Assessing ionic transport during apple juice electro-acidification: influence on system efficiency

Journal of Membrane Science 246 (2005) 217–226

Assessing ionic transport during apple juice electro-acidification:influence on system efficiency

M. Mondora, A. Lam Quocb, F. Lamarchea,∗, D. Ippersiela, J. Makhloufb

a Food Research and Development Center, Agriculture and Agri-Food Canada, Saint-Hyacinthe, Que., Canada J2S 8E3b Department of Food Sciences and Nutrition, Laval University, Sainte-Foy, Que., Canada G1K 7P4

Received 8 May 2003; received in revised form 6 August 2004; accepted 15 August 2004

Abstract

The purpose of this work was to assess ionic transport, during electro-acidification of apple juice, and its influence on the acidification rateand energy usage. In order to fulfill this objective, advanced current–voltage characterization of monopolar (cationic and anionic) and bipolarmembranes was carried out. Experiments were conducted using two electrodialysis configurations (bipolar-cationic membranes and bipolar-a uice wasa ed to the ionict nsmembranep boundaryl fit from thea n,a thef©

K

1

cqdicibic

f

d-bly.tialionor.them--A).the

outwasnfig-Hared

0d

nionic membranes) with two different systems (KCl–KCl and KCl–juice). For the bipolar-anionic configuration, the system HCl–jlso considered. The characteristic values of the transmembrane potential and of the estimated membrane resistance were correlat

ransport through the membranes, and to the energy usage of the systems. Influence of the membrane boundary layers on the traotential was also investigated by working at different feed flow rates. It was shown that for the present operating conditions, the

ayers do not affect the transmembrane potential. Furthermore, use of HCl with the bipolar-anionic configuration enables to benedvantages of each configuration: a small anionic membrane resistance due to the transport of H+, as for the bipolar-cationic configurationd a high acidification rate of the apple juice due the neutralization of OH− by HCl. It was for this system that the acidification rate was

astest.2004 Elsevier B.V. All rights reserved.

eywords:Electrodialysis; Foods; Ionic transport; Membrane potentials; Membrane resistance

. Introduction

Demand for cloudy or unclarified apple juice has in-reasing market potential due to its sensory and nutritionalualities. However, the production of high-quality juice isifficult because of its sensory instability. Cloudy apple juice

s very sensitive to enzymatic browning because it containsonsiderable quantities of polyphenols and polyphenol ox-dases (PPO). Enzymatic browning reactions are catalyzedy PPO and result in the oxydation of phenolic compounds

nto o-quinones, which then polymerize into complex dark-olored pigments[1].

∗ Corresponding author. Tel.: + 1 450 773 1105x222;ax: +1 450 773 8461.

E-mail address:[email protected] (F. Lamarche).

Acidification of cloudy apple juice to pH 2.0 and reajusting the pH to its initial value was found to irreversiinhibit PPO activity and stabilize juice color[2]. Tronc et al[3,4] and Lam Quoc et al.[5] have demonstrated the potenof electrodialysis with bipolar membrane for the inhibitof PPO in cloudy apple juice, without altering juice flavTwo electrodialysis configurations were used to modifypH of cloudy apple juice: (A) bipolar and cationic mebranes (BP-C); (B) bipolar and anionic membranes (BPIn the BP-C configuration, addition of exogenous KCl tojuice was required to obtain a pH of 2.0[3,4]. However, withthe BP-A configuration, a pH of 2.0 was reached withaddition of exogenous KCl, moreover acidification rateincreased by a factor of 3, when compared to the BP-C couration[5]. A yield of 10 L of juice/m2 membrane/min at p2.0 was obtained with the BP-A configuration, as comp

376-7388/$ – see front matter © 2004 Elsevier B.V. All rights reserved.oi:10.1016/j.memsci.2004.08.018

218 M. Mondor et al. / Journal of Membrane Science 246 (2005) 217–226

to 3.3 L of juice/m2 membrane/min for the BP-C configura-tion.

Although these previous works have demonstrated the fea-sibility of using electrodialysis with bipolar membranes forthe prevention of enzymatic browning in cloudy apple juice,the fundamental information concerning the dynamic trans-port in the membranes and in the boundary layers remainslimited. Nevertheless, a better understanding of ion transportthrough monopolar and bipolar membranes is the first steptoward the optimization of electrodialysis conditions.

In the past, transient-state current or voltage curves wereused to study the ionic transport in the diffusion-controlledboundary layers next to ion permeable membranes[6], andto explain overlimiting current behavior[7,8]. Bipolar mem-brane current–voltage characterization has been investigatedbefore, mainly focusing on the dynamic development of theelectrical potential difference[9], but also to investigate thetransport in and the energy requirements of different bipolarmembranes under water splitting conditions[10,11]. In thiscontribution, the current–voltage behavior of the monopolar(cationic and anionic) and bipolar membranes was investi-gated for ideal system of KCl–KCl, and complex systemsof KCl–juice and HCl–juice. Characteristics of voltage re-sponse curves and of membranes resistances for the differentsystems are analyzed with focus on electrodialysis cell per-f

2

2

ion-e of twoo es ared ueouss trixi ingt rtp atrixa the-s ble toe em-b aren bilityo hant tion-a ta ism)[

2b

inep ason,

ionic transport in electrodialysis system is often described atthe membrane scale[13]. However, in some circumstances,the boundary diffusion layers may also significantly influencethe ionic transport and must be considered in the treatment[15].

There are three contributions to the transmembrane po-tential in absence of an external electrical field, that is to sayDonnan, diffusion and streaming potentials. Donnan poten-tials result from the difference of ion concentration betweenmembrane and adjacent bulk solutions, which leads to a po-tential difference at the interfaces[16].

In addition, if the bulk solutions have different concentra-tions, a transfer of ions will take place from the concentratedsolution to the diluted solution. Since ions are moving at dif-ferent velocities in the membrane, a separation of chargestakes place setting up an electrical field[16,17]. Thus eventhough no external electrical field is imposed on the sys-tem, an electrical potential exists through the monopolarmembrane (diffusion potential). When osmotic transfer ispresent, another potential (streaming) is observed. This po-tential arises from the influence of water transport on thepermeation of counter-ions in the membrane[18].

In presence of an external electrical field, there is a fourthcontribution to the transmembrane potential, since ions trans-fer will not only take place by diffusion but also by electro-m f them

lacedi term rm yers,t polarm

2

re atl nce:a s so-l a-t n oft ucha us-i e ofe em-b brane[

p ut ane ance:

R

w sayt f thei e

ormance (acidification rate and energy usage).

. Theory

.1. Monopolar membrane and ions transport

Monopolar membranes used in electrodialysis act asxchange membranes, and are considered a systemr several phases. In literature, ion-exchange membranescribed as a combination of a gel phase being an aqolution of fixed and mobile ions with the polymer mancluded, and an equilibrium electroneutral solution fillhe intergel spaces[12,13]. There is likewise a third inehase formed from hydrophobic parts of the polymer mnd/or from the inert binder introduced during the synis stage. Monopolar membranes are in theory permeaither anions or cations. However, in practice, anionic mranes shows a high permeability for protons, while theyearly impermeable for other cations, and the permeaf hydroxyl ions in a cationic membrane is much higher t

hat of other anions. This behavior is due to the exceplly high mobility of H+ and OH− ions, which have differennd more rapid transport mechanism (tunneling mechan

14].

.2. Transmembrane potential through monopolar andipolar membranes

Ion and water transport in the membranes will determredominately the process performances. For this re

igration. This potential is due to the ohmic resistance oembrane (ohmic potential)[18].Concerning the bipolar membranes, when they are p

nto an external electrical field, they will split waolecules, into protons and hydroxyl ions[19]. Since bipolaembrane consists of a cation and an anion selective la

he aforementioned potentials presented for the monoembrane still applied.

.3. Membrane resistance

Due to membranes microheterogeneousity, there aeast two different contributions to the membrane resista

“gel phase resistance” who is function of the aqueouution of fixed and mobile ions including the polymer mrix, and an “intergel phase resistance” who is functiohe electroneutral solution filling the intergel spaces ss fissures or inner electroneutral parts. It is possible

ng structure-kinetics models to describe the influencach fraction (conducting and non-conducting) on the mrane resistance and ionic transport through the mem

13,20].However, as pointed out by Wilhelm et al.[10,11], it is also

ossible from the potential measurement with and withoxternal electrical field, to estimate the membrane resist

= Rgel + Rintergel = �φE − �φ

I(1)

here R is the membrane resistance, that is tohe summation of the gel phase resistance, and ontergel phase resistance,�φE is the transmembran

M. Mondor et al. / Journal of Membrane Science 246 (2005) 217–226 219

Fig. 1. Electrodialysis experimental systems: (A) bipolar-cationic membranes configuration; (B) bipolar-anionic membranes configuration. The Ag/AgClelectrodes are denoted as: (), ( ), ( ). Bipolar membrane: BP; cationic membrane: C; and anionic membrane: A.

potential in presence of an external electrical field (Don-nan + diffusion + streaming + ohmic),�φ is the instanta-neous transmembrane potential, when the current is switchedoff (Donnan + diffusion + streaming), andI is the current. Forthe situation where the boundary diffusion layers signifi-cantly affect the ionic transport, their resistances will alsocontribute to the membrane resistance (Eq.(1)).

3. Materials and methods

3.1. Experimental systems

A six-compartment membrane module, as presented inFig. 1, was used for these experiments. The module used wasan ED-1-BP electrodialysis cell (100 cm2 of effective elec-trode surface) from Electrosynthesis Co. Inc. (Lancaster, PA).The anode, a dimensionally stable electrode (DSA), and thecathode, a 316 stainless steel electrode, were supplied withthe cell. The experiments were performed with two mem-brane configurations: bipolar-cationic (BP-C) and bipolar-anionic (BP-A). A total of five membranes are used: threebipolar membranes (BP-1 from Tokuyama Soda Ltd.) and twomonopolar membranes (Neosepta CMX cationic membranesor Neosepta AMX anionic membranes from Tokuyama SodaLtd.). The intermembrane gap is 0.75 mm, the membranet MXm ted ina them lacedi ranehw ms:K he

electro-acidification was studied with three different systems:KCl–KCl, KCl–juice and HCl–juice.

This arrangement defines three closed loops containing thesolution to alkalinize (0.2 M KCl solution or 0.2 M HCl so-lution), the solution to acidify (0.2 M KCl or the apple juice)and a 0.2 M Na2SO4 solution used as rinsing solution for theelectrodes of the electrodialysis cell. The apple juice (Oasis)was given by Lassonde Inc. (Rougemont, Que.). Each closedloop was connected to a separate external reservoir, allowingfor continuous recycling. The electro-acidification was car-ried out with electrolytes volumes of 5 L for the solution toalkalinize (KCl or HCl), 5 L for the KCl solution or 3 L forthe apple juice to acidify and 5 L for the Na2SO4 solution.The flow rate was controlled at 0.75 or 1.50 L/min, usingpanel-mount flow meters, and the temperature of the elec-trolytes was maintained at 15◦C. Electro-acidification pro-cess was monitored with a YSI conductivity meter (model35, Yellow Springs, OH) and a pH meter (model AP61,Fisher Scientific, Montreal, Que.). After each experiment,the membranes are restored by rinsing the compartmentswith water (three wash cycles of 10 min each) and immer-sion in the appropriate salt or acid solution. The energyrequired for the electro-acidification was estimated as de-scribed by Lam Quoc et al.[5]. Experiments were conductedrandomly and each experiment was conducted in dupli-c

3

piralA opo-l intot o not

hicknesses are 0.17 and 0.16 mm for the CMX and Aembranes, respectively. The membranes are equilibra0.2 M KCl or HCl solution for at least 24 h outsideembrane module. After then, the membranes are p

n alternance in the electrodialysis cell. Each membas an effective surface area of 100 cm2. The acidificationith the BP-C configuration was performed for two systeCl–KCl and KCl–juice. For the BP-A configuration, t

ate.

.2. Transmembrane potential measurement

The transmembrane potential was measured with sg/AgCl electrodes placed on each side of the mon

ar or bipolar membranes. The electrodes are insertedhe separators/turbulence promoters, such that they d

220 M. Mondor et al. / Journal of Membrane Science 246 (2005) 217–226

disturb the flow in the electrodialysis cell. The transmem-brane potential was measured with two RadioShack multi-meters model 22-178 (RadioShack, Inter Tan Ltd., Canada).Measurements through the monopolar and bipolar mem-branes are performed simultaneously, as a function of time,under a constant cell potential of 10 V. In a first set ofexperiments, the residual potential (potential free from IReffects: Donnan + diffusion + streaming potentials) was mea-sured every 5 min, as the instantaneous potential value ob-tained upon interruption of the electrical field. At the end ofthe experiment, the evolution of the residual potential wasfollowed until the system reaches the steady state (approxi-mately 10 min).

Furthermore, in order to determine if the boundary lay-ers present at the membranes have a significant influence onthe transmembrane potential and on the value of the mem-brane resistance, a second set of experiments was performed.Every 5 min, the flow rate of the different solutions was in-creased from 0.75 to 1.50 L/min, until transmembrane poten-tial reached steady state. The variation of the transmembranepotential resulting from this increase was recorded.

3.3. Electrodes preparation

The Ag/AgCl electrodes were prepared by chlorination ofs thodd wm e in-fl firsts andw d ina wast to ana a plat-i het sityo re-m mityo

4

4e

4ar

m hta rma-t ely( ytg theH con-

Fig. 2. Electrodialysis with bipolar-cationic membranes (system KCl–KCl).Temperature: 15◦C; cell potential: 10 V: (A) (♦) pH–KCl acid; (�): �V−Cwith current; ( ) �V−C without current; (B) (♦) pH–KCl acid; (�)�V− BP with current; ( ) �V− BP without current; (C) (©) resistanceBP; and (�) resistanceC.

figuration, the overall result is the formation of HCl and KOH(Fig. 1B).

From Fig. 2 (BP-C) and 3 (BP-A), it was observed thatpH evolution in the acidic compartment was similar for bothconfigurations. The acidification rate was in the order of6× 10−4 mol of H+ accumulated in the solution to acidifyby minute. However, the BP-C configuration required 80%more energy to obtain a pH of 1.25 (9.20 kW h/m3 of KCl toacidify versus 5.10 kW h/m3). This is certainly due to the lossof H+ through the cationic membrane, which is more signif-icant than the loss of OH− through the anionic membrane inthe BP-A configuration.

In order to confirm this hypothesis, the K+ ionic fluxfor the BP-C configuration was estimated from the acid

ilver wires having a diameter of 0.65 mm, using a meerived from Shoemaker et al.[21]. Such electrodes alloeasurement of the transmembrane potential without th

uence of polarization voltages or over-potentials. In thetep, the silver wires are rinsed with methanol, acetoneater to remove the impurities. Then, they are immersemmoniac solution to remove any oxide traces. The wire

hen placed in an electrolysis bath and was connectednode of a continuous current source. The cathode was

num wire and the electrolyte was a 0.25 M KCl solution. Treatment was performed for 30 min with a current intenf 10–12 mA/cm2. Scanning electron microscopy measuents were done to verify the homogeneity and uniforf the AgCl deposit.

. Results and discussion

.1. Acidification rate and energy usage duringlectro-acidification

.1.1. System KCl–KClIn the BP-C configuration, the H+ generated by the bipol

embrane gradually replace the K+ which migrate throughe cationic membrane and form KOH with the OH− gener-ted by the bipolar membrane. The overall result is the fo

ion of HCl and KOH in compartment A and B, respectivFig. 1A). In the BP-A configuration, the OH− generated bhe bipolar membrane gradually replace the Cl− which mi-rate through the anionic membrane and form HCl with+ generated by the bipolar membrane. As for the BP-C

M. Mondor et al. / Journal of Membrane Science 246 (2005) 217–226 221

accumulation in the KCl solution to acidify, knowing thatthe K+ permeation through the cationic membrane was simi-lar to the rate of H+ accumulated in the acidic compartment.In the same manner, for the BP-A configuration, the accu-mulation of base into the compartment of KCl to alkalinizewas used to estimate the ionic flux of Cl− through the an-ionic membrane. These ionic fluxes were compared to thetotal ionic flux in order to establish the “useful flux” for eachspecies:

J = nI

Fz(2)

whereJ is the total ionic flux (mol/s),n is the number ofprimary stack (n= 2), I is the current (A),F is the Faradayconstant (96,486 A s/mol) andz is the valence absolute value(z= 1).

Based on the results, the permselectivity of the Cl− in theBP-A configuration reached 90.2%, as compared to 58.9%for the K+ in the BP-C configuration. This explains the higherenergy consumption for the BP-C configuration.

4.1.2. Systems KCl–juice and HCl–juiceThe acidification performance is function of the nature

and of the ionic concentration of the solution to acid-ify. From an electro-chemical point of view, apple juice ism antc arep2 i-n rings

thea se oft ingc i-t icew -b teadys ent[

asm thep re-s et e wasm tion(

so-l Hg e theca ratew d byH ilarf

4.2. Evolution of the transmembrane potential formonopolar and bipolar membranes, with and without anexternal electrical field

In this study, the variation with time of the transmem-brane potential, in presence of an external electrical field,was small. Only the systems where the transport of H+ leadsto a significant decrease of the membrane resistance (BP-C configuration (Figs. 2A and 4A) and BP-A configurationwith the system HCl–juice (Fig. 6A)) have resulted in amonopolar transmembrane potential decreasing with time.It was also observed that there is no clear difference be-tween the potential curves for the flow rate at 0.75 L/minversus 1.50 L/min (Figs. 4–6), which means that the contri-bution of the boundary layers resistance to the membraneresistance is small. Thus it is expected that they do not limitthe ion transport and will not be considered in this treat-ment.

In the absence of an external electrical field, the measuredtransmembrane potential corresponds to the summation ofthe Donnan, diffusion and streaming potentials through themembrane. Based on the reported results (Figs. 2A–6A), thetransmembrane potential was always small, a few millivolts,through the monopolar membranes. This residual potentialis not function of the solutions under treatment (KCl, juice,H sys-t

ab-s -n ntiali tro-d anda tiallyw f1 or-r em-b

endo raned po-t tran-s thiss thet utions si cl -i isr oryr ntra-t tsidet m 0( ns-m ipo-l senceo ons

ainly constituted of potassium which is the predomination (764± 45 mg/L), while magnesium and calciumresent but in significantly lower concentration (35± 5 and7± 3 mg/L, respectively)[5]. Malic acid is the predomant anion, it forms a potassium salt resulting in a buffeystem.

For both configurations, replacing the KCl solution incidic compartment by apple juice resulted in a decrea

he acidification rate which was controlled by the bufferapacity of malic acid (Fig. 4(BP-C) and 5 (BP-A)). In addion for the BP-C configuration, the acidification of the juas complicated by the loss of H+ through the cationic memrane. Both phenomena together explain the pseudo-state pH value of 2.25 reached after 90 min of treatm5].

For the system KCl–juice, the BP-A configuration wore performant than the BP-C configuration, due toermeation of Cl− anions through the anionic membraneulting in the retention of H+ in the acidic compartment. Tharget pH 2.0 was easily reached and the energy usagore than two times less than with the BP-C configura

5.35 kW h/m3 of juice versus 12.40 kW h/m3).Furthermore, the use of HCl instead of KCl, as the

ution to alkalinize, enables the neutralization of the O−enerated by the bipolar membrane and would eliminatompetition between both anions Cl− and OH− for perme-tion through the anionic membrane. The acidificationas increased by 25%, when KCl solution was replaceCl solution, while the energy requirement remains sim

or both systems.

Cl) since similar results were reported for the differentems under consideration.

However, for the bipolar membrane, the potential inence of an electrical field (Figs. 2B–6B) represents a sigificant fraction (around 60%) of transmembrane pote

n presence of an external electrical field. During elecialysis, the thin transitory region between the cationicnionic layers of the bipolar membrane contains essenater having a concentration of H+ and OH− in the order o× 10−7 M. In absence of an external electrical field, this c

esponds to a theoretical potential through the bipolar mrane of 0.72 V[19].

When the external electrical field is switch-off at thef the treatment, ions diffuse inside the bipolar membue to the concentration gradient. During that time, the

ential decreases while the concentration of ions in theitory region increases. For all systems considered intudy, except for the system HCl–juice, at the end ofreatment, we have an acid solution and an alkaline soleparated by the bipolar membrane. The H+ and other cation

n presence (mainly K+) will diffuse through the cationiayer of the bipolar membrane, while the OH− and other anons (mainly Cl−) will diffuse through the anionic layer. Thesults in the formation of water and KCl in the transitegion of the bipolar membrane. Since the ionic conceion inside the membrane was always different than ouhe membrane, the bipolar membrane potential differs froFigs. 2B–5B). This behavior of the steady state bipolar traembrane potential, which was different from zero for b

ar membrane placed between an acid and a base, in abf an external electrical field was also reported by Sim

222 M. Mondor et al. / Journal of Membrane Science 246 (2005) 217–226

Fig. 3. Electrodialysis with bipolar-anionic membranes (system KCl–KCl).Temperature: 15◦C; cell potential: 10 V: (A) (♦) pH–KCl acid; (�) �V−Awith current; ( ) �V−A without current; (B) (♦) pH–KCl acid; (�)�V− BP with current; ( ) �V− BP without current; (C) (©) resistanceBP; and (�) resistanceA.

[22]. Clearly, the value of the transmembrane potential atsteady state will be function of the ionic environment andconcentration.

However, for the system HCl–juice (Fig. 6B), this is theH+ and Cl− that diffuse inside the bipolar membrane andwill form an HCl solution in equilibrium with the bulk solu-tions. It was observed that for this system the bipolar resid-ual potential decreased to 0 in two steps: a slow first phasehaving a length of 10 min and a second quicker phase hav-ing a length of 5 min (Fig. 6B). When the external electricalfield is switch-off, the anionic layer of the bipolar membranecontains principally OH− ions. Similarly, the cationic layercontains principally H+ ions. The first diffusion step results

Fig. 4. Electrodialysis with bipolar-cationic membranes (systemKCl–juice). Temperature: 15◦C; cell potential: 10 V: (A) (♦) pH–juice; (�)�V−C with current: 0.75 L/min; (�) �V− C with current: 1.50 L/min;( ) �V− C without current; (B) (♦) pH–juice; (�) �V− BP with current:0.75 L/min; (©) �V− BP with current: 1.50 L/min; ( ) �V− BP withoutcurrent; (C) (©) resistance BP: 0.75 L/min; (�) resistance BP: 1.50 L/min;(�) resistanceC: 0.75 L/min; and (�): resistanceC: 1.50 L/min.

in the formation of water through the reaction of OH− andH+, during this step the difference of concentration betweenthe inside and the outside of the bipolar membrane decreasesslowly, as for the transmembrane potential. Then during thesecond step, that is to say when all the OH− ions have re-acted with the H+ ions, it is mainly the Cl− and H+ ionsthat will diffuse to form HCl in the transitory region. At thismoment the difference of concentration between the insideand outside of the bipolar membrane becomes less signifi-cant and ultimately the system reaches the equilibrium andthe transmembrane potential becomes 0. Thus the evolutionof the transmembrane potential is a quick and simple method

M. Mondor et al. / Journal of Membrane Science 246 (2005) 217–226 223

Fig. 5. Electrodialysis with bipolar-anionic membranes (system KCl–juice).Temperature: 15◦C; cell potential: 10 V: (A) (♦) pH–juice; (�) �V−Awith current: 0.75 L/min; (�) �V− A with current: 1.50 L/min; ( ) �V−Awithout current; (B) (♦) pH–juice; (�) �V− BP with current: 0.75 L/min;(©)V− BP with current: 1.50 L/min; ( )�V− BP without current; (C) (©)resistance BP: 0.75 L/min; (�) resistance BP: 1.50 L/min; (�) resistanceA:0.75 L/min; and (�) resistanceA: 1.50 L/min.

to characterize the exchange properties and ionic diffusion ofmonopolar and bipolar membranes.

4.3. Evolution of the membrane resistance

In Fig. 7, the evolution of the cell resistance is presented forthe different systems, under a cell potential of 10 V and a flowrate of 0.75 L/min. Despite the fact that important variationof solutions conductivities is observed during the treatment,the cell resistance remains relatively stable. It is only for theBP-A configuration with the system HCl–juice that a signif-icant decrease (35%) was observed. Clearly, the membraneresistance is the main factor affecting the cell resistance and

Fig. 6. Electrodialysis with bipolar-anionic membranes (system HCl–juice).Temperature: 15◦C; cell potential: 10 V: (A) (♦) pH–juice; (�) �V−Awith current: 0.75 L/min; (�) �V−Awith current: 1.50 L/min; ( ) �V−Awithout current; (B) (♦) pH–juice; (�) �V− BP with current: 0.75 L/min;(©) �V− BP with current: 1.50 L/min; ( ) �V− BP without current; (C)(©) resistance BP: 0.75 L/min; (�) resistance BP: 1.50 L/min; (�) resistanceA: 0.75 L/min; and (�) resistanceA: 1.50 L/min.

this parameter can be estimated only from the transmembranepotentials (Eq.(1)).

4.3.1. Cationic membrane resistanceFor the BP-C configuration with the system KCl–KCl,

an increase of 50% of the conductivity for the KCl solutionto acidify and approximately 30% for the KCl solution toalkalinize was observed, while the cationic membrane showsa decrease of the membrane resistance of only 33% (Fig. 2C)).It was also conceivable to observe a significant decrease ofthe cationic membrane resistance due to the increase of thepresence of H+ in the ionic flux with a decrease of pH. Inthis case, it is conceivable that another phenomenon occured

224 M. Mondor et al. / Journal of Membrane Science 246 (2005) 217–226

Fig. 7. Cell resistance. Temperature: 15◦C; cell potential: 10 V; flow rate: 0.75 L/min: (♦) BP-C KCl–KCl; (�) BP-A KCl–KCl; (©) BP-C KCl–juice; (�)BP-A KCl–juice; and (�) BP-A HCl–juice.

inside the cationic membrane and limited the decrease of themembrane resistance.

For the BP-C configuration with the system KCl–juice(Fig. 4C), a decrease of the cationic membrane resistancewas observed at the beginning of the treatment. This decreasewas attributable to the fact that the juice is poor in cations([K+] = 0.02 M) and the H+ have very quickly replaced theK+ in the ionic flux through the cationic membrane[4,5].For the juice, which is less conductive than the KCl solu-tion, the acidification resulted in a larger cationic membraneresistance.

4.3.2. Anionic membrane resistanceFor the BP-A configuration with the system KCl–KCl

(Fig. 3C), the resistance of the anionic membrane was ap-proximately seven times larger than the resistance of thecationic membrane with the BP-C configuration (a resistanceof 0.53 as compared to 0.075) (Fig. 2C). Based on thedata from the membrane manufacturer, the resistance of bothmembrane types is similar when operated with KCl solution.Furthermore, membrane thickness is similar for both mem-brane types, suggesting that this parameter should not have asignificant influence on the membrane resistance.

For the BP-A configuration, an increase of the anionicmembrane resistance from 0.53 to 0.9 was observed, whenthe KCl solution was replaced by the juice (Figs. 3C and 5C).T lessc

ion( nsp of thero paredt .F the

HCl solution, when compared to the KCl solution, for a givenconcentration of 0.2 M, would decrease the membrane resis-tance. Both phenomena have inverse effects on the membraneresistance and it was expected that they would compensateeach other. In practice, a significant decrease of the anionicmembrane resistance was observed for the system HCl–juice,when compared to the system KCl–juice (membrane resis-tance of 0.31 as compared to 0.9). Furthermore, althoughthe conductivity of the HCl solution decreases during thetreatment due to a decrease of the acid concentration, a de-crease of the anionic membrane resistance from 0.4 to 0.2

was observed (Fig. 6C).

4.3.3. Neutralization interfaceFor the situation where the pH of the bulk solutions are

significantly different, as it was the case for the present exper-iments, it was not a negligible counter-current transport of H+

and OH− that was observed, but a significant flux[14]. Forexample, in the production of acid and base by electrodialy-sis, some authors have reported a transport of H+ through theanionic membrane in the order of 66%[23].

The water concentration inside the membrane is also afactor that will influence the membrane resistance[24]. Inorder to explain the results obtained for the resistances, wepropose the notion of “neutralization interface”. It is a zoneinside the membrane, where water molecules are formed duet lli terf raner

Cl,w em-b g the4 BP-A in

his is certainly attributable to the fact that the juice isonductive than the KCl solution.

For the system HCl–juice with the BP-A configuratFig. 6C), by comparing the conductivity of both solutioermeating the membrane, we can predict an increaseesistance of the anionic membrane when the Cl− are thenly ions transported through the membranes, as com

o the situation where both Cl− and OH− are transportedrom another point of view, the larger conductivity of

o the reaction between H+ and OH−. This phenomenon winduce an effect of ionic dilution attributable to the waormation and this will result in an increase of the membesistance.

For the BP-C configuration with the system KCl–Kater formation may explain the fact that the cationic mrane resistance did not decrease as expected followin0% increase in KCl solutions conductivities. For theconfiguration with the system KCl–KCl, this resulted

M. Mondor et al. / Journal of Membrane Science 246 (2005) 217–226 225

an increase of the anionic membrane resistance followed bya stabilization of this resistance, instead of a decrease withthe increase of the conductivity of both solutions during thetreatment (Fig. 3C).

By developing in more details the notion of “neutralizationinterface”, it is clear that the exact position of this interfacewill also influence the effect of the dilution. The closer theinterface is from the downstream of the membrane, the largerthe zone of dilution in the membrane. If the zone of dilu-tion is close to the membrane upstream, its effect will be lesssignificant. The water formed inside the membrane by thereaction of H+ and OH− will become part of the main ionicflux, which leads to the conclusion that the membrane resis-tance is function of the position of the neutralization inter-face.

For the BP-A configuration with the system KCl–juice, itwas believe that the accumulation of OH− in the KCl solution,which was more significant than the accumulation of H+ in thejuice at the half time of the treatment, has contributed to movethe neutralization interface from the membrane downstreamto the membrane upstream, which decreased the effect ofdilution. This resulted in a decrease of the anionic membraneresistance after the half time of the treatment (Fig. 5C).

For the HCl–juice system, the resistance of the anionicmembrane is smaller than for the KCl–KCl and KCl–juices ore,t sig-n n oft rv-a f thed

4the

n tantt -t rnalem ulest

5

tionc nor-g ons.T he est entse ystemp

w andt f thec yr to a

low electro-acidification performance and to a high-energyconsumption. For the case of a biological solution that ispoor in ions, especially the cations, such as the apple juice,this configuration is not appropriate when a large pH variationis desired. A limit pH is imposed by the low concentration inorganic acid of the juice[5].

The cell resistances of the BP-A configuration were largerthan the resistances of the BP-C configuration (Fig. 7) dueto the water formation, which was more significant for theanionic membrane. However, the migration of anions (Cl−),which are responsible of the solution acidification, representsa larger fraction of the ionic flux than the K+ in the BP-C configuration. This resulted in a more efficient treatmentwhen compared to the BP-C configuration.

The use of HCl instead of KCl, for the BP-A configurationenables to benefit from the advantages of each configuration:a small anionic membrane resistance due to the transport ofH+, as for the BP-C configuration, and a high acidificationrate due the neutralization of OH− by the HCl. It was for thissystem that the acidification rate was the highest.

The measurement of the transmembrane potential in ab-sence of an external electrical field, enabled to establish thequalitative contribution of Donnan, diffusion and streamingpotentials to the total transmembrane potential. In absenceof an external electrical field, only the bipolar transmem-b tiala bledt thet flu-e sportw tes.I , theb portt

ts oftc ld ber ed int

A

s enP .,A Sci-e is ac-k

R

ess,

cti-990)

ystems due to the absence of water formation. Furthermhe formation of HCl inside the membrane decreasesificantly the anionic membrane resistance as a functio

reatment evolution (Fig. 6C). This phenomenon was obseble even when the concentration and the conductivity oownstream HCl solution are decreasing.

.3.4. Bipolar membrane resistanceThe variation of the bipolar membrane resistance, with

ature of the bulk solutions, was similar but less imporhan for monopolar membranes (Figs. 2C–6C). This is cerainly attributable to the fact that in presence of an extelectrical field, it is mainly the H+ and OH− ions that willigrate out of the transition layer and the water molec

hat will diffuse inside the membrane[19].

. Conclusion

In this study, the performance of two electro-acidificaonfigurations (BP-C and BP-A) was compared using ianic solutions and apple juice under different conditihe measurement of the transmembrane potential and t

imation of the membrane resistance during the treatmnabled to assess ion transport and its influence on serformance.

The cell resistances of the BP-C configuration (Fig. 7)ere lower than for the BP-A counterpart configuration,

his was principally attributable to the small resistance oationic membrane. However, the H+ who are not effectiveletained in the acidic compartment, have contributed

-

rane potential is significant. The evolution of the potenfter the external electrical field is switched-off also ena

o improve our understanding of the ionic diffusion intoransition layer of the bipolar membrane. Furthermore, innce of the membrane boundary layers on the ionic tranas also investigated by working at different feed flow ra

t was shown that for the present operating conditionsoundary layers do not significantly affect the ionic trans

hrough the membranes.However, further results including the measuremen

ransport number of anion (Cl−) and cations (K+, H+) and aomplete mass balance on electrodialysis system wouequired to fully demonstrate the assumptions proposhis paper.

cknowledgements

The financial support of the “Conseil des Rechercheeche et en Agroalimentaire du Quebec”, A. Lassonde incgriculture and Agri-Food Canada and of the Naturalnces and Engineering Research Council of Canadanowledged.

eferences

[1] J.J. Macheix, A. Fleuriet, J. Billot, Fruits Phenolics, CRC PrBoca Raton, FL, 1990.

[2] G.P. Zemel, C.A. Sims, M.R. Marshall, M. Balaban, Low pH inavation of polyphenol oxidase in apple juice, J. Food Sci. 55 (1562–565.

226 M. Mondor et al. / Journal of Membrane Science 246 (2005) 217–226

[3] J.S. Tronc, F. Lamarche, J. Makhlouf, Enzymatic browning inhibi-tion in cloudy apple juice by electrodialysis, J. Food Sci. 62 (1997)75–78.

[4] J.S. Tronc, F. Lamarche, J. Makhlouf, Effect of pH variation byelectrodialysis on the inhibition of enzymatic browning in cloudyapple juice, J. Agric. Food Chem. 46 (1998) 829–833.

[5] A. Lam Quoc, F. Lamarche, J. Makhlouf, Acceleration of pH incloudy apple juice using electrodialysis with bipolar membranes, J.Agric. Food Chem. 48 (2000) 2160–2166.

[6] R. Audinos, G. Pichelin, Characterization of electrodialysis mem-branes by chronopotentiometry, Desalination 68 (1988) 251–263.

[7] H.W. Rosler, F. Maletzki, E. Staude, Ion transfer across electrodial-ysis membranes in the overlimiting current range: chronopotentio-metric studies, J. Membr. Sci. 72 (1992) 171–179.

[8] J.J. Krol, M. Wessling, H. Strathmann, Chronopotentiometry andoverlimiting ion transport through monopolar ion exchange mem-branes, J. Membr. Sci. 162 (1999) 155–164.

[9] R.E. Moussaoui, G. Pourcelly, M. Maeck, H.D. Hurwitz, C. Gavach,Co-ion leakage through bipolar membranes influence onI–V re-sponses and water splitting efficiency, J. Membr. Sci. 90 (1994)283–292.

[10] F.G. Wilhelm, N.F.A. van der vegt, M. Wessling, H. Strathmann,Chronopotentiometry for the advanced current–voltage characteriza-tion of bipolar membranes, J. Electro. Chem. 502 (2001) 152–166.

[11] F.G. Wilhelm, N.F.A. van der vegt, H. Strathmann, M. Wessling,Comparison of bipolar membranes by means of chronopotentiometry,J. Membr. Sci. 199 (2002) 177–190.

[12] V.I. Zabolotsky, V.V. Nikonenko, Effect of structural membraneinhomogeneity on transport properties, J. Membr. Sci. 79 (1993)181–198.

[ lly,, De-

[14] H. Strathmann, G.-H. Koops, Process economics of the electro-dialytic water dissociation for the production of acid and base,in: A.J.B. Kemperman (Ed.), Handbook on Bipolar MembraneTechnology, Twenty University Press, Enschede, 2000, pp. 191–219.

[15] S. Mokrani, L. Dammak, G. Bulvestre, C. Larchet, B. Auclair, Ex-perimental and theoretical studies of the crossed ionic fluxes througha cation-exchange membrane, J. Membr. Sci. 199 (2002) 147–160.

[16] J.-P. Brun, Transport, in: Procedes de separation par membrane, Mas-son, Paris, 1989, pp. 1–82.

[17] V. Suendo, M. Minagawa, A. Tanioka, Membrane potential of abipolar membrane: the effect of concentration perturbation of theintermediate phase around a certain value, J. Electro. Chem. 520(2002) 29–39.

[18] B. Auclair, V. Nikonenko, C. Larchet, M. Metayer, L. Dammak, Cor-relation between transport parameters of ion-exchange membranes,J. Membr. Sci. 195 (2002) 89–102.

[19] H. Stratmann, J.J. Krol, H.-J. Rapp, G. Eigenberger, Limiting currentdensity and water dissociation in bipolar membranes, J. Membr. Sci.125 (1997) 123–142.

[20] J.-H. Choi, S.-H. Kim, S.-H. Moon, Heterogeneity of ion-exchangemembranes: the effects of membrane heterogeneity on transportproperties, J. Coll. Inter. Sci. 241 (2001) 120–126.

[21] D.P. Shoemakerand, C.W. Garland, J.I. Steinfeld, J.W. Nibler, Mis-cellaneous Procedures, Experiments in Physical Chemistry, fourthed, McGraw-Hill, New York, 1981, Chapter XIX, pp. 694–695.

[22] R. Simons, Preparation of a high performance bipolar membrane, J.Membr. Sci. 78 (1993) 13–23.

[23] C. Gavach, D. Raucp, G. Pourcelly, Production of sulfuric acid andcaustic soda from sodium sulphate by electromembrane processes,

[ able.

13] V. Nikonenko, V. Zabolotsky, C. Larchet, B. Auclair, G. PourceMathematical description of ion transport in membrane systemssalination 147 (2002) 369–374.

Desalination 91 (1993) 163–175.24] T.A. Davis, J.D. Genders, D. Pletcher, A first course in ion perme

membranes, The Electrochemical Consultance, Romsey, 1997