Collection Mémoires et thèses électroniques
AccueilÀ proposNous joindre

Chapitre 2 Effet du pH du concentrat et de la composition minérale d’une solution de protéines de lactosérum sur le colmatage des membranes en cours d’ÉD conventionnelle

Table des matières

Il apparaît dans la littérature que la majorité des travaux de recherche ont été effectués sur les membranes anioniques et que très peu de travaux ont ciblé l’étude de la présence de protéines sur la formation de colmatage en ÉD. Ce chapitre, rédigé sous forme d’article, avait comme but l’identification des conditions générales de traitement menant à la formation d’un colmatage, aussi bien minéral que protéique.

L’objectif du travail présenté dans cet article était l’étude de l’effet du pH du concentrat et de la composition en calcium, carbonate et protéines de lactosérum du diluat sur la formation de colmatage sur les membranes en cours d’ÉD conventionnelle. Afin de décrire l’évolution du procédé de déminéralisation, la conductivité, la résistance électrique du système, le pH du diluat et la migration des cations majeurs ont été suivis. Le suivi des paramètres électrodialytiques de l’empilement ont permis d’évaluer l’impact du colmatage sur le procédé. La migration totale des cations était similaire pour toutes les conditions, bien que différents colmatages aient été identifiés après trois traitements d’ÉD consécutifs de 100 minutes chacun. La composition du colmatage et la surface des membranes où le colmatage s’est formé dépendait du pH du concentrat et de la composition minérale du diluat. Une configuration alternative de l’empilement est proposée pour éviter la formation de colmatage.

The aim of this work was to study the effect of the concentrate solution pH and the composition in calcium, carbonate and protein of the diluate solution to be treated by conventional electrodialysis on the fouling of ion-exchange membranes. Conductivity, system resistance, pH of the diluate and cation migration were monitored to follow the evolution of the demineralization. Total cation migration was similar for all conditions although different forms of fouling were identified after 3 consecutive 100-minute electrodialysis treatments. The nature of fouling and the membrane surface fouled depended on the concentrate pH value, the diluate mineral composition and the intrinsic composition of the whey isolate. Once conditions leading to membrane fouling were identified, an alternative configuration for our electrodialysis stack is proposed to prevent fouling onset.

Membrane separation processes are continuously growing and finding new applications. Electrodialysis (ED) is not an exception, since potential uses are being evaluated for diverse food processes, such as acidity modification, concentration, extraction, fractionation and purification (Bazinet et al., 1998a). ED differs from most membrane processes since it separates according to electrical charge instead of particle size (Bazinet et al., 1998a). Two different types of ion-exchange membranes (IEM) are used in conventional electrodialysis: cation-exchange (CEM) and anion-exchange (AEM) membranes, which are permeable to cationic and anionic species, respectively.

Fouling formation is among the most important limitations in ED processes. Build-up of fouling film causes an increase in resistance (Bleha et al., 1992), which deteriorates the performance of process and can eventually lead to membrane integrity alteration. Under certain conditions, fouling and integrity alterations are irreversible and membranes must be replaced. Process interruptions to perform cleaning or membrane replacement can be extremely expensive, so research in order to avoid or prevent fouling is justified. Numerous studies have been done on the identification of species causing fouling (Korngold et al., 1970; Bleha et al., 1992; Linstrand et al., 2000b), but a great majority of these works achieved on membrane fouling have been usually directed to AEM, since their fouling susceptibility is higher than that of CEM (Korngold et al., 1970; Davis, 1990; Lindstrand et al., 2000a; 2000b). Recently, Bazinet and Araya-Farías (2005b) have reported the formation of a mineral fouling on CEM and AEM during conventional ED of different solutions of CaCl2 and Na2CO3. In the same way, a calcium deposit, presumably in the form of carbonate and hydroxide, was identified on CEM used during the bipolar membrane electroacidification (BMEA) of skim milk to produce casein isolates (Bazinet et al., 2001b). Such deposits would have been formed by calcium precipitating with electrogenerated OH- (Bazinet et al., 1999; 2000b; 2001b). Furthermore, the formation of a slight protein fouling on the CEM was observed when the stream in contact with the other side of the membrane was strongly acidified (pH ≈ 0.6) in order to neutralize the electrogenerated OH- during BMEA of skim milk (Bazinet et al., 2003). Calcium and carbonate are among the major fouling contributors found in food systems (Bazinet and Araya-Farías, 2005b), especially in dairy products, where they are naturally present. This limits the performance of demineralization, fractionation and/or purification of such foodstuffs. Furthermore, not much work characterizing protein-caused fouling of ED membranes has been found.

This study was part of a broader research project aimed at understanding factors leading to fouling formation, according to the composition of the treated solution and of its nature during conventional and bipolar membrane electrodialysis of food and nutraceutical products. A better description of protein fouling phenomenon and its relationship with mineral fouling is required in order to improve and develop ED applications for the food and nutraceutical industry. In the present work, our objective was to investigate the effect of the pH of the concentrate solution and the effect of the composition in calcium and carbonate of a whey protein isolate solution to be demineralized by conventional electrodialysis on the onset of protein and mineral fouling formation on both CEM and AEM.

The electrodialysis cell was a MicroFlow type cell (ElectroCell AB, Täby, Sweden) with two Neosepta CMX-S CEM and two Neosepta AMX-SB AEM (Tokuyama Soda Ltd., Tokyo, Japan). This arrangement defined three closed loops containing the diluate, concentrate and electrode-rinsing solution (Figure 4). Respectively, the solutions were the model diluate solution (330 ml), a 2 g/l aqueous KCl solution at a defined pH (330 ml) and a 20 g/l NaCl solution (500 ml) serving as the electrode-rinsing solution. The membranes tested, with an effective surface of 10 cm2, were both in contact with the model salt and protein solution (i.e. diluate) on one side and with the pH-controlled KCl solution (i.e. concentrate) on the other side. Each closed loop was connected to a separate external plastic reservoir, allowing continuous recirculation. The electrodialysis system was not equipped to maintain constant temperature, but, due to the low value of the applied current density, the temperature was expected to vary only in a small range.

A current density of 13 mA/cm2 was fixed for all conditions in order to start the demineralization under the limiting current density value. Preliminary tests had been previously performed to determine the limiting current density (LCD) values in the different demineralization conditions according to the method described by Cowan and Brown (1959). The current density used was equivalent to 75 % of the lowest LCD value obtained from all 9 conditions. Electrodialysis was carried out three times in 100-minute consecutive batches to ensure the fouling or not of the membranes. In order to maintain the system under the initial current density of 13mA/cm2, the voltage of the power supply (model HPD 60-5SX, Xantrex, Burnaby, BC, Canada) was adjusted accordingly at the beginning of each batch. The voltage necessary to reach such initial current density was then kept constant and the current was allowed to drop freely until the end of each demineralization batch.

The mineral composition of the different model diluate solutions treated by ED is presented in Table 1. A WPI concentration of 32.7g/l and a KCl concentration of 800mg/l were used for all diluate solutions. The pH value of the diluate solution (pH(diluate)) was adjusted at 6.5 and kept refrigerated overnight to ensure proper hydration of proteins. The pH value of the concentrate solution (pH(concentrate)) was maintained constant during the demineralization process at pH of 2, 7 and 12 by the addition of HCl and/or NaOH. Each pH(concentrate) value was used for each mineral condition: CaCl2 added alone, Na2CO3 added alone and CaCl2+Na2CO3. Conductivity and pH of the diluate solution were recorded at the beginning and after every ten minutes. 5 ml-samples of the diluate solution were taken at the beginning and at every 20 minutes during electrodialysis. All samples were then frozen until inductively coupled plasma (ICP) analysis was performed.

After the three consecutive 100-minute demineralization batches the ED stack was dismounted and both sides of each AEM and CEM were photographed. A surface of about 5 cm2 of the membranes in contact with the solution was cut and ashed overnight. The ashes were solubilized in a 1.0 N HNO3 solution and the concentrations of sodium, potassium, calcium and phosphorous were determined by ICP.

The pH(diluate) varied according to the nature of salts present in the diluate. According to the analysis of variance, the effect of Na2CO3 addition was significant (P<0.0001), but no effect was found for CaCl2 addition (P=0.4090). The effect of the pH(concentrate) (P<0.0001) also showed a significant effect on the total variation of the pH(diluate).

In presence of CaCl2 alone, the small pH variation ranging from 0.1 to 0.2 pH units was mainly due to the pH(concentrate) value. During the demineralization process, H+ or OH- leaching (i.e. migration of H+ across the anion-exchange membrane or OH- through the cation-exchange membrane) could account for such pH variation due to the low buffering capacity of CaCl2 and a limited buffering capacity of proteins present. Presence of Na2CO3 with or without CaCl2 added contributed mainly to the buffer capacity of the solution by enhancing the protein buffering capacity. This conducts to a general increase of pH(diluate) during the ED process. However, this increase depended on the pH(concentrate) value and the presence or not of CaCl2. With pH(concentrate) value of 12, the pH(diluate) varied from 0.40 to 0.50 pH units from the beginning to the end of the demineralization due to the leaching of OH- from the concentrate compartment. At pH(concentrate) value of 7, a variation of 0.15 to 0.20 pH units was observed. This smaller variation would be due to the balance between H+ and OH- leaching. At pH(concentrate) value of 2, the variations of 0.1 to 0.3 could no be explained by leaching since the low pH(concentrate) would have accounted a pH(diluate) decrease. This difference could probably be explained by the formation of a fouling on the AEM hindering the entrance of H+ into the diluate stream. Furthermore, these results are in accordance with those reported by Bazinet and Araya (2005b), where the presence of carbonate influenced the buffering capacity of the diluate solution and thus required a larger quantity of H+ to lower its pH.

It appeared from the results of the analysis of variance that the pH(concentrate) value and CaCl2 addition did not influence the conductivity drop (P=0.8470 and P=0.4720, respectively). On the opposite way, Na2CO3 addition had a significant effect (P<0.0001) on conductivity drop.

Naturally, pH(concentrate) values did not affect the initial conductivity of the diluate and had no effect on the final demineralization rate. As expected, initial conductivity of the solutions varied according to their respective salt composition and concentration corresponding to 2588 ± 86, 3195 ± 57 and 4322 ± 126 µS/cm for CaCl2 added alone, Na2CO3 added alone and Na2CO3 + CaCl2, respectively with all pH(concentrate) values averaged for each mineral condition. However, the conductivity variation was lower when CaCl2 was alone. In the presence of CaCl2, the variation in conductivity from the beginning to the end of the process was 1579 ± 132 µS/cm, while for Na2CO3 alone and Na2CO3 + CaCl2 the variations were similar with respective values of 1978 ± 167 µS/cm and 1944 ± 110 µS/cm. Such conductivity variations corresponded to demineralization rates, after 100 minutes, of 61.1 ± 5.9, 61.9 ± 4.9 and 45.0 ± 3.2 %, respectively. The difference could be explained by the slower migration of calcium ions, due to their hydration shell, than those of other monovalent cations (Hiraoka et al., 1979; Pérez et al., 1994; Bazinet et al., 2003). In fact, for the case of CaCl2 alone, when the main part of the K+ ions from the diluate solution would have migrated, the less mobile Ca2+ ions had to migrate in order to maintain the electroneutrality of the solution. On the other hand, when Na2CO3 was present, it was Na+ that would have migrated. In the case of Na2CO3 + CaCl2, the lower demineralization rate would be caused, as expected, by the higher salt concentration at the beginning of the demineralization process.

Since ion mobility and concentration determine the conductivity of the solution (Bard and Faulkner, 1983; Bazinet et al., 1997) an increase in salt concentrations caused an expected increase in the initial conductivity of the diluate solution. In these experiments, the migration rates were similar for conditions where Na2CO3 was present and different for CaCl2 alone. Since the initial intensity applied was the same for all conditions and then was allowed to drop freely, the different demineralization rates would depend on the ionic nature of the media, while the final demineralization level would depend on the velocity of ion migration and the presence or absence of fouling.

Results obtained from the analysis of variance showed a significant effect of both CaCl2 (P<0.0001) and Na2CO3 (P<0.0001) addition to the diluate solutions on system resistance. No significant effect was found for pH(concentrate) on system resistance.

In general, a slight decrease of the resistance ranging from 5 to 15 Ohms (Figure 5) was observed at the beginning of the ED process. This tendency could be explained by the reduction of voltage at the beginning of the demineralization, as mentioned in the protocol. Subsequent increase in the resistance occurred after voltage was kept constant and current density was allowed to drop freely. In the case of CaCl2 addition alone, an overall increment of 55 ± 16.7 Ohms was measured for the different demineralizations. This increase of resistance (Figure 5) would appear following the depletion of potassium ions.

At this point, the lowered conductivity of the diluate and the hindered demineralization of Ca2+ ions (i.e. given their slower migration) would have been responsible for the resistance increase of the ED system. For demineralization with only Na2CO3 added, a resistance increase of 42.6 ± 10.7 Ohms was found (Figure 5). Presumably, the depletion of rapidly demineralizing Cl- ions and the lower mobility of the carbonate species would probably account for such resistance increase of the system. However, chlorine and carbonate concentrations were not determined. In the case of simultaneous addition of CaCl2 and Na2CO3, the resistance varied by only 5.7 ± 6.9 Ohms (Figure 5). By the end of the demineralization, resistance values remained very close to the original ones. The Na+ and Cl- enriched diluate could explain the minimal change in the resistance of the system. Such ions, coming from the added CaCl2 and Na2CO3, would prolong an unhindered demineralization just as KCl. Thus, an increase of the resistance did not occur after 100 minutes of demineralization.

The correlation between the system resistance and the conductivity evolution was in accordance with previous studies (Pérez et al., 1994; Bazinet, et al., 2004). The higher increase in resistance (Figure 5) observed at the end of the ED treatment was due to the high demineralization rate of the diluate solution. The global system resistance could also have been influenced by the presence of a fouling, increasing the resistance at the membrane surface level (Lindstrand et al., 2000b).

After the three 100-minute-demineralization batches were completed, the ED stack was dismounted and photographs were taken of the membrane surfaces in contact with the diluate and with the concentrate. This was done for both AEM and CEM. The surface of the AEM in contact with the diluate (Figure 6) did not present any visible fouling for treatments with a pH(concentrate) value of 12 in any mineral condition. However, a clearly visible heterogeneous deposit was observed on this side of the AEM treated with pH(concentrate) values of 2 and 7 for all mineral conditions.

This deposit, which was like a gel on the surface of the membrane, would constitute a protein fouling. The thick layer of deposit, which partially filled the diluate compartment, easily detached from the membranes as they were manipulated for further analysis. Although no conclusive results could derive from quantitative nitrogen analysis, finding an overall concentration of 28.70 ± 1.83 mg N / g of dry membrane (Table 3), a permanent residue remained fixed to the surface of the membrane, forming a fouling layer. Apparently, H+ leakage and/or electrogenerated H+ at the interface of the AEM in contact with the diluate solution would have accounted for protein precipitation. H+ leakage through the AEM occurs due to the presence of proton acceptors (e.g. water molecules and acidic species contained in the feed solution) (Roualdes et al., 2002). Furthermore, water splitting processes take place when the LCD is exceeded, and may even occur at lower values (Korngold et al., 1970). During the ED procedure, the diluate was depleted of ionic species, lowering the initial value of the LCD, leading to the electrogeneration of H+ by secondary and tertiary amino groups. Such amino groups catalyze water splitting process and originate after the conversion of tetra-alkyl ammonium groups fixed to the AEM during current flow (Simons, 1984; 1985). This phenomenon of H+ electrogeneration may also occur in CEM, since carboxylic groups have water splitting activity, as well as sulphonic acid groups when ionizable molecules, such as aminoacids, are present inside the membrane (Simons, 1985).

Surface of AEM in contact with the concentrate (Figure 7) presented a visual fouling only at pH(concentrate) value of 12 and 7 for mineral conditions of the diluate solution where CaCl2 alone was added.

Apparently, this mineral-like fouling on this side of the AEM would be constituted of deposits of calcium, since a ten-fold concentration of this element (2.05 ± 0.66 mg Ca / g of dry membrane) for these two treatments, in comparison to all the other (0.22 ± 0.21 mg Ca / g of dry membrane), was quantified via membrane ash analysis (Table 3). Phosphorous was also present in higher concentrations (Table 3) in these two membranes (1.06 ± 0.06 mg P / g of dry membrane), when compared to the average of other seven (0.23 ± 0.09 mg P / g of dry membrane). Phosphorous was naturally present in WPI as a residual mineral, as mentioned in the materials section. Calcium would have migrated through the CEM and, due to recirculation of the concentrate, would have precipitated in the form of Ca(OH)2 and/or Ca3(PO4)2 as it entered in contact with the AEM. The precipitation of these minerals would have been caused by unfavourable pH conditions for Ca2+ solubility given by the combination of pH(concentrate) and the electrogenerated OH- production influencing the local basification of the interface AEM-concentrate solution. Furthermore, since magnesium was present in WPI as a residual mineral, calcium precipitation could have been induced at a pH value as low as 7.38 when traces of Mg2+ was present, as previously shown by Shaposhnik et al. (2002). No fouling was detected for any condition with a pH(concentrate) value of 2 nor with Na2CO3 added to the diluate solution since an acidic pH(concentrate) would have been able to neutralize the OH- at the AEM, avoiding calcium precipitation. In the case of conditions where Na2CO3 was present, the presence of carbonate species could have exerted a buffering protection against extreme basification, hence impeding the precipitation of calcium. As expected for AEM membranes, low concentrations were found for potassium (0.31 ± 0.22 mg K / g of dry membrane) and sodium (0.67 ± 0.21 mg Na / g of dry membrane), but no remarkable differences were observed for any treatment (Table 3).

In the case of the surface of CEM in contact with the diluate solution (Figure 8), only conditions combining the presence of CaCl2 and Na2CO3 simultaneously at a pH(concentrate) value of 2 presented a very slight whitish deposit.

This deposit would apparently have come from precipitation of protein or by contact with the protein fouling formed on the AEM. No other deposit was visually detected for any other CEM surface in contact with the diluate. These results agree with those reported by Bazinet et al. (2003), where a very slight protein fouling formed on cationic membranes when a medium with a very low pH was present on the opposite side of the membrane, in this case the concentrate. However, as previously for the AEM, nitrogen analysis did not allow to detect any difference (Table 3). The lower overall-nitrogen concentration found for CEM (13.43 ± 1.09 mg N / g of dry membrane) with respect to AEM (28.70 ± 1.83 mg N / g of dry membrane) would be given, as expected, by the different nature of the ionic sites of each membrane type (Davis et al., 2001).

Surface of CEM in contact with the concentrate (Figure 9) did not present any deposit for conditions with pH(concentrate) value of 2 and 7. For Na2CO3 added alone, no deposit was found at any pH(concentrate) value. However, an evident fouling appeared for conditions with a pH(concentrate) value of 12 and CaCl2 added.

The results of CEM ash analysis, showed an extremely high concentration of calcium for conditions with a pH(concentrate) value of 12 CaCl2 added alone (120 mg Ca / g of dry membrane) and CaCl2 + Na2CO3 added simultaneously (65 mg Ca / g of dry membrane), as compared to the average of the other seven conditions (6.92 ± 6.14 mg Ca / g of dry membrane) (Table 3). This fouling would be likely constituted of calcium, precipitating as a hydroxide and/or a carbonate, in the case when Na2CO3 was present. This fouling was accompanied of a severe deformation of the membranes. As expected for CEM membranes, given their ionic affinity, higher concentrations with respect to AEM were found for potassium (4.93 ± 1.64 mg K / g of dry membrane) and sodium (18.92 ± 7.31 mg Na / g of dry membrane). The higher variations in these two elements would be given by their relatively low concentrations in those cases where calcium concentration was extremely high (Table 3), as already mentioned. No phosphorous (0.00 ± 0.00 mg P / g of dry membrane) was detected in the CEM ashes.

According to MANOVA results, time (P<0.0001 for K+) and Na2CO3 addition (P<0.0001 for K+, P<0.0001 for Na+, P=0.0129 for Ca2+) had a significant effect on cation concentration in the diluate solution. The dual interaction Na2CO3-time had also a significant effect (P=0.0019 for K+, P<0.0001 for Na+, P=0.0110 for Ca2+) on cation concentration evolution during demineralization.

Whatever the pH(concentrate) value, the cation concentration evolution was different according to the nature of the cation and to the salt added. For K+ , all pH(concentrate) averaged, its concentration decreased (Figure 10) of approximately 272 ± 60 mg/l, 209 ± 24 mg/l and 206 ± 10 mg/l, respectively for CaCl2 alone, Na2CO3 alone and CaCl2 + Na2CO3, from the beginning to the end of the demineralization.

As expected, Na+ concentration decrease, all pH(concentrate) averaged, was more pronounced (Figure 11) for both conditions with Na2CO3 added, being 129 ± 80 mg/l, 254 ± 46 mg/l and 287 ± 41 mg/l, respectively for CaCl2 alone, Na2CO3 alone and CaCl2 + Na2CO3. In the case of Ca2+, all pH(concentrate) averaged, its migration (Figure 12) was close to 0 mg/l when Na2CO3 was present, while in the case of CaCl2 alone, the migration was approximately of 105 ± 42 mg/l.

In accordance with electrical mobility and their hydration shell, K+ would be the first cation to migrate through the cationic membrane, followed by Na+ and finally Ca2+. These results agree with those found in literature, which report faster migration rates for monovalent species than divalent species (Kabay et al., 2003), following the order K+ > Na+ > Ca2+ (Pérez et al., 1994). Furthermore, CEM ash analyses showed a certain pattern according to the presence or not of carbonate: For conditions where CaCl2 was added, the solely presence of Na+ ion from the Na2CO3 addition decreased the content of Ca2+ in the CEM membrane of about 39.4 ± 10.0%, independently to whether the membrane was fouled or not (Table 3). In the cases where only Na2CO3 was added, calcium traces originating from the WPI were found (0.80 ± 0.13 mg Ca /g of dry membrane).

In the case of CaCl2 alone, sodium ions present would come from their intrinsic presence as a residual salt in the WPI, as presented previously in the materials and methods section. Then, the Ca2+ migration observed was to compensate for the K+ depletion in order to maintain the electroneutrality of the solution. When Na2CO3 was added, there was no need for Ca2+ to migrate since Na+ concentration was high enough to compensate K+ depletion and, hence, delay the migration of Ca2+. In fact, competition of Na+ with K+ would account for the lower demineralization of the latter for both conditions with Na2CO3 added. The residual presence of Ca2+ in the condition of Na2CO3 alone would also originate from its intrinsic presence as a residual mineral present in the WPI.

The migration of the cations through the cation membrane could be expressed as equivalents of migrated mass (Emass) given by Equation 1 (Bazinet et al., 2000a),

Emi = n i z i (Equation 1)

where Emi is the migrated mass expressed in equivalents, n i is the number of moles of the ionic specie i and z i the charge of the ionic species i.

The total equivalents of migrated mass (Emt) for each mineral condition j (CaCl2 added alone, Na2CO3 added alone, CaCl2 + Na2CO3) is then given by Equation 2.

Emtj = Emij = (n ij z ij ) (Equation 2)

If molarity of the ionic specie i in the mineral condition j (M ij = n ij / v) is substituted to introduce a volumetric ratio into Equation 2, expressed in milliequivalents, then

Emtj / v = Emij / v = (M ij z ij ) (Equation 3)

where Emtj /v, given in milliequivalents per litre (mEq/l), is the total milliequivalents of mass migrated per litre of diluate with mineral conditions j, M ij is the molarity of the ion i for the condition j and v is the volume of diluate solution.

By calculating the total migration of cations at the end of all demineralizations with Equation 3, no apparent differences were observed among the different mineral conditions when all the pH(concentrate) conditions were averaged (Table 4): 17.8 ± 2.6 mEq/l for CaCl2 added alone, 16.0 ± 3.1 mEq/l for Na2CO3 added alone and 19.0 ± 4.9 mEq/l CaCl2 + Na2CO3 added simultaneously.

Actually, this agrees with the fact that calcium would have been forced to migrate once the concentration of monovalent ions decreased in the diluate solution, in order to compensate for the most easily-demineralizing ions. The respective resistance increase caused by calcium migration, as discussed earlier, would have lead to the formation of H+ and OH- ions. These ions would have promoted the formation of deposits on the surface of membranes.

This would imply, at the same pH(concentrate) value, that, once the monovalent ions were depleted, a higher potential would have been necessary to force the migration of calcium in order to maintain the same demineralization level as confirmed by cationic milliequivalent balance (Table 4). Since potential was kept constant, the migration of the divalent ions would have caused an increase of the global resistance of the system, leading to a lower value of limiting current density (LCD) and, hence, to water splitting to compensate for the lack of mobile ions (Simons, 1979; Rubinstein et al., 1984; Bazinet et al., 1998a). If demineralization took place above the new value of limiting current density, dissociation of water to produce H+ ions at the interface of the AEM would precipitate proteins, initiating an autocatalytic fouling. In addition to this fouling, In the case of calcium migration into a concentrate with pH(concentrate) value of 12, the pH profile of the membrane interface given by OH- ions leakage and as well as their high concentration in the concentrate solution would account for the precipitation of calcium. In the case of the AEM a white deposit was also detected on the surface that was in contact with concentrate. This would also be a calcium deposit, which would have been caused by the recirculation of the concentrate, allowing contact of Ca2+ migrated through the CEM to enter in contact with the AEM. Further studies are presently taking place in order to identify the nature of such deposits.

Once the main part of the cations migrated, the concentration of mobile ions would have been very low, and the global resistance of the system was increased by demineralization. Although the initial voltage and current values were under the limiting current density, the latter was probably reached and surpassed during the process, leading to water splitting at the AEM to compensate for the depleted mobile ions and membrane fouling would have occurred (Simons, 1979; Rubinstein et al., 1984; Bazinet et al., 1998a). This would explain the pH increase in certain conditions, as well as the presence of a presumable protein fouling in the surface in contact with the diluate of the AEM. A protein fouling would be responsible for retaining the H+ from reaching the diluate solution, hence lowering its pH value. This way, the H+ would precipitate more protein, in an autocatalytic process that would increase the thickness of the fouling layer.

Nitrogen quantification did not allow the confirmation of a protein fouling, so further surface analysis will follow to confirm the protein-nature of the deposits. The large variations found for the mineral compositions of membranes would be partially explained by the fact that only one membrane section was used for the ash analysis and also to the uneven distribution of the deposits on the surface of the membranes. However, further surface analysis are currently in progress to identify the nature of the mineral deposits.

It appeared from these results that acidic and neutral conditions lead to protein film formation over the diluate side of the AEM, but basic conditions prevented its formation. Protein fouling on CEM was not visually apparent. CEM presented mineral fouling only in basic concentrate conditions when calcium was present, which would precipitate as calcium hydroxide. Mineral fouling on the concentrate side of AEM was due to recirculation and mixture of both anion and cation-receiving streams. The stack configuration used allowed calcium to migrate through CEM and, by recirculation, to be in contact with the AEM and thus to precipitate on its surface, possibly as calcium hydroxide.

According to these results, the separation of the concentrate stream in two different loops would prevent formation of both, mineral and protein foulings. A neutral pH condition should be maintained for the cation-receiving stream in order to prevent mineral fouling on the concentrate side of CEM. This loop must remain independent to the anion receiving one. A basic pH should be maintained for the latter to prevent formation of a protein film over the AEM surface. Mineral fouling on AEM will no longer form, as calcium will not enter in contact with the concentrate side of such membrane.

© Erik Ayala Bribiesca, 2005