Pervaporative Separation of Ethylene Glycol/Water Mixtures by Using Cross-linked Chitosan Membranes

Transcription

1 Ind. Eng. Chem. Res. 2007, 46, Pervaporative Separation of Ethylene Glycol/Water Mixtures by Using Cross-linked Chitosan Membranes P. Srinivasa Rao,*, S. Sridhar, Ming Yen Wey, and A. Krishnaiah EnVironmental Restoration and Disaster Reduction Research Center, Department of EnVironmental Engineering, National Chung-Hsing UniVersity, Taichung 402, Taiwan, Membrane Separations Group, Chemical Engineering DiVision, Indian Institute of Chemical Technology, Hyderabad , India, and Department of Chemistry, Sri Venkateswara UniVersity, Tirupati , India Chitosan (CS) is one of the most widely used pervaporation membranes in the world today. A novel method for cross-linking CS membranes using phosphoric acid in alcohol baths was investigated in this study for the separation of ethylene glycol (EG)/water mixtures. The cross-linked membranes were subjected to sorption studies to evaluate the extent of interaction and degree of swelling in pure as well as binary mixtures of the two liquids. In order to gain a more detailed picture of the molecular transport phenomenon, we have performed sorption gravimetric experiments at 30 C to compute diffusion, swelling, sorption, and permeability coefficients of phosphorylated chitosan (P-CS) membranes in the presence of ethylene glycol and water. The effects of experimental parameters such as feed composition, membrane thickness, and permeate pressure on separation performance were determined. The membranes were characterized before and after cross-linking by Fourier transform infrared (FTIR) analysis and thermogravimetric analysis (TGA) to verify cross-linking and to observe the thermal degradation range of the polymer. The membrane appears to have a good potential for breaking the boiling mixture of ethylene glycol/water since a moderately good selectivity of 234 was obtained at a reasonable flux of 0.37 kg/m 2 h. The separation factor was found to improve with decreasing feedwater concentration whereas flux decreased correspondingly. Increasing the membrane thickness decreased the flux but had a less profound effect on the separation factor. Higher permeate pressure caused a reduction in flux and an increase in selectivity. 1. Introduction The pervaporation (PV) technique, which is economical, safe, and ecofriendly, is considered to be a promising alternative to conventional energy intensive technologies such as extractive or azeotropic distillation in liquid mixture separation. This method has been useful particularly for the separation of azeotropes, close boiling mixtures, isomers, and heat sensitive and hazardous compounds. 1 Hydrophilic membranes have been extensively used in PV based dehydration of organic solvents. 2-4 It is well accepted that the permeable components are transported in dense membranes according to the solution-diffusion process 5 in which the component sorption in the membrane plays a major role. Ethylene glycol (EG) is an important chemical widely used for nonvolatile antifreeze and coolant as well as an intermediate in the manufacture of polyesters. 6 At present, the main commercial route of ethylene glycol production is the direct oxidation of ethylene to ethylene oxide followed by the hydrolysis of ethylene oxide. Although ethylene glycol and water do not form an azeotrope over the entire composition range, the separation of water from ethylene glycol by distillation has proved to be costly because high-pressure steam is required for the reboiler due to the high boiling point of ethylene glycol (198 C). In fact, ethylene glycol-water separation by distillation is ranked as the eighth most energy intensive distillation operation in the chemical process industry. 7 Therefore distillation * To whom correspondence should be addressed. popurishrinu@gmail.com. Fax: Telephone: National Chung-Hsing University. Indian Institute of Chemical Technology. Sri Venkateswara University. and dehydration columns can be substituted by pervaporation processes which have been known to be energy saving and economical processes. Extensive studies have been carried out to investigate pervaporation for the separation of various azeotropic and close boiling mixtures. However, not as much effort has been directed at pervaporation application for dehydrating nonazeotropic systems such as ethylene glycol/water systems. A very few reports are available in the literature for the PV separation of EG/water mixtures. Feng et al. 8 reported that neutralized chitosan/polysulfone composite membranes had good separation performance for the selective removal of water from aqueous ethylene glycol solutions by pervaporation. They showed that pervaporation from chitosan membranes worked best (0.3 kg/m 2 h of water flux) for a feed with low water concentrations (10 wt %). From this result, they demonstrated the potential of pervaporation as an alternative process to conventional distillation for the separation of water/ethylene glycol mixtures. Chen and Chen 9 investigated that pervaporation of water/eg solutions with cross-linked poly(vinyl alcohol) (PVA) membranes was affected by membrane preparation conditions. They showed that the PVA membrane exhibited a separation factor of 231 and a permeation flux of kg/m 2 h for 82.5 wt % ethylene glycol at 80 C. Reid et al. 10 compared several methods including pervaporation to separate glycol and water from liquids. In order to recover the glycol from liquid coolants, the mechanical separation of solid and light weight materials is required for pretreatment. Texaco 11 holds a patent for the dehydration of glycol using a cross-linked PVApolysulfone composite membrane and reported enhancement of water concentration in permeate to 99.6 wt % with a permeate flux of kg/m 2 h for an 85 wt % ethylene glycol concentration at 80 C. Jehle et al. 12 used several GFT composite /ie061268n CCC: $ American Chemical Society Published on Web 02/27/2007

2 2156 Ind. Eng. Chem. Res., Vol. 46, No. 7, 2007 membranes for pervaporation. A GFT1001 membrane showed the highest selectivity and the lowest permeate flux among the membranes tested. GFT1510 membranes showed a high permeate flux of 1.7 kg/m 2 h and good selectivity with 98.5 wt % water concentration in permeate. Nam and Lee prepared ionically cross-linked chitosan composite membranes for EG/ water separation. They showed a very high water concentration in permeate (99.5 wt %) and a permeate flux of 1.13 kg/m 2 h measured at 80 wt % feed EG concentration. 13 Burshe et al. 14 prepared the interpenetrating polymer network membranes of PAA-PVA for the dehydration of EG and concluded that these membranes are suitable for the dehydration of EG/water mixtures in the range of water concentration wt % in the feed. Sekulic et al. 15 synthesized nonselective microporous titania membrane for the separation of p-dioxane/water and EG/ water mixtures. This membrane showed a remarkably high selectivity in the separation of EG/water due to the formation of a hydrogen-bonded network of ethylene glycol in the micropores, which blocks the transport of EG, while water can still permeate through. Chitosan (CS) or poly(d-glucosamine), a natural biopolymer obtained by the deacetylation of chitin, has many inherent characteristics such as hydrophilicity, biocompatibility, antibacterial properties, and a remarkable affinity for many substances. Chitosan can be used to develop blends with PVA, since it can form highly selective and permeable blends with another hydrophilic polymer like PVA. Because of the widespread availability of purified chitosans, coupled with its ability to be used in a variety of forms including powder, gel, film, fiber, solution, and membrane, it is used in many commercial applications. Chitosan has many biomedical applications, but in dehydration studies on aqueous-organic mixtures, its hydroxyl and amino groups can be modified easily. Chitosan swells in water, and hence, it can be cross-linked by a suitable agent to improve its mechanical strength and selectivity during PV experiments. Dense CS membranes that are ionically crosslinked by sulfuric acid showed the highest selectivity with lower permeation rates. Lee and co-workers investigated the pervaporation performance for water/alcohol mixtures using chitosan and modified chitosan membranes 20 and recently reported pervaporation of water/alcohol mixtures through chitosan composite membranes. 21 Chitosan can be cross-linked by various methods to increase separation characteristics. Chitosan was cross-linked by glutaraldehyde, 17 organic acid with several functional sites, 22 metals, 19 and inorganic acids. 23 Cross-linked chitosan had strong mechanical properties 24 and revealed moderate membrane characteristics and excellent biocompatibility. 25,26 In the present study, CS was cross-linked with phosphoric acid to overcome the drawbacks of the former for dehydrating ethylene glycol/water mixtures. The work also explores the effect of varying water concentration in the binary feed mixture on membrane flux and selectivity. Characterization by Fourier transform infrared (FTIR) analysis, wide-angle X-ray diffraction (WAXD), thermogravimetric analysis (TGA), tensile strength measurements, and sorption studies has been used to explain interaction mechanisms between the polymer and PV results. Effects of membrane thickness and permeate pressure on separation performance were evaluated. 2. Experimental Materials. Chitosan used for blending was purchased from Aldrich chemical Co., USA, having an average molecular weight of The degree of deacetylation was found to be 84%. Figure 1. Structural representation of the ionic cross-linking of chitosan with phosphoric acid. The cross-linking agent phosphoric acid and the reagents isopropyl alcohol (IPA) and hydrochloric acid were purchased from S.D. Fine Chemicals, Mumbai, India. Solvent ethylene glycol (EG) of purity 99% used in the study was of a reagent grade sample and was purchased from Ranbaxy fine chemicals Ltd, New Delhi, India. Double distilled water of conductivity 0.02 S/cm was used for the preparation of feed solutions. Membrane Preparation. Dense chitosan membranes were prepared by the solution casting technique. A 3% (w/v) chitosan solution was prepared by dissolving chitosan flakes in an aqueous acetic acid (2 v/v %) solution. The polymer solution was then filtered to remove a trace amount of nondissolved residual solids. The chitosan solution was then cast onto a horizontally positioned glass plate under ambient conditions, and solvent was allowed to evaporate slowly at room temperature for 24 h. The resultant membrane was removed from the glass plate and modified chemically with phosphoric acid for 4 h in a mixture of IPA (400 ml), water (100 ml), and a phosphoric acid (15 ml) bath and then vacuum-dried for 6hat 60 C. The thickness of the entire membrane was determined as a mean value from at least ten single measurements taken around the circumference of a membrane disk using a peacock dial thickness gauge (Model G, Ozaki Mfg. Co., Japan) with an accuracy of (5 µm. Figure 1 is a structural representation of the reaction that takes place in the preparation of an ionically cross-linked CS membrane. Pervaporation Procedure. Experiments were carried out on a 100 ml batch level with an indigenously constructed pervaporation manifold (Figure 2) operated at a vacuum as low as 0.25 mmhg in the permeate line. The membrane area in the pervaporation cell assembly (Figure 2b) was approximately 20 cm 2. The experimental procedure is described in detail elsewhere. 27 Permeate was collected for a duration of 6-8 h. Tests were carried out at room temperature (30 ( 2 C) and repeated twice using fresh feed solution for reproducibility. The collected permeate was weighed after allowing it to attain room temperature in a Sartorius electronic balance (accuracy: 10-4 g) to

3 Ind. Eng. Chem. Res., Vol. 46, No. 7, Figure 2. Schematic of laboratory pervaporation unit. determine the flux, and it was then analyzed by gas chromatography to evaluate the membrane selectivity. Flux and Selectivity Equations. In pervaporation, the flux J of a given species, say faster permeating component i of a binary liquid mixture comprised of i (water) and j (EG), is given by J i ) W i At Where, W i represents the mass of water in the permeate (kg), A is the membrane area (m 2 ), and t represents the evaluation time (h). In the present study, though different membrane thicknesses were utilized, the flux has been normalized and reported for a thickness of 10 µm. 28 The membrane selectivity is the ratio of permeable coefficients and can be calculated from the respective concentrations in feed and permeate as follows: y(1 - x) R) x(1 - y) Where, x and y represent the feed and permeate concentrations of the faster permeating component i which is water in the present case. Quite often, the pervaporation separation index (PSI) is used to describe the overall performance of a membrane for a selected feed mixture. This can be calculated from the product of water flux, j, and selectivity, R. Often, an enrichment factor, β, calculated from eq 4 has been used to test the membrane performance. 29,30 Here, C w p and C w F are concentrations of water in the permeate and feed sides. (1) (2) PSI ) J i R (3) β ) C w p /C w F (4) Membrane Characterization. All the membranes were characterized by FTIR, WAXD, and TGA along with tensile testing. The FTIR spectra of unmodified and cross-linked membranes were obtained by using a Perkin-Elmer-283B FTIR spectrometer. Thermal stability of the polymer films was examined, using a Seiko 220TG/DTA analyzer, from 25 to 600 C heated at 10 C/min and flushed with nitrogen at 200 ml/min. The samples were subjected to TGA both before and after blending to determine the thermal stability and decomposition characteristics. A Siemens D 5000 powder X-ray diffractometer was used to study the solid-state morphology of the cross-linked CS membrane in a powdered form. X-rays of Å wavelength were generated by a Cu K source. The angle of diffraction was varied from 0 to 65 to identify the change in the crystal structure and intermolecular distances between the intersegmental chains after cross-linking. The equipment used to study the mechanical properties of films was a universal testing machine (UTM) (Shimadzu make, model AGS-10kNG) with an operating head load of 5 kn. The crosssectional area of the sample of known width and thickness was calculated. The films were then placed between the grips of the testing machine. The grip length was 5 cm, and the speed of testing was set at the rate of 12.5 mm/min. Tensile strength was calculated using the equation: Max load Tensile strength ) Cross-sectional area (N/mm2 ) (5) Ion Exchange Capacity [IEC]. In order to determine the effect of cross-linking on the blend membranes, the ion exchange capacities (IECs) of the blends were estimated. The IEC indicates the number of groups present before and after crosslinking, which gives an idea about the extent of cross-linking. Thus, IEC gives the number of milliequivalents of ions in 1 g of dry polymer. To determine IEC, specimens of identical weights were soaked in 50 ml of 0.01 N NaOH solution for about 12 h at ambient temperature. Then, 10 ml of sample was titrated against 0.01 N H 2 SO 4. The sample was regenerated with

4 2158 Ind. Eng. Chem. Res., Vol. 46, No. 7, 2007 Table 1. Sorption, Diffusion, and Permeability Coefficients of the P-CS Membrane solvent sorption coefficient (S) diffusion coefficient D 10 7 (cm 2 /s) permeability coefficient P 10 7 (cm 2 /s) swelling coefficient (R) ethylene glycol water EG/water (92/8) a a For water. 1M HCl, washed with water, and dried to constant weight. IEC was then calculated as or IEC ) (B - P)M NaOH 5 m (B - P)(M H2 SO 4 2)5 IEC ) m where B is the amount of sulfuric acid used to neutralize the blank sample and P is the amount of sulfuric acid used to neutralize the pervaporation membranes. Here, 5 is the factor corresponding to the ratio of the amount of NaOH taken to dissolve the polymer to the amount used for titration. Also, m is the sample mass in grams. Dynamic Swelling. Dynamic and equilibrium swelling of the known weight circular pieces of cross-linked polymer films (3 cm in diameter) were carried out at 30 C until equilibrium was attained in the presence of ethylene glycol/water mixtures and as well as individual components. The films were taken out after different soaking periods and quickly weighed after carefully wiping out excess liquid to determine the amount absorbed at the particular time t. The films were then quickly placed back in the solvent. During this period, the total time spent by the membrane outside the solvent medium was kept minimum (25-35 s) to minimize the experimental error due to solvent evaporation. This error was negligible when the time spent by the membrane outside the test bottles is compared to the time spent inside the solvent medium. The process was repeated until the films attained steady state as indicated by constant weight after a certain period of soaking time. Weight gain, Q t, of the soaked polymer membranes is expressed in mole percent units (i.e., number of moles of solvent sorbed by 100 g of the polymer), which is calculated as (8) M s Q t ) ( W t - W i W i ) where W t and W i are, respectively, weight gains at time t and the initial weight and M s is the molecular weight of the sorbed liquid. When equilibrium was reached, Q t was taken as Q Since permeation of a solvent through a polymeric membrane occurs due to diffusion of molecules, permeability coefficient, P, can be calculated as where D is diffusion coefficient and S is the solubility or sorption coefficient. The sorption coefficient is simply equilibrium sorption, which was calculated as (6) (7) P ) DS (9) S ) M M 0 (10) where M is the equilibrium mass of the membrane and M 0 is its initial mass. The diffusion coefficient was computed from sorption results using the equation 32 D ) π( hθ 4Q ) 2 (11) where θ is slope of the linear portion of the sorption curve before attainment of 60% equilibrium and h is the initial film thickness. Estimated values of D, S, and P are given in Table 1. Swelling coefficient R was calculated using the equation 33 R)( M - M 0 M 0 ) 1/F s (12) where M and M 0 are the masses of the swollen membrane and the initial membrane and F s is density of the solvent. The percent sorption is obtained from % sorption ) M s - M d M d 100 (13) where M s ) the mass of the swollen polymer in grams and M d ) the mass of the dry polymer in grams. Analytical Procedure. The feed and permeate samples were analyzed using a Nucon gas chromatograph (model 5765) installed with a thermal conductivity detector (TCD) and a Tenax packed column of 2 m in length. The oven temperature, which was programmed, was maintained initially at 70 C for 3 min followed by an increase in temperature at 25 C/min up to 220 C. The injector and detector temperatures were maintained at 230 C each. The sample injection size was 1 µl, and pure hydrogen was used as the carrier gas at a pressure of 1 kg/cm 2. The GC response was calibrated for this particular column and conditions with known compositions of EG-water mixtures and the calibration factors were fed into the software to obtain the correct analysis for unknown samples. 3. Results and Discussion The chitosan membrane was chosen for the PV studies of EG-water mixtures on the basis of the close proximity of its Hansen s solubility parameter value (43.04 J 1/2 /cm 3/2 ) 34 to that of water (47.9 J 1/2 /cm 3/2 ) 35 as well as other useful features, such as hydrophilicity, good mechanical strength, and chemical resistance. Figure 1 represents the cross-linking reaction between amino groups of chitosan with the phosphate group of phosphoric acid. Upon reaction with phosphoric acid, the chitosan chains form an antiparallel structure, resulting in strong ionic bonding between NH + 3 functional groups of the polymer and PO 3-4 of the acid. This is possible because the amino group of chitosan is a stronger nucleophile than its hydroxyl group. An estimation of the number of groups present before and after cross-linking gives an idea of the extent of cross-linking. To the best of our knowledge, it is the first kind of study wherein phosphoric acid is employed as a cross-linking agent, and the

5 Ind. Eng. Chem. Res., Vol. 46, No. 7, Figure 3. FTIR spectra of (a) chitosan and (b) phosphorylated chitosan. Figure 4. XRD pattern of (a) chitosan and (b) phosphorylated chitosan. membrane could withstand the solvent environment and PV conditions employed in this study. FTIR Studies. Figure 3 shows the FTIR spectra of the plain and cross-linked chitosan membranes. The FTIR spectrum of chitosan (Figure 3a) shows the prominent peaks of hydroxyl and amide at 3459 and /cm, respectively. 36 After crosslinking, the amide band shifts to the lower wavenumber caused by the cross-linking of chitosan chains. Amino groups in chitosan and phosphoric acid have a Columbic interaction, which cross-link the chitosan main chains ionically as shown in Figure 1. The FTIR analysis of modified chitosan (Figure 3b) confirms that cross-linking has taken place by the reaction of the amine group of chitosan with the hydroxyl group of phosphoric acid. Further, the peak at /cm can be attributed to the NH 3 + deformation, while a peak at /cm corresponds to PdO stretching and OH deformation occurring on cross-linking the polymer. 37 The Coulombic interactions between amino groups in chitosan and phosphate ions strengthen the charge density in chitosan membranes, leading to an increase in the affinity of a membrane toward water and thus resulting in high selectivities of a membrane toward water. 38 X-ray Diffraction (XRD) Studies. From the spectra obtained for pure and cross-linked CS shown in Figure 4, it is observed that XRD patterns of both pure and P-CS membranes appear to be semicrystalline. The broad peaks observed in the XRD pattern around 10 of 2θ indicate the average intermolecular distance of the amorphous part and relatively sharp semicrystalline peaks are centered at around 20 of 2θ. From these observations, it can be concluded that the average intermolecular distances in CS and cross-linked CS are the same. On comparing the spectra of CS and cross-linked CS membranes, it can be seen that there are two distinct bands with their maxima at 2θ ) and 2θ at 20, which are related to two types of crystals: crystal 1 and crystal Crystal 1, which corresponds to the peak at 9, is responsible for the separation, since it comprises functional groups like -NH 2 and -OH and has undergone significant change after cross-linking. The reduction in the effective d-spacing value from 4.59 Å (at 2θ ) 21 ) for pure CS to 4.37 Å (at 2θ ) 20 ) for the cross-linked CS gives a clear picture of the shrinkage in cell size or intersegmental spacing, which would improve the selective permeation of the CS membrane.

6 2160 Ind. Eng. Chem. Res., Vol. 46, No. 7, 2007 Figure 5. TGA curves of (a) chitosan and (b) phosphorylated chitosan. Table 2. Sorption and Diffusion Coefficients and Pervaporation Parameters of the P-CS Membrane at Different Feed Concentrations feed concentration EG water sorption coefficient (S) flux (kg/m 2 h) % sorption selectivity (R) enrichment factor (β) diffusion coefficient (water) D 10 7 (cm 2 /s) PSI TGA Analysis. Thermal degradation of CS and phosphorylated CS was examined by TGA, and these results are displayed in Figure 5. It was observed that pure CS exhibited weight loss in the range of 180 C followed by a final decomposition of the polymer that began around 260 C. Weight loss could be attributed to splitting of the main polymer chain before its final decomposition. The cross-linked polymeric blend exhibited weight loss in the range of 249 C followed by a final decomposition at 330 C. An increase in thermal stability of the membrane was observed, which is due to the more exothermic nature of the phosphorylated CS. It can thus be said that cross-linked chitosan can be safely used for dehydration of ethylene glycol/water mixtures at room temperature. Tensile Strength and Elongation. The results of tensile strength and elongation of CS and cross-linked CS blend membranes exhibited tensile strengths of and N/mm 2 and percents of elongation at break of 3.48 and 2.21, respectively, and it is clear that after cross-linking the membrane with phosphoric acid, tensile strength increases. This increase in mechanical strength is due to the formation of strong ionic interactions in the membrane. From these values, it is evident that the cross-linking plays an important role in enhancing the mechanical strength and decreasing the elongation of the P-CS membrane. It is well-known that cross-linking polymeric chains create chemical cross-link structures in the membranes as a result of entanglement of polymeric chains. This chemical cross-link generated by entanglement among the polymer chains opposes deformation which decreases the elongation. Ion Exchange Capacity. Residual ionic groups are generally estimated by ion exchange capacity. Ion exchange capacity plays an important role on the values of pervaporation parameters such as flux and selectivity. IEC studies of the phosphorylated chitosan membrane give the residual amine and hydroxyl groups present in the membrane after cross-linking. The IEC values of chitosan and phosphorylated chitosan are 2.61 and 1.08 mequiv/ g, respectively. More than half (58%) of the groups are crosslinked with phosphoric acid, and there are still some amine and hydroxyl groups left for sorption and diffusion of water molecules through hydrogen bonding. Sorption Phenomenon. The pervaporation transport mechanism can be well interpreted by the solution-diffusion model. Thus, preferential sorption characteristics of the membrane layer were explored. Sorption, namely solubility of the membrane, is caused by the interaction of penetrating species and the membrane. The water sorption of the P-CS membrane for the ethylene glycol/water mixture is shown in Table 2, where the water amount sorbed in the membrane increases with water content in the mixture, i,e., with feedwater concentrations from 0 to 35 wt %. From Table 2, it is observed that the membranes showed a high sorption percentage to water at equilibrium (79.8%). The hydrophilic groups in these membranes are responsible for the preferential water sorption. It is also shown that the sorption coefficient increases with increasing water concentration in the mixture. Results of S, P, and D for ethylene glycol, water, and the EG/water mixture (92/8 wt %) are presented in Table 1 at 30 C. From the table, it is observed that the values of S, P, and D are higher for water compared to ethylene glycol and it shows that the P-CS membrane preferentially allows the water molecules over ethylene glycol due to its hydrophilic nature. As membrane swelling increases with increasing water content, more ethylene glycol molecules are sorbed together with water molecules. This is called sorption coupling. The P-CS membrane shows high water selectivity with the maximum occurring at 97 wt %

7 Ind. Eng. Chem. Res., Vol. 46, No. 7, content of ethylene glycol in the feed mixture at 30 C. From these data, it is apparent that P-CS is an excellent material for the dehydration of ethylene glycol mixtures. Influence of Operating Conditions. It is well-known that separation characteristics of a membrane depend upon the interaction between the solvent to be separated and the membrane matrix. For a somewhat hydrophilic membrane like CS that can develop hydrogen bond interactions, water could be extracted from the organic solvent by PV. The hydrogen bond interaction between water and ethylene glycol forms a cluster, which has the formula (ROH) x yh 2 0, so that separation by a hydrophobic membrane is difficult due to the relatively large coupling of the diffusion. The influence of feed composition, membrane thickness, and permeate pressure have been examined in order to study in detail the PV separation. Diffusion Coefficient. Transport of molecules in PV experiments has been explained by the solution-diffusion model, 40 since molecular transport in PV depends upon sorption and diffusion of liquid molecules through the barrier membrane, and hence, attempts have been made to compute the diffusion coefficients of liquids through P-CS membranes. Diffusion in PV experiments occurs as a result of a concentration gradient and driving force across the membrane created as a result of pressure differential due to the application of high vacuum (lower pressure compared to feed side) on the permeate side. The diffusion coefficient, D i, of J i solvent molecules was computed from the PV results using the equation. 41,42 J i ) (D i C i )/h (14) where C i is concentration of water or ethylene glycol in the feed mixture after completion of the PV process, J i is the flux of the given species, and h is the membrane thickness. Computed values of D i at 30 C are also included in Table 2. Notice that diffusion coefficients calculated from eq 11, i.e., before completion of 60% equilibrium sorption are quite different than those computed from eq 14 due to different processes (see Tables 1 and 2). However, these cannot be compared on an absolute scale due to the fact that the experimental setup and the systems chosen for the study are different. In order to ascertain the mechanism of liquid diffusion through P-CS membranes, we have analyzed the mole percent sorption data using the empirical equation 43 log( Q t ) log k + n log t (15) Q ) where Q t and Q are mole percent uptake values at time t and at equilibrium, respectively and k and n are the system parameters that depend upon the structural characteristics of P-CS. These values give valuable information about the nature of interactions between liquid components and P-CS membranes. Values of k and n have been estimated by the method of regression analysis by fitting log (Q t /Q )vslogt. The value of n indicates the type of diffusion mechanism. It was found that n values varied from 1 to 1.7, indicating relaxationcontrolled transport, i.e., case II type transport. However, the following n values were found: 1.16 for water, 1.04 for ethylene glycol, and 1.72 for ethylene glycol/water (92/8 wt %). These indicate that diffusion follows the relaxation controlled mechanism. Effect of Feed Concentration on Flux and Selectivity. Pervaporation experiments were performed with solvent systems such as ethylene glycol/water at 30 C and a wide range of concentrations to study the separation behavior of the P-CS Figure 6. Effect of feedwater concentration of EG/water on flux and selectivity. membrane. For this study, the membrane thickness and permeate pressure were kept constant at 50 µm and 0.1 mmhg, respectively. Ionic cross-linking leads to membranes with higher separation factors and less permeation flux than those of an uncross-linked chitosan membrane. Figure 6 shows the effect of the feedwater concentration on the total flux and selectivity for ethylene glycol/water. The flux for the system is observed to vary from 0.37 to 0.82 kg/m 2 h as the feedwater concentration is increased from 4 to 35 wt %. It was interesting to note that the total flux for the system increases substantially from 4 to 35 wt % water in the feed. This is due to the fact that the P-CS membrane has a high affinity for water and is very hydrophilic in nature. The ionically cross-linked membrane structure effectively excludes the permeation of ethylene glycol, while the total and water permeabilities remain almost unchanged. This phenomenon can be explained by the fact that ionized groups hydrate easily and exclude organic solvents (the salting-out effect). 44 Moreover, residual -OH and -NH 2 groups of chitosan will also be available for interaction with water molecules through hydrogen bonding. In pervaporation, it is essential to obtain reasonably good flux but higher selectivity in order to attain a permeate free of the organic component. At a 97 wt % composition of ethylene glycol, a separation factor of 234 is achieved for the P-CS membrane. It means that chitosan is an extremely water permselective material. The separation factor of the P-CS membrane decreases from 234 to 4.18 with increasing feedwater concentration from 3 to 35 wt %. This membrane is hydrophilic, and it swells linearly with water content. Thus, more ethylene glycol molecules coupled with water molecules diffuse through the membrane, resulting in decreased separation efficiency. The separation performance of the membranes is evaluated in terms of the pervaporation separation index (PSI), which gives the combined result of selectivity and flux. The variation of PSI with the concentration of water in the feed is also shown in Table 2. From this table, it is evident that the P-CS membrane yields a higher value of PSI at the 3 wt % concentration of water in the feed, and it decreases with an increasing feedwater concentration. According to the PSI values, the P-CS membrane performed a better separation at a higher feed ethylene glycol concentration. It is realized that the performance of a pure CS membrane is not satisfactory due to larger free volume between the molecular chains. Effect of Membrane Thickness. The effects of membrane thickness on water flux and selectivity were evaluated at a constant feed composition (92/8 wt %) and permeate pressure (0.25 mmhg) by casting membranes of thicknesses ranging from 30 to 160 µm. With an increase in membrane thickness, a gradual reduction in flux from 0.55 to kg/m 2 h can be

8 2162 Ind. Eng. Chem. Res., Vol. 46, No. 7, 2007 Figure 7. Effect of membrane thickness on flux and selectivity for EG/ water. Conclusions Pervaporation of EG/water mixtures through ionically crosslinked phosphorylated chitosan membranes is conducted. At 30 C and 90 wt % EG feed concentration, a permeation flux of kg/m 2 h and water concentration of permeate higher than 93.5 wt % were achieved. The effects of operational conditions including feed EG concentration, permeate pressure, and thickness of P-CS membranes on the pervaporation performance is investigated. In all tested EG concentrations, the present chitosan membrane showed permselectivity toward water. The permeate rate increases as the concentration of water in the feed increases; this mainly includes the contribution of both the increase of the degree of swelling, which enhances the expansion of the free volume, and the active effect of the diffusion coefficient of water through a membrane. In this case, pervaporation is proven as the most promising, alternative technology for the separation of aqueous-organic mixture of EG/water, particularly for the separation of close boiling mixtures. Literature Cited Figure 8. Effect of permeate pressure on flux and selectivity for EG/water. clearly evidenced from Figure 7. Even though the availability of polar groups enhances with an increase in membrane thickness, the flux decreases since diffusion of feed is retarded due to increased resistance to mass transfer. The permeate concentration of water varied from to wt %, which means that selectivity has increased from 56.5 to In the PV experiment, the upstream layer of the membrane is swollen and plasticized due to sorption of feed liquid, thus allowing the unrestricted transport of feed components. In contrast, the downstream layer is virtually dry due to continuous evacuation in the permeate side, and therefore, this layer forms the restrictive barrier, which allows only the interacting and smaller size molecules such as water to pass through. It is expected that the thickness of the dry layer would increase with an increase in the overall membrane thickness, thereby resulting in an improved selectivity as observed in the present case. Effect of Permeate Pressure. The effect of permeate pressure on membrane performance of the cross-linked chitosan membranes was studied in the range of 0.5 to 9 mmhg at a constant membrane thickness of 50 µm. As the permeate pressure decreases, the driving force for diffusing molecules increases, resulting in high permeate rates. Figure 8 exhibits a considerable lowering of flux from 0.45 to 0.17 kg/m 2 h. Similarly, selectivity increased from 61.7 to With the increasing permeate pressures, diffusion of the feed molecules through the membrane, which is the rate-determining step, becomes slow, whereas high vacuum exerts a larger driving force. Under lower vacuum conditions, the volatility of the feed components governs the separation selectivity of the membrane. Ethylene glycol, being less volatile than water, permeates slowly, thus increasing the selectivity. (1) Feming, H. L.; Slater, C. S. Membrane Handbook; Ho, W. S. W., Sirkar, K. K., Eds.; Van Mastrand Reinhold: New York, 1992; pp (2) Lee, Y. M.; Oh, B. K. In-situ Complex Membrane: Dehydration of Pyridine Aqueous Solution by Pervaporation. Macromol. Symp. 1997, 118, (3) Doguparthy, S. P. Pervaporation of aqueous alcohol mixtures through a photopolymerised composite membrane. J. Membr. Sci. 2001, 185, (4) Kim, K. J.; Park, S. H.; So, W. W.; Moon, S. J. Pervaporation separation of aqueous organic mixtures through sulfated zirconia-poly(vinyl alcohol) membrane. J. Appl. Polym. Sci. 2001, 79, (5) Soltanieh, M.; Ghoreyshi, S. A. A.; Farhadpour. F. A. Multicomponent transport across nonporous polymeric membranes. Desalination 2002, 144, (6) Greek, B. F. Ethylene-glycol supplies will be sufficient to meet rising demand. Chem. Eng. News 1991, 69 (33), 11. (7) Bravo, J. L.; Fair, J. R.; Humphery, J. L.; Martin, C. L.; Saibert, A. F.; Joshi, S. Fluid Mixture Separation Technologies for Cost Reduction and Process ImproVement; Noyes Data Corp.: Park Ridge, NJ, (8) Feng, X.; Huang, R. Y. M. Pervaporation with chitosan membranes. I. Separation of water from ethylene glycol by a chitosan/polysulfone composite membrane. J. Membr. Sci. 1996, 116, (9) Chen, F. R.; Chen, H. F. Pervaporation separation of ethylene glycolwater mixtures using crosslinked PVA-PES composite membranes. Part I. Effects of membrane preparation conditions on pervaporation performances. J. Membr. Sci. 1996, 109, (10) Reid, R. C.; Prausnitz, J. M.; Polimg, B. E. The Properties of Gases and Liquid; McGraw-Hill: New York, (11) Bartels, C. B.; Reale, J. Dehydration of glycols, Texaco Inc. US Patent 4,802,988, (12) Jehle, W.; Staneff, T. H.; Wagner, B.; Steinwandel, J. Separation of glycol and water from coolant liquids by evaporation, reverse osmosis and pervaporation. J. Membr. Sci. 1995, 102, (13) Nam, S. Y.; Lee, Y. M. Pervaporation of ethylene glycol-water mixtures: I. Pervaporation performance of surface crosslinked chitosan membranes. J. Membr. Sci. 1999, 153, (14) Burshe, M. C.; Sawant, S. B.; Joshi, J. B.; Pangarkar, V. G. Dehydration of ethylene glycol by pervaporation using hydrophilic IPNs of PVA, PAA and PAAM membranes. Sep. Purif. Technol. 1998, 13, (15) Sekulic, J.; Elshof, J. E.; Blank, D. H. A. Selective Pervaporation of Water through a Nonselective Microporous Titania Membrane by a Dynamically Induced Molecular Sieving Mechanism. Langmuir 2005, 21, (16) Ito, H.; Shibata, T.; Noishiki, Y.; Inagaki, H. Formation of polyelectrolyte complexes between cellulose derivatives and their blood compatibility. J. Appl. Polym. Sci. 1986, 31, (17) Kim, J. H.; Kim, J. Y.; Lee, Y. M.; Kim, K. Y. Properties and swelling characteristics of cross-linked poly (vinyl alcohol)/chitosan blend membrane. J. Appl. Polym. Sci. 1992, 45,

9 Ind. Eng. Chem. Res., Vol. 46, No. 7, (18) Kim, J. H.; Lee, Y. M. Synthesis and properties of diethylaminoethyl chitosan. Polymer 1993, 34, (19) Mochizuki, A.; Sato, Y.; Ogawara, H.; Yamashita, S. Pervaporation separation of water/ethanol mixtures through polysaccharide membranes. II. The permselectivity of chitosan membrane. J. Appl. Polym. Sci. 1989, 37, (20) Lee, Y. M.; Shin, E. M. Pervaporation Separation of Water-Ethanol through Modified Chitosan Membrane: 1. Chitosan-Acetic Acid and-metal Ion Complex Membranes. Polymer (Korea) 1991, 15, (21) Lee, Y. M.; Nam, S. Y.; Woo, D. J. Pervaporation of ionically surface crosslinked chitosan composite membranes for water-alcohol mixtures. J. Membr. Sci. 1997, 133, (22) Wu, L.; Zhu, C.; Liu, M. Study of a new pervaporation membrane: Part 2. Performance test and analysis of the new membrane. J. Membr. Sci. 1994, 90, (23) Mochizuki, A.; Amiya, S.; Sato, Y.; Ogawara, H.; Yamashita, S. Pervaporation separation of water/ethanol mixtures through polysaccharide membranes. III. The permselectivity of the neutralized chitosan membrane and the relationships between its permselectivity and solid state structure. J. Appl. Polym. Sci. 1989, 37, (24) Lee, Y. M.; Nam, S. Y.; Kim, J. H. Pervaporation of water-ethanol through poly(vinyl alcohol)/chitosan blend membrane. Polym. Bull. 1992, 29, (25) Kim, J. H.; Kim, J. Y.; Lee, Y. M.; Kim, K. Y.; Cho, C. S.; Sung, Y. K. Controlled-Release Drug Delivery through Crosslinked Poly(vinyl alcohol)/chitosan Blend Membrane. Polymer (Korea) 1991, 15 (6), (26) Kifune, K. Clinical Application of Chitin Artificial Skin. In AdVances in Chitin and Chitosan, Proceedings of the 5th International Conference on Chitin and Chitosan; Princeton, NJ, 1991; p 9. (27) Sridhar, S.; Ravindra, R.; Khan, A. A. Recovery of Monomethylhydrazine Liquid Propellant by Pervaporation Technique. Ind. Eng. Chem. Res. 2000, 39, (28) Devi, D. A.; Smitha, B.; Sridhar, S.; Aminabhavi, T. M. Pervaporation separation of dimethylformamide/water mixtures through poly(vinyl alcohol)/poly(acrylic acid) blend membranes. Sep. Purif. Technol. 2006, 51, (29) Lu, J.; Nguyen, Q.; Zhou, L.; Xu, B.; Ping, Z. Study of the role of water in the transport of water and THF through hydrophilic membranes by pervaporation. J. Membr. Sci 2003, 226, (30) Billard, P.; Nguyen, Q. T.; Leger, C.; Clement, R. Diffusion of organic compounds through chemically asymmetric membranes made of semi-interpenetrating polymer networks. Sep. Purif. Technol. 1998, 14, (31) Harogoppad, S. B.; Aminabhavi, T. M. Diffusion and sorption of organic liquids through polymer membranes. II. Neoprene, SBR, EPDM, NBR, and natural rubber versus n-alkanes. J. Appl. Polym. Sci. 1991, 42, (32) Aminabhavi, T. M.; Khinnavar, R. S. Diffusion and sorption of organic liquids through polymer membranes: 10. Polyurethane, nitrilebutadiene rubber and epichlorohydrin versus aliphatic alcohols (C 1-C 5). Polymer 1993, 34, (33) Unnikrishnan, G.; Thomas, S. Diffusion and transport of aromatic hydrocarbons through natural rubber. Polymer 1994, 35, (34) Ravindra, R.; Kameswara Rao, K.; Khan, A. A. Solubility parameter of chitin and chitosan. Carbohydr. Polym. 1998, 36, (35) Barton A. F. M., Ed. CRC Handbook of Solubility Parameters and other CohesiVe Parameters; CRC Press: Boca Raton, Florida, (36) Yao, A. A.; Peng, T.; Goosen, M. F. A.; Min, J. M.; He, Y. Y. ph-sensitivity of hydrogels based on complex forming chitosan: Polyether interpenetrating polymer network. J. Appl. Poly. Sci. 1993, 48, (37) Aimoli, C. G.; Beppu, M. M. Precipitation of calcium phosphate and calcium carbonate induced over chitosan membranes: A quick method to evaluate the influence of polymeric matrices in heterogeneous calcification, Colloid Surf., B: Biointerfaces 2006, 53, (38) Lee, Y. M.; Nam, S. Y.; Woo, D. J. Pervaporation of ionically surface crosslinked chitosan composite membranes for water-alcohol mixtures. J. Membr. Sci. 1997, 133, (39) Dhanuja, G.; Sridhar, S.; Smitha, B. Pervaporation of isopropanolwater mixtures through polyion complex membranes. Sep. Purif. Technol. 2004, 44, (40) Wijmans, J. G.; Baker, R. W. The solution-diffusion model: a review. J. Membr. Sci. 1995, 107, (41) Uragami, T.; Takigawa, K. Permeation and separation characteristics of ethanol-water mixtures through chitosan derivative membranes by pervaporation and evapomeation. Polymer 1990, 31, (42) Kusumocahyo, S. P.; Sudoh, M. Dehydration of acetic acid by pervaporation with charged membranes. J. Membr. Sci. 1999, 161, (43) Lucht, L. M.; Peppas, N. A. Transport of penetrants in the macromolecular structure of coals. V. Anomalous transport in pretreated coal particles, J. Appl. Polym. Sci. 1987, 33, (44) Maeda, I.; Kai, M. PerVaporation Membrane Process; Huang, R. Y. M., Ed.; Elsevier: Amsterdam, ReceiVed for review October 2, 2006 ReVised manuscript received January 6, 2007 Accepted January 15, 2007 IE061268N