- Email: [email protected]
Effect of calcium and carbonate concentrations on anionic membrane fouling during electrodialysis
Journal of Colloid and Interface Science 296 (2006) 242–247 www.elsevier.com/locate/jcis
Effect of calcium and carbonate concentrations on anionic membrane fouling during electrodialysis Monica Araya-Farias, Laurent Bazinet ∗ Institute of Nutraceuticals and Functional Foods (INAF), Department of Food Sciences and Nutrition, Pavillon Paul Comtois, Université Laval, Sainte-Foy (QC), G1K 7P4, Canada Received 30 June 2005; accepted 19 August 2005 Available online 22 September 2005
Abstract A previous study on electrodialysis of calcium and carbonate high concentration solutions demonstrated that calcium migrated through the cation-exchange membrane (CEM) was blocked by the anion-exchange membrane (AEM) where it formed another fouling. The aim of the present work was to complete the identification of the deposit formed on AEM during electrodialysis and to characterize its physical structure at the interface of the membrane. No fouling was found on the anionic membranes treated without calcium chloride in presence of sodium carbonate, while membranes used during ED process of solutions containing calcium chloride and sodium carbonate were slightly fouled. A thin layer of precipitates was observed on the anionic membrane surface. The appearance of precipitates was typical of a crystalline substance. The size and form of crystal increased in proportion to the concentration of calcium chloride in solution. Large and cubic crystals were the best defined on the membrane treated at 1600 mg/L of CaCl2 . The precipitate was identified as calcium hydroxide. However, this fouling was not found to affect significantly the electrical conductivity and the thickness of the membranes. Furthermore, the fouling formed was reversible. © 2005 Elsevier Inc. All rights reserved. Keywords: Electrodialysis; Fouling; Anionic membranes; Sodium carbonate; Calcium chloride
1. Introduction Electrodialysis (ED) is a membrane process in which ionized species are transported through an ion-exchange membrane under the influence of an electric field. The ED processes have been used for demineralization for more than two decades. In the food industry, one of the principal applications is the demineralization of whey [1]. However, the major factor limiting the use of ED in many applications is membrane fouling. Fouling is the accumulation of undesired deposits on the membrane surface which affects its performance. The causes of these deposits can be multiple: scaling (salt precipitation), clogging (deposit of macromolecules, colloids) or poisoning (blocking of the membranes by tensioactive agents) [2]. Studies carried-out on membrane fouling reported that anionic membranes were more affected by this phenomenon [3,4]. Membrane fouling is * Corresponding author. Fax: +1 (418) 656 3353.
E-mail address: [email protected] (L. Bazinet). 0021-9797/$ – see front matter © 2005 Elsevier Inc. All rights reserved. doi:10.1016/j.jcis.2005.08.040
typically characterized by an increase in the membrane resistance and can also cause a loss in the permselectivity of the membranes [4,5]. These changes in membrane properties often continue throughout the ED process and sometimes require extensive cleaning or replacement of the membranes. This has economic detrimental effects on the process. In our previous work [6], we showed that in the absence of sodium carbonate in the solution, no fouling were visually observed on anion-exchange membranes (AEM) and only at 1600 mg/L CaCl2 on cation-exchange membrane (CEM), while at only 800 mg/L CaCl2 with sodium carbonate, a deposit was observed on both membranes. The precipitates on the membrane surface increased in proportion to the concentration of the salts in the solution treated. It was proposed that due to the basic pH on the side of the membrane in contact with the NaCl solution, the calcium would precipitate while the calcium migrated through the CEM was blocked by the AEM where it formed another fouling. On the side of the CEM membrane in contact with the NaCl solution, the precipitate was identified recently as calcium hydroxide [7].
M. Araya-Farias, L. Bazinet / Journal of Colloid and Interface Science 296 (2006) 242–247
The aim of the present study was to complete the identification of the deposit observed on the AEM membrane in contact with the NaCl basic solution and to characterize its physical structure at the interface of the membrane. The fouling was characterized by membrane parameters, optical microscopy and microscopic membrane surface elemental analysis and mapping.
243
Table 1 Ionic composition and concentration of the different salts in the solutions treated by electrodialysis Salt
Solution A
B
C
D
E
F
KCl (mg/L) CaCl2 (mg/L) Na2 CO3 (mg/L)
800 400 0
800 800 0
800 1600 0
800 400 1000
800 800 1000
800 1600 1000
2. Experimental 2.1. Material Neosepta AMX-SB anion-exchange membranes from Tokuyama Soda Ltd. (Tokyo, Japan) used in a previous work [6] were analyzed. Electrodialysis were carried out in batch process using a current density of 40 mA/cm2 and after reaching 30 V, the voltage was maintained constant at 30 V, in order to not surpass the total power of the power supply (Model HPD 3010SX, Xantrex, Burnaby, BC, Canada). In preliminary tests we observed that whatever the initial solution compositions, the limiting current density was never reached for anode/cathode voltage differences up 30 V at the beginning of the process. The electrodialysis was stopped after the pH reached 4.0. The initial pH of salt solution was adjusted at 6.5. CaCl2 ·2H2 O, NaCl, and KCl were obtained from Laboratoire MAT (Québec, Canada). Na2 CO3 was obtained from BDH (BDH Inc., Toronto, Canada). 2.2. Electrodialysis cell The electrodialysis cell was a MicroFlow type cell (ElectroCell AB, Täby, Sweden) with two Neosepta CMX-S cationic membranes and two Neosepta AMX-SB anionic membranes (Tokuyama Soda Ltd., Tokyo, Japan). This arrangement defines three closed loops containing the model salt solution (300 mL), a 2 g/L aqueous KCl solution (300 mL), and a 20 g/L NaCl solution (500 mL) [6]. The anode, a dimensionally stable electrode (DSA), and the cathode, a 316 stainless-steel electrode were supplied with the MicroFlow cell. The membranes tested were both in contact with the model salt solution on one side and with the NaCl solution on the other side. The pH of the KCl and NaCl solutions were maintained constant at 11.0–12.0. 2.3. Protocol Original AMX-SB membrane characteristics were measured on unused membranes freshly cut from the sheet and were compared to those of membranes used for electrodialysis. The ionic composition and concentration of the different salt solutions treated by ED are presented in Table 1. For each condition, four ED treatments were carried-out to ensure the fouling or not of the membrane. After the four ED, photographs were taken with a binocular microscope, and the membrane characteristics were determined by measuring their electrical conductivity and thickness. The surface of the membrane in contact with the solution, and corresponding to the electrode effective surface (10 cm2 ) was then cut and separated into two pieces of 5 cm2 .
The first 5 cm2 of membrane was soaked in 1.0 N HCl solution overnight to evaluate the reversibility of the fouling. On the second 5-cm2 membrane sample, electron microscopy photographs and elemental analysis were performed to complete the physical and chemical measurements. 2.4. Analysis methods 2.4.1. Membrane electrical conductivity The membrane electrical conductivity was measured with specially designed cell (conductivity clip) from the Laboratoire des Matériaux Echangeurs d’Ions (Créteil, France) as previously described by Bazinet and Araya-Farias [7]. The membrane electrical conductivity κ (mS/cm) was calculated as follows [8,9]: κ=
l , Rm A
where l is the membrane thickness (cm), Rm the transversal electric resistance of the membrane (), and A the electrode area (1 cm2 ). 2.4.2. Thickness The membrane thickness was measured using a Mitutoyo Corp. IDC type digimatic indicator with absolute encoder (Model ID-C112 EB, Kanagawa-Ken, Japan), specially devised for plastic film thickness measurements, with a resolution of 1 µm and a range of 12.7–0.001 mm. The digimatic indicator was equipped with a 10-mm diameter flat contact point. The membrane thickness value was averaged from six to ten measurements at different locations on the membrane. 2.4.3. Optical microscopy Photographs were taken at 100× magnitude on a binocular microscope (Laborlux S, Ernst Leitz Wetzlar GmbH., Wetzlar, Germany) equipped with a color video camera (model hyper HAD, CCD-IRIS/RGB, Sony, Toronto, ON, Canada). The images were processed with Matrox Inspector software (version 3.1, Matrox Electronics Systems Ltd., Dorval, QC, Canada). 2.4.4. Electron microscopy and elemental analysis Images were taken on uncoated sample at 5 kV with an electron microscope (S-3000 N, Hitachi, Japan). Elemental analysis were performed on an X-ray energy-dispersive spectrometer (EDS) (Oxford Link Isis, Oxford Instruments Microanalysis Group, Concord, MA, USA). The EDS conditions were 20-kV accelerating voltage, 14.7-mm working distance.
244
M. Araya-Farias, L. Bazinet / Journal of Colloid and Interface Science 296 (2006) 242–247
3. Results and discussion 3.1. Physical and chemical membrane parameters: Membrane electrical conductivity and thickness 3.1.1. Membrane electrical conductivity Membrane fouling is typically registered as an increase in the membrane resistance or a decrease in the membrane electrical conductivity. The original membrane conductivity was 4.97 ± 0.12 mS/cm. Similar results of electrical conductivity were reported previously in the literature for original AMX membranes [10,11]. For membranes used during ED treatments at different salt concentrations, a similar small decrease in the conductivity was observed whatever the CaCl2 concentration and the presence or not of sodium carbonate. The averaged values of electrical conductivity for the membranes A, B, and C were 4.10 ± 0.14 mS/cm in comparison to 4.18 ± 0.24 mS/cm for membranes D, E, and F. No significant difference in the electrical conductivity of membranes was observed after HCl treatments overnight, and the averaged values were 4.10 ± 0.13 mS/cm for treatments A, B, and C, and 4.18 ± 0.23 mS/cm for treatments D, E, and F. According to these results, the fouling effect was insignificant and appeared to not affect the membrane integrity, since the electrical conductivity was similar to the one of membranes treated without sodium carbonate and with the one of membranes soaked overnight in HCl. However, after 4 ED treatments, when the cell was dismounted and the membrane inspected, a thin layer of precipitates was observed on the membranes surface treated with sodium carbonate [6]. Furthermore, the small decrease in conductivity observed between original membrane freshly cut from the sheet and used membranes would be due to the replacement of original counterion by new counterions less mobile or/and by the increase of the coion concentration in the membrane. A fraction of current is transported by the coion, which migrates in opposite direction and this phenomenon is called escape of the coion [8]. Without sodium carbonate present in the solution and whatever the CaCl2 concentration, due to the high difference in concentration between NaCl solution and model solution, calcium and potassium ions migrated through the anionic membrane. Then, decrease in conductivity would be due to this increase of the coion concentration, as observed by Lebrun et al. [8] and confirmed by the small increase in potassium and calcium concentration observed in these membranes [6]. While, when sodium carbonate was present in the model solution, the decrease in the membrane conductivity would be associated with the replacement of part of the Cl− counterion by ions from sodium carbonate present in the solution according to their equilibrium constant [12]. In fact, in its original form the electrical charges of the AMX-SB membrane are neutralized by Cl− ions, while during the treatment with sodium carbonate part of these mobile charges will be exchanged for HCO− 3 ions which are less mobile (40 vs 68 cm2 /V s) and consequently less conductive. It was demonstrated that the membrane conductivity varies with the limiting ionic conductivity of counterions [13,14].
3.1.2. Thickness There is no difference in the thickness for the membrane treated with solution containing no sodium carbonate in comparison with the original membrane. In fact, the average value for the membranes A, B, and C was 0.132 ± 0.002 mm while the thickness for the original membrane was 0.131 ± 0.002 mm. However, in presence of sodium carbonate, the thickness of the membranes increased quite proportionally to the increase in CaCl2 from 400 to 1600 mg/L. The thickness values for treatments D, E, and F were 0.138 ± 0.003, 0.139 ± 0.003, and 0.145 ± 0.006 mm, respectively. At 1600 mg/L CaCl2 , the increase in thickness was significant (11% increase in comparison with original membrane), and would correspond to a fouling of 0.0142 mm. After soaking overnight in HCl, the average values of thickness were 0.131 ± 0.001 mm for membranes A, B, and C, and 0.132 ± 0.003 mm for membranes D, E, and F. These values were very close to those of original membrane and suggested that the fouling formed on the membranes would be reversible and soaking in HCl solution allowed the recovery of membrane integrity. Thickness values for original AMX membranes were in accordance with results already published in the literature [10, 11]. The fact that no change in the thickness for membranes used during ED treatments of solutions without sodium carbonate was in accordance with the fact that no fouling was visually observed and small increases in cation concentrations were determined in AEM membranes [6]. For membranes used with solutions containing sodium carbonate, the increase in the thickness of the membranes was also in accordance with these previous results; a visual deposit was observed at 800 and 1600 mg/L of CaCl2 on the AEM membrane and a significant increase from 0.74 to 10.27 mg/g dry membrane in calcium concentration was reported in the membranes. Furthermore, in comparison with our previous work [7] on the cationic membranes used during these experiments, the deposit thickness values for the anionic membranes was lower than those previously reported for the CMX-S membrane: 0.014 vs 0.052 mm after 4 ED treatments of solution containing sodium carbonate and 1600 mg/L CaCl2 . The fouling of the CEM membrane was 3 times higher, while it is well known that anionic membrane are more susceptible to fouling than cationic membrane. However, in the literature the susceptibility to fouling of AEM is due to the fact that most of the colloids present in natural water are negatively charges [3,4,15]. 3.2. Membrane physical properties and fouling identification Whatever the side, the original AMX-SB membrane presented a plane and clean surface, and at a higher magnification only small imperfection on the surface of the membrane has been observed (Figs. 1A and 1B). For membranes used during ED in different conditions of salt concentrations, only results concerning the surface of the AMX-SB membranes in contact with the basified NaCl stream was discussed, since no fouling were observed visually and with the optical microscope on the other side.
M. Araya-Farias, L. Bazinet / Journal of Colloid and Interface Science 296 (2006) 242–247
(A)
245
(B)
Fig. 1. Original AMX-SB membrane photographs at (A) optical microscope magnitude of 100× and (B) surface electron microscope magnitude of 45×.
(A)
(B)
(C)
(D)
(E)
(F)
Fig. 2. Optical microscope photographs (magnitude of 100×) of basified side of AMX-SB membranes used during electrodialysis of solutions containing 800 mg/L KCl and (A) 400 mg/L CaCl2 , (B) 800 mg/L CaCl2 , (C) 1600 mg/L CaCl2 , (D) 400 mg/L CaCl2 and 1000 mg/L Na2 CO3 , (E) 800 mg/L CaCl2 and 1000 mg/L Na2 CO3 , (F) 1600 mg/L CaCl2 and 1000 mg/L Na2 CO3 .
The electronic and optical images of membranes treated without carbonate showed a clean surface (Figs. 2A–2C and 3A–3C). For those membranes no deposit or fouling was found and their appearance was similar to the one observed for the original membrane AMX-SB (Figs. 1A and 1B). In fact, no real difference between those membranes was observed. This was confirmed by elemental analysis performed at higher magnification. The original membrane spectra showed the presence of three major elements: C: 82.24, Cl: 15.18, and O: 2.87% of relative percentage (see Table 2). In comparison, the spectra of membranes A, B, and C showed the presence of the same elements. The averaged relative percentage for C, Cl, and O were 82.19, 14.66, and 3.40%, respectively (Table 3).
Table 2 Elemental analysis results (%) of original AMX-SB membrane Element
Relative percentage (%)
Mg K Ca Na S Cl O C
0.00 0.00 0.02 0.00 0.00 15.18 2.87 82.24
For membranes used during ED treatment of solutions containing sodium carbonate, the optical microscope photographs
246
M. Araya-Farias, L. Bazinet / Journal of Colloid and Interface Science 296 (2006) 242–247
(A)
(B)
(C)
(D)
(E)
(F)
Fig. 3. Electronic microscope photographs (magnitude of 350×) of basified side of AMX-SB membranes used during electrodialysis of solutions containing 800 mg/L KCl and (A) 400 mg/L CaCl2 , (B) 800 mg/L CaCl2 , (C) 1600 mg/L CaCl2 , (D) 400 mg/L CaCl2 and 1000 mg/L Na2 CO3 , (E) 800 mg/L CaCl2 and 1000 mg/L Na2 CO3 , (F) 1600 mg/L CaCl2 and 1000 mg/L Na2 CO3 .
Table 3 Elemental analysis results (%) of AMX-SB membranes used during electrodialysis of solutions containing 800 mg/L KCl and (A) 400 mg/L CaCl2 , (B) 800 mg/L CaCl2 , (C) 1600 mg/L CaCl2 , (D) 400 mg/L CaCl2 and 1000 mg/L Na2 CO3 , (E) 800 mg/L CaCl2 and 1000 mg/L Na2 CO3 , (F) 1600 mg/L CaCl2 and 1000 mg/L Na2 CO3 Element Mg K Ca Na S Cl O C
Membrane A
B
C
D
E
F
0.00 0.01 0.01 0.00 0.00 15.56 2.50 82.17
0.00 0.00 0.00 0.00 0.00 14.39 3.18 82.72
0.00 0.01 0.01 0.00 0.00 14.02 4.53 81.69
0.01 0.01 3.02 0.00 0.00 10.38 26.72 60.10
0.00 0.00 2.31 0.00 0.00 12.52 17.24 68.15
0.00 0.00 2.79 0.00 0.00 9.94 21.08 66.37
showed the presence of precipitates dispersed on the surface of the membranes in contact with the basified NaCl solution (Figs. 2D–2F and 3D–3F). The elemental analysis showed that the surface of those membranes was composed of three main elements: carbon, chloride, and oxygen. A marked difference in the relative percentage of oxygen was observed in comparison with the original membrane (Tables 2 and 3); the relative percentage of oxygen increased 10.0 times for membrane D, 6.8 times for membrane E, and 8.4 times for membrane F. Furthermore, calcium was also present on the membranes surface. The calcium percentage was 3.02, 2.31, and 2.79% for membrane D, E, and F, respectively (Table 3).
The precipitates on membranes AMX-SB would be mainly composed of calcium and oxygen since those elements were present at the same place on the surface of membranes. This precipitate would be identified as calcium hydroxide. However, this identification is limited by the fact that the elemental analysis does not allow the identification of the presence of H. According to the electronic images the appearance of precipitates was typical of a crystalline substance. When the concentration of CaCl2 in solution increase, the size and form of the crystals become more apparent (Figs. 2D–2F and 3D–3F). Morphologically, large crystals very well defined where observed on membrane F (1600 mg/L CaCl2 ). The structure of crystals was characteristic of a rhombohedra or cubic substance. It appears from these results that the fouling observed on the surface of the anionic membrane in contact with the basified NaCl solution during treatment of solution containing Na2 CO3 and CaCl2 was formed of Ca(OH)2 . As suggested previously [6], calcium migrated through the cationic membrane could make contact with OH− ions from the basified solution and formed an insoluble deposit of Ca(OH)2 on the surface of anionic membranes. The crystals of precipitates grow gradually with the time and block the surface of AMX-SB membrane. It was documented in the literature that the alkaline-earth-metal ion such Ca2+ , Mg2+ may form precipitates when they are in basic solutions [16]. Ogata et al. [17] showed that the alkaline earth ions precipitate in the membrane as hydroxide and that the accumulation of these precipitates was a major factor in the degradation of the membrane. Calcium was encountered
M. Araya-Farias, L. Bazinet / Journal of Colloid and Interface Science 296 (2006) 242–247
the most harmful followed by Mg, Sr, and Ba. In this work, during ED treatment, the pH of the basified compartment was maintained constant at pH 11–12 with 1.0 N NaOH solution. Furthermore, after soaking with HCl the precipitates were dissolved and it would be due to the reaction of Ca(OH)2 with HCl which generate CaCl2 which is a soluble substance. Concerning the morphological characteristics of the precipitate formed, Tomazic et al. [18] noted that the morphological quality of crystalline precipitates of calcium hydroxide from aqueous solutions is dependent upon variables, such as supersaturation, chemical composition of the solution, including impurities and the method used to induce precipitation. Specifically, the growing of crystals results from their particles solution environment. Tomazic et al. [19] have observed a pronounced increase on growth kinetics of Ca(OH)2 with an increase in NaCl concentration at constant supersaturation (σ = 0.44 ± 0.01). Indeed, an increase of NaCl concentration increased the overall crystallization rate and after 10 min the weights of crystal were 90, 200, and 330 mg in 0.1, 0.2, and 0.3 M NaCl, respectively. On this account, similar behavior would be expected in our work when the concentration of NaCl in the concentrating compartment is approximately 0.3 M and is expected to be related with the growth of crystals especially when the CaCl2 concentration is highest (Figs. 3D–3F). Furthermore, in this work, an important difference was observed between the appearance of fouling observed on the anionic membranes and that of the cationic membranes treated with sodium carbonate [7]. The deposits on the anionic membranes had a cubic defined form while big nonmorphologically defined spot of deposits were observed on the surface of cationic membranes. Once again, as observed previously for CEM [7] no formation of calcium carbonate was observed despite the presence of high concentration of sodium carbonate and as previously suggested the difference in fouling of the membrane treated with and without sodium carbonate could be explained by the different treatment duration (68 vs 21 min, respectively). Carbonate has a high buffer capacity, and the time to reach pH 4.0 was then longer than the one without carbonate where there was no buffer capacity. 4. Conclusion In the present work, the presence of fouling was only observed on the membranes treated with a solution containing 400, 800, and 1600 mg/L of CaCl2 in presence of sodium carbonate. The appearance of precipitates on the anionic membranes was typical of a cubic and crystalline substance. The largest crystals were best defined when the membranes were treated with 1600 mg/L of CaCl2 . However, the cubic form of precipitates was only observed on the anionic membranes. In
247
fact, in our previous work, the cationic membranes showed in general a spotted and blotchy surface. The crystals would be identified as calcium hydroxide. However, this fouling was not found to affect significantly the conductivity and the thickness of the membranes and it was easily dissolved after the soaking with HCl. The analysis of anionic membranes used during this study complete the evaluation of the impact of the calcium and carbonate concentrations during ED treatments on the anionic and cationic membranes. Acknowledgments The authors thank Diane Montpetit from AAC-FRDC (SaintHyacinthe, QC, Canada) for her technical help on electron microscopy and elemental analysis. The financial support of the Natural Sciences and Engineering Research Council of Canada (NSERC) is acknowledged. References [1] B. Saint-Pierre, in: Symposium sur les applications alimentaires de l’électrodialyse. Fondation des Gouverneurs, Saint-Hyacinthe, Québec, 23 et 24 Mars 1999. [2] C. Gavach, in: Symposium sur les applications alimentaires de l’électrodialyse. Fondation des Gouverneurs, Saint-Hyacinthe, Québec, 23 et 24 Mars 1999. [3] E. Korngold, F. De Körösy, R. Rahav, M.F. Taboch, Desalination 8 (1970) 195. [4] V. Lindstrand, G. Sundström, A.S. Jönsson, Desalination 128 (2000) 91. [5] V. Lindstrand, A.S. Jönsson, G. Sundström, Desalination 130 (2000) 73. [6] L. Bazinet, M. Araya-Farias, J. Colloid Interface Sci. 286 (2005) 639. [7] L. Bazinet, M. Araya-Farias, J. Colloid Interface Sci. 281 (2005) 188. [8] L. Lebrun, E. Da Silva, G. Pourcelly, M. Métayer, J. Membr. Sci. 227 (2003) 95. [9] R. Lteif, L. Dammark, C. Larchet, B. Auclair, Eur. Polym. J. 35 (1999) 1187. [10] N. Pismenskaya, E. Laktionov, V. Nikonenko, A. El Attar, B. Bouclair, G. Pourcelly, J. Membr. Sci. 181 (2001) 185. [11] A. Ellatar, A. Elmidaoui, N. Pismenskaia, C. Gavach, G. Pourcelly, J. Membr. Sci. 143 (1998) 249. [12] A. Lounis, C. Gavach, J. Membr. Sci. 54 (1990) 63. [13] P.H. Barry, J.W. Lynch, J. Membr. Biol. 121 (1991) 101. [14] G. Milazzo, in: Électrochimie. Bases théoriques. Applications analytiques. Électrochimie des colloïdes, vol. 1, Dunod, Paris, 1969, p. 375. [15] T.A. Davis, J.D. Genders, D. Pletcher, The Electrochemical Consultancy, Romsey, 1997, p. 189. [16] K. Okada, M. Mamoru Tomita, Y. Tamura, Milchwissenschaft 31 (10) (1976) 602. [17] Y. Ogata, T. Kojima, S. Uchiyama, M. Yasuda, F. Hine, J. Electrochem. Soc. (1989) 136. [18] B. Tomazic, R. Mohanty, M. Tadros, J. Estrin, J. Cryst. Growth 75 (1986) 339. [19] B. Tomazic, R. Mohanty, M. Tadros, J. Estrin, J. Cryst. Growth 75 (1986) 329.
Effect of calcium and carbonate concentrations on anionic membrane fouling during electrodialysis Monica Araya-Farias, Laurent Bazinet ∗ Institute of Nutraceuticals and Functional Foods (INAF), Department of Food Sciences and Nutrition, Pavillon Paul Comtois, Université Laval, Sainte-Foy (QC), G1K 7P4, Canada Received 30 June 2005; accepted 19 August 2005 Available online 22 September 2005
Abstract A previous study on electrodialysis of calcium and carbonate high concentration solutions demonstrated that calcium migrated through the cation-exchange membrane (CEM) was blocked by the anion-exchange membrane (AEM) where it formed another fouling. The aim of the present work was to complete the identification of the deposit formed on AEM during electrodialysis and to characterize its physical structure at the interface of the membrane. No fouling was found on the anionic membranes treated without calcium chloride in presence of sodium carbonate, while membranes used during ED process of solutions containing calcium chloride and sodium carbonate were slightly fouled. A thin layer of precipitates was observed on the anionic membrane surface. The appearance of precipitates was typical of a crystalline substance. The size and form of crystal increased in proportion to the concentration of calcium chloride in solution. Large and cubic crystals were the best defined on the membrane treated at 1600 mg/L of CaCl2 . The precipitate was identified as calcium hydroxide. However, this fouling was not found to affect significantly the electrical conductivity and the thickness of the membranes. Furthermore, the fouling formed was reversible. © 2005 Elsevier Inc. All rights reserved. Keywords: Electrodialysis; Fouling; Anionic membranes; Sodium carbonate; Calcium chloride
1. Introduction Electrodialysis (ED) is a membrane process in which ionized species are transported through an ion-exchange membrane under the influence of an electric field. The ED processes have been used for demineralization for more than two decades. In the food industry, one of the principal applications is the demineralization of whey [1]. However, the major factor limiting the use of ED in many applications is membrane fouling. Fouling is the accumulation of undesired deposits on the membrane surface which affects its performance. The causes of these deposits can be multiple: scaling (salt precipitation), clogging (deposit of macromolecules, colloids) or poisoning (blocking of the membranes by tensioactive agents) [2]. Studies carried-out on membrane fouling reported that anionic membranes were more affected by this phenomenon [3,4]. Membrane fouling is * Corresponding author. Fax: +1 (418) 656 3353.
E-mail address: [email protected] (L. Bazinet). 0021-9797/$ – see front matter © 2005 Elsevier Inc. All rights reserved. doi:10.1016/j.jcis.2005.08.040
typically characterized by an increase in the membrane resistance and can also cause a loss in the permselectivity of the membranes [4,5]. These changes in membrane properties often continue throughout the ED process and sometimes require extensive cleaning or replacement of the membranes. This has economic detrimental effects on the process. In our previous work [6], we showed that in the absence of sodium carbonate in the solution, no fouling were visually observed on anion-exchange membranes (AEM) and only at 1600 mg/L CaCl2 on cation-exchange membrane (CEM), while at only 800 mg/L CaCl2 with sodium carbonate, a deposit was observed on both membranes. The precipitates on the membrane surface increased in proportion to the concentration of the salts in the solution treated. It was proposed that due to the basic pH on the side of the membrane in contact with the NaCl solution, the calcium would precipitate while the calcium migrated through the CEM was blocked by the AEM where it formed another fouling. On the side of the CEM membrane in contact with the NaCl solution, the precipitate was identified recently as calcium hydroxide [7].
M. Araya-Farias, L. Bazinet / Journal of Colloid and Interface Science 296 (2006) 242–247
The aim of the present study was to complete the identification of the deposit observed on the AEM membrane in contact with the NaCl basic solution and to characterize its physical structure at the interface of the membrane. The fouling was characterized by membrane parameters, optical microscopy and microscopic membrane surface elemental analysis and mapping.
243
Table 1 Ionic composition and concentration of the different salts in the solutions treated by electrodialysis Salt
Solution A
B
C
D
E
F
KCl (mg/L) CaCl2 (mg/L) Na2 CO3 (mg/L)
800 400 0
800 800 0
800 1600 0
800 400 1000
800 800 1000
800 1600 1000
2. Experimental 2.1. Material Neosepta AMX-SB anion-exchange membranes from Tokuyama Soda Ltd. (Tokyo, Japan) used in a previous work [6] were analyzed. Electrodialysis were carried out in batch process using a current density of 40 mA/cm2 and after reaching 30 V, the voltage was maintained constant at 30 V, in order to not surpass the total power of the power supply (Model HPD 3010SX, Xantrex, Burnaby, BC, Canada). In preliminary tests we observed that whatever the initial solution compositions, the limiting current density was never reached for anode/cathode voltage differences up 30 V at the beginning of the process. The electrodialysis was stopped after the pH reached 4.0. The initial pH of salt solution was adjusted at 6.5. CaCl2 ·2H2 O, NaCl, and KCl were obtained from Laboratoire MAT (Québec, Canada). Na2 CO3 was obtained from BDH (BDH Inc., Toronto, Canada). 2.2. Electrodialysis cell The electrodialysis cell was a MicroFlow type cell (ElectroCell AB, Täby, Sweden) with two Neosepta CMX-S cationic membranes and two Neosepta AMX-SB anionic membranes (Tokuyama Soda Ltd., Tokyo, Japan). This arrangement defines three closed loops containing the model salt solution (300 mL), a 2 g/L aqueous KCl solution (300 mL), and a 20 g/L NaCl solution (500 mL) [6]. The anode, a dimensionally stable electrode (DSA), and the cathode, a 316 stainless-steel electrode were supplied with the MicroFlow cell. The membranes tested were both in contact with the model salt solution on one side and with the NaCl solution on the other side. The pH of the KCl and NaCl solutions were maintained constant at 11.0–12.0. 2.3. Protocol Original AMX-SB membrane characteristics were measured on unused membranes freshly cut from the sheet and were compared to those of membranes used for electrodialysis. The ionic composition and concentration of the different salt solutions treated by ED are presented in Table 1. For each condition, four ED treatments were carried-out to ensure the fouling or not of the membrane. After the four ED, photographs were taken with a binocular microscope, and the membrane characteristics were determined by measuring their electrical conductivity and thickness. The surface of the membrane in contact with the solution, and corresponding to the electrode effective surface (10 cm2 ) was then cut and separated into two pieces of 5 cm2 .
The first 5 cm2 of membrane was soaked in 1.0 N HCl solution overnight to evaluate the reversibility of the fouling. On the second 5-cm2 membrane sample, electron microscopy photographs and elemental analysis were performed to complete the physical and chemical measurements. 2.4. Analysis methods 2.4.1. Membrane electrical conductivity The membrane electrical conductivity was measured with specially designed cell (conductivity clip) from the Laboratoire des Matériaux Echangeurs d’Ions (Créteil, France) as previously described by Bazinet and Araya-Farias [7]. The membrane electrical conductivity κ (mS/cm) was calculated as follows [8,9]: κ=
l , Rm A
where l is the membrane thickness (cm), Rm the transversal electric resistance of the membrane (), and A the electrode area (1 cm2 ). 2.4.2. Thickness The membrane thickness was measured using a Mitutoyo Corp. IDC type digimatic indicator with absolute encoder (Model ID-C112 EB, Kanagawa-Ken, Japan), specially devised for plastic film thickness measurements, with a resolution of 1 µm and a range of 12.7–0.001 mm. The digimatic indicator was equipped with a 10-mm diameter flat contact point. The membrane thickness value was averaged from six to ten measurements at different locations on the membrane. 2.4.3. Optical microscopy Photographs were taken at 100× magnitude on a binocular microscope (Laborlux S, Ernst Leitz Wetzlar GmbH., Wetzlar, Germany) equipped with a color video camera (model hyper HAD, CCD-IRIS/RGB, Sony, Toronto, ON, Canada). The images were processed with Matrox Inspector software (version 3.1, Matrox Electronics Systems Ltd., Dorval, QC, Canada). 2.4.4. Electron microscopy and elemental analysis Images were taken on uncoated sample at 5 kV with an electron microscope (S-3000 N, Hitachi, Japan). Elemental analysis were performed on an X-ray energy-dispersive spectrometer (EDS) (Oxford Link Isis, Oxford Instruments Microanalysis Group, Concord, MA, USA). The EDS conditions were 20-kV accelerating voltage, 14.7-mm working distance.
244
M. Araya-Farias, L. Bazinet / Journal of Colloid and Interface Science 296 (2006) 242–247
3. Results and discussion 3.1. Physical and chemical membrane parameters: Membrane electrical conductivity and thickness 3.1.1. Membrane electrical conductivity Membrane fouling is typically registered as an increase in the membrane resistance or a decrease in the membrane electrical conductivity. The original membrane conductivity was 4.97 ± 0.12 mS/cm. Similar results of electrical conductivity were reported previously in the literature for original AMX membranes [10,11]. For membranes used during ED treatments at different salt concentrations, a similar small decrease in the conductivity was observed whatever the CaCl2 concentration and the presence or not of sodium carbonate. The averaged values of electrical conductivity for the membranes A, B, and C were 4.10 ± 0.14 mS/cm in comparison to 4.18 ± 0.24 mS/cm for membranes D, E, and F. No significant difference in the electrical conductivity of membranes was observed after HCl treatments overnight, and the averaged values were 4.10 ± 0.13 mS/cm for treatments A, B, and C, and 4.18 ± 0.23 mS/cm for treatments D, E, and F. According to these results, the fouling effect was insignificant and appeared to not affect the membrane integrity, since the electrical conductivity was similar to the one of membranes treated without sodium carbonate and with the one of membranes soaked overnight in HCl. However, after 4 ED treatments, when the cell was dismounted and the membrane inspected, a thin layer of precipitates was observed on the membranes surface treated with sodium carbonate [6]. Furthermore, the small decrease in conductivity observed between original membrane freshly cut from the sheet and used membranes would be due to the replacement of original counterion by new counterions less mobile or/and by the increase of the coion concentration in the membrane. A fraction of current is transported by the coion, which migrates in opposite direction and this phenomenon is called escape of the coion [8]. Without sodium carbonate present in the solution and whatever the CaCl2 concentration, due to the high difference in concentration between NaCl solution and model solution, calcium and potassium ions migrated through the anionic membrane. Then, decrease in conductivity would be due to this increase of the coion concentration, as observed by Lebrun et al. [8] and confirmed by the small increase in potassium and calcium concentration observed in these membranes [6]. While, when sodium carbonate was present in the model solution, the decrease in the membrane conductivity would be associated with the replacement of part of the Cl− counterion by ions from sodium carbonate present in the solution according to their equilibrium constant [12]. In fact, in its original form the electrical charges of the AMX-SB membrane are neutralized by Cl− ions, while during the treatment with sodium carbonate part of these mobile charges will be exchanged for HCO− 3 ions which are less mobile (40 vs 68 cm2 /V s) and consequently less conductive. It was demonstrated that the membrane conductivity varies with the limiting ionic conductivity of counterions [13,14].
3.1.2. Thickness There is no difference in the thickness for the membrane treated with solution containing no sodium carbonate in comparison with the original membrane. In fact, the average value for the membranes A, B, and C was 0.132 ± 0.002 mm while the thickness for the original membrane was 0.131 ± 0.002 mm. However, in presence of sodium carbonate, the thickness of the membranes increased quite proportionally to the increase in CaCl2 from 400 to 1600 mg/L. The thickness values for treatments D, E, and F were 0.138 ± 0.003, 0.139 ± 0.003, and 0.145 ± 0.006 mm, respectively. At 1600 mg/L CaCl2 , the increase in thickness was significant (11% increase in comparison with original membrane), and would correspond to a fouling of 0.0142 mm. After soaking overnight in HCl, the average values of thickness were 0.131 ± 0.001 mm for membranes A, B, and C, and 0.132 ± 0.003 mm for membranes D, E, and F. These values were very close to those of original membrane and suggested that the fouling formed on the membranes would be reversible and soaking in HCl solution allowed the recovery of membrane integrity. Thickness values for original AMX membranes were in accordance with results already published in the literature [10, 11]. The fact that no change in the thickness for membranes used during ED treatments of solutions without sodium carbonate was in accordance with the fact that no fouling was visually observed and small increases in cation concentrations were determined in AEM membranes [6]. For membranes used with solutions containing sodium carbonate, the increase in the thickness of the membranes was also in accordance with these previous results; a visual deposit was observed at 800 and 1600 mg/L of CaCl2 on the AEM membrane and a significant increase from 0.74 to 10.27 mg/g dry membrane in calcium concentration was reported in the membranes. Furthermore, in comparison with our previous work [7] on the cationic membranes used during these experiments, the deposit thickness values for the anionic membranes was lower than those previously reported for the CMX-S membrane: 0.014 vs 0.052 mm after 4 ED treatments of solution containing sodium carbonate and 1600 mg/L CaCl2 . The fouling of the CEM membrane was 3 times higher, while it is well known that anionic membrane are more susceptible to fouling than cationic membrane. However, in the literature the susceptibility to fouling of AEM is due to the fact that most of the colloids present in natural water are negatively charges [3,4,15]. 3.2. Membrane physical properties and fouling identification Whatever the side, the original AMX-SB membrane presented a plane and clean surface, and at a higher magnification only small imperfection on the surface of the membrane has been observed (Figs. 1A and 1B). For membranes used during ED in different conditions of salt concentrations, only results concerning the surface of the AMX-SB membranes in contact with the basified NaCl stream was discussed, since no fouling were observed visually and with the optical microscope on the other side.
M. Araya-Farias, L. Bazinet / Journal of Colloid and Interface Science 296 (2006) 242–247
(A)
245
(B)
Fig. 1. Original AMX-SB membrane photographs at (A) optical microscope magnitude of 100× and (B) surface electron microscope magnitude of 45×.
(A)
(B)
(C)
(D)
(E)
(F)
Fig. 2. Optical microscope photographs (magnitude of 100×) of basified side of AMX-SB membranes used during electrodialysis of solutions containing 800 mg/L KCl and (A) 400 mg/L CaCl2 , (B) 800 mg/L CaCl2 , (C) 1600 mg/L CaCl2 , (D) 400 mg/L CaCl2 and 1000 mg/L Na2 CO3 , (E) 800 mg/L CaCl2 and 1000 mg/L Na2 CO3 , (F) 1600 mg/L CaCl2 and 1000 mg/L Na2 CO3 .
The electronic and optical images of membranes treated without carbonate showed a clean surface (Figs. 2A–2C and 3A–3C). For those membranes no deposit or fouling was found and their appearance was similar to the one observed for the original membrane AMX-SB (Figs. 1A and 1B). In fact, no real difference between those membranes was observed. This was confirmed by elemental analysis performed at higher magnification. The original membrane spectra showed the presence of three major elements: C: 82.24, Cl: 15.18, and O: 2.87% of relative percentage (see Table 2). In comparison, the spectra of membranes A, B, and C showed the presence of the same elements. The averaged relative percentage for C, Cl, and O were 82.19, 14.66, and 3.40%, respectively (Table 3).
Table 2 Elemental analysis results (%) of original AMX-SB membrane Element
Relative percentage (%)
Mg K Ca Na S Cl O C
0.00 0.00 0.02 0.00 0.00 15.18 2.87 82.24
For membranes used during ED treatment of solutions containing sodium carbonate, the optical microscope photographs
246
M. Araya-Farias, L. Bazinet / Journal of Colloid and Interface Science 296 (2006) 242–247
(A)
(B)
(C)
(D)
(E)
(F)
Fig. 3. Electronic microscope photographs (magnitude of 350×) of basified side of AMX-SB membranes used during electrodialysis of solutions containing 800 mg/L KCl and (A) 400 mg/L CaCl2 , (B) 800 mg/L CaCl2 , (C) 1600 mg/L CaCl2 , (D) 400 mg/L CaCl2 and 1000 mg/L Na2 CO3 , (E) 800 mg/L CaCl2 and 1000 mg/L Na2 CO3 , (F) 1600 mg/L CaCl2 and 1000 mg/L Na2 CO3 .
Table 3 Elemental analysis results (%) of AMX-SB membranes used during electrodialysis of solutions containing 800 mg/L KCl and (A) 400 mg/L CaCl2 , (B) 800 mg/L CaCl2 , (C) 1600 mg/L CaCl2 , (D) 400 mg/L CaCl2 and 1000 mg/L Na2 CO3 , (E) 800 mg/L CaCl2 and 1000 mg/L Na2 CO3 , (F) 1600 mg/L CaCl2 and 1000 mg/L Na2 CO3 Element Mg K Ca Na S Cl O C
Membrane A
B
C
D
E
F
0.00 0.01 0.01 0.00 0.00 15.56 2.50 82.17
0.00 0.00 0.00 0.00 0.00 14.39 3.18 82.72
0.00 0.01 0.01 0.00 0.00 14.02 4.53 81.69
0.01 0.01 3.02 0.00 0.00 10.38 26.72 60.10
0.00 0.00 2.31 0.00 0.00 12.52 17.24 68.15
0.00 0.00 2.79 0.00 0.00 9.94 21.08 66.37
showed the presence of precipitates dispersed on the surface of the membranes in contact with the basified NaCl solution (Figs. 2D–2F and 3D–3F). The elemental analysis showed that the surface of those membranes was composed of three main elements: carbon, chloride, and oxygen. A marked difference in the relative percentage of oxygen was observed in comparison with the original membrane (Tables 2 and 3); the relative percentage of oxygen increased 10.0 times for membrane D, 6.8 times for membrane E, and 8.4 times for membrane F. Furthermore, calcium was also present on the membranes surface. The calcium percentage was 3.02, 2.31, and 2.79% for membrane D, E, and F, respectively (Table 3).
The precipitates on membranes AMX-SB would be mainly composed of calcium and oxygen since those elements were present at the same place on the surface of membranes. This precipitate would be identified as calcium hydroxide. However, this identification is limited by the fact that the elemental analysis does not allow the identification of the presence of H. According to the electronic images the appearance of precipitates was typical of a crystalline substance. When the concentration of CaCl2 in solution increase, the size and form of the crystals become more apparent (Figs. 2D–2F and 3D–3F). Morphologically, large crystals very well defined where observed on membrane F (1600 mg/L CaCl2 ). The structure of crystals was characteristic of a rhombohedra or cubic substance. It appears from these results that the fouling observed on the surface of the anionic membrane in contact with the basified NaCl solution during treatment of solution containing Na2 CO3 and CaCl2 was formed of Ca(OH)2 . As suggested previously [6], calcium migrated through the cationic membrane could make contact with OH− ions from the basified solution and formed an insoluble deposit of Ca(OH)2 on the surface of anionic membranes. The crystals of precipitates grow gradually with the time and block the surface of AMX-SB membrane. It was documented in the literature that the alkaline-earth-metal ion such Ca2+ , Mg2+ may form precipitates when they are in basic solutions [16]. Ogata et al. [17] showed that the alkaline earth ions precipitate in the membrane as hydroxide and that the accumulation of these precipitates was a major factor in the degradation of the membrane. Calcium was encountered
M. Araya-Farias, L. Bazinet / Journal of Colloid and Interface Science 296 (2006) 242–247
the most harmful followed by Mg, Sr, and Ba. In this work, during ED treatment, the pH of the basified compartment was maintained constant at pH 11–12 with 1.0 N NaOH solution. Furthermore, after soaking with HCl the precipitates were dissolved and it would be due to the reaction of Ca(OH)2 with HCl which generate CaCl2 which is a soluble substance. Concerning the morphological characteristics of the precipitate formed, Tomazic et al. [18] noted that the morphological quality of crystalline precipitates of calcium hydroxide from aqueous solutions is dependent upon variables, such as supersaturation, chemical composition of the solution, including impurities and the method used to induce precipitation. Specifically, the growing of crystals results from their particles solution environment. Tomazic et al. [19] have observed a pronounced increase on growth kinetics of Ca(OH)2 with an increase in NaCl concentration at constant supersaturation (σ = 0.44 ± 0.01). Indeed, an increase of NaCl concentration increased the overall crystallization rate and after 10 min the weights of crystal were 90, 200, and 330 mg in 0.1, 0.2, and 0.3 M NaCl, respectively. On this account, similar behavior would be expected in our work when the concentration of NaCl in the concentrating compartment is approximately 0.3 M and is expected to be related with the growth of crystals especially when the CaCl2 concentration is highest (Figs. 3D–3F). Furthermore, in this work, an important difference was observed between the appearance of fouling observed on the anionic membranes and that of the cationic membranes treated with sodium carbonate [7]. The deposits on the anionic membranes had a cubic defined form while big nonmorphologically defined spot of deposits were observed on the surface of cationic membranes. Once again, as observed previously for CEM [7] no formation of calcium carbonate was observed despite the presence of high concentration of sodium carbonate and as previously suggested the difference in fouling of the membrane treated with and without sodium carbonate could be explained by the different treatment duration (68 vs 21 min, respectively). Carbonate has a high buffer capacity, and the time to reach pH 4.0 was then longer than the one without carbonate where there was no buffer capacity. 4. Conclusion In the present work, the presence of fouling was only observed on the membranes treated with a solution containing 400, 800, and 1600 mg/L of CaCl2 in presence of sodium carbonate. The appearance of precipitates on the anionic membranes was typical of a cubic and crystalline substance. The largest crystals were best defined when the membranes were treated with 1600 mg/L of CaCl2 . However, the cubic form of precipitates was only observed on the anionic membranes. In
247
fact, in our previous work, the cationic membranes showed in general a spotted and blotchy surface. The crystals would be identified as calcium hydroxide. However, this fouling was not found to affect significantly the conductivity and the thickness of the membranes and it was easily dissolved after the soaking with HCl. The analysis of anionic membranes used during this study complete the evaluation of the impact of the calcium and carbonate concentrations during ED treatments on the anionic and cationic membranes. Acknowledgments The authors thank Diane Montpetit from AAC-FRDC (SaintHyacinthe, QC, Canada) for her technical help on electron microscopy and elemental analysis. The financial support of the Natural Sciences and Engineering Research Council of Canada (NSERC) is acknowledged. References [1] B. Saint-Pierre, in: Symposium sur les applications alimentaires de l’électrodialyse. Fondation des Gouverneurs, Saint-Hyacinthe, Québec, 23 et 24 Mars 1999. [2] C. Gavach, in: Symposium sur les applications alimentaires de l’électrodialyse. Fondation des Gouverneurs, Saint-Hyacinthe, Québec, 23 et 24 Mars 1999. [3] E. Korngold, F. De Körösy, R. Rahav, M.F. Taboch, Desalination 8 (1970) 195. [4] V. Lindstrand, G. Sundström, A.S. Jönsson, Desalination 128 (2000) 91. [5] V. Lindstrand, A.S. Jönsson, G. Sundström, Desalination 130 (2000) 73. [6] L. Bazinet, M. Araya-Farias, J. Colloid Interface Sci. 286 (2005) 639. [7] L. Bazinet, M. Araya-Farias, J. Colloid Interface Sci. 281 (2005) 188. [8] L. Lebrun, E. Da Silva, G. Pourcelly, M. Métayer, J. Membr. Sci. 227 (2003) 95. [9] R. Lteif, L. Dammark, C. Larchet, B. Auclair, Eur. Polym. J. 35 (1999) 1187. [10] N. Pismenskaya, E. Laktionov, V. Nikonenko, A. El Attar, B. Bouclair, G. Pourcelly, J. Membr. Sci. 181 (2001) 185. [11] A. Ellatar, A. Elmidaoui, N. Pismenskaia, C. Gavach, G. Pourcelly, J. Membr. Sci. 143 (1998) 249. [12] A. Lounis, C. Gavach, J. Membr. Sci. 54 (1990) 63. [13] P.H. Barry, J.W. Lynch, J. Membr. Biol. 121 (1991) 101. [14] G. Milazzo, in: Électrochimie. Bases théoriques. Applications analytiques. Électrochimie des colloïdes, vol. 1, Dunod, Paris, 1969, p. 375. [15] T.A. Davis, J.D. Genders, D. Pletcher, The Electrochemical Consultancy, Romsey, 1997, p. 189. [16] K. Okada, M. Mamoru Tomita, Y. Tamura, Milchwissenschaft 31 (10) (1976) 602. [17] Y. Ogata, T. Kojima, S. Uchiyama, M. Yasuda, F. Hine, J. Electrochem. Soc. (1989) 136. [18] B. Tomazic, R. Mohanty, M. Tadros, J. Estrin, J. Cryst. Growth 75 (1986) 339. [19] B. Tomazic, R. Mohanty, M. Tadros, J. Estrin, J. Cryst. Growth 75 (1986) 329.