References

1 Kober, P.A. (1917) Pervaporation, Perstillation and Percrystallization in 1. Am. Chem. Sot. 39,944-948 2 Maeda, Y. and Kai, M. (1991) Recent Progress in Pervaporation for Water/Ethanol Separation in Pervaporation Membrane Separation Processes (Huang, R.Y.M., ed.), pp. 391-435, Elsevier 3 Mulder, M. (1992) Basic Prmciples ofMembrane Technology, Kluwer Academic Press 4 Karlsson, H.O.E. and Tr&@dh, C. (1993) Aroma Compound Recovery with Pervaporation - Feed Flow Effects in 1. Membr. Sci. 81, 163-171 5 Hiimpler, C. and Bode, E. (1995) Single Permeant Pervaporation Through Surface Modified and Multiple Layers Membranes in Proceedings of the Seventh International Conference on Pervaporation Processes in the Chemical Industry (Bakish, R., ed.), pp. 182-l 92, Bakish Materials Corporation, Englewood, NJ, USA 6 Koops, C.H. and Smolders, C.A. (1991) Estimation and Evaluation of Polymeric Materials for Pervaporation Membranes in Pervaporation Membrane Separation Processes (Huang, R.Y.M., ed.), pp. 253-278, Elsevier 7 Karlsson, H.O.E. and Tr;igdrdh, G. (19931 Pervaporation of Dilute Organic-Water Mixtures. A Literature Review on Modelling Studies and Applications to Aroma Compound Recovery in 1. Membr. Sci. 76, 121-l 46 8 Sulc, D. (1984) Fruchtsaftkonzentrierung und Fruchtaromaseparierung in Confructa Studien 28, 258-318 9 Mannheim, C.H. and Passy, N. (1975) Aroma Recovery and Retention in Fruit Juices in fnt. F/avows Food Addit. 6, 323-328 10 Karlsson, H.O.E. and Tr;igdrdh, G. (1994) Aroma Compound Recovery with Pervaporation - The Effect of High Ethanol Concentrations in 1. Membr. Sci. 91, 189-198 11 Bomben, J.L., Kitson, J.A. and Morgan, A.I. (1966) Vacuum Stripping of Aromas in Food Jechnol. 20, 125-l 28 12 Escoudier, J.L., Le Bouar, M., Moutounet, M., Jouret, C. and Barillere, J.M. (1988) Application and Evaluation of Pervaporation for the Production of Low 13

14 15

16

17 18

19

20

21

Alcohol Wines in Proceedings of the Third Jnternational Conference on Pervaporation Processes in the Chemical Industry (Bakish, R., ed.), pp. 387-397, Bakish Materials Corporation, Englewood, NJ, USA Hickey, P.J. and Cooding, C.H. (1994) The Economic Optimization of Spiral Wound Membrane Modules for the Pervaporative Removal of VOCs from Water in /. Membr. Sci. 97, 53-70 BrOschke, H.E.A. (1990) Removal of Ethanol from Aqueous Streams by Pervaporation in Desalination 77, 323-329 Kimmerle, K. and Gudernatsch, W. (1991) Pilot Dealcoholization of Beer by Pervaporation in Proceedings of the Fifth International Conference on Pervaporation Processes in the Chemical /ndustry (Bakish, R., ed.), pp. 291-307, Bakish Materials Corporation, Englewood, NJ, USA Lee, E.K. (1993) in Science for the Food /ndustry of the Zlst Century, Biotechnology, Supercritical Fluids, Membranes and Other Advanced Technologies for low Calorie, Healthy Food Alternatives (Yalpani, M., ed.), pp. 195-212, ATL Press Lee, E.K., Kalyani, V.J. and Matson, S.L. (1991) Process for Treating Alcoholic Beverages by Vapor-arbitrated Pervaporation, US Patent 5 013 447 Koseoglu, S.S., Hernandez, E., Shah, V. and Tuohey, D. (1995) Opportunities for Pervaporation: Processing Edible Oils and Fats in Proceedings ofthe Seventh International Conference on Pervaporation Processes in the Chemical Industry (Baklsh, R., ed.), pp. 263-270, Bakish Materials Corporation, Englewood, NJ, USA Rautenbach, R., Klatt, S. and Vier, I. (1992) State of the Art of Pervaporation 10 Years of Industrial PV in Proceedings oithe Sixth International Conference on Pervaporation Processes in the Chemical Industry (Bakish, R., ed.), pp. 2-l 5, Bakish Materials Corporation, Englewood, NJ, USA van Gemert, R.W. and Cuperus, F.P. (1995) Newly Developed Ceramic Membranes for Dehydration and Separation oi Organic Mixtures by Pervaporation in 1. Membr. Sci. 105,287-291 Sano, T., Hasegawa, M., Kawakami, Y. and Yanagishita, H. (1995) Separation of Methanol/Methyl-terf-butyl Ether Mixture by Pervaporation Usmg Zeolite Membrane in 1. Membr. Sci. 107, 193-l 96

Review

Lipid oxidation in
Lipid oxidation food emulsions. depends is a major cause of quality deterioration in The design of foods with improved quality on a better understanding of the physicochemical

food emulsions
John N. Coupland and D. JulianMcClements

mechanisms of lipid oxidation

in these systems. The oxidation

of emulsified lipids differs from that of bulk lipids, because of the presence of the droplet membrane, the interactions between the ingredients, and the partitioning regions. of ingredients between the oil, aqueous and interfacial

Lipid oxidation is of great concern to the food industry because it leads to the development of undesirable offflavors (rancidity) and potentially toxic reaction products. Food manufacturers must therefore develop methods of preventing, or at least retarding, lipid oxidation in foods. To do this effectively, it is necessary to have a

John N. Coupland
the Department 01003,

and D. Julian McClements University

(corresponding of Massachusetts,

author) Amherst,

are at MA

of Food Science,

USA (fax: +l-413-545-1262;

e-mail: mcclements@foodsci.umass.edu).

thorough understanding of the mechanisms of lipid oxidation, and how these are affected by the physicochemical environment of the lipids. Lipid oxidation has been extensively studied in bulk fats and oils, and there is now a fairly good understanding of the factors that affect oxidation in such systems. Research in this area has attempted to elucidate the mechanisms by which lipid oxidation proceeds under various conditions, and the resultant reaction products. Understanding of the factors that affect lipid oxidation in systems in which the fat is dispersed as emulsion
83

Trends

in Food Science

& Technology

March

1996

[Vol. 71

01996,

Elsewer

Scmce

Ltd

droplets is still fairly poor, and currently this is an active area of researchzw6. This lack of understanding is surprising considering the large number of foods that consist either wholly or partially of emulsions, or that have been in an emulsified form at some time during their production; these include milk, cream, cheese, mayonnaise, margarine, butter, ice cream, soups, sauces, cream liqueurs, baby foods and coffee whiteners. There are significant differences between lipid oxidation in bulk fats and that in emulsified fats, and the aim of this article is to highlight these differences. As there still appears to be some confusion about the nature of the colloidal systems used in many previous oxidation studies, we begin by describing the basic features of micellar and emulsion systems.

Food colloids
Emulsions An emulsion consists of two immiscible liquids (usually oil and water), one dispersed in the other in the form of small spherical droplets (Box 1). In most foods, the diameter of these droplets varies between -0.1 pm and 50p,m (Ref. 7). A system that consists of oil droplets dispersed in an aqueous phase is known as an oil-in-water (or o/w) emulsion (e.g. mayonnaise, milk, cream and soups), whereas a system that consists of water droplets dispersed in an oil phase is known as a water-in-oil (or w/o) emulsion (e.g. margarine, butter and spreads). The focus in this article is exclusively on lipid oxidation in o/w emulsions. It has been suggested that lipid oxidation in w/o emulsions occurs at a rate similar to that in bulk oils2. This may be the case for some w/o emulsions, but we believe there may be others in which the interfacial oxidation mechanisms described in this article are also important. An emulsion is a thermodynamically unstable system because of the positive free energy needed to increase the surface area between the oil and water phases. With time, emulsions tend to separate into a system that consists of a layer of oil (lower density) on top of a layer of water (higher density). This is clearly seen if one tries to homogenize pure oil and pure water together: initially an emulsion is formed, but after a few minutes phase separation occurs. To form emulsions that are kinetically stable for a reasonable period of time (a few weeks, months or even years, depending on the application), chemical substances known as emulsifiers must be added before homogenization. Emulsifiers are surface-active molecules that adsorb to the surface of freshly formed droplets during homogenization, forming a protective membrane that prevents the droplets from coming close enough together to aggregate7. The most common emulsifiers used in the food industry are: amphiphilic proteins (e.g. from casein, whey, soy or egg), phospholipids (e.g. egg or soy lecithin) and small-molecule surfactants (e.g. Spans, Tweens or fatty acids). The nature of the membrane surrounding the droplets may have important implications .for lipid oxidation processes, as will be discussed later.
84

An emulsion can be considered to consist of three regions: the interior of a droplet, the continuous phase and the interfacial region. The interface consists of a narrow region surrounding each emulsion droplet that consists of a mixture of oil, water and emulsifier molecules. Typically, the interfacial region has a thickness of only a few nanometers*. In fine emulsions, the molecules at the interface make up a significant proportion of the total number of molecules present in the droplet. The effect of emulsion droplet size on the percentage of oil molecules at the surface of a droplet is summarized in Table 1. The various molecules in an emulsion partition themselves between the three different regions according to their polarity and surface activity. Nonpolar molecules are located predominantly in the oil phase, polar molecules in the aqueous phase, and amphiphilic molecules at the interface. The precise molecular environment of a molecule may have a significant effect on its chemical reactivity or other properties in a systemlO. Another factor that will also be important with regard to lipid oxidation in emulsions is the orientation of the lipid molecules in the interfacial region; for example, whether they are parallel or perpendicular to the interface, because this will affect their accessibility to attack by water-soluble, reactive oxygen species (e.g. hydrogen peroxide, hydroxyl radicals and perhydroxyl radicals). Many of the physicochemical properties of emulsions can be understood only with reference to their highly dynamic nature. Emulsion droplets are in continual motion and frequently collide with one another. Some types of emulsifiers are capable of rapidly exchanging between the interfacial region and the aqueous (or oil) phase, whereas others exchange slowly, if at all. Nonpolar molecules may be transferred between droplets as a result of droplet-droplet collisions, molecular diffusion through the continuous phase, or binding to micelles or protein molecules*. All of these phenomena may have a significant influence on lipid oxidation in colloidal systems. Micelles and vesicles Small-molecule surfactants aggregate spontaneously in solution to form a variety of thermodynamically stable structures including micelles, bilayers and vesicles (Box 1). Such structures are formed because they minimize the contact area between polar and nonpolar regions: the so-called hydrophobic effect9. The precise nature of the molecular aggregates formed is governed by the molecular geometry and interactions of the surfactant head groups and tails8,9. In both micelles and vesicles, the surfactant molecules arrange themselves so that the nonpolar tails are located in the interior (away from the water) and the hydrophilic head groups are located at the surface (in contact with the water). Micelles and vesicles are capable of solubilizing nonpolar molecules into their hydrophobic cores. Solubilization has a number of important practical implications:
Trends in Food Science & Technology March 1996 [Vol. 71

the concentration of nonpolar molecules in the aqueous phase is increased;

l

nonpolar molecules can be transported across an aqueous phase in which they are normally insoluble; the rate of certain chemical reactions is different in micelles or vesicles than in a bulk phase.

Many previous studies of lipid oxidation in colloids have used model systems consisting of lipid molecules incorporated into surfactant micelles, vesicles or microemulsions12-17. Such model systems are used to mimic the characteristics of lipids dispersed in real foods, and have several advantages: their composition and structure can be carefully controlled, they are thermodynamically stable, and a large proportion of the lipid molecules are at an oil-water interface (which speeds up any interfacial reactions). To interpret measurements accurately in these systems, it is important to appreciate the structure and dynamics of the micelles and vesicles used, and how these are affected by changes in environmental conditions and the addition of other components (such as antioxidants or radical generators). In addition, it should be stressed that although these systems represent useful models for oxidation in biological membranes, they may not be such good models for food emulsions, because a larger fraction of the lipids in emulsions are present in the interior of the emulsion droplets (Table 1). Thus, some of the factors that affect lipid oxidation in emulsions might be expected to be similar to those in bulk oils, whereas others might be similar to those in micellar or vesicle systems.

lipid oxidation
Lipid oxidation proceeds via a complex, radical chain reaction, which is generally separated into initiation, propagation and termination stages. As these reactions have been widely and thoroughly reviewed18v9, only a brief statement of the types of reactions involved will be given. Lipid autoxidation is a radical reaction between unsaturated oils and oxygen. It is initiated by the abstraction of a hydrogen atom from a methylene group of a polyunsaturated fatty acid. A range of oxidation initiators are important in food systems, including UV light, transition metal ions and certain enzymes. It is worthwhile noting that, apart from those generated by light, food radicals are usually generated in the aqueous
Trends

Table 1. Droplet-size dependence of the surface area per gram of oil, and the percentage of oil molecules at the droplet surface, for a typical oil-in-water emulsion Droplet radius (pm)
100 10
1 0.1 Values were calculated

Droplet surface area (m*/g)

0.03 0.3 3 30
assuming that the oil molecules

Oil molecules at surface (%)

0.02 0.2
1.8

18
are typical triacylglycerols, length of 6 nm. Data taken

with an oil density of 920 kg/m and an end-to-end from Ref. 9

molecular

/ i
85

in Food Science

&Technology

March

1996

[Vol. 71

phase, which has important implications for the oxidation of emulsified oils. Once formed, the lipid radical reacts rapidly with oxygen to form a lipid peroxide. The peroxide extracts another hydrogen atom to form a hydroperoxide, which propagates the radical reaction. Most food oils naturally contain enough lipid peroxides to promote lipid oxidation, even if other sources of radical generators are rigorously eliminated. Lipid peroxides are relatively stable but decompose rapidly in the presence of heat or in the presence of transition metal ions via the Fenton reaction. During lipid oxidation, a number of decomposition reactions occur simultaneously, eventually leading to the formation of a complex mixture of reaction products, including aldehydes, ketones, alcohols and hydrocarbons, which are responsible for the characteristic physicochemical properties of oxidized oils. Many oxidation products are smaller and contain more oxygen atoms than the starting oil. These molecules are likely to be more surface active than the initial lipid, and some of them may even be relatively water soluble (e.g. propanal), such that they leach into the aqueous phase.

required for bulk oils because the lipid component usually has to be separated from the aqueous phase before analysis. This is most often achieved by mixing the emulsion with an appropriate nonpolar solvent (e.g. hexane) to extract the lipid. Occasionally other chemicals must be added to facilitate the breakdown of the emulsion. The efficiency of the extraction process depends on the partitioning of the lipid breakdown products between the extracting solvent and the aqueous phase. Problems may therefore be encountered in the latter stages of lipid oxidation because many of the reaction products formed are surface active or water soluble, and will not therefore be extracted by nonpolar solvents. Thus, it may be necessary to use a combination of extraction procedures, or analytical techniques, to provide information about the reaction products in both the aqueous and oil phases. Emulsion properties To better understand the factors that determine lipid oxidation in food emulsions, it is important to adequately characterize the physical properties of the systems used. The most important properties are: the size, concentration and physical state of the emulsion droplets, the nature of the emulsifier membrane (i.e. thickness, electrical charge, degree of packing, composition), and the extent of any droplet-droplet interactions. A variety of experimental techniques has been developed to measure these properties8*2. Droplet concentration and size distribution can be measured using traditional techniques such as light or electron microscopy, or using more modern methods such as dynamic or static light scattering, electrical conductivity, ultrasound or NMR. The degree of crystallinity of emulsion droplets can be measured using dilatometry (density measurements), ultrasound, NMR or differential scanning calorimetry. It is difficult to study the nature of the droplet interface in situ, because of the small size of the droplets and their dynamic nature. Nevertheless, several methods (e.g. surface tension meters, Langmuir troughs, neutron reflection and radioactive tracers) are available for studying the nature of planar oil-water interfaces, and the results from such studies can often be extended to emulsions. The physicochemical properties of the oil and aqueous phases change during oxidation, which can have a pronounced influence on the accuracy of many of the experimental techniques mentioned above. For example, static light scattering is routinely used to measure the droplet size distributions of emulsions, and the accuracy of the measurements is extremely sensitive to the refractive indices of the oil and aqueous phases. As lipid oxidation proceeds, the refractive index of the oil may change significantly, and may lead to large errors in the measured droplet size distribution if not taken into account22. Similar caveats may be relevant for other commonly used experimental techniques. In most previous studies, researchers have carefully characterized the chemical changes that occur during lipid oxidation in emulsions, but have largely ignored
Trends in Food Science & Technology March 1996 [Vol. 71

Experimental techniques
Lipid oxidation A variety of analytical techniques has been developed to study lipid oxidation in bulk fats and oils1g*20. Many of these techniques can also be used to monitor lipid oxidation in emulsions, although it is often necessary to extract the oil phase before analysis. The techniques measure changes in the concentration of molecules within a system that are indicative of the progress of lipid oxidation. Some techniques measure the loss of initial reactants (such as oxygen or lipid), others the formation of intermediates (such as hydroperoxides or conjugated dienes), and others the formation of later products (such as alcohols, aldehydes, hydrocarbons or ketones). A number of the most useful analytical techniques are mentioned below. Oxidation can be followed by measuring the extent of oxygen consumption using an oxygen meter, whereas the loss of oxidizable lipid, or the formation of specific reaction products, can be measured by gas chromatography or high-performance liquid chromatography. By combining these chromatographic techniques with mass spectroscopy or NMR spectroscopy it is possible to obtain information about the molecular structure of the various reaction products. The formation of hydroperoxides can be determined by measuring conjugated dienes using UV-visible spectroscopy. Simple chemical tests are available to determine the concentration of specific classes of reaction products or types of chemical changes such as the TBA (thiobarbituric acid) value or peroxide value. Because of the chemical complexity of lipid oxidation great care must be taken when selecting an appropriate technique to characterize its progress. It is always advisable to use at least two, if not more, different analytical techniques to obtain an adequate description of the process. The preparation of samples for the analysis of lipid oxidation in emulsions is often more difficult than that
86

the physical structure and dynamics of the systems used. We believe that a better understanding of lipid oxidation in such systems can be achieved only by using a combination of both physical and chemical techniques.

aqueous phase of the emulsion increased the stability of the lipid against oxidation. This is probably because at higher surfactant concentrations, the packing of surfactant molecules at the oil-water interface is tighter, and therefore the membrane acts as a more efficient barrier Control of lipid oxidation in colloidal systems to the diffusion of lipid oxidation initiators into the oil Much research into methods of controlling lipid oxi- droplets. Studies with phospholipid-stabilized emulsions dation in colloidal systems has been carried out by have shown that the diffusion of molecules across an researchers involved in medical or pharmaceutical interface can be effectively retarded by increasing the research using micellar or vesicle systems. Lipid oxi- tightness of the packing of phospholipids at the dation in food emulsions is influenced by many of the interfacez5. same factors as in non-food systems, but a number of The membrane surrounding a droplet may also conadditional factors are also important. In this section, tain molecules that are capable of scavenging free radidata from previous studies of lipid oxidation in micelles, cals and therefore of retarding lipid oxidation. Aqueous vesicles and emulsions will be presented, to highlight solutions of certain sugars and amino acids are known methods of retarding lipid oxidation in food emulsions. to be able to scavenge free radicals. Many emulsifier molecules (e.g. Tweens) contain sugar or amino acid Control of droplet membranes moieties that may also act as free-radical scavengers. Emulsion droplets are surrounded by a membrane of Such emulsifier molecules are likely to be particularly emulsifier molecules that prevents them from coalesc- effective when they form part of a droplet membrane ing. In foods, this membrane usually consists of small- because of their relatively high local concentration. molecule surfactants or proteins (or sometimes a combiProteins are the most widely used emulsifiers in nation of the two). In addition to enhancing the physical foods, and so it is important to understand their roles in stability of emulsion droplets, these membranes may determining the overall properties of food emulsions. also protect lipids from oxidation by acting as a barrier Proteins are amphiphilic molecules that tend to accumuto the penetration and diffusion of molecular species late at oil-water interfaces because of their surface acthat initiate lipid oxidation into the droplets. Lipid oxi- tivity. However, only a finite interfacial area is present dation may therefore be controlled by engineering the in any given emulsion and once this has been saturated properties of the membranes surrounding the droplets, any additional protein remains in the aqueous phase for example by altering their charge or permeability, or (although some proteins are capable of forming multiby utilizing specific interactions between membrane layers). Proteins can be displaced from an interface by molecules and lipid oxidation initiators. more-surface-active ingredients, such as small-molecule An important method of controlling lipid oxidation in surfactants or other proteins. The conformation of some colloidal systems is to alter the electrical charge of the proteins (e.g. P-lactoglobulin) is altered when they are membrane. For example, ascorbic acid (which is nega- adsorbed to the surface of emulsion droplets in response tively charged) is -lOOO-fold less effective as an anti- to the change in their thermodynamic environment. The oxidant when the lipid is contained inside negatively partitioning of proteins between the droplet membrane charged micelles (sodium dodecyl sulfate, SDS) than and the aqueous phase, the composition of the memwhen it is contained inside positively charged micelles brane and the conformation of protein molecules are all (hexadecyl trimethylammonium bromide)]. This is be- likely to influence the effectiveness of proteins at retardcause there is an electrostatic repulsion between ascor- ing lipid oxidation in emulsions. bic acid and the negatively charged micelles, such that Proteins are commonly regarded as being effective the antioxidant is effectively excluded from the region antioxidants in emulsions; for example, proteins derived where the oxidation reaction takes place. Conversely, from meat, fish, milk and cereals are all capable of there is an electrostatic attraction between ascorbic acid retarding lipid oxidation in oil-in-water emulsionszh. and positively charged micelles, such that the antioxiHowever, some proteins have been shown to be prodant is present in the region where oxidation occurs and oxidants under certain conditions27~28. number of factors A is therefore more effective. Another study, which high- contribute to the antioxidant activity of proteins. Many lights the importance of the charge on the membrane, proteins are capable of forming relatively thick viscoshowed that EDTA (ethylenediaminetetraacetic acid) elastic membranes, which would be expected to restrict can act as a pro-oxidant in emulsions because of its abil- the penetration and diffusion of lipid oxidation initiators ity to chelate iron catalysts, making them less polar, and into the interior of emulsion droplets. Proteins, in the therefore more able to penetrate the droplet membranez3. droplet membrane or dispersed in the aqueous phase, Similar considerations are likely to be important for are capable of scavenging free radicals, perhaps being lipids contained in emulsion droplets. preferentially oxidized themselves. An important examAnother method of altering lipid oxidation rates in ple of this is the antioxidant activity of thiol groups in emulsions is to alter the effectiveness of the packing of proteins**. Cysteine and glutathione inhibit the haemthe emulsifier molecules in the droplet membrane. In a catalyzed oxidation of emulsified ethyl linoleate, but study of the autoxidation of emulsified safflower oilz4, it lose their antioxidant activity when their sulfhydryl was observed that the addition of extra surfactant to the groups are blocked or removed**. However, under
Trends in Food Science & Technology March 1996 [Vol. 71 a7

certain conditions, sulfhydryl groups can also act as pro-oxidants, for example by promoting superoxide generation by a copper catalystz8. The major proteins naturally present in milk, especially the caseins, have antioxidant activity in oil-inwater emulsions of triolein and of sunflower oil, both stabilized by lysophosphatidylcholinezg~30. However, it was also observed that the metalloproteins in milk could act as significant pro-oxidants, especially in the presence of small amounts of metal ions2g*30. The prooxidant effect of xanthine oxidase and also of lactoperoxidase in the presence of metal ions was greatly reduced by heating (at 80C for 2Os), suggesting that enzyme activity may have been responsible for at least some of the pro-oxidant effect. It is interesting to note that the presence of a small-molecule surfactant in this model system would have prevented the accumulation of protein at the interface and so the results may not be representative of a real milk sample. Homogenization protects milk fat from oxidation catalyzed by metal complexes3*. In raw milk the fat droplets are surrounded by a milk fat globule membrane that consists primarily of lipoproteins and small amounts of phospholipids, diacylglycerols, sterols and enzymes7. Homogenization increases the inter-facial surface area and alters the composition of the droplet membrane7. Casein, which was originally dispersed in the aqueous phase, adsorbs to the droplet surface forming a protective layer, which enhances the oxidative stability because it is the most efficient antioxidant protein in milkzg. Antioxidants The most commonly used method of retarding lipid oxidation in fatty foods is by the addition of antioxidantsig. Antioxidants that are effective in bulk oils are also usually effective in micelles, vesicles and emulsions12-5*17,32-36. However, the effectiveness of an antioxidant in an emulsion may be considerably different from that of the same antioxidant in a bulk oiP3. The effectiveness of nonpolar and polar antioxidants in suppressing the oxidation of corn oil in bulk and in emulsions has been investigated%. Predominantly nonpolar antioxidants (a-tocopherol and ascorbyl palmitate) were found to be more effective in an oil-in-water emulsion than in bulk oil, whereas the opposite was observed for predominantly polar antioxidants (Trolox and ascorbic acid). Differences in the effectiveness of the antioxidants can be attributed to their different affinities for the air-oil or water-oil interfaces in the two systems. Polar antioxidants are more effective in bulk oils because they form a protective membrane at the air-oil interface, which presumably reduces the accessibility of the lipid substrate to oxygen. In contrast, predominantly nonpolar antioxidants (which have some surface activity) are more effective in emulsions because they form a protective membrane around the droplets. In a similar study13, with soybean-oil-in-water emulsions stabilized by lecithin, a-tocopherol was shown to be much more effective as an antioxidant when it was solubilized in a lecithin membrane surrounding the droplets, than when aa

it was dissolved in the interior of the droplets. This is because the free radicals responsible for initiating lipid oxidation are generated in the aqueous phase, and must cross the droplet membrane before interacting with the oil in the interior of the droplet. The studies mentioned in this and the previous section indicate that the effectiveness of antioxidants at slowing down lipid oxidation in colloidal systems depends not only on their chemical interactions with components of the oxidation pathway, but also on their location and orientation in a structured system. Furthermore, they demonstrate the need for careful consideration of the precise nature of the system being investigated when selecting an appropriate antioxidant system: an antioxidant that works well for a bulk oil may not do so for the same oil in an emulsified state. It should also be noted that the effectiveness of antioxidants will not only depend on their partitioning into the interfacial region, but also on their molecular orientation and interactions in that region. Most of the studies of these effects have concentrated on the interactions between vitamins E and C in models of living tissue; this interesting structurereactivity relationship is described further in Box 2. Control of oxygen Undesirable lipid oxidation reactions can also be retarded by other means besides using natural or synthesized chemical antioxidants. One method is to reduce the concentration of oxygen in the food, for example by packing foods under vacuum or nitrogen. Oxygen is about three times more soluble in food oils than in water (-0.6mM and 0.2mM at 25C respectively4), which indicates that if measures are not taken to exclude oxygen from the aqueous phase there is likely to be sufficient present in the oil phase (as a result of diffusion from the aqueous phase) to cause lipid oxidation. The effect of low oxygen pressure on the kinetics of lipid oxidation in linoleic acid oil-in-water emulsions stabilized by Tween20 has been studied42,43.At low oxygen concentrations, the rate-limiting step for lipid oxidation was the diffusion of oxygen through the aqueous phase. At higher concentrations, the rate of oxygen diffusion was much faster than the rate of lipid oxidation, thus diffusion was not limiting. Lipid oxidation was found to be particularly sensitive to the speed at which the reaction vessel was shaken and to temperature in lowoxygen systems. The more vigorously a sample was agitated, the faster the rate of oxidation, because more oxygen is incorporated into the emulsion. In contrast, agitation had little effect on the rate of oxidation in those emulsions in which oxygen was not limiting. The rate of lipid oxidation decreased as the holding temperature of the sample was increased, because the solubility of oxygen in water decreases as the temperature is raised. Partitioning of reactants/products There is an increasing tendency for food manufacturers to produce low-fat products to meet consumer preferences for healthier foods. However, lowering the
Trends in Food Science & Technology March 1996 (Vol. 71

fat content of foods can result in the loss of important quality attributes, such as texture, mouthfeel and flavor. The effects of low concentrations of oil droplets on the autoxidation and enzymatic oxidation of linoleic acid and various vegetable oils have recently been investigated4v5. Oxidation was measured in model systems that consisted of lipid substrate solubilized in surfactant micelles (Tween 20) either in the presence or absence of low concentrations of inert oil droplets (O-3%, w/w, hexadecane). The rate of lipid oxidation in the emulsions decreased as the concentration of inert oil droplets was increased. The reason for this could be that some of the substrate (linoleic acid) moved from the surfactant micelles into the interior of the emulsion droplets, and therefore became inaccessible to direct interaction with the free radicals in the aqueous phase. The presence of inert oil also acted as a nonpolar reservoir for the products of lipid oxidation, thus reducing the amount of volatiles released into the headspace. The partitioning of lipid breakdown products between the oil and aqueous phases may affect the sensory perception of many real foodP. For example, flavor components are perceived much more strongly in an aqueous phase than in an oil phase. This may be of particular importance in the oxidation of emulsified oil because many of the final reaction products are more water soluble and volatile than the starting material. Ingredient interactions Most of the model systems discussed above consisted of oil droplets stabilized by an emulsifier dispersed in an aqueous phase. The only other additives were antioxidants and radical-generating systems. However, most food systems actually contain a large number of additional ingredients in the aqueous phase. These ingredients may act as either pro-oxidants or antioxidants depending on their chemical properties, the environmental conditions, and their interactions with the lipid component. It is therefore important to determine the effects that the various ingredients commonly found in foods have on the susceptibility of lipids to oxidation. The addition of sucrose (O-67%) to the aqueous phase of safflower-oil-in-water emulsions has been shown to reduce the rate of lipid oxidation24. It was suggested that the sugar decreased the concentration of oxygen in the aqueous phase, and decreased the diffusion coefficient of oxygen (because of the increased viscosity of the aqueous phase) 45 High concentrations of sucrose are . also capable of effectively scavenging free radicals. The extent of lipid oxidation was also shown to depend on the concentration of oil droplets present in the
Trends in Food Science & Technology March 1996 [Vol. 71

emulsionz4. For a given sugar concentration, there was an increase in the fraction of oil that was oxidized as the concentration of oil droplets decreased. This is probably because the number of radicals generated per droplet increases as the concentration of oil droplets decreases. Oil droplet composition Food oils consist primarily of a mixture of polyunsaturated fatty acids (which are highly susceptible to lipid oxidation) and saturated and monounsaturated fatty acids (which are relatively stable to lipid oxidation) either as free fatty acids or as glycerol esters, plus various other minor components. Recently, a study was carried out to ascertain the influence of droplet composition (ratio of polyunsaturated to saturated oil) on the rate of lipid oxidation in emulsions46. Oil-in-water emulsions were prepared using different ratios of ethyl linoleate (the polyunsaturated substrate) and hexadecane (the saturated inert diluent) as the oil phase. The rate of lipid oxidation was dependent on the concentration of oxidizable substrate in the droplets. Initially, oxidation occurred more rapidly in droplets containing low concentrations of substrate; however, during later stages oxidation was more rapid in droplets containing high concentrations. It is hypothesized that oxidation is probably initiated more rapidly at low substrate concentrations because a greater percentage of the surfaceactive substrate molecules is present at the surface of the droplet, and therefore more easily accessible to attack by the free radicals generated in the aqueous phase. However, the spread of oxidation within a droplet
89

is slower at low substrate concentrations because collisions between reactive species and substrate molecules will be less frequent. Future directions The importance of the precise physical as well as chemical environment of lipids in foods on their susceptibility to oxidation has been realized for many years. Nevertheless, few systematic studies of the roles of the physical state and structure of lipids in foods on lipid oxidation have been carried out. If food manufacturers are going to engineer high-quality foods that are nutritious, desirable and healthy, it is important that they have a better understanding of the complex factors that influence lipid oxidation. In this article we have reviewed some of the previous work in this area and attempted to provide suggestions, based on colloid science, to facilitate a better understanding of lipid oxidation in emulsions. All food emulsions are stabilized by an interfacial membrane. Food scientists could therefore attempt to engineer this membrane to protect the lipids from chemical as well as physical deterioration, perhaps by using charged surfactants, or by chemically modifying the best antioxidant proteins in such a way that they accumulate at the interface. It may be particularly profitable to copy the strategies that living cells use to cope with similar problems. Much more basic research still needs to be carried out in this area; for example, the influence of such factors as the physical state of the droplets, the molecular structures of the lipid and antioxidant components, the effects of various food ingredients, pH, the solubilities and diffusivities of the various intermediates, and the structure of the interfacial membrane on oxidation. Finally, although this review has concentrated on the effects of physical structure on oxidation reactions, it seems likely that oxidation reactions may alter physical structure, including changing the interfacial tension, and causing oxidative damage to proteins at the interface, which may lead to changes in the physical properties or stability of emulsions. Acknowledgements We are grateful to Nestle Research for partially funding this research, and to P. Chinachoti, E. Decker, W.W. Nawar and L. Ponginebbi for helpful advice and discussion. References
1 Halliwell, B., Murcia, M.A., Chirico, S. and Aruoma, 0.1. (1995) Free Radicals and Antioxidants in Food and In Viva: What They Do and How They Work in Crit. Rev. FoodSci. Mr. 35, 7-20 2 Fritsch, C.W. (1994) Lipid Oxidation -The Other Dimensions in Inform 5,423-436 3 Frankel, E.N. (1991) Recent Advances in Lipid Oxidation in /. Sci. Food Agric. 54, 495-551 4 Roozen, J.P., Frankel, E.N. and Kinsella, J.E. (1994) Enzymic and Autoxidation of Lipids in Low Fat Foods: Model of Linoleic Acid in Emulsified Hexadecane in Food Chem. 50,33-38 5 Roozen, J.P., Frankel, E.N. and Kinsella, J.E. (1994) Enzymic and Autoxidation of Lipids in Low Fat Foods: Model of Linoleic Acid in Emulsified Triolein and Vegetable Oils in Food Chem. 50, 39-43

6 Huang, SW., Frankel, E.N. and German, J.B. (1994) Antioxidant Activity of a- and y-Tocopherols in Bulk Oils and in Oil-in-water Emulsions in /. Agric. FooaChem.42,2108-2114 7 Dickinson, E. (1992) An Introduction to Food Colloids, Oxford University Press 8 Dickinson, E. and McClements, D.J. (1995) Advances in Food Colloids, Blackie 9 Israelachvili, J.N. (1992) /ntermo/ecu/ar andSurface Forces, Academic Press 10 Wedzicha, B.L. (1988) Distribution of Low Molecular-weight Food Additives in Dispersed Systems in Advances in Food Emulsions and Foams (Dickinson, E. and Stainsby, G., eds), pp. 329-371, Royal Society of Chemistry 11 Hiemenz, PC. (1986) Principles ofCo//oid and Surface Chemistry (2nd edn), Marcel Dekker 12 Niki, E., Kawakami, A., Yamamoto, Y. and Kamiya, Y. (1985) Oxidation of Lipids. VIII. Synergistic Inhibition of Oxidation of Phosphatidylcholine Liposome in Aqueous Dispersion by Vitamin E and C in Bull. Chem. Sot. Ipn 58, 1971-1975 13 Moberger, L. and Larsson, K. (1987) A Study of Fat Oxidation in a Microemulsion System in 1. Dispersion Sci. Jechnol. 8, 207-215 14 Motoyama, T., Miki, M., Mino, M., Takahashi, M. and Niki, E. (1989) Synergistic Inhibition of Oxidation in Dispersed Phosphatidylcholine Liposomes by a Combination of Vitamin E and Cysteine, in Arch. Biochem. Biophys. 270,655-661 15 Miyashita, K., Nara, E. and Toru, 0. (1993) Oxidative Stability of Polyunsaturated Fatty Acids in Aqueous Solution in Biosci. Biotechnol. Biochem. 57,1638-l 640 16 Pryor, W.A., Strickland, T. and Church, D.F. (1988) Comparison of the Efficiencies of Several Natural and Synthetic Antioxidants in Aqueous Sodium Dodecyl Sulphate Micelle Solutions in /. Am. Chem. Sot. 110, 2224-2229 17 Pryor, W.A., Cornicelli, J.A., Devall, L.J., Tait, B., Trivedi, B.K., Witiak, D.T. and Wu, M. (1993) A Rapid Screening Test to Determine the Antioxidant Potencies of Natural and Synthetic Antioxidants in /. Org. Chem. 58, 3521-3532 18 Halliwell, B. and Gutteridge, J.M.C. (1991) Free Radicals in Biology and Medicine (2nd edn), Clarendon Press 19 Nawar, W.W. (1985) Lipids in Food Chemistry(2nd edn) (Fennema, O.R., ed.), pp. 139-254, Marcel Dekker 20 Pike, O.A. (1994) Fat Characterization in Introduction to the Chemical Analysis of Foods (Nielsen, S.S., ed.), pp. 193-205, Jones and Bartlett 21 Dickinson, E. (1995) New Physico-chemical Techniques for the Characterisation of Complex Food Systems, Blackie 22 Zhang, H.J. and Xu, C.D. (1992) The Effect of Particle Refractive Index on Size Measurement in Powder Techno/. 70, 189-l 92 23 Ruben, C. and Larsson, K. (1985) Relations Between Antioxidant Effect of cu-Tocopherol and Emulsion Structure in 1. Dispersion Sci. Technol. 6, 213-221 24 Sims, R.J., Fioriti, ].A. and Turnbetas, J. (1979) Effect of Sugars and Sugar Alcohols on Autoxidation of Safflower Oil in Emulsions in /. Am. Chem. Sot. 56,742-745 25 Kabalanov, AS. and Shchukin, E.D. (1992) Ostwald Ripening Theory: Application to Fluorocarbon Emulsion Stability in Adv. Colloidfmul. Sci. 38,69-97 26 Lin, C.G., Fujimoto, K. and Hwang, L.S. (1993) The Antioxidative Effect of Protein on the Hemoglobin-catalyzed Oxidation of Sardine Oil in an Emulsion System in Nippon Shokuhin Kogyo Gakkai-Shi 40, 602-608 27 Yee, J.J., Shipe, W.F. and Kinsella, J.E. (1980) Antioxidant Effects of Soy Protein Hydrolysates on Copper-catalyzed Methyl Linoleate Oxidation in J Food Sci. 45,1082-l 083 28 Yee, 1.1.and Shipe, W.F. (1982) Effects of Sulphydryl Compounds on Lipid Oxidations Catalyzed by Copper and Haem in /. Dairy.%. 65, 1414-1420 29 Allen, J.C. and Wierden, W.L. (1982) Influence of Milk Proteins on Lipid Oxidation in Aqueous Emulsion. I. Casein, Whey Protein, and a-Lactalbumin in 1. Dairy Res. 49, 239-248 30 Allen, J.C. and Wierden, W.L. (1982) Influence of Milk Proteins on Lipid Oxidation in Aqueous Emulsion. II. Lactoperoxidase, Lactoferrin, Superoxide Dismutase and Xanthine Oxidase in 1. Dairy Res. 49, 249-263 31 Hegenauer, J., Saltman, P., Ludwig, D., Ripley, L. and Baja, P. (1979) Effects of Supplemental Iron and Copper on Lipid Oxidation in Milk. I. Comparison of Metal Complexes in Emulsified and Homogenized Milk in 1. Agric. Food Chem. 27,8604367 32 Barclay, L.R.C., Locke, 5.1. and MacNeil, J.M. (1983) The Autoxidation of Unsaturated Lipids in Micelles. Synergism of Inhibitors Vitamins C and E in Can. 1. Chem. 61,1288-l 290 33 Porter, W.L. (1993) Paradoxical Behavior of Antioxidants in Food and Biological Systems in Toxicol. Ind. Health 9, 93-122 34 Frankel, E.N., Huang, S.W., Kanner, J. and German, J.B. (1994) Interfacial Phenomena in the Evaluation of Antioxidants: Bulk Oils vs. Emulsions in 1. Agric. Food Chem. 42, 1054-l 059 35 Toyosaki, T., Yamamoto, A. and Mineshita, T. (1987) Antioxidant Effect of Riboflavin Tetrabutyrate in Emulsions in 1. FoodSci. 52, 1387-1390

90

Trends

in Food Science

& Technology

March

1996

[Vol. 71

36 Cillard, I., Cillard, P., Cormier, M. and Girre, L. (1980) u-Tocopherol Pro-oxidant Effect in Aqueous Media: Increased Autoxidation Rate of Linoleic Acid in /. Am. Oil Chem. Sot. 57,252-255 37 Machlin, L.J. (1995) Critical Assessment of the Epidemiological Data Concerning the Impact of Antioxidant Nutrients on Cancer and Cardiovascular Disease in Crit. Rev. Food Sci. Jechnol. 35, 41-50 38 Perly, B., Smith, I.C.P., Hughes, L., Burton, C.W. and Jngold, K.U. (1985) Estimation of the Location of Natural a-Tocopherol in Lipid Bilayers by C-NMR Spectroscopy in Biochim. Biophys. Acta 819,131-l 35 39 Cillard, J. and Cillard, P. (1986) Inhibitors of the Pro-oxidant Activity of a-Tocopherol in /. Am. Oil Chem. Sot. 63, 1165-l 169 40 Buettner, C.R. (1993) The Pecking Order of Free Radicals and Antioxidants: Lipid Peroxidation, a-Tocopherol and Ascorbate in Arch. Biochem. Biophys. 300,535-543 41 Ke, P.J. and Ackman, R.G. (1973) Bunsen Coefficient for Oxygen in Marine Oils at Various Temperatures Determined by an Exponential Dilution

42

43

44

45 46

Method with a Polarographic Oxygen Electrode in 1. Am. Oil Chem. Sot. 50, 429-435 Marcuse, R. and Fredriksson, P.O. (1968) Fat Oxidation at tow Oxygen Pressure. I. Kinetic Studies on the Rate of Fat Oxidation in Emulsions in /. Am. Oil Chem. Sot. 45,400-+07 Marcuse, R. and Fredriksson, P.O. (1969) Fat Oxidation at Low Oxygen Pressure. II. Kinetic Studies on Linoleic Acid Oxidation in Emulsions in the Presence of Antioxidants in /. Am. Oil Chem. Sot. 468262-260 McNuJty, P.B. (1987) Flavour Release - Elusive and Dynamic in Food Structure and Behaviour (Blanshard, J.M.V. and Lillford, P., eds), pp. 245-258, Academic Press Sims, R. (1994) Oxidation of Fats in Food Products in Inform 5, 1020-l 028 Coupland, J.N., Zhu, Z., McClements, DJ., Nawar, W.W. and Chinachoti, P. Droplet Composition Affects the Rate of Oxidation of Emulsified Ethyl Linoleate rn /. Am. Oil Chem. Sot. fin press)

Viewpoint

Preservation microbiology
Bacteria present in food systems are frequently stresses that are purposefully barriers to their growth. are pointing role in preventing placed Studies using modern of pathogenic subjected to genetic tools bacteria, in in the food systems as such barriers play a vital

and safety: Evidence that stress enhances virulence and triggers adaptive mutations
Douglas1. Archer

to the fact that while the outgrowth

certain cases, the virulence

of the stressed pathogens may in-

crease. In short, the bacteriums environment may arm that bacterium to survive similar stresses encountered in the human host. In this Viewpoint ability adaptive of environmental mutations article, I wish to focus on the bacterial virudriving force promoting stress to modulate

lence, and also to act as a potential even more virulent.

that may serve to select strains that are

In a recent Viewpoint article in this journal, Knochel and Gould addressed the fact that consumers are demanding foods that are more convenient, less processed, and more natural, and also that the use in foods of multiple barriers to bacterial growth is affecting the physiology of the bacteria present in the food, including pathogenic bacteria. I would like to concur with Knochel and Goulds statements regarding the effects of stresses on the physiology of foodbome pathogenic bacteria, but with one variation. In the summary section of their article, they pose the question, With such a reduction taking place in the preservation of foods, how are we going to achieve a real reduction in the incidence of food

Douglas L. Archer
University

is at the Food Science and Human FL 3261 l-0370,

Nutrition

Department, 157;

of Florida, Gainesville,

USA (fax: +1-904-846-l

e-mail: fos@gnv.ifas.ufl.edu).

poisoning from manufactured foods? I wonder if a reduction in preservation might not in fact lead to a reduction in the immediate virulence of certain pathogens, and, additionally, to a lowering of the rate of emergence of new or better host-adapted pathogens. The use in foods of barriers to microbial growth has served as an excellent means of preventing certain foodborne diseases such as staphylococcal food poisoning and botulism. In the case of such diseases, the toxinproducing bacteria need to achieve relatively high numbers in order for sufficient toxin to be formed and hence human intoxication to occur. The prevention of the growth of either Staphylococcus aureus or Clostridium botulinum was synonymous with keeping the food safe. Barriers with a long history of use include lowered temperatures, natural or human-directed acidification,
01996, Elsewer Science Ltd

Trends

in Food Science

& Technology

March

1996

[Vol.

71

91