Table 6 presents the obtained results for this thermodynamic modeling. Figures 1 and 2 shows qualitatively, the tendency of the experimental data and their correlation with the NRTL model. The average absolute relative deviation was determined for each case to compare and evaluate the accuracy.
(6) As was done when determining the thermodynamic functions of dissolution, the lower temperature solubility (288 K) was neglected in the three systems with [emim][EtSO4], while modeling the solubility data with NRTL and UNIQUAC equations, improving the fitness of the correlation with such models. Solubilities from UNIQUAC model were not plotted against experimental data in Figures 1 and 2 due to visualization issues, since their values are very close to those from NRTL model. Thus, the two models are able to correlate the solubility data with a good accuracy, showing average deviations smaller than 4 % in almost all the cases. There is no significant difference whether predicting solubility of these systems with NRTL or with UNIQUAC. 4. Conclusions
The solubility of four monosaccharides was measured over the temperature range of 288–328 K in two ionic liquids, using an isothermal technique combined with HPLC analysis. The results showed that solubility values for [emim][EtSO4] were higher than for Aliquat®336, which is explained by the fact of hydrofobicity and larger non-polar alkyl chains on the latter which have not much affinity with the hydroxyl groups of the monosaccharides. On the other hand the ethylsulfate anion has oxygen atoms giving extra hydrogen acceptor capability to the [emim][EtSO4], combined with its short alkyl chained cation make it a good IL to dissolve polar molecules such as the carbohydrates. It was also verified a kind of ranking on the solubility of the monosaccharides, independently of the IL: D-(-)Fructose > D-(+)-Xylose > D-(+)-Glucose > D-(+)-Galactose, which could be justified with the melting properties of these solutes, since they are very similar compounds. The apparent standard thermodynamic functions of dissolution were also determined in this work from a Van’t Hoff based equation, which allows to support that all of dissolution processes are non spontaneous, , requiring heat to occur, and favourable under an entropic point of view, . NRTL and UNIQUAC local composition models were applied to correlate the experimental data, the binary interaction parameters for both models were determined, and the models were used to compute the solubility of the studied binary systems, showing very good agreement with the experimental data, and no significant difference between the two thermodynamic models accuracy.
Ac
ce pt
ed
M
an
us
cr
ip t
Page 9 of 22
List of symbols f k Np P R T x fugacity yy axis intercept number of experimental points pressure (bar) ideal gas constant (8.314 J∙mol-1∙K-1) temperature (K) mole fraction
Acknowledgements Financial support for this work was in part provided by LSRE, for which the authors are thankful. A.C. acknowledges the financial support (Grant SFRH/BD/62105/2009) from Fundação para a Ciência e a Tecnologia (FCT, Portugal).
Ac
Abbreviations AARD Average absolute relative deviation HPLC High performance liquid chromatography NRTL Non Random Two Liquid ILs Room temperature ionic liquids UNIQUAC Universal Quasi Chemical
ce pt
Superscripts 0 standard state 0,L pure liquid state 0,S pure solid state exp experimental L liquid phase model calculated from the model
ed
M
Subscripts 2 dissol. fus hm i
related to the solute related to the dissolution process fusion harmonic mean related to the i compound
an
Page 10 of 22
us
Greek letters γ activity coefficient ΔCp heat capacity change (J∙mol-1∙K-1) ΔG Gibbs energy change (J∙mol-1) ΔH enthalpy change (J∙mol-1) ΔS entropy change (J∙mol-1∙K-1)
cr
ip t
References [1] T.P. Vispute, H. Zhang, A. Sanna, R. Xiao, G.W. Huber, Science 330 (2010) 12221227. [2] T.A. Werpy, G. Petersen, Top Value Added Chemicals From Biomass: I. Results of Screening for Potential Candidates from Sugars and Synthesis Gas, in, Pacific Northwest National Laboratory, 2004. [3] S.S.Y. Tan, D.R. MacFarlane, Top. Curr. Chem. 290 (2009) 311-339. [4] J.H. Clark, F.E.I. Deswarte, T.J. Farmer, Biofuel Bioprod Bior. 3 (2009) 72-90. [5] J.J. Bozell, Clean-Soil Air Water 36 (2008) 641-647. [6] B. Kamm, M. Kamm, Appl. Microbiol. Biotechnol. 64 (2004) 137-145. [7] B. Kamm, C. Hille, P. Schönicke, G. Dautzenberg, Biofuel Bioprod Bior. 4 (2010) 253-262. [8] L.R. Lynd, C.E. Wyman, T.U. Gerngross, Biotechnol. Prog. 15 (1999) 777-793. [9] Y.H.P. Zhang, J. Ind. Microbiol. Biotechnol. 35 (2008) 367-375. [10] J.F. Brennecke, E.J. Maginn, AIChE J. 47 (2001) 2384-2389. [11] K.R. Seddon, J. Chem. Technol. Biotechnol. 68 (1997) 351-356. [12] J.G. Huddleston, H.D. Willauer, R.P. Swatloski, A.E. Visser, R.D. Rogers, Chem. Commun. (1998) 1765-1766. [13] T. Welton, Chem. Rev. 99 (1999) 2071-2083. [14] T.C.R. Brennan, S. Datta, H.W. Blanch, B.A. Simmons, B.M. Holmes, Bionerg. Res. 3 (2010) 123-133. [15] F.R. Tao, H.L. Song, L.J. Chou, Carbohydr. Res. 346 (2011) 58-63. [16] Y. Fukaya, K. Hayashi, M. Wada, H. Ohno, Green Chem. 10 (2008) 44. [17] R.P. Swatloski, S.K. Spear, J.D. Holbrey, R.D. Rogers, J. Am. Chem. Soc. 124 (2002) 4974-4975. [18] I. Kilpelainen, H. Xie, A. King, M. Granstrom, S. Heikkinen, D.S. Argyropoulos, J. Agric. Food Chem. 55 (2007) 9142-9148. [19] G. Moyna, R.C. Remsing, Z.W. Liu, I. Sergeyev, J. Phys. Chem. B 112 (2008) 7363-7369. [20] A.A. Rosatella, L.C. Branco, C.A.M. Afonso, Green Chem. 11 (2009) 1406. [21] M.E. Zakrzewska, E. Bogel-Łukasik, R. Bogel-Łukasik, Energ. Fuel. 24 (2010) 737-745. [22] Q. Liu, M.H.A. Janssen, F. van Rantwijk, R.A. Sheldon, Green Chem. 7 (2005) 39. [23] R.R. Krug, W.G. Hunter, R.A. Grieger, J. Phys. Chem. 80 (1976) 2341-2351. [24] E.K. Panteli, E.K. Voutsas, J. Chem. Eng. Data 54 (2009) 812-818. [25] H. Renon, J.M. Prausnitz, AIChE J. 14 (1968) 135-144. [26] D.S. Abrams, J.M. Prausnitz, AIChE J. 21 (1975) 116-128. [27] J.D. Holbrey, W.M. Reichert, R.P. Swatloski, G.A. Broker, W.R. Pitner, K.R. Seddon, R.D. Rogers, Green Chem. 4 (2002) 407-413. [28] A.M. Peres, E.A. Macedo, Ind. Eng. Chem. Res. 36 (1997) 2816-2820. [29] F.L. Mota, A.R. Carneiro, A.J. Queimada, S.P. Pinho, E.A. Macedo, Eur. J. Pharm. Sci. 37 (2009) 499-507. [30] L.A. Ferreira, E.A. Macedo, S.P. Pinho, Chem. Eng. Sci. 59 (2004) 3117-3124. [31] H. Zhao, C.L. Jones, J.V. Cowins, Green Chem. 11 (2009) 1128-1138. [32] I.M. Al Nashef, M.H. Gaily, S.M. Al-zahrani, A.E. Abasaeed, Method for separating fructose and glucose, King Saud University (Riyadh, SA), United States, 2011. [33] E.K. Karakatsani, I.G. Economou, M.C. Kroon, M.D. Bermejo, C.J. Peters, G.-J. Witkamp, Phys. Chem. Chem. Phys. 10 (2008) 6160.
Ac
ce pt
ed
M
an
us
cr
ip t
Page 11 of 22
List of figures
Ac
ce pt
ed
Figure 3. Linear representation of ln x2 against (1/T-1/Thm). a) [emim][EtSO4] as solvent, b) Aliquat®336 as solvent.
M
Figure 2. Solubility of the carbohydrates in [emim][EtSO4] and NRTL correlation
an
Figure 1. Solubility of the carbohydrates in Aliquat®336 and NRTL correlation
us
cr
[34] F.M. Maia, O. Rodriguez, E.A. Macedo, Fluid Phase Equilib. 296 (2010) 184-191. [35] L.D. Simoni, Y. Lin, J.F. Brennecke, M.A. Stadtherr, Ind. Eng. Chem. Res. 47 (2008) 256-272. [36] J.M. Prausnitz, R.N. Lichtenthaler, E.G. Azevedo, Molecular Thermodynamics of Fluid-Phase Equilibria third ed., Prentice Hall, New Jersey, 1999. [37] G.M. Kontogeorgis, G.K. Folas, Thermodynamic Models for Industrial Applications, John Wiley & Sons Ltd, 2010. [38] U. Domanska, L. Mazurowska, Fluid Phase Equilib. 221 (2004) 73-82. [39] W. Feng, H. Van der Kooi, J. de Swaan Arons, Chem. Eng. Sci. 60 (2005) 617624. [40] O. Ferreira, E.A. Brignole, E.A. Macedo, Ind. Eng. Chem. Res. 42 (2003) 62126222. [41] S.O. Jonsdottir, S.A. Cooke, E.A. Macedo, Carbohydr. Res. 337 (2002) 1563-1571. [42] V.H. Alvarez, M. Aznar, J. Chin. Inst. Chem. Eng. 39 (2008) 353-360. [43] S.A. Cooke, S.O. Jonsdottir, P. Westh, Fluid Phase Equilib. 194 (2002) 947-956.
ip t
Page 12 of 22
Table(s)
Table 1. Sample description Initial Mass Purification Final Mole Analysis Method Fraction Purity Method Fraction Purity Alfa Aesar 0.96 none Karl-Fisher [emim][EtSO4]b Synthesis not measured Distillation 0.999 Titration , 1HNMR,13C-NMRc D-(+)-glucose Merck 0.99 none D-(-)-fructose Merck 0.99 none D-(+)-xylose Fluka 0.99 none D-(+)-galactose Fluka 0.99 none Source Chemical Name Aliquat®336a
Ac
ce pt
ed
M
an
Page 13 of 22
us
cr
a
Commercial name, a mixture of methyltrioctylammonium chloride and methyltridecylammonium chloride, with the first as predominant. b[emim][EtSO4] - 1-ethyl-3-methylimidazolium ethylsulfate c Nuclear magnetic ressonance
ip t
Table 2. Melting properties of the monosaccharides used in this work
Solute D-(+)-Glucose D-(-)-Fructose D-(+)-Xylose D-(+)-Galactose
M (g∙mol-1) 180.2 180.2 150.1 180.2
Tfus (K) 423.15 378.15 416.15 436.15
ΔfusH ΔfusCp -1 (kJ∙mol ) (J∙mol-1∙K-1) 32.4 26.0 31.7 43.8 120 99 97 139
Ref. [39,40] [39] [41] [41]
Ac
ce pt
ed
M
an
Page 14 of 22
us
cr
ip t
Table 3. Water contents of ionic liquids
Solvent [emim][EtSO4] Aliquat®336
mass % of Water Initial Final 0.1 0.2 4.2 3.8
Ac
ce pt
ed
M
an
Page 15 of 22
us
cr
ip t
Table 4. Experimental solubilities of carbohydrates in the ILs
1-ethyl-3-methylimidazolium ethylsulfate - [emim][EtSO4]
D-(+)-Glucose
T* / K Sol. mass % SD**
D-(-)-Fructose
T/K Sol. mass % SD
D-(+)-Xylose
T/K Sol. mass % SD
D-(+)-Galactose
T/K Sol. mass % SD
±0.01
D-(+)-Glucose
T/K Sol. mass % SD
±0.01
D-(-)-Fructose
T/K Sol. mass % SD
±0.02 Aliquat®336
T/K
cr
SD T/K
average standard relative uncertainty, u(Sol.)/Sol.
D-(+)-Xylose
Sol. mass %
us
12.6 0.9 14 2 15.4 0.8 16.70 0.04
ip t
298.3 308.6 318.3 329.2 4.2 6.3 7.4 8.6 ±0.003
288.2 298.2 308.2 318.6 328.3
10.4 15.5 19.1 21.8 23.7
0.1 0.2 0.6 0.7 0.4
288.2 298.2 308.0 318.4 329.3
25.7 29 33.8 37.4 43.5
0.4 2 0.3 0.1 0.1
288.1 298.1 308.3 318.0 328.5
15.0 0.1 20.5 0.1 23.0 0.2 24.51 0.03 27.41 0.03
288.8 298.2 308.5 318.2 328.2
4.0 6.88 9.5 11.8 14.0
0.1 0.03 0.1 0.1 0.2
±0.005
D-(+)-Galactose
Sol. mass % SD
average standard relative uncertainty, u(Sol.)/Sol.
±0.01
±0.02
M
297.9 307.8 318.4 327.9
9.0 10.4 11.6 12.5
0.2 0.3 0.3 0.1
298.2 307.7 318.3 328.2
15.4 18.4 21.8 25.2
0.5 0.1 0.2 0.4
298.5 308.8 318.9 330.0
0.2 0.2 0.4 0.2
Ac
ce pt
ed
*The uncertainty associated to the temperature measurement is: u(T)= ±0.01 K ** Standard deviation
an
±0.01
Page 16 of 22
Table 5. Apparent thermodynamic functions of dissolution
Solvent [emim][EtSO4] [emim][EtSO4] [emim][EtSO4] [emim][EtSO4] Aliquat®336 Aliquat®336 Aliquat®336 Aliquat®336
Solute D(+)Glucose D(-)Fructose D(+)Xylose D(+)Galactose D(+)Glucose D(-)Fructose D(+)Xylose D(+)Galactose
Np 4 5 4 4 4 4 4 4
Thm (K) 312.87 307.69 312.77 312.83 312.57 312.65 313.56 313.13
Ac
ce pt
ed
M
an
Page 17 of 22
us
cr
5.7 17.0
ip t
2.86 5.06
(kJ∙mol-1) 10.9 9.1 6.7 18.6 7.6 10.3
(kJ∙mol-1) 3.67 2.37 2.90 5.31 3.83 2.53
(J∙mol-1) 23.0 21.7 12.1 42.6 12.2 24.7 9.2 38.2
Table 6. Structural parameters used in UNIQUAC model
Compound [emim][EtSO4] Aliquat®336 D-(+)-Glucose D-(-)-Fructose D-(+)-Xylose D-(+)-Galactose
M (g.mol-1) 236.3 442 180.2 180.2 150.1 180.2
r
q
Ref. [42] [43] [41] [41] [41]
Ac
ce pt
ed
M
an
Page 18 of 22
us
cr
ip t
23.12 14.36 14.60 11.68 5.80 4.84 5.80 4.92 4.83 4.03 5.80 4.84
Table 7. NRTL and UNIQUAC modeling
Binary system Solvent Solute Np 4 5 4 4 4 4 4 4
NRTL (α=0,3) Δg12 (J/mol) -6337 -2190 -2241 -6765 1490 -461 -282 -7688 Δg21 (J/mol) 4454 -2057 -4730 2580 -4672 -2840 -6913 4090 AARD (%) 3.02 0.96 0.93 3.52 2.33 1.50 0.96 7.08
UNIQUAC Δu12 Δu21 (J/mol) (J/mol) -2327 2161 -2635 3533 -3622 8610 -1662 572 AARD (%) 3.21 1.14 0.91 3.64 2.43 1.42 0.85 7.28
glucose Aliquat®336 fructose xylose galactose
Ac
ce pt
ed
M
an
Page 19 of 22
us
cr
ip t
-2803 -2278 -3516 -2428 5333 3674 96187 2334
glucose fructose [emim][EtSO4] xylose galactose
Figure(s)
Figure 1
Ac
ce pt
ed
M
an
Page 20 of 22
us
cr
ip t
Figure 2
Ac
ce pt
ed
M
an
Page 21 of 22
us
cr
ip t
-0.500
ln x2
ln x2
-3.000 -2.50E-04
D-(+)-Glucose
-2.500 1.50E-04 3.50E-04 -2.50E-04
D-(+)-Galactose D-(+)-Glucose
-5.00E-05
-5.00E-05
cr
b)
ip t us
1/T-1/Thm D-(-)-Fructose D-(+)-Galactose
1.50E-04
3.50E-04
D-(+)-Xylose
1/T-1/Thm D-(-)-Fructose D-(+)-Xylose
Figure 3
Ac
ce pt
ed
M
an
a)
Page 22 of 22
Title: Solubility of monosaccharides in ionic liquids Experimental data and modeling Authors: Aristides P. Carneiro, Oscar Rodr´ ıguez, Eug´ enia A. Macedo PII: DOI: Reference: To appear in: Received date: Revised date: Accepted date: S0378-3812(11)00490-0 doi:10.1016/j.fluid.2011.10.011 FLUID 9009 Fluid Phase Equilibria 30-7-2011 11-10-2011 13-10-2011
Please cite this article as: A.P. Carneiro, O. Rodr´ ıguez, E.A. Macedo, Solubility of monosaccharides in ionic liquids - Experimental data and modeling, Fluid Phase Equilibria (2010), doi:10.1016/j.fluid.2011.10.011 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
*Highlights
Highlights Ionic Liquids as solvents for biomass carbohydrates Solubility of glucose, fructose, xylose and galactose in two ionic liquids Correlation of the experimental data with NRTL and UNIQUAC Determination of the apparent thermodynamic functions of dissolution
Ac
ce pt
ed
M
an
Page 1 of 22
us
cr
ip t
*Revised Manuscript
Solubility of monosaccharides in ionic liquids – Experimental data and modeling
Aristides P. Carneiro, Oscar Rodríguez, Eugénia A. Macedo*
Laboratory of Separation & Reaction Engineering, Associate Laboratory LSRE/LCM Faculdade de Engenharia da Universidade do Porto, Rua Dr. Roberto Frias, s/n, 4200-465 Porto, Portugal.
* To whom correspondence should be addressed. Tel.: +351 22 5081653. Fax: +351 22 508 1674. E-mail: [email protected].
Abstract
Keywords: Ionic liquids, Carbohydrates, Solubility, Thermodynamic models
Ac
Biomass represents nowadays one of the more sustainable alternatives as a source of fuels and chemicals. The capability of ionic liquids to act as selective solvents and catalysts for biomass processing has already been proven. Thus they are a serious alternative to conventional solvents, provided that phase equilibria with biomass derived compounds is studied. To overcome the lack of experimental data on phase equilibria of biomass carbohydrates in Ionic liquids, the solubilities of monosaccharides such as D-(+)-Glucose, D-(-)-Fructose, D-(+)-Xylose and D-(+)Galactose in two ILs were measured in a temperature range from 288 K to 328 K. The ionic liquids selected for this work were the 1-etyhl-3-methylimidazolium ethylsulfate, [emim][EtSO4], and the Aliquat®336. The experimental technique used a combination of an isothermal method to attain the solid-liquid equilibrium, and quantitative analysis by HPLC with refractive index detector. For the two ionic liquids, the ranking of solubility follows the series: D-(-)-Fructose > D-(+)-Xylose > D-(+)-Glucose > D-(+)-Galactose. The correlation of the solubility data was achieved through local composition thermodynamic models NRTL and UNIQUAC. For both models the average relative deviations are almost all below 4 %, which is a satisfactory result. The apparent standard thermodynamic functions of dissolution were also determined from the experimental data using an approach based on a modified Van’t Hoff equation. The results show good accuracy.
ce pt
ed
M
an
us
Page 2 of 22
cr
ip t
1. Introduction The human exploitation of nature has been somehow chaotic over the years. Feedstocks, mainly the fossil ones, are being extracted at increasing rates, probably faster than they can be regenerated. Direct consequences are problems raised in several fields of society such as politics, economics, social and environment. In order to attain equilibrium in our planet’s energy and mass balances, other sources for feedstocks must be explored. The output needs to be the same as from petroleum-based production, but using a greener and more sustainable path. Biomass is an abundant and renewable source of carbon, which can be used to produce energy, fuel and chemicals[1]. Lignocellulosic biomass feedstocks are composed mainly of cellulose, hemicellulose, starch and lignin. These carbohydrates can be easily converted into fermentable sugars or phenolic acids, which can be further used to produce fuel or as a source of added-value chemicals[2]. The depolymerisation of cellulose into fermentable sugars can be attained by gasification, pyrolysis, liquefaction and acid hydrolysis of biomass[3]. While sugars can be further processed through reactions of fermentation, dehydration, acetylation and hydrogenation producing compounds such as 5hydroxymethyl furfural, sorbitol or bioethanol (among others), which are potential "building blocks" for the production of a large number of chemicals. Connecting the production of these three fundamental basic resources (energy, fuel and chemicals) on an optimized perspective is the goal of newer integrated biorefineries[4]. On one hand, high volume and low value products can be obtained (energy and fuel); on the other hand, low volume and high value chemicals are produced to maintain the economic feasibility of the biorefinery[5]. Biorefining consists on a set of chemical and biological processes, including pretreatment of the biomass, conversion of biopolymers into other compounds (e.g., fermentable sugars), fermentation reactions, separations, etc[6-9]. All of these processes have a common feature: the need to use solvents. Whether acting as reaction media or as an extraction agent, solvents play a key role in this biorefining concept. Room temperature ionic liquids (ILs) are a novel class of green solvents: salts which are liquid near to room temperature[10-13]. As a kind of molten salts, they are composed only of ions, usually a combination of a larger organic cation and a smaller anion. Most of them exhibit interesting properties such as negligible vapour pressure, high thermal stability, high conductivity, high heat capacities, nonflammability, etc. Small modifications on the cation or the anion can often produce a large variation on physical properties such as density and viscosity, and also in their chemical affinity with other compounds. This behaviour makes them able to dissolve both polar and non-polar compounds, and even polymers. They are often labelled as “designer solvents” due to this ability to change their properties and chemistry with small modifications on the ions (properties can be tuned for each desired application). With all of these characteristics, ILs have the potential to act as solvents in many applications and replace common organic solvents avoiding losses by evaporation and atmospheric contamination. The use of ILs as solvents for biomass processing has been recently reported by several authors[3,14,15]. Their capacity to dissolve natural polymers present in biomass, like cellulose [16,17] and even wood[18] has also been proved. Such capacity is related mainly with the ionic liquid anion’s ability to establish hydrogen bonds
Ac
ce pt
ed
M
an
us
cr
ip t
Page 3 of 22
2. Materials and methods
2.1. Materials
Aliquat®336 containing 4.18 mass % of water (measured through KarlFisher titration) and with average molecular weight of 442, was purchased from Alfa Aesar. When drying this IL, it became solid at ~ 1 mass % of water in mass, so it was decided to work with it as received, without further purification. 1-Ethyl-3methylimidazolium ethylsulfate was prepared in our laboratory, following a procedure available in the literature[27]. 1-methylimidazole and diethylsulfate used in this synthesis were purchased from Merck with purities higher than 99% in both cases. Toluene was purchased from Fisher Scientific (99.99 %, HPLC grade). D-(+)-Glucose and D-(-)-Fructose were purchased from Merck. D-(+)Xylose and D-(+)-Galactose were purchased from Fluka. All these carbohydrates have purities greater than 99.0 % and were dried at 318 K for 24 h, before the experiments. A description of the compounds used in this work is presented in Table 1.
2.2. Preparation of 1-ethyl-3-methylimidazolium ethylsulfate - [emim][EtSO4]
Approximately 0.83 mol of 1-methylimidazole were diluted in 250 mL of toluene. Then, an equimolar amount of diethylsulfate was added dropwise at room temperature. The reaction was performed under nitrogen atmosphere and an ice bath was used to maintain temperatures below 313 K due to the highly exothermic reaction. The mixture was stirred for 24 h and after that, the two liquid phases were separated by decantation. The lower IL-rich phase was washed with toluene (3 x 200 mL). The excess toluene was removed by distillation under reduced pressures at about 323 K in a rotary evaporator (IKA RV 10 D S40). Finally the
Ac
ce pt
ed
M
with the hydroxyl groups of carbohydrates[19], while the cation’s structure also affects solvation properties, however with a smaller effect. Therefore, and despite the majority of them has still high purchase and production costs nowadays, which is a serious drawback to the scale-up of their applications, ILs can play an important role in the development of new biorefinery processes, either used as reaction media, catalysts, extraction solvents, or entrainers, among other functions. Solubility data of biomass-derived compounds such as carbohydrates in ionic liquids is an important issue in order to develop chemical or bio-processes involving these compounds. In the literature there are some data reported already[20-22], however it is far from being enough to have a good knowledge on phase equilibria of these systems. In most cases, these data are also unavailable in a sufficiently large temperature range, which turns their modeling a very difficult task. In this work, solubility of several monosaccharides, D-(+)-Glucose, D-(-)-Fructose, D-(+)-Xylose and D-(+)-Galactose, in two ILs has been measured in a temperature range from 288 K to 328 K using an isothermal method. The 1-ethyl-3methylimidazolium ethylsulfate, [emim][EtSO4] and the Aliquat®336 (a mixture of methyltrioctylammonium chloride and methyltridecylammonium chloride) were the two ionic liquids used as solvents. The apparent thermodynamic functions of dissolution, , and , have been also determined from a modified Van’t Hoff equation[23,24],and the data have been successfully correlated with local composition thermodynamic models: NRTL[25] and UNIQUAC[26].
an
us
cr
ip t
Page 4 of 22
resultant liquid was dried for 48 h under vacuum (P < 1×10-3 bar) at about 353 K, to remove the volatile compounds, including water. A pale yellow and slightly viscous liquid was obtained (yield greater than 90%), its water content (960 ppm) was measured by Karl Fisher titration in a Metrohm 870 KF Titrino plus titrator. The chemical structure was checked by 1H-NMR, with results very similar to those reported in the literature[27]. H (400 MHz, CDCl3) 1.239 (3H, t, NCH2CH3), 1.524 (3H, t, OCH2CH3), 3.981 (3H, s, NCH3), 4.047 (2H, q, OCH2), 4.284 (2H, q, NCH2), 7.504 (2H, s, C(4,5)H), 9.442(1H, s, C(2)H). The solubility measurements were performed using an isothermal technique based on the research group experience on solid-liquid equilibria[28]. First, to obtain saturated solutions, the carbohydrate was added, in a slight excess of the estimated solubility, to a specified amount of ionic liquid (10-15 g) in a 25 cm3 glass cell. Preliminary experiments were made to estimate the amounts to be added of each component. The equilibrium cells were equipped with a jacket where water from a thermostatic bath (JULABO F12-ED) was flowing to maintain the desired temperature, with a precision of ±0.1 K. The internal temperature of the cells was also measured with calibrated platinum probes, and its history was recorded by data acquisition software. Mixtures were magnetically stirred for at least 48 h, the time considered to be sufficient to reach the equilibrium, based on some preliminary tests and according to the experience of the team[28-30]. Once the solutions were saturated, the stirring was stopped and the excess solid gradually deposited in the bottom, resulting in a clear top liquid phase. Sampling was made in triplicate, using 2 mL syringes equipped with filter membranes (0.45 μm). Quantitative analysis was performed by reverse phase HPLC, using a LichroCART®250 mm x 4 mm column with Purospher®STAR RP-18e (5µm) as stationary phase. The eluent was water (100%) flowing at 0.7 mL∙min-1, and the detection was made by refractive index at 313 K using a HITACHI L-2490, with an injection volume of 20 µL of diluted sample. All samples were diluted conveniently and analysed three times each. Concentrations were calculated from calibration curves, previously prepared for each carbohydrate analyzed using the adopted chromatographic method. In the case of Aliquat®336, which is hydrophobic, the carbohydrate was re-extracted to an aqueous phase after precipitation with an antisolvent, dichloromethane[20]. This operation enables the quantification using the previous chromatographic method. The precision of the analytical method was estimated to be 0.88 %. 3. Results and discussion
3.1. Solubility data
Dissolving a solid in a liquid can be approximated, under some assumptions, to a melting process. Hence, solute melting properties are relevant to this study and they are presented in Table 2 for all the solutes used in this work. Another important issue is the water content of ionic liquids, which affects both the physical properties and their phase equilibria with other compounds. Thus, the water content of ILs used in this work was measured by Karl-Fisher titration, before and after the solubility measurements, as presented in Table 3.
Ac
ce pt
ed
M
an
us
cr
ip t
Page 5 of 22
2.3. Experimental determination of solubilities
In the case of 1-ethyl-3-methylimidazolium ethylsulfate, there is a small increase in the water content, which does not affect the properties neither the solubility since the ionic liquid is over 99.8% pure before and after measurements. On the other hand, Aliquat®336 shows a decrease in its water content; this is caused by the fact that the commercial product has 4.18 mass% of water and when submitted to temperatures around 328 K, there is a tendency to evaporate some water and its content decreases. However, these variations did not appear to be sufficiently large to produce effects on the solubility, as will be shown in results tendency (Table 4). As can be observed from Figures 1 and 2, the expected linear increase on solubility is verified in all cases, due to the relatively short range of temperatures studied. [emim][EtSO4] has shown a higher capacity to dissolve the carbohydrates than Aliquat®336. This can be explained, on one hand, with the higher ability of ethylsulfate anion to establish hydrogen bonds with the hydroxyl groups of carbohydrates. On the other hand, Aliquat®336 has a cation with large alkyl chains (hydrophobic), which reduces the affinity with polar substances such as monosaccharides. A solubility ranking could be drawn based on the melting properties of the solutes (independent of the ionic liquid), since all the solutes are chemically identical. Direct inspection of Tables 2 and 4 shows that D-(-)-Fructose is the most soluble carbohydrate, whereas D-(+)-Galactose is the less soluble. Their melting temperature and their melting enthalpy are related with this solubility ranking. The larger the melting temperature and enthalpy, the lower the solubility. Literature values for the studied binary systems are scarce. Solubility of glucose in Aliquat®336 at 308.15 K has been measured against the water content in the ionic liquid by Rosatella et al.[20] The data presented by the authors, for the same water content as the one used in this work, was also plotted in Figure 1 for comparison purposes. There is a significant deviation from the value obtained in this work, as observed in Figure 1. Actually, Aliquat®336 is a mixture of two ILs, the trioctylmethylammonium chloride and the tridecylmethylammonium chloride. Therefore, a possible reason for this deviation could be attributed to different ratios of C10 and C8 alkyl chains between ionic liquids used in the two works. A larger content of C10 chains could cause a decrease on glucose solubility due to an increase of the hydrophobic effect. Another possible reason could be related with the experimental techniques. However, the authors of the mentioned work did not present sufficient information, for example in respect to the time needed to reach the solid-liquid equilibrium, which is a very important issue in solubility experiments. Concerning the [emim][EtSO4], glucose solubility at 333 K has been published elsewhere[31]. The reported solubility can be considered in good agreement with the value obtained in this work, extrapolating to that temperature (see Figure 2). However, no information has been given about the water content in the ionic liquid, and this could be related with the deviation. A recent patent[32], also reported solubility values for glucose and fructose in [emim][EtSO4] at room temperature .Those values are far from those obtained in this work and [31], as shown in Figure 2, but a reliable comparison is not possible since the authors do not mention how the measurements were carried out neither the water content on the ionic liquid.
Ac
ce pt
ed
M
an
us
cr
ip t
Page 6 of 22
3.2. Thermodynamic functions of dissolution
Apparent thermodynamic functions of dissolution, , and can be obtained from the experimental solubility data. These thermodynamic functions are useful to provide information about the dissolution process to explain the molecular behaviour during the dissolution, and even to obtain energetic parameters for molecular-based equations of state to model the phase equilibria[33]. The approach proposed by Krug et al.[23], based on a modified Van´t Hoff equation, was used in this work,
us
where represents the mole fraction of the solute at saturated conditions (solubility), R is the ideal gas constant, T is the temperature in K, and is the harmonic average of mean of the experimental temperatures given by:
cr
where, Np is the number of experimental data points. The apparent standard enthalpy change of dissolution, , is taken from the slope of the linear plot To obtain the apparent standard change of Gibbs energy during the dissolution, , this approach uses the following expression:
ed
M
of
against
, Figure 3.
an
ce pt
where the constant k, is the yy axis intercept of the linear plot represented on Figure 3. Finally, through the definition of Gibbs energy, one can calculate the apparent standard entropy change of dissolution, . (4)
Figure 3 explains why not all the solubility data were used to calculate the thermodynamic functions for three systems. The data point neglected, the one which presented larger deviations to the linear behaviour, was in all cases at the lowest temperature of the range, 288 K. This may be related to the difficulty in efficient stirring due to the high viscosities of ionic liquids at this temperature. Table 5 presents the thermodynamic functions for each binary system studied. The obtained values for are positive in all cases, showing an endothermic behaviour in these dissolution processes. The lower this enthalpy, the easier it is for the system to overcome the energetic barrier to the dissolution. However, we cannot focus only in enthalpy to justify values of solubility. Entropy also plays an important role as an indicator if the configuration of solvent molecules after dissolution, is more or less favourable than the configuration in pure solvent, from an entropic point of view. Hence, positive values in all of the studied systems indicate more degrees of freedom to the ionic liquids when they
Ac
ip t
(1) (2) (3)
Page 7 of 22
are solvating molecules of carbohydrate than they have in a pure ionic liquid state. Finally, the positive values of show that, these dissolution processes are all non-spontaneous; requiring a continuous energy supply to occur.
3.3. Modeling solubilities through local composition models
M
In this work, two local composition models were used, to represent the solubility data, NRTL and UNIQUAC. Although these models were initially developed for non-electrolytes, they have been used with a good accuracy in many studies involving phase equilibria with ILs[24,34,35]. The solubility of the carbohydrates in the ILs, were determined from an expression based on the symmetric convention for the calculation of the activity coefficients, i.e., the pure liquid at the solution temperature as a standard state for the carbohydrate, together with their fusion entalpy, , melting temperature, , and heat capacity difference between pure liquid and solid states, . Table 2, shows these properties for the carbohydrates used in this work. For a general case, this equation can be derived through an idealized thermodynamic cycle[36] between the solid and liquid carbohydrate phase states, under some assumptions: (i) The solvent does not appear in the solid phase. (ii) The difference on the heat capacity between the pure liquid and the pure solid carbohydrate is considered temperature independent and equal to such difference at melting temperature, . Thus, the resulting expression to the calculation of the solubilities is:
an
us
cr
Where, is the mole fraction of solute (subscript 2) in the saturated solution, the activity coefficient of solute in the liquid phase, T, the temperature and R, the ideal gas constant. In order to compute solubility values, must be known. Its values are obtained from the models mentioned above, since they are explicit for activity coefficients. The binary parameters of NRTL and UNIQUAC models were obtained for each binary system fitting the model to the experimental data. Beyond the binary parameters, NRTL model has also a parameter, α, related with the nonrandomness of the mixture. When the number of available points for solubility data are reduced, this parameter has to be fixed and it can be made using some heuristics[37]. Thus, in this work a value of 0.3 was adopted for α. UNIQUAC model is characterized with structural parameters of volume and area, r and q, respectively, for each pure component. Available values in the literature were used in this work, except for Aliquat®336 which were estimated through an empiric correlation proposed by Domanska and Mazurowska, 2004[38] using the molar volume of the ionic liquid at 298.15 K (0.8865 g∙cm-3, measured experimentally on an Anton Paar DMA 4500M densimeter). Table 6, presents the structural parameters used for each compound and also the melting properties of the carbohydrates. Hence, the binary parameters have been obtained for both models and binary systems studied, using a Matlab™ framework, and the solubilities were calculated using these parameters to generate the activity coefficients in eq 5.
Ac
ce pt
ed
ip t
(5)
Page 8 of 22
Table 6 presents the obtained results for this thermodynamic modeling. Figures 1 and 2 shows qualitatively, the tendency of the experimental data and their correlation with the NRTL model. The average absolute relative deviation was determined for each case to compare and evaluate the accuracy.
(6) As was done when determining the thermodynamic functions of dissolution, the lower temperature solubility (288 K) was neglected in the three systems with [emim][EtSO4], while modeling the solubility data with NRTL and UNIQUAC equations, improving the fitness of the correlation with such models. Solubilities from UNIQUAC model were not plotted against experimental data in Figures 1 and 2 due to visualization issues, since their values are very close to those from NRTL model. Thus, the two models are able to correlate the solubility data with a good accuracy, showing average deviations smaller than 4 % in almost all the cases. There is no significant difference whether predicting solubility of these systems with NRTL or with UNIQUAC. 4. Conclusions
The solubility of four monosaccharides was measured over the temperature range of 288–328 K in two ionic liquids, using an isothermal technique combined with HPLC analysis. The results showed that solubility values for [emim][EtSO4] were higher than for Aliquat®336, which is explained by the fact of hydrofobicity and larger non-polar alkyl chains on the latter which have not much affinity with the hydroxyl groups of the monosaccharides. On the other hand the ethylsulfate anion has oxygen atoms giving extra hydrogen acceptor capability to the [emim][EtSO4], combined with its short alkyl chained cation make it a good IL to dissolve polar molecules such as the carbohydrates. It was also verified a kind of ranking on the solubility of the monosaccharides, independently of the IL: D-(-)Fructose > D-(+)-Xylose > D-(+)-Glucose > D-(+)-Galactose, which could be justified with the melting properties of these solutes, since they are very similar compounds. The apparent standard thermodynamic functions of dissolution were also determined in this work from a Van’t Hoff based equation, which allows to support that all of dissolution processes are non spontaneous, , requiring heat to occur, and favourable under an entropic point of view, . NRTL and UNIQUAC local composition models were applied to correlate the experimental data, the binary interaction parameters for both models were determined, and the models were used to compute the solubility of the studied binary systems, showing very good agreement with the experimental data, and no significant difference between the two thermodynamic models accuracy.
Ac
ce pt
ed
M
an
us
cr
ip t
Page 9 of 22
List of symbols f k Np P R T x fugacity yy axis intercept number of experimental points pressure (bar) ideal gas constant (8.314 J∙mol-1∙K-1) temperature (K) mole fraction
Acknowledgements Financial support for this work was in part provided by LSRE, for which the authors are thankful. A.C. acknowledges the financial support (Grant SFRH/BD/62105/2009) from Fundação para a Ciência e a Tecnologia (FCT, Portugal).
Ac
Abbreviations AARD Average absolute relative deviation HPLC High performance liquid chromatography NRTL Non Random Two Liquid ILs Room temperature ionic liquids UNIQUAC Universal Quasi Chemical
ce pt
Superscripts 0 standard state 0,L pure liquid state 0,S pure solid state exp experimental L liquid phase model calculated from the model
ed
M
Subscripts 2 dissol. fus hm i
related to the solute related to the dissolution process fusion harmonic mean related to the i compound
an
Page 10 of 22
us
Greek letters γ activity coefficient ΔCp heat capacity change (J∙mol-1∙K-1) ΔG Gibbs energy change (J∙mol-1) ΔH enthalpy change (J∙mol-1) ΔS entropy change (J∙mol-1∙K-1)
cr
ip t
References [1] T.P. Vispute, H. Zhang, A. Sanna, R. Xiao, G.W. Huber, Science 330 (2010) 12221227. [2] T.A. Werpy, G. Petersen, Top Value Added Chemicals From Biomass: I. Results of Screening for Potential Candidates from Sugars and Synthesis Gas, in, Pacific Northwest National Laboratory, 2004. [3] S.S.Y. Tan, D.R. MacFarlane, Top. Curr. Chem. 290 (2009) 311-339. [4] J.H. Clark, F.E.I. Deswarte, T.J. Farmer, Biofuel Bioprod Bior. 3 (2009) 72-90. [5] J.J. Bozell, Clean-Soil Air Water 36 (2008) 641-647. [6] B. Kamm, M. Kamm, Appl. Microbiol. Biotechnol. 64 (2004) 137-145. [7] B. Kamm, C. Hille, P. Schönicke, G. Dautzenberg, Biofuel Bioprod Bior. 4 (2010) 253-262. [8] L.R. Lynd, C.E. Wyman, T.U. Gerngross, Biotechnol. Prog. 15 (1999) 777-793. [9] Y.H.P. Zhang, J. Ind. Microbiol. Biotechnol. 35 (2008) 367-375. [10] J.F. Brennecke, E.J. Maginn, AIChE J. 47 (2001) 2384-2389. [11] K.R. Seddon, J. Chem. Technol. Biotechnol. 68 (1997) 351-356. [12] J.G. Huddleston, H.D. Willauer, R.P. Swatloski, A.E. Visser, R.D. Rogers, Chem. Commun. (1998) 1765-1766. [13] T. Welton, Chem. Rev. 99 (1999) 2071-2083. [14] T.C.R. Brennan, S. Datta, H.W. Blanch, B.A. Simmons, B.M. Holmes, Bionerg. Res. 3 (2010) 123-133. [15] F.R. Tao, H.L. Song, L.J. Chou, Carbohydr. Res. 346 (2011) 58-63. [16] Y. Fukaya, K. Hayashi, M. Wada, H. Ohno, Green Chem. 10 (2008) 44. [17] R.P. Swatloski, S.K. Spear, J.D. Holbrey, R.D. Rogers, J. Am. Chem. Soc. 124 (2002) 4974-4975. [18] I. Kilpelainen, H. Xie, A. King, M. Granstrom, S. Heikkinen, D.S. Argyropoulos, J. Agric. Food Chem. 55 (2007) 9142-9148. [19] G. Moyna, R.C. Remsing, Z.W. Liu, I. Sergeyev, J. Phys. Chem. B 112 (2008) 7363-7369. [20] A.A. Rosatella, L.C. Branco, C.A.M. Afonso, Green Chem. 11 (2009) 1406. [21] M.E. Zakrzewska, E. Bogel-Łukasik, R. Bogel-Łukasik, Energ. Fuel. 24 (2010) 737-745. [22] Q. Liu, M.H.A. Janssen, F. van Rantwijk, R.A. Sheldon, Green Chem. 7 (2005) 39. [23] R.R. Krug, W.G. Hunter, R.A. Grieger, J. Phys. Chem. 80 (1976) 2341-2351. [24] E.K. Panteli, E.K. Voutsas, J. Chem. Eng. Data 54 (2009) 812-818. [25] H. Renon, J.M. Prausnitz, AIChE J. 14 (1968) 135-144. [26] D.S. Abrams, J.M. Prausnitz, AIChE J. 21 (1975) 116-128. [27] J.D. Holbrey, W.M. Reichert, R.P. Swatloski, G.A. Broker, W.R. Pitner, K.R. Seddon, R.D. Rogers, Green Chem. 4 (2002) 407-413. [28] A.M. Peres, E.A. Macedo, Ind. Eng. Chem. Res. 36 (1997) 2816-2820. [29] F.L. Mota, A.R. Carneiro, A.J. Queimada, S.P. Pinho, E.A. Macedo, Eur. J. Pharm. Sci. 37 (2009) 499-507. [30] L.A. Ferreira, E.A. Macedo, S.P. Pinho, Chem. Eng. Sci. 59 (2004) 3117-3124. [31] H. Zhao, C.L. Jones, J.V. Cowins, Green Chem. 11 (2009) 1128-1138. [32] I.M. Al Nashef, M.H. Gaily, S.M. Al-zahrani, A.E. Abasaeed, Method for separating fructose and glucose, King Saud University (Riyadh, SA), United States, 2011. [33] E.K. Karakatsani, I.G. Economou, M.C. Kroon, M.D. Bermejo, C.J. Peters, G.-J. Witkamp, Phys. Chem. Chem. Phys. 10 (2008) 6160.
Ac
ce pt
ed
M
an
us
cr
ip t
Page 11 of 22
List of figures
Ac
ce pt
ed
Figure 3. Linear representation of ln x2 against (1/T-1/Thm). a) [emim][EtSO4] as solvent, b) Aliquat®336 as solvent.
M
Figure 2. Solubility of the carbohydrates in [emim][EtSO4] and NRTL correlation
an
Figure 1. Solubility of the carbohydrates in Aliquat®336 and NRTL correlation
us
cr
[34] F.M. Maia, O. Rodriguez, E.A. Macedo, Fluid Phase Equilib. 296 (2010) 184-191. [35] L.D. Simoni, Y. Lin, J.F. Brennecke, M.A. Stadtherr, Ind. Eng. Chem. Res. 47 (2008) 256-272. [36] J.M. Prausnitz, R.N. Lichtenthaler, E.G. Azevedo, Molecular Thermodynamics of Fluid-Phase Equilibria third ed., Prentice Hall, New Jersey, 1999. [37] G.M. Kontogeorgis, G.K. Folas, Thermodynamic Models for Industrial Applications, John Wiley & Sons Ltd, 2010. [38] U. Domanska, L. Mazurowska, Fluid Phase Equilib. 221 (2004) 73-82. [39] W. Feng, H. Van der Kooi, J. de Swaan Arons, Chem. Eng. Sci. 60 (2005) 617624. [40] O. Ferreira, E.A. Brignole, E.A. Macedo, Ind. Eng. Chem. Res. 42 (2003) 62126222. [41] S.O. Jonsdottir, S.A. Cooke, E.A. Macedo, Carbohydr. Res. 337 (2002) 1563-1571. [42] V.H. Alvarez, M. Aznar, J. Chin. Inst. Chem. Eng. 39 (2008) 353-360. [43] S.A. Cooke, S.O. Jonsdottir, P. Westh, Fluid Phase Equilib. 194 (2002) 947-956.
ip t
Page 12 of 22
Table(s)
Table 1. Sample description Initial Mass Purification Final Mole Analysis Method Fraction Purity Method Fraction Purity Alfa Aesar 0.96 none Karl-Fisher [emim][EtSO4]b Synthesis not measured Distillation 0.999 Titration , 1HNMR,13C-NMRc D-(+)-glucose Merck 0.99 none D-(-)-fructose Merck 0.99 none D-(+)-xylose Fluka 0.99 none D-(+)-galactose Fluka 0.99 none Source Chemical Name Aliquat®336a
Ac
ce pt
ed
M
an
Page 13 of 22
us
cr
a
Commercial name, a mixture of methyltrioctylammonium chloride and methyltridecylammonium chloride, with the first as predominant. b[emim][EtSO4] - 1-ethyl-3-methylimidazolium ethylsulfate c Nuclear magnetic ressonance
ip t
Table 2. Melting properties of the monosaccharides used in this work
Solute D-(+)-Glucose D-(-)-Fructose D-(+)-Xylose D-(+)-Galactose
M (g∙mol-1) 180.2 180.2 150.1 180.2
Tfus (K) 423.15 378.15 416.15 436.15
ΔfusH ΔfusCp -1 (kJ∙mol ) (J∙mol-1∙K-1) 32.4 26.0 31.7 43.8 120 99 97 139
Ref. [39,40] [39] [41] [41]
Ac
ce pt
ed
M
an
Page 14 of 22
us
cr
ip t
Table 3. Water contents of ionic liquids
Solvent [emim][EtSO4] Aliquat®336
mass % of Water Initial Final 0.1 0.2 4.2 3.8
Ac
ce pt
ed
M
an
Page 15 of 22
us
cr
ip t
Table 4. Experimental solubilities of carbohydrates in the ILs
1-ethyl-3-methylimidazolium ethylsulfate - [emim][EtSO4]
D-(+)-Glucose
T* / K Sol. mass % SD**
D-(-)-Fructose
T/K Sol. mass % SD
D-(+)-Xylose
T/K Sol. mass % SD
D-(+)-Galactose
T/K Sol. mass % SD
±0.01
D-(+)-Glucose
T/K Sol. mass % SD
±0.01
D-(-)-Fructose
T/K Sol. mass % SD
±0.02 Aliquat®336
T/K
cr
SD T/K
average standard relative uncertainty, u(Sol.)/Sol.
D-(+)-Xylose
Sol. mass %
us
12.6 0.9 14 2 15.4 0.8 16.70 0.04
ip t
298.3 308.6 318.3 329.2 4.2 6.3 7.4 8.6 ±0.003
288.2 298.2 308.2 318.6 328.3
10.4 15.5 19.1 21.8 23.7
0.1 0.2 0.6 0.7 0.4
288.2 298.2 308.0 318.4 329.3
25.7 29 33.8 37.4 43.5
0.4 2 0.3 0.1 0.1
288.1 298.1 308.3 318.0 328.5
15.0 0.1 20.5 0.1 23.0 0.2 24.51 0.03 27.41 0.03
288.8 298.2 308.5 318.2 328.2
4.0 6.88 9.5 11.8 14.0
0.1 0.03 0.1 0.1 0.2
±0.005
D-(+)-Galactose
Sol. mass % SD
average standard relative uncertainty, u(Sol.)/Sol.
±0.01
±0.02
M
297.9 307.8 318.4 327.9
9.0 10.4 11.6 12.5
0.2 0.3 0.3 0.1
298.2 307.7 318.3 328.2
15.4 18.4 21.8 25.2
0.5 0.1 0.2 0.4
298.5 308.8 318.9 330.0
0.2 0.2 0.4 0.2
Ac
ce pt
ed
*The uncertainty associated to the temperature measurement is: u(T)= ±0.01 K ** Standard deviation
an
±0.01
Page 16 of 22
Table 5. Apparent thermodynamic functions of dissolution
Solvent [emim][EtSO4] [emim][EtSO4] [emim][EtSO4] [emim][EtSO4] Aliquat®336 Aliquat®336 Aliquat®336 Aliquat®336
Solute D(+)Glucose D(-)Fructose D(+)Xylose D(+)Galactose D(+)Glucose D(-)Fructose D(+)Xylose D(+)Galactose
Np 4 5 4 4 4 4 4 4
Thm (K) 312.87 307.69 312.77 312.83 312.57 312.65 313.56 313.13
Ac
ce pt
ed
M
an
Page 17 of 22
us
cr
5.7 17.0
ip t
2.86 5.06
(kJ∙mol-1) 10.9 9.1 6.7 18.6 7.6 10.3
(kJ∙mol-1) 3.67 2.37 2.90 5.31 3.83 2.53
(J∙mol-1) 23.0 21.7 12.1 42.6 12.2 24.7 9.2 38.2
Table 6. Structural parameters used in UNIQUAC model
Compound [emim][EtSO4] Aliquat®336 D-(+)-Glucose D-(-)-Fructose D-(+)-Xylose D-(+)-Galactose
M (g.mol-1) 236.3 442 180.2 180.2 150.1 180.2
r
q
Ref. [42] [43] [41] [41] [41]
Ac
ce pt
ed
M
an
Page 18 of 22
us
cr
ip t
23.12 14.36 14.60 11.68 5.80 4.84 5.80 4.92 4.83 4.03 5.80 4.84
Table 7. NRTL and UNIQUAC modeling
Binary system Solvent Solute Np 4 5 4 4 4 4 4 4
NRTL (α=0,3) Δg12 (J/mol) -6337 -2190 -2241 -6765 1490 -461 -282 -7688 Δg21 (J/mol) 4454 -2057 -4730 2580 -4672 -2840 -6913 4090 AARD (%) 3.02 0.96 0.93 3.52 2.33 1.50 0.96 7.08
UNIQUAC Δu12 Δu21 (J/mol) (J/mol) -2327 2161 -2635 3533 -3622 8610 -1662 572 AARD (%) 3.21 1.14 0.91 3.64 2.43 1.42 0.85 7.28
glucose Aliquat®336 fructose xylose galactose
Ac
ce pt
ed
M
an
Page 19 of 22
us
cr
ip t
-2803 -2278 -3516 -2428 5333 3674 96187 2334
glucose fructose [emim][EtSO4] xylose galactose
Figure(s)
Figure 1
Ac
ce pt
ed
M
an
Page 20 of 22
us
cr
ip t
Figure 2
Ac
ce pt
ed
M
an
Page 21 of 22
us
cr
ip t
-0.500
ln x2
ln x2
-3.000 -2.50E-04
D-(+)-Glucose
-2.500 1.50E-04 3.50E-04 -2.50E-04
D-(+)-Galactose D-(+)-Glucose
-5.00E-05
-5.00E-05
cr
b)
ip t us
1/T-1/Thm D-(-)-Fructose D-(+)-Galactose
1.50E-04
3.50E-04
D-(+)-Xylose
1/T-1/Thm D-(-)-Fructose D-(+)-Xylose
Figure 3
Ac
ce pt
ed
M
an
a)
Page 22 of 22
