Anh Huong Cua PH Den Luu Luong Loc Sua

Original article

On the origin of flux dependence in pH-modifiedskim milk filtration

Murielle RABILLER-BAUDRY1,2*, Habib BOUZID

1,2,3, Bernard CHAUFER1,2,

Lydie PAUGAM1,2, David DELAUNAY1,2, Omar MEKMENE

4,Sarfraz AHMAD

4, Frédéric GAUCHERON4

1 Université Rennes 1, UMR CNRS 6226, Sciences Chimiques de Rennes, CS74205, Case 1011,35042 Rennes Cedex, France

2 Université Européenne de Bretagne, France3 Université de Mostaganem, Département de Chimie, Faculté des Sciences et Sciences de l’Ingénieur,

BP 227 Route de Bel-Hacel, 27000 Mostaganem, Algeria4 INRA, UMR1253, Sciences et Technologie du Lait et de l’Œuf, 65 rue de Saint-Brieuc,

35042 Rennes Cedex, France

Received 31 October 2008 – Accepted 3rd April 2009

Abstract – The aim of this study was to contribute to identify the physico-chemical origin of fluxvariations, namely the limiting flux (maximum) and the critical flux (sustainable), in ultrafiltration(UF), nanofiltration (NF) and reverse osmosis (RO) of skim milks modified by addition of HCl orNaOH. Both the limiting and the critical fluxes varied in a nonpredictable way with pH but withclose similar trends in UF, NF and RO, highlighting the leading role of the fluid behaviour. Physico-chemical characteristics of caseins, such as size and electrophoretic mobility, were measured in thepH range 1.9–11.5 and were correlated to fluxes in UF and to a less extent in NF and RO. Availabledata on the electrophoretic mobility of α-lactalbumin and β-lactoglobulin allowed to suggest thatserum proteins would also participate in the flux variations with a possible specific impact ofα-lactalbumin as an internal foulant in UF. The role of calcium and inorganic phosphate over thewide pH range was discussed by taking calculated salt equilibrium of milk as a function of acidicpH using a new software and new additional analyses in alkaline media. Results underlined thedetermining role of Ca2+ in the inorganic irreversible fouling of organic membranes allowingproposal of a simplified cleaning protocol over a wide pH range between 6.7 and 11.5.

ultrafiltration / nanofiltration / reverse osmosis / limiting flux / critical flux

摘要 – 不同 pH 的脱脂乳对膜过滤中通量的影响○ 本文研究了酸化和碱化的脱脂乳(添加盐酸或者氢氧化钠)进行超滤、纳滤及反渗透操作过程中通量变化的影响因素，研究结果表明极限通量及临界通量都随着脱脂乳 pH 值呈现不规则变化，但是变化趋势相近，其中流体的特性起着决定性的作用○ 在 pH 1.9~11.5 时，酪蛋白的粒径及电泳迁移率与超滤的通量相关，但是对纳滤及反渗透的影响不大○ α-乳白蛋白和 β-乳球蛋白电泳迁移率的分析结果表明乳清白蛋白对通量的变化也有影响，主要是 α-乳白蛋白是超滤过程膜的主要堵塞物○ 了计算出钙及无机磷酸盐的浓度与 pH 值之间的函数关系○ 在 pH 6.7~11.5 范围内简单的清洗步骤即可以有效去除钙对有机膜造成的不可逆污染○

超滤 / 纳滤 / 反渗透 / 极限通量 / 临界通量

*Corresponding author (通讯作者): [email protected]

Dairy Sci. Technol. 89 (2009) 363–385© INRA, EDP Sciences, 2009DOI: 10.1051/dst/2009018

Available online at:www.dairy-journal.org

Article published by EDP Sciences

http://dx.doi.org/10.1051/dst/2009018

http://www.dairy-journal.org

http://www.edpsciences.org/

Résumé – Sur l’origine de la dépendance des flux en filtration de lait écrémé à pHmodifié. L’objectif de cette étude était de contribuer à identifier les origines physico-chimiquesdes variations des flux critiques (maximum) et limites (durables) en ultrafiltration (UF), nano-filtration (NF) et osmose inverse (OI) de laits écrémés modifiés par ajout de HCl ou NaOH. Lesflux limites et critiques variaient de façon imprévisible en fonction du pH mais avec destendances similaires en UF, NF et OI, soulignant le rôle déterminant du fluide. Les caractéristi-ques physico-chimiques des caséines (taille, mobilité électrophorétique) ont été mesurées pourdes pH entre 1,9 et 11,5 et corrélées aux variations de flux en UF et dans une moindre mesureen NF et OI. Les données disponibles sur la mobilité électrophorétique de l’α-lactalbumine et laβ-lactoglobuline permettent de supposer que les protéines sériques participent également auxvariations de flux avec un impact spécifique de l’α-lactalbumine comme colmatant interne enUF. Le rôle des minéraux (calcium et phosphate inorganique) a été appréhendé en utilisant unlogiciel de calcul des équilibres salins en milieu acide, et analyses complémentaires en milieualcalin. Les résultats soulignent le rôle déterminant de Ca2+ dans le colmatage irréversible desmembranes organiques, permettant finalement de proposer un protocole simplifié de nettoyagepour une large gamme de pH entre 6,7 et 11,5.

ultrafiltration / nanofiltration / osmose inverse / flux limite / flux critique

1. INTRODUCTION

Membrane processes, namely ultrafiltra-tion (UF), nanofiltration (NF) and reverseosmosis (RO), are widely used in the dairyindustry, for the concentration of wholemilk or target components, depending onthe use of the obtained fractions [7]. Produc-tion at industrial scale as well as effluenttreatment is performed by membrane pro-cesses well identified as clean, sober andsafe in accordance with requirements of sus-tainable production. The main developmentlimitation of such processes at industrialscale is due to the scientific and technolog-ical bottleneck of fouling governing fluxesand thus, productivity and cleaning aimingat restoring membrane performances andensuring the safety of produced fluids andproduction equipments [5]. For a long time,the unique objective was to reach the max-imum flux (i.e. the limiting flux), involvinga high transmembrane pressure (TMP). Thiscriterion has many disadvantages amongwhich is the buildup of a thick fouling layer,closely packed and thus cohesive and hardto remove during the cleaning step. More-over, in industrial production filtering pro-cess units, working at constant permeate

flux, fouling increase is balanced by anincrease in TMP to attain the target fluxlevel, entering in a vicious cycle leadingto shortened production times and importantcleaning difficulties. To avoid such prob-lems, an original theoretical concept wasproposed in the 1990s: filtration in criticalor sub-critical conditions [12]. The aimwas to propose nonempirical rules in orderto understand how to manage membraneprocesses for a long-term use, and thushow to ensure a sustainable productionusing such processes. For this purpose, aboundary value of flux, the so-called criticalflux, is proposed. Below this value, revers-ible deposit can be built up on the mem-brane surface and above this value anirreversible deposit appears that can onlybe removed by chemical cleaning. Conse-quently, a range of stable conditions for fil-tration could be suggested to the user forchecking the production. This concept wasearlier proposed on a hydrodynamics back-ground and mainly described in microfiltra-tion and UF of model solutions (latex,silicates, etc.) designed for the purpose [2,16, 32], and has also been shown successfulfor the filtration of milk products [13, 14,17, 18, 33–35]. Globally, these works

364 M. Rabiller-Baudry et al.

mainly concerned parametric studies ofhydrodynamics conditions of filtration,highlighting flux variations and sometimesthose of retentions [13, 14, 17, 33–35].Whatever the fluids, the impact of the phys-ico-chemical environment was rarely takeninto account and mainly through small vari-ations of pH and ionic strength [4, 11, 13,14, 17, 26, 33–35].

With the aim of a better mastering ofstandard milk filtration and new dairy prod-ucts and/or effluent treatments (acid, tran-sient and alkaline), this paper contributesto study wide pH variations from 2.7 to11.5 on the limiting and critical fluxes mea-sured with spiral membranes either in UF,NF or RO. The behaviour of the whole foul-ing is only considered through the destabil-isation of the fouling layer due to pHvariations. From a scientific point of view,the main question arising during this studyconcerns the identification of the origin offlux dependence vs. the pH of modifiedmilks. Several parameters are a priori con-sidered for a better understanding of theinterdependency of physico-chemical char-acteristics of filtered fluids and membraneprocesses. Our approach considers the roleof the physico-chemical characteristics ofcaseins (size and charge through the electro-phoretic mobility measurements) either inmicelle or in aggregate forms, as they areknown to be responsible for flux limitationat natural milk pH. The possible impact ofserum proteins is taken into account in thediscussion, using previous characterisationsof their charges [23, 24], knowing that mostof them are no more soluble and can beassociated to caseins after the ultra hightemperature (UHT) treatment.

The role of different forms of calciumsalts is also considered as they can enterin the composition and cohesion of the foul-ing layer through interactions with variousproteins. As speciation of salts is notdirectly possible in such complex media,we have used theoretical calculations froma new software dedicated to salts’ equilib-

rium in skim milk and developed in INRA(Rennes, France) [20] but only availablefor acidic conditions up to now. Thus, pre-cipitate, free and soluble-linked species ofcalcium and phosphate salts are calculated.Because the speciation in alkaline media isnot well known and no software is avail-able, we take experimental data from [1].The latter are less detailed and only distinc-tion of soluble and insoluble forms can betaken into account.

2. MATERIALS AND METHODS

2.1. Fluids

The skim milk (UHT, Lait de Montagne,Carrefour, Levallois, France) contained, onaverage, 31.5 g·L−1 proteins (out of which27 g·L−1 caseins) and 48 g·L−1 carbohy-drates (mainly lactose, MW = 342 g·mol−1)(according to the provider) and various salts(on average 8 g·L−1) according to [7]. Thenatural pH of skim milk was close to pH =6.6–6.7, as expected. Adjustments of pH inthe range 1.9–11.5 were carried out by theslow addition of HCl or NaOH, under mag-netic stirring and the pH remained stableduring the whole filtration time of nearly6 h. HCl (concentrated, for analysis, Acros,Geel, Belgium) as well as NaOH (pellets,Prolabo, Fontenay-sous-Bois, France, recta-pur 97%) were used for preparing the stocksolutions (1 or 2 mol·L−1) used. Deionisedwater used for solutions and membrane rins-ing was filtered (1 μm).

2.2. Filtration experimental set-upand membranes

Three different spiral wounded mem-branes as well as two different filtrationexperimental set-ups were used in thisstudy. Common procedures concerned:

� The use of a unique membrane ofeach type that was cleaned according

UF, NF and RO flux of pH-modified milk 365

to the appropriate procedure (see below)until recovery of the hydraulic resis-tance (Rm, calculated from the well-known Darcy’s relationship) after eachexperiment. The accuracy on measuredfluxes was ± 5%.

� The choice of a constant flow rate dur-ing the cross-flow filtration leading toa constant cross-flow velocity (v)depending on the membrane process(see below).

� The temperature was kept constant at25 ± 2 °C during filtration.

� All experiments were performed inbatch mode with a volume reductionratio (VRR) of 1 corresponding to atotal recycling of both retentate and per-meate in the feed tank.

� The increase in flux was graduallyachieved by increasing the TMP in theappropriate range for each filtration type(see below).

2.2.1. Spiral UF

The spiral UF membrane used in thisstudy was made of polyethersulphone(PES, HFK-131 from Koch, Lyon,France, 6.5 m2, molecular weight cut off(MWCO) = 5–10 kg·mol−1). The hydraulicresistance of the virgin membrane wasRm = 2.1 ± 0.3 × 1013 m−1.

The UF was performed on a pilotdesigned by TIA (Bollène, France) with25 L of fluid being processed at each filtra-tion. All the experiments were conducted ata constantflow rate of 10 m3·h−1 correspond-ing to an average cross-flow velocity (v)close to v = 0.3 m·s−1. TMP increased from1.2 to 5.2 bar with an accuracy of ± 0.1 bar.

2.2.2. Spiral NF

The spiral NF membrane used in thisstudy was made of polyamide (PA, Desal 5DL from Osmonics, Delft, The Netherlands,2.5 m2, MWCO = 150–300 g·mol−1, Rm =

6.5 ± 0.3 × 1013 m−1). The NF was per-formed on a second pilot designed by TIA(Bollène, France) with 10 L of fluid pro-cessed at each filtration. All the experimentswere conducted at a constant flow rate of750 L·h−1 corresponding to v close to0.1 m·s−1. TMP increased from 5.0 to 25.0bar with an accuracy of ± 0.5 bar.

2.2.3. Spiral RO

The spiral RO membrane was a compos-ite organic one (TFC HR from Koch, Lyon,France, 2.5 m2, NaCl retention of 99.5%,Rm = 27 ± 2 × 1013 m−1). The RO wasperformed on the same pilot as NF at790 L·h−1 (v close to 0.1 m·s−1). TMPincreased from 7.5 to 30.0 bar with an accu-racy of ± 0.5 bar.

2.2.4. Irreversible foulingand cleaning

Part of the overall fouling is generatedby a sub-layer strongly bounded to themembrane corresponding to the so-calledirreversible fouling. Hydraulic resistancedue to irreversible fouling established atlimiting flux (Rirrev. limiting) was determinedfrom water flux after the rinsing step follow-ing the “milk” filtration. Rirrev. limiting corre-sponds to the target value to removeduring the cleaning step. Efficient cleaningsteps depend on both membranes and mod-ified skim milks. For membranes used inUF, NF and RO, similar cleanings were per-formed (50 °C, 1 h, VRR = 1) according totwo different procedures depending on thepH of the modified milk. HNO3 (pH 1.6)and an alkaline formulated detergent (P3-Ultrasil 10 at 0.4 wt.% pH 12.0, containingchelatant(s) and surfactant(s) provided byEcolab, Issy-les-Moulineaux, France) wereused as single step or in cascade.

The procedures were first established forUF membrane and then applied to NF andRO membranes with similar efficiency. Thereversiblepart of foulingof spiralmembranes


in UF, NF and RO were first removed bywater rinsing whatever the pH, before thechemical cleaning steps began.

� For pH > 5.0, UF, NF and RO spiralmembranes were cleaned (50 °C, 1 h)with the alkaline formulated detergent.This procedure is efficient as more than90% of initial flux is recovered what-ever the membrane. This procedurewas first established for UF membraneand proved to be efficient for more than25 filtration cycles that alternated skimmilk UF at natural pH and cleaning.

� For acidic pH between 3.0 and 4.7, theprevious procedure was not efficientenough. So a cascade of acid and alka-line detergents was tested. Due to salts’presence in the irreversible foulinglayer, the use of the acid step for solu-bilisation of minerals is needed. In orderto avoid the precipitation of hydroxidein/on the membrane, we started withthe acid step (HNO3, pH = 1.6, 50 °C,1 h) prior to the alkaline, one describedbefore.

� For pH close to 2.8, UF membranes weredramatically fouled. One acidic and fivealkaline cleaning steps were needed torestore the membrane initial flux.

Thus, too acidic pH do not reveal to becompatible with sustainable requirements.Then, the following experiments were lim-ited to pH ≥ 3.7. Finally, taking into accountthe cleaning problematic leads to a limitationin the pH range of modified skim milk thatcan be acceptable for UF, NF and RO withorganic membranes. A sustainable processonly corresponds to the filtration of slightlyacidic to neutral and alkaline pH, whateverthe obtained retentions.

2.2.5. Limiting and critical fluxdetermination

To determine both fluxes, UF, NF orRO of a given fluid was performed by

increasing step by step the TMP, with val-ues depending on the membrane. Eachincrease of TMP was realised after reachingthe plateau value of flux (J) for the previousTMP. Filtration for a given TMP was per-formed for about 30–100 min dependingon the TMP. Typical plot of J at plateauvalue vs. TMP is shown in Figure 1.A linear relationship is observed below thecritical point (Jcritical and TMPcritical) corre-sponding to critical conditions of filtration.Beyond the limiting pressure (TMPlimiting)the flux does not change anymore and aconstant maximal value corresponding tothe limiting flux (Jlimiting) is reached. Thisprocedure can be used only if the occur-rence of an irreversible deposit at limitingflux had been previously checked. Thiscan be done by a stepwise decrease inTMP from the limiting flux to a value lowerthan the critical flux. Occurrence of the crit-ical flux is checked if a hysteresis is evi-denced on the J vs. TMP graph due to anevolution of the irreversible fouling uponwhich the reversible fouling is built.According to Wu et al., classification[32] all experimental curves led to define

0

5

10

15

20

25

30

35

0 1 2 3 4 5 6TMP (bar)

Jp (L

. h –1

. m–2

) limiting values

critical values

Figure 1. Experimental determination of crit-ical and limiting fluxes in UF: a linearrelationship was observed below the criticalpoint (Jcritical and TMPcritical). Beyond thelimiting pressure (TMPlimiting) the constant fluxcorresponds to the limiting flux (Jlimiting) withTMP, transmembrane pressure and J, flux.


“critical flux of weak form”, meaning thatthe permeability (slope) in sub-critical con-ditions is always lower than the permeabil-ity to pure water, whatever the membraneprocess and pH. Limiting fluxes were deter-mined with a precision close to 5% but crit-ical fluxes were obtained by extrapolation ofthe J vs. TMP curve with a poorer precisionclose to 10%.

2.3. Physico-chemicalcharacterisation of caseinand aggregates

2.3.1. Size distribution of caseinmicelles and aggregates

The size distribution of particles andaggregates was first determined by laserlight scattering (LLS), using a Mastersizer2000 granulometer (Malvern Instruments,Worcestershire, UK) at two differentwavelengths (He/Ne laser: 633 nm andelectroluminescent diode: 466 nm) at roomtemperature. Five hundred microlitres ofunfiltered modified skim milks wereinjected in the cell containing about100 mL of Milli-Q water at 1500 rpm stir-ring at 25 °C. Refractive indexes were setat 1.34 and 1.59 for solution and particles,respectively.

2.3.2. Size and electrophoreticmobility of casein micellesand aggregates

Hydrodynamic particle diameters weremeasured by dynamic light scattering(DLS) on a Zetasizer 3000 HS (MalvernInstruments, Worcestershire, UK). Measure-ments were carried out at 90° scatteringangle, 633 nm wavelength and 25 °C tem-perature. The refractive index solution wasset at 1.341 and the viscosity of dispersivesolution at 1.02 × 10−3 Pa·s. Diluted milkswere 1.2 μm filtered prior to analysis, allow-ing characterisation of the smaller particles.An average size (Zaverage) from electropho-

retic measurement (electrophoretic dynamiclight scattering (EDLS), simultaneouslymeasured as size from DLS) was also deter-mined with an accuracy of ± 5 nm.

Electrophoretic mobility (μ, velocity toelectric field ratio of casein micelles oraggregates) was also measured with aZetasizer 3000 HS (Malvern Instruments,Worcestershire, UK) after 1.2 μm filtration.To discuss the change in casein charge withthe pH, the zeta potential is commonly used[6, 21]. It is inferred from the measurementof μ that is a proportional value. However,recent theoretical developments ([30] andrelated papers) pointed out that a single zetapotential value is not so clearly defined for aporous particle, even if the charge exists. Ascasein micelles can be considered as porousparticles, we select the experimental μ valuerather than a questionable inferred zetapotential. μ was measured with an accuracyof 0.4 × 10−8 m2·V−1·s−1 close to the iso-electric point and with a better accuracy of0.2 × 10−8 m2·V−1·s−1 for higher values.

For both size and μ measurements, dilu-tion of modified and unmodified skim milkswas achieved with the respective aqueousphase (ultrafiltrate at the same pH) to avoidequilibrium displacement according to pre-viously validated protocol ([26] and relatedpapers).

2.4. Chemical oxygen demandanalysis

The chemical oxygen demand (COD)was measured with a 160 DCO kit(Macherey Nagel, Molsheim, France) withan accuracy better than 5% after appropriatedilution by water if necessary. Finally, accu-racy on retentions, defined according to

Retention of COD ¼ 1� permeate COD

retentate CODð1Þ

was lower than 10%.


2.5. Calculation for salt speciationwith the pH variations

Only modelling is possible to deal withthe various forms of salts that cannot bemea-sured. As a first attempt of correlationbetween fluxes and milk composition, webase our discussion on the available informa-tion on salt speciationover thewide pH rangeof this study, using skim milk composition.Several complementary (very heavy) analy-ses might be performed in order to evidencethe exact composition of UHTmilk salt frac-tions that could be slightly different from thatof the un-treated skim milk.

2.5.1. Theoretical calculations in theacidic pH range

Theoretical calculations were madeusing a new software based on dissociationconstants proposed by Mekmene et al. [20]allowing the calculation of mineral distribu-tion, in unmodified and modified skimmilks, in various forms in the aqueousphase as minerals or linked to casein. Con-centrations of ionic species, namely free,complexed (corresponding to soluble formsof calcium associated to phosphate, citrateand chloride) or precipitated forms (corre-sponding to nanoclusters of calcium phos-phate included in casein micelles and notto free calcium phosphate precipitate in sus-pension in the aqueous phase), can be calcu-lated according to the pH adjusted by HCladdition. In this study we mainly focus oncalcium and phosphate soluble salts in theaqueous phase of milk. For calculation,selected concentrations are chosen as thoseof the average composition of skim milk:Ca, Mg, Na, K, inorganic phosphate andCl as 30, 5, 22, 38, 21 and 27, respectively(in mmol·L−1). Caseins are taken at 25 g·L−1

and their binding with Ca is based on phos-phoserine and carboxylate sites of proteins.Figure 2 shows the predicted concen-tration for calcium in the pH range 2.0–6.7.

The inorganic phosphate is fully soluble at apH lower than 5.5 and then precipitateswith calcium in the nanoclusters contributingto the casein micelles’ or aggregates’ cohe-sion. Three zones can be distinguishedfor calcium forms that are not linked tocaseins:

(i) 2.0 < pH < 3.0Two main forms exist, free (Ca2+) andcomplex ones, in a roughly constantamount with a free to complex formsratio close to 2.

(ii) 3.5 < pH < 5.5Ca2+ amount decreases slightly withthe pH and conversely complex formsincrease until a free to complex formsratio is close to 1.

(iii) 5.5 < pH < 6.7Ca2+ amount decreases stronglywith the pH and complex formsdecreased slightly as calcium phos-phate precipitate increases in the na-noclusters linked to caseins.

2.5.2. In the alkaline pH range

At alkaline pH the release of calcium fromcaseins was observed during the destructur-ation ofmicelles connected to the progressivemilk transparency increase. In the absence ofproteins Ca2+ would immediately enter invarious possible combinations with phos-phate, citrate and hydroxyl ions, possiblyleading either to soluble or precipitated com-ponents. Among the latter are calcium phos-phate and calcium hydroxide precipitates,calcium citrate and di-citrate complexes.

The presence of proteins in the aqueousphase strongly modifies this equilibrium,andnoprecipitate canbe evidenced asprovedby the milk transparency observation. Amodel of the complex evolution of salts andassociation with caseins was recently pro-posed [1]. To summarise, one can considerin a first attempt that no more Ca2+ exists inthe modified milks at alkaline pH and that


whole calcium is engaged in combina-tions either in complex soluble forms orwith casein aggregates. The pH close to 9appeared as a key point in this system, corre-sponding more or less to the end of thecalcium leakage in the aqueous phase.Adapting experimental results of [1] for solu-ble and non-soluble forms of calcium, empir-ical correlations were established (Fig. 3) forfurther calculation of soluble forms (withoutdescription of various complexed forms) andassociatedwith caseins for calciumandphos-phate.At this time, itwas not possible to enterin amore detailed speciation of soluble formscorresponding mainly to citrate and phos-phate complexes.

3. RESULTS

3.1. Physico-chemicalcharacterisation of caseinsand casein aggregates

Figure 4 shows the size distribution ofcaseins and casein aggregates measured by

LLS on modified milk without prior filtra-tion of the sample. At natural pH of milk,particles of about 2.2 μm size probably cor-respond to residual fat particles in the UHTskim milk.

Size is then determined from DLS on1.2 μm filtered milks. At natural pH of milk,casein micelles exhibit a size distributionbetween 20 and 700 nm (data not shown),in good agreement with the literature data,whatever the calculation mode or the tech-nique itself [4, 8, 15, 19, 29]. Zaverage (fromEDLS) is found close to 210 nm, in goodagreement with part of the literature data[4, 15, 19] even if sometimes a value closeto 100 nm has been reported ([8] and relatedpapers) in good agreement with our ownresults from the mean size calculated onthe particle number (calculation mode high-lighting small particles); this is not contra-dictory with having an average size ofdifferent value. Decreasing the pH from6.7 to 5.1, slight changes are evidenced inthe size distribution but Zaverage remainsclose to 210 nm. Increasing the pH from6.7 to 11.0 induces the disappearance of

0

5

10

15

20

25

2pH

Con

cent

ratio

n (1

0–3 m

ol. L–1

)

Ca2+ Ca-linked to caseins Ca "with Phosphate in nanoclusters" Ca "complexes" P "soluble"

2.5 3 3.5 4 4.5 5 5.5 6 6.5 7

Figure 2. Theoretical calculations on Ca and P of milk according to the software in the pH range2.0–6.7.


native casein micelles and the appearance ofaggregates, mainly of micrometre magni-tude. Zaverage of aggregates lower than0.8 μm·s decreased first up to pH close to9 and then increased (Fig. 5).

Figure 6 shows the variation in the elec-trophoretic mobility of particles made ofcaseins (and lower than 1.2 μm) vs. pH.As expected, μ tends towards zero atpH 4.6, which corresponds to the casein iso-electric pH. μ of smaller particles is notmodified by pH increase up to 11.5, evenif casein micelles are denatured andreplaced by “casein aggregates”.

As a conclusion for physico-chemicalcharacterisations, it can be drawn that sizeand electrophoretic mobility are two inde-pendent and complementary characteristics

of particles made from caseins. As caseinaggregates may vary quite strongly withpH, further discussion can be divided intothree parts:

(i) High acidic pH range (3.7 < pH < 4.7)The electrophoretic mobility is close tozero as caseins are at their isoelectricpH. They formed micrometric aggre-gates that precipitate. Such unchargedcaseins are able to build up a depositat the membrane wall during furtherUF, NF or RO.

(ii) Low acidic pH range (5.1 < pH < 6.7)The electrophoretic mobility increaseslinearly with pH. The average diameterof caseins, whatever their forms,remains close to Zaverage = 210 nm.

y = 0.6404 x2 – 12.997 x + 66.191R2 = 0.9993

y = –1.5174 x2 + 32.752 x – 151.5R2 = 0.95

0

5

10

15

20

25

30

6

pH

Con

cent

ratio

n C

a (1

0–3 m

ol. L

–1)

P soluble P linked to caseins

y = –2.4831 x + 27.343R2 = 0.99

y = –1.1082 x2 + 24.575 x – 117.72R2 = 0.98

0

5

10

15

20

25

30

pH

Con

cent

ratio

n P

(10–3

mol

. L–1

)

7 8 9 10 11 12

6 7 8 9 10 11 12

a

b

Figure 3. Calcium (a) and inorganic phosphate (b) forms at alkaline pH adapted from [1] – ( )soluble form and ( ) linked to caseins.


Accordingly, repulsive electrostaticinteractions involving casein aggregatesand charged membrane may occur. Allthe flux and retention variationsobserved during filtrations are then cor-related with variations of the μ as theaverage size is roughly constant.

(iii) Neutral to alkaline pH range(6.7 < pH < 11.0)The electrophoretic mobility is constantand equal to −1.9 × 10−8 m2·V−1·s−1.

A bimodal size distribution is evidenced,one centred around 2.4 μmand the otherone around 80–250 nm. By increasingthe pH, the smaller particles tend to dis-appear and a plateau minimum value isobserved at pH close to 9. The smallestparticles, native casein micelles (Zaverage= 210 nm at natural milk pH), are pro-gressively replaced by smaller aggre-gates up to Zaverage = 90 nm at pH 9.Then, the average size again increases

0

1

2

3

4

5

6

7

8

9

0.01

Size (µm)

Volu

me

(%)

pH 6.7 pH 11.0

0

1

2

3

4

5

6

7

8

9

Size (µm)

Volu

me

(%)

pH 3.7

pH 4.7

pH 1.9

0.1 1 10 100

0.01 0.1 1 10 100 1000

Figure 4. Size distribution by LLS of particles of natural and modified skim milks in alkaline andin acidic pH range (for both, no filtration before measurements).


up to Zaverage = 150 nm with increasingpH, with a concomitant calcium leakagefromcaseins up to pH9. Secondly aggre-gation of smaller aggregates occurs,which may be due to the formation ofnew calcium phosphate bonds with

caseins. As the electrophoretic mobilityis constant and highly negative, electro-static interactions involving caseins areexpected during filtration and may playa significant role in both flux andretention of charged solutes, whatever

0

50

100

150

200

250

pH

Size

(nm

)

6 7 8 9 10 11 12

Figure 5. Size of casein aggregates (Zaverage) vs. pH of modified skim milk (sample after 1.2 μmfiltration).

–2.50

–2.00

–1.50

–1.00

–0.50

0.00

2pHC

asei

n el

ectr

opho

retic

mob

ility

(10–8

m2 . V

–1 . s–1

)

3 4 5 6 7 8 9 10 11 12

Figure 6. Electrophoretic mobility (μ) of caseins whatever their forms vs. pH (μ ± 0.2 ×10−8 m2·V−1·s−1).


the pH. So, all theflux and retention vari-ations observed during UF, NF or RO atalkaline pH are then correlated to thecasein sizes’ variations.

3.2. Filtration of skim milks in the3.7–11.5 pH range

3.2.1. Limiting and critical fluxvariations with pH

Figure 7 shows both critical and limitingfluxes vs. pH in UF, NF and RO. The lim-iting and critical fluxes behave similarly,whatever the considered membrane process,meaning that the behaviour is mainly mon-itored by the filtered fluid in comparisonwith the membrane type.

3.2.2. Fouling

As it is the case for the fluxes, complexvariations of the resistance due to the wholefouling at limitingflux (Rf) are observedwithpH (data not shown). Whatever the pH, Rfis always lower than 20 × 1013 m−1 in UF,whereas it ranges from 60 × 1013 to200 × 1013 m−1 in both NF and RO.

Resistance due to the irreversible part(Rirrev. limiting only at limiting flux) is alwayslower than 6 × 1013 m−1 in UF and lowerthan 13 × 1013 m−1 in both NF and RO.Generally, the irreversible fouling in NF isintermediate between those of UF and RO(Fig. 8).Rirrev. limiting varies with pH in a non-predictable linear way for both UF and RO.Further investigations are needed to under-stand the origin of this linear correlation.The increase inRirrev. limiting for pH lower than5may be related to the occurrence ofmineralirreversible fouling at very low pH (seebelow).

3.2.3. Cleaning needs

Another approach to describe the foulingis to consider the needs for cleaning. For all

pH and membranes, an alkaline cleaningstep is needed, meaning that an organic irre-versible fouling exists in UF, NF and RO.

The membrane resistance is recoveredafter the first alkaline cleaning step for allfiltrations performed with milk whosepH > 5, meaning that mainly organic com-ponents are in the irreversible fouling in/onthe membrane.

Our previous experiments on UF mem-brane cleaning fouled by skim milk at natu-ral pH showed that no acid step is neededfor a long-term use (more than 30 consecu-tive cycles of skim milk UF followed by analkaline cleaning and already partial valida-tion at industrial scale) [9] that is fullycoherent with the analysis of the irreversiblefouling, evidencing only proteins on themembrane in this particular case [3, 22,27]. Further analysis of irreversible foulingof UF membranes fouled by pH modifiedskim milk in the range 5.0–6.7 would bedone to confirm whether some smallamount of inorganic components enter inthe irreversible fouling or not. Whateverthe conclusion, it is already shown that thealkaline cleaning step is efficient enoughand that inorganic compounds could bepresent only in small quantities.

For UF, NF and RO filtrations performedwith acid-modified milks, with pH lowerthan 5.0, the alkaline step is not sufficientand an acid one is needed to restore themembrane flux. This need clearly showsthat minerals are in the irreversible foulingfor such pH (the role of origin of mineralswill be discussed in the following section).

3.2.4. Retention of COD

Retention of COD is measured at bothlimiting and critical fluxes for UF, NF andRO experiments. Table I shows that reten-tion is roughly independent of the flux.For effluents’ treatment purpose, typicalCOD retentions are in good agreement withretentions observed for dairy effluents atnatural pH [31]. Consequently, for effluent


treatment, whatever the membrane process,it can be drawn that the productivity ismainly due to the flux mastering as theretentions are constant.

Considering that pH variation of milkinduces ionic strength variation that mightbegreater than73 × 10−3 mol·L−1 (accordingto the software calculations the ionic strength

UF

0

5

10

15

20

25

30

35

40

pH

J (L

. h–1

. m–2

)

J limitingJ critical

NF

0

2

4

6

8

10

12

14

16

pH

J (L

. h–1

. m–2

)


RO

0

2

4

6

8

10

12

14

16

pH

J (L

. h–1

. m–2

)


2 3 4 5 6 7 8 9 10 11 12

2 3 4 5 6 7 8 9 10 11 12

2 3 4 5 6 7 8 9 10 11 12

Figure 7. Behaviour of critical and limiting fluxes with pH in UF, NF and RO of modified skimmilk.


will be in the 151–73 × 10−3 mol·L−1 rangefor pH from2.0 to 6.7, respectively), the ques-tion arises “why the ionic strength modifica-tion does not affect the membraneperformances?”. A partial answer can begiven looking at the components responsi-ble for the COD, namely mainly proteinsand lactose. Proteins are charged solutesfor which an increase in the ionic strengthmight provoke a decrease in retention dueto an increasing screening of charges.Experimentally, the observed retention is100% in NF and RO as expected andgreater than 90% in UF, whatever the pH.Due to the low cut-off of the UF membraneone can consider that retention mechanismis thus mainly due to size exclusion andthat only a minor part of the retention canbe modulated by the ionic strength.

The main part of COD in permeate isdue to lactose, an uncharged solute, theretention of which is thus independent ofthe ionic strength. This is confirmed byquantitative analyses not shown here.

4. DISCUSSION

Here is a first attempt to propose possiblephysico-chemical origins of flux variations;their is a global approach common to UF,NF and RO in a wide pH range, since asalready mentioned above, the filtered fluidcharacteristics are most reliable parametersto be taken into account.

4.1. Possible components of thewhole fouling and their possiblelocation

As a general approach, one can considerfour types of components that could a prioribe involved in the membrane fouling: case-ins, serum proteins, lactose and minerals.According to the ability of components tocross the membrane towards the permeate,one can distinguish the ability to foul themembrane inside the pore or at the milk/membrane interface or in the accumulatedlayer outside of the membrane.

y = –0.57x + 6.68R2 = 0.95

y = –1.35 x + 17.42R2 = 0.95

0

2

4

6

8

10

12

14

16

pH

Res

ista

nce

(1013

m–1

)

2 3 4 5 6 7 8 9 10 11 12

Figure 8. Irreversible hydraulic resistance at limiting flux in UF ( ), NF (▲) and RO ( ) vs. the pHof modified skim milk.

Table I. COD retentions at limiting and criticalfluxes during UF, NF and RO.

Retention(%)

pHrange

At limitingflux

At criticalflux

UF 3.7–6.0 56 ± 6 50 ± 4UF 6.5–11.5 54 ± 4 40 ± 5NF 3.7–11.5 99.2 ± 0.8 99.2 ± 0.7RO 3.7–11.5 > 99.9 > 99.9


4.1.1. Small components as lactoseand minerals

Lactose as well as minerals are able tocross the membranes, whatever their type,consequently, they would be able to partici-pate in the internal as well as in the externalfouling. As lactose is not a pH-dependentcomponent, its role is not discussed here.As it is easy to solubilise in water, it wasalready proved that lactose does not enterin the composition of the irreversible foul-ing at natural pH of milk in UF [6, 22,27]. For the mineral fraction of milk, thepH-dependent components mostly corre-spond to calcium and phosphate salts, therole of which will be discussed in the fol-lowing section.

4.1.2. Caseins

Whatever the membrane process, case-ins, either in micelles or in aggregatedforms, cannot cross the membranes. Conse-quently, they can only contribute to theexternal fouling, meaning that they can bepresent in the accumulated layer or at theinterface. Accordingly, their behaviour willbe discussed according to a commonapproach for UF, NF and RO.

4.1.3. Serum proteins

In NF and RO, proteins cannot cross themembrane and the corresponding foulingcan only be on the membrane surface atthe interface with milk or in the accumu-lated layer. UF is a different case, since atnatural milk pH, only α-lactalbumin canpermeate, as evidenced by reverse-phaseHPLC (RP-HPLC) (data not shown). Theirreversible fouling was previously studiedby streaming potential measurements, bothalong the membrane surface and throughthe pores [25]. It was evidenced thatthe membrane isoelectric point (iep) ofthe surface moved towards the iep ofβ-lactoglobulin, whereas the membrane iep

inside the pore is shifted towards that ofthe α-lactalbumin. Thus, one may think thatinternal fouling due to proteins correspondsto α-lactalbumin, whereas external foulingat the membrane interface is mainly madeof β-lactoglobulin during UF of skim milkat natural pH.

Protein retention during UF is roughlyconstant and > 98% when decreasing thepH of milk from 6.7 to 3.7, as evidencedby whole protein quantification performedby FTIR-ATR (data not shown). Therefore,only part of α-lactalbumin probably crossthe membrane, whatever the pH and mayconstitute the internal fouling.

At alkaline pH in the range 8.0–11.0, pro-tein retention during UF is roughly constantand close to 90%. The increase of proteinsin the permeate is incompatible with only fullα-lactalbumin transmission, meaning thatother small denatured and soluble proteinsare able to cross themembrane. Identificationof these proteins has not been performed yetand numerous and delicate quantitative anal-yses would be necessary [28] to be affirma-tive on this particular point. Consequently,the internal fouling can be modified, regard-less of the natural pH of milk.

Finally, whatever the pH, serum proteinfouling can form a priori on the UF mem-brane surface, at the interface as well as inthe pores but the composition would bemodified according to the pH range: anacidic to neutral one and an alkaline one.

4.2. Possible impact of proteinson fluxes

As explained before, variations of case-ins properties can only modify the accumu-lated layer.

4.2.1. Role of charge in the 5.1–6.7pH range

In the 5.1–6.7 pH range, casein averagesize (Zaverage) is constant, whereas the


charge significantly varies as evidenced byμ variations. Limiting and critical fluxesare plotted against those of casein aggre-gates in this pH range (Fig. 9).

In UF, both fluxes vary linearly withcasein electrophoretic mobility. The fluxesare high when the absolute value of μ ishigh, meaning that repulsive electrostaticinteractions, involving caseins (whatevertheir forms), may be involved in the fluxvariations. Of course, the free casein aggre-gates, in suspension in the aqueous phase,do not correspond directly to the foulingat the membrane wall, but they can governthe cohesion of a more or less compact layerthat regulates fluxes. For comparison, limit-ing fluxes of a UF zirconia membrane (witha MWCO of 10 kg·mol−1, M5 Carbosep,Orelis, Miribel, France) adapted from [26]are reported in Figure 9. Looking at theJlimiting vs. μ graph for both UF organicand inorganic membranes, it appears thatthe slope is independent of the membranematerial with similar cut-off or pore size.

Looking at the plot of both limiting andcritical fluxes vs. casein electrophoreticmobility for NF and RO, it appears thatthe slope is roughly zero, highlighting thatthe fluxes’ dependency on electrostaticinteractions is much more important for amembrane with larger pores (UF) than fora membrane with smaller pores and densemembrane (NF and RO).

A question that arises is then “even if theaccumulated layer provokes a flux limita-tion, is the difference in flux variation withpH really due to the accumulated layer orto the internal fouling occurring in UF andnot in NF and RO?”.

The linear decrease of flux can beexplained taking into account the internalfouling of the membrane by salts: one canassume that when the pH increases from5.0 to 6.7, the free calcium decreases inmilk, leading to a decrease in the internalfouling due to calcium. This is in goodagreement with the occurrence of a moreor less important mineral irreversible

fouling due to calcium with pH (see below).Such a phenomenon can occur in UF, NFand RO since small ions are able to enterthe three membrane types. According toan important difference in the slopes ofJ vs. pH, it is not possible to attest that saltspresent in the internal fouling solely contrib-ute to flux variations.

An important difference between thethree membranes can be related to α-lactal-bumin internal fouling (perhaps not nativehere, see below), which can enter in theUF membrane but cannot do so in NF andRO membranes.

Thus, caseins, whatever their form, canbe only partly responsible for fluxvariations.

4.2.2. Role of casein size in the6.7–11.0 pH range

For pH > 6.7, the charge of caseinaggregates is high and constant, meaningthat electrostatic interactions can beinvolved in the fouling layer cohesion.Roughly the size of the particles increasesprogressively in this pH range up to2.2 μm, but the average behaviour of thesmaller particles is quite complicated(Fig. 5). Looking at the size evolution ofthe smaller particles in the filtration media(Fig. 5) and the flux behaviour in this pHrange (Fig. 7), it is clear that flux increaseis related to size increase in the smaller par-ticles, even if they are the minor part andwhatever the filtration process. So, the cohe-sion of the fouling layer is partly controlledby the size of particles entering in its com-position: bigger particles favour higher andless cohesive fouling deposit, and thushigher fluxes contrary to smaller particles.Once again, slopes of J vs. pH in this rangeare low in NF and RO compared to UF,meaning that only part of the accumulatedlayer can be responsible for flux variationand that other components of milk playa role.


4.2.3. Possible impact of serumproteins on fluxes in the 4.6–6.7pH range

Electrophoretic mobility of nativeα-lactalbumin and β-lactoglobulin was the-oretically determined from their primary

sequence and then measured at pH 7 by cap-illary electrophoresis in previous studies [23,24]. These studies concluded that theoreticalvalues are not correct in the caseof serumpro-tein behaviour at pH 7 in borate as well asin phosphate buffer, including ionic strengtheffects in the 10–200 × 10−3 mol·L−1 range.

y = −0.8 x + 7.6R2 = 0.98

y = −8.0 x + 9.3R2 = 0.999

y = −8.4 x + 18.2R2 = 0.99

y = −2.3 x + 6.5R2 = 0.90

0

5

10

15

20

25

30

35

40

–2.5 –2.0 –1.5 –1.0 –0.5 0.0

–2.5 –2.0 –1.5 –1.0 –0.5 0.0

Casein electrophoretic mobility (10–8 m2.V–1.s–1)

J (L

. h–1

. m–2

)Limiting flux

PES-HFK131- UF

Zirconia- M5 - UF

PA- Desal 5DL - NF

PA- TFCHR- RO

y = −1.0 x + 4.7R2 = 0.98

y = −0.6 + 7.1R2 = 0.96

y = −5.2 x + 7.9R2 = 0.96

0

2

4

6

8

10

12

14

16

18

20

Casein electrophoretic mobility (10–8 m2.V –1.s–1)

J (L

. h–1

. m–2

)

Critical flux

PES-HFK131- UF

PA- Desal 5DL - NF

PA- TFCHR- RO

Figure 9. Correlation between limiting flux and critical flux and casein electrophoretic mobility fordifferent membrane types in the pH range 5.1–6.7. The results for zirconia UF membrane areadapted from [26].


Nevertheless, experimental mobility of−1.1 × 10−8 m2·V−1·s−1 for α-lactal-bumin and −1.6 × 10−8 m2·V−1·s−1 and−1.5 × 10−8 m2·V−1·s−1 was measured forthe A and B variants of β-lactoglobulin,respectively. Knowing the isoelectric pointof each protein, 4.2–4.5 and 5.1 for α-lactal-bumin and β-lactoglobulin, respectively, itcan be concluded that the electrophoreticmobility should be zero at such pH.Thus when decreasing the pH from6.7 to 4.6, the electrophoretic mobilityof α-lactalbumin would increase fromabout −1.1 × 10−8 m2·V−1·s−1 to 0. Simi-larly, μ would increase from about−1.5 × 10−8 m2·V−1·s−1 to 0 for β-lactoglob-ulin in the same pH range. These μ variationsas well as the order of magnitude of the elec-trophoretic mobility of serum proteins areclose to those determined here for caseins.Thus, when filtering un-treated skim milk itwill be difficult to infer whether flux differ-ences with pH are due to caseins and caseinaggregates and/or serum proteins. Moreover,as serum proteins are partly denatured inUHT skim milk, discussion about theirrespective impact on fluxes is of courselimited. Nevertheless, it can be assumedthat effect of charges of serum protein onflux behaviour might be superimposed tothat of caseins, if any. Thus, the role ofα-lactalbumin as an internal foulant of theUF membrane proposed in the previous sec-tion would be in good agreement with suchconsiderations.

4.3. Possible impact of mineralson fluxes

Minerals can a priori participate in thewhole fouling, whatever its location forUF, NF and RO membranes. Empirical cor-relations between the limiting flux observedin UF and different forms of both calciumand phosphate are suggested and shown inthe following section. As flux variationsare significantly different mainly for

pH > 9, only conclusions at very alkalinepH can be extrapolated as well in NF andRO.

4.3.1. Neutral to acidic pH

For pH lower than 7.0, empirical linearcorrelations can be found between limitingflux in UF and soluble calcium concentra-tion (Fig. 10). The higher the soluble cal-cium, the lower the limiting flux. Similartrends are evidenced with phosphate(Fig. 10). Precipitates of calcium phosphatein the form of nanoclusters embedded incasein aggregates do not increase the overallfouling contrary to soluble calcium thatfavours in-situ fouling of membrane. Thus,it can be drawn that soluble calcium actsas a promoter of inorganic fouling in/onthe membrane during filtration. In order toclarify the role of various forms of the solu-ble calcium fraction on the membrane foul-ing, empirical linear correlations are foundmainly with Ca2+ (Fig. 11) and to a lessextent for complexed forms.

4.3.2. Alkaline pH

Roughly linear correlations between UFlimiting flux and calcium forms either solu-ble or associated to caseins, are obtained indifferent regions of the alkaline pH range(Fig. 12). This suggests a role of soluble cal-cium ions in the overall fouling, thenaccording to the pH, a role of casein aggre-gates containing calcium and phosphate inassociation. For pH lower than 8.2, wheresoluble calcium (only complexed forms) ispredominant, the higher the soluble cal-cium, the lower the limiting flux, highlight-ing that soluble complexes do not limitfluxes. For pH > 8.2, where calcium andphosphate, associated to casein aggregates,are predominant, the higher fluxes arerelated to the higher content of associatedions, with probably a similar role as the na-noclusters in casein aggregates at acidic pH.A more detailed explanation based on


calcium complex forms cannot be proposedas the respective amounts of these com-plexes are not available at that time.

4.3.3. Conclusion on whole foulingand minerals

As a conclusion, all these empirical cor-relations suggest that calcium salts play arole in the overall fouling, both in solubleforms and those associated to casein aggre-gates, in addition to the impact of the caseinaggregates themselves and to that ofα-lactalbumin in UF. Similar conclusionscan be drawn for NF and RO, except for

α-lactalbumin. Further investigations areneeded for a better understanding of thewhole phenomenon and the origin of varia-tion of cohesion of casein aggregates mustbe studied looking at the role of associatedcalcium and phosphate either in nanoclus-ters at acidic pH or in associated forms thathave to be quantified at alkaline pH.

4.3.4. Irreversible fouling at limitingflux and consequenceon cleaning

For pH lower than 5.0 (Figs. 2 and 7),when considering the requirement of an

0

5

10

15

20

25

30

35

40

0 5 10 15 20 25 30

Concentration Ca (10–3 mol.L–1)

( L.

h–1. m

–2 )

Ca "in nanoclusters"

Ca "soluble"

10

15

20

25

30

35

40

0 5 10 15 20 25

(L. h

–1. m

–2)

P "in nanoclusters"

P "soluble"

Concentration P (10–3 mol.L–1)

J UF

limiti

ngJ U

F lim

iting

Figure 10. Empirical correlation between theoretical Ca (a) and P (b) concentrations in variousforms and limiting flux in UF for pH lower than 7.0.


acid step during membrane cleaning,it indirectly highlights the presence ofan irreversible fouling due to inorganiccomponents. When pH increases from 3.7to 5.0, the Ca2+ amount in modifiedmilk strongly decreases and the acidcleaning step is always needed. So, it canbe deduced that Ca2+ amount is related to

severe inorganic irreversible fouling in/onorganic membrane, whatever the technique(UF, NF and RO) and the organic activelayer of either PES (UF) or various PA(NF and RO).

Considering the cleaning needs for filtra-tion of unmodified and alkaline modifiedmilks, we have observed that no acid

15

20

25

30

35

40

0 5 10 15 20 25 30


J UF

limiti

ng (

L.h–1

. m–2

)

Ca "complexed" Ca2+ "free"

Ca "soluble"

Figure 11. Empirical correlations between limiting flux in UF and theoretical calcium concen-tration in various soluble forms for pH lower than 7.0.

0

5

10

15

20

25

30

35

40

0 5 10 15 20 25 30


(L. h

–1. m

–2)

Ca solublepH 6.7–9.2

Ca linked to caseinspH 9.2–11.5 J U

F lim

iting

Figure 12. Empirical correlation between soluble calcium and calcium associate phosphate incasein aggregates with the limiting flux in UF in the pH range 6.7–11.5.


cleaning step is needed. For neutral toalkaline pH, Ca2+ is always lower than2 × 10−3 mol·L−1 and soluble calcium ismainly engaged in soluble complexes withphosphate and citrate. Then, Ca2+ appearsas the main promoter of the inorganicirreversible fouling in/on the organic mem-branes, whereas calcium phosphate engagedwith casein in micelles or aggregates or insoluble complexes has no severe conse-quence on the cleaning step as calcium isnot disposable.

This phenomenon suggests that if thesolubility limit of calcium salts is notreached in the bulk solution, meaning thatCa2+ exists in solution, it is not the samecase at the membrane surface and probablyin the pores. It can be suggested that con-centration polarisation occurs in close vicin-ity of the membrane wall, playing the roleof a destabilising surface provoking themineral nucleation, and thus their precipita-tion, regardless of the salt accumulation. Anunanswered question concerns also thebehaviour of proteins (and probably mainlyβ-lactoglobulin) at the milk/membrane inter-face as the salting out effect can also occur,favouring gel formation. Nevertheless, wethought that this second eventuality wouldlead to a reversible fouling as gel formedbefore the membrane pore entrance duringUF of skim milk has already been shownto be easily removed by water rinsing[10]. Further studies are needed to definitelyconclude on this particular point.

Complementary in-depth study of thePES-UF membrane cleaning when fouledby unmodified skim milk has confirmed thatminerals are not discernable in the irrevers-ible part of the fouling layer, and thus wehave suggested that acid cleaning step isnot needed for such filtrations [10, 22].Then, a single step of alkaline cleaningcan be sufficient for filtrations performedwith modified milks in the 6.7–11.5 pH(an additional disinfection step must beadded for the safety of equipment of foodindustry). This proposal (without the disin-

fection step) is successfully tested, and thusvalidated during this study in UF, NF andRO. Moreover, partial validation at indus-trial scale is already done in PES-UF fornatural pH and a systematic validation isin progress at industrial scale.

5. CONCLUSION

The aim of this study was to identify thepossible physico-chemical origins of fluxvariations, namely the limiting flux (maxi-mum) and the critical flux (sustainable), inUF, NF and RO of skim milks modifiedby the addition of HCl or NaOH. Both thelimiting and the critical fluxes are pHdependent in a nonpredictable way withclose similar trends in UF, NF and RO,highlighting the leading role of the fluidbehaviour. From physico-chemical charac-teristics of caseins (size and electrophoreticmobility), we have shown that variationsof electrostatic interactions and aggregates’size partly control fluxes. Charges of thetwo main serum proteins, α-lactalbuminand β-lactoglobulin, vary similarly as caseincharge in the 4.6–6.7 pH range. Then,impact of serum proteins, if any, wouldbe superimposed to that of caseins in anondistinguishable way. Some differencesobserved between UF, NF and RO may bedue to internal fouling by α-lactalbumin thatonly exists in UF. The role of minerals (cal-cium and inorganic phosphate) is taken intoaccount through empirical linear correla-tions between fluxes and soluble elementforms, using a software dedicated to themineral speciation in acid media togetherwith additional analyses in alkaline media.Results suggest that soluble minerals areinvolved in the whole fouling, regardlessof the pH. A careful analysis of cleaningprotocols of membranes vs. pH indirectlyshows this particular role of minerals inthe irreversible fouling. Results evidence thedetermining role of free calcium (Ca2+) inthe inorganic fouling of organic membranes.


Although already associated calcium wouldnot be involved in the irreversible fouling ofmembranes, free calcium is prone to precipi-tate in/on the membrane during filtration,probably due to a higher concentration at themembrane interface owing to concentrationpolarisation and a destabilising role playedby themembrane surface. Finally, on the basisof thismineral role,wehave shown that a sim-plified cleaning protocol with only one singlealkaline step is efficient enough for UF, NFand RO fouled by skim milks in the6.7–11.5 pH range.

Acknowledgements: The authors acknowl-edge the PROFAS program between Franceand Algeria for the financial support for Ph.D.of H.B. as well as the French National Agencyof Research (ANR) for the financial support ofthe ECOPROM project of the National Programfor Research in Nutrition (PNRA).

REFERENCES

[1] Ahmad S., Piot M., Rousseau F., GrongnetJ.F., Gaucheron F., Physico-chemical changesin casein micelles of buffalo and cow milksas a function of alkalinization, Dairy Sci.Technol. 89 (to appear).

[2] Bacchin P., Aimar P., Field R.W., Criticaland sustainable fluxes: theory experimentsand applications, J. Membr. Sci. 281 (2006)42–69.

[3] Bégoin L., Rabiller-Baudry M., Chaufer B.,Faille C., Blanpain-Avet P., Bénezech T.,Doneva T.A., Methodology of analysis of aspiral wound module. Application to PESmembrane of ultrafiltration of skimmedmilk, Desalination 192 (2006) 40–53.

[4] Bouzid H., Rabiller-Baudry M., Paugam L.,Rousseau F., Derriche Z., Bettahar N.E.,Impact of zeta potential and size of caseinsas precursors of fouling deposit on limitingand critical fluxes in spiral ultrafiltration ofmodified skim milks, J. Membr. Sci. 314(2008) 67–75.

[5] CFM and ADEME, Inquiry: brake andbottlenecks for membrane diffusion in indus-try (2005) www.ademe.fr.

[6] Dalgleish D.G., Measurement of electropho-retic mobilities and zeta potentials of particlesfrom milk using laser Doppler electrophore-sis, J. Dairy Res. 51 (1984) 425–438.

[7] Daufin G., René F., Aimar P., Les séparationsparmembrane dans les procédés de l’industriealimentaire, Lavoisier, Paris, 1998.

[8] De Kruif C.G., Supra-aggregates of caseinmicelles as a prelude to coagulation, J. DairySci. 81 (1998) 3019–3028.

[9] Delaunay D., Nettoyage éco-efficace demembranes planes et spirales d’ultrafiltrationde lait écrémé. Approches physico-chimi-ques et hydrodynamiques concertées, Ph.D.Thesis, Université Rennes 1, France, 2007.

[10] Delaunay D., Rabiller-Baudry M., Gozálvez-Zafrilla J.M., Balannec B., Frappart M.,Paugam L., Mapping of protein fouling byFTIR-ATR as experimental tool to studymembrane fouling and fluid velocity profilein various geometries and validation by CFDsimulation, Chem. Eng. Process 47 (2008)1106–1117.

[11] Espinasse B., Bacchin P., Aimar P., On anexperimental method to measure critical fluxin ultrafiltration, Desalination 146 (2002)91–96.

[12] Field R.W., Wu D., Howell J.A., Gupta B.B.,Critical 1 flux concept for microfiltrationfouling, J. Membr. Sci. 100 (1995) 259–272.

[13] Gésan-Guiziou G., Boyaval E., Daufin G.,Critical stability conditions in crossflowmicrofiltration of skimmed milk: transitionto irreversible deposition, J. Membr. Sci. 158(1999) 211–222.

[14] Gésan-Guiziou G., Jimenez A., Arcelin C.,Cake properties in dead-end ultrafiltration ofcasein micelles: determination of criticaloperating conditions, Desalination 199(2006) 20–22.

[15] Horne D.S., Casein micelle structure: mod-els and muddles, Curr. Opin. Coll. InterfaceSci. 11 (2006) 148–153.

[16] Huisman I.H., Vellenga E., Tragardh G.,Tragardh C., The influence of the membranezeta potential on the critical flux for cross-flow microfiltration of particles suspension,J. Membr. Sci. 156 (1999) 153–158.

[17] Jimenez-Lopez A.J.E., Leconte N.,Dehainault O., Geneste C., Fromont L.,Gésan-Guiziou G., Role of milk constituentson critical conditions and deposit structure


in skim milk microfiltration (0.1 μm), Sep.Purif. Technol. 61 (2008) 33–43.

[18] Le Berre O., Daufin G., Microfiltration(0.1 μm) of milk: effect of protein size andcharge, J. Dairy Res. 65 (1998) 443–455.

[19] Marchin S., Puteaux J.L., Pignon F., Léonil J.,Effects of the environmental factors on thecasein micelle structure studied by cryo-transmission electron microscopy and smallangle X-ray scattering/ultra-small angleX-ray scattering, J. Chem. Phys. 126 (2007)doi: 10.1063/1.2409933.

[20] Mekmene O., Le Graët Y., Gaucheron F., Amodel for predicting salt equilibria in milk andmineral-enrichedmilks, J. FoodChem. (2009).doi: 10.1016/j.foodchem.2009.02.039.

[21] Michalsky M.C., Michel F., Sainmont D.,Briard V., Apparent zeta potential as a tool toassess mechanical damages to the milk fatglobule membrane, Coll. Surf. B: Biointer-faces 23 (2001) 23–30.

[22] Rabiller-Baudry M., Bégoin L., DelaunayD., Paugam L., Chaufer B., A dual approachof membrane cleaning based on physico-chemistry and hydrodynamics. Applicationto PES membrane of dairy industry, Chem.Eng. Process 47 (2008) 267–275.

[23] Rabiller-Baudry M., Bouguen A., Lucas D.,Chaufer B., Physico-chemical characteriza-tion of proteins by capillary electrophoresis,J. Chromatogr. B 706 (1998) 23–32.

[24] Rabiller-Baudry M., Chaufer B., Specificadsorption of 1 phosphate ions on proteinsevidenced by capillary electrophoresis andreversed-phase high performance liquidchromatography, J. Chromatogr. B 753(2001) 67–77.

[25] Rabiller-Baudry M., Delaunay D., Paugam L.,Pihlajamäki A., Nyström M., Complemen-tary characterisations by streaming potentialand FTIR-ATR of surface of virgin andfouled PES ultrafiltration membrane: Whatkind of information on fouling occurrence?,in: Szymczyk A. (Ed.), Surface ElectricalPhenomena in Membranes and Microchan-nels, Transworld Research Network Edi-tions, Kerala, India, 2008.

[26] Rabiller-Baudry M., Gésan-Guiziou G.,Roldan-Calbo D., Beaulieu S., Michel F.,

Limiting flux in skimmed milk ultrafiltra-tion: impact of electrostatic repulsion due tocasein micelles, Desalination 175 (2005)49–59.

[27] Rabiller-Baudry M., LeMauxM., Chaufer B.,Bégoin L., Characterisation of cleaned andfouled membrane by ATR-FTIR and EDXanalysis coupled with SEM: application to UFof skimmed milk with a PES membrane,Desalination 146 (2002) 123–128.

[28] Resmini P., Pellegrino L., Andreini R., Prati F.,Determinazione delle sieroproteine solubili dellatte per HPLC (cromatographia liquida ad altaprestazione) in fase inversa, Sci. E. TechnicaLatterio-Casearia 40 (1989) 7–23.

[29] Schmidt D.G., Both P., Van Markwijk B.W.,Buchheim W., The determination of size andmolecular weight of casein micelles bymeans of light scattering and electronmicroscopy, Biochim. Biophys. Acta Prot.Struct. 365 (1974) 72–79.

[30] Van Leeuwen H.P., Duval J.F.L., Faradaicdouble layer depolarization in electrokinet-ics: Onsager relations and substrate limita-tions, J. Coll. Interface Sci. 309 (2007)350–359.

[31] Vourch M., Balannec B., Chaufer B.,Dorange G., Nanofiltration and reverseosmosis of model process waters from thedairy industry to produce water for reuse,Desalination 172 (2005) 245–256.

[32] Wu D., Howell J.A., Field W.R., Criticalflux measurement for model colloids, J.Membr. Sci. 152 (1999) 89–98.

[33] Youravong W., Grandison A.S., Lewis M.J.,The effect of physico-chemical changes oncritical flux of skimmed milk ultrafiltration,Songklanakarin J. Sci. Technol. 24 (2002)929–939.

[34] Youravong W., Grandison A.S., Lewis M.J.,Effect of hydrodynamic and physico-chem-ical changes on critical flux of milk proteinsuspensions, J. Dairy Res. 69 (2002)443–455.

[35] Youravong W., Lewis M.J., Grandisons A.S.,Critical flux in ultrafiltration of skimmed milk,Trans. IChemE, part C 81 (2003) 303–308.


Documents

Anh Huong Cua PH Den Luu Luong Loc Sua