Supported Ionic Liquid Membranes for Separation of Lignin Aqueous Solutions

Lignin valorization is a key aspect to design sustainable management systems for lignocellulosic biomass. The successful implementation of bio-refineries requires high value added applications for the chemicals derived from lignin. Without effective separation processes, the achievement of this purpose is difficult. Supported ionic liquid membranes can play a relevant role in the separation and purification of lignocellulosic components. This work investigated different supported ionic liquid membranes for selective transport of two different types of technical lignins (Kraft lignin and lignosulphonate) and monosaccharides (xylose and glucose) in aqueous solution. Although five different membrane supports and nine ionic liquids were tested, only the system composed by [BMIM][DBP] as an ionic liquid and polytetrafluoroethylene (PTFE) as a membrane support allowed the selective transport of the tested solutes. The results obtained with this selective membrane demonstrated that lignins were more slowly transferred from the feed compartment to the stripping compartment through the membrane than the monosaccharides. A model was proposed to calculate the effective mass transfer constants of the solutes through the membrane (values in the range 0.5–2.0 × 10−3 m/h). Nevertheless, the stability of this identified selective membrane and its potential to be implemented in effective separation processes must be further analyzed.


Introduction
Among renewable raw materials, wood must be highlighted, because more effective, cost-competitive and sustainable alternatives have not been identified for some applications.Forest exploitation provides economic and social values from this natural resource and promotes a sustainable development chance for rural areas [1].The wood processing industrial sector obtains forest products such as lumber, engineered wood, and pulp.Traditional wood pulping has been focused on the fractionation of the main lignocellulosic components (cellulose, hemicellulose, and lignin), but paying special attention to the cellulosic fraction, which is the relevant one for the production of paper.In this framework, hemicellulose and lignin have been only employed for energy recovery by direct combustion [2].However, recent research interest is being targeted to hemicellulose and lignin as raw materials for renewable chemicals to replace those derived from petroleum.Therefore, the integral use of the lignocellulosic biomass must take into account the valorization of hemicellulose and lignin [3][4][5].For example, lignin must be considered the most promising renewable source to produce aromatic chemicals at a real industrial scale because of its structure (Figure 1) and its abundance in nature [6].In this new scenario, biorefineries have been introduced to provide an alternative to traditional petroleum refineries.A biorefinery is a facility that integrates the biomass conversion processes to produce bioenergy, biofuels, and bio-based chemicals from biomass [7].While the valorization of cellulose and hemicellulose has been successfully implemented in biorefineries [8], the optimal valorization of lignin remains as a great challenge to be solved.Enzymatic hydrolysis of cellulose and hemicellulose results in fermentable sugars, which can be easily transformed to biofuels (bioethanol) or precursors for production of valuable bio-based chemicals.Delignification is a necessary prerequisite for enzymatic hydrolysis, as lignin interferes the reaction and blocks the process [9,10].Therefore, lignin must be separated during the pretreatment of lignocellulosic biomass, which facilitates its posterior valorization.
The production of commercially available lignin-derived chemicals has been very limited until recent days: only dispersing and emulsifying agents obtained from lignosulphonates and plywood panels can be mentioned.This lack of commercial application can be justified by the heterogeneous structure of lignin, which, unlike cellulose or hemicellulose, is not formed by the systematic series of regular monomers.Despite this irregular and complex structure, research efforts have been applied to find the most suitable options for lignin conversion to valuable products [11][12][13][14][15][16].Although some possibilities have been identified for direct valorization of raw lignin, lignin depolymerization is a more promising route.On the one hand, aggressive unselective depolymerization to break C-C and C-O linkages results in aromatic compounds mixtures like benzene, toluene, xylene, and phenol [17,18].In addition, some short aliphatic (C1-C3) and, in less extent, longer cycloaliphatic (C6-C7) hydrocarbons can be obtained.On the other hand, highly selective depolymerization processes are based on the cleavage of only determined links.This way, products that are not easily produced by traditional petrochemical routes can be obtained, like substituted coniferols, aromatic polyols, or oxidized monomers [19].
However, classic fractionation processes for lignocellulosic biomass (Kraft process, Organosolv process, alkaline treatment, steam explosion ...) result in low-purity lignin, mostly because of the presence of impurities derived from cellulose and hemicellulose.Consequently, new research efforts must be applied to develop an efficient and selective process to obtain a purified lignin fraction.Membrane separation technologies have been implemented in biorefineries because they show very advantageous properties: no phase change, no heat requirements, low energy consumption, and In this new scenario, biorefineries have been introduced to provide an alternative to traditional petroleum refineries.A biorefinery is a facility that integrates the biomass conversion processes to produce bioenergy, biofuels, and bio-based chemicals from biomass [7].While the valorization of cellulose and hemicellulose has been successfully implemented in biorefineries [8], the optimal valorization of lignin remains as a great challenge to be solved.Enzymatic hydrolysis of cellulose and hemicellulose results in fermentable sugars, which can be easily transformed to biofuels (bioethanol) or precursors for production of valuable bio-based chemicals.Delignification is a necessary prerequisite for enzymatic hydrolysis, as lignin interferes the reaction and blocks the process [9,10].Therefore, lignin must be separated during the pretreatment of lignocellulosic biomass, which facilitates its posterior valorization.
The production of commercially available lignin-derived chemicals has been very limited until recent days: only dispersing and emulsifying agents obtained from lignosulphonates and plywood panels can be mentioned.This lack of commercial application can be justified by the heterogeneous structure of lignin, which, unlike cellulose or hemicellulose, is not formed by the systematic series of regular monomers.Despite this irregular and complex structure, research efforts have been applied to find the most suitable options for lignin conversion to valuable products [11][12][13][14][15][16].Although some possibilities have been identified for direct valorization of raw lignin, lignin depolymerization is a more promising route.On the one hand, aggressive unselective depolymerization to break C-C and C-O linkages results in aromatic compounds mixtures like benzene, toluene, xylene, and phenol [17,18].In addition, some short aliphatic (C1-C3) and, in less extent, longer cycloaliphatic (C6-C7) hydrocarbons can be obtained.On the other hand, highly selective depolymerization processes are based on the cleavage of only determined links.This way, products that are not easily produced by traditional petrochemical routes can be obtained, like substituted coniferols, aromatic polyols, or oxidized monomers [19].
However, classic fractionation processes for lignocellulosic biomass (Kraft process, Organosolv process, alkaline treatment, steam explosion ...) result in low-purity lignin, mostly because of the presence of impurities derived from cellulose and hemicellulose.Consequently, new research efforts must be applied to develop an efficient and selective process to obtain a purified lignin fraction.Membrane separation technologies have been implemented in biorefineries because they show very advantageous properties: no phase change, no heat requirements, low energy consumption, and compact and easily scalable design [20].Research works have investigated the potentiality of membrane techniques for lignin separation and purification.Lignin-rich liquors can be treated with ultrafiltration and nanofiltration membranes for purification of lignin and separation of other inorganic compounds [21][22][23][24][25][26].Other authors have applied membrane separations for different tasks aimed to lignin valorization, such as concentration of lignin solutions and elimination of lower molecular weight impurities [27], fractionation of lignin fragments according to their molecular weight [28,29], or separation of different lignin derivatives [30][31][32].
Since lignin is not easily solubilized in conventional solvents, the use of ionic liquids (ILs) for fractionation of lignocellulosic biomass has been deeply investigated.ILs are organic salts formed by high-volume organic cations and smaller organic or inorganic anions.Some of the most interesting ILs have melting points below 100 • C, so they are liquid at room temperature.These ILs present some common characteristics, like negligible vapor pressure or high thermal and mechanic stabilities [33].The physicochemical properties can be customized by an optimal combination of the most convenient cations and anions for each application.The viability of ILs for dissolution, separation, and recovery of the main components of lignocellulosic biomass has been investigated [34][35][36][37][38][39][40][41][42].Imidazolium based ILs (with different radicals joined to the central ring and combined with simple inorganic and more complex organic anions) have been deeply investigated for the selective dissolution of lignin, since they are not good solvents for cellulose or hemicellulose [43][44][45].The ILs based on 1,3-dialkyl-imidazolium have been object of most research works.However, the limited stability of these ILs in alkaline and oxidant media must be taken into account because posterior treatment of lignin can take place under these conditions [12].Therefore, alternative ILs non-based on imidazolium have been tested for lignin processing, paying special attention to ILs based on ammonium, phosphonium, pyridinium, and pyrrolidinium [46][47][48][49].Nevertheless, the high economic costs derived from the employment of this type of IL remains the main drawback for real-scale implementation in the biorefinery processes [50].
Supported liquid membranes (SLMs) consist of porous supports that have been impregnated with a specific solvent to get it imbedded in the pores.The solvent is kept there by capillary forces and forms a three-phase system, since it separates the feed and stripping phases [51,52].When this solvent is an IL, a supported ionic liquid membrane (SILM) is obtained (Figure 2) [53].SILMs require the existence of three simultaneous processes to be applied for effective separation of solutes: the extraction from the feed solution to the SILM, the diffusion through the SILM and the re-extraction from the SILM to the stripping solution.SILMs present some advantages over SLMs, mainly because of their improved stability, since the use of ILs reduces the solvent losses from the support by evaporation or dispersion in the feed and stripping phases [54].Moreover, ILs can provide very high specificity for the solutes to be separated and the small amount of IL required in a SILM can reduce the economic costs significantly.
Although scientific information about the use of membranes and ILs for lignocellulosic biomass fractionation and further processing is abundant, the integration of both tools as SILMs has not been deeply investigated.SILMs have been successfully applied for extraction of minor components that appear during vegetal biomass fractionation (for example, lipophilic compounds linked to resin, such as fatty acids or sterols) for analytical purposes [55].The potential of these systems for extraction and purification of lignin must be studied, since SILMs can be preferred over other technologies for extraction and purification of lignin.Since the solute extraction from the feed phase and the re-extraction to the stripping phase occur in a unique stage, very simple designs are possible, avoiding complex configurations or high energy requirements (separation can be carried out without heat or pressure application).The employment of highly porous materials to support the SILMs provides a very high interfacial area for mass transfer, which allows very compact equipment.Moreover, the coupling of the extraction and stripping results in mass transport without limitations due to the solubility limits [56].The small amount of IL required to implement a SILM allows the selection of non-traditional ILs, which can offer better permeability and specificity for lignin without compromising the chemical structure and physicochemical properties of lignin or interfere in the characteristics of the rest of the lignocellulosic components.Lastly, the most relevant disadvantage of the use of ILs (their high economic costs) is minimized when SILMs are implemented, since the total volume of ILs required is greatly reduced [51].A previous work began the analysis of the application of SILMs for lignin separation, but the preliminary results revealed that the selective transport of lignin was not easily obtained [57].The main objective of this work was a complete study of the potentiality of different SILMs for fractionation and separation of lignocellulosic biomass components, with special attention to lignin extraction and purification.The flow of two lignin types (Kraft lignin and lignosulphonates) from the feed compartment to the stripping one was characterized and compared with the flow of monosaccharides (glucose and xylose were selected because cellulose is made with repeated glucose units and xylose is the main sugar monomer in the structure of hemicellulose) to determine their potential for selective transport of the developed SILMs.

Chemicals and Materials
Nine different ILs were selected to prepare SILMs (Figure 3

). Six imidazolium-based ILs ([BMIM]MeSO4, [BMIM][DBP], [BMIM][OTf] and [HMIM][OTf] from Iolitec, and [EMIM]EtSO4 and [EMIM]
Ac from Sigma-Aldrich, Munich, Germany), two phosphonium-based ILs (CYPHOS 101 and CYPHOS 108 from Cytec, Woodland Park, NJ, USA) and a mixture of quaternary ammonium salts (Aliquat 336 from Sigma-Aldrich, Munich, Germany) were used as supplied.Kraft lignin (low sulfonate content), D-(+)-xylose (>99%) and D-(+)-glucose (>99.5%) were provided by Sigma-Aldrich, Munich, Germany, while sodium lignosulphonate was purchased from TCI Chemicals, Tokyo, Japan.The employed water was obtained by an Elix purification system (Millipore, Darmstadt, Germany).A previous work began the analysis of the application of SILMs for lignin separation, but the preliminary results revealed that the selective transport of lignin was not easily obtained [57].The main objective of this work was a complete study of the potentiality of different SILMs for fractionation and separation of lignocellulosic biomass components, with special attention to lignin extraction and purification.The flow of two lignin types (Kraft lignin and lignosulphonates) from the feed compartment to the stripping one was characterized and compared with the flow of monosaccharides (glucose and xylose were selected because cellulose is made with repeated glucose units and xylose is the main sugar monomer in the structure of hemicellulose) to determine their potential for selective transport of the developed SILMs.

Chemicals and Materials
Nine different ILs were selected to prepare SILMs (Figure 3

SILMs Preparation
The SILMs were prepared using the different polymeric membranes and ILs.Firstly, the corresponding membrane and IL were introduced in a vacuum oven (<35 mbar and 70 °C) within separate Petri dishes to remove the humidity, gases, and any other traces of volatile compounds.Later, the membrane was soaked in the IL, keeping the vacuum for 24 h to promote a proper impregnation by removal of air from the membrane pores.Finally, the liquid excess over the membrane surface was removed by allowing dripping overnight.This way, the SILM was ready to be employed.

Installation and Analytical Procedures
The experimental tests were carried out in a membrane cell designed for this specific purpose (Figure 4).The glass cell was composed of two identical compartments (volume lower than 150 mL each compartment), one for the feed solution and the other for the stripping one, which were separated by the SILM.The feed and stripping solutions were poured into the cell at the same time and the compartments were closed without inlet or outlet streams in the system.Samples were taken As membrane supports, membrane disc filters were employed.Five different polymeric materials were tested: PP (polypropylene) and PTFE (polytetrafluoroethylene) from Filter-Lab, Barcelona, Spain, PCTE (polycarbonate) from Sterlitech, Kent, WA, USA, PVDF (hydrophobic polyvinylidene fluoride), and HPVDF (hydrophilic polyvinylidene fluoride) from Millipore, Darmstadt, Germany.All the membranes had the same diameter (47 mm) and pore diameter (0.45 µm), except PCTE (0.40 µm).

SILMs Preparation
The SILMs were prepared using the different polymeric membranes and ILs.Firstly, the corresponding membrane and IL were introduced in a vacuum oven (<35 mbar and 70 • C) within separate Petri dishes to remove the humidity, gases, and any other traces of volatile compounds.Later, the membrane was soaked in the IL, keeping the vacuum for 24 h to promote a proper impregnation by removal of air from the membrane pores.Finally, the liquid excess over the membrane surface was removed by allowing dripping overnight.This way, the SILM was ready to be employed.

Installation and Analytical Procedures
The experimental tests were carried out in a membrane cell designed for this specific purpose (Figure 4).The glass cell was composed of two identical compartments (volume lower than 150 mL each compartment), one for the feed solution and the other for the stripping one, which were separated by the SILM.The feed and stripping solutions were poured into the cell at the same time and the compartments were closed without inlet or outlet streams in the system.Samples were taken at regular time intervals from both compartments.All the experiments were carried out at room temperature.It was decided to stop the experiments after the capture of enough samples to characterize the transport.Kraft lignin, lignosulphonate, glucose and xylose concentrations were determined by a ultraviolet-visible (UV-VIS) spectrophotometer DR 5000 (Hach, Düsseldorf, Germany), using a wavelength of 280 nm for Kraft lignin and lignosulphonate [58,59] and of 575 nm for monosaccharides, according to the dinitrosalicylic acid method for determination of reducing sugars [60,61].The calibration curves for the different solutes are compiled in Figure 5.  Kraft lignin, lignosulphonate, glucose and xylose concentrations were determined by a ultraviolet-visible (UV-VIS) spectrophotometer DR 5000 (Hach, Düsseldorf, Germany), using a wavelength of 280 nm for Kraft lignin and lignosulphonate [58,59] and of 575 nm for monosaccharides, according to the dinitrosalicylic acid method for determination of reducing sugars [60,61].The calibration curves for the different solutes are compiled in Figure 5.  Kraft lignin, lignosulphonate, glucose and xylose concentrations were determined by a ultraviolet-visible (UV-VIS) spectrophotometer DR 5000 (Hach, Düsseldorf, Germany), using a wavelength of 280 nm for Kraft lignin and lignosulphonate [58,59] and of 575 nm for monosaccharides, according to the dinitrosalicylic acid method for determination of reducing sugars [60,61].The calibration curves for the different solutes are compiled in Figure 5.The calibration curves showed good linearity in the defined intervals (until 0.05 g/L for Kraft lignin, 0.10 g/L for lignosulphonate and 5.00 g/L for monosaccharides).From the obtained lines, the corresponding absorptivity coefficients a λ were calculated according to the Beer-Lambert law: where A is the absorbance, b is the length of the path and C the solute concentration.The obtained absorptivity coefficients are compiled in Table 1.

Membrane Transport Characterization
The previously published paper with preliminary results presented a transport model for the selective transport of solutes through the prepared SILMs [57].However, the preliminary results revealed that the transport through the SILMs was not selective.Therefore, in this section the previously prepared model for the selective transport is summarized and the adaptations required to consider non-selective transport are added.
The amount of a solute that passes by selective transport through a SILM per unit of time and surface is called flux J and it is proportional to the gradient of concentration C between both solutions: where k is the proportionality constant that can be defined as the permeability.To model the evolution of the concentration of the solute in the feed compartment C F , a mass balance to the feed compartment can be applied: where V is the volume of the feed solution and A M the active surface of the SILM.The equation can be reorder and modified considering the gradient between the feed (F) and stripping (S) compartments, and the total mass balance with C 0 as initial feed concentration All the constant can be grouped into a unique constant K, called effective mass transfer constant: Integration can be applied and the evolution of the concentration in the feed compartment C F can be assessed: In a totally equivalent way, the evolution of the concentration in the stripping compartment C S can be obtained: For the case of non-selective transport, the corresponding mass balance can be formulated, just by replacement of the proportionality constant k by the permeate flux F P : This way, the evolution of the concentrations in both compartments can be derived, defining a new effective mass transfer constant K P :

Results and Discussion
During the preparation of the SILMs, some experimental drawbacks were identified.For example, when lignosulphonate solutions were mixed with Aliquat 336 or CYPHOS 101, unstable systems were obtained, since sticky precipitated solids appeared (Figure 6a,b).Under these conditions, a SILM with CYPHOS 101 in PVDF support did not result in effective results for Kraft lignin transport, because of the precipitation of solids on the membrane surface (Figure 6c).
Processes 2018, 8, x FOR PEER REVIEW 8 of 18 For the case of non-selective transport, the corresponding mass balance can be formulated, just by replacement of the proportionality constant k by the permeate flux FP: This way, the evolution of the concentrations in both compartments can be derived, defining a new effective mass transfer constant KP:

Results and Discussion
During the preparation of the SILMs, some experimental drawbacks were identified.For example, when lignosulphonate solutions were mixed with Aliquat 336 or CYPHOS 101, unstable systems were obtained, since sticky precipitated solids appeared (Figure 6a,b).Under these conditions, a SILM with CYPHOS 101 in PVDF support did not result in effective results for Kraft lignin transport, because of the precipitation of solids on the membrane surface (Figure 6c).Another identified problematic issue was the employment of [EMIM]Ac for the preparation of SILMs.After the contact between [EMIM]Ac and the PCTE support for 24 h at 70 °C in the vacuum oven, the membrane was totally dissolved in the IL (Figure 7a).Even when the temperature in the vacuum oven was reduced to 30 °C, the PCTE membrane suffered severe degradation after contact with this IL (Figure 7b).Other membranes, like PVDF or HPVDF, were highly modified (color changed) after impregnation with [EMIM]Ac at 70 °C (Figure 7c), but a decreased value of the temperature in the vacuum oven (30 °C) resulted in stable SILMs.Another identified problematic issue was the employment of [EMIM]Ac for the preparation of SILMs.After the contact between [EMIM]Ac and the PCTE support for 24 h at 70 • C in the vacuum oven, the membrane was totally dissolved in the IL (Figure 7a).Even when the temperature in the vacuum oven was reduced to 30 • C, the PCTE membrane suffered severe degradation after contact with this IL (Figure 7b).Other membranes, like PVDF or HPVDF, were highly modified (color changed) after impregnation with [EMIM]Ac at 70 • C (Figure 7c), but a decreased value of the temperature in the vacuum oven (30  The only virgin membrane that was permeable to Kraft lignin solution and allowed the transport of the solute from the feed to the stripping compartment was the HPVDF membrane.The evolution of the normalized concentration of Kraft lignin in the stripping compartment of the cell with respect to time is shown in Figure 8, where the transport through the virgin membrane was compared to the transport through some SILMs based on HPVDF.As can be observed in the graph, the virgin membrane showed the fastest transport, while the performance of the permeable SILMs was very similar and only minor differences could be identified among the different ILs.Higher permeation of Kraft lignin occurred during the initial experimental phase, when the solute concentration gradient was maximum between both compartments of the cell.Then, the transport was slowed down due to reduced gradient for concentrations getting closer to the equilibrium.These experimental results fitted satisfactorily with the proposed model for nonselective transport through permeable membranes.The corresponding equations were employed for direct assessment of the effective mass transfer constants KP (values compiled in Table 2).The R 2 values of the fitting of the experimental results ranged from 0.936 to 0.982.As expected from the experimental results in Figure 8, the maximal KP value corresponded to the virgin HPVDF membrane (0.0334 h −1 ), with lower values for the HPVDF-based SILMs.Taking into account the effective membrane area in the cell (14.6 cm 2 ) and the volume of each cell compartment (120 mL), the The only virgin membrane that was permeable to Kraft lignin solution and allowed the transport of the solute from the feed to the stripping compartment was the HPVDF membrane.The evolution of the normalized concentration of Kraft lignin in the stripping compartment of the cell with respect to time is shown in Figure 8, where the transport through the virgin membrane was compared to the transport through some SILMs based on HPVDF.As can be observed in the graph, the virgin membrane showed the fastest transport, while the performance of the permeable SILMs was very similar and only minor differences could be identified among the different ILs.The only virgin membrane that was permeable to Kraft lignin solution and allowed the transport of the solute from the feed to the stripping compartment was the HPVDF membrane.The evolution of the normalized concentration of Kraft lignin in the stripping compartment of the cell with respect to time is shown in Figure 8, where the transport through the virgin membrane was compared to the transport through some SILMs based on HPVDF.As can be observed in the graph, the virgin membrane showed the fastest transport, while the performance of the permeable SILMs was very similar and only minor differences could be identified among the different ILs.Higher permeation of Kraft lignin occurred during the initial experimental phase, when the solute concentration gradient was maximum between both compartments of the cell.Then, the transport was slowed down due to reduced gradient for concentrations getting closer to the equilibrium.These experimental results fitted satisfactorily with the proposed model for nonselective transport through permeable membranes.The corresponding equations were employed for direct assessment of the effective mass transfer constants KP (values compiled in Table 2).The R 2 values of the fitting of the experimental results ranged from 0.936 to 0.982.As expected from the experimental results in Figure 8, the maximal KP value corresponded to the virgin HPVDF membrane (0.0334 h −1 ), with lower values for the HPVDF-based SILMs.Taking into account the effective membrane area in the cell (14.6 cm 2 ) and the volume of each cell compartment (120 mL), the Higher permeation of Kraft lignin occurred during the initial experimental phase, when the solute concentration gradient was maximum between both compartments of the cell.Then, the transport was slowed down due to reduced gradient for concentrations getting closer to the equilibrium.These experimental results fitted satisfactorily with the proposed model for non-selective transport through permeable membranes.The corresponding equations were employed for direct assessment of the effective mass transfer constants K P (values compiled in Table 2).The R 2 values of the fitting of the experimental results ranged from 0.936 to 0.982.As expected from the experimental results in Figure 8, the maximal K P value corresponded to the virgin HPVDF membrane (0.0334 h −1 ), with lower values for the HPVDF-based SILMs.Taking into account the effective membrane area in the cell (14.6 cm 2 ) and the volume of each cell compartment (120 mL), the corresponding permeate flux F P were calculated from the effective mass transfer constants K P (Table 2).
Table 2. Values of effective mass transfer constants K P , permeate fluxes F P , and resistances (attributable to the membrane support and the IL) for virgin HPVDF membranes and SILMs based on these membranes (viscosity and density values of the ILs are included).The maximal permeate flux of the virgin membrane can be directly justified taking into consideration the resistances in the series model, since the ILs supported in the membranes provide an additional resistance to the permeate transport: The value of the membrane resistance R MEMB was calculated from the F P value of the virgin membrane and the resistance values attributable to the ILs were derived from the corresponding F P values once R MEMB was known.According to the figures in Table 2, the R IL values were lower than the R MEMB value for all the SILMs but the ones with [BMIM][DBP] and [HMIM] [OTf].The high resistance exhibited by [BMIM][DBP] can be directly attributed to its high viscosity, but the value corresponding to [HMIM][OTf] was not easily related to the viscosity or density of the IL because other more viscous or denser ILs showed lower resistance [62][63][64][65][66][67].
Apart from HPVDF, the rest of the virgin membranes were impermeable to water or aqueous solutions.The SILMs prepared with these impermeable membranes were initially tested with Kraft lignin and lignosulphonate solutions.The results obtained with some of these SILMs are graphed in Figure 9, where the evolution of the solute concentration in both compartments is presented.The differences among the experimental data for SILMs with different ILs and solutes were not obvious and very similar results were obtained for all the SILMs that exhibited solute transport.When additional experiments were carried out with monosaccharides (glucose and xylose) as solutes, the performance of the tested SILMs was comparable and the evolution of the monosaccharides concentration in the cell compartments followed a similar trend to the identified one for Kraft lignin or lignosulphonates (Figure 10).A further analysis of the experimental results shown in Figures 8-10 suggested non-selective transport through the SILMs, as the different investigated membranes, ILs or solutes had not relevant influence on the performance of the SILMs.For example, the performance of the SILMs based on impermeable membranes could be compared to the SILMs based on permeable HPVDF.The lack of selectivity of the studied SILMs was previously reported during the preliminary works of this research group [57].Those results demonstrated that the performance of the freshly prepared SILMs was totally similar to SILMs used repeatedly in consecutive cycles, even when the IL had been apparently lost from the membrane support.The mass loss of the SILMs once the first cycle was finished corresponded to the amount of IL previously immobilized in the membrane.Scanning electron microscope (SEM) images confirmed this fact, since the appearance of the membrane before the second cycle was much more similar to the virgin membrane than the SILM before the first cycle (Figure 11).Therefore, the transport through the SILMs seemed to be more closely related to the modification of the membrane supports than the presence of ILs in the membranes.Further experiments were carried out to understand the transport through the SILMs and tests to analyze the permeability of the SILMs to water and aqueous solutions were proposed.While the feed compartment was filled with water or an aqueous solution of Kraft lignin, the  The lack of selectivity of the studied SILMs was previously reported during the preliminary works of this research group [57].Those results demonstrated that the performance of the freshly prepared SILMs was totally similar to SILMs used repeatedly in consecutive cycles, even when the IL had been apparently lost from the membrane support.The mass loss of the SILMs once the first cycle was finished corresponded to the amount of IL previously immobilized in the membrane.Scanning electron microscope (SEM) images confirmed this fact, since the appearance of the membrane before the second cycle was much more similar to the virgin membrane than the SILM before the first cycle (Figure 11).Therefore, the transport through the SILMs seemed to be more closely related to the modification of the membrane supports than the presence of ILs in the membranes.Further experiments were carried out to understand the transport through the SILMs and tests to analyze the permeability of the SILMs to water and aqueous solutions were proposed.While the feed compartment was filled with water or an aqueous solution of Kraft lignin, the The lack of selectivity of the studied SILMs was previously reported during the preliminary works of this research group [57].Those results demonstrated that the performance of the freshly prepared SILMs was totally similar to SILMs used repeatedly in consecutive cycles, even when the IL had been apparently lost from the membrane support.The mass loss of the SILMs once the first cycle was finished corresponded to the amount of IL previously immobilized in the membrane.Scanning electron microscope (SEM) images confirmed this fact, since the appearance of the membrane before the second cycle was much more similar to the virgin membrane than the SILM before the first cycle (Figure 11).Therefore, the transport through the SILMs seemed to be more closely related to the modification of the membrane supports than the presence of ILs in the membranes.Further experiments were carried out to understand the transport through the SILMs and tests to analyze the permeability of the SILMs to water and aqueous solutions were proposed.While the feed compartment was filled with water or an aqueous solution of Kraft lignin, the stripping compartment remained empty.Surprisingly, nearly all the tested SILMs that exhibited solute transport were identified as permeable and the only SILM that maintained its impermeability was that one based on [BMIM][DBP] supported in PTFE (Figure 12).stripping compartment remained empty.Surprisingly, nearly all the tested SILMs that exhibited solute transport were identified as permeable and the only SILM that maintained its impermeability was that one based on [BMIM][DBP] supported in PTFE (Figure 12).As a summary, Table 3 compiles all the calculated KP values of the tested SILMs.Moreover, the same table identifies the SILMs that were not tested because of the technical drawbacks during their preparation and the SILMS that did not show any solute transport.The reported KP values ranged between 0.0139 and 0.0284 h −1 for [BMIM][DBP] in PP and [BMIM]MeSO4 in PCTE, respectively.These figures fell in the range of the SILMS prepared with HPVDF (0.0141-0.0263 h −1 ).This concordance once again confirmed the common transport mechanisms between both types of SILMs: convective transport of the solute as consequence of the permeation of the solutions.stripping compartment remained empty.Surprisingly, nearly all the tested SILMs that exhibited solute transport were identified as permeable and the only SILM that maintained its impermeability was that one based on [BMIM][DBP] supported in PTFE (Figure 12).As a summary, Table 3 compiles all the calculated KP values of the tested SILMs.Moreover, the same table identifies the SILMs that were not tested because of the technical drawbacks during their preparation and the SILMS that did not show any solute transport.The reported KP values ranged between 0.0139 and 0.0284 h −1 for [BMIM][DBP] in PP and [BMIM]MeSO4 in PCTE, respectively.These figures fell in the range of the SILMS prepared with HPVDF (0.0141-0.0263 h −1 ).This concordance once again confirmed the common transport mechanisms between both types of SILMs: convective transport of the solute as consequence of the permeation of the solutions.As a summary, Table 3 compiles all the calculated K P values of the tested SILMs.Moreover, the same table identifies the SILMs that were not tested because of the technical drawbacks during their preparation and the SILMS that did not show any solute transport.The reported K P values ranged between 0.0139 and 0.0284 h −1 for [BMIM][DBP] in PP and [BMIM]MeSO 4 in PCTE, respectively.These figures fell in the range of the SILMS prepared with HPVDF (0.0141-0.0263 h −1 ).This concordance once again confirmed the common transport mechanisms between both types of SILMs: convective transport of the solute as of the permeation of the solutions.
The SILM with [BMIM][DBP] supported in PTFE membrane, which was the only one that showed impermeability to the tested solutions, was the only case that exhibited selective solute transport.As can be observed in Figure 13, the evolution of the concentration in the stripping compartment was very different for the two graphed solutes: the transport of glucose is much faster than the transport of lignosulphonate.While in the case of glucose, 70 h was time enough to achieve a concentration that could be considered an equilibrium situation; after 100 h the normalized concentration of lignosulphonate in the stripping compartment was still below 0.38.least easily transported solute (0.67 × 10 −3 m/h), while glucose was the most easily transported one (1.86 × 10 −3 m/h).Nevertheless, the stability of this identified selective SILM and its applicability to separation processes must be investigated more deeply and further work will be carried out.

Processes 2018, 8 ,
x FOR PEER REVIEW 4 of 18 of the use of ILs (their high economic costs) is minimized when SILMs are implemented, since the total volume of ILs required is greatly reduced[51].

Figure 2 .
Figure 2. Schematic representation of transport through a SILM (Supported Ionic Liquid Membrane).

Figure 2 .
Figure 2. Schematic representation of transport through a SILM (Supported Ionic Liquid Membrane).
at regular time intervals from both compartments.All the experiments were carried out at room temperature.It was decided to stop the experiments after the capture of enough samples to characterize the transport.

Figure 4 .
Figure 4. Image of the experimental cell selected for testing SILMs.

Figure 5 .
Figure 5. Calibration curves for the determination by ultraviolet-visible (UV-VIS) spectrophotometry of the concentration of the different tested solutes.

Figure 4 .
Figure 4. Image of the experimental cell selected for testing SILMs.
at regular time intervals from both compartments.All the experiments were carried out at room temperature.It was decided to stop the experiments after the capture of enough samples to characterize the transport.

Figure 4 .
Figure 4. Image of the experimental cell selected for testing SILMs.

Figure 5 .
Figure 5. Calibration curves for the determination by ultraviolet-visible (UV-VIS) spectrophotometry of the concentration of the different tested solutes.

Figure 5 .
Figure 5. Calibration curves for the determination by ultraviolet-visible (UV-VIS) spectrophotometry of the concentration of the different tested solutes.

Figure 8 .
Figure 8. Evolution of the concentration of Kraft lignin in the stripping compartment with virgin HPVDF membrane and three SILMs supported in HPVDF.

Figure 8 .
Figure 8. Evolution of the concentration of Kraft lignin in the stripping compartment with virgin HPVDF membrane and three SILMs supported in HPVDF.

Figure 8 .
Figure 8. Evolution of the concentration of Kraft lignin in the stripping compartment with virgin HPVDF membrane and three SILMs supported in HPVDF.

Figure 9 .
Figure 9. Evolution of the concentration of Kraft lignin (KL) or lignosulphonate (LS) in the feed and stripping compartments with three different SILMs supported in PCTE.

Figure 10 .
Figure 10.Evolution of the concentration of lignosulphonate (LS), glucose (G), or xylose (X) in the feed compartment with three different SILMs.

[Figure 9 . 18 Figure 9 .
Figure 9. Evolution of the concentration of Kraft lignin (KL) or lignosulphonate (LS) in the feed and stripping compartments with three different SILMs supported in PCTE.

Figure 10 .
Figure 10.Evolution of the concentration of lignosulphonate (LS), glucose (G), or xylose (X) in the feed compartment with three different SILMs.

[Figure 10 .
Figure 10.Evolution of the concentration of lignosulphonate (LS), glucose (G), or xylose (X) in the feed compartment with three different SILMs.

Figure 11 .
Figure 11.Scanning electron microscope (SEM) microphotographs of virgin PVDF membrane (a), SILM with CYPHOS 108 in PVDF before the transport test (b) and SILM with CYPHOS 108 in PVDF after the transport test (c).

Figure 11 .
Figure 11.Scanning electron microscope (SEM) microphotographs of virgin PVDF membrane (a), SILM with CYPHOS 108 in PVDF before the transport test (b) and SILM with CYPHOS 108 in PVDF after the transport test (c).

Figure 11 .
Figure 11.Scanning electron microscope (SEM) microphotographs of virgin PVDF membrane (a), SILM with CYPHOS 108 in PVDF before the transport test (b) and SILM with CYPHOS 108 in PVDF after the transport test (c).

Table 1 .
The absorptivity coefficients for determination by UV-VIS spectrophotometry of the concentration of the different tested solutes.