DOI: 10.1021/la8006684 6649Langmuir 2009, 25(12), 6649–6657 Published on Web 05/11/2009

© 2009 American Chemical Society

pubs.acs.org/Langmuir

Theoretical Stability Maps for Guiding Preparation of Emulsions Stabilized

by Protein-Polysaccharide Interfacial Complexes

Young-Hee Cho and David Julian McClements*

Biopolymers and Colloids Laboratory, Department of Food Science, University of Massachusetts, Amherst,Massachusetts 01003

Received March 3, 2008. Revised Manuscript Received March 21, 2008

The purpose of this study was to evaluate the usefulness of a simple theoretical model at predicting the stability ofemulsions containing lipid droplets and polyelectrolytes. The influence of droplet concentration,mean droplet diameter,droplet charge and polyelectrolyte concentration on the aggregation stability of the emulsions was examined. Emulsionsstabilized by a globular protein (β-lactoglobulin) were prepared with different oil droplet (0.5-10 wt %) and pectin(0-0.5 wt %) concentrations at pH 7 (where lipid droplets and pectin molecules were both anionic) and pH 3.5 (wherelipid droplets were cationic and pectin molecules anionic). The particle charge, size, and creaming stability of theemulsions were then measured, and stability maps were constructed at pH 3.5 and 7. At pH 7, there was no evidence ofpectin adsorption to droplet surfaces, and the emulsions were stable to bridging f locculation, but depletion f locculationoccurred when the pectin concentration exceeded about 0.1 wt % (independent of droplet concentration). At pH 3.5,pectin adsorbed to the droplet surfaces, and the emulsions were unstable to bridging flocculation at intermediate pectinconcentrations (dependent on droplet concentration) and unstable to depletion flocculation at high pectin concentra-tions. At certain droplet and pectin concentrations, stable emulsions could be formed consisting of protein-coated lipiddroplets surrounded by a pectin layer. The information gained from this study would be useful for optimizing theproduction of emulsions stabilized by protein-polysaccharide interfacial complexes.

Introduction

Recently, it has been shown that an electrostatic layer-by-layer(LbL) deposition technique can be used to improve the stability ofprotein-coated lipid droplets to a variety of environmentalstresses, such as pH extremes, ionic strength, thermal processing,freezing, and dehydration.1-14 This technique is based on theelectrostatic deposition of charged polymers onto oppositelycharged colloidal particles,15-18 in this case charged polysacchar-ides onto charged protein-coated lipid droplets. These kinds ofinterfacial layers can be formed around lipid droplets using asimple procedure which is well-established in the literature.19

Initially, a “primary emulsion” containing lipid droplets coated

with a layer of protein molecules is prepared by homogenization.Subsequently, a “secondary emulsion” is formed by adsorbing anoppositely charged polysaccharide onto the droplet surfaces sothat each droplet is covered by a protein-polysaccharide coating.These two-component interfacial coatings often provide lipiddroplets with improved stability to environmental stresses suchas thermal processing, ionic strength, pH, freezing, and dehydra-tion than single-component coatings.7,10,12,13,20 In addition, theability to systematically control the properties of the interfacialcoatings can be used to develop delivery systems with novelcontrolled or triggered release properties.18 A major challengeassociated with utilizing this type of emulsion industrially is thatthey are highly susceptible to flocculation, and therefore, it isimportant to establish the optimum conditions required for theirpreparation.19,21

The purpose of the present study was to examine some of themajor factors that impact the preparation of stable oil-in-wateremulsions containing lipid droplets coated with globular pro-tein-anionic polysaccharide interfacial layers. These systemswere formed by mixing an anionic polysaccharide (pectin) withan emulsion containing lipid droplets coatedby a globular protein(β-lactoglobulin). We examined the influence of droplet concen-tration, droplet size, droplet charge, and polysaccharide concen-tration on the formation and stability of these emulsions.A special emphasis was placed on comparing the experimentallydetermined stability maps with theoretical stability mapscalculated using a recently developed theory.21

Theoretical Prediction of Emulsion Stability Maps

A simple theoretical model was recently developed to predictthe influence of various factors on the stability of emulsions

*To whom correspondence should be addressed. Tel: (413) 545-1019. Fax:(413) 545- 1262. E-mail: mcclements@foodsci.umass.edu.(1) Dickinson, E. Food Hydrocolloids 2003, 17, 25–39.(2) Dickinson, E. Colloids Surf., A 2006, 288, 3–11.(3) Dickinson, E.; Pawlowsky, K. J. Agric. Food Chem. 1997, 45, 3799–3806.(4) Gu, Y.; Decker, E.; McClements, D. Langmuir 2006, 22, 7480–7486.(5) Gu, Y. S.; Decker, A. E.; McClements, D. J. Langmuir 2005, 21, 5752–5760.(6) Gu, Y. S.; Decker, E. A.; McClements, D. J. Food Hydrocolloids 2005, 19,

83–91.(7) Gu, Y. S.; Regnier, L.; McClements, D. J. J. Colloid Interface Sci. 2005, 286,

551–558.(8) Guzey, D.; Kim, H. J.; McClements, D. J. FoodHydrocolloids 2004, 18, 967–

975.(9) Guzey, D.; McClements, D. J. Food Hydrocolloids 2006, 20, 124–131.(10) Guzey, D.; McClements, D. J. Food Biophysics 2006, 1, 30–40.(11) Guzey, D.; McClements, D. J. J. Agric. Food Chem. 2007, 55, 475–485.(12) Harnsilawat, T.; Pongsawatmanit, R.; McClements, D. J. Agric. Food

Chem. 2006, 54, 5540–5547.(13) Harnsilawat, T.; Pongsawatmanit, R.; McClements, D. Biomacromolecules

2006, 7, 2052–2058.(14) Harnsilawat, T.; Pongsawatmanit, R.; McClements, D. FoodHydrocolloids

2006, 20, 577–585.(15) Caruso, F. Adv. Mater. 2001, 13, 11–+.(16) Caruso, F.; Mohwald, H. J. Am. Chem. Soc. 1999, 121, 6039–6046.(17) Decher, G. Science 1997, 277, 1232–1237.(18) Decher, G.; Schlenoff, J. B. Multilayer Thin Films: Sequential Assembly of

Nanocomposite Materials; Wiley-VCH: Weinheim, 2003; xix, 524.(19) Guzey, D.; McClements, D. J. Adv. Colloid Interface Sci. 2006, 128, 227–

248.

(20) Aoki, T.; Decker, E. A.; McClements, D. J. Food Hydrocolloids 2005, 19,209–220.

(21) McClements, D. J. Langmuir 2005, 21, 9777–9785.

6650 DOI: 10.1021/la8006684 Langmuir 2009, 25(12), 6649–6657

Article Cho and McClements

prepared by mixing spherical droplets with oppositely chargedpolyelectrolytes.21 This model predicts that the stability of anemulsion can be divided into a number of different regimesdepending on the polyelectrolyte concentration (C).I. C = 0. In the absence of polyelectrolyte, the stability of

the droplets to aggregation is governed by the relative strength ofthe attractive (usually van der Waals and hydrophobic) andrepulsive (usually electrostatic and steric) interactions betweenthem. If the attractive interactions dominate, then the dropletswill aggregate, but if the repulsive interactions dominate, then thedroplets will remain as individual entities.II. 0 < C < CSat. Bridging flocculation occurs when the

polyelectrolyte concentration is insufficient to completely satu-rate the particle surfaces (CSat). This is because there are bothpositive and negative patches on the droplet surfaces, whichpromotes bridging flocculation due to sharing of single polyelec-trolyte molecules between neighboring droplets. These aggregatescannot subsequently be disrupted by the application of mechan-ical stresses, since there is always insufficient polyelectrolyteavailable to coat all of the droplets.III. CSat < C < CAds. Bridging flocculation also occurs

when the polyelectrolyte concentration is sufficient to completelysaturate the droplets surfaces, but it is too low to ensure that thedroplets are saturated with polyelectrolyte before a dropletcollision occurs (CAds). In principle, it should be possible todisrupt these aggregates by the application of mechanical stressesbecause there is sufficient polyelectrolyte present to saturate allthe droplet surfaces once the flocs have been broken.IV. CAds < C < CDep. Emulsions containing droplets

coated by a polyelectrolyte layer can be formed when theirsurfaces are rapidly and completely saturated with polyelectro-lyte, and when there is not enough free polyelectrolyte present inthe continuous phase to promote depletion flocculation. Underthese circumstances, it should be possible to prepare stableemulsions consisting of droplets completely surrounded by apolyelectrolyte layer (provided the net droplet repulsion over-comes the net droplet attraction).V. C>CDep.When the concentration of free polyelectrolyte

exceeds some critical value (CDep), depletion flocculationoccurs because the depletion forces make the overall attractiveforces strong enough to overcome the overall repulsive forces(e.g., electrostatic and steric).

The following expressions were derived for the criticalpolyelectrolyte concentrations mentioned above:21

CSat ¼ 3φΓSat

rð1Þ

CAds ¼ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi60ΓSat

2rPEφ

r3

sð2Þ

CDep ¼ M

NA

-1 þ ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi1-8vX

p

4v

!ð3Þ

where

X ¼ wDep

kBT

� �Crit

1

2πrPE2 r þ 23rPE

� �whereφ is the volume fraction of the particles,ΓSat is the surface

load of the polyelectrolyte at saturation (in kgm-2), r is the radius

of the spherical particles (in m), rPE is the radius of the polyelec-trolyte molecules in solution (in m),M is the molecular weight ofthe polyelectrolyte (in kg mol-1), v (=4rπPE

3/3) is the effectivemolar volume of the polyelectrolyte in solution (in m3), NA isAvogadro’s number, wDep is the strength of the depletion attrac-tion at droplet contact, which depends on droplet sizeand nonadsorbing polyelectrolyte concentration, kB is theBoltzmann’s constant, and T is the absolute temperature.

These equations can be used to develop “stabilitymaps”, whichindicate the region where stable emulsions can potentially becreated.21 To produce a colloidal dispersion that is stable toflocculation, it is necessary to ensure that (CSat andCAds) <C<CDep, i.e., that there is enough polyelectrolyte to completelysaturate the surfaces of the particles, but not too much freepolyelectrolyte to promote depletion flocculation. Using thevarious equations derived above for CSat, CDep, and CAds, it ispossible to generate stability maps for formation of polyelectro-lyte-coated droplets without promoting bridging flocculation anddepletion flocculation. Further details about the construction ofstability maps are given in an earlier reference.21 Stability mapspredicted using the above expressions will be compared withexperimental data on a system consisting of protein-coateddroplets and pectin (see below).

It should be noted that the equations given above weredeveloped assuming that emulsion droplets and polymer mole-cules encountered each other through Brownian motion. Inpractice, a droplet-polymer mixture may be stirred during itspreparation, which may change the relative frequencies of dro-plet-droplet and droplet-polymer collisions, thereby alteringthe tendency for bridging flocculation to occur. In addition,shearing an emulsion during mixing may disrupt any flocs thathave formed. It would therefore be useful in the future to extendthe above theory to include the effects of shear forces on collisionfrequencies and floc breakup.

Materials and Methods

Materials. Powdered β-Lg was obtained fromDavisco FoodsInternational (lot no JE 001-1-922, Le Sueur,MN). The proteincontent was reported to be 98.3% (dry basis) by the supplier, withβ-Lgmaking up 95.5% of the total protein. The moisture contentof the protein powder was reported to be 4.9%. The fat, ash, andlactose contents of this product are reported to be 0.3( 0.1, 2.5(0.2, and <0.5 wt %, respectively. Pectin extracted from citrusfruit was purchased from Sigma Chemical Company (lot no016K0713, St. Louis, MO). The degree of esterification (DE) ofthe pectin was reported to be 60% by the supplier. The averagemolecular weight of the pectin was determined to be 310 kDaby static light scattering (NanoZS, Malvern Instruments,Worcestershire, UK). Corn oil was purchased from a foodsupplier (Mazola, ACH Food Companies, Inc., Memphis, TN,USA) and used without further purification. The manufacturerreported that the corn oil contained approximately 14.3, 28.6, and57.1 wt % of saturated, monounsaturated, and polyunsaturatedfats, respectively. Analytical grade hydrochloric acid (HCl)and sodium hydroxide (NaOH) were purchased from theSigma Chemical Company (St. Louis, MO). Purified waterfrom a Nanopure water system (Nanopure Infinity, BarnsteadInternational, Iowa) was used for the preparation of all solutions.

Solution Preparation.An emulsifier solution containing 1 wt% protein was prepared by dispersing powdered β-Lg into 5 mMphosphate buffer (pH 7.0) and stirring for at least 2 h to ensurecomplete hydration. A 1 wt % pectin solution was prepared bydispersing powdered pectin into the same phosphate buffer andstirring for at least 4 h to ensure complete hydration.

Emulsion Preparation. A stock emulsion (d32 = 0.27 μm)was prepared by homogenizing 20 wt % corn oil with 80 wt %

DOI: 10.1021/la8006684 6651Langmuir 2009, 25(12), 6649–6657

Cho and McClements Article

aqueous emulsifier solution (1 wt % β-Lg, pH 7.0) with a high-speed blender (M133/1281-0, Biospec Products, Inc., ESGC,Switzerland) for 2 min followed by 5 passes through a two-stagehigh-pressure valve homogenizer (LAB 1000, APV-Gaulin,Wilmington, MA): the first stage at 3000 psi, the second stage at300 psi. Inone study, the effect of initial droplet sizewas examinedby using different homogenization pressures and/or number ofpasses to prepare three emulsions with different mean dropletdiameters: d32=0.26, 0.47, 0.62 μm. Emulsions containing differ-ent droplet concentrations (0.5 to 10 wt%) and pectin concentra-tions (0 to 0.5 wt %) were prepared by mixing differentproportions of stock emulsion, pectin solution (1 wt %, 5 mMphosphate, pH 7) and buffer solution (5 mM phosphate, pH 7).The pH of the resulting emulsions was then adjusted to eitherpH 7.0 or pH 3.5 usingHCl. To investigate the disruption of flocsformed during preparation, some of the emulsions were subjectedto an ultrasound treatment using a high-intensity ultrasonicgenerator (model 500, Sonic Disembrator, Fisher Scientific,Pittsburgh, PA) with a titanium alloy horn.ζ-Potential Measurements. To determine the electrical

charge on lipid droplets, emulsions were diluted to a dropletconcentration of approximately 0.01 wt % using an appropriatebuffer solution (at the same pHas the sample) and placed into themeasurement chamber of a microelectrophoresis instrument(ZEM 5300, Zetamaster, Malvern Instruments, Worcestershire,UK). This instrument determines the electrical charge (ζ-poten-tial) on the particles in an emulsionbymeasuring the direction andvelocity of particle movement in an applied electric field. Theζ-potential measurements are reported as the average andstandard deviation of measurements made on two freshlyprepared samples, with five readings made per sample.Particle Size Analysis. Emulsion samples were diluted to a

droplet concentration of approximately 0.005 wt % using anappropriate buffer solution (at the same pH as the sample) andplaced into the measurement chamber of the laser diffractioninstrument (Mastersizer X, Malvern Instruments Ltd., Malvern,UK). This instrument finds the particle size distribution of anemulsion that gives the best fit between the experimentalmeasure-ments and predictionsmadeusing light scattering theory (i.e.,Mietheory). A refractive index ratio of 1.08 was used by the instru-ment to calculate the particle size distributions.Measurements arereported as the volume weighted mean diameter: d43 = Σnidi

4/Σnidi

3 or as the volume-surfacemeandiameter: d32=Σnidi3/Σnidi

2

where ni is the number of droplets of diameter di. The particle sizemeasurements are reported as the average and standard deviationofmeasurementsmade on two freshly prepared samples, with tworeadings made per sample.Creaming Stability Measurements. Ten grams of each

emulsion were transferred into a test tube (with 15 mm internaldiameter and 125 mm height) and then stored for 24 h at roomtemperature. After storage, a number of emulsions separated intoan optically opaque “cream layer” at the top and a transparent(or turbid) “serum layer” at the bottom. The total height of theemulsions (HE) and the height of the cream layer (HC) weremeasured. The extent of creaming was characterized by the creamlayer thickness = (HC/HE) � 100. The cream layer thicknessprovided indirect information about the extent of droplet aggrega-tion in the emulsions: themore aggregation, the larger the particles,the faster the creaming, and the thicker the cream layer. Allmeasurements weremade on at least two freshly prepared samples.

Microstructure Measurements. Emulsion microstructurewas examined by optical microscopy (Nikon microscope EclipseE 400, Nikon Corporation, Japan). After 7 days storage, emul-sions were mixed in a glass test tube using a vortexer to prepare ahomogeneous sample, and then, a dropof emulsionwas placedona microscope slide and covered by a coverslip. Microstructureimages of emulsions were then obtained using a CCD camera(CCD-300-RC,DAGE-MTI,MichiganCity, IN), and the imageswere processed by Digital Image Processing Software (MicroVideo Instruments Inc., Avon, MA).

Statistical Analysis.Experimentswere performed twiceusingfreshly prepared samples. Averages and standard deviations werecalculated from these duplicate measurements.

Results and Discussions

Stability Map at pH 7. The droplet size, electrical charge,and creaming stability of β-lactoglobulin stabilized oil-in-wateremulsions was measured at pH 7.0 as a function of dropletconcentration (0.5 to 10 wt %) and pectin concentration (0 to0.5 wt %), and an experimental stability map was constructed(data not shown). This stability map was based on observationsof the creaming stability of the emulsions after 24 h storage atroom temperature. The creaming stability of the emulsions wascharacterized according to the degree of visible phase separationthat had occurred: stable (S), emulsions appeared homogeneousthroughout; partially unstable (U*), a cream layer was observedat the top of the tubes and a highly turbid or opaque serum layerwas observed at the bottom; highly unstable (U), a cream layerwas observed at the top of the tubes and either a transparent orslightly turbid serum layer was observed at the bottom. All theemulsions were stable to creaming when the pectin concentrationwas e0.06 wt %, were partially unstable from 0.08 to 0.1 wt %pectin, and were highly unstable at 0.3 and 0.5 wt % pectin,irrespective of the initial droplet concentration. This data in-dicated that therewas a critical flocculation concentration (CFC)for the emulsions somewhere between 0.08 and 0.1 wt % pectin,which did not depend strongly on droplet concentration. Weattribute the creaming instability of the emulsions at high pectinconcentrations to depletion flocculation caused by the presenceof nonadsorbed pectin molecules in the aqueous phase, as hasbeen reported by many other workers.1,8

At pH 7, the electrical charge on both the pectin molecules(ζ = -44.1 ( 0.6 mV) and the protein-coated lipid droplets(ζ=-61.1( 0.4mV)was negative, and therefore, one would notexpect the pectin molecules to adsorb to the droplet surfacesbecause of a strong electrostatic repulsion. The nonadsorbedpectin will therefore be excluded from a narrow region surround-ing each droplet, which is approximately equal to the radius ofhydration of the pectin molecules. Consequently, there will be apectin concentration gradient between the “exclusion zone”surrounding each droplet (Cpectin ≈ 0) and the bulk aqueousphase (Cpectin ≈ Cbulk).

22 This concentration gradient leads to thegeneration of an osmotic pressure that tends to drive the dropletsinto close proximity so as to reduce the total volume of the regionfrom where pectin molecules are excluded. The magnitude of thisosmotic pressure increases with increasing pectin concentration,until the depletion attraction is sufficiently strong to overcome therepulsive interactions operating between the droplets, and so thedroplets flocculate. The minimum pectin concentration whereflocculation is first observed is the CFC.

Measurements of the creaming stability of the emulsionsshowed that the thickness of the cream layer at the top of theunstable emulsions increased with decreasing droplet concentra-tion, and with increasing pectin concentration above the CFC(Figure 1). This result can be attributed to differences in theconcentration and packing of the droplets within the cream layer.At higher initial droplet concentrations, there are more dropletsavailable to pack into the creamed layer.22 In addition, at higherinitial droplet concentrations flocculation is more likely to lead tothe formation of a three-dimensional particle network thatextends throughout the volume of the container and therefore

(22) McClements, D. J. Food Emulsions: Principles, Practice, and Techniques,2nd ed.; CRC Press: Boca Raton, 2005.

6652 DOI: 10.1021/la8006684 Langmuir 2009, 25(12), 6649–6657

Article Cho and McClements

restricts further creaming.22As the pectin concentration increases,the strength of the depletion attraction between the dropletsincreases, which means that they are more likely to stick stronglytogether and form an open-structured particle network.23 Inaddition, the viscosity of the continuous phase increases withincreasing pectin concentration, which would slow down dropletmovement. After storage, themeasuredmean particle diameter ofall the emulsions was the same as the initial value (d32 = 0.27 (0.01 μm; d43 = 0.37 ( 0.01 μm), which indicated that theemulsions were stable to droplet coalescence, and that flocsformed in the emulsions above the CFC were disrupted whenthe emulsions were diluted for particle size measurements. Thedisruption of flocs upon dilution is commonly seen with emul-sions that have been flocculated through a depletion mechanismand can be attributed to the fact that the concentration ofnonadsorbed biopolymer in the continuous phase is reduced tobelow the CFC when the emulsion is diluted for the particle sizemeasurements.

At pH7, there should not be any bridging flocculation, becausethe pectin does not adsorb to the droplet surfaces (ΓSat = 0);hence,Csat andCads do not need to be calculated. Consequently, arelatively simple theoretical stability map can be calculated usingthe equations given above, since onlyCdep needs tobe determined.Theoretical calculations of the dependence of Cdep (eq 3) ondroplet concentration were made using experimentally deter-mined parameters for the pectin molecules and oil droplets usedin this study (Figure 2). Three droplet radii (r = 150, 200, and250 nm) were used in these calculations to examine the effects ofemulsion polydispersity. The median droplet radii of the emul-sions used in this study was 180 nm, with 50 vol% of the particlesbeing within the range 120-290 nm. The mean molecular weightof pectin (310 kDa) was measured by static light scattering in thisstudy, and the radius of gyration of pectin (rPE = 20 nm) wastaken from a previous study of a number of pectins.24 Thetheoretical predictions are in reasonable agreement with theexperimental measurements, indicating that depletion floccula-tion should occur when the pectin concentration exceeds about0.2 wt % (depending on droplet diameter and concentration).Nevertheless, the theoretical model predicts that the criticalflocculation concentration (CFC) for depletion should decreasesignificantly with increasing droplet concentration (Figure 2). Wedid not observe this phenomenon in our study (Figure 1), possibly

because of the polydispersity of both the emulsion droplets andthe pectin molecules (see below).

The experimental measurements indicated that there was not adistinct transition at a particular pectin concentration from astable emulsion (i.e., homogeneous whitish appearance through-out) to an unstable emulsion (i.e., a white cream layer on top of atransparent serum layer). Instead, there was a range of pectinconcentrations where a cream layer was observed on top of aturbid serum layer, with the turbidity of the serum layer decreas-ing with increasing pectin concentration (Figure 3). This phenom-enon can largely be attributed to the polydispersity of both theemulsion droplets and the pectin molecules. Theory and experi-ment have shown that the CFC for depletion flocculationincreases with decreasing droplet size and increasing polymermolecular weight.25-28 Hence, in a polydisperse emulsion agreater amount of pectin is required to induce depletion floccula-tion (and therefore creaming instability) for the larger dropletsthan the smaller ones. This effect is seen in Figure 2, whichpredicts the dependence ofCdep on droplet concentration and sizefor emulsions similar to the ones used in this study. The threevalues of the particle radius used in these predictions are in therange of the values for the emulsion used in this study: 50 vol%ofthe droplets in the emulsions had radii between 120 and 290 nm.These predictions show that there should actually be a rangeof pectin concentrations where depletion flocculation andcreaming should occur, which is what we observed experimentally(Figure 3). It should be noted that the theoretical calculationsassume that the polymer chains all have the same chain length(and therefore radius of gyration), but in practice, pectin samplesnormally contain a distribution of different chain lengths, whichwill affect the strength of the depletion attraction.29

Stability Map at pH 3.5. The droplet size, electrical charge,and creaming stability of β-lactoglobulin stabilized oil-in-wateremulsions was also measured as a function of droplet concentra-tion (0.5 to 10 wt %) and pectin concentration (0 to 0.5 wt %) atpH 3.5 (Figures 4-6). This information was used to construct an

Figure 1. Influence of droplet and pectin concentrations onthe thickness of the creamed layer in β-lactoglobulin stabilizedoil-in-water emulsions (pH 7). Figure 2. Predicted stability map of oil-in-water emulsions with

different droplets sizes and concentrations due to depletion floccu-lation. The stability map shows phase boundaries plotted aspolymer concentration (C) versus droplet concentration (φ). Thepredictions were made assuming a polymer molecular weight of310 kDa and a radius of gyration of 20 nm.

(23) Blijdenstein, T. B. J.; Veerman, C.; van der Linden, E. Langmuir 2004, 20,4881–4884.(24) Fishman, M. L.; Chau, H. K.; Kolpak, F.; Brady, J. J. Agric. Food Chem.

2001, 49, 4494–4501.

(25) Chanamai, R.; Herrmann, N.; McClements, D. J. J. Phys. D: Appl. Phys.1998, 31, 2956–2963.

(26) Jang, W.; Nikolov, A.; Wasan, D. T. J. Dispersion Sci. Technol. 2004, 25,817–821.

(27) McClements, D. J. Food Hydrocolloids 2000, 14, 173–177.(28) Radford, S. J.; Dickinson, E. Colloids Surf., A 2004, 238, 71–81.(29) Kleshchanok, D.; Tuinier, R.; Lang, P. R. Langmuir 2006, 22, 9121–9128.

DOI: 10.1021/la8006684 6653Langmuir 2009, 25(12), 6649–6657

Cho and McClements Article

experimental stability map (Table 1). At this pH, the electricalcharge on the pectin molecules is negative (ζ=-13.8( 0.7 mV)and on the protein-coated lipid droplets is positive (ζ=+54.6(0.5 mV), and therefore, one would expect the pectin molecules toadsorb to the droplet surfaces because of a strong electrostaticattraction.8,11 The emulsions were stable to creaming in theabsence of pectin, which indicated that the electrical charge onthe protein-coated droplets was high enough to generate a strongelectrostatic repulsion that prevented droplet flocculation.

In the presence of pectin, the emulsions were either stable orunstable to creaming depending on the amount of pectin addedand the droplet concentration. For example, the 3 wt % oil-in-water emulsion was stable to creaming at 0 wt % pectin, unstableat 0.01 to 0.04 wt % pectin, stable from 0.06 to 0.10 wt % pectin,and unstable again at 0.30 and 0.50 wt % pectin (Table 1,Figures 4-6). We attribute this fairly complicated dependenceof emulsion creaming stability on pectin concentration to acombination of bridging and depletion flocculation:I. No Pectin. In the absence of pectin, the emulsions were

stable to creaming because the relatively strong electrostaticrepulsion between the droplets prevented them from coming intoclose proximity and aggregating.II. Low Pectin Concentrations (Bridging Flocculation).

The poor creaming stability observed at relatively low pectinconcentrations (e.g., 0.01 to 0.04 wt % pectin for the 3 wt % oilemulsion) can be attributed to charge neutralization and bridgingflocculation caused by sharing of anionic pectin molecules be-tween two or more cationic protein-coated droplets. The range ofpectin concentrations where this kind of creaming instabilitywas observed increased with increasing droplet concentration(Table 1), as would be expected because the amount of polysac-charide needed to completely cover all the droplet surfaces presentshould increase with increasing droplet concentration (eq 1).III. IntermediatePectinConcentrations (Stable).The good

creaming stability observed at intermediate pectin concentrations

Figure 3. Photograph of a 5 wt % oil-in-water emulsion contain-ing different pectin concentrations after 24 h storage at ambienttemperature (pH 7). Some emulsions separated into an opaquewhite cream layer and a turbid serum layer,whichwas attributed todepletion flocculation.

Figure 4. Influence of pectin concentration on the ζ-potential ofβ-lactoglobulin stabilized 3 wt % oil-in-water emulsions (pH 3.5).The initial d32 of the primary emulsion was 270 nm. The symbolsrepresent the experimental data points, and the line represents thebest-fit to the data using eq 4.

Figure 6. Influence of pectin and droplet concentrations on thethickness of the creamed layer formed in β-lactoglobulin stabilizedoil-in-water emulsions (pH 3.5).

Table 1. Stability Map of β-Lactoglobulin Stabilized Oil-in-Water

Emulsions As a Function of Droplet and Pectin Concentration at

pH 3.5a

droplet concentration (wt %)

pectin concentration (wt%) 0.5 1 3 5 8 10

0.00 S S S S S S0.01 U* U U U* S S0.02 S U U U S U*0.04 S S U U U U*0.06 S S S U U U0.08 S S S U U U0.10 S S S U* U U0.30 U* U* U* S S S0.50 U* U* U* S S S

aThe stability was defined in terms of their creaming stability: S =stable (no visible separation);U*= (visible cream layer, highly turbid oropaque serum layer); U = (visible cream layer, clear or slightly turbidserum layer).

Figure 5. Influence of pectin and droplet concentrations on themean particle diameters of β-lactoglobulin stabilized oil-in-wateremulsions (pH 3.5).

6654 DOI: 10.1021/la8006684 Langmuir 2009, 25(12), 6649–6657

Article Cho and McClements

(e.g., 0.06 to 0.1 wt % for the 3 wt % oil emulsion) canbe attributed to the fact that the lipid droplets were completelycoated with pectin molecules, and there was a strong electro-static and steric repulsion between the coated droplets.Thus, stable emulsions can be formed in this range of pectinconcentrations.IV. Relatively High Pectin Concentrations (Depletion

Flocculation). The poor creaming stability observed at relativelyhigh pectin concentrations (e.g., 0.3 and 0.5 wt % for the 3 wt %oil emulsion) can be attributed to depletion flocculation caused bythe presence of a large quantity of nonadsorbed pectin in theaqueous phase. No creaming instability was observed in thepH 3.5 emulsions with the highest droplet concentrations (5 to10 wt %) at relatively high pectin concentrations, even thoughcreaming instability was observed in pH 7.0 emulsions with thesame droplet concentrations. The most likely reason for thisobservation is that there was insufficient nonadsorbed pectin topromote depletion flocculation in the pH 3.5 emulsions, becausean appreciable fraction was absorbed to the droplet surfaces(rather than free in the aqueous phase). Alternatively, the dropletconcentration in these emulsions may have been sufficient toprevent droplet creaming due to network formation.

Evidence for adsorption of pectin molecules to the protein-coated droplet surfaces at pH 3.5 was obtained from ζ-potentialversus pectin concentration measurements (Figure 4). In theabsence of pectin, the protein-coated droplets had a ζ-potentialof around +55 mV because the adsorbed proteins were belowtheir isoelectric point (pI = 5). The ζ-potential of the dropletsbecame increasingly less positive and then became negative as thepectin concentration was increased, eventually reaching a plateauvalue at around 0.1 wt% pectin. These data indicate that anionicpectin molecules adsorbed to cationic protein-coated dropletsuntil the droplet surfaces became saturated with pectin. Thecritical polysaccharide concentration where the droplet surfacesbecame saturated with pectin was established by modeling theζ-potential versus polysaccharide concentration curves using anempirical equation to model the data:11

ζðcÞ-ζSatζ0 -ζSat

¼ exp -c

3cSat

� �ð4Þ

where ζ(c) is the ζ-potential of the emulsion droplets atpolysaccharide concentration c, ζ0 is the ζ-potential in the absenceof polysaccharide, ζSat is the ζ-potential when the droplets aresaturatedwith polysaccharide, and cSat is theminimumamount ofpolysaccharide required to completely cover the droplet surfaces.The binding of a polysaccharide to the droplet surfaces cantherefore be characterized by ΔζSat (=ζ0 - ζSat) and cSat. ValuesforΔζSat and cSat were calculated by fitting the above equation tothe experimental ζ-potential versus pectin concentration curve forthe 3 wt % oil-in-water emulsion: ΔζSat = -71 mV and cSat =0.069 wt % pectin. The good agreement between the model andexperiment is shown in Figure 4. The saturation concentrationdepends on the size and concentration of emulsion droplets used(since this determines the exposed surface area), and therefore it isuseful to calculate a more system-independent quantity for theamount of adsorbed pectin. The surface load at saturation can becalculated using the following expression:

ΓSat ¼ CSatd32

6φð5Þ

Here, CSat is defined as the mass of polyelectrolyte (pectin)adsorbed to the surface of the droplets per unit volume of

emulsion (kg m-3), d32 is the volume-surface mean dropletdiameter, and φ is the droplet volume fraction. For thisstudy: d32 = 0.27 μm, φ = 0.03 (3%), and CSat ≈ 0.69 kg m-3

(0.069 wt %); hence, ΓSat = 1.03 mg m-2. This value is in goodagreement with those found for the surface loads of other anionicpolysaccharides adsorbed to the surfaces of protein-coated oildroplets: ΓSat ∼ 1.3, 1.6, 2.1, and 5.1 mg m-2 for alginate, pectin,carrageenan, and gum arabic adsorbed to β-lactoglobulin coatedoil droplets, respectively.11-13

The influence of droplet concentration on particle aggregationand creaming stability of the emulsions was measured (Table 1,Figures 5 and 6). The increase in mean particle diameter andcreaming instability due to bridging flocculation at intermediatepectin concentrations are clearly seen in Figures 5 and 6. As thedroplet concentration in the emulsions increased, the range ofpectin concentrations where bridging flocculation was observedincreased, which can be attributed to the increase in the totaldroplet surface area that needs to be covered by pectin (eq 1).Using the information for the saturation concentration givenabove, we calculated CSat = 0.012, 0.023, 0.069, 0.12, 0.18, and0.23 wt % pectin for the 0.5, 1, 3, 5, 8, and 10 wt % emulsions.This would account for the fact that emulsions stable to bridgingflocculation could only be formed when the pectin concen-tration exceeded these critical values, e.g., 0.01, 0.02, 0.06,>0.1, >0.1, and >0.1 wt % pectin for the 0.5, 1, 3, 5, 8, and10 wt % emulsions, respectively (Table 1, Figures 5 and 6).At higher droplet concentrations, the mean particle diameter(d32 < 0.4 μm) of the emulsions was fairly similar to the initialvalues, which indicated that the emulsions were stable to dropletcoalescence, and that any flocs formed in the emulsions weredisrupted when the emulsions were diluted for particle sizemeasurements. Nevertheless, the formation of a thin creamedlayer was observed on top of some of the emulsions at higherpectin concentrations (Table 1), which indicated that flocculationdid occur. This kind of droplet aggregation can be attributed todepletion flocculation due to the presence of relatively highconcentrations of nonadsorbed pectin in the continuous phase,as discussed in the previous section.

As mentioned above, at pH 3.5 the droplets are positivelycharged while the pectin is negatively charged, and so, one wouldexpect that both bridging flocculation (at intermediate pectinconcentrations) and depletion flocculation (at high pectin con-centrations) to occur. The theoretical approach described earlierwas used to construct a stability map based on calculations of theCsat, Cads, and Cdep values given in eqs 1-3 and the followingexperimentally determined parameters: r=180 nm, rPE=20 nm,

Figure 7. Predicted stability map of oil-in-water emulsions withdifferent droplet and pectin concentrations. The stability mapshows phase boundaries plotted as polymer concentration (C)versus droplet concentration (φ). The predictions were madeassuming a polymer molecular weight of 310 kDa and a radius ofgyration of 20 nm, and a droplet radius of 180 nm.

DOI: 10.1021/la8006684 6655Langmuir 2009, 25(12), 6649–6657

Cho and McClements Article

M=310 kgmol-1, and ΓSat= 1.03� 10-6 kgm-2 (Figure 7). hetheoretical stability map predicts that stable emulsions can onlybe produced over a narrow range of droplet and pectin concen-trations. In particular, the stability map suggests that stablepectin-coated emulsion droplets can only be produced at dropletconcentrations less than about 3 wt %, with the range of pectinconcentrations where stable emulsions can be produced increas-ing with decreasing droplet concentration (Figure 7). For exam-ple, for 1 wt % emulsions it should only be possible to createstable pectin-coated emulsion droplets at pectin concentrationsbetween about 0.15 and 0.26 wt %. Below 0.02 wt % pectin, onewould expect to get irreversible bridging flocculation becausetherewas insufficient pectin to cover the droplet surfaces; between0.02 and 0.14 wt %, one might expect to get reversible bridgingflocculation because the pectin did not adsorb rapidly enoughto the droplet surfaces (even though there was sufficient present tosaturate the surfaces); and above 0.26 wt %, one would expect toget depletion flocculation.

The theoretical predictions are in fairly good qualitativeagreement with the experimental measurements (Table 1,Figures 5 and 6). For example, the experimental stability mapshows: (i) bridging flocculation occurs over a range of inter-mediate pectin concentrations; (ii) depletion flocculation occursat high pectin concentrations in some of the samples; (iii) therange of pectin concentrations where emulsions stable to bridgingflocculation can be formed increases with decreasing dropletconcentration. The theoretical stability maps therefore seem toprovide a good qualitative prediction of the stability of theemulsions, which may prove useful when designing optimumconditions for coating lipid droplets with polyelectrolytes. Never-theless, there are quantitative differences between the experi-mental measurements and theoretical predictions, which can beattributed to limitations in the assumptions underlying the theory,as well as the effects of droplet and pectin polydispersity.

Microscopic images of the emulsions indicated that differentkinds of aggregate structures were formed in the emulsions in theintermediate pectin range (bridging flocculation) and the highpectin range (depletion flocculation) (Figure 8). At intermediatepectin concentrations, all of the droplets appeared to be present inthe form of large aggregates, which appeared to contain somecoalesced droplets. At high pectin concentrations, a smallerfraction of the droplets appeared to be aggregated, and theaggregates formed were much smaller. These differences inaggregate structure can be attributed to the different mechanismsresponsible for droplet flocculation: bridging flocculation (strong,irreversible) and depletion flocculation (weak, reversible).

In this study, pectin molecules were mixed with protein-coateddroplets at pH7where they are both negatively charged, and thenthe pHwas adjusted to pH3.5 to promote pectin adsorption. Thisapproach was used since previous studies in our laboratory haveshown that less bridging flocculation occurs than when thepolysaccharide and protein-coated droplets are mixed togetherat a pHwhere they have opposite charges.8 Themost likely reasonfor the reduction in bridging flocculation when pectin adsorptionis induced by a pH reduction is that the pectin moleculessurrounding the protein-coated droplets are already evenly dis-tributed throughout the aqueous phase. On the other hand, whendroplets and pectin are mixed together at a pH where they haveopposite charges then there will be local variations in droplet andpectin concentrations, aswell as additional shearing effects, whichmay impact the rate of droplet-droplet and droplet-pectincollisions. This difference highlights the potential impact ofpreparation conditions on the calculation of stability maps. Itwould therefore be useful to extend the above theory to take intoaccount the effects of shearing on the rate of droplet-droplet anddroplet-pectin collisions, as well as on the potential breakup offlocs.Influence of Initial Droplet Size. The expressions for CSat,

CAds, and CDep given above (eqs 1-3) indicate that the stabilitymap of an emulsion containing oppositely charged droplets andpolymer molecules should depend on the mean droplet diameter.We therefore investigated the influence of initial droplet diameteron the properties of emulsions as a function of pectin concentra-tion (Figures 9 and 10). Three 1 wt % oil-in-water emulsions(pH3.5)withdifferentmeandroplet diameters (d32=0.26( 0.01,0.47 ( 0.02, 0.62 ( 0.04 μm) and pectin concentrations (0 to0.5 wt %) were prepared using the approach described earlier.

Initially, we calculated the dependence ofCSat, CAds, and CDep

on droplet diameter for a 1%oil-in-water emulsion using eqs 1-3and the values of themeanmolecularweight (310 kDa) and radiusof gyration (rPE= 20 nm) of pectin given earlier (Figure 11). Thevalue of CSat should be inversely proportional to d32. Never-theless, we did not observe an appreciable difference in the pectindependence of the ζ-potential of the three emulsionswith differentmean droplet diameters (data not shown). Indeed, the surfaceloads of the pectin molecules calculated from the ζ-potentialversus pectin profiles using eqs 4 and 5 were ΓSat = 1.1, 1.6, and2.0 mg m-2 for the emulsions with d32= 0.26, 0.47, and 0.62 μm,respectively. These values suggest that the amount of pectinadsorbed to the droplet surfaces decreased with decreasing mean

Figure 8. Photographs and microscopy images of 1 wt % oil-in-water emulsions containing different pectin concentrations (shownbeneath each image) at pH 3.5. The microscopy images were takenafter 7 days storage at a magnification of 100� (the scale bar isshown in the 0% pectin figure). The creaming measurements weremadeafter emulsionswere stored inglass test tubes (15mminternaldiameter, 125 mm height) for 24 h.

Figure 9. Influence of pectin concentration on mean particlediameter of β-lactoglobulin stabilized 1 wt % oil-in-water emul-sions with different mean droplet diameters (pH 3.5).

6656 DOI: 10.1021/la8006684 Langmuir 2009, 25(12), 6649–6657

Article Cho and McClements

droplet diameter. This phenomenon may be due to the effectsof droplet curvature on pectin adsorption, since theoreticalcalculations indicate that adsorption of a charged polyelectrolyteto an oppositely charged surface becomes less favorable as thecurvature of the surface increases.30 Alternatively, it may be dueto differences in the type of pectin molecules that adsorb to thedroplet surfaces, e.g., the molecular weight and packing of pectinmolecules may be different on small and large droplets. Furtherresearch is needed to identify the origin of the effects of dropletsize on pectin surface load.

There were some differences between the stability of the threeemulsions to aggregation and creaming (Figures 9 and 10).Extensive droplet aggregation and creaming instability wereobserved in all three emulsions at 0.01 and 0.02 wt % pectinconcentrations, which can be attributed to bridging flocculation.Nevertheless, the extent of droplet aggregation was greatest in theemulsion with the smallest droplets (d32= 0.26 μm) in this pectinrange (Figure 9). This effect can be accounted for by the fact thatCAds is expected to increase rapidly with decreasing droplet size(Figure 11), which occurs because the time between dropletcollisions decreases more rapidly than the time required for thedroplet surfaces to be saturated with pectin, leading to morebridging flocculation.Distinguishing Irreversible and Reversible Bridging Floc-

culation. The theoretical model predicts that bridging floccula-tion may have two potential origins: (i) insufficient pectin presentto cover all of the droplet surfaces present; (ii) relatively slowadsorption of pectin molecules to droplet surfaces compared to

the time between droplet collisions. In the first case, bridgingflocculation should be irreversible since therewill never be enoughpectin to entirely cover the droplet surfaces. In the second case, itmay be possible to disrupt flocs once they have formed byapplying mechanical agitation, since there should be sufficientpectin present in the system to cover all the droplet surfaces.Previous studies have shown that mechanical treatment, such ashomogenization, high-speed blending, or high-intensity ultra-sound, can be used to disrupt the flocs formed during preparationof polyelectrolyte-coated emulsion droplets under some circum-stances.19,20,31 In these studies, it was suggested that this improve-ment in emulsion stability was due to the presence of sufficientpolymer to adsorb to the droplet surfaces and provide a strongrepulsion between the droplets after the flocs had been disrupted.The data presented in the previous section showed that there wasonly a narrow range of pectin concentrations where stable pectin-coated emulsion droplets could be produced at pH 3.5, which wasattributed to bridging flocculation caused by the twomechanismsdescribed above.We postulated that we might be able to broaden

Figure 10. Influence of pectin concentration on appearance ofβ-lactoglobulin stabilized 1 wt % oil-in-water emulsions withdifferent mean droplet diameters (pH 3.5).

Figure 11. Predicted dependence of critical flocculation concen-trations (CSat, CDep, and CAds) on droplet size for oil-in-wateremulsions. The predictions were made assuming a polymer mole-cular weight of 310 kDa and a radius of gyration of 20 nm, and adroplet concentration of 1%.

Figure 12. Influence of sonication on the mean particle diameterand appearance of β-lactoglobulin stabilized 1 wt % oil-in-wateremulsions containing different levels of pectin (pH 3.5).

(30) Stoll, S.; Chodanowski, P. Macromolecules 2002, 35, 9556–9562.(31) Ogawa, S.; Decker, E. A.;McClements, D. J. J. Agric. Food Chem. 2003, 51,

2806–2812.

DOI: 10.1021/la8006684 6657Langmuir 2009, 25(12), 6649–6657

Cho and McClements Article

this range by applying ultrasound treatment to break down anyflocs formeddue to slowpectin adsorptionduring the preparationof the emulsions.

The influence of sonication on floc disruption in 1 wt% oil-in-water emulsions containing different levels of pectin is shown inFigure 12. In the absence of sonication, extensive droplet floccu-lation occurred at 0.01 and 0.02 wt% pectin, as demonstrated byan increase in mean particle diameter and creaming index. Afterthe application of sonication, there appeared to be a slightimprovement in stability of the emulsions to droplet aggregationat 0.02 wt % pectin, although a visible creaming layer was stillobserved. This result suggests that applying mechanical agitationto emulsions may be able to improve their stability to bridgingflocculation under certain conditions, but not under conditionswhere there is insufficient pectin to saturate the droplet surfaces.

Conclusions

This study has shown that a simple theoretical model can beused to predict the stability of protein-stabilized oil-in-wateremulsions containing charged polysaccharides. Experimentalstability maps of the influence of droplet and pectin concentra-tions on emulsion stability were established at pH 7 (where pectindoes not adsorb to the droplet surfaces) and at pH 3.5 (wherepectin does adsorb). At pH 7, droplet flocculation and creamingwere observed when the pectin concentration exceeded a parti-cular level, which was attributed to depletion flocculation. AtpH 3.5, the stability of the emulsions was much more complex,going from stable, to unstable, to stable, to unstable withincreasing pectin concentration. In these systems, the instabilityat low pectin concentrations was attributed to bridging floccula-tion, whereas the stability at high pectin concentrations wasattributed to depletion flocculation. At intermediate pectin con-centrations stable emulsions could be formed, which consisted of

lipid droplets coatedwith a pectin layer.Weusedone specific kindof polysaccharide in this study (citrus pectin). Pectin varies inmolecular structure depending on its biological origin andextraction procedure and so it would be useful to examine theinfluence of pectin’s molecular properties (e.g., molecular weight,charge density, branching, and hydrophobicity) on the nature ofthe stability maps in future studies. In addition, our experimentswere carried out at relatively low ionic strengths (no added salt).Ionic strength affects the range and magnitude of electrostaticinteractions between electrically charged lipid droplets and poly-mer molecules, as well as the conformation of charged polymermolecules. Consequently, ionic strength is likely to affect thestability maps of systems containing charged lipid droplets andpolymermolecules. The impact of ionic strength on stabilitymapswould be an interesting area for future research.

The simple theoretical model used in this work was able toqualitatively predict this behavior andmay therefore prove usefulfor optimizing the preparation conditions for forming stableemulsions containing lipid droplets coated by functional poly-electrolytes. Nevertheless, there are a number of areas where thecurrentmodel could be extended and tested in order to improve itsrange of applicability. For example, it would be useful to includethe effects of droplet and polymer polydispersity (which affect thecritical flocculation concentrations) and shearing (which affectdroplet-droplet and droplet-pectin collisions, as well as flocdisruption).

Acknowledgment. This material is based upon work sup-ported by the Cooperative State Research, Extension, EducationService, United State Department of Agriculture, MassachusettsAgricultural Experiment Station (Project No. 831) and UnitedStates Department of Agriculture, CREES, NRI Grants (AwardNumbers 2005-01357, 2008-02191 and 2008-01368).