Structural change of ion-exchange membrane surfaces under high electric fields and its effects on membrane properties

Journal of Colloid and Interface Science 265 (2003) 93–100 www.elsevier.com/locate/jcis

Structural change of ion-exchange membrane surfaces under high electric fields and its effects on membrane properties Jae-Hwan Choi 1 and Seung-Hyeon Moon ∗ Department of Environmental Science and Engineering, Kwangju Institute of Science and Technology (K-JIST) 1, Oryong-dong, Buk-gu, Gwangju, 500-712, South Korea Received 15 February 2002; accepted 31 January 2003

Abstract Structural change of an ion-exchange membrane under a high electric field was investigated by comparing water dissociation and the FTIR spectra between the virgin membrane and that used at an overlimiting current density. From a series of water dissociation experiments at overlimiting current densities, it was observed that water dissociation in an anion-exchange membrane used at an overlimiting current density was higher than that in a virgin membrane at the same current density. The FTIR study revealed that the tertiary amine groups are formed from the quaternary ammonium groups on the anion-exchange membrane surface where ion depletion occurs under the influence of the applied strong electric field. The occurrence of increased water dissociation is considered to be caused by the protonation and deprotonation of the tertiary amine groups in the anion-exchange membrane. On the other hand, there was no structural change for the cation-exchange membrane under the electric field investigated in this study, which is coincident with the results of water dissociation experiments for the CMX membrane. In addition, we found that membrane resistance, permselectivity, and plateau length of the current–voltage curve were affected by the converted tertiary amine groups depending on the solution pH.  2003 Elsevier Inc. All rights reserved. Keywords: Ion-exchange membrane; Water dissociation; FTIR spectrum; Quaternary ammonium; Tertiary amine

1. Introduction Electrodialysis has been widely used in numerous processes from wastewater treatment to applications in the food industries [1–9]. Since the electrodialysis process cost depends on the membrane area, it is desired to operate at the highest practicable current density to get the maximum ion flux per unit membrane area [10,11]. Operating current levels are, however, restricted by concentration polarization. Based on the theory, no current higher than the limiting current can be expected since the concentration near the membrane has reached zero [2,10,12]. In fact, however, the current density can be increased beyond the limiting one. Nevertheless, operation at overlimiting current densities is considered to be undesirable because the dissociation of water may lead to a pH change on both sides of the * Corresponding author.

E-mail address: [email protected] (S.-H. Moon). 1 Current address: Department of Chemical Engineering, Kongju Na-

tional University, 182, Shinkwan-dong, Gongju, Chungnam 314-701, South Korea. 0021-9797/$ – see front matter  2003 Elsevier Inc. All rights reserved. doi:10.1016/S0021-9797(03)00136-X

membrane. As a result scaling on or in the membrane may occur, and the membrane may deteriorate under extreme pH conditions. Further, water dissociation is an energyconsuming process and reduces the current efficiency [10]. Often the overlimiting current has been related to the occurrence of water dissociation and many works have focused on the occurrence of water dissociation with ionexchange membranes [13–20]. The mechanism of water dissociation at the overlimiting current density, however, has not been identified clearly. Of various mechanisms proposed, two are catalytic theory and electric field theory. The former suggests that H+ and OH− ions may be produced in proton transfer reactions between charged groups in the membrane and water molecules [14–17]. The electric field theory explains that the water dissociation constant, Kd , can be increased as much as 107 times because of a second Wien effect due to the strong electric field (108 V/m) at the membrane–solution interface under severe concentration polarization [19]. However, many researchers have found that water dissociation is more pronounced with anion-exchange membranes, whereas it seems to be of minor importance with cation-exchange membranes [13–17]. With

94

J.-H. Choi, S.-H. Moon / Journal of Colloid and Interface Science 265 (2003) 93–100

regard to the differences of water dissociation between cation and anion-exchange membranes, Simons suggested that the dissociation of water is bound to a thin layer at the surface of an anion-exchange membrane and is caused by a reversible protonation of weakly basic groups such as tertiary amines, which must be present in the membrane [14– 16]. This idea has been favored by Rubinstein et al., who found that water dissociation was reduced when anionexchange membranes with crown ether groups were used instead of anion-exchange membrane containing quaternary ammonium groups [17]. In this study, structural changes in an ion-exchange membrane at high electric field was investigated. After a series of water dissociation experiments at the overlimiting current density the structural changes of the membranes used were identified by FIIR spectrum using the internalreflection spectroscopy technique. In addition, the effects of structural change on the membrane properties were studied by measuring the current–voltage curves, permselectivity, and membrane resistance.

2. Experimental 2.1. Water splitting in an ion-exchange membrane Water-splitting experiments at various current densities were carried out using an electrodialysis cell consisting of five compartments made of Plexiglas. The arrangement of the cell and its functions for the anion-exchange membrane are schematically shown in Fig. 1. The two outer compartments contain the working electrodes, i.e., a platinized titanium anode and a stainless steel cathode. The membranes used in the experiments were a Neosepta CMX cation-exchange membrane and a Neosepta AMX anionexchange membrane that were purchased from Tokuyama Soda Co., Japan. The effective membrane area of the cell was 25 cm2 . The Neosepta CMX is a cation-exchange membrane containing sulfonic acid groups as fixed charges and the Neosepta AMX is an anion-exchange membrane containing quaternary ammonium groups as fixed charges. The

Table 1 The major properties of the ion-exchange membranes used in this study Properties Electric resistancea ( cm2 ) Thickness (mm) Exchange capacity (meq/g dry membrane) Transport numberb Water contentc Characteristics

CMX

AMX

2.5–3.5

2.5–3.5

0.17–0.19

0.16–0.18

1.5–1.8

1.4–1.7

0.98 < 0.25–0.30 High mechanical strength cation membrane

0.98 < 0.25–0.30 High mechanical strength anion membrane

a Equilibrated with 0.5 N NaCl solution at 25 ◦ C. b Measured by electrophoresis with sea water at a current density of 2 A/dm2 . c Equilibrated with 0.5 N NaCl solution.

main properties of the ion-exchange membranes used in this study are listed in Table 1 [21]. The anion membrane between compartments 1 and 2 was used to block proton transport from the anolyte solution while the cation membrane between compartment 4 and 5 blocked the flow of OH− ions from catholyte solution. Water-splitting experiments were carried out with 500 ml of 0.05 M NaCl solution. Analytical grade chemicals and doubly distilled pure water were used in preparing the electrolyte solutions. To avoid the pH change caused by the dissolution of CO2 from air the solution was placed in a solution reservoir, which was sealed. The solution was circulated through the central compartment (compartment 3) with a flow rate of 150 ml/min. In the electrode compartments (compartment 1 and 5) 0.5 M Na2 SO4 was used. The solutions in compartments 2 and 4 were 1.0 L of 0.5 M NaCl so that no interference occurred due to concentration polarization on any membrane other than the test membrane. Watersplitting experiments were conducted at various current densities. A fixed current was supplied (HP6613C, 0–1 A, 0–50 V) through the stack for 1 h. The pH (Orion 250A) of the central compartment solution was measured every 5 min. All the solutions were replaced for each experiment. In order to determine the limiting current density, experiments were conducted in the cell shown in Fig. 1 in 0.05 M NaCl solution. The voltage drop across the test membrane was measured by Ag/AgCl electrodes. Current– voltage curves were obtained by a stepwise increase of the current density (steps of 0.8 mA/cm2) through the cell. More information about the experiment can be obtained in the previous study [22]. 2.2. FTIR spectroscopic characterization

Fig. 1. Schematic diagram of water dissociation experimental setup for the AMX membrane. The central two AMX membranes between the outer AMX and CMX membrane were replaced by the CMX membranes for the water dissociation experiment for the cation-exchange membrane.

While infrared spectroscopy is an instrumental method often applied to structural determination of organic compounds [23], the ordinary FTIR spectroscopic technique is not proper to identify organic compounds in commercial

J.-H. Choi, S.-H. Moon / Journal of Colloid and Interface Science 265 (2003) 93–100

ion-exchange membranes because the thickness of the membranes is generally in the range 150–200 µm. On the other hand, internal-reflection spectroscopy is a technique for obtaining infrared spectra of samples that are difficult to deal with, such as films, threads, pastes, adhesives, and powders [24]. So the technique is an effective tool for the investigation of membrane surface. Using the internal-reflection spectroscopy technique FTIR spectrums were obtained for the virgin CMX and AMX membranes and those used in water-splitting experiments. The samples were dried for 4 h under vacuum at 50 ◦ C. FTIR spectra were taken at 1 cm−1 resolution in the range 4000–400 cm−1 with a JASCO FTIR-460 infrared spectrometer.

95

equation [2,3] Em =

RT C1 (2t¯+ − 1) ln , F C2

(1)

where Em is the concentration cell potential, C1 and C2 are the electrolyte concentrations in the cells, and R and T are the gas constant and temperature, respectively. 2.5. Electrical resistance of membrane Electric resistances of membranes were measured at 1000 Hz using a lab-made clip cell and an LCZ meter (NF Electronic Instruments, Model 2321). The electrical resistance of the membrane is the difference between the measurements performed with and without the membrane.

2.3. Current–voltage curves For the virgin and used membranes in water splitting experiment, current–voltage curves were obtained using the two-compartment cell shown in Fig. 2. This electrodialytic cell was composed of two equal-volume (200-cm3) compartments. The membrane was placed in a circular hole between the compartments. The effective area of the membrane was 0.785 cm2 . The potential difference across the membrane was measured using two Ag/AgCl electrodes immersed into Luggin capillaries. The electrical current was supplied at a current scanning rate of 1 µA/s by a potentiostat/galvanostat (AutoLab, Model PGSTAT 30) connected to one pair of Ag/AgCl electrode plates. To minimize the water dissociation reactions at the electrodes, which may affect the composition of electrolytes in the compartment, two Ag/AgCl electrodes were used. 2.4. Transport number The apparent transport numbers of the counterion in the membranes were determined by the emf method using Ag/AgCl electrodes for the virgin and used membranes in the water-splitting experiment. To ignore the change of activity coefficient of the electrolyte, dilute solutions (0.005 M and 0.05 M of NaCl) were used. The transport number, t¯+ , for each membrane was calculated by the

Fig. 2. Schematic diagram of the two-compartment electrolytic cell used in current–voltage measurements. (1) Membrane tested; (2) rubber; (3) Luggin capillary; (4) cathode; (5) anode; (6) reference electrode (Ag/AgCl).

3. Results and discussion 3.1. Observation of water dissociation at overlimiting current densities Water dissociation experiments were carried out with increasing current density in successive steps (4, 15, 18, and 21 mA/cm2). The same experiment was repeated at current density of 15 mA/cm2 with the used membrane. Accumulation of water dissociation products within the membrane can affect the pH change in the following experiment. In order to avoid the accumulation, the test membrane was equilibrated with 0.5 M NaCl solution by circulating the solution for 30 min through compartments 2, 3, and 4 prior to each experiment. The pH of the central compartment solution (see Fig. 1) was measured as a function of time at different fixed current densities, and the results are shown in Figs. 3a and 3b. The limiting current density for the CMX membrane in 0.05 M NaCl was 5.1 mA/cm2, that for the AMX membrane being 8.2 mA/cm2 . Figure 3 shows that the pH of the solutions remains constant when the current density is lower than the limiting current density. With the increasing applied current density the change in pH of the solutions become more prominent. It is noticeable that there is a significant difference in pH change between the first and second runs of the water dissociation experiment for the AMX membrane at a current density of 15 mA/cm2. Water dissociation of the second run was enhanced compared to that of the first even under the same experimental conditions. In contrast no noticeable pH change was observed for the CMX membrane between the first and the second run at a current density of 15 mA/cm2. In order to calculate the transport number of the watersplitting products in the membrane, it is necessary to convert the pH values into H+ or OH− concentrations, which can be calculated according to [10] t¯H+ =

F V dCH+ iA dt

or t¯OH− =

F V dCOH− , iA dt

(2)

96

J.-H. Choi, S.-H. Moon / Journal of Colloid and Interface Science 265 (2003) 93–100

Fig. 3. Change in solution pH in central compartment as a function of time for the AMX membrane (a) and for the CMX membrane (b) at various current densities.

Fig. 4. Transport numbers of protons and hydroxide ions in the AMX and CMX membrane as a function of applied current density.

where t¯H+ and t¯OH− are the transport numbers of protons and hydroxyl ions in the membrane, i is the current density, V is the solution volume, and A is the membrane area. The determined transport numbers are shown in Fig. 4. It can be seen that the contribution of water dissociation for the AMX anion-exchange membrane is an order of magnitude larger than that for the CMX cation-exchange membrane. Even for the AMX membrane, however, the determined transport number is very low, smaller than 0.01. This

result indicates that most of the current is still carried by salt ions, and water dissociation is not responsible for the overlimiting current. The experimental results for water splitting agree reasonably with those that obtained by Krol et al. [10]. They showed that the contribution of water dissociation was greater for an anion-exchange membrane than for a cation-exchange membrane. Also, the contribution of water dissociation to the overlimiting current in the anionexchange membrane remained smaller than 0.03, even when a current density 10 times the limiting current density was applied. Although the classical theory of concentration polarization for an ion-exchange membrane predicts a plateau of the current–voltage curve, it does not explain the overlimiting current. It has been reported by previous researchers that the overlimiting current was due to mechanisms such as water splitting [25], electroosmosis, or loss of permselectivity at a high voltage [26]. Recent works, however, have shown conclusively that no such mechanisms are responsible for the overlimiting currents. All these effects were found to be side effects. Researchers have investigated the main mechanism to explain the reason for the overlimiting current [10,18,22,27–29]. Rubinstein et al. [28] studied the role of the membrane surface in concentration polarization at ion-exchange membranes and reported that the ion conductance of the membrane surfaces is not uniform. If a membrane surface comprises patches of varying ion conductance, the electric field lines in the adjacent solution layer are distorted. The interaction of space charges with the electric field gives rises to a spatially inhomogeneous bulk force that is bound to set the fluid in the depletion diffusion layer in motion, which is called electroconvection. It was thought that the significant concentration polarization formed in the diffusion boundary layer was disturbed by turbulent convection when the current exceeded the limiting current density, leading to hydrodynamic mixing. As a consequence the concentration at the membrane surface increased to a sufficient level for generation of the overlimiting current. Electroconvection has been described in detail elsewhere [27]. In addition, it can be observed that the transport number of proton ions for the second run with an AMX membrane is about two times higher than that for the first run at a current density of 15 mA/cm2. This interesting experimental result could be explained by the catalytic mechanism of water dissociation in an anion-exchange membrane. Simons suggested that when a strong electric field is involved in the anion-exchange membrane containing quaternary ammonium groups, tertiary groups can be created by decomposition of the quaternary groups, and this changes cause to enhance water dissociation in the anion-exchange membrane [15]. To prove this suggestion, the surface changes of the membranes used in water dissociation experiments (coded “used AMX” and “used CMX”) were characterized by FTIR spectra, and the spectra were compared with those of virgin membranes.

J.-H. Choi, S.-H. Moon / Journal of Colloid and Interface Science 265 (2003) 93–100

3.2. FTIR spectrum study The FTIR spectra were obtained on both sides of the used membrane, i.e., the front side where ion depletion occurs in water splitting experiments and the back side where ions are concentrated. Figure 5 shows the FTIR spectrum for the virgin AMX and used AMX membrane. The absorptions of interest in the infrared spectra of amines are those associated with N–H vibrations. Primary and secondary amines are characterized by absorption bands in the region 3000– 3500 cm−1 that are ascribed to N–H stretching vibrations. Primary amines give two bands in this region; secondary amines generally give only one. Tertiary amines, since they have no N–H group, do not absorb the source in this region. Absorption bands arising from C–N stretching vibrations of aliphatic amines occur in the region 1020–1220 cm−1 [23]. As can be seen in Fig. 5, there are no peaks in the range 3000–3500 cm−1 . However, a sharp absorption peak appears for the front side of the used AMX membrane at a wave number of 1020 cm−1 . This indicates that tertiary amine groups exist on the front side of the used AMX membrane, and there are no primary and secondary amine groups. On the other hand, no significant change in FTIR spectra was observed between the virgin AMX membrane and the back side of the used AMX membrane. Through the FTIR study it can be concluded that the quaternary ammonium groups in the virgin AMX membrane are converted to tertiary amine groups under the strong electric field. The enhanced water dissociation in the used AMX membrane may be caused by reversible protonation and deprotonation of the converted tertiary amine groups. According to Simons’ catalytic theory [14–16], water dissociation in membranes containing weak base groups can take place by the reactions B + H2 O ⇔ BH+ + OH− , +

+

BH + H2 O ⇔ B + H3 O ,

(3) (4)

97

considered that the enhanced water dissociation in the used AMX membrane is caused by reversible protonation and deprotonation of the converted tertiary amine groups. Comparing the FTIR spectra of the front and back sides of the used AMX membrane shows that structural change occurred only on the front side, while there was no structural change on the back side of the membrane. It has been reported that the fixed charges of anion-exchange polymers can be split off the polymeric backbone in a strong basic (higher than pH 12) environment at a high temperature [30,31]. Sata et al. [31] examined the properties of commercial anion-exchange membranes after the membranes had been immersed in an aqueous sodium hydroxide solution of high temperature up to 75 ◦ C. They found that quaternary ammonium groups decompose in the alkali solution by the Hofmann degradation reaction. When concentration polarization becomes very severe, the electrical potential gradient at the membrane–solution interface becomes large enough to cause water dissociation. It is generally believed that the protons and hydroxyl ions originate from water dissociation reactions in a very thin (10–100 Å) region (space charge layer) where the concentration of counterions decreases and uncompensated fixed charges exist [14]. The water-splitting products must then be transported through the membrane and diffusion boundary layer. The water dissociation at the space charge layer is illustrated in Fig. 6 with the expected pH profile. Although it is nearly impossible to measure the pH inside the membrane, it is predictable that the solution inside an anion-exchange membrane becomes an alkaline condition that is preferable to structural change from quaternary to tertiary amine. One of the operation characteristics of the electrodialysis process is the temperature increase of solutions. The temperature increase is caused by ion transport through the membrane because of the electrical resistance of the membrane, i.e., Joulean heat. Mavrov et al. [13] reported that

where B is a neutral base. Based on the known kinetic constants, the author reported that this process should be substantially faster than direct water dissociation. Thus it is

Fig. 5. FTIR spectrum of the AMX membrane: (a) virgin membrane, (b) back side of the used AMX membrane, (c) front side of the used AMX membrane.

Fig. 6. Schematic drawing of water dissociation site (space charge layer) at the anion-exchange membrane surface and expected pH (dotted line) and temperature profile (solid line).

98

J.-H. Choi, S.-H. Moon / Journal of Colloid and Interface Science 265 (2003) 93–100

Fig. 7. FTIR spectrum of the CMX membrane: (a) virgin membrane, (b) back side of the used CMX membrane, (c) front side of the used CMX membrane.

heat generation depends on the current density, the types of the solutions, and the membranes. They found that temperature increases were negligible at current densities below the limiting current density and temperature increased steeply beyond the limiting current density. When the overlimiting current is passed through the membrane the electric field intensity within the membrane is not uniform but has an extremely high field strength (108 V/m) at the space charge layer [19]. This implies that the Joulean heat generated in the space charge layer is significantly proportional to the electric field strength. Consequently the temperature in the space charge layer could be increased significantly. Thus it is explained that the structural change at the front side of the AMX membrane is caused by the increase in pH and temperature in the space charge layer when the overlimiting current is supplied. It is complicated to find a temperature profile in the membrane because heat is transported through the membrane and diffusion boundary layer. Considering that the space charge layer is very thin (a few nanometers) compared to the membrane thickness (about 200 µm) and heat can be dissipated into the solution immediately, the temperature at the edge of the back side of the membrane may be nearly equal to that of bulk solution (see Fig. 6). Thus it is expected that the structural change did not occur at the back side of the used AMX membrane. On the other hand, no noticeable change was observed in the FTIR spectrum for the used CMX membrane as shown in Fig. 7. The result implies that the CMX membrane is more stable than the AMX membrane under a strong electric field. It is generally known that many organic compounds are more stable in acid conditions than in alkaline conditions [27]. When the cation-exchange membranes are placed in severe environments, the fixed charges of membranes can be split off the polymeric backbone, or the polymeric backbone can be destroyed. As an example, sulfonic acid groups in the cation-exchange membrane can be split off under high temperatures (much higher than ambient temperature) and strongly acid (less than pH 0) environments [30]. However, these conditions hardly occur in electrodialysis operations.

Fig. 8. Current–voltage curves for the used AMX membrane in 0.02 M of NaCl solution.

3.3. Characterization of the used AMX membrane The characteristics of used AMX membrane, such as current–voltage curve, permselectivity, and membrane resistance, were analyzed and compared with those of virgin AMX membrane. Figure 8 shows the current–voltage curves obtained for the virgin and used AMX membranes in 0.02 M of NaCl solution. Current–voltage measurements were performed for the front and back sides of the used AMX membrane, the membrane surface investigated facing the cathode. All the curves show a typical pattern of current–voltage relations: a first region of approximately ohmic behavior and a second region corresponding to the plateau, followed by a third region of rapid current increase and electrical noise. The limiting current densities, about 2.5 mA/cm2, were observed from the current–voltage curves shown in Fig. 8. According to the classical concentration polarization theory, the limiting current density is expressed by the equation [2,10] ilim =

zi CF D , δ(t¯i − ti )

(5)

where ilim is limiting current density, zi is the valence of species i, D is a diffusion coefficient, δ is a diffusion boundary layer thickness, and t¯i , ti are the transport numbers in the membrane and solution phases, respectively. The equation indicates that the limiting current density is proportional to the concentration and diffusion coefficient and to the reciprocal of diffusion boundary layer thickness and the transport number in the membrane phase. In this study the limiting current density for the AMX membrane was determined as 8.2 mA/cm2 in 0.05 M NaCl solution. If all the parameters in Eq. (5) are the same, the limiting current density in a 0.02 M NaCl solution has to be 3.28 mA/cm2, which is much higher than that obtained from Fig. 8. The discrepancy is due to the different hydrodynamic conditions, which affect the diffusion boundary layer thickness significantly. In this study two different cells (shown in Figs. 1 and 2) were used to investigate the water splitting and characterization of

J.-H. Choi, S.-H. Moon / Journal of Colloid and Interface Science 265 (2003) 93–100

the membrane. Current–voltage curves were obtained with a flow rate of 150 ml/min with the cell shown in Fig. 1, but without stirring with the cell shown in Fig. 2. This different hydrodynamic condition changed diffusion boundary layer thickness at the membrane surface. It has been proved that the current plateau (limiting current density) forms as a result of increasing resistance due to concentration polarization in the boundary layer. There is a distinct difference in the plateau length of the current– voltage curve between the virgin AMX membrane and the front side of the used AMX membrane. In contrast, no significant difference in the current–voltage curve is observed between the virgin AMX membrene and the back side of the used AMX membrane. It is considered that the discrepancy in plateau length is caused by the structural change at the membrane surface. According to the electroconvection theory, convection in the diffusion boundary layer at the overlimiting current region is affected by the separation distance between conductive sites at the membrane surface [28]. The plateau length can be regarded as a minimum potential that can cause electroconvective mixing in the diffusion boundary layer. Therefore it can be inferred that plateau length increases when the inhomogeneity of the membrane (or the distance between conductive sites) increases. As identified in the FTIR study, the front side of the used AMX membrane contains a tertiary amine group that is a weak base. The membrane samples used in the current–voltage experiment were equilibrated with a 0.1 M NaCl solution with pH about 5.7. In the solution tertiary amine exists in ionized forms because their pKa values are approximately in the range of 9–11 depending on the compounds substituted at the nitrogen atom [23]. When the current density approaches the limiting current, however, water dissociation occurs at the membrane–solution interface. Then the dissociated hydroxyl ions move to the anion-exchange membrane so that a part of the tertiary amine may be converted to neutral amine, resulting in a decrease of conducting sites in the membrane and a greater plateau length. From the current– voltage curves obtained at the front and back sides of the used AMX membrane, it can be inferred that the membrane surface plays an important role in the transport phenomena of an ion-exchange membrane system. Figure 8 shows that the limiting current density for the virgin AMX membrane is slightly smaller than that for the used AMX membrane. This is due to the decrease in permselectivity of the used AMX membrane. The permselectivities of the used AMX and virgin AMX membranes were measured using Eq. (1), and the values were 0.953 and 0.961, respectively. The transport number of the used AMX was estimated to be slightly lower than that of the virgin AMX membrane. The permselectivity of the ion-exchange membrane is affected by the concentration of the counterions in the membrane phase [32]. The concentration of counterions of the virgin membrane may be greater than that of the used membrane because of the electroneutrality requirement. The difference is attributed to the structural change of membrane

99

surface. In the used AMX membrane, quaternary ammonium groups, which are a strong basic group, were changed partially into weakly basic anion exchange groups. As the transport numbers for the virgin and used membranes have little difference this can be negligible. However, the effect on the permselectivity of structural change of the ion-exchange membrane caused by water dissociation can be significant if the ion-exchange membrane processes are operated at the overlimiting current condition for a long time. A primary effect of water dissociation is to increase the resistance of a weakly basic membrane. Most commercially available anion-exchange membranes contain quaternary ammonium (–R3 N+ ) groups (R = alkyl or aryl group) as functional groups [2]. These groups represent strong bases and are completely ionized over the entire pH range. When the tertiary amine groups exist at the membrane surface, however, the increase in membrane resistance can be caused by the conversion of amine groups from the charged to the neutral form. This feature was well explained in Fig. 9, showing the membrane resistance at different pH values. After the immersion of the virgin and used AMX membranes in 0.1 M NaCl solutions of various pH values for 24 h, the membrane resistances were measured by a clip cell. It is observed that the resistance of the used AMX membrane begins to increase at pH 9 and remains constant from pH 11. The result of membrane resistance with the solution pH for the used AMX membrane is coincident with the pKa values of tertiary amines, which are in the range of 9–11. In contrast, the membrane resistance for the virgin AMX membrane does not increase with the solution pH because the quaternary ammoniums are completely ionized over the pH range. The decrease in resistance at pH 12 is due to the hydroxyl ions added to adjust the solution pH. Although the concentration of hydroxyl ions (10−2 M) is an order of magnitude smaller than that of chloride ions (10−1 M) at pH 12, the mobility of hydroxyl ions is about three times higher than that of chloride ions, resulting in a decrease in the membrane resistance [33].

Fig. 9. Membrane resistance of the used AMX membrane in various pH solutions.

100

J.-H. Choi, S.-H. Moon / Journal of Colloid and Interface Science 265 (2003) 93–100

4. Conclusion The structural change of an ion-exchange membrane at high electric field was investigated by comparing the FTIR spectra between the virgin membrane and that used at the overlimiting current. An argument about the mechanism of water dissociation originated from the observation that the rate of water dissociation with anion-exchange membranes is much greater than cation-exchange membranes. Through this study it was proved that the enhanced water dissociation of the anion-exchange membrane is caused by the structural change from quaternary ammonium to tertiary amines at the membrane surface, which results from the increase in pH and temperature in the space charge layer. In addition, it was observed that the membrane resistance, permselectivity, and plateau length of the current–voltage curve were affected by the converted tertiary amine groups, depending on the solution pH. For economic reasons, it is desired to operate electrodialysis systems at a current density as high as possible in order to obtain fast desalination with a given membrane area. Many researchers have observed that the overlimiting current is still carried dominantly by the electrolyte, and the loss of permselectivity or water dissociation is not responsible for the overlimiting current [10,18,22,27–29]. Those findings may lead to the design and operation of a high-currentdensity electrodialysis unit in the future. In this study it is considered that the conventional anion-exchange membrane containing quaternary ammonium groups is not an adequate electrodialysis membrane when the processes are operated in the overlimiting current region. To overcome the problem it is recommended that an alternative functional group, which is stable under the high electric field condition, be used for the required positive charges of an anion-exchange membrane.

Acknowledgment This work was supported by the National Research Laboratory (NRL) Program of the Korea Institute of Science and Technology Evaluation and Planning (Project 2000-NNL-01-C-185).

References [1] S.J. Parulekar, J. Membr. Sci. 148 (1998) 91. [2] M. Mulder, Basic Principles of Membrane Technology, Kluwer Academic, Dordrecht, 1996. [3] G.S. Solt, Membrane Separation Process, Elsevier, Amsterdam, 1976. [4] B.T. Batchelder, FIL-IDF Bull. 212 (1987) 84. [5] S. Itoi, I. Nakamura, T. Kawahara, Desalination 32 (1980) 383. [6] S. Novalic, F. Jagschits, J. Okwor, K.D. Kulbe, J. Membr. Sci. 108 (1995) 201. [7] P.J. Moon, S.J. Parulekar, S.-P. Tsai, J. Membr. Sci. 141 (1998) 75. [8] E.G. Lee, S.H. Moon, Y.K. Chang, I.K. Yoo, H.N. Chang, J. Membr. Sci. 145 (1998) 53. [9] K. Lee, J. Hong, J. Membr. Sci. 75 (1992) 107. [10] J.J. Krol, M. Wessling, H. Strathmann, J. Membr. Sci. 162 (1999) 145. [11] A.J. Makai, J.C.R. Turner, Trans. IchemE 60 (1982) 88. [12] V.M. Barragan, C. Ruiz-Bauza, J. Colloid Interface Sci. 205 (1998) 365. [13] V. Mavrov, W. Pusch, O. Kominek, S. Wheelwright, Desalination 91 (1993) 225. [14] R. Simons, Electrochim. Acta 29 (1984) 151. [15] R. Simons, Nature 280 (1979) 824. [16] R. Simons, Desalination 28 (1979) 41. [17] I. Rubinstein, A. Warshawsky, L. Schechtman, O. Kedem, Desalination 51 (1984) 55. [18] J.H. Choi, H.J. Lee, S.H. Moon, J. Colloid Interface Sci. 238 (2001) 188. [19] L. Jialin, W. Yazhen, Y. Changying, L. Guangdou, S. Hong, J. Membr. Sci. 147 (1998) 247. [20] H. Strathmann, J.J. Krol, H.-J. Rapp, G. Eigenberger, J. Membr. Sci. 125 (1997) 123. [21] Tokuyama Soda Inc., Neosepta Ion Exchange Membranes, Product brochure,1998. [22] J.H. Choi, Transport Phenomena in Ion-Exchange Membrane at Under- and Over-Limiting Current Regions, Ph.D. thesis, K-JIST, Gwangju, Korea, 2002. [23] F.A. Carey, Organic Chemistry, McGraw–Hill, New York, 1992. [24] A.A. Skoog, J.J. Leary, in: Principles of Instrumental Analysis, Saunders, New York, 1992. [25] B.A. Cooke, Electrochim. Acta 4 (1961) 179. [26] J.A. Manzanares, K. Kontturi, S. Mafe, V.M. Aguilella, J. Pelicer, Acta Chem. Scand. 45 (1991) 115. [27] I. Rubinstein, Phys. Fluids 3 (1991) 2301. [28] I. Rubinstein, E. Staude, O. Kedem, Desalination 69 (1988) 101. [29] I. Rubinstein, F. Maletzki, J. Chem. Soc. Faraday Trans. 87 (1991) 2079. [30] J. Kerres, in: Advanced Course on Electro-membrane Processes, Lecture Note, Univ. Stuttgart, 2000. [31] T. Sata, M. Tsujimoto, T. Yamaguchi, K. Matsusaki, J. Membr. Sci 112 (1996) 161. [32] F. Helfferich, Ion Exchange, Dover, New York, 1962. [33] W.J. Moore, Physical Chemistry, Prentice Hall, Englewood Cliffs, 1972.