Journal of Colloid and Interface Science 389 (2013) 1–9

Contents lists available at SciVerse ScienceDirect

Journal of Colloid and Interface Science

www.elsevier .com/locate / jc is

Editor’s Choice

Effects of emulsifier hydrophile–lipophile balance and emulsifier concentrationon the distributions of gallic acid, propyl gallate, and a-tocopherol in corn oilemulsions

Sonia Losada-Barreiro, Verónica Sánchez-Paz, Carlos Bravo-Díaz ⇑Universidad de Vigo, Fac. Química, Dpto Química Física, 36200 Vigo, Spain

a r t i c l e i n f o

Article history:Received 11 June 2012Accepted 11 July 2012Available online 25 July 2012

Dedicated to Prof. Luis Carballeira-Ocaña onthe occasion of his retirement.

Keywords:Oil-in-water emulsionsAntioxidantsPseudophase modelPartition constants

0021-9797/$ - see front matter � 2012 Elsevier Inc. Ahttp://dx.doi.org/10.1016/j.jcis.2012.07.036

⇑ Corresponding author. Fax: +34 986 812556.E-mail address: cbravo@uvigo.es (C. Bravo-Díaz).

a b s t r a c t

We evaluated the effects of the hydrophile–lipophile balance (HLB) and emulsifier concentration on thedistribution of the antioxidants gallic acid (GA), propyl gallate (PG), and a-tocopherol (TOC) between theaqueous, interfacial, and oil regions of food-grade emulsions composed of stripped corn oil, acidic water,and a mixture of the non-ionic surfactants Tween 20, 40, 80, and Span 20. The distribution of the antiox-idants (AOs) is described by two partition constants, that between the oil-interfacial region, PI

O, and thatbetween the aqueous and interfacial region, PI

W, of the emulsions. The partition constants were deter-mined from the kinetic analyses of the variation in the observed rate constant, kobs, for the reactionbetween the AOs and the hydrophobic 4-hexadecylbenzenediazonium ions, 16-ArNþ2 , with the emulsifiervolume fraction. The effects of emulsifier HLB on the second-order rate constants in the interfacial regionkI were also evaluated for each antioxidant. Results show that an increase in emulsifier concentrationpromotes the incorporation of AOs to the interfacial region of the emulsions, so that at surfactant volumefractions of 0.04, more than 90% of GA and PG and more than 50% of TOC are located in that region. Adecrease in the HLB favors the incorporation of PG and TOC to the interfacial region of the emulsionsbut has a negligible effect on the fraction of GA in that region. The %AOs in the interfacial region of theemulsions does not correlate with the polarity of the antioxidant, so that GA and PG are predominantlylocated in the aqueous–interfacial regions of the emulsion rather that in the oil droplet interior; mean-while, TOC is mostly located in the oil-interfacial regions. Results should aid to understand how antiox-idants are distributed in food-grade emulsions and their relative efficiency in inhibiting lipid oxidation.

� 2012 Elsevier Inc. All rights reserved.

1. Introduction because of the existing differences in the physical nature of the

Addition of antioxidants (AOs) to food products is one of severalstrategies commonly used in the food industry to retard or inhibitlipid oxidation, which is a chemical reaction major that may occurat any stage of food processing and storage, causing a major food-quality deterioration [1–3]. Thus, knowledge concerning the effi-ciency or activity of AOs in inhibiting lipid oxidation in food emul-sions is a subject of paramount importance for developing specificstrategies to prepare healthier and nutritional foods with longershelf-life [3–7].

The rate at which lipid oxidation takes place depends on severalfactors such as the structure of the lipids, the temperature, the con-centration of pro-oxidants, and antioxidants [8,9]. The first qualita-tive theory attempting to predict the antioxidant efficiency, the‘‘polar paradox’’, was formulated by Porter [10,11] as early as1980, who stated that AOs added to oil-in-water emulsions exhibitdifferent effectiveness compared to when included in bulk oils

ll rights reserved.

two systems. Subsequently, different researchers [8,12–15] con-firmed that lipophilic antioxidants inhibit lipid oxidation more effi-ciently than hydrophilic antioxidants in emulsions because theyare primarily located in the interfacial regions of emulsions. In con-trast, in bulk oils, polar (hydrophilic) antioxidants would be ad-sorbed in the air–oil interface and would be more efficient ininhibiting lipid oxidation than the lipophilic counterparts.

The validity of the so-called polar paradox has been, however,questioned recently by several researchers [16–20] because theyfound that the antioxidant efficiency of hydrophobic antioxidantswith alkyl chains ranging 12–20 carbon atoms is lower than thosewith alkyl chains of less than 12 carbon atoms, that is, more polarantioxidants, which reverse the predictions of the polar paradox,highlighting the importance of the location of the antioxidantwithin the system, that is, its distribution between the different re-gions of the emulsion [17,18,20,21].

It is, however, currently recognized that the antioxidant effi-ciency depends not only on the polarity (i.e., location) of the anti-oxidant in the different regions of the emulsion but also on anumber of other factors including the chemical properties of the

Scheme 1. Basic representation of the different regions of binary and emulsifiedsystems showing the distribution of an antioxidant AO and the partition constantsneeded to describe its distribution. Parenthesis indicates concentrations in molesper liter of the volume of a particular region.

Scheme 2. Basic illustration of the interfacial region of an emulsion showing the oil,water, and interfacial regions. UI indicates the volume fraction of a region, PI

W andPI

O are the partition constants describing the distribution of an antioxidant AObetween the aqueous, oil and interfacial regions, and kI is the interfacial rateconstant for reaction between 16-ArNþ2 and the GA in the interfacial region.

2 S. Losada-Barreiro et al. / Journal of Colloid and Interface Science 389 (2013) 1–9

AOs, the emulsifier and oil types, and the environmental conditions(acidity, temperature, electrolyte concentration, etc.) [12,21,22].All these parameters have an effect on the AO distribution in theemulsion and thus somehow control the percentage of AO locatedin a particular region. As a consequence, in the last years, there hasbeen an upsurge in the efforts to understand how different param-eters control the antioxidant distribution in emulsified systemsand how the antioxidant partitioning affects the lipid oxidationchemistry [17,18,20,23].

The distribution of an antioxidant (and that of any other com-pound) in binary oil–water systems in the absence of emulsifier,Scheme 1, is usually described by the partition constant betweenthe oil and water phases, PO

W, which can be calculated from the ra-tio of the analyte concentrations in each phase [24–26], Eq. (1),where VW and VO are the aqueous and oil region volumes, respec-tively, and concentrations in them are represented withparentheses.

POW ¼

ðAOOÞðAOWÞ

¼ ð%AOOÞð%AOWÞ

VW

VO¼

PIW

� �

PIO

� � ð1Þ

Addition of emulsifiers to prepare kinetically stable emulsionscreates a new region in the system, the interfacial region, Scheme 1,and the AOs may distribute between the oil, interfacial and aque-ous regions. Consequently, two partition constants, PI

O and PIW,

are now necessary to describe the distribution of the antioxidantbetween the oil and interfacial regions PI

O and between the aque-ous and interfacial regions (PI

W). Note that the partition constant(PO

W) is equal to the ratio PIW=PI

O as indicated in Eq. (1), where PIW

and PIO are PI

W ¼ ðAOIÞ=ðAOWÞ and PIO ¼ ðAOIÞ=ðAOOÞ, Scheme 1.

The analytical methods [25] commonly employed to determinePO

W in binary systems (i.e., the analyte concentrations in eachphase) cannot be employed in emulsified systems because it isphysically impossible to isolate the interfacial region from theaqueous or the oil ones. Schwarz and coworkers [12,14,22,27–30]attempted to approach the problem primarily by isolating the oiland water regions of AO containing emulsions by determiningthe AO concentration in each region using different analyticalmethods. However, this approach cannot provide information onthe percentage of AO in the interfacial regions because the methodrequires breaking emulsion prior to any analytical determination.It is worth noting that the lack of reliable methods for determiningtheir distributions hindered, in part, the development of reliablemethods to establish correlations between distribution in the sys-tem and their efficiency [3,18,20].

Thus, determining the AO distributions in emulsified systemsneeds to be done in the intact emulsion and new strategies toassess the antioxidant distribution in model emulsions were

developed. We proceeded in a different way [31–36]. Rather thanfocusing in developing analytical methods to measure analyte con-centrations in the different regions of the emulsion, we focused inbuilding up methods to determine the partition constants of theantioxidants in the intact emulsion. The choice was a kinetic meth-od that exploits the reaction between the antioxidant and a hydro-phobic arenediazonium ion, Scheme 2. Because emulsions areopaque, we also developed protocols to monitor the reaction inthe intact emulsion to calculate the corresponding observed rateconstants kobs. The kobs values depend on both concentration andmedium effects, and their relative contribution can be determinedby employing the pseudophase kinetic model for chemical reactiv-ity as described elsewhere [32,34,35,37].

In previous work, we used this methodology to investigate thedistribution of a number of antioxidants and to evaluate the rela-tive effects of external parameters (temperature and acidity) ontheir distribution in the different regions of the food emulsions[31–36]. Here, we expand our previous investigations and reporton the effects of the emulsifier concentration and of the natureof the emulsifier, characterized by the hydrophile–Lipophile Bal-ance (HLB), on the distribution of three selected antioxidants, gallicacid (GA), propyl gallate (PG), and a-tocoppherol (TOC). These anti-oxidants constitute a set of very efficient antioxidants widely em-ployed in the food industry [38–42] ranging from the veryhydrophilic, oil insoluble, GA to the very hydrophobic, water insol-uble, TOC. PG is both oil and water soluble.

The HLB of the emulsifier has an enormous impact on the nat-ure and stability of emulsions [43–45], but its effect on the antiox-idant distributions has not been determined up to date. The HLB isan empirical number proposed by Griffin [46] and Davies [47] as away of predicting the emulsion type from the emulsifier molecularcomposition [43,48]. Emulsifiers (or blends of emulsifiers) withHLB numbers ranging 3–6 are usually employed to preparewater-in-oil emulsions, meanwhile those with HLB 8–18 are em-ployed to prepare oil-in-water emulsions. Because most of thecommon food products (e.g., beverages, dressings, deserts, soups,etc.), exist as oil-in-water emulsions [3,44,48,49], we preparedfood-grade oil-in-water emulsions by mixing stripped corn oil,

Table 1Proportions (%) of the emulsifiers employed to prepare emulsions of different HLB.Reported density values (average of three runs) were determined at ambienttemperature by weighing a known volume of the surfactant blend.

Tween 20 Tween 40 Tween 80 Span 20 HLB q (g mL�1)

100 – – – 16.7 1.095– 100 – – 15.6 1.085– – 100 – 15.0 1.07675 – – 25 14.7 1.07050.7 – – 49.3 12.7 1.06725.2 – – 74.8 10.6 1.048– – – 100 8.6 1.032

S. Losada-Barreiro et al. / Journal of Colloid and Interface Science 389 (2013) 1–9 3

acidic water, and different amounts of the non-ionic surfactantssorbitan fatty acid esters (Spans) and the polyoxyethylene sorbitanfatty acid esters (Tweens) of HLB values ranging 8–17, Table 1.

Corn oil has traditionally been considered a premium vegetableoil widely used for cooking, margarine, spreads, and salad dress-ings. It is primarily a mixture of monounsaturated and polyunsat-urated fatty acids, containing high levels of linolenic acid (60% molfraction), relatively low levels (<15%) of saturated fatty acids andvery low levels of linolenic acid, which is susceptible to oxidation[50]. Tweens and Spans are relatively small molecule surfactantscommonly used in food emulsions [44] and in pharmaceuticalapplications [51] to stabilize water-in-oil, oil-in-water, andwater-in-oil-in-water double emulsions. The chemical structuresof some Tweens, Spans, and of the main components of the cornoil are shown in Scheme 3.

2. Experimental section

2.1. Materials

Gallic acid (GA, Riedel de Haën), propyl gallate (PG, Fluka), a-tocopherol (TOC, Sigma), and the emulsifiers Tween 20, Tween

Scheme 3. Structures of the 16-ArNþ2 ions, the antioxidants GA, PG and TOC, the non-ion

40, Tween 80, and Span 20 (Fluka) were of the highest purity avail-able and used as received. Aqueous buffered solutions were pre-pared from acetic acid – sodium acetate (final concentration0.04 M) and their pH was determined from potentiometric mea-surements. Stripped corn oil (Acros Organics) was used as receivedand kept at low temperature in the dark to minimize lipidperoxidation.

4-Hexadecylbenzenediazonium tetrafluoroborate, 16-ArN2BF4,was prepared in high yield and purity under nonaqueous condi-tions from commercial 4-hexadecylaniline (Aldrich, 97%) by diazo-tization following a published method [52,53] and stored in thedark at low temperature to minimize its decomposition.

All aqueous solutions were prepared with Milli-Q grade water.The acidity of aqueous phase in the emulsions was controlled usingacetic acid/acetate buffer (0.04 M). Solutions of the coupling re-agent N-(1-naphthyl)ethylenediamine (NED (Aldrich) were pre-pared in a 50:50 (v/v) BuOH/EtOH mixture to give[NED] = 0.02 M. The chemical structures of 16-ArNþ2 ions and theantioxidants are shown in Scheme 3.

2.2. Methods

2.2.1. Emulsion preparationEmulsions of 1:9 or 1:4 (v:v) oil to water (Vtotal = 10 mL) and dif-

ferent HLB were prepared by mixing the necessary amounts ofstripped corn oil, acidic water (acetic acid buffer, 0.04 M, measuredpH = 3.7), and different proportions of weighed amounts of thenon-ionic surfactants to obtained the required HLB. The HLBvalue was calculated by employing the following equation,HLBblend = X � HLBA + (1 � X) � HLBB, where X is the weight fractionof the emulsifier. The calculated values are indicated in Table 1.The volume fraction of surfactant, UI, defined hereafter asUI = Vsurf/Vemulsion was varied from UI = 0.005 up to UI = 0.04. Therequired amounts of the antioxidants GA and PG were dissolvedin the bulk water meanwhile those of TOC were dissolved in the

ic emulsifiers Tween 20, Tween 80, Span 20, and of the main components of corn oil.

Scheme 4. Coupling of 16-ArNþ2 with N-(1-naphthyl)ethylenediamine, NED, to yield a stable azo dye whose concentration is determined spectrometrically at k ¼ 572 nm.

4 S. Losada-Barreiro et al. / Journal of Colloid and Interface Science 389 (2013) 1–9

stripped corn oil, so that final concentrations of the antioxidant inthe emulsion were [AO] � 4 � 10�3 M. The mixture was stirredwith a high-speed rotor (Polytron PT 1600 E) for 1 min andtransferred to a continuously stirred thermostated cell. No phaseseparation was observed within 4–6 h, a much longer time periodthan required for complete reaction between AO and 16-ArNþ2 ions.

2.3. Determining kobs values for the reaction between 16-ArNþ2 and theantioxidants in emulsions of different HLB. Derivatization method: azodye formation

The reaction between the antioxidant and the probe moleculewas initiated by adding an aliquot (16 lL) of a 0.17 M stock16-ArNþ2 solution in acetonitrile to a thermostated emulsion. Ki-netic data were obtained by employing our derivatization method(azo dye formation) as described in detail elsewhere [33]. Thismethodology exploits the rapid reaction of ArNþ2 ions with a suit-able coupling agent such as N-(1-naphthyl)ethylenediamine,NED, to yielding an stable azo dye, Scheme 4, that is diluted witha 50:50 (v:v) BuOH/EtOH mixture to yield an optically transparent,homogeneous solution whose absorbance is measured spectromet-rically [33–35]. Control experiments showed that the absorbanceof the azo dye at k ¼ 572nm follows Beers law and is a linear func-tion of the concentration of 16-ArNþ2 . Thus, the decrease in theabsorbance of the azo dye with time can be used to determine kobs.

The reaction between 16-ArNþ2 and NED was monitored for atleast 2–3t1/2, with typical correlation coefficients >0.995. Fig. 1shows typical kinetic plots of absorbance versus time atk ¼ 572 nm for the reaction between GA and PG in an emulsionof HLB = 12.7. Fig. 1 also displays the corresponding ln versus timeplots based on the integrated rate equation for first-order

Fig. 1. Typical kinetic curves for the reaction of 16-ArNþ2 ions with GA (A) and PG (B) cabuffer, 0.04 M, pH = 3.7), and a blend of Tween 20 and Span 20 (HLB = 12.7). [AO] � 4.2

reactions. Duplicate or triplicate experiments gave kobs valueswithin ±7–9%.

2.4. Relationships between the partition and the observed rateconstants (kobs) in emulsions: pseudophase kinetic model

The protocols for applying the pseudophase kinetic model toemulsions have been described in detail [32,33], and only the rel-evant parts are summarized here. As in pseudophase kinetic mod-els, antioxidants and other reactive components partitionthemselves among the different regions according to their solubil-ity characteristics and surface activity, which are in turn deter-mined by their chemical structures and polarity.

The distribution of antioxidants or other reactive components ina stirred, fluid, emulsion is assumed to be in dynamic equilibriumbecause their transfer rates (and those of other components) be-tween oil, interfacial and aqueous regions, Scheme 1, are much fas-ter than its rate of reaction with 16-ArNþ2 .

When oil-in-water emulsions are stabilized with small mole-cule surfactants such as Tweens or Spans, they do absorb ontothe emulsion droplet surface [43]. If the concentration of the sur-factants is beyond the interfacial layer saturation threshold, theremaining surfactant partitions into the continuous phase wheresurfactant micelles may form if the surfactant concentration ex-ceeds its critical micelle concentration [44]. Thus, if micelles areformed, then they should solubilize a portion of the oil (i.e., swol-len or reverse micelles) and their interfacial regions should haveproperties as a reaction medium that are very similar to that ofthe interfacial region of the large emulsion droplets because theirinterfacial region contains the same molecules and their composi-tion is similar. That is, the rate constant for reaction in micelles or

rried out in an emulsion composed of stripped corn oil, acidic water (acetic acetate� 10�3 M, [16-ArNþ2 ] � 2.7 � 10�4 M, T = 25 �C.

Table 2Values of PO

W (for binary corn oil–water mixtures in the absence of emulsifier) and ofPI

W, PIO, and kI for GA, PG and TOC in emulsions of different HLB.

Antioxidant POW

HLB PIW PI

O102kI (M�1 s�1)

GA 0.03 16.7 123 – 5.315.6 133 – 5.315.0 148 – 5.712.7 150 – 1.78.05 141 – 2.1

PG 0.84 16.7 176 209 19.715.0 203 242 14.012.7 362 431 10.510.6 522 622 8.5

TOC – 16.7 – 11.3 1715.6 – 10.2 1415.0 – 8.4 1814.7 – 6.1 238.6 – 6.1 5

S. Losada-Barreiro et al. / Journal of Colloid and Interface Science 389 (2013) 1–9 5

in emulsion droplets should be very similar, and because the distri-bution of reactants between all aggregates is in dynamic equilib-rium, the measured rate constant kobs is the average value forreaction in all droplets.

Under these conditions, the observed rate, v, for a bimolecularreaction in an emulsion is the sum of the rates in each region ofthe macroemulsion [31–34]. The 16-ArNþ2 ions are insoluble inthe water region because of its hydrophobicity, a 16 carbon tailand an aromatic ring, and also in the oil region because of its cat-ionic headgroup. Thus, their concentrations in the aqueous and oilregions are effectively zero and there is no reaction between theantioxidant and 16-ArNþ2 in these regions as illustrated in Scheme 2.As a consequence, the interfacial and total concentrations of16-ArNþ2 are equal and v (and kobs) are directly proportional to theconcentration of the antioxidant in the interfacial region, equation:

v ¼ k2½16-ArNþ2 � ¼ kobs½16-ArNþ2T� ¼ kIð16-ArNþ2IÞðAOIÞUI ð2Þ

where subscript T stands for the stoichiometric or total amount inthe emulsion; subscript I represents the interfacial region; kobs

(s�1), k2 (M�1 s�1) and kI (M�1 s�1) are, respectively, the observedfirst and second order rate constants, and the second order rate con-stant in the interfacial region. Square brackets, [ ], denote concen-tration in moles per liter of total emulsion volume; parentheses,(), indicate concentration in moles per liter of the volume of a par-ticular region and UI is the surfactant volume fraction, which is as-sumed to be equal to that of the interfacial region [32,34]. Thevolume fractions of the oil, UO, and aqueous, UW, regions are de-fined in a similar way as UI as the ration between the volume ofthe particular region and that of the emulsion.

The distribution of PG, which is both oil and water soluble,(PO

W ¼ 0:84, see below), is described by two partition constants, PIW

and PIO, because significant amounts of PG are in all regions, oil,

interface, and water, Scheme 1, but for very hydrophobic (waterinsoluble) or very hydrophilic (oil insoluble) antioxidants, thedescription of their distribution can be simplified. For instance, forvery hydrophilic antioxidants such as ascorbic acid (vitamin C) orGA (PO

W ¼ 0:03, see below), their concentrations in the oil regioncan be considered effectively zero [35,54] and only PI

W is needed todescribe their distributions. For very hydrophobic antioxidants suchas TOC, only PI

O is needed to describe their distributions because itseffective concentration in the water region is negligible [34]. Detailson these approximations can be found elsewhere [32,34,35].

Eqs. (3)–(5) describe the dependence of kobs on the surfactant,water and oil volume fractions and the medium effect of the inter-facial region, for antioxidants that are very hydrophilic and parti-tion primarily between water and the interfacial regions [35] (Eq.(3)), very hydrophobic and partition primarily between the oiland interfacial regions [34] (Eq. (4)), and for antioxidants of inter-mediate hydrophobicity (or hydrophilicity) (Eq. (5)) and partitionbetween all three regions.

kobs ¼kI½AO�TPI

W

UIPIW þUW

ð3Þ

kobs ¼kI½AO�TPI

O

UIPIO þUO

ð4Þ

kobs ¼kI½AO�TPI

OPIW

UWPIO þUIP

IWPI

O þUOPIW

ð5Þ

The values of the partition constants are obtained from the re-ciprocal forms of Eqs. (3)–(5) (not shown), that is, from plots of1/kobs versus UI should be linear with positive intercepts. The reci-procal forms were used to determine the values of PI

W, PIO and kI for

GA, PG, and TOC as indicated elsewhere [32–35].

Once the partition constants are known, determining the per-centage of the antioxidant in the interfacial region of the emulsionis straightforward. Eqs. (6)–(8), which were derived previously[32–35], were used to calculate the percentage of the antioxidantsin the interfacial region of the emulsions. Similar expressions,which can be found elsewhere [31–36], were employed to calcu-late the percentage of the antioxidant in the aqueous and oilregions.

%AOI ¼100UIP

IW

UIPIW þUW

ð6Þ

%AOI ¼100UIP

IO

UIPIO þUO

ð7Þ

%AOI ¼100UIP

IOPI

W

UOPIW þUIP

IOPI

W þUIPIO

ð8Þ

3. Results

3.1. Partition constant POW of GA, PG, and TOC between stripped oils

and water in the absence of emulsifier

Calculations of PIW and PI

O from the kinetic data requires valuesof the bulk phase partition constant between the oil and waterphases in the absence of emulsifier, PO

W, for GA, PG, and TOC. Valuesof PO

W were determined in a 1:9 (v:v) oil to water mixture, in theabsence of emulsifier, by employing a modified shake-flask method[25]. GA and PG were dissolved in the water phase, and TOC wasdissolved in the oil phase. The oil/water mixture was stirred witha high-speed rotor for 1 min and allowed stand for 30 min to per-mit phase separation and thermal equilibrium at ambient temper-ature. The phases were separated by centrifugation and the AOconcentrations in the aqueous and in the oil phase were deter-mined by UV spectrometry with the aid of previously prepared cal-ibration curves. Values of PO

W were calculated from Eq. (1) and arethe average of duplicate or triplicate runs, with standard deviationsof <5%.

TOC has a very low solubility in water, <5 � 10�5 M [55]. Thus,the values of (TOCW) in the oil/water mixture will be very small,and small errors in the denominator of Eq. (1) may have a large ef-fect on the value of PO

W. Because the values of POW values for TOC are

not reliable, they are not reported in Table 2. Conversely, GA isessentially oil insoluble and an average value (six runs) ofPO

W ¼ 0:03 at T = 25 �C was determined. This value is similar to thatfound by Huang et al. [13] in unbuffered 1:9 corn oil–water binary

Fig. 2. Illustrative variations of kobs with UI in selected 1:9 (v:v) corn oil to acidic water emulsions of different HLB. (A) GA, h HLB = 16.7, s HLB = 15.0,4 HLB = 8.06. (B) PG, h

HLB = 16.7, s HLB = 15.0, 4 HLB = 12.7, } HLB = 10.6. The lines were obtained by fitting the kobs �UI pairs of data to Eq. (3) (GA) and 4 (PG) and to their reciprocals.[GA] � [PG] = 4.2 � 10�2 M, [16-ArNþ2 ] � 2.8 � 10�4, T = 25 �C.

Fig. 3. Effects of increasing UI on kobs (A) and 1/kobs versus UI (B) for reaction of 16-ArNþ2 with TOC in 1:9 (v:v) corn oil to water emulsions of different HLB. The solid lines areobtained by fitting the kobs values to Eq. (7) or its reciprocal. h HLB = 16.7, s HLB = 15.0, 4 HLB = 12.7, } HLB = 8.6. Experimental conditions: [a-Tocoferol] = 4 � 10�3 M,[16-ArNþ2 ] = 3 � 10�4 M, [NED] = 0.02 M. T = 25 �C, pH = 3.7 (acetate buffer 0.04 M).

6 S. Losada-Barreiro et al. / Journal of Colloid and Interface Science 389 (2013) 1–9

mixtures, POW ¼ 0:02. The PO

W value indicates that >99% of GA is inthe aqueous phase, in keeping with the hydrophilic nature of thisantioxidant. PG is both oil and water soluble, and an average value(four runs) of PO

W ¼ 0:84 at T = 25 �C was obtained, in close agree-ment with that of PO

W ¼ 0:85 estimated by Huang et al. [13].

3.2. Effects of emulsifier concentration and HLB on kobs

Figs. 2 and 3 illustrate the variation of kobs with UI at constanttemperature in 1:9 corn oil–water emulsions of different HLB.For any antioxidant and HLB, values of kobs decrease asymptoticallywith increasing UI at constant antioxidant concentration, consis-tent with the predictions of Eqs. (3)–(5). Note that kobs decreasesby a factor of 4–6 on going from up to UI = 0.04, a variation similarto that found for other antioxidants.

For any UI, kobs (GA) < kobs (TOC) < kobs (PG), and the measuredkobs value depends on the HLB of the emulsion so that whenUI = 0.005, the kobs values decrease by a factor of 3–5 on decreasingthe HLB from 16.7 to 8.6 (Figs. 2 and 3). A change in the oil to waterratio (from 1:9 to 1:4) of the emulsion did not result in significantchanges in the kobs values for GA and TOC (differences less than10–15%, results not shown), indicating that for these antioxidants,kobs values are essentially independent of the oil to water ratio,

probably as a consequence of the very hydrophilic (GA) and veryhydrophobic (TOC) nature of the antioxidants, which is reflectedin the PO

W values reported in Table 2.The straight lines shown in Figs. 2 and 3 are plots of 1/kobs ver-

sus UI, Eqs. (3)–(5), and these plots were used to calculate PIW, PI

O

and kI values for the different emulsions. The results are summa-rized in Table 2. Note that the reported PI

W and PIO are the macro-

scopic (or measured) values in solution and not the values in theinterfacial regions.

4. Discussion

Results in Table 2 show that it is possible to evaluate the contri-butions of concentration and medium effects on kobs by employingthe assumptions of the pseudophase model for chemical reactivityin homogeneous microemulsions [37]. The concentration effectsare reflected in the antioxidant distribution (i.e., PI

O and PIW); mean-

while, medium effect is reflected in the values of the rate constantkI at the reaction site, in this case the interfacial region of the emul-sion, Scheme 2.

The similarities in the kI values in emulsions prepared withTweens (HLB 16.7–15) for each of the antioxidants GA, PG, andTOC suggest that the antioxidants are experiencing a similar

Fig. 4. Percentage of GA in the interfacial (A) and aqueous (B) regions of selected corn oil emulsions prepared with emulsifiers of different HLB. d HLB = 16.7, j HLB = 15.0, �HLB = 8.05.

S. Losada-Barreiro et al. / Journal of Colloid and Interface Science 389 (2013) 1–9 7

chemical environment in the interfacial region of the emulsionscomposed of these surfactants. However, in emulsions preparedwith Spans, the kI values are much lower, decreasing by factorsof 2–3 with respect to the values in Tween emulsions. Such a dif-ference indicates a different reaction medium, as can be expectedfrom the substitution of the bulky polyoxyethylene groups inTweens by the hydroxyl groups on the sorbitan ring of Spans,decreasing the hydrophilicity of the molecule, and thus decreasingthe interfacial polarity.

Determining kI values is not necessary to assess the antioxidantdistribution; however, their knowledge may be relevant becausecomparison of kI values for a number of antioxidants could laythe basis for a scale of antioxidant activity that is independent ofthe antioxidant distribution in the emulsion [32]. In addition, theycan afford insights on the kinetics of the reaction in the interfacialregion of the emulsions, providing evidences to substantiatewhether emulsions change, or not, the reaction mechanism. Upto date, we found that emulsions do not change the mechanismof the reaction [31,35,36,56,57]. Moreover, we have recently deter-mined, for the first time, the activation parameters for the reactionbetween 16-ArNþ2 ions and GA [35], TOC [34], PG (unpublished re-sults) [58], and ascorbic acid (Vitamin C) [54] in the interfacial re-gion of corn oil emulsions.

A preliminary analyses of the partition constant values inTable 2 show that PI

W and PIO values depend on the particular anti-

oxidant employed, and their variation with the HLB does not followthe same trend for all the antioxidants. For instance, PI

W values forGA are essentially independent of the HLB of the emulsion, with anaverage value of PI

W ¼ 139. However, for PG, both PIW and PI

O values

Fig. 5. Effects of emulsifier HLB and emulsifier concentration on the distribution of PG bHLB = 16.7, � HLB = 15.0, j HLB = 12.7, N HLB = 10.6.

increase upon decreasing the HLB and, on the contrary, the PIO val-

ues for TOC decrease by a factor of �2 upon decreasing the HLBfrom 16.7 to 8.6.

The calculated value for GA in emulsions of HLB = 16.7,PI

W ¼ 123, is very similar to that previously reported at the samepH [35]. The small variation in the PI

W values upon decreasingthe emulsifier HLB is reflected in the distribution of GA, Fig. 4,showing that, for any given HLB, the percentage of GA in the inter-facial or aqueous regions is quite similar. Results in Fig. 4 indicatethat at low surfactant volume fractions of UI = 0.005, the percent-age of GA in the interfacial region is �40%, and this percentage in-creases upon increasing UI to %GA � 90% at UI = 0.04. Thus, thefraction of GA in the interfacial region depends on the particularsurfactant volume fraction, and for UI values >0.04, GA is almostcompletely located in the interfacial region of the emulsions.

PG is much more hydrophobic and less soluble in water than GAbecause of the acylation of the carboxylic group [56]. Note that theworking pH = 3.7 is nearly 105 times lower than the pKa value ofthe phenolic group in the 4-position of the pyrogallol moiety,pKa � 8.1 [59,60], which means that the 4-OH group of PG is essen-tially completely protonated. This is reflected in the calculated PI

W

value, Table 2, which is higher than that of GA as a consequence ofthe higher hydrophobicity.

Bearing in mind the definition of PIO and PI

W, Scheme 1, results inTable 2 also indicate that more and more PG is being displaced tothe interfacial region from the oil and aqueous regions upondecreasing the HLB of the emulsion as shown in Fig. 5. At surfac-tant volume fractions low of UI = 0.005, %PG increases from �40%to �90% upon decreasing the HLB from 16.7 to 10.6. The

etween the interfacial (A), aqueous (B), and oil (C) regions of corn oil emulsions. d

Fig. 6. Distribution of TOC between the interfacial (A) and oil (B) regions of corn oil emulsions of different HLB. d HLB = 16.7, j HLB = 15.6, N HLB = 15.0, . HLB = 14.7, �HLB = 8.6.

8 S. Losada-Barreiro et al. / Journal of Colloid and Interface Science 389 (2013) 1–9

differences in the percentages of PG in the interfacial region arelower at high UI = 0.04 values, ranging from �85% (HLB = 16.7) to�95% (HLB 10.6).

TOC is not water soluble and thus only distributes between theoil and interfacial regions of the emulsions, Fig. 6. The determinedPI

O values are much lower than those of PG, showing that, for thesame surfactant concentration, the amount of TOC in the interfacialregion is much lower than that of PG, as illustrated in Fig. 6. Indeed,TOC is an oil soluble vitamin, and although their maximum solubil-ity in corn oils is currently unknown, soybean oil solutions with[TOC] � 2 M have been prepared [61].

The large differences in the measured PIO values between PG and

TOC are reflected in the remarkable differences in their distribu-tion. Results in Fig. 6 show that a large fraction of TOC is locatedin the oil region, %TOC = 70–90%, depending on the HLB value, atUI = 0.004 and decrease up to %TOC = 20–50% at UI = 0.04. Our re-sults are in keeping with those found by Panya et al. [62] for ros-marinate alkyl esters, who were able to show, by using our16-ArNþ2 chemical probe, that a fraction of the antioxidants are lo-cated in the oil region.

Taken together, our distribution results clearly show that themain parameter controlling the distribution of antioxidants is theemulsifier concentration; meanwhile, the HLB of the emulsifierhas a modest effect on the distribution of moderate to high hydro-phobicity AOs and an almost negligible effect on the distribution ofhydrophilic antioxidants. For a given UI, the percentage of GA andPG in the interfacial region is much higher than that of TOC be-cause of the much more solubility of TOC in the oil. The resultsare consistent with literature reports [17,20,63], showing thatthere seems to be an upper limit in the fraction of antioxidants lo-cated in the interfacial region of emulsions as demonstrated by theparabolic effect found in the efficiency of hydrophobic antioxidantswith alkyl chains ranging 12–20 carbon atoms, which is lower thanthose with alkyl chains of less than 12 carbon atoms, that is, morepolar antioxidants, which is contrary to the predictions of the polarparadox [18]. For gallic acid esters, the upper limit is found formedium chain length (between 4 and 8 carbon atoms) [58].

5. Conclusions

Results in Figs. 4 and 6 support several conclusions. (1) The frac-tion of GA in the interfacial region of stripped corn oils is not af-fected by the HLB of the emulsifier, contrary to that of PG andTOC. (2) For a given UI value, the fraction of PG and TOC in theinterfacial region increases upon decreasing the HLB of the

emulsifier. (3) At low UI (0.01) values, a significant fraction,>40%, of GA and PG is located in the interfacial region of the cornoils, but the amount of the much more hydrophobic antioxidantTOC is substantially lower. (4) The %AO in the interfacial region in-creases with increasing UI such that 80% or more of the AOs are lo-cated in the interfacial region at UI = 0.04 and they follow the order%PG > %GA > %TOC. (5) The intrinsic rate constants kI of each anti-oxidant decrease upon decreasing the HLB of the emulsifier.

The results obtained here are basic to understand how emulsi-fiers influence the activity of antioxidants in emulsions, providingfundamental information on the factors that control antioxidantdistributions. Lipid oxidation is initiated in the interfacial regionof emulsions [3,18] and, hence, knowledge on the effects of param-eters having an effect on the partitioning of antioxidants betweenthe different regions of food-grade emulsions (emulsion composi-tion, type of oil and emulsifier, acidity, temperature, added salts,etc.) is crucial to understand the efficiency of AOs in inhibiting lipidoxidation.

Partition constants describe the relative solubility of an antiox-idant (or other component) between two regions, for example, oiland water phases, and oil and interfacial and water and interfacialregions, and the percentage of the antioxidant in each region de-pends on its relative solubility in each region and on the volumeof the region. Thus, the assessment of antioxidant partitioning be-tween different phases is an important factor to select antioxidantsto favor their distribution toward the microenvironment that ismost susceptible to oxidation. The oxidative stability of foods var-ies according to the nature and location of lipid components, pro-oxidants and antioxidants [3], and prediction of the efficiency ofantioxidants in food emulsions is complex as a result of the parti-tioning of the antioxidants into the different regions of the emul-sions, each being characterized by specific properties.

The non-destructive chemical method employed here to assessthe antioxidant distribution is robust and permits examination ofthe parameters affecting antioxidant distribution, for example,the nature of the oil and of the emulsifier, the HLB of the antioxi-dant, and factors such as acidity and temperature. The methodol-ogy is also useful to obtain estimates of thermodynamicparameters controlling the antioxidant distribution and efficiency,which otherwise are very difficult to obtain. For instance, the anal-ysis of the temperature dependence on the partition constants ofantioxidants provided, for the first time, estimates of the thermo-dynamic parameters (DG, DH, and DS) for transfer of antioxidantsfrom the aqueous to the interfacial regions and from the oil to theinterfacial regions [34,35]. In addition, the analyses of effects of

S. Losada-Barreiro et al. / Journal of Colloid and Interface Science 389 (2013) 1–9 9

temperature on the rate constants kI provide estimates of the acti-vation energies in the interfacial region of emulsions [34,35].

By comparing the effects of different parameters, one can eval-uate the relative contributions of some of the main parameterscontrolling the distribution of the antioxidant in the interfacial re-gion of emulsified systems (temperature, acidity, oil to water ratio,or emulsifier concentration). Our investigations suggest that theemulsifier concentration is the main parameter controlling the dis-tribution of antioxidants, but other parameters such as the acidity[54], temperature [34,58] or the HLB of the emulsifier, as shownhere, may also have a significant influence depending on the natureof the antioxidant. The method also provides an estimate of thesecond order rate constant for the reaction in emulsions, and com-parison of these rate constants provides an estimate of antioxidantefficiency. Currently, no other method provides reliable estimatesof these parameters.

Finally, it is worth noting that the pseudophase kinetic modelemployed here makes no assumption on the type of emulsion(o/w or w/o) employed and our methodology can be, in principle,employed to interpret the chemical reactivity and determine theantioxidant distribution in w/o emulsions. The important assump-tions of the method are two: (1) the reactive components are in dy-namic equilibrium between the oil, interfacial, and aqueous regionthroughout the time course of the reaction, that is, the diffusivity ofthe reactive components is orders of magnitude faster than thechemical reaction, and (2) the distributions of components aredescribed by partition constants between regions. Extension ofour method to determine the distributions of antioxidant seriesor other components of interest in w/o emulsions is thus war-ranted and important to evaluate the importance of the partition-ing between the different regions of emulsions.

Acknowledgments

We thank Prof. Larry Romsted (Rutgers, the State University ofNew Jersey) for extremely helpful discussions and encouragement.Financial support from the following institutions is acknowledged:Xunta de Galicia (10TAL314003PR), Ministerio de Educación yCiencia (CTQ2006-13969-BQU), Fondo Europeo de Desarrollo Re-gional (FEDER) and Universidad de Vigo.

References

[1] S.J. Jadhav, S.S. Nimbalkar, A.D. Kulkarni, D.L. Madhavi (Eds.), FoodAntioxidants, Technological, Toxicological and Health Perspectives, MarcellDekker, Inc., NY, 1996.

[2] C. Genot, A. Meynier, A. Riaublanc, in: Afaf Kalman-Eldin (Ed.), Lipid Oxidationin Emulsions, Lipid Oxidation Pathways, AOC Press, USA, 1999.

[3] E.N. Frankel, Lipid Oxidation, The Oily Press, PJ Barnes & Associates,Bridgwater, England, 2005.

[4] E.M. Becker, L.R. Nissen, L.H. Skibsted, Eur. Food Res. Technol. 219 (2004).[5] T. Basu, N.J. Temple, M.L. Garg, Antioxidants in Human Health and Disease,

CABI Publ., Wallingford, UK, 1999.[6] E. Choe, D.B. Min, Compr. Rev. Food Sci. Food Saf. 8 (2009) 345–358.[7] J.W. Finley, A. Kong, K.J. Hintze, E.H. Jeffery, L.L. Ji, X.G. Lei, J. Agric. Food Chem.

59 (2011) 6837–6846.[8] E.N. Frankel, S.W. Huang, J. Kanner, J.B. German, J. Agric. Food Chem. 42 (1994)

1054–1059.[9] E.N. Frankel, Food Chem. 57 (1996) 51–55.

[10] W.L. Porter, Recent Trends in Food Applications of Antioxidants, Plenum Press,New York, 1980.

[11] W.L. Porter, Paradoxical behavior of antioxidants in food and in biologicalsystems, Priceton Scientific, Princeton, NJ, 1993.

[12] A. Heins, D.B. Donald, T. Skolowski, H. Stöckman, K. Schwarz, Lipids 42 (2007)573–582.

[13] S.W. Huang, E.N. Frankel, R. Aeschbach, J.B. German, J. Agric. Food Chem. 45(1997) 1991–1994.

[14] C. Jacobsen, K. Schwarz, H. Stöckmann, A.S. Meyer, J. Adler-Nissen, J. Agric.Food Chem. 47 (1999) 3601–3610.

[15] K. Oehlke, A. Heins, H. Stöckmann, K. Schwarz, Food Chem. 118 (2010) 48–55.[16] M. Pazos, A. Alonso, I. Sánchez, I. Medina, J. Agric. Food Chem. 56 (2008) 3334–

3340.[17] M. Laguerre, L.J. López-Giraldo, J. Lecomte, M.J. Figueroa-Espinoza, B. Baréa, J.

Weiss, E.A. Decker, P. Villeneuve, J. Agric. Food Chem. 57 (2009) 11335–11342.[18] F. Shahidi, Y. Zhong, J. Agric. Food Chem. 59 (2011) 3499–3504.[19] Y. Zhong, F. Shahidi, J. Agric. Food Chem. 60 (2012) 4–6.[20] M.M. Laguerre, L.J. López-Giraldo, J. Lecomte, M.J. Figueroa-Espinoza, B. Baréa,

J. Weiss, E.A. Decker, P. Villeneuve, J. Agric. Food Chem. 58 (2010) 2869–2876.[21] C. Berton, M.H. Ropers, M. Viau, C. Genot, J. Agric. Food Chem. 59 (2011) 5052–

5061.[22] A. Heins, T. Sokolowski, T. Stöckmann, K. Schwarz, Lipids 42 (2007) 561–572.[23] M. Laguerre, E.A. Decker, J. Lecomte, P. Villeneuve, Curr. Opin. Clin. Nutr.

Metab. Care 13 (2010) 518–525.[24] A. Leo, C. Hansch, D. Elkins, Chem. Rev. 71 (1971) 525–616.[25] A. Berthod, S. Carda-Bosch, J. Chromatogr., A 1037 (2004) 3–14.[26] J. Sangster, Octanol–Water Partition coefficients, Fundamentals and Physical

Chemistry, J. Wiley & Sons, Chichester, 1997.[27] K. Schwarz, E.N. Frankel, J.B. German, Fett/Lipid 98 (1996) 115–121.[28] K. Schwarz, S.W. Huang, J.B. German, B. Tiersch, J. Hartmann, E.N. Frankel, J.

Agric. Food Chem. 48 (2000) 4874–4882.[29] H. Stöckman, K. Schwarz, Langmuir 15 (1999) 6142–6149.[30] H. Stöckman, K. Schwarz, T. Huynh-Ba, J. Am. Oil Chem. Soc. 77 (2000) 535–

542.[31] K. Gunaseelan, L.S. Romsted, E. González-Romero, C. Bravo-Díaz, Langmuir 20

(2004) 3047–3055.[32] K. Gunaseelan, L.S. Romsted, M.J. Pastoriza-Gallego, E. González-Romero, C.

Bravo-Díaz, Adv. Colloid Interface Sci. 123–126 (2006) 303–311.[33] V. Sánchez-Paz, M.J. Pastoriza-Gallego, S. Losada-Barreiro, C. Bravo-Diaz, K.

Gunaseelan, L.S. Romsted, J. Colloid. Interface Sci. 320 (2008) 1–8.[34] M.J. Pastoriza-Gallego, V. Sánchez-Paz, S. Losada-Barreiro, C. Bravo-Diaz, K.

Gunaseelan, L.S. Romsted, Langmuir 25 (2009) 2646–2653.[35] S. Losada-Barreiro, V. Sánchez Paz, C. Bravo Díaz, F. Paiva Martins, L.S.

Romsted, J. Colloid Interface Sci. 370 (2012) 73–79.[36] M.J. Pastoriza-Gallego, V. Sánchez-Paz, L.S. Romsted, C. Bravo-Diaz, Prog.

Colloid Polym. Sci. 138 (2011) 33–38.[37] R. da Rocha Pereira, D. Zanette, F. Nome, J. Phys. Chem. 94 (1990) 356–361.[38] S.M. Barlow, Toxicological aspects of antioxidants used as food additives, in:

B.J.F. Hudson (Ed.), Food Antioxidants, Elsevier, London, 1990.[39] G.O. Adegoke, M. Kumar, M. Vijay, A.G. Krishna, M.C. Vadaraj, K. Sambaiah, B.R.

Lokesh, J. Food Sci. Technol. 35 (1998) 283.[40] F. Shahidi, Y. Zhong, Antioxidants: Regulatory Status, vol. 6, J. Wiley & Sons,

NY, 2005.[41] J. Pokorny, N. Yanishlieva, M. Gordon, Antioxidants in Food: Practical

Applications, CRC press, Boca Raton, 2001.[42] T. Waraho, D.J. McClements, E.A. Decker, Trends Food Sci. Technol. 22 (2011)

3–13.[43] P. Binks, Modern Aspects of Emulsion Science, The Royal Society of Chemistry,

Cambridge, 1998.[44] D.J. McClements, Food Emulsions, CRC Press, Boca Raton, USA, 2005.[45] D.T. Wasan, A.D. Nikolov (Eds.), Encyclopedic Handbook of Emulsion

Technology, Marcel Dekker, NY, USA, 2001.[46] W.C. Griffin, J. Soc. Cosmet. Chem. 1 (1949) 311.[47] J.T. Davies (Ed.), Proceedings of the Second International Congress of Surface

Activity, Butterworths Scientific Publications, London, 1957.[48] S.E. Friberg, K. Larsson, Food Emulsions, Marcel Dekker, NY, 1997.[49] K. Larsson, S.E. Fribera, Food Emulsions, MARCELL DEKKER INC., NY, 1990.[50] R.A. Moreau (Ed.), Vegetable Oils in Food Technology-Composition, Properties

and Uses, Blackwell Publishing, Sussex, UK, 2011 (Chapter 10).[51] G. Marti-Mestres, F. Nielloud (Eds.), Pharmaceutical Emulsions and

Suspensions, Marcel Dekker, NY, 2000.[52] M.P. Doyle, W.J. Bryker, J. Org. Chem. 44 (1979) 1572–1574.[53] M.C. Garcia-Meijide, C. Bravo-Diaz, L.S. Romsted, Int. J. Chem. Kinet. 30 (1998)

31–39.[54] M.J. Pastoriza-Gallego, S. Losada-Barreiro, C. Bravo Díaz, J. Phys. Org. Chem., in

press. doi: 10.1002-poc2949.[55] M.D. Dubbs, R.B. Gupta, J. Chem. Eng. Data 43 (1998) 590–591.[56] L.S. Romsted, J. Zhang, J. Agric. Food Chem. 50 (2002) 3328–3336.[57] M.J. Pastoriza-Gallego, A. Fernández-Alonso, S. Losada-Barreiro, V. Sánchez-

Paz, C. Bravo-Diaz, J. Phys. Org. Chem. 21 (2008) 524–530.[58] S. Losada-Barreiro, PhD Thesis, Universidad de Vigo, 2009.[59] M.M. Delgado-Zamarreño, A. Sánchez-Pérez, I. González-Maza, J. Hernández-

Méndez, J. Chromatogr. A 871 (2000) 403–414.[60] S. Losada-Barreiro, C. Bravo-Diaz, Helv. Chim. Acta 92 (2009) 2009–2023.[61] J. Velasco, M.C. Dobarganes, G. Márquez-Ruiz, Eur. J. Lipid Sci. Technol. 106

(2004) 325.[62] A. Panya, M. Laguerre, C. Bayrasy, J. Lecomte, P. Villeneuve, D. McClements, E.A.

Decker, J. Agric. Food Chem. 60 (2012) 2692–2700.[63] A. Torres de Pinedo, P. Peñalver, I. Pérez-Victoria, D. Rondón, J.C. Morales, Food

Chem. 105 (2007) 657–665.