Relative contributions of charged species to conductivity changes in skim milk during electrochemical acidification

Journal of Membrane Science 352 (2010) 32–40

Contents lists available at ScienceDirect

Journal of Membrane Science journal homepage: www.elsevier.com/locate/memsci

Relative contributions of charged species to conductivity changes in skim milk during electrochemical acidification Laurent Bazinet a,b,c,∗ , Yves Pouliot a,b,c , Franc¸ois Castaigne c a b c

Institute of Nutraceuticals and Functional Foods (INAF), Pavillon Paul Comtois, Université Laval, Sainte-Foy (QC), Canada G1V 0A6 Centre de Recherche en Science et Technologies du Lait (STELA), Pavillon Paul Comtois, Université Laval, Sainte-Foy (QC), Canada G1V 0A6 Department of Food Sciences and Nutrition, Pavillon Paul Comtois, Université Laval, Sainte-Foy (QC), Canada G1V 0A6

a r t i c l e

i n f o

Article history: Received 19 October 2009 Received in revised form 11 December 2009 Accepted 27 January 2010 Available online 4 February 2010 Keywords: Electrochemical acidification Bipolar membrane Model Contribution to conductivity Charged species Milk

a b s t r a c t Bipolar membrane electroacidification (BMEA) is an acidification technique that approximates chemical acidification. However, differences in the casein precipitation profile have been observed between these two procedures and correlated with differences in conductivity. In this work, ionic and protein contributions to conductivity during BMEA were mathematically modeled. Chloride, potassium, calcium and sodium made major contribution during electrochemical acidification. The evolution of protein contribution was the same as observed previously for chemical acidification, but with slightly higher values. During pH decrease, chloride contribution to conductivity was greater during chemical acidification while the decrease in potassium contribution was more marked during BMEA. Since the contribution of proteins during BMEA was greater, electrostatic repulsions between proteins would also be greater than during chemical acidification, explaining their precipitation at a higher pH value. © 2010 Elsevier B.V. All rights reserved.

1. Introduction Electrodialysis is a well known electrochemical separation process by which electrically charged species are transported from one solution to another, across one or more cationic and anionic membranes, driven by direct current [1,2]. Anionic and cationic membranes are monopolar, meaning that they are permeable to only one type of ion. A newer type of membrane, called a bipolar membrane, was introduced at the end of the 1980s. Bipolar membranes bring about the dissociation of water in an electric field. They are composed of three parts: an anion-exchange layer, a cationexchange layer and a hydrophilic interface at the anionic–cationic junction. When a direct current is applied, water molecules migrate into the hydrophilic layer where they are split into H+ and OH− , assuring electron transport through the cationic and anionic layers respectively [2–5]. Bipolar membrane electroacidification (BMEA), which is a technology that uses the property of bipolar membranes to acidify solutions by splitting water and the action of monopolar mem-

∗ Corresponding author at: Department of Food Sciences and Nutrition, Pavillon Paul Comtois, Université Laval, Sainte-Foy (QC), Canada G1V 0A6. Tel.: +1 418 656 2131x7445; fax: +1 418 656 3353. E-mail addresses: [email protected] (L. Bazinet), [email protected] (Y. Pouliot), [email protected] (F. Castaigne). 0376-7388/$ – see front matter © 2010 Elsevier B.V. All rights reserved. doi:10.1016/j.memsci.2010.01.056

branes for demineralization, has been successfully applied to acid casein production [6–8], soy isolate production [9,10], fractionation of soy proteins [11], chitosan oligomers production [12,13], delipidation of a whey protein concentrate [14], inhibition of enzymatic browning [15,16] and passion fruit juice deacidification [17]. BMEA is an acidification technique that approximates chemical acidification since both procedures are based on the addition of hydrogen ions. However, differences in the casein precipitation profile were observed between these two procedures and correlated with differences in conductivity. Conductivity increased during chemical acidification, increasing ionic strength and decreasing electrostatic repulsions between proteins (salting-in effect) whereas it decreased by BMEA decreasing ionic strength and increasing electrostatic repulsions [18]. The purpose of the present work was to model the contribution of charged species in milk to the change in conductivity during electrochemical acidification by BMEA process. In a previous article [19], various mathematical functions were presented, which allowed the determination of protein and ionic contributions to conductivity increases during chemical acidification. The main ionic species contributing to the conductivity change during HCl acidification were potassium and chloride as well as calcium solubilized from the micelle and the contribution of protein to conductivity was in agreement with the changes in ␨-potential reported in the literature [19]. In the present paper, the ion and protein contributions to conductivity during electrochemical acid-

L. Bazinet et al. / Journal of Membrane Science 352 (2010) 32–40

33

ification of skim milk are presented. The postulate of this work is that milk is in a nonsteady state during electroacidification due to its complexity and composition changes as a function of pH. Furthermore, the interactions of cations with phosphate, citrate and casein phosphoserines were not taken into account to simplify this first tentative approach. The physico-chemical phenomena taken into account to evaluate ion concentrations as a function of pH during electrochemical acidification include: (1) dissociation of casein micelle resulting in solubilization of calcium and phosphate, (2) migration of cations across the cationic membrane, (3) acidification without addition of chloride, (4) evolution of phosphorus forms as pH decreases, (5) changes in viscosity affecting electrical mobility, and (6) binding of added H+ by proteins. 2. Theoretical approach Fig. 1. Principle of bipolar membrane electroacidification and ionic balance.

During electrochemical acidification of milk, addition of H+ occurs by splitting of water molecules at the interface of the bipolar membrane. Ionic balance is maintained by cationic migration from the milk across the cationic membrane, leading to a decrease in pH and conductivity [6,20]. As is the case during chemical acidification, the decrease in pH leads to dissolving of colloidal calcium phosphate and small amounts of magnesium [21–23] and dissociation of casein from micelles [24]. The conductivity at a given pH is the net result of H+ addition, total soluble protein charge, dissolving of calcium, phosphate and magnesium from micelles, potassium, sodium, chloride, hydrogen ions and citrate already present in the soluble phase, and migration of potassium, sodium, calcium and magnesium. The contributions of iron, zinc, carbonate, sulphate, etc., may be neglected due to their low concentrations in milk [19]. The general equation for electrical conductivity s (mS cm−1 ) of a solution is given by Lopez Leiva [25]: s = F



 

(Ci . zi  .i )

(1)

i

where F is the Faraday constant (C mol−1 ), zi the valence, Ci the concentration (mol L−1 ) and i the electrical mobility (cm2 s−1 V−1 ) of ion i. The conductivity of milk at a given pH can thus be written in the following general form:

⎡ ⎢

determined arithmetically from Eq. (4), after calculation of ti (pH) for each ionic species [19]. 3. Calculation for milk during BMEA treatment The different Ci (pH) functions for milk ionic species were calculated from values reported in the literature or obtained experimentally by Bazinet et al. [20]. Sigmaplot 5.0 for Windows (Jandel Scientific, Corte Madera, CA) was used to fit curves to each function. 3.1. CH+ (pH) determination During electrochemical acidification, as observed for chemical acidification, only part of the H+ electrogenerated was used to decrease the pH [19]. When milk is electroacidified, a large number of hydrogen ions are added but these ions are almost all bound to amino groups in the side chains of proteins [26]. The equation for the number of electrogenerated H+ effectively used for pH decrease was the same as for chemical acidification: CH+ (pH) = 10−pH

(6)

⎤

CH+ (pH) · zH+ · H+ + CCl− (pH) · zCl− · Cl− + Cprot (pH) · zprot · prot

s (pH) = F × ⎣ +CCa2+ (pH) · zCa2+ · Ca2+ + CP x− (pH) · zP x− · P x− + CMg2+ (pH) · zMg2+ · Mg2+ i

i

i

⎥ ⎦ (2)

+CK+ (pH) · zK+ · K+ + CNa+ (pH) · zNa+ · Na+ + CCit x− (pH) · zCit x− · Cit x− where s (pH) is the conductivity (mS cm−1 ) at the specified pH, CH+ (pH), CCl− (pH), Cprot (pH), CCa2+ (pH), CP x− (pH), CMg2+ (pH), (mol L−1 )

i

are concentrations, zH+ , CK+ (pH), CNa+ (pH), CCit x− (pH) zCl− , zprot , zCa2+ , zP x− , zMg2+ , zK+ , zNa+ , zCit x− the corresponding i

valences, and H+ , Cl− , prot , Ca2+ , P x− , Mg2+ , K+ , Na+ , Cit x− i

the electrical mobilities (cm2 s−1 V−1 ) of hydrogen ion, chloride, soluble protein, calcium, phosphate, magnesium, sodium and citrate. Since the transport number ti (pH) = si (pH)/s (pH) (Eq. (3)) also represents the relative contribution of a given

ion (si ) to the total conductivity (s ) of the solution [19] and ti = 1 (Eq. (4)) [25], the evolution of the transport number of the ion during a pH decrease can be written as follows:

 

ti (pH) = F · zi  · i ·

Ci (pH) s (pH)

(5)

Ci (pH) and s (pH) are functions to be evaluated experimentally for each ion or charged species i. The contribution of the protein mixture to the variation in conductivity during a pH decrease will be

3.2. CCl− (pH) determination During electrochemical acidification by BMEA, there is no addition of Cl− with electrogenerated H+ . Thus CCl− (pH) = CCl−

present

CCl−

present

(pH)

(7)

(pH) corresponds to the initial concentration of chloride in

milk or 0.029 mol L−1 , calculated from White and Davies [27] and Holt [28]. Since chlorides are negatively charged, their concentration is not affected by electromigration. The configuration of BMEA, combining bipolar and cationic membranes, theoretically does not allow migration of negative charges from the skim milk solution (Fig. 1). The chloride present in the soluble phase was considered completely soluble and contributing entirely to the conductivity. The total chloride concentration as a function of pH during electrochemical acidification is thus a constant: CCl− (pH) = 0.029

(8)

34

L. Bazinet et al. / Journal of Membrane Science 352 (2010) 32–40

3.3. CMg2+ (pH), CCa2+ (pH) and CNa+ (pH) determination Two kinetics had to be taken into account to determine the total concentrations of magnesium, calcium and sodium as a function of pH during electrochemical acidification: solubilization from casein micelles and electromigration across the cationic membrane from the skim milk to the adjacent compartment. The ion has to dissolve from its casein micelle in order to electromigrate as a free ion. Thus Ci (pH) = Ci dissolved (pH) − Ci migrated (pH)

(9)

For these three ions, Ci dissolved (pH) was assumed to be the same as calculated previously for chemical acidification [19]. The concentrations of magnesium, calcium and sodium migrating as a function of pH were modeled from experimental data obtained by Bazinet et al. [20]. These functions were fitted with sigmoidal curves (Fig. 2), since from pH 6.6 to 5.4, no significant statistical differences were observed in the migration of these species. Ions began to migrate just after the majority of them had dissolved from the casein micelles. Substitution of numerical values (Table 1) obtained from statistical analysis, into the model curve gave the following equations: CMg2+ (pH) =



0.0012



1 + exp (pH − 4.94)/0.1961

(10)

Fig. 2. Concentration of potassium, calcium, magnesium and sodium migrated during bipolar membrane electroacidification.

CCa2+ (pH) = CNa+ (pH) =



0.0089



0.0152



(11)



(12)

1 + exp (pH − 4.83)/0.2251 1 + exp (pH − 4.74)/0.2134

Table 1

  Statistical results for (a) CMg2+ migrated (pH), (b) CCa2+ migrated (pH), and (c) CNa+ migrated (pH): general equation fit: Ci (pH) = a/1 + exp − (pH − x0 )/b . (a) CMg2+ migrated (pH) R = 0.9911, R2 = 0.9823, Adjusted R2 = 0.9735 Standard error of estimate = 0.0001 Coefficient a 0.0012 b −0.1961 4.9445 x0 Analysis of variance: DF Regression 2 Residual 4 Total 6 Normality test Passed Constant variance test Passed (b) CCa2+ migrated (pH) R = 0.9944, R2 = 0.9888, Adjusted R2 = 0.9832 Standard error of estimate = 0.0005 Coefficient a 0.0089 b −0.2251 4.8351 x0 Analysis of variance: DF Regression 2 Residual 4 Total 6 Normality test Passed Constant variance test Passed (c) CNa+ migrated (pH) R = 0.9985, R2 = 0.9970, Adjusted R2 = 0.9955 Standard error of estimate = 0.0004 Coefficient a 0.0152 b −0.2134 4.7384 x0 Analysis of variance: DF Regression 2 Residual 4 Total 6 Normality test Passed Constant variance test Passed

Std. error 0.0001 0.0602 0.0710

t 11.2838 −3.2574 69.6681

P 0.0004 0.0312 <0.0001

SS 0.0000 0.0000 0.0000 (P = 0.5090) (P = 0.0735)

MS 0.0000 0.0000 0.0000

F 111.32

Std. error 0.0001 0.0000 0.0699

t 11.2936 −4.5031 69.2104

P 0.0004 0.0108 <0.0001

SS 0.0001 0.0000 0.0001 (P = 0.6883) (P = 0.0956)

MS 0.0000 0.0000 0.0000

F 177.35

Std. error 0.0007 0.0237 0.0344

t 20.7933 −8.9968 137.6764

P <0.0001 0.0008 <0.0001

SS 0.0002 0.0000 0.0002 (P = 0.2433) (P = 0.2166)

MS 0.0001 0.0000 0.0000

F 675.81

P 0.0003

P 0.0001

P <0.0001

L. Bazinet et al. / Journal of Membrane Science 352 (2010) 32–40

The sum of the dissolved and migrated ions thus gave the following equations:

−

0.0012

0.0199



1 + exp (pH − 5.67)/0.1956



0.0089



CNa+ (pH) = 0.0215 + 9.4 × 10−4 × (6.7 − pH)



0.0152

(pH) = 2.888 − 2.198 × pH + 0.621 × pH2 − 0.076 × pH3 (20)

(13)

(14)

1 + exp (pH − 4.83)/0.2251

−

migrated

+ 3.45 × 10−3 × pH4



1 + exp (pH − 4.94)/0.1961



fitted to a quartic equation (R2 = 0.9976) in order to account for the increase in concentration at the end of BMEA (Fig. 2): CK+



1 + exp (pH − 5.52)/0.134



CCa2+ (pH) = 0.00189 + −

0.00143



CMg2+ (pH) = 0.00060 +

35





(15)

1 + exp (pH − 4.74)/0.2134

Bazinet et al. [20] attributed this increase in potassium after pH 5.0–5.4 was reached to a leakage across the bipolar membrane. Potassium ions would be carried across the anionic layer of the BPM with water molecules. Since the potassium concentration on the anionic side of the BPM is high, it would override the Donnan effect, which controls the selectivity of the anionic layer. Anion-exchange membranes have been shown not to be completely selective for anions and to allow a small leakage of cations on the order of 1–2% [29]. According to Eq. (19), the evolution of the concentration of potassium as a function of pH in the soluble phase was:



CK+ (pH) = 0.0229 + 9.9 × 10−4 × (6.7 − pH)

3.4. CP x− (pH) determination

− 2.888 + 2.198 × pH − 0.621 × pH2 + 0.076×, pH3

As mentioned for Cl− , since the three charged forms of phosphate in the milk, H2 PO4 − , HPO4 2− and PO4 3− , are negatively charged, their concentrations during electrochemical acidification are not affected by electromigration. The mathematical relations obtained for CH PO− (pH) and CHPO2− (pH) during chemical acidifi-

− 3.45 × 10−3 × pH4

i

2

4

4

cation were thus applicable to electrochemical acidification. The contribution of PO4 3− was neglected due to its low concentration in the pH range considered [19]. The phosphorus concentration as a function of pH, considering H2 PO4 − /HPO4 2− distribution, was thus given in the following equation: CP x− (pH) = CH

− 2 PO4

i

with CH



(pH) =

 (0.0149 × 0.95) +



0.0146



1 + exp (pH − 5.47)/0.1766

−436.49 + 438.10 pH − 134.90 pH2 + 18.56 pH3 − 0.96 pH4

(17)

100

and



CHPO2− (pH) =

(0.0149 × 0.60) +





0.0146



100

CK+ (pH) = −2.858 + 2.197 × pH − 0.621 × pH2 + 0.076 × pH3 − 3.45 × 10−3 × pH4

3.6. CCit x− (pH) determination Since 94% of the citrate in milk is in the soluble phase [28], it was considered entirely in the soluble phase at an initial concentration in milk of 0.0098 mol L−1 [30]. In addition, only 3% of citrate is present as free ion [30] and the conversion to free ion is very slow. Furthermore, its concentration is not affected by electromigration due to its negative valence. Its concentration contributing to conductivity was evaluated at 3 × 10−4 mol L−1 and assumed to be constant during the pH decrease. CCit 3− (pH) = 3 × 10−4

(18)

3.5. CK+ (pH) determination As for magnesium, calcium and sodium, potassium concentration as a function of pH can be expressed as dissolved

For this cation, CK+

(pH) − CK+

dissolved

(pH)

(23)

s (pH) = 103.14 − 73.64 × pH + 20.38 × pH2 − 2.501 × pH3 + 0.1154 × pH4

CK+ (pH) = CK+

(22)

The evolution of conductivity was calculated as a function of pH from previous experimental data [31] and was strongly correlated (R2 = 0.9976) to the pH (Fig. 3):

1 + exp (pH − 5.47)/0.1766

1043.22 − 812.64 pH + 237.58 pH2 − 30.94 pH3 + 1.51 pH4

After simplifications:

3.7. s (pH) determination



4

×

(16)

4



− 2 PO4

×

(pH) + CHPO2− (pH)

(21)

(19)

migrated

(pH) was assumed to be the same as calcu-

lated previously for chemical acidification [19]. The evolution of the concentration in moles per liter of potassium having migrated from the soluble phase as a function of pH was calculated from data obtained by Bazinet et al. [20] and was

(24)

The conductivity of the skim milk acidified by bipolar membrane electroacidification decreased from 4.8 to 3.9 mS cm−1 [31]. In this study, no statistical difference was observed between the initial conductivity values of skim milk chemically and electrochemically acidified, respectively 5.15 and 4.80 mS cm−1 , due to large standard deviations. However, differences in the initial values of conductivity between chemical and electrochemical acidification may be explained by a slight dilution during BMEA treatment, due to the dead volume of the electrodialysis system. To compare the contribution of each charged species during chemical and electrochemical acidification, a correction factor of 0.35 was added to

36

L. Bazinet et al. / Journal of Membrane Science 352 (2010) 32–40

chemical acidification model in order to better reflect the true ion mobility in skim milk. The use of these correction factors nevertheless tended to minimize the true effect of viscosity [19] Fc = 0.995 +



1.3048

2

(26)

1 + (pH − 4.919)/0.2384

The correction factor Fc was calculated from data obtained for skim milk by Attia et al. [23]. F c was the ratio of consistency coefficients of skim milk and water and was equal to 1.59 [19]. 4. Results and discussion 4.1. Electrochemical acidification model The contributions of ionic species to conductivity were calculated with electric mobility values of Milazzo [32], Barry and Lynch [33] and Dean [34] used previously [19]. For proteins, tprot (pH) was calculated as follows:

⎡

Fig. 3. Evolution of skim milk conductivity during bipolar membrane electroacidification.

the initial electrochemical conductivity value to reach the value of 5.15 observed for chemical acidification. Consequently, Eq. (24) became: s (pH) = 103.49 − 73.64 × pH + 20.38 × pH2 − 2.501 × pH3 + 0.1154 × pH4

(25)

3.8. Evolution of viscosity and impact on electrical mobility The viscosity of milk during electroacidification was assumed to be the same as during chemical acidification. However, the demineralization of skim milk during BMEA should slightly increase its viscosity as demonstrated for soybean during BMEA treatment [9], but no data are available for milk. Mobility values of ionic species were divided by Fc and F c , the same correction factors used for the

⎢ ⎣

F×

tprot (pH) = 1 − ⎢

  ⎤ zi  · i · Ci (pH) ⎥ i−prot ⎥ ⎦ s (pH)

(27)

The results for transference numbers of each charged species are presented in Table 2. As observed for chemical acidification, the contribution of each ion to conductivity was very different from one species to another and varied mainly with the pH. Chloride made one of the more significant contributions throughout acidification, with values ranging from 0.1223 to 0.2423. Its contribution increased slightly over pH 6.6–6.0, decreased over pH 6.0–4.9 and then increased again to a value equivalent to its initial level. Its constant concentration was not affected by electromigration phenomena although the viscosity increase over pH 5.5–5.0 [23] decreased chloride mobility and hence its contribution to conductivity. Chlorine relative contribution was maximal at pH 6.0, but can be considered constant in the pH range 6.6–5.5, due to its constant concentration in electroacidified milk. Its contribution changed when calcium and phosphate were removed from casein

Table 2 Correction factor (Fc ), conductivity of milk (s in mS cm−1 ), relative contribution (t) of each ionic species to conductivity, sum of ionic species contribution and contribution of protein to conductivity as a function of pH during electrochemical acidification of reconstituted skim milk (10%, w/v). pH

Fc

s

tH+

tCl−

tCa2+

tP x−

tMg2+

tK+

tNa+

tcit 3−

Sum

tprot

6.6 6.5 6.4 6.3 6.2 6.1 6.0 5.9 5.8 5.7 5.6 5.5 5.4 5.3 5.2 5.1 5.0 4.9 4.8 4.7 4.6 4.5 4.4

1.02 1.02 1.03 1.03 1.04 1.05 1.06 1.07 1.08 1.11 1.14 1.18 1.25 1.36 1.54 1.82 2.16 2.29 2.04 1.70 1.46 1.31 1.22

5.15 5.04 4.95 4.86 4.79 4.73 4.67 4.62 4.58 4.54 4.50 4.46 4.42 4.39 4.35 4.32 4.29 4.26 4.24 4.23 4.22 4.22 4.24

0.9 × 10−5 1.2 × 10−5 1.5 × 10−5 1.9 × 10−5 2.5 × 10−5 3.2 × 10−5 4.0 × 10−5 5.0 × 10−5 6.3 × 10−5 7.8 × 10−5 9.7 × 10−5 11.8 × 10−5 14.2 × 10−5 16.5 × 10−5 18.5 × 10−5 19.8 × 10−5 21.2 × 10−5 25.4 × 10−5 36.0 × 10−5 54.6 × 10−5 80.2 × 10−5 112.2 × 10−5 151.3 × 10−5

0.2269 0.2314 0.2351 0.2380 0.2402 0.2417 0.2423 0.2421 0.2408 0.2381 0.2336 0.2265 0.2158 0.2000 0.1782 0.1518 0.1287 0.1223 0.1381 0.1661 0.1937 0.2154 0.2307

0.0256 0.0275 0.0302 0.0344 0.0409 0.0509 0.0657 0.0863 0.1125 0.1421 0.1704 0.1920 0.2025 0.1993 0.1824 0.1555 0.1291 0.1182 0.1272 0.1451 0.1608 0.1711 0.1770

0.0443 0.0446 0.0450 0.0454 0.0460 0.0468 0.0478 0.0494 0.0517 0.0548 0.0588 0.0630 0.0661 0.0664 0.0628 0.0559 0.0488 0.0473 0.0542 0.0659 0.0776 0.0871 0.0944

0.0066 0.0068 0.0069 0.0070 0.0071 0.0072 0.0075 0.0079 0.0087 0.0101 0.0121 0.0144 0.0159 0.0159 0.0144 0.0118 0.0093 0.0081 0.0082 0.0089 0.0095 0.0099 0.0101

0.1780 0.1667 0.1532 0.1384 0.1230 0.1076 0.0927 0.0788 0.0664 0.0557 0.0469 0.0399 0.0346 0.0307 0.0276 0.0249 0.0230 0.0244 0.0307 0.0411 0.0526 0.0630 0.0710

0.1118 0.1144 0.1167 0.1186 0.1202 0.1213 0.1219 0.1220 0.1215 0.1200 0.1174 0.1131 0.1066 0.0971 0.0843 0.0692 0.0558 0.0497 0.0517 0.0565 0.0594 0.0596 0.0579

0.0066 0.0068 0.0069 0.0070 0.0070 0.0071 0.0071 0.0071 0.0070 0.0070 0.0068 0.0066 0.0063 0.0058 0.0052 0.0044 0.0038 0.0036 0.0040 0.0049 0.0057 0.0063 0.0067

0.5998 0.5980 0.5939 0.5889 0.5844 0.5825 0.5850 0.5936 0.6087 0.6279 0.6462 0.6557 0.6479 0.6155 0.5551 0.4737 0.3987 0.3737 0.4145 0.4889 0.5601 0.6136 0.6492

0.4002 0.4020 0.4061 0.4111 0.4156 0.4175 0.4150 0.4064 0.3913 0.3721 0.3538 0.3443 0.3521 0.3845 0.4449 0.5263 0.6013 0.6263 0.5855 0.5111 0.4399 0.3864 0.3508

i

L. Bazinet et al. / Journal of Membrane Science 352 (2010) 32–40

micelles at pH 5.4. Potassium, calcium and sodium also made significant contributions to conductivity, ranging from 0.0230 to 0.1780, 0.0256 to 0.2025 and 0.0497 to 0.1220 respectively, although their evolutions during BMEA treatment differ. Potassium contribution decreased from pH 6.6 to 5.0 and reached its lowest value and then increased slightly while sodium and calcium contributions increased until pH 5.9 and 5.4 respectively were reached, decreased until pH 4.9 and then increased again. The contribution of potassium was maximum at pH 6.6 with tK+ = 0.1780 since it is the main free cation in milk, and it migrated through the CEM until pH 5.4 where its concentration was very low in milk: more than 76% migration [20]. As potassium concentration decreased, sodium contribution increased. At pH 6.1, when the concentration of sodium became higher than that of potassium, sodium became the predominant free cation. The maximum relative contribution of sodium with tNa+ = 0.1220 was at pH 5.9, when 60% of the potassium have migrated. In fact according to the linear migration of potassium [20], the pH of 5.94 would represent the pH value at which 50% of the potassium has migrated and pH 5.97, the pH value where the potassium concentration is equal to sodium [20]: these values are close to the theoretical value of 6.1 calculated by the model and corresponding to the maximal relative contribution of sodium. For calcium, its maximum relative contribution was at pH 5.4; this value corresponds to the beginning of its migration in milk and its removal from the casein micelles since no significant amount of proteins was precipitated between pH ranging from 6.6 to 5.4 [20]. As previously noted for chemical acidification, the contribution of phosphate regardless of its form, was quite constant throughout electrochemical acidification from pH 6.6 to 5.0, with an average value of 0.0528 ± 0.008, while at the end of the process its contribution increased exponentially from 0.0488 to 0.0944. Magnesium and citrate had little impact on the variation of skim milk conductivity. Hydrogen ion contribution during electrochemical acidification was twice the contribution observed for chemical acidification but its highest value was 151.3 × 10−5 at the end of the procedure. Its contribution was thus extremely low and could be neglected. These evolutions confirm the fact that potassium is the main electromigrating ion due to its greater mobility and relatively high concentration in skim milk [20,27,28]. Near pH 5.4, however its concentration is no longer sufficient to counterbalance electrogenerated H+ , forcing cations such as sodium and calcium to maintain ionic balance [20]. From the model, it appeared that sodium would be the first ion mobilized to counterbalance H+ . This would begin near pH 5.9, when the contribution of sodium to conductivity began to decrease. Thereafter, at lower pH, both calcium and sodium would counterbalance electrogenerated H+ . For chloride, since its concentration in milk was constant, we could bring out from the model, that it was its mobility which was affected by the viscosity of the solution, decreasing consequently its contribution to the global conductivity of skim milk. Proteins made a major contribution to conductivity, depending on their charges, with values ranging from 0.3443 to 0.6263. The evolution of protein contribution to conductivity could be separated into four phases. From pH 6.6 to 6.1, protein contribution was quite constant at 0.4070 ± 0.0063. From pH 6.1 to 5.5, protein contribution decreased linearly to 0.3521 and then from pH 5.5 it increased exponentially to its maximum of 0.6263 at pH 4.9. In the fourth phase, beginning at pH 4.9, the protein contribution dropped quickly to its minimum of 0.3508 at pH 4.4. These protein contribution trends are in agreement with results reported in the literature and particularly with those published for ␨-potential evolution [35–39]. A decrease in ␨-potential, which represents the surface charge of colloids, may be related to a decrease in contribution to conductivity and vice versa. However, milk is composed of different proteins, divided into two major groups, namely caseins (␣s1 -, ␣s2 -, ␤-, and ␬-), and whey proteins dominated by ␤-lactoglobulin

37

and ␣-lactalbumin [26,39,40] having different isoelectric points, sizes and volumes. Since the protein physico-chemical characteristics are completely different, the change in the number of free ions and also the kind of free ions could have also influenced the protein relative contribution to conductivity. Indeed, the milk composition in terms of free ␣-lactalbumin and ␤-lactoglobulin did not really change during pH variation while the composition in free caseins varied since caseins are removed from the micelles changing consequently their relative conductivity contributions. Furthermore, in the pH range between 5.2 and 6.7, ␤-lactoglobulin is present in milk as dimer, and these are stable with a molecular weight of 36,700 Da [39]. However, in the pH range of 4–5.2, the dimers of ␤-lg aggregate via physico-chemical interactions to form octamers of 147,000 Da with an optimum at 4.65 [39]. In addition, theoretically, according to its isoelectric point range at pH values under 5.3–5.5 [41], the ␤-lactoglobulin would become positively charged and could no more interact with cations but with anions. Hence, at pH under 5.3–5.5, the composition of the double layer at the interface of this protein would have changed in terms of composition from negatively charged protein/cations/anions to positively charged protein/anions/cations. Such phenomenon did not affect ␣-lactalbumin in this work since its isoelectric pH range is between 4.2 and 4.5 [41]. Consequently, the octamerization phenomenon coupled or not with the progressive neutralization of ␤-lactoglobulin could explain, during milk electroacidification, the decrease in protein relative contribution observed between pH 6.0 and 5.5, in parallel with no precipitation of casein [42]. Whatever the casein type, during bipolar membrane electroacidification, the removal of the different caseins from micelles and their precipitations begin at a pH close to 5.4 [42]. This would correspond to the point where the protein relative contribution to conductivity increased drastically. The caseins removed from the micelles would therefore interact with ions present in the medium, and probably calcium since ␣s1 -casein and ␣s2 -casein are known to interact with this cation. As described by Walstra [43], under the pH value of 5.2, the hairy structure of the casein micelles disappears due to the decrease in electrostatic repulsions and then, the attractions between micelles would be mainly due to ionic interactions. Parker and Dagleish [44] and Horne [45] reported that calcium adsorption on ␣s1 -casein resulted in the reduction of its electrical charge and the creation of reactive sites by which the aggregation was initiated. At pH 4.9, which would correspond to the precipitation of 50% of protein during electrochemical acidification [46], the relative contribution of protein would decrease due to an increase in the voluminosity of the casein aggregates directly linked to an increase of potential protein-protein interactions. In fact, Banon and Hardy [47] mentioned that during chemical acidification of milk, the diameters of the casein superstructures increased drastically with a decrease in pH from 5.2 to 4.6. 4.2. Comparison between chemical and electrochemical model results The relative contributions by chloride, potassium, calcium, sodium, phosphate and protein to the conductivity of skim milk as a function of acidification by HCl according to the previous chemical model [19] and by BMEA as calculated by the proposed model are shown in Fig. 4. The curves obtained for chemical and electrochemical acidification had the same general tendencies for chloride, calcium and phosphate. However, chloride contribution to conductivity was consistently higher during chemical acidification while calcium and phosphate contributions were higher during electrochemical acidification. For the other charged species, the curves had more divergent shapes. The decrease in potassium contribution was more marked during electrochemical acidification, with a first

38

L. Bazinet et al. / Journal of Membrane Science 352 (2010) 32–40

Fig. 4. Relative contributions to conductivity of the major charged species present in skim milk during electrochemical (䊉) and chemical () acidification.

inflection point and greatest separation from chemical acidification occurring at about pH 5.6. After the second inflection point at pH 4.9, the difference between both curves was constant. The contribution of sodium to conductivity was higher during electrochemical acidification than chemical acidification almost throughout the pH decrease, although the tendencies of both curves were different. From the beginning of the acidification, the difference increases until pH 5.5, then decreased and converged at the end of the process. For proteins, electrochemical acidification produced a greater contribution than chemical acidification over the pH 6.2–5.5 range, after which the difference decreased considerably and the two curves were of almost identical form. The differences observed between electrochemical and chemical acidifications in the tendencies of charged species contributions

to conductivity may be explained by two main phenomena of opposite nature: the addition of chloride during chemical acidification and the migration of potassium during electrochemical acidification. The significant addition of chloride during chemical acidification explains why the contributions of sodium, calcium and phosphate were lower than observed for these species during electroacidification as well as the large difference observed for chloride between both procedures. The migration of potassium during electrochemical acidification also contributed to increasing the differences for sodium, calcium and phosphate between the two procedures. The models also indicated that the increase in conductivity in the case of chemical acidification was due to the chloride added by HCl and calcium solubilized from casein micelles and not to hydrogen ion added by acidification. In addition, although large

L. Bazinet et al. / Journal of Membrane Science 352 (2010) 32–40

differences were observed between electrochemical and chemical acidification for each major ionic species, the differences in contribution by protein were not so significant. These observations warrant the investigation of two hypotheses. The first is that the difference between the two acidification treatments for proteins was real because there was a difference in their precipitation behaviour. The second hypothesis is that the difference was not real, merely an artefact of the correction factor calculated for chemical acidification. According to this hypothesis, regardless of the procedure, the contribution by proteins would be the same and would not differ as a function of the ionic species present. However, the first hypothesis would be supported by the observation by Bazinet et al. [31] of a delay in protein precipitation during chemical compared to electrochemical acidification and by the fact that demineralization of skim milk prior to either chemical or electrochemical treatment increases the pH at which all caseins are precipitated [18]. This shift reflects the differences in all relative contributions to conductivity observed between the two procedures at pH 5.5 just before the sharp increase in solution viscosity. In fact, below this pH, these differences were quite constant until the end of the acidification. The difference observed for protein near pH 5.5 was also quite constant during the precipitation phase. The differing ionic strengths produced by chemical and electrochemical acidification affect electrostatic repulsion between proteins, as demonstrated by Bazinet et al. [18]. Since the greater contribution of protein to conductivity during electrochemical acidification was still present higher at pH 5.5, before precipitation and formation of precipitated particles, electrostatic repulsions would be greater than with chemical acidification; this would explain the precipitation of protein during electrochemical acidification at a higher pH value, pH 4.6 vs 4.4–4.5. The difference in protein contribution may have been larger, the correction factor calculated for chemical acidification Fc and used for electrochemical acidification having probably minimized this difference. 5. Conclusion According to our model, chloride, potassium, calcium and sodium made similar contributions to conductivity in skim milk during electrochemical acidification. Sodium would be the first ion to decrease in order to counterbalance H+ after the potassium concentration became insufficient due to transmembrane migration. The contribution of chloride increased regularly, since its concentration was not affected by electromigration. Protein contributions to conductivity, depending on their charges, would range from 0.3443 to 0.6263. The evolution of protein contribution to conductivity was the same as observed previously for chemical acidification, but with slightly higher values. The chemical and electrochemical models differ somewhat in relative contributions to conductivity of the ionic charges in skim milk and in the evolution of these contributions. Throughout the pH decrease, chloride contribution to conductivity was greater during chemical acidification while the decrease in potassium contribution was more marked during electrochemical acidification. Since the contribution of proteins during electrochemical acidification was higher at pH 5.5, before precipitation and formation of precipitated particles, ionic strength was lower and the electrostatic repulsions higher than with chemical acidification, explaining the precipitation of protein during electrochemical acidification at a higher pH value. A limitation of the estimations carried out in this work is the use of the correction factor Fc , calculated for skim milk viscosity evolution during chemical acidification. Its application to the electrochemical model would cause an underestimation of the protein contribution to conductivity. The difference in protein contribution

39

to conductivity between chemical and electrochemical acidification should thus be larger. Since the viscosity of electroacidified milk has never been published, we will measure the viscosity of milk during BMEA and thereby to increase the accuracy of the model. Furthermore, to investigate the possible influence of chloride on the size of casein particles formed during acidification, work on the evolution of particle size and changes in ␨-potential during electrochemical acidification will be carried out.

References [1] D. Gardais, Les Procédés électriques de traitement des rejets industriels, in: Environnement et Electricité. Electra, Doppee difffusion, Avon, 1990, pp. 200–310. [2] L. Bazinet, Electrodialytic phenomena and their applications in the dairy industry: a review, CRC Crit. Rev. Food Sci. Nutr. 45 (2005) 307–326. [3] K.N. Mani, Electrodialysis water splitting technology, J. Membr. Sci. 58 (1991) 117–138. [4] K. Nagasubramanian, F.P. Chandla, K.J. Liu, Use of bipolar membranes for generation of acid and base—an engineering and economic analysis, J. Membr. Sci. 2 (1977) 109–124. [5] K.N. Mani, F.P. Chlanda, K. Nagasubramanian, Aquatech membrane technology for recovery of acid/base values from salt stream, Desalination 68 (1988) 149–167. [6] L. Bazinet, F. Lamarche, D. Ippersiel, J. Amiot, Bipolar membrane electroacidification to produce bovine milk casein isolate, J. Agric. Food Chem. 47 (1999) 5291–5296. [7] J. Balster, L. Pünt, H. Lammers, A.B. Verver, D.F. Stamatialis, M. Wessling, Electrochemical acidification of milk by whey desalination, J. Membr. Sci. 303 (2007) 213–220. [8] M.P. Mier, R. Ibanez, I. Ortiz, Influence of process variables on the production of bovine milk casein by electrodialysis with bipolar membranes, Biochem. Engin. J. 40 (2008) 304–311. [9] L. Bazinet, F. Lamarche, R. Labrecque, R. Toupin, M. Boulet, D. Ippersiel, Electroacidification of soybean proteins for the production of isolate, Food Technol. 51 (1997), 52–56, 58, 60. [10] M. Mondor, D. Ippersiel, F. Lamarche, J.I. Boye, Production of soy protein concentrates using a combination of electroacidification and ultrafiltration, J. Agric. Food Chem. 52 (2004) 6991–6996. [11] L. Bazinet, D. Ippersiel, R. Labrecque, F. Lamarche, Effect of temperature on the separation of soybean 11S and 7S protein fractions during bipolar membrane electroacidification, Biotechnol. Prog. 16 (2000) 292–295. [12] F. Lin Teng Shee, J. Arul, S. Brunet, L. Bazinet, Chitosan solubilization by bipolar membrane electroacidification: reduction of membrane fouling, J. Membr. Sci. 290 (2007) 29–35. [13] F. Lin Teng Shee, J. Arul, S. Brunet, L. Bazinet, Performing a three-step process for conversion of chitosan to its oligomers using an unique bipolar membrane electrodialysis system, J. Agric. Food Chem. 56 (2008) 10019–10026. [14] F. Lin Teng Shee, P. Angers, L. Bazinet, Delipidation of a whey protein concentrate by electroacidification with bipolar membranes (BMEA), J. Agric. Food Chem. 55 (2007) 3985–3989. [15] J.-S. Tronc, F. Lamarche, J. Makhlouf, Enzymatic browning inhibition in cloudy apple juice by electrodialysis, J. Food Sci. 62 (1997) 75–78. [16] A. Lam Quoc, M. Mondor, F. Lamarche, D. Ippersiel, L. Bazinet, J. Makhlouf, Effect of a combination of electrodialysis with bipolar membranes and mild heat treatment on the browning and opalescence stability of cloudy apple juice, Food Res. Int. 39 (2006) 755–760. [17] E. Vera Calle, J. Ruales, J. Sandeaux, M. Dornier, F. Persin, M. Reynes, G. Pourcelly, Comparison of different methods for deacidification of clarified passion fruit juice, J. Food Eng. 59 (2003) 361–367. [18] L. Bazinet, D. Ippersiel, C. Gendron, J. René-Paradis, J. Beaudry, M. Britten, B. Mahdavi, J. Amiot, F. Lamarche, Bipolar membrane electroacidification of demineralized skim milk, J. Agric. Food Chem. 49 (2001) 2812–2818. [19] L. Bazinet, F. Castaigne, Y. Pouliot, Relative contribution of proteins to conductivity changes in skim milk during chemical acidification, Appl. Engin. Agric. 21 (2005) 455–464. [20] L. Bazinet, D. Ippersiel, C. Gendron, J. Beaudry, B. Mahdavi, J. Amiot, F. Lamarche, Cationic balance in skim milk during bipolar membrane electroacidification, J. Membr. Sci. 173 (2000) 201–209. [21] Y. Graet (Le), G. Brulé, Les équilibres minéraux du lait: influence du pH et de la force ionique, Lait 73 (1993) 51–60. [22] A.J.R. Law, J. Leaver, Effects of acidification and storage of milk on dissociation of bovine casein micelles, J. Agric. Food Chem. 46 (1998) 5008– 5016. [23] H. Attia, N. Kherouatou, J. Ayadi, Acidification chimique directe du lait: corrélation entre la mobilité du matériel micellaire et les micro et macrostructures des laits acidifiés, Sciences des Aliments 20 (2000) 289–307. [24] A.C.M. Van Hooydonk, H.G. Hagerdoorn, I.J. Boerrigter, pH-induced physicochemical changes of casein micelles in milk and their effect on renneting. I. Effect of acidification on physichochemical properties, Neth. Milk Dairy J. 40 (1986) 281–296.

40

L. Bazinet et al. / Journal of Membrane Science 352 (2010) 32–40

[25] M.H. Lopez Leiva, The use of electrodialysis in food processing. Part I. Some theoretical concepts, Lebensm. -Wiss. Techn. 21 (1988) 119–125. [26] H.E. Swaisgood, Characteristics of edible fluids of animal origin: milk, in: O.R. Fennema (Ed.), Food Chemistry, 2nd ed., Marcel Dekker Inc., New York, 1985, pp.791–827. [27] J.C.D. White, D.T. Davies, The relation between the chemical composition of milk and the stability of the caseinate complex. I. General introduction, description of samples, methods and composition, J. Dairy Res. 25 (1958) 236– 255. [28] C. Holt, The milk and salts: their secretion, concentrations and physical chemistry, in: P.F. Fox (Ed.), Developments in Dairy Chemistry, vol. 3, Elsevier Applied Sciences Publishers, London, 1985, pp.143–181. [29] T.A. Davis, J.D. Genders, D. Pletcher, A First Course in Ion Permeable Membranes, The Electrochemical Consultancy, Romsey, 1997. [30] C. Holt, D.G. Dalgleish, R. Jenness, Calculation of the ion equilibria in milk diffusate and comparison with experiment, Anal. Biochem. 113 (1981) 154– 163. [31] L. Bazinet, F. Lamarche, D. Ippersiel, C. Gendron, B. Mahdavi, J. Amiot, Comparison of electrochemical and chemical acidification of skim milk, J. Food Sci. 65 (2000) 1303–1307. [32] G. Milazzo, Table II-9. In Électrochimie: Bases théoriques. Applications analytiques. Électrochimie des colloïdes. Tome 1. Dunod, Paris, (1969) p. 375. [33] P.H. Barry, J.W. Lynch, Topical Review. Liquid junction potentials and small cell effects in patch clamp analysis, J. Membr. Biol. 121 (1991) 101–117. [34] J.A. Dean, Lange’s Handbook of Chemistry, 14th ed., McGraw-Hill, New York, 1992. [35] D.G. Dalgleish, Measurements of electrophoretic mobilities and zeta-potentials of particules from milk using Doppler electrophoresis, J. Dairy Res. 51 (1984) 425–436. [36] T. Wade, J.K. Beattie, W.N. Rowlands, M.A. Augustin, Electroacoustic determination of size and zeta potential of casein micelles in skim milk, J. Dairy Sci. 63 (1996) 387–404.

[37] H.J. Vreeman, B.W. Van Markwijk, P. Both, The structure of casein micelles between pH 5.5 and 6.7 as determined by light-scattering, electron microscopy and voluminosity experiments, J. Dairy Res. 56 (1989) 463– 470. [38] H.J. Vreeman, B.W. Van Markwijk, P. Both, Size distribution and average size parameters of casein micelles determined by electron microscopy in bovine milk between pH 5.5 and 6.7. A comparison of several evaluation methods, J. Dairy Res. 58 (1991) 299–312. [39] P. Cayot, D. Lorient, Structures et technofonctions des protéines du lait. Arilait Recherches, Technique et Documentation Lavoisier, Paris, (1998) 363 p. [40] J. Amiot, S. Fournier, Y. Lebeuf, P. Paquin, R. Simpson, Composition, propriétés physicochimiques, valeur nutritive, qualité technologique et techniques d’analyses du lait, in: C.L. Vignola (Ed.), Science et Technologie du lait: Transformation du lait, Presses Internationales Polytechnique. Fondation de Technologie Laitière du Québec Inc., 2002, pp. 1–73. [41] K.R. Marshall, Industrial isolation of milk proteins: whey proteins, in: F. Fox (Ed.), Developments in Dairy Chemistry, vol. 1, Applied Science Publishers, London and New York, 1982, pp. 339–373 (Chapter 11). [42] L. Bazinet, D. Ippersiel, C. Gendron, B. Mahdavi, J. Amiot, F. Lamarche, Effect of added salt and increase in ionic strength on skim milk electroacidification performances, J. Dairy Res. 68 (2001) 237–250. [43] P. Walstra, On the stability of casein micelles, J. Dairy Sci. 73 (1990) 1965–1979. [44] T.G. Parker, D.G. Dalgleish, The potential application of the theory of branching processes to the association of milk protein, J. Dairy Res. 44 (1977) 79– 84. [45] D.S. Horne, The kinetics of the precipitation of chemically modified ␣s1 -casein by calcium, J. Dairy Res. 46 (1979) 265–269. [46] L. Bazinet, D. Ippersiel, C. Gendron, C. Tétreault, J. René-Paradis, J. Beaudry, M. Britten, B. Mahdavi, J. Amiot, F. Lamarche, Comparison between reconstituted and fresh skim milk chemical and electrochemical acidification, J. Sci. Food Agric. 82 (2002) 1356–1364. [47] S. Banon, J. Hardy, A colloidal approach of milk acidification by glucono-deltalactone, J. Dairy Sci. 75 (1992) 935–941.