Lait 87 (2007) 287–300 Available online at:c© INRA, EDP Sciences, 2007 www.lelait-journal.orgDOI: 10.1051/lait:2007021

Original article

The stabilization mechanism of acidified milkdrinks induced by carboxymethylcellulose

Baiqiao Dua, Jing Lia, Hongbin Zhanga*, Ping Chenb, Long Huangb,Jianjun Zhoub

a Department of Polymer Science and Engineering,Shanghai Jiao Tong University, Shanghai, 200240, China

b Danisco (China) Co., Ltd., Kunshan, 215300, China

Abstract – In the present work carboxymethylcellulose (CMC) was used as a stabilizer to avoid theflocculation of milk proteins in acidified milk drinks. The particle diameter and ζ-potential evolu-tion during acidification of casein micelles in CMC were studied. The experimental results indicatethat the adsorption of CMC takes place at and below pH 5.2 and that electrosorption may be thedriving force for the adsorption of CMC onto the casein micelles. The stability of acidified milkdrinks induced by CMC could be explained presumably by steric stabilization caused by the an-chor of CMC onto the casein micelles’ surface rather than electrostatic repulsion. Above pH 5.2phase separation of the casein/CMC mixture corresponding to thermodynamic incompatibility wasfound at high CMC concentration, while below pH 5.2 the adsorption of CMC led to either sta-bilization or bridging among casein micelles depending on the CMC concentration. In addition,the non-adsorbed CMC (serum CMC) increased the viscosity of the serum and thus contributed topreventing casein micelles from precipitating.

acidified milk drink / carboxymethylcellulose (CMC) / casein micelle / electrosorption /stability

摘摘摘要要要 –羧羧羧甲甲甲基基基纤纤纤维维维素素素钠钠钠稳稳稳定定定酸酸酸性性性乳乳乳饮饮饮料料料的的的机机机理理理研研研究究究。。。羧甲基纤维素钠 (CMC)可作为稳定剂,阻止酸性乳饮料中乳蛋白发生絮凝。本文研究了酪蛋白/CMC混合体系中,酪蛋白粒径和ζ-电位随体系pH降低的变化。研究表明,当体系 pH�5.2时, CMC因静电作用在酪蛋白胶粒上发生吸附。酸性乳体系的稳定性可能源于CMC在酪蛋白表面上发生吸附产生的空间位阻效应而非静电排斥作用。若体系 pH>5.2,当CMC浓度较高时,体系因热力学不相容而发生相分离。而 pH<5.2时,根据CMC浓度的不同,吸附在酪蛋白表面的CMC可使体系稳定,也可使体系因酪蛋白胶粒间的架桥絮凝而发生失稳。此外,由于体系中非吸附的CMC可增加体系的粘度,从而降低酪蛋白胶束的沉降速度,这也有利于提高酸性乳饮料的稳定性。

酸酸酸性性性乳乳乳饮饮饮料料料 /羧羧羧甲甲甲基基基纤纤纤维维维素素素钠钠钠(CMC) /酪酪酪蛋蛋蛋白白白胶胶胶束束束 /静静静电电电吸吸吸附附附 /稳稳稳定定定性性性

Résumé – Mécanisme de stabilisation de boissons laitières acidifiées induite par la carboxy-méthylcellulose. Dans le présent travail, la carboxyméthylcellulose (CMC) a été utilisée commestabilisant pour éviter la floculation des protéines laitières dans des boissons laitières acidifiées.Le diamètre des particules et l’évolution du potentiel zeta au cours de l’acidification des micelles

* Corresponding author (通讯作者): hbzhang@sjtu.edu.cn

Article published by EDP Sciences and available at http://www.lelait-journal.org or http://dx.doi.org/10.1051/lait:2007021

288 B. Du et al.

de caséine dans la CMC ont été étudiés. Les résultats expérimentaux indiquent que l’adsorption deCMC a lieu au pH inférieur ou égal à 5,2 et que l’électrosorption est la force motrice de l’adsorptionde CMC sur les micelles de caséine. La stabilité des boissons laitières acidifiées induite par la CMCpourrait s’expliquer probablement par la stabilisation stérique provoquée par l’ancrage de CMC surla surface des micelles de caséine plutôt que par répulsion électrostatique et le comportement dephase du mélange caséine/CMC correspondait à l’incompatibilité thermodynamique au dessus depH 5,2 et à l’adsorption, conduisant soit à la stabilisation, soit à la liaison entre les micelles decaséine, à pH inférieur à 5,2. La CMC non adsorbée (CMC du sérum) augmente la viscosité dusérum et contribue ainsi à prévenir la précipitation des micelles de caséine.

lait acidifié / carboxyméthylcellulose /micelle de caséine / électrosorption / stabilité

1. INTRODUCTION

Acidified milk drinks are popular dairyproducts worldwide. This beverage can bedescribed as an acidified protein liquidsystem with stability and viscosity simi-lar to natural milk. Such drinks are usu-ally composed of an acid dairy phase (fer-mented base) or a neutral base (milk) withan acidic medium (fruit phases such aspulp, fruit concentration, etc.) which canbe flavored. Upon lowering the pH ofmilk, as is done during the preparation ofacidified milk drinks, the native stabiliza-tion mechanism of casein micelles fails.This failure is believed to be related tothe decrease in steric repulsion after thecollapse of the extended conformation ofκ-casein which is located on the caseinmicelle surface [24]. Therefore, acidifiedmilk drinks need the addition of a stabilizerto avoid the flocculation of milk proteinsand subsequent macroscopic whey sepa-ration. Anionic polysaccharides, such ashigh-methoxyl pectin, propyleneglycoalgi-nate (PGA), soybean soluble polysaccha-rides (SSPS) and carboxymethylcellulose(CMC) are often used to achieve this [22].The stabilizer can prevent aggregation ofcasein micelles or be the cause of it. Thesephenomena are believed to be related to theinteraction between polysaccharides andproteins; the outcome depends on whetherthe polysaccharides adsorb onto the caseinparticles or not.

A lot of work has been reported on theinteraction between pectin and milk pro-tein [11, 15, 16]. Pectin has a chain struc-ture of α-(1→4)-linked D-galacturonicacid units interrupted by the insertionof α-(1→2)-linked L-rhamnopyranosylresidues. High-methoxyl pectin has beenwidely used as a stabilizer in acidifiedmilk drinks to prevent flocculation ofmilk proteins. Pectin is a non-adsorbingpolymer when it is in solution with caseinmicelles at pH 6.7. The adsorption ofpectin onto casein micelles is multilayeredand takes place at and below pH 5.0 [25].The mechanism of stabilization by pectinwas postulated to involve the blockwisedistribution of charges along the pectinchain. Adsorption of the pectin chain ontothe micelle surface would take place onlyat the charged blocks, while the unchargedstretches in between form entropy-richloops that extend into the solution. Thesteric repulsion caused by these loopsmaintains the pectin-coated casein mi-celles stable at low pH [10, 23]. However,in a stable system, not all of the pectinaddition adsorbs onto the casein micelles.A weak gel or network structure composedof non-adsorbed pectin in the serum andpectin-coated casein micelles promotesthe acidified milk drinks’ long-termstability [3].

SSPS, extracted from soybean cotyle-dons, has recently been used as a stabilizerin acidified milk drinks [2]. SSPS, likepectin, is composed of D-galactose,

Stabilization of acidified milk drinks 289

L-arabinose, D-galacturonic acid andL-rhamnose. However, the content ofneutral monosaccharides in SSPS is muchhigher than that in pectin [13]. The sta-bilizing ability and mechanism of SSPShave been investigated in comparison withpectin [12, 14]. SSPS, just like pectin,stabilizes acidified milk drinks by a steric-stabilizing effect due to the adsorbed SSPSlayer. However, because of the thick layerof neutral sugar side chains of SSPS onthe surface of the protein particles, thestability of acidified milk drinks inducedby SSPS is different from that inducedby pectin. Less SSPS is more effectivethan pectin at stabilizing and dispersingprotein particles. SSPS could give riseto better stability at pH < 4.2 and thestability was not affected by pH betweenpH 4.2 and 3.2, while acidified milkdrinks homogenized with pectin showeda particle size distribution that dependedon pH.

CMC, one of the important cellulosederivatives, is a typically anionic polysac-charide and has been widely used in phar-maceuticals, cosmetics and foods as anemulsifying agent, stabilizer or rheolog-ical control, etc. [4, 17]. CMC chainsare linear β-(1→4)-linked glucopyranoseresidues. A maximum degree of substitu-tion (DS) of 1.5 is permitted, but more typ-ically DS is in the range 0.6–0.95 for foodapplications. CMC is generally found insodium salt form, a water-soluble productfor DS > 0.5. One of the important charac-teristics of the CMC is that it can be dis-solved in both hot and cold water and hascertain viscosity. CMC aqueous solutionsof higher DS exhibit pseudoplastic behav-ior [4]. On the other hand, CMC is a polaradhesive and as such may allow the forma-tion of complexes with proteins such as ca-seins at, or around, the isoelectric region ofthe protein [17]. It has been reported thatCMC interacts with β-casein and hindersthe thermal or Ca2+-induced aggregation ofthe protein [8].

CMC is commonly used as a stabiliz-ing agent in acidified milk drinks instead ofpectin in Asia, especially in China. CMC ischosen because of its large application inthe food industry and low cost. The mainaim of the present work is to investigate theinteraction between CMC and casein mi-celles as a function of the pH. The mech-anism by which CMC stabilizes the acidi-fied milk drinks is addressed.

2. MATERIALS AND METHODS

2.1. Materials

CMC with different molecular weight(700 000; 250 000) and a degree of sub-stitution (DS) of 0.9 were purchased fromAcros Organics (New Jersey, USA). Skimmilk powders and whole milk powderswere given by the Fonterra Co. Ltd., NewZealand. A 500 g·kg−1 citric acid solu-tion was prepared by mixing citric acidmonohydrate (from the Shanghai Chemi-cal Reagent Co. Ltd., China) with distilledwater.

2.2. Acidified milk drink preparation

The samples were prepared on a pilotscale through the following successive pro-cess steps. The 80 g·kg−1 milk solids notfat (MSNF) reconstituted milk was pre-pared by mixing milk powders and distilledwater at 45 ˚C for 30 min. Meanwhile,10 g·kg−1 CMC was dissolved in distilledwater at 75 ˚C by stirring for 20 min. Thenthe CMC solution was added to reconsti-tuted milk and the MSNF was adjusted to40 g·kg−1. The pH of this solution was ad-justed to various values with 500 g·kg−1

citric acid. For some measurements in thepresent work described in the followingsections, additional post-treatment (heat-ing and homogenization) based on ac-tual processing was applied to the samples

290 B. Du et al.

in order to make the final samples simi-lar to commercial products. All measure-ments were performed after the sample wasstored at room temperature overnight.

2.3. Dynamic light scattering (DLS)experiments

DLS experiments were carried out witha particle size analyzer (Malvern Zetasizer3000 HSA, Malvern Instruments, Worces-tershire, UK). The system consisted ofan optics unit with a 10 W max outputHe-Ne laser, and a Malvern K7132 cor-relation used in serial configuration. TheZetasizer 3000 HSA worked at a fixed scat-tering angle of 90˚ and the wavelength ofthe laser beam was 633 nm. The PS dis-posable cuvette (15 mm, 1.5 mL) (Plas-tibrand, Wertheim, Germany) containingthe sample was thermostatted by a Joule-Peltier thermostat at 20 ˚C. The samplewas made by dispersing 80 g·kg−1 recon-stituted skim milk in SMUF (1:100). Then0.5% CMC was added to the mixture atabout neutral pH (6.6–6.7). All solutionsin this measurement were prepared withultra-pure water with 18.2 MΩ·cm−1 (Mil-lipore, Bedford, MA, USA), and filteredthrough 0.22-μm membrane filters priorto use. Before measurement, the sampleswere filtrated (0.45 μm; Millipore). The ap-parent diameter of the micelles was mon-itored while acidifying the diluted skimmilk with citric acid. All measurementswere performed three times. The biggestvariance of the measurement is below 8%.

2.4. Measurement of zeta potential

The ζ-potential of casein micelles wasdetermined using a particle electrophoresisinstrument (Zetasizer Nano ZS, MalvernInstruments, Worcestershire, UK) whichmeasures the direction and velocity ofdroplet movement in an applied electric

field at 20 ˚C. The ζ-potential provides anestimate of the net charge of a particle mea-sured at the ‘shear plane’, which dependson the charge on the actual particle (in thiscase casein micelle and polysaccharides)plus the charge associated with any ionsthat move along with the particle in theelectric field. An individual ζ-potential wasdetermined from the average of three read-ings taken on the same sample.

2.5. Confocal scanning lasermicroscopy (CSLM)

The obtained samples were preheatedto 65 ˚C and then homogenized at 200bar with a two-stage value Rannie TYPE8.30 H homogenizer (APV Rannie A/S,Denmark). Imaging was performed onthe sample using a Leica TCS SP2 con-focal scanning laser microscope, config-ured with a Leica DM IRE2 inverted mi-croscope (Leica Microsystems, Germany).The protein and fat were stained by FITCand Nile Red, respectively. The 488-nmlaser beam was used for excitation, in-ducing a fluorescent emission of FITC,detected between 503 and 543 nm. The548 nm laser beam was used for excita-tion of Nile Red, detected between 605 and673 nm.

2.6. Dynamic rheological properties

The samples were preheated to 65 ˚Cand homogenized at 200 bar, then pasteur-ized at 90 ˚C for 30 min, and finally cooledto room temperature. Small deformationoscillatory rheology was done at 25 ˚C us-ing an adapted stirrer vessel geometry ona stress-controlled rheometer (TA Instru-ments AR 2000, New Castle, USA). To testfor the presence of a weak gel structure, os-cillation was set at 0.1 Hz and 4.1×10 −3 Paafter shearing of the sample at 100 s−1 for3 min.

Stabilization of acidified milk drinks 291

Figure 1. Diameter of casein micelles, measured with DLS, as a function of pH during acidificationwith citrate acid for 80 g·kg−1 skim milk diluted 100 times in SMUF. The filled squares refer tocasein micelles without CMC, the open circles refer to casein micelles and 30 mg·kg−1 CMC, theopen diamonds refer to casein micelles and 45 mg·kg−1 CMC, the open triangles to casein micellesand 80 mg·kg−1 CMC, and the crosses to casein micelles and 400 mg·kg−1 CMC. The Mw of CMCis 250 000.

3. RESULTS AND DISCUSSION

3.1. Electrosorption of CMC ontocasein micelles

3.1.1. Effect of the concentration ofCMC on the diameter of caseinmicelles during acidification

A similar experiment following Tuinieret al. [25] was done as shown in Figure 1,the variation in particle diameter of ca-sein micelles with different concentrationsof CMC added during acidification withcitric acid. The sample was made by dis-persing 80 g·kg−1 reconstituted skim milkin SMUF (1:100). The filled squares referto casein micelles without the addition ofCMC. Just before citric acid is added, the

diameter is close to 230 nm, which is inagreement with earlier DLS results regard-ing the size of the casein micelle [7, 25].Upon decreasing the pH from 5.8 to 5.0 thecasein micelle diameter decreases due tothe shrinking of the casein micelle becauseof the solution of some fraction of α- andβ-caseins [25]. At about pH 5.0, the ap-parent casein micelle size greatly increasesdue to the formation of aggregates. Thesuspension becomes unstable and macro-scopic flocculation is observed after a fewminutes. This observation is the same asthat of Tuinier et al. [25].

The open circles refer to the same sam-ple but with 30 mg·kg−1 CMC added. Atneutral pH, we find a small increase insize, as compared with the bare casein mi-celle size. A similar phenomenon was also

292 B. Du et al.

observed in the size evolution during acidi-fication of pectin/casein mixtures [25]. Theincrease in size has been interpreted basedon the Stokes-Einstein relation as follows:

D = kT/3πηmdeff (1)

where ηm is the medium viscosity anddeff is the effective hydrodynamic diameter.In diluted dispersions, the self-diffusioncoefficient, D, is related to the effectivehydrodynamic diameter of the sphere, deff

according to equation (1). The viscos-ity of SMUF was taken as the mediumviscosity, ηm. However, upon addition ofCMC, the actual medium viscosity willincrease. It is thus supposed that the in-crease in the particle diameter is traceablefrom the calculation method of DLS [25].Alternatively, Nakamura et al. [14] alsofound the initial diameter of skim milk in-creased with increasing pectin concentra-tion at neutral pH. Taking into account thatthe DLS measurements were carried outafter the samples were extensively dilutedin the corresponding ultrafiltration perme-ate, and assuming that the measurement ofparticle diameter was not affected by theviscosity of the pectin, they attributed thisincrease in size to an association of pectinto casein micelles at neutral pH, possiblybecause of the existence of calcium ions. Inour case, the increase in diameter of caseinmicelles upon addition of CMC at neutralpH needs further clarification. The additionof CMC has no effect on the decrease in thecasein micelle size. Upon decreasing thepH to close to 5.2, we find that the diam-eter of casein micelles in the mixture withCMC increases in comparison with that inthe absence of CMC. This is caused by theadsorption of CMC onto casein micelles,which leads to effectively larger casein mi-celles. The negative charge might be statis-tically distributed along the CMC chains,presumably yielding a conformation withmany loops [9]. For adsorbed CMC, theseconceivable loops might extend into thesolution and cause a repulsive interaction

between the casein micelles at low pH inthe same way as κ-casein chains do at neu-tral pH. The adsorption of CMC onto ca-sein micelles takes place just before caseinmicelle aggregation would have started inthe absence of CMC. The apparent diam-eter of casein micelles increased with de-creasing pH and seemed to level off toca. 300 nm at a pH range of 4.6 to 4.4in the presence of 30 mg·kg−1 CMC. Atstill lower pH values it was observed thatthe diameter of casein micelles with ad-sorbed CMC decreased between pH 4.4and 4.0. With further decreasing pH, thenegative charge on CMC chains becomesless and less while the positive charge oncasein micelles becomes more and more.More positively charged casein micelleswould adsorb more CMC chains. At thesame time, CMC chains might absorb in alikely flatter conformation with small loopsfor the lower charge on CMC when lower-ing the pH. The collapse of the adsorbedlayer might be the reason that the diame-ters of casein micelles with adsorbed CMCdecrease in this pH range. The aggrega-tion of casein micelles exceeding 1 μmwas observed at pH 3.9 with 30 mg·kg−1

CMC. Probably, a CMC concentration of30 mg·kg−1 is not sufficient to provide athick enough adsorbed layer on casein mi-celles in the system. The collapse of theCMC adsorbed layer results in a decreasein the steric repulsion between CMC-coated casein micelles, and thus the aggre-gation occurs at pH 3.9. This phenomenonis similar to the collapse of κ-casein whichleads to aggregation of native casein mi-celles when lowering the pH [24].

The size evolution during acidificationof 45 mg·kg−1 CMC in diluted casein mi-celle suspension in SMUF was monitoredas shown, referred to by open diamonds.Upon decreasing the pH, the size evolu-tion is similar to that upon addition of30 mg·kg−1 CMC. The only difference isthat the aggregation takes place at a lowerpH of 3.6. The increasing concentration of

Stabilization of acidified milk drinks 293

CMC might be assumed to lead to a thickeradsorption, and consequently the stabilityof the mixture at the lower pH of 3.6.

Upon further increasing the CMC con-centration, the size evolution of casein mi-celles during acidification in the presenceof 80 mg·kg−1 and 400 mg·kg−1 CMC isalso shown in Figure 1 (open trianglesand crosses, respectively). Casein micelleswith 80 mg·kg−1 CMC are stable until thepH is decreased to 3.0. The casein mi-celles maintain stability and level off to c.a.300 nm in size even at the low pH of 3.0.The particle diameter with 400 mg·kg−1

CMC added level off to c.a. 420 nm atpH 3.0. It seems that upon lowering the pHin the range 4.3 to 3.0 there is a balance be-tween the incessant adsorption of CMC be-cause of the more positively charged caseinmicelles and collapse of the CMC adsorp-tion layer due to protonation. Adsorptiontakes place efficiently when more CMC ispresent in casein micelles. And the thicklayer of CMC is able to maintain the sta-bility of casein micelles at low pH.

3.1.2. Effect of the concentrationof CMC on the ζ-potentialof casein micelles duringacidification

The ζ-potential of casein micelles in thesimilar samples described in Figure 1 isshown as a function of pH in Figure 2. Asthe net surface charge of casein micellesis negative, the ζ-potential is also negative.The changes reported refer to the absolutemagnitude of the ζ-potential.

The ζ-potentials for native casein mi-celles, measured in SMUF buffer, rangedfrom about –16 to –8 mV over the pH rangeof 6.7 to 4.6. Considering only the proteinpart of casein micelles, a steady increasein ζ-potential with increasing pH would beexpected [20]. However, the observed re-lationship between the ζ-potential of na-tive casein micelles from milk dispersed

in SMUF buffer and pH is considerablymore complex. Overall, the ζ-potential pHprofile shown in Figure 2 is similar tothat reported by Schmidt and Poll [18]and Anema and Klostermeyer [1]. The ζ-potential at pH 4.6 was higher than zero(Fig. 2). This contrasts with the resultsof Darling and Dickson [5, 6], who re-ported a ζ-potential of zero at the isoelec-tric pH of casein (pH 4.6). Due to the com-plete solubilization of the colloidal calciumphosphate, the calcium and phosphate con-centrations, the calcium ion activity andthe ionic strength of natural milk serumat pH 4.6 would be considerably higherthan in the SMUF buffer used in these ex-periments, which simulates milk serum atpH 6.6. It is possible that these differencesresult in a higher ζ-potential of the ca-sein micelles in the SMUF buffer systemat low pH.

The addition of CMC to the casein mi-celles seems to cause some increase in theζ-potential of casein micelles. A remark-able feature in Figure 2 is the increase inζ-potential between pH 5.2 and 4.8 uponaddition of CMC. The ζ-potential of ca-sein micelles in this pH range became morenegative when CMC was added, whichconfirms that anionic CMC molecules ad-sorbed onto the surfaces of casein mi-celles [19, 21]. This result may correspondto the particle size evolution in Figure 1.The fact that negatively charged CMCchains adsorbed onto the surface of ca-sein micelles depending on pH suggeststhat there are electrostatic attractions be-tween anionic carboxylate groups on theCMC molecules and cationic patches (i.e.– NH+3 ) on casein micelles. In the pH range4.8–4.3 the ζ-potential shows a plateau.Nevertheless, the results of the particle sizeevolution in Figure 1 show that the parti-cle size increased, especially at high con-centrations of CMC. As the pH is lowered,in the pH range 4.8 to 4.3, casein micelleswill take more positive charge and adsorbmore CMC molecules, which leads to the

294 B. Du et al.

Figure 2. Zeta potential of casein micelles as a function of pH during acidification with citric acidfor 80 g·kg−1 skim milk diluted 100 times in SMUF. The filled squares refer to casein micelleswithout CMC, the open circles refer to casein micelles and 30 mg·kg−1 CMC, the open diamondsrefer to casein micelles and 45 mg·kg−1 CMC, the open triangles to casein micelles and 80 mg·kg−1

CMC, and the crosses to casein micelles and 400 mg·kg−1 CMC. The Mw of CMC is 250 000.

increase in particle size. When the surfacecharge reached a certain value, there wasa strong electrostatic repulsion betweenthe surface and similarly charged CMC inthe aqueous phase, which limited furtheradsorption of the polyelectrolyte [21]. Itseems that the adsorption of CMC ontocasein micelles was not only induced butalso controlled by the electrostatic force.The decrease in the ζ-potential was clearlyobserved between pH 4.3 and 3. Thisphenomenon might be explained by thefact that in this pH range the adsorptionlayer is thick enough to prevent the fur-ther adsorption of CMC and the carboxy-late groups on the adsorbed CMC layerwere protonated at the lower pH. There-

fore, the ζ-potential decreases with lower-ing pH. The reason that the reduction inζ-potential is higher at lower CMC concen-trations (30 mg·kg−1 and 45 mg·kg−1) thanat higher CMC concentrations (80 mg·kg−1

and 400 mg·kg−1) is probably attributed tothe fact that more CMC adsorbs effectivelyonto the casein micelle at higher CMC con-centrations when lowering the pH.

The ζ-potential of CMC-coated caseinmicelles in the system with 80 mg·kg−1

and 400 mg·kg−1 CMC is low at pH 3.0 incomparison with that at neutral. Since thereare no aggregations if the adsorbed CMClayer of casein micelles is thick enough athigh CMC concentrations, as indicated inFigure 1, this result suggests that the steric

Stabilization of acidified milk drinks 295

stability generated by the CMC layer playsa main role in stabilizing casein micelles atlow pH values.

3.2. The stability of acidified milkdrinks

The neutral base composed of 4 g·kg−1

CMC/40 g·kg−1 MSNF (skim milk pow-der) was acidified to different pH val-ues. The stability of the system observed15 days after preparation is shown in Fig-ure 3. Below pH 5.2, there was no obviousphase separation and the system showedstability. This fact is in good agreementwith the abovementioned experimental re-sults, in which the effective interaction be-tween CMC and casein micelles is revealedto take place below pH 5.2. As for thephase separation above pH 5.2, it mightbe supposed to originate from a depletionflocculation between casein micelles andnon-adsorbing polysaccharides [22, 26].

Figure 4 shows the confocal micro-scopic protein distributions in acidifiedmilk drinks with 4 g·kg−1 CMC and40 g·kg−1 MSNF (whole milk powder)acidified to pH 6.7, 5.0 and 4.1, respec-tively. The systems acidified to pH 6.7and 5.0 have a number of obvious clus-ters. The degree of aggregation at pH 5.0was lower than that at pH 6.7. CMC couldadsorb onto casein micelles at pH 5.0and the adsorbed layer would maintain thestability of acidified milk drinks. How-ever, the systems show macroscopic floc-culation, as seen in the CSLM image.This might be related to the thin CMClayer adsorbed at this pH. The effect ofadding different concentrations of CMC,at pH 4.1, on the microstructure of acidi-fied milk drinks is illustrated in Figure 5.Particles of acidified milk drinks with1 g·kg−1 CMC showed an obvious aggre-gation. No obvious clusters were observedin the acidified milk drinks which were sta-bilized by 2 g·kg−1 and 4 g·kg−1 CMC;

the two samples showed a homogeneousappearance.

When CMC adsorbs onto the casein mi-celles at below pH 5.2, CMC may give riseto bridging among the casein micelles ormake the casein micelles stable, dependingon the CMC concentration. Adding a littleCMC might consequently result in bridg-ing flocculation, in which a single polymerchain may absorb onto two or more caseinmicelles, thereby connecting the particles.Above full coverage, the high surface cov-erage of CMC generally has a stabilizingeffect on casein micelles [25]. Our resultsindicate that the stability of casein micellesin acid medium induced by CMC dependsnot only on the interaction between thembut also on the CMC concentration.

3.3. The role of non-adsorbedCMC (serum CMC)in CMC-stabilizedacidified milk drinks

The results of particle size evolutionand ζ-potential of casein micelles withdifferent CMC concentrations as a func-tion of pH indicate that the steric stabilitywhich was generated by the adsorbed CMClayer maintained the stability of casein mi-celles at low pH. Here the role of serumCMC is further discussed. Dynamic rheo-logical properties of acidified milk drinkswere measured to determine whether thereare weak network structures composed ofCMC-coated casein micelles/serum CMC.Figure 6 shows the small deformationrheological properties of acidified milkdrinks after shearing the sample at 100 s−1

for 3 min. After the applied shearing isremoved, the system shows an obviousliquid-like character; as can be seen, theloss modulus, G′′ is always higher thanthe storage modulus, G′ with a small losstangent, tanδ of ca. 2.7. The effect of theserum CMC is to increase the viscosity ofthe acidified milk drinks. The increase in

296 B. Du et al.

Fig

ure

3.V

isua

lap

pear

ance

ofph

ase

beha

vior

ofac

idifi

edm

ilk

drin

ksat

diff

eren

tpH

valu

es15

days

afte

rpr

epar

atio

n(t

heco

ncen

trat

ion

ofC

MC

(Mw=

700

000)

is4

g·kg−

1).

Stabilization of acidified milk drinks 297

Fig

ure

4.C

onfo

cals

cann

ing

lase

rm

icro

grap

hsof

acid

ified

mil

kdr

inks

stab

iliz

edby

40g·k

g−1

MS

NF

and

4g·k

g−1

CM

C(M

w=

700

000)

atpH

(a)

6.7,

(b)

5.0,

(c)

4.1.

The

scal

eba

ris

16μ

m.

298 B. Du et al.

Fig

ure

5.C

onfo

cal

scan

ning

lase

rm

icro

grap

hsof

acid

ified

mil

kdr

inks

stab

iliz

edby

(a)

1g·k

g−1

CM

C,

(b)

2g·k

g−1

CM

C,

(c)

4g·k

g−1

CM

Cat

pH4.

10.T

hesc

ale

bar

is16μ

m.T

heM

wof

CM

Cis

700

000.

Stabilization of acidified milk drinks 299

Figure 6. Time evolution of the storage, loss moduli and loss tangent of a CMC-stabilized acidifiedmilk drink after shearing at 100 s−1for 3 min at 0.1 Hz, 4.1×10−3 Pa and 25 ˚C. The system contained40 g·kg−1 MSNF, 4 g·kg−1 CMC (Mw= 700 000) and 80 g·kg−1 sucrose.

viscosity of the system will decrease thesedimentation velocity of casein micelles.Therefore, the serum CMC contributes tothe stability by increasing the viscosity ofacidified milk drinks.

4. CONCLUSION

The stabilization mechanism of acid-ified milk drinks induced by CMC wasinvestigated. The present study indicatesthat steric stability generated by the ad-sorbed CMC layer may play a key rolein the stability of casein micelles at lowpH. There is no network composed of theCMC-coated casein micelles and the non-adsorbed CMC. The serum CMC increasesthe viscosity of the acidified milk drinks,which also contributes to the stability ofthe system.

Acknowledgements: The authors thank F.Madsen for help with the CSLM experiments,and X. Jia for useful discussions. H. Zhangthanks Danisco (China) Co., Ltd. for financialsupport. The authors are also indebted to the re-viewers for their constructive comments.

REFERENCES

[1] Anema S.G., Klostermeyer H., ζ-potentialsof casein micelles from reconstituted skimmilk heated at 120 ˚C, Int. Dairy J. 6 (1996)637–687.

[2] Asai I., Watari Y., Iida H., Masutake K.,Ochi T., Ohashi S., Furuta H., Maeda H.,Effect of soluble soybean polysaccharideon dispersion stability of acidified milkprotein, in: Nishinari K., Doi E. (Eds.),Food Hydrocolloids: Structure, propertiesand functions, Plenum Press, New York,1994, pp. 151–156.

[3] Boulenguer P., Laurent M.A., Comparisonof the stabilization mechanism of acid

300 B. Du et al.

dairy drinks (ADD) induced by pectin andsoluble soybean polysaccharide (SSP), in:Voragen F., Schols H., Visser R. (Eds.),Advanced in pectin and pectinase research,Kluwer Academic Publishers, Dordrecht,The Netherlands, 2003, pp. 467–480.

[4] Clasen C., Kulicke W.M., Determination ofviscoelastic and rheo-optical material func-tions of water-soluble cellulose derivatives,Prog. Polym. Sci. 26 (2001) 1839-1919.

[5] Darling D.F., Dickson J., The determinationof the zeta potential casein micelles, J. DairyRes. 46 (1979) 329–332.

[6] Darling D.F., Dickson J., Electrophoreticmobility of casein micelles, J. Dairy Res. 46(1979) 441–451.

[7] De Kruif C.G., Casein micelles: diffusivityas a function of renneting time, Langmuir 8(1992) 2932–2937.

[8] Delben F., Stefancich S., Interaction of foodproteins with polysaccharides, I. Propertiesupon mixing, J. Food Eng. 31 (1997)325–346.

[9] Fujimoto J., Petri D.F.S., Adsorption be-havior of carboxymethylcellulose on amino-terminated surfaces, Langmuir 17 (2001)56–60.

[10] Laurent M.A., Boulenguer P., Stabilizationmechanism of acid dairy drinks (AMD) in-duced by pectin, Food Hydrocoll. 17 (2003)445–454.

[11] Maroziene A., De Kruif C.G., Interaction ofpectin and casein micelles, Food Hydrocoll.14 (2000) 391-394.

[12] Nakamura A., Furuta H., Maeda H.,Takao T., Nagamatsu Y., Structural studiesby stepwise enzymatic degradation ofthe main backbone of soybean solublepolysaccharides consisting of galacturonanand rhamnogalacturonan, Biosci. Biotech.Bioch. 66 (2002) 1301–1313.

[13] Nakamura A., Furuta H., Kato M., MaedaN.Y., Effect of soybean soluble polysaccha-rides on the stability of milk protein underacidic conditions, Food Hydrocoll. 17 (2003)333–343.

[14] Nakamura A., Yoshida R., Maeda H.,Corredig M., The stabilizing behaviour ofsoluble polysaccharide and pectin in acidi-fied milk beverages, Int. Dairy J. 16 (2006)361–369.

[15] Parker A., Boulenguer P., Kravtchenko T.P.,Effect of the addition of high methoxy pectin

on the rheology and colloidal stability of acidmilk drinks, in: Nishinari K., Doi E. (Eds.),Food Hydrocolloids: Structure, propertiesand functions, Plenum Press, New York,1994, pp. 307–312.

[16] Pereyra R., Schmidt K.A., Wicker L.,Interaction and stabilization of acidified ca-sein dispersions with low and high methoxylpectins, J. Agr. Food Chem. 45 (1997)3448–3451.

[17] Phillips G.O., Williams P.A., Celllulosics,in: Handbook of hydrocolloids. Woodheadpublishing Ltd, Abington, England, 2000,pp. 219–229.

[18] Schmidt D.G., Poll J.K., Electrokinetic mea-surements on unheated and heated casein mi-celle systems, Nerth. Milk Dairy J. 40 (1986)269–280.

[19] Sejersen M.T., Salomonsen T., Ipsen R.,Clark R., Rolin C., Engelsen S.B., Zeta po-tential of pectin – stabilised casein aggre-gates in acidified milk drinks, Int. Dairy J.4 (2007) 302–307.

[20] Shaw D.J., Introduction to Colloid andSurface Chemistry, Butterworth and Co.,London, 1980.

[21] Surh J., Decker E.A., Mcclements D.J.,Influence of pH and pectin type on propertiesand stability of sodium-caseinate stabilizedoil-in-water emulsions, Food Hydrocoll. 20(2006) 607–618.

[22] Syber A., Bauer W.J., Klostermeyer H.,Polymer science concepts in dairy systems– an overview of milk protein and food hy-drocolloid interaction, Int. Dairy J. 8 (1998)179–193.

[23] Tromp R.H., De Kruif C.G., Van Eijk M.,Rolin C., On the mechanism of stabiliza-tion of acidified milk drinks by pectin, FoodHydrocoll. 18 (2004) 565–572.

[24] Tuinier R., Kruif C.G., Stability of caseinmicelles in milk, J. Chem. Phys. 17 (2002)1290–1295.

[25] Tuinier R., Ten Grotenhuis E., De KruifC.G., The effect of depolymerised guargum on the stability of skim milk, FoodHydrocoll. 14 (2000) 1–7.

[26] Tuinier R., Rolin C., De Kruif C.G.,Electrosorption of pectin onto caseinmicelles, Biomacromolecules 3 (2002)632–638.