Some Food Gels

Some Food Gels The fairly theoretical points discussed earlier will now be illustrated by discussing some food gels. 3.5.3.1 Plastic Fats [60,77] If a triacylglycerol oil is cooled, fat crystals can form and these will aggregate. There is van der Waals attraction between the crystals and no repulsion (except at very small distance). The crystals form a fractal gel, at least under some conditions [75]. Subsequently, the aggregated crystals may to some extent grow together (fuse) by local accretion of solid material, because it may take a long time before fat crystallization has reached an equilibrium state. This results in a rigid and brittle network. The modulus is fairly high (e.g., 106 Pa), the proportionality between stress and strain extends only for a strain of at most 0.01, and yielding occurs at a strain of about 0.1; the yield stress in this case is about 105 Pa. These values greatly increase as the fraction solid increases. Permeability ranges from about 10-16 to 10-13 m2 , according to crystal size and fraction solid. These properties are characteristic of what is known as a plastic fat. 3.5.3.2 Caseinate Gels [70,82,86] Milk contains casein micelles, proteinaceous aggregates of about 120 nm average diameter, each containing some 104 casein molecules (see Chap. 14). The micelles can be made to aggregate, by lowering the pH to about 4.6 (thereby decreasing electric repulsion) or by adding a proteolytic enzyme that removes the parts of k-casein molecules that protrude into the solvent (thereby decreasing steric repulsion). Fractal gels are formed with a fractal dimensionality of about 2.3. The permeability, which is about 2 × 10-13m2 for average casein concentration (c), is thus strongly dependent on the latter, being about proportional to c -3. For casein gels (Fig. 21a) a linear relation exists between log modulus and log casein concentration, in accordance with their fractal nature. The different slopes imply a difference in structure. Apparently, the initially tortuous strands in rennet gels become straightened soon after formation, whereas the strands in acid gels remain tortuous. The building blocks of the gel, that is, the casein micelles, are themselves deformable, and FIGURE 21 Dependence of the modulus of various g els on mass concentration of g elling material. Note that some of the abscissas are log arithmic, others linear. (a) Casein g els, made by slow acidification or by renneting . (b) Ag ar and g elatin g els. (c) Gels of k-carrag eenan of two molar masses in 0.1 molar KCI. (d) Heat-set g els of bovine serum albumin; the fig ures near the curves denote pH/added NaCl (molar). Approximate results, only meant to illustrate trends. From various sources. Pag e 130 the junctions between them are flexible. Thus, the gel is rather weak and deformable. For acid casein gels, fracture stress is about 100 Pa, and fracture strain about 1.1; for rennet gels, these values are about 10 Pa and 3, respectively. The acid gel is thus shorter (more brittle). These results apply to long time scales, say 15 min; at shorter time scales, the fracture stress is much greater. Applying a stress of slightly over 10 Pa to a rennet gel will cause flow (there is no detectable yield stress), and after considerable time fracture will occur. Applying a stress of 100 Pa leads to fracture within 10 sec. Similar behavior is found for some other types of particle gels but is by no means universal. All these values depend on conditions applied, especially temperature. It is seen that the modulus of a casein gel is larger at lower temperature (Fig. 22a). This may appear strange, since hydrophobic bonds between casein molecules are considered to play a major part in keeping the gel together, and these bonds decrease in strength with decreasing temperature. Presumably, a decrease of hydrophobic bond strength (low T) leads to swelling of the micelles, to a large contact area between adjoining micelles, and to a greater number of bonds per junction. Conversely, a higher temperature of gel formation leads to a larger modulus (at least for acid gels; Fig. 22a), and this is due to a somewhat different network geometry, not to a difference in type of bonds. At temperatures above about 20°C, rennet gels show syneresis. Syneresis accompanies rearrangement of the network of particles, which implies that some deaggregation occurs. In a region where no liquid can be expelled, that is, internally in the gel, rearrangement occurs as well, leading to both denser and less dense regions. This is called microsyneresis; it goes along with an increase in permeability and causes the straightening of the network strands mentioned above. 3.5.3.3 Gelatin [10,34] Of all food gels, gelatin is closest to an ideal entropic gel. The flexible molecular strands between cross-links are long and this causes the gel to be very extensible. It is also predominantly elastic because the cross-links are fairly permanent (at least at low temperature). The dependence of modulus on concentration (Figure 21b) is in reasonable agreement with Equation 19 (data not FIGURE 22 Dependence of the modulus of various g els on temperature of measurement. Arrows indicate the temperature sequence. (a) Acid casein g els (2.5%); fig ures on curves denote temperature of formation and ag eing . (b) Gelatin (2.5%); the broken line indicates the relation according to Equation 19. (c) k-carrag eenan (1%); fig ures on curves denote concentration of CaCl 2. (d) b-Lactog lobulin (10%) at two pH values. Data represent trends; results may vary g reatly depending on heating or cooling rate and other conditions. After various sources. Pag e 131 shown), but the temperature dependence is not (see Fig. 22b). This discrepancy stems from the mechanism of cross-linking. Despite the severe treatment of the collagen during preparation of gelatin, the molecules retain much of their length and produce highly viscous aqueous solutions. Upon cooling, the molecules tend to form triple helices (broadly speaking proline helices) like those in collagen. This applies to only part of the gelatin, and the helical regions are relatively short. Presumably, a gelatin molecule sharply bends at a so-called b-turn, and then forms a short double helix. Subsequently, a third strand may wind around this helix, thereby completing it. If the third strand is part of another molecule, a cross-link is formed; if not, there is no crosslink. Regardless of how this develops, groups of triple helices tend to align themselves in a parallel array, forming microcrystalline regions, and these constitute cross-links (Fig. 23a). It would follow from this mechanism that the kinetics of gelation are intricate, which is indeed the case. Also, rheological properties can be strongly dependent on temperature history. 3.5.3.4 Polysaccharides [40,49] Despite the wide range of polysaccharide types (see Chap. 4), some general rules governing their gelling behavior exist. Unlike gelatin, most polysaccharide chains are fairly stiff; appreciable bending can occur only if chain length exceeds about 10 monomers (monosaccharide residues). This characteristic causes polysaccharides to produce highly viscous solutions; for example, 0.1% xanthan will increase the viscosity of water by at least a factor of 10. Some polysaccharides can form gels. Broadly speaking, the gel cross-links consist of microcrystalline regions (see Fig. 17b), involving an appreciable portion of the material. This means that the length of strands between cross-links is not very long. Combined with chain stiffness, this leads to rather short gels, very unlike rubber gels. Actually, they are intermediate between entropic and enthalpic gels. Of course, there is considerable variation among polysaccharides in this regard. Cross-links among polysaccharide molecules can be any of the following three types, each illustrated by one example: Type 1. Single helices. These helices, as found in amylose, arrange themselves in microcrystalline regions and, if the concentration is sufficiently great, gelation will occur. With amylopectin, similar behavior is observed. Type 2. Double helices. These occur in k-carrageenan below a sharply defined temperature. Each helix generally involves one molecule. Double helix formation between two molecules is very unlikely for geometric reasons. As soon as even a very short helix would have formed, the rotational freedom of the remaining part of each molecule involved would be very much reduced. Consequently, double helices involving FIGURE 23 Various types of junctions in polymer g els. (a) Partly stacked triple helices as in g elatin. (b) Stacked double helices as in carrag eenans. (c) “Eg g box” junction as in alg inate, dots denote calcium ions. (d) Swollen starch g rains in a concentrated starch g el. Helices are schematically indicated by crosslines. The scales of a-c and d are very different. Pag e 132 two molecules will be rare. Cross-links arise because the stiff helices form microcrystalline regions (Fig. 23b). Helix formation is very rapid (milliseconds), whereas gelation takes longer (several seconds). As soon as the helices “melt,” so does the gel. Type 3. “Egg-box” junctions. These occur with some charged polysaccharides, such as alginate, when divalent cations are present (Fig. 23c). Alginate has negative charges, often spaced at regular distances, allowing divalent cations, such as Ca2+, to establish bridges between two parallel polymer molecules. In this way, fairly rigid junctions are formed. It is likely that the junctions further arrange themselves in microcrystalline regions. The junctions do not readily “melt” unless the temperature is near 100°C. Many factors may affect gelation and gel properties of polysaccharides. These include molecular structure, molar mass (Fig. 21c), temperature (Fig. 22c), solvent quality, and, for polyelectrolytes, pH and ionic strength (Fig. 22c). 3.5.3.5 Globular Proteins [11,56,57] Several globular proteins form irreversible heat-setting gels (Fig. 22d). The proteins denature, that is, unfold, and then mutually react. The gel bonds formed may be -S-S- linkages (or possibly other covalent bonds), salt bridges, and/or hydrophobic bonds. A fairly large concentration of protein is needed to obtain stiff gels (Fig. 21d). The structures, and thereby rheological properties of the gels, vary greatly with type of protein, pH, ionic strength and rate of heating. Although proteins are polymers, the gels are not typical polymer gels, nor are they quite like casein gels. At pH values far from the isoelectric pH, fairly weak fine-stranded gels are typically formed. The strands of these gels have a thickness of ~ 10 nm and their structure is somewhat irregular. Near the isoelectric pH, stiffer gels form and they look like particle gels. The particles are about 1 mm in diameter or more, and thus contain numerous molecules, at least 106 . These gels are often fractal-like [74]. The two types should differ greatly in permeability, though substantiating determinations appear not to have been made. Structure formation during extrusion is comparable to heat-setting of globular proteins. An important example involves proteinrich products from soya [35]. Spinning of proteins, however, may involve different mechanisms [73]. 3.5.3.6 Concentrated Starch Gels [30] Starch (see Chap. 4) occurs naturally in rigid granules, mostly between 5 and 100 mm in diameter, that are insoluble in water. Part of the amylopectin is present in microcrystalline regions, which gives the granules considerable rigidity. If starch granules are heated in excess water, they gelatinize. This involves swelling, as water up to several times their own weight is taken up; melting of microcrystallites; and leaching of some amylose. The swollen granules remain intact upon cooling, unless vigorous agitation is applied. Dilute gelatinized starch suspensions gel in roughly the same way as does amylose; that is, leached amylose forms a gel network in which the swollen granules are trapped. A concentrated, gelatinized starch solution—above 5–15%, depending on starch type—forms a very different kind of gel. The granules swell until they virtually fill the whole volume, deforming each other in the process (Fig. 23d). Especially for potato starch, the swollen granules interlock, as in a jig-saw puzzle. A very thin layer of gelled amylose solution is present between the granules and may act as intergranule glue. The rheological properties of the gel are dominated by those of the granules. A freshly made gel can be deformed very much and it is purely elastic (like a rubber ball) if deformation is very fast. Slower deformation may lead to fracture. Presumably, the individual granules are rubber-like. During storage of concentrated starch gels, considerable rearrangement of the molecules Pag e 133 occurs. This is like partial crystallization and happens faster at a lower nonfreezing temperature. This leads to an increase in number and particularly strength of microcrystalline junctions. The granules and the gel become stiffer (the modulus increases). It also results in an increase of the fracture stress and a decrease of the fracture strain of the gel upon aging. These “retrogradation” phenomena are of great importance in the staling of bread. 3.5.3.7 Mixed Gels It should now be clear that the structure and properties of gels can vary greatly. The modulus of a 1% gel can vary by almost 5 orders of magnitude and the strain at fracture by a factor of 100. The preceding discussion of starch gels gives one example of how the various rheological and fracture properties can be related to each other. In that case most relations can be explained, even semiquantitatively. However, nearly every system exhibits specific relations that are often poorly understood. The situation becomes even more complicated when mixed gels are considered. Some mixtures of polysaccharides may show enhanced gelation at a lower concentration [9,41]. For instance, dilute xanthan or locust bean gum solutions do not show an appreciable yield stress, but dilute mixed solutions do. Filled systems may also have greatly altered properties from unfilled [9,38]. Phase separation can also occur if the polymers are thermodynamically incompatible. For instance, mixed solutions of a highly soluble polysaccharide and a protein can separate into two phases, one rich in protein and the other rich in polysaccharide [61,62]. In other cases, a mixed polymer solution forms a coacervate, that is, a phase containing a high concentration of both, leaving another phase that is depleted of polymers. This can occur if the polymers have opposite charge. Complete foods are still more complicated than the systems so far discussed [26]. Nevertheless, knowledge of the principles provided can be of great help in understanding food behavior and for designing experiments to study it. 3.6 Emulsions 3.6.1 Description Emulsions are dispersions of one liquid in another. The most important variables determining emulsion properties are: 1. Type, that is, o/w or w/o. This determines, among other things, with what liquid the emulsion can be diluted (Section 3.1.2). The o/w emulsions are most common, examples being milk and milk products, sauces, dressings, and soups. Butter and margarine are w/o emulsions, but they contain other structural elements as well. 2. Droplet size distribution. This has an important bearing on physical stability, smaller drops generally giving more stable emulsions. Also the energy and the amount of emulsifier needed to produce the emulsion depend on the droplet size desired. A typical mean droplet diameter is 1 mm, but it can range between 0.2 and several micrometers. Because of the great dependence of stability on droplet size, the width of the size distribution is also important. 3. Volume fraction of dispersed phase (j). In most foods, j is between 0.01 and 0.4. For mayonnaise it may be 0.8, which is above the value for maximum packing of rigid spheres, roughly 0.7; this means that the oil droplets are somewhat distorted. 4. Composition and thickness of the surface layer around the droplets. This determines interfacial tension, colloidal interaction forces, etc. (Section 3.2.7). Pag e 134 5. Composition of the continuous phase. This determines solvent conditions for the surfactant and thereby colloidal interactions. The viscosity of the continuous phase has a pronounced effect on creaming. Unlike the solid particles in a suspension, emulsion droplets are spherical (greatly simplifying many predictive calculations) and deformable (allowing droplet disruption and coalescence), and their interface is fluid (allowing interfacial tension gradients to develop). Nevertheless, in most conditions, emulsion droplets behave like solid particles. From Equation 9, the Laplace pressure of a droplet of 1 mm radius and interfacial tension g=5 mN.m-1 is 104 Pa. For a liquid viscosity of h=10-3 Pa.sec (water) and a velocity gradient achieved by stirring of G=106 sec-1 (this is extremely vigorous), the shear stress hG acting on the droplet would be 103 Pa. This implies that the droplet would be deformed only slightly. Also, the surfactant at the droplet surface allows this surface to withstand a shear stress (Section 3.2.6). For the conditions mentioned, an interfacial tension difference between two sides of the droplet of 1 mN . m-1 would suffice to prevent lateral motion of the interface, and a difference of this magnitude can be achieved readily. It can be concluded that emulsion droplets behave like solid spheres, unless stirring is extremely vigorous or droplets are very large. 3.6.2 Emulsion Formation [76,80] In this section the size of the droplets and the surface load obtained during making of emulsions are discussed, especially when protein is used as a surfactant. To make an emulsion, one needs oil, water, an emulsifier (i.e., a surfactant), and energy (generally mechanical energy). Making drops is easy, but to break them up into small droplets mostly is difficult. Drops resist deformation and thereby break up because of their Laplace pressure, which becomes larger as droplet size decreases. This necessitates a large input of energy. The energy needed can be reduced if the interfacial tension, hence the Laplace pressure, is reduced by adding an emulsifier, though this is not the latter’s main role. The energy needed to deform and break up droplets is generally provided by intense agitation. Agitation can cause sufficiently strong shear forces if the continuous phase is very viscous. This is common when making w/o emulsions, resulting in droplets with diameters down to a few micrometers (which is not very small). In o/w emulsions the viscosity of the continuous phase tends to be low, and to break up droplets inertial forces are needed. These are produced by the rapid, intensive pressure fluctuations occurring in turbulent flow. The machine of choice is the high-pressure homogenizer, which can produce droplets as small as 0.2 mm. The average droplet size obtained is about proportional to the homogenization pressure to the power -0.6. When using high-speed stirrers, faster stirring, longer stirring, or stirring in a smaller volume result in smaller droplets; however, average droplet diameters below 1 or 2 mm usually cannot be obtained. However, this is not the whole story. Figure 24 shows the various processes that occur during emulsification. Besides disruption of droplets (Fig. 24a), emulsifier has to be transported to the newly created interface (Fig. 24b). The emulsifier is not transported by diffusion but by convection, and this occurs extremely quickly. The intense turbulence (or the high shear rate, if that is the situation) also leads to frequent encounters of droplets (Fig. 24c and d). If these are as yet insufficiently covered by surfactant, they may coalesce again (Fig. 24c). All these processes have their own time scales, which depend on several conditions, but a microsecond is fairly characteristic. This means that all processes occur numerous times, even during one passage through a homogenizer valve, and that a steady state—where break-up and coalescence balance each other—is more or less obtained. The role of the emulsifier is not yet fully clear, but it is fairly certain that the Gibbs- Pag e 135 FIGURE 24 Important processes occurring during emulsification. The drops are depicted by thin lines and the emulsifier by heavy lines and dots. Hig hly schematic and not to scale. Marangoni effect is crucial (Fig. 25). If two drops move toward each other (Fig. 25a), which they frequently do at great speed during emulsification, and if they are insufficiently covered by emulsifier, they will acquire more emulsifier during their approach, but the least amount of emulsifier will be available where the film between droplets is thinnest. This lead to an interfacial tension gradient, with interfacial tension (g) being largest where the film is thinnest (Fig. 25b). FIGURE 25 Diag ram of the Gibbs-Marang oni effect acting on two approaching emulsion droplets during emulsification. Emulsifier molecules indicated by Y. See text for further discussion. (After Ref. 80.) Pag e 136 The gradient causes the emulsifier, and thereby the droplet surfaces, to move in the direction of large g, carrying liquid along with it (Fig. 25b). This is the Marangoni effect (see also Fig. 9). The streaming liquid will thus drive the droplets away from each other (Fig. 25c), thereby providing a self-stabilizing mechanism. The magnitude of the effect depends on the Gibbs elasticity of the film, which is twice the surface dilational modulus (Sec. 3.2.5). The Gibbs elasticity generally increases with increasing molar concentration of surfactant in the continuous phase. It therefore decreases during emulsification, because the interfacial area increases, whereby the adsorbed amount of surfactant increases. The Gibbs-Marangoni effect occurs only if the emulsifier is in the continuous phase, since otherwise a g gradient cannot develop. This is the basis of Bancroft’s rule: The phase in which the emulsifier is (most) soluble becomes the continuous phase. Hence the need for a surfactant with high HLB number for o/w emulsions and a low HLB for w/o emulsions (Sec. 3.2.2). Proteins are the emulsifiers of choice for o/w food emulsions because they are edible, surface active, and provide superior resistance to coalescence [78]. They cannot be used for w/o emulsions because of their insolubility in oil. Proteins do not give a very low interfacial tension (Section 3.2.2, Fig. 5) and their Gibbs-Marangoni effect is not very strong, presumably because of their low molar concentration. Therefore, the droplets obtained are typically not very small for given conditions. However, droplets can be made smaller by applying more intense emulsification, such as a greater homogenization pressure. Some examples of average droplet size (dvs) are given in Figure 26. At large emulsifier concentration, dvs reaches a plateau value. This value is smaller for the nonionic surfactant than for the proteins, because the former produces a lower interfacial tension. The nonionic is also more effective at low concentration than the proteins, presumably because of a stronger Gibbs-Marangoni effect. It is also seen that the various proteins give about the same plateau value for dvs. This is not strange because they produce comparable values for the interfacial tension (about 10 mN.m-1). However, at low concentrations large differences in dvs are apparent. Several tests FIGURE 26 Influence of emulsifier concentration (%w/w) on volume/surface averag e droplet diameter dvs for various emulsifiers. Emulsions contained about 20% triacylg lycerol oil in water and were prepared under constant conditions (moderate emulsification intensity). B, blood protein; C, sodium caseinate; N, nonionic small-molecule surfactant; S, soya protein; W , whey protein. Approximate results from various sources. The broken line indicates approximately the conditions during determination of the so-called emulsifying activity index. Pag e 137 have been developed to evaluate the suitability of proteins as emulsifiers. The well-known emulsifying activity index (EAI) involves emulsifying a large quantity of oil in a dilute protein solution [46]. This test corresponds roughly to the conditions indicated by the broken line in Figure 26. This is not realistic for most practical situations, because the protein/oil ratio normally would be much larger. Consequently, EAI values are often irrelevant. Attempts have been made to explain differences in EAI of various proteins by differences in their surface hydrophobicity [29]. However, the correlation is poor and other workers have refuted this concept [e.g., 52]. In the author’s view, proteins differ primarily in emulsifying efficiency because they differ in molar mass. For a larger molar mass and the same mass concentration, the molar concentration is smaller, and the latter is presumably the most important variable in providing a strong GibbsMarangoni effect. Proteins of a smaller molar mass would thus be more efficient emulsifiers. (However, a very small molar mass, as obtained, for instance, by partial hydrolysis of the protein, would not be desirable, because emulsions obtained with fairly small peptides usually show rapid coalescence.) It should be realized that several protein preparations, especially industrial products, contain molecular aggregates of various size, thereby greatly increasing effective molar mass and decreasing emulsifying efficiency. By and large, protein preparations that are poorly soluble are poor emulsifiers. As a rule, well-soluble proteins are about equally suitable to make an emulsion (i.e., obtain small droplets), if present at not too low a concentration. Another important variable is the surface load (G). If an emulsifier tends to give a high G, relatively much of it is needed to produce an emulsion. Compare, for instance, whey protein and soya protein in Figures 26 and 27. Moreover, a fairly high G is usually needed to obtain a stable emulsion. In the case of small-molecule surfactants, equilibrium is reached between G and bulk concentration of surfactant according to the Gibbs relation (Eq. 5). Consequently, knowledge of total surfactant concentration, of o/w interfacial area, and of the adsorption isotherm (e.g., Fig. 5, lower part) allows calculation of G, irrespective of the manner of formation of the emulsion. This FIGURE 27 Protein surface load ( G) as a function of protein concentration (c) per unit oil surface area (A) created by emulsification. The broken line indicates 100% adsorption. (From Ref. 85.) Pag e 138 is not the case when a protein (or other polymer) is the emulsifier, because thermodynamic equilibrium is not reached (Sec. 3.2.2). The surface load of a protein can depend on the way of making the emulsion, in addition to the variables mentioned. It has been observed that plots as given in Figure 27 are more suitable to relate G to protein concentration. Figure 27 gives some examples of results obtained for various proteins. If c/A is very small, some proteins presumably almost fully unfold at the o/w interface, and form a stretched polypeptide layer, with a G of about 1 mg.m-2. Many highly soluble proteins give a plateau value of about 3 mg.m-2. Aggregated proteins can yield much larger values. It may also be noted that any large protein aggregates present tend to become preferentially adsorbed during emulsification, thereby further increasing G. An emulsifier is needed not only for the formation of an emulsion, but also for providing stability of the emulsion once made. It is of importance to clearly distinguish between these main functions, since they often are not related. An emulsifier may be very suitable for making small droplets, but may not provide long-term stability against coalescence, or vice versa. Evaluation of proteins merely for their ability to produce small droplets is therefore not very useful. Another, usually desirable, feature of a surfactant is to prevent aggregation under various conditions (pH near the isoelectric pH, high ionic strength, poor solvent quality, high temperature). Types of emulsion instability and means of prevention are discussed next. 3.6.3 Types of Instability [17,43,59,78,83] Emulsions can undergo several types of physical change as illustrated in Figure 28. The figure pertains to o/w emulsions; the difference with w/o emulsions is that downward sedimentation rather than creaming would occur. Ostwald ripening (Fig. 28a) does not normally occur in o/w emulsions, because triacylglycerol oils are commonly used and they are insoluble in water. When essential oils are present (e.g., citrus juices), some have sufficient solubility so that smaller droplets gradually disapFIGURE 28 Types of physical instability for oil-in-water emulsions. Hig hly schematic. The size of the contact area denoted in (d) may be g reatly exag g erated; the short heavy lines in (e) denote triacylg lycerol crystals. Pag e 139 pear [19]. W/o emulsions may exhibit Ostwald ripening. Data in Table 4 show only a very small solubility excess for a 1-µm droplet, but it would be sufficient to produce marked Ostwald ripening during prolonged storage. This can easily be prevented by adding a suitable solute to the water phase, that is, one that is insoluble in oil. A low concentration of salt (say, NaCl) will do: As soon as a small droplet shrinks, its salt concentration and osmotic pressure increase, thereby producing a driving force for water transport in the opposite direction. The net result is a stable droplet size distribution. The other instabilities are discussed in other sections: creaming in 3.4.2, aggregation in 3.3 and 3.4.3, coalescence in 3.6.4, and partial coalescence in 3.6.5. The various changes may affect each other. Aggregation greatly enhances creaming, and if this occurs, creaming further enhances aggregation rate, and so on. Coalescence can only occur if the droplets are close to each other, that is, in an aggregate or in a cream layer. If the cream layer is more compact, which may occur if individual, fairly large droplets cream, coalescence will be faster. If partial coalescence occurs in a cream layer, the layer may assume characteristics of a solid plug. It is often desirable to establish what kind of instability has occurred in an emulsion. Coalescence leads to large drops, not to irregular aggregates or clumps. Clumps due to partial coalescence will coalesce into large droplets when heated sufficiently to melt the triacylglycerol crystals. A light microscope can be used to establish whether aggregation, coalescence, or partial coalescence has occurred. Section 3.4.4 gives some hints for distinguishing among various causes of aggregation. It is fairly usual that coalescence or partial coalescence leads to broad size distributions, and then the larger droplets or clumps cream very rapidly. Agitation can disturb creaming and may disrupt aggregates of weakly held droplets, but not clumps formed by partial coalescence. Slow agitation tends to prevent any true coalescence. If air is beaten in an o/w emulsion, this may lead to adsorption of droplets onto air bubbles. The droplets may then be disrupted into smaller ones, due to spreading of oil over the a/w interface (Sec. 3.2.3). If the droplets contain crystalline fat, clumping may occur; beating in of air thus promotes partial coalescence. This is what happens during churning of cream to make butter and also during whipping of cream. In the latter case, the clumped, partially solid droplets form a continuous network that encapsulates and stabilizes the air bubbles and lends stiffness to the foam. A way to prevent or retard all changes except Ostwald ripening is to cause the continuous phase to gel (Sec. 3.5). Examples are butter and margarine. Here, the water droplets are immobilized by a network of fat crystals. Moreover, some crystals become oriented at the oil/water interface because of the favorable contact angle (Sec. 3.2.3). In this way, the droplets cannot closely encounter each other. If the product is heated to melt the crystals, the aqueous droplets readily coalesce. Often, a suitable surfactant is added to margarine to prevent rapid coalescence during heating, since this would cause undesirable spattering. 3.6.4 Coalescence [21,78,83] This discussion will focus on o/w emulsions. The theory is still in a state of some confusion. As a preliminary remark it should be noted that factors governing coalescence during emulsion formation (Sec. 3.6.2) differ completely from those that are important in “finished” emulsions. The latter are considered here. Coalescence is induced by rupture of the thin film (lamella) between close droplets. If a small hole is somehow formed in the film, the droplets will immediately flow together. Film rupture is a chance event and this has important consequences: (a) The probability of coalescence, if possible, will be proportional to the time that the droplets are close to each other. Hence, Pag e 140 it is especially likely in a cream layer or in aggregates. (b) Coalescence is a first-order rate process, unlike aggregation, which is in principle second order with respect to time and concentration. (c) The probability that rupture of a film occurs will be proportional to its area. This implies that flattening of the droplets on approach leading to formation of a greater film area, will promote coalescence. Oil droplets normally found in food emulsions do not show such flattening, because their Laplace pressures are too high (Fig. 28d is thus misleading in this respect). There is no generally accepted coalescence theory, but the author feels that the following principles are valid. Coalescence is less likely for: 1. Smaller droplets. They lead to a smaller film area between droplets, and hence a lower probability of rupture of the film; more coalescence events are needed to obtain droplets of a certain size; the rate of creaming is decreased. In practice, average droplet size often is the overriding variable. 2. A thicker film between droplets. This implies that strong or far-reaching repulsive forces between droplets (Sec. 3.3) provide stability against coalescence. For DLVO-type interactions, it turns out that coalescence will always occur if the droplets are aggregated in the primary minimum (Fig. 11). Steric repulsion is especially effective against coalescence, because it keeps the droplets relatively far apart (Fig. 12). 3. A greater interfacial tension. This may appear strange, because a surfactant is needed to make an emulsion and a surfactant decreases g. Moreover, a smaller g implies a smaller surface free energy of the system. However, it is the activation free energy for film rupture that counts and that is larger for a larger g. This is because a larger g makes it more difficult to deform this film (bulging, development of a wave on it), and deformation is needed to induce rupture. Based on these principles, proteins appear to be very suitable for preventing coalescence, and this agrees with observation. Proteins do not produce a very small g, and they often provide considerable repulsion, both electric and steric. Figure 29 shows results of experiments in which FIGURE 29 Time (tc) for coalescence of droplets of various diameter (d) with a planar o/w interface, in 20-min-old 1 ppm protein solutions. , b-Casein; , k-casein; lysozyme. (From Ref. 21.) Pag e 141 small droplets in a dilute protein solution were allowed to cream to planar o/w interfaces of various age and the time needed for coalescence was observed. A strong effect of droplet size is apparent. The results were obtained under conditions (protein concentration and adsorption time) where the surface load of the proteins would have been only about 0.5 mg.m-2, implying very weak repulsion. In cases where a thicker adsorbed layer was allowed to form, the authors found virtually no coalescence of small droplets. Figure 29 shows no significant difference between proteins in their ability to prevent coalescence. This is also the general experience in practice, except for gelatin, which is less effective than most other proteins. Under severe conditions (see later), differences among proteins may be observed, with caseinates tending to be superior. Partial hydrolysis of proteins can severely impair their ability to prevent coalescence. Attempts have been made to relate the coalescence inhibiting ability of proteins (and of other surfactants) to various properties, particularly surface shear viscosity of the adsorbed protein layer (Section 3.2.5). Unfortunately, this quantity often is named “film strength,” a misleading term. In some cases a positive correlation between surface shear viscosity and coalescence stability is observed, but there are several instances where deviation from this relationship are very large. Some positive correlations may have been due to a relation between surface shear viscosity and surface load. Most small-molecule surfactants yield a small interfacial tension. Because a small g favors coalescence, surfactants that provide considerable steric repulsion, such as the Tweens© , are among the most effective. Ionic surfactants are effective against coalescence only at very low ionic strength. Small-molecule surfactants present in (added to) protein-stabilized emulsions tend to displace protein from the droplet surface (Sec. 3.2.2, Fig. 6), and this generally leads to less resistance to coalescence. If coalescence is desired, this provides a method to achieve it; for example, add sodium lauryl sulfate and some salt (to lower double layer thickness), and rapid coalescence will usually occur. Food emulsions may exhibit coalescence under extreme conditions. For example, during freezing, formation of ice crystals will force the emulsion droplets closer together, often causing copious coalescence on thawing. Something similar happens upon drying and subsequent redispersion; here coalescence is alleviated by a relatively high concentration of “solids not fat.” In such cases, best stability is obtained by having small droplets and a thick protein layer, such as of Na caseinate. Another extreme condition is centrifuging. This causes a cream layer to form rapidly, a pressing of droplets together with sufficient force to cause considerable flattening even of small droplets, and likely coalescence. This implies that centrifugation tests to predict coalescence stability of emulsions are usually not valid, since the conditions during centrifugation differ so greatly from those during handling of emulsions. (This does not mean that centrifugation tests to predict creaming are useless. They can be quite helpful, provided that the complications discussed in Sec. 3.4.2 are taken into account.) Predicting coalescence rate is always very difficult. The best approach is to use a sensitive method to estimate average droplet size (e.g., turbidity at a suitable wavelength) and establish the change over time (perhaps 1 day). Even then, extrapolation may result in error: protein layers around droplets tend to age (Sec. 3.2.2) and become better inhibitors of coalescence. 3.6.5 Partial Coalescence [5–7,12,64] In many o/w food emulsions part of the oil in the droplets can crystallize. The proportion of fat solid, y, depends on the composition of the triacylglycerol mixture and on temperature (Chap. 5). In emulsion droplets, y can also depend on temperature history, since a finely Pag e 142 emulsified oil can show considerable and long-lasting supercooling [77]. If an emulsion droplet contains fat crystals, they usually form a continuous network (Sec. 3.5.3). These phenomena greatly affect emulsion stability. Oil droplets containing a network of fat crystals cannot fully coalesce (Fig. 28e). If the film between them ruptures, they form an irregular clump, held together by a “neck” of liquid oil. True and partial coalescence thus have rather different consequences. Partial coalescence causes an increase in the apparent volume fraction of dispersed material, and if the original volume fraction is moderate or high and shear rate is fairly small, a solid or gel-like network of partially coalesced clumps can form. This is desired, for instance, in ice cream, where it gives the product a desirable texture (“dryness,” lack of stickiness, melt-down resistance). Rupture of the film between close droplets containing some solid fat can be triggered by a crystal that protrudes from the droplet surface and pierces the film. This happens particularly in a shear field, and then it may occur very much faster, for instance, about six orders of magnitude faster than true coalescence (same emulsion, no fat crystals). This implies that partial coalescence is far more important than true coalescence in o/w emulsions that are subject to fat crystallization. The kinetics of particle coalescence are complicated and variable. This is due to the numerous variables affecting it. In many emulsions, large particles (original droplets or clumps already formed) are more prone to partial coalescence than small ones, leading to a self-accelerating process, which soon leads to very large clumps that cream very rapidly. The remaining layer may then exhibit a decreased average droplet size. Other emulsions may simply show a gradual increase in average particle size with time. The most important factors affecting the rate of partial coalescence generally are (Fig. 30): 1. Shear rate. This has various effects. (a) Because of the shear, droplets encountering each other tend to roll over each other, thereby greatly enhancing the probability that FIGURE 30 Schematic depiction of rate of partial coalescence (Q) in protein-stabilized emulsions as influenced by: (a) shear rate G (sec -1); (b) volume fraction j; (c) proportion of fat solid y; (d) mean droplet diameter d ( mm); (e) protein surface load G (mg .m -2); and (f) concentration of added small-molecule surfactant c (%), on partial coalescence rate Q. Various sources. Pag e 143 a protruding crystal can pierce the film between them. (b) The encounter rate between particles is proportional to shear rate (Section 3.4.3). (c) The shear force tends to press approaching droplets closer together, thereby enhancing the possibility that a protruding crystal can pierce the film. Thus, shear rate has a very large effect on the rate of partial coalescence, and this influence is accentuated when flow is turbulent. 2. Volume fraction of droplets. For a higher j, the rate of partial coalescence is obviously greater, being about second order with respect to j. 3. Fat crystallization. If the fraction solid, y, is 0 partial coalescence is impossible, and it also can not occur for y=1. For fairly low y, partial coalescence rate generally increases with increasing y, as this causes more crystals to protrude. However, the relation between y and rate is variable, and the curve in Figure 30c only provides an example. The variability is largely due to variation in crystal size and arrangement. An important aspect is that the crystals must form a network throughout the droplet to support protruding crystals. The minimum y needed for such a network often is of the order 0.1. If most of the oil is crystallized and the crystals are very small, the crystal network may tenaciously hold the remaining oil, thereby preventing partial coalescence, even if the film is pierced. The protrusion distance may also depend on y, temperature history, crystal size, and crystal shape. 4. Droplet diameter. A relation as depicted in Figure 30d is usually observed, but the scale of droplet size may vary among emulsions. The effect of d presumably is due to (a) larger droplets sensing a larger shearing force; (b) larger droplets exhibiting a larger film area between two droplets; and (c) a positive correlation between droplet size and crystal size. 5. Surfactant type and concentration. Two effects are of major importance. First, these variables will determine the oil-crystalwater contact angle (Sec. 3.2.3) and thereby affect the distance a given crystal can protrude. Second, these variables determine repulsion (strength and range) between the droplets. The weaker the repulsion, the easier it is for two droplets to closely approach each other, thus increasing the likelihood that a protruding crystal will pierce the film between them. The repulsion, together with the droplet size, will thus determine what minimum shear rate is needed for partial coalescence to occur; values between 5 and 120 sec-1 have been observed. Some emulsions show no partial coalescence at all at the shear rates studied. The best type of surfactant to achieve this is, again, a protein, if the surface load is large enough (Fig. 30e). Addition of a smallmolecule surfactant generally leads to displacement of the protein from the interface (Sec. 3.2.2), thereby greatly enhancing partial coalescence (Fig. 30f). This approach is commonly used in ice cream mixes. 3.7 Foams In a sense, foams are much like o/w emulsions; both are dispersions of a hydrophobic fluid in a hydrophilic liquid. However, because of considerable quantitative differences, their properties also are qualitatively different. Quantitative information is given in Table 8. It is evident that bubble diameter is so large as to exclude foam bubbles from the realm of colloids. Large diameters, combined with large density difference, cause foam bubbles to cream faster than emulsion droplets by several orders of magnitude. The relatively high solubility of air in water causes rapid Ostwald ripening (often called disproportionation in foams). If the gas phase is CO2, as it is in some foods (bread, carbonated beverages), the solubility is even higher, by a factor of about 50. The characteristic time scales during formation are two or three orders of magnitude longer for Pag e 144 TABLE 8 Comparison of Foams and Emulsions: Order of Mag nitude of Some Quantities Property Foam Emulsion o/w Emulsion w/o Units Drop/bubble diameter 10-3 10-6 10-6 m Volume fraction 0.9 0.1 0.1 – Drop/bubble number 109 1017 1017 m-3 Interfacial tension 0.04 0.006 0.006 N.m-1 Laplace pressure 102 104 104 Pa Solubility D in C 2.2 0 0.15 vol% Density difference D-C -103 -102 102 kg.m-3 Viscosity ratio D/C 10-4 102 10-2 – Time scalea 10-3 10-6 10-5 sec Key: D, dispersed phase (air in the case of foam); C, continuous phase. aCharacteristic times during formation. foams than for emulsions. Because creaming and Ostwald ripening occur so fast, formation and instabilities often cannot be separated, which makes the study of foams rather difficult. Several aspects of foams will be briefly discussed here. A few have already been mentioned in Section 3.1.2. Surface phenomena are of overriding importance to foam production and properties, and background information is provided in Section 3.2. Further sources of literature are Refs. 23, 48, and 79. For churning and whipping of cream see Ref. 43. Also, some of the books mentioned in the bibliography, especially Dickinson’s, have chapters on foam. 3.7.1 Formation and Description In principle, foams can be made in two ways, by supersaturation or mechanically. 3.7.1.1 Via Supersaturation A gas, usually CO2 or N2O because of their high solubility, is dissolved in the liquid at high pressure (a few bars). When the pressure is released, gas bubbles form. They do not form by nucleation; a spontaneously formed gas bubble would have a radius of, say, 2 nm, which would imply a Laplace pressure (Eq. 9) of about 108 Pa, or 103 bar. To achieve this, the gas would have to be brought to this pressure, which is, of course, impractical. Instead, gas bubbles always grow from small air pockets that are already present at the wall of the vessel or on boiling chips. The contact angle gas/water/solid may be as high as 150 degrees for a fairly hydrophobic solid, and this allows small air pockets to remain in crevices or sharp dents in the solid (Fig. 7d). For a negative curvature, air could even remain there if less than saturated. To give an example, if a pressurized bottle of a carbonated liquid is opened, the overpressure is released, CO2 becomes supersaturated, and it diffuses toward small air pockets in the bottle wall. These grow, and become dislodged when large enough, leaving a remnant from which another bubble can grow. The bubbles rise while growing further, and a creamed layer of bubbles is formed, that is, a foam. These bubbles always are fairly large, say 1 mm. Another example is the formation of CO2 in a leavened dough. Excess CO2 collects at sites of tiny entrapped small air bubbles, and these sites grow in size. Some of them grow to form visible gas cells, creating a macroscopic foam structure. Pag e 145 3.7.1.2 By Mechanical Forces A gas stream can be led through narrow openings into the aqueous phase (sparging); this causes bubbles to form, but they are fairly large. Smaller bubbles can be made by beating air into the liquid. At first, large bubbles form, and these are broken up into progressively smaller ones. Shear forces are typically too weak to obtain small bubbles and the breakup mechanism presumably involves pressure fluctuations in a turbulent field, as is true during formation of o/w emulsions (Sec. 3.6.2). Bubbles of about 0.1 mm can be obtained in this way, which is the method of choice in industrial processing. The method also enables the amount of air incorporated to be controlled. This is often expressed as percent overrun, that is, the relative increase in volume, which is equal to 100j/(1-j), where j is the volume fraction of air (or gas) in the system. 3.7.1.3 Foam Formation and Structure To make a foam, a surfactant is needed. Almost any type will do, since the only criterion for its functionality is that a certain g gradient be created. This does not mean that any surfactant is suitable to make a stable foam, as will be discussed later. A fairly low concentration of surfactant suffices. During beating, a surface area of, say 105 m2 per m3 aqueous phase is produced. Assuming a surface load of 3 mg surfactant per m2 , about 0.1% of surfactant would be more than sufficient. Figure 31 illustrates the stages in foam formation after creation of initial bubbles. For most surfactants, Ostwald ripening will be substantial even during foam formation, implying that very small bubbles will soon disappear and the bubble size distribution will become fairly narrow. As soon as beating stops, bubbles rise rapidly and form a foam layer (unless liquid viscosity is quite high). The buoyancy force soon is sufficient to cause mutual deformation of bubbles, causing the formation of flat lamellae between them. The stress due to buoyancy is roughly equal to rwatergH, where H is the height in the foam layer, about 100 Pa for H=1 cm. However, there is marked stress concentration as spherical bubbles come into contact, and this means that bubbles with a Laplace pressure of 103 Pa would become significantly flattened. Further drainage of interstitial liquid causes the bubbles to attain a polyhedral shape. Where three lamellae meet (never > 3, because that would yield an unstable conformation), a prismshaped water volume, bounded by cylindrical surfaces, is formed. This structural element is called a Plateau border. Residual small bubbles usually disappear by Ostwald ripening. In this way, a fairly regular polyhedral foam is formed, not unlike a honeycomb structure. In the lower part of a foam layer, bubbles remain more or less spherical. FIGURE 31 Subsequent stag es (a, b, and c) in the formation of a foam, once bubbles have been created. Thickness of the lamellae between bubbles is too small to be shown on this scale (bubble diameter <1 mm). Pag e 146 As the foam keeps draining (see also the following section), the volume fraction of air increases. The Laplace pressure in the Plateau borders is lower that in the lamellae, and this causes liquid to flow to the Plateau borders. Because the latter are interconnected, they provide pathways through which the liquid can drain. As drainage continues, a j value of 0.95 can readily be reached, which corresponds to an “overrun” of 2000%. Such a foam is not very substantial as a food. To avoid excessive drainage, small filler particles may be incorporated, but they should be hydrophilic; otherwise, considerable coalescence of bubbles can occur (Sec. 3.7.2). Small protein-coated emulsion droplets will function well, and they are incorporated in several whipped toppings. Another approach is gelation of the aqueous phase. This is employed in many aerated food products, such as meringues, foam omelets, bavarois, bread, and cakes. By letting the system gel in an early stage, it is also possible to make a foam with spherical rather than polyhedral bubbles (“Kugelschaum”). Otherwise, only small bubbles in a fairly thin layer of foam can retain a spherical shape. A polyhedral foam itself may be considered a gel. Deformation of the foam causes an increase in curvature of bubbles, a corresponding increase in Laplace pressure, and elastic behavior at small deformation. Then, at greater stress, bubbles slip past each other and viscoelastic deformation occurs. There is thus a yield stress, which is readily observed, since even fairly large portions of foam retain shape under their own weight. The yield stress usually exceeds 100 Pa. 3.7.2 Stability Foams are subject to three main types of instability: 1. Ostwald ripening (disproportionation), which is the diffusion of gas from small to larger bubbles (or to the atmosphere). This occurs because the pressure in a small bubble is greater than in larger ones. 2. Drainage of liquid from and through the foam layer, due to gravity. 3. Coalescence of bubbles due to instability of the film between them. These changes are to some extent interdependent: Drainage may promote coalescence, and Ostwald ripening or coalescence may affect drainage rate. These instabilities are governed by fundamentally different factors, as will be made clear next. Unfortunately, most studies have failed to distinguish between the three types of instability. One reason for this may be the lack of suitable methods to follow bubble size distribution. 3.7.2.1 Ostwald Ripening* This is often by far the most important type of instability. Within minutes after formation, noticeable coarsening of the bubble size distribution often occurs. It happens most rapidly at the top of a foam layer, because the air can diffuse directly to the atmosphere and the layer of water between bubbles and atmosphere is very thin. Ostwald ripening also occurs inside a foam at a significant rate. The classical treatment of Ostwald ripening rate, based on Equation 10 and the diffusion laws, is by de Vries [14]. He considered a small bubble of radius r0 surrounded by very much larger bubbles at an average distance d. The change in radius with time t then would be given by (24) where D is the diffusion coefficient of the gas in water, is solubility for (mol · m-3 · P-1), g is interfacial tension (mostly about 0.05 N.m-1) and p is ambient pressure (Pa). It follows *See Sec. 3.2 for fundamental aspects. Pag e 147 from Equation 24 that a bubble will shrink ever faster as it becomes smaller. Furthermore, since g and solubility for most gases in water are high, shrinkage is fast as illustrated by the following examples. A nitrogen bubble of radius of 0.1 mm at d=1 mm in water will disappear in about 3 min, and a CO2 bubble in about 4 sec. This is not quite realistic, since the geometric assumptions underlying Equation 24 are not fully met in practice, and also because the process is somewhat slower if a mixture of gases, like air, is present. Moreover, as the remaining bubbles become larger, the rate of change decreases with time. Nevertheless, Ostwald ripening can occur very fast. Can Ostwald ripening be stopped or retarded? If a bubble shrinks, its area decreases and its surface load (G) increases, provided the surfactant does not desorb. If no desorption occurs, g is lowered, the Laplace pressure (Eq. 9) is lowered, and the driving force for Ostwald ripening decreases. It will even stop as soon as the surface dilational modulus, Esd, which is a measure of the change in g with change in area (see Eq. 11), becomes equal to g/2. However, surfactant normally desorbs and Esd therefore decreases, at a rate that depends on several factors, especially surfactant type. For a foam made with small-molecule surfactants, desorption occurs readily and retardation of Ostwald ripening is never great. Proteins, however, desorb very sluggishly (see Sec. 3.2.2), Esd remains high (Section 3.2.6), and Ostwald ripening will be greatly retarded [50]. Some proteins produce tenacious layers at the a/w interface because of cross-linking reactions between adsorbed molecules (Section 3.2.2). Egg white especially is a good foam stabilizer. During foaming, strong surface denaturation occurs, leading to fairly large protein aggregates. These remain irreversibly adsorbed, resulting in a very strong resistance to Ostwald ripening. Something similar can be achieved with solid particles, if they have a suitable contact angle (Fig. 7). An example is provided by the partially solid fat globules in whipped cream, which completely coat the air bubbles, and also form a network throughout the system (see Sec. 3.6.5). Many complex systems contain at least some solid particles that act in this manner (being small and fairly hydrophilic). Bubble shrinkage occurs until the adsorbed solid particles touch each other. Then a small but stable bubble remains. This is presumably the cause of many undesirable persistent foams. Another example is the gas cells in bread dough [69]. They show extensive Ostwald ripening, and the number of visible cells in the final product is less than 1% of those originally present. This does not mean that all the others have disappeared. In fact, many tiny cells remain, presumably stabilized by solid particles. They are not visible but scatter light sufficiently to give the bread its white appearance. It should be mentioned that Ostwald ripening can be prevented by a yield stress in the aqueous phase, but it would need to be high, at least about 104 Pa. 3.7.2.2 Drainage As mentioned in Sec. 3.2.6, immobilization of the a/w interface by means of a g gradient is essential to prevent almost instantaneous drainage (Figure 9C and D). The maximum height that a vertical film (lamella) between two bubbles can have while preventing motion of the film surfaces is given by Hmax = 2 Dg/rgd (25) The maximum value that Dg (between top and bottom of a vertical film) can assume equals the surface pressure P, which would be about 0.03 N.m-1. For an aqueous film of thickness d=0.1 mm, Hmax would be 6 cm, far more than needed in food foams (6 cm is, indeed, about the height of the largest foam bubbles floating on a detergent solution). he drainage time of a single vertical film with immobilized surfaces is given by (26) where t(d) is the time needed for the film to drain to a given thickness d. For a water film of a 1 mm height, only 6 sec of drainage is required to achieve a thickness of 10 mm. However, the drainage rate diminishes with decreasing thickness, and it would take 17 days of drainage to achieve d=20 nm. The latter is the approximate thickness at which colloidal interaction forces between the two film surfaces come into play, and it appears that this point is usually not reached within the lifetime of a foam. Predicting the drainage rate in a real foam is far more difficult, and accurate calculations cannot be made. Equation 26 will serve to provide approximate (order of magnitude) values. Drainage can, of course, be slowed down considerably by increasing the viscosity. For this purpose, viscosity should be measured at fairly low shear stress. A yield stress of about gHrwater (where H is the height of the foam layer) will also arrest drainage. 3.7.2.3 Coalescence The mechanism differs with circumstances. Three cases are important. 1. Thick films. This refers to films thick enough so that colloidal interaction between the two surfaces is negligible. In this section the Gibbs stabilizing mechanism is essential (Sec. 3.2.6, especially Fig. 9E). Film rupture, and thereby bubble coalescence, will occur only when surfactant concentration is very low, and only during foam formation. If a film is rapidly stretched, as will always occur during whipping, rupture occurs more readily (it causes Esd to decrease). Thus, it is observed that there is an optimum whipping speed for foam formation, that is, one that achieves greatest air incorporation. 2. Thin films. As mentioned earlier, drainage will almost never lead to films thin enough for colloidal interactions to become important. However, water may evaporate from the film, especially at the top of a foam. If so, film rupture may, and often will, occur. The considerations given in Section 3.6.4 roughly apply. Compared to emulsions, g is large (more stable), but film area is very large (less stable). Again, proteins may yield the most stable films, especially if they form thick adsorbed layers. 3. Films containing extraneous particles. It is often observed that the presence of extraneous particles, especially lipids, is very detrimental to foam stability. Such particles can cause rupture of relatively thick films, and possible mechanisms are indicated in Figure 32. In cases a and b, a hydrophobic particle becomes trapped in a thinning film, coming into contact with both air bubbles. Because of the obtuse contact angle (Fig. 7), the curvature of the film surface where it reaches the particle results in a high Laplace pressure. Hence, the water flows away from this high pressure region and the film breaks. Hydrophilic particles, on the other hand, may adsorb and not induce film rupture. In cases c and d, contact with one air bubble would suffice. An oil droplet reaching the a/w interface (case c) will suddenly alter its shape. It may greatly flatten, depending on the three interfacial tensions (Fig. 7). This induces water to flow away from the flattening droplet, and if the film is fairly thin, it can lead to film rupture. In case d the contact angle as measured in the oil is zero. This is equivalent to the existence of a positive spreading pressure Ps. Oil then spreads over the a/w interface, dragging water along with it, which may, again, lead to film rupture. Pag e 149 FIGURE 32 Mechanisms of rupture of an aqueous film (lamella) between air bubbles, as induced by extraneous particles; 1, 2, and 3 denote subsequent stag es. (a) Solid particle; (b and c) oil droplets; (d) oil droplet or composite particle. A is air; W is water; q is contact ang le (measured in the water phase); h is viscosity; P s is spreading pressure. See text for further discussion. The surface pressure usually is negative if the foaming agent is a small-molecule surfactant, but for many proteins it can be positive—hence the detrimental effect of fat on the stability of beer foam, which is primarily stabilized by proteins and other polymers. The situation can even be more complicated. For a droplet to come into contact with air, the thin aqueous film between droplet and air bubble has to break. Since both bubble and droplet are covered by surfactants that may provide colloidal repulsion, rupture of this film may not readily occur. If the droplet contains fat crystals, something similar to the induction of partial coalescence may happen (Sec. 3.6.5)—that is, a protruding crystal can pierce the film. It is indeed observed, that partial crystallization of the oil in the droplets may greatly enhance their destabilizing activity. Finally, the number concentration of particles should be considered. A typical foam (Table 8) contains about 1011 lamellae per cubic meter of liquid phase. Presumably, 1010 particles per cubic meter would suffice to cause overwhelming coalescence, provided the particles can induce film rupture. If the concentration of extraneous particles is much smaller, the foam may remain fairly stable. In a typical whipping cream, the number of particles, that is, partially solid fat globules, is very large, about 1016/m3 . These globules can induce film rupture according to mechanism d (Fig. 32). However, their large concentration causes almost simultaneous adsorption of many globules very close to each other. Spreading of liquid oil over any distance then is not possible, and film rupture will rarely occur. However, if whipping goes on, the fat globules undergo extensive partial coalescence, large clumps are formed, and eventually their number becomes so small that film rupture can occur. In other words, overwhipping destroys the foam made at an earlier stage. Pag e 150 Acknowledgments The author gratefully acknowledges the cooperation on subjects as discussed in this chapter with his Wageningen colleagues Drs. A. Prins and T. van Vliet. Frequently Used Symbols A (Specific) surface area (m-1, m2 ) Hamaker constant (J) a Thermodynamic activity (mole fraction) Acceleration (m. sec-2) B Permeability (m2 ) c Concentration (kg. m-3; mol. m-3; moles per liter) D Diffusion coefficient (m-1.sec-1) Fractal dimensionality (—) d Particle diameter (m) Esd Surface dilational modulus (N.m-1) F Force (N) G Velocity gradient, shear rate (sec-1) Elastic shear modulus (Pa) g Acceleration due to gravity (9.81 m.sec-2) H Height (m) h Interparticle distance (m) I Ionic strength (moles per liter) k Boltzmann constant (1.38×10-23 J. K-1) l Distance, length (m) m Concentration (moles per liter) N Total number concentration (m-3) ni Number of particles in class i (m-3) p Pressure (Pa) pL Laplace pressure (Pa) Q Volume flow rate (m3 .sec-1) R Universal gas constant (8.314J. mol-1 . K-1) Radius of aggregate (floc) (m) Rcr Critical radius (m) Rg Radius of gyration (m) r Particle radius (m) s Solubility of gas (mol.m-3.Pa-1) T Temperature (absolute) (K) t Time (sec) t0.5 Halving time (sec) V Interaction free energy (J) VA Part of V due to van der Waals attraction VE Part of V due to electrostatic repulsion VR Part of V due to steric repulsion v Velocity (m.sec-1) vs Stokes velocity of particle (m.sec-1) W Stability ratio (—) x Distance (m) z Valence (—) Pag e 151 Greek Symbols G Surface excess (load) (mol.m-2, kg.m-2) g Surface (interfacial) tension (N . m-1) d Layer (film) thickness (m) e Strain (relative deformation) (—) efr Strain at fracture (—) q Contact angle (rad) k Reciprocal Debye length (m-1) h Viscosity (Pa . sec) hss Surface shear viscosity (N.sec.m-1) P Surface pressure (N.m-1) Ps Spreading pressure (N .m-1) Posm Osmotic pressure (Pa) r Mass density (kg.m-3) s Stress (Pa) sfr Fracture stress (Pa) sy Yield stress (Pa) j Volume fraction (—) y Fraction solid (—) Subscripts A Air C Continuous phase D Dispersed phase O Oil S Solid W Water (aqueous phase) Bibliography Blanshard, J. M. V. and P. Lillford, eds. (1987). Food Structure and Behaviour, Academic Press, London. Blanshard, J. M. V., and J. R. Mitchell, eds. (1988). Food Structure—Its Creation and Evaluation, Butterworth, London. Dickinson, E. (1992). An Introduction into Food Colloids, Oxford Science, Oxford. Dickinson, E., and G. Stainsby, (1982). Colloids in Foods, Applied Science Publishers, London. Larsson, K., and S. E. Friberg, eds. (1990). Food Emulsions, 2nd. ed., Marcel Dekker, New York. Mitchell, J. R., and D. A. Ledward, eds. (1986). Functional Properties of Food Macromolecules, Elsevier, London. P. Walstra (In preparation). Physical Chemistry of Foods, Marcel Dekker, New York. References 1. Adamson, A. W. (1976). Physical Chemistry of Surfaces, 3rd ed., Interscience, New York. 2. Allen, T. (1981). Particle Size Measurement, 3rd ed., Chapman & Hall, London. 3. Bergenståhl, B. (1988). Gums as stabilizers of emulsifier covered emulsion droplets, in Gums and Stabilizers for the Food Industry, Volf. 4 (G. O. Phillips, P. A. Williams, and D. J. Wedlock, eds.), IRL Press, Oxford, pp. 363–372. 4. Blanshard, J. M. V. (1982). Hydrocolloid water interactions. Progr. Food Nutr. Sci. 6:3–20. 5. Boode, K., and P. Walstra (1993). Kinetics of partial coalescence in oil-in-water emulsions, in Food Pag e 152 Colloids and Polymers: Stability and Mechanical Properties (eds. E. Dickinson and P. Walstra), Royal Society Chemistry, Cambridge, pp. 23–30. 6. Boode, K., and P. Walstra (1993). Partial coalescence in oil-in-water emulsions 1. Nature of the aggregation. Colloids Surf. 81:121–137. 7. Boode, K., P. Walstra, and A. E. A. de Groot-Mostert (1993). Partial coalescence in oil-in-water emulsions 2. Influence of the properties of the fat. Colloids Surf 81:139–151. 8. Bremer, L. G. B., P. Walstra, and T. van Vliet (1995). Aggregation kinetics related to the observation of instability of colloidal systems. Colloids Surf. A99:121–127. 9. Brownsey, G. J., and V. J. Morris (1988). Mixed and filled gels—Models for foods, in Functional Properties of Food Macromolecules (J. R. Mitchell and D. A. Ledward, eds.), Elsevier, London, pp. 7–23. 10. Busnell, J. P., S. M. Clegg, and E. R. Morris (1988). Melting behaviour of gelatin: Origin and control, in Gums and Stabilizers for the Food Industry, Vol. 4 (G. O. Phillips, D. J. Wedlock, and P. A. Williams, eds.), IRL Press, Oxford, pp. 105–115. 11. Clark, A. H., and C. D. Lee-Tufnell (1986). Gelation of globular proteins, in Functional Properties of Food Macromolecules (J. R. Mitchell and D. A. Ledward, eds.), Elsevier, London, pp. 203–272. 12. Darling, D. F. (1982). Recent advances in the destabilization of dairy emulsions. J. Dairy Res. 49:695–712. 13. De Feijter, J. A., J. Benjamins, and M. Tamboer (1987). Adsorption displacement of proteins by surfactants in oil-in-water emulsions. Colloids Surf. 27:243–266. 14. De Vries, A. J. (1958). Foam stability. II. Gas diffusion in foams. Rec. Trav. Chim. 77:209–225. 15. Dickinson, E. (1988). The role of hydrocolloids in stabilising particulate dispersions and emulsions, in Gums and Stabilizers for the Food Industry, Vol. 4 (G. O. Phillips, D. J. Wedlock, and P. A. Williams, eds.), IRL Press, Oxford, pp. 249–263. 16. Dickinson, E. (1992). Structure and composition of adsorbed protein layers and the relation to emulsion stability. J. Chem. Soc. Faraday Trans. 88: 2973–2983. 17. Dickinson, E. (1994). Protein-stabilized emulsions. J. Food Eng. 22:59–74. 18. Dickinson, E., and L. Eriksson (1991). Particle flocculation by adsorbing polymers. Adv. Colloid Interf. Sci. 34: 1–29. 19. Dickinson, E., V. B. Galazka, and D. M. W. Anderson (1991). Emulsifying behaviour of gum arabic. Part 1:Effect of the nature of the oil phase on the emulsion droplet-size distribution. Carbohydrate Polym. 14:373–383. 20. Dickinson, E., and Y. Matsumura (1991). Time-dependent polymerization of b-lactoglobulin through disulphide bonds at the oil-water interface in emulsions. Int. J. Biol. Macromol. 13:26–30. 21. Dickinson, E., B. S. Murray, and G. Stainsby (1988). Coalescence stability of emulsion-sized droplets at a planar oil-water interface and the relation to protein film surface rheology. J. Chem. Soc., Faraday Trans. 1 84:871–883. 22. Fleer, G. J., M. A. Cohen Stuart, J. M. H. M. Scheutjens, T. Cosgrove, and B. Vincent (1993). Polymers at Interfaces, Chapman & Hall, London. 23. Garrett, P. R. (1993). Recent developments in the understanding of foam generation and stability. Chem. Eng. Sci. 48:367–392. 24. Garti, N., and D. Reichman (1994). Surface properties and emulsification activity of galactomannans. Food Hydrocolloids 8:155–173. 25. Graham, D. E., and M. C. Phillips (1980). Proteins at liquid interfaces IV. Dilational properties. J. Colloid Interface Sci. 76:227–239. 26. Hermansson, A.-M. (1988). Gel structure of food biopolymers, in Functional Properties of Food Macromolecules (J. R. Mitchell and D. A. Ledward, eds.), Elsevier, London, pp. 25–40. 27. Hough, D. B., and L. R. White (1980). The calculation of Hamaker constants from Lifshitz theory with application to wetting phenomena. Adv. Colloid Interface Sci. 14:3–41. 28. Israelachvilli, J. N. (1992). Intermolecular and Surface Forces, 2nd ed., Academic Press, London. 29. Kato, A., and S. Nakai (1980). Hydrophobicity determined by a fluorescent probe method and its correlation with surface properties of proteins. Biochim. Biophys. Acta 624:13–20. 30. Keetels, C., T. van Vliet, and H. Luyten (1995). The effect of retrogradation on the structure and Pag e 153 mechanics of concentrated starch gels, in Food Macromolecules and Colloids (E. Dickinson and D. Lorient, eds.), Roy. Soc. Chem., Cambridge, pp. 472–479. 31. Kimura, H., S. Morikata, and M. Misaki (1973). Polysaccharide 13140: A new thermo-gelable polysaccharide. J. Food Sci. 38:668–670. 32. Krog, N. J. (1990). Food emulsifiers and their chemical and physical properties, in Food Emulsions (K. Larsson and S. E. Friberg, eds.), 2nd. ed., Marcel Dekker, New York, pp. 127–180. 33. Larsson, K., and P. Dejmek (1990). Crystal and liquid crystal structure of lipids, in Food Emulsions (K. Larsson and S. E. Friberg, eds.), 2nd ed., Marcel Dekker, New York, pp. 97–125. 34. Ledward, D. A. (1986). Gelation of Gelatin, in Functional Properties of Food Macromolecules (J. R. Mitchell and D. A. Ledward, eds.), Elsevier, London, pp. 171–201. 35. Ledward, D. A. and J. R. Mitchell (1988). Protein extrusion—More questions than answers?, in Food Structure—Its Creation and Evaluation (J. M. V. Blanshard and J. R. Mitchell, eds.), Butterworths, London, pp. 219–229. 36. Lucassen-Reynders, E. H. (1981). Adsorption at fluid interfaces, in Anionic Surfactants: Physical Chemistry of Surfactant Action (E. H. Lucassen-Reynders, ed.), Marcel Dekker, New York, pp.1–54. 37. Lucassen-Reynders, E. H. (1981). Surface elasticity and viscosity in compression/dilation, in Anionic Surfactants: Physical Chemistry of Surfactant Action (E. H. Lucassen-Reynders, ed.), Marcel Dekker, New York, pp. 173–216. 38. Luyten, H., and T. van Vliet (1990). Influence of a filler on the rheological and fracture properties of food materials, in Rheology of Foods, Pharmaceutical and Biological Materials with General Rheology (R. E. Carter, ed.), Elsevier, London, pp. 43–56. 39. Luyten, H., T. van Vliet, and W. Kloek (1991). Sedimentation in aqueous xanthan + galactomannan mixtures, in Food Polymers, Gels and Colloids (E. Dickinson, ed.), Royal Society of Chemistry, Cambridge, pp. 527–530. 40. Morris, V. J. (1986). Gelation of polysaccharides, in Functional Properties of Food Macromolecules (J. R. Mitchell, and D. A. Ledward, eds.), Elsevier, London, pp. 121–170. 41. Morris, V. J. (1992). Designing polysaccharides for synergistic interactions, in Gums and Stabilizers for the Food Industry, Vol. 6 (G. O. Phillips, P. A. Williams and D. J. Ledward, eds.) IRL Press, Oxford, pp. 161–172. 42. Muhr, A. H., and J. M. V. Blanshard (1982). Diffusion in gels. Polymer 23 (suppl.):1012–1026. 43. Mulder, H., and P. Walstra (1974). The Milk Fat Globule: Emulsion Science as Applied to Milk Products and Comparable Foods, PUDOC, Wageningen, The Netherlands, and CAB, Farnham Royal. 44. Norde, W., and J. Lyklema (1991). Why proteins prefer interfaces. J. Biomater. Sci. Polymer Ed. 2: 183–202. 45. Oguntunde, A. O., P. Walstra, and T. van Vliet (1989). Physical characterization of soymilk, in Trends in Food Biotechnology (Ang How Ghee, ed.), Proc. 7th World Congr. Food Sci. Technol., 1987, pp. 307–308. 46. Pearce, K. N., and J. E. Kinsella (1978). Emulsifying properties of proteins: Evaluation of a turbidimetric technique. J. Agric. Food Chem. 26: 716–723. 47. Peleg, M. (1987). The basics of solid food rheology, in Food Texture (H. R. Moskowitz, ed.), Marcel Dekker, New York, pp.3–33. 48. Prins, A. (1988). Principles of foam stability, in Advances in Food Emulsions and Foams (E. Dickinson and G. Stainsby, eds.), Elsevier, London, pp. 91–122. 49. Rinaudo, M. (1992). The relation between the chemical structure of polysaccharides and their physical properties, in Gums and Stabilizers for the Food Industry, vol. 6 (G. O. Phillips, P. A. Williams, and D. J. Wedlock, eds.), IRL Press, Oxford, pp. 51–61. 50. Ronteltap, A. D., and A. Prins (1990). The role of surface viscosity in gas diffusion in aqueous foams. II. Experimental. Colloids Surf. 47:285–298. 51. Shaw, D. J. (1970). Introduction to Colloid and Surface Chemistry, 2nd. ed., Butterworth, London. 52. Shimizu, M., M. Saito, and K. Yamauchi (1986). Hydrophobicity and emulsifying activity of milk proteins. Agric. Biol. Chem. 50:791–792. 53. Shinoda, K., and H. Kunieda (1983). Phase properties of emulsions: PIT and HLB, in Encyclopedia of Emulsion Technology, Vol. 1, Basic Theory (P. Becher, ed.), Marcel Dekker, New York, pp. 337–367. Carbohydrates comprise more than 90% of the dry matter of plants. They are abundant, widely available, and inexpensive. They are common components of foods, both as natural components and as added ingredients. Their use is large in terms of both the quantities consumed and the variety of products in which they are found. They have many different molecular structures, sizes, and shapes and exhibit a variety of chemical and physical properties. They are amenable to both chemical and biochemical modification, and both modifications are employed commercially in improving their properties and extending their use. They are also safe (nontoxic). Starch, lactose, and sucrose are digestible by humans, and they, along with D-glucose and D-fructose, and human energy sources, providing 70–80% of the calories in the human diet worldwide. In the United States, they supply less than that percentage, in fact only about 50%, of the human caloric intake. Health organizations recommend that the percentage of the total calories the average American consumes in the form of fat (about 37%) be reduced to no more than 30% and that the difference be made up with carbohydrate, especially starch. The term carbohydrate suggests a general elemental composition, namely Cx(H2O)y,) which signifies molecules containing carbon atoms along with hydrogen and oxygen atoms in the same ratio as they occur in water. However, the great majority of natural carbohydrate compounds produced by living organisms do not have this simple empirical formula. Rather, most of the natural carbohydrate is in the form of oligomers (oligosaccharides) or polymers (polysaccharides) of simple and modified sugars.* Lower molecular weight carbohydrates are often obtained by depolymerization of the natural polymers. This chapter begins with a presentation of the simple sugars and builds from there to larger and more complex structures. 4.1 Monosaccharides [8,27] Carbohydrates contain chiral carbon atoms. A chiral carbon atom is one that can exist in two different spatial arrangements (configurations). Chiral carbon atoms are those that have four different groups attached to them. The two different arrangements of the four groups in space (configurations) are nonsuperimposable mirror images of each other. In other words, one is the *Carbohydrates that cannot be broken down to lower molecular weig ht carbohydrates by hydrolysis are monosaccharides, a term that indicates that they are the monomeric building units of the olig o- and polysaccharides. Monosaccharides are commonly referred to simply as sug ars. However, as will be explained later, table sug ar (sucrose) is not a monosaccharide reflection of the other that we would see in a mirror, with everything that is on the right in one configuration on the left in the other and vice versa (Fig. 1). D-Glucose, the most abundant carbohydrate and the most abundant organic compound (if all its combined forms are considered), belongs to the class of carbohydrates called monosaccharides. Monosaccharides are carbohydrate molecules that cannot be broken down to simpler carbohydrate molecules by hydrolysis, so they are sometimes referred to as simple sugars. They can be joined together to form larger structures, namely, oligosaccharides and polysaccharides (see Secs. 4.2 and 4.3), that can be converted into monosaccharides by hydrolysis. D-Glucose is both a polyalcohol and an aldehyde. It is classified as an aldose, a designation for sugars containing an aldehyde group (Table 1). The ending -ose signifies a sugar; and-signifies an aldehyde group. When D-glucose is written in an open or vertical, straight-chain fashion, known as an acyclic structure to organic chemists, with the aldehyde group (position 1) at the top and the primary hydroxyl group (position 6) at the bottom, it is seen that all secondary hydroxyl groups are on carbon atoms having four different substituents attached to them. TABLE 1 Classification of Monosaccharides Kind of carbonyl g roup Number of carbon atoms Aldehyde Ketone 3 Triose Triulose 4 Tetrose Tetrulose 5 Pentose Pentulose 6 Hexose Hexulose 7 Heptose Heptulose 8 Octose Octulose 9 Nonose Nonulose Pag e 160 These carbon atoms are therefore chiral. Glucose has four chiral carbon atoms: C-2, C-3, C-4, and C-5. Naturally occurring glucose is designated as the D form, specifically D-glucose. It has a molecular mirror image, termed the L form, specifically Lglucose. Since each chiral carbon atom has a mirror image, there are 2n arrangements for these atoms. Therefore, in a sixcarbon aldose, there are 24 or 16 different arrangements of the carbon atoms containing secondary hydroxyl groups, allowing formation of 16 different six-carbon sugars with an aldehyde end. Eight of these belong to the D-series (see Fig. 3); eight are their mirror images and belong to the L-series. All sugars that have the hydroxyl group on the highest numbered chiral carbon atom (C-5 in this case) positioned on the right-hand side are arbitrarily called D-sugars, and all with a left-hand positioned hydroxyl group on the highest numbered chiral carbon atom are designed L-sugars. Two structures of D-glucose in its openchain, acyclic form (called the Fischer projection) with the carbon atoms numbered in the conventional manner are given in Figure 2. In this convention, each horizontal bond projects outward from the plane of the page and each vertical bond projects into or below the plane of the page. It is customary to omit the horizontal lines for covalent chemical bonds to the hydrogen atoms and hydroxyl groups as in the structure on the right. Because the lowermost carbon atom is nonchiral, it is meaningless to designate the relative positions of the atoms and groups attached to it. Thus, it is usually written as -CH2OH. D-Glucose and all other sugars containing six carbon atoms are called hexoses, the most common group of aldoses. The categorical names are often combined, with a six-carbon-atom aldehyde sugar being termed an aldohexose. There are two aldoses containing three carbon atoms. They are D-glycerose (D-glyceraldehyde) and L-glycerose (Lglyceraldehyde), each possessing only one chiral carbon atom. Aldoses with four carbon atoms, the tetroses, have two chiral carbon atoms; aldoses with five carbon atoms, the pentoses, have three chiral carbon atoms and comprise the second most common group of aldoses. Extending the series above six carbon atoms gives heptoses, octoses, and nonosese, which is the practical limit of naturally occurring sugars. Development of the eight D-hexoses from D-glycerose is shown in Figure 3. In this figure, the circle represents the aldehyde group; the horizontal lines designate the location of each hydroxyl group on its chiral carbon atom, and at the bottoms of the vertical lines are the terminal, nonchiral primary hydroxyl groups. This FIGURE 2 D-Glucose (open-chain or acyclic structure). Pag e 161 FIGURE 3 Rosanoff structure of the D-aldoses containing from three to six carbon atoms. shorthand way of indicating monosaccharide structures is called the Rosanoff method. Sugars whose names are in italics in Figure 3 are commonly found in plants, almost exclusively in combined forms. They therefore are present in our diets in combined froms. D-Glucose is the only free aldose usually present in natural foods, and then only in small amounts. L-Sugars are less numerous and less abundant in nature than are the D forms but nevertheless have important biochemical roles. Two L-sugars found in foods are L-arabinose and L-galactose, both of which occur as units in carbohydrate polymers (polysaccharides). In the other type of monosaccharide, the carbonyl function is a ketone group. These sugars are termed ketoses. (Ket- signifies the ketone group.) The suffix designating a ketose in systematic carbohydrate nomenclature is -ulose (Table 1). D-Fructose (Fig. 4) is the prime example of this sugar group. It is one of the two monosaccharide units of the disaccharide sucrose (see Sec. 4.2.3) and makes up about 55% of high-fructose corn syrup and about 40% of honey. D-Fructose has only three chiral carbon atoms, C-3, C-4, and C-5. Thus, there are only 23 or 8 D-ketohexoses. D-Fructose is the principal commercial ketose and the only one found free in natural foods, but, like D-glucose, only in small amounts. 4.1.1 Monosaccharide Isomerization Simple aldoses and ketoses containing the same number of carbon atoms are isomers of each other; that is, a hexose and a hexulose both have the empirical formula C6H12O6 and can Pag e 162 FIGURE 4 D-Fructose (open-chain or acyclic structure). be interconverted by isomerization. Isomerization of monosaccharides involves both the carbonyl group and the adjacent hydroxyl group. By this reaction, an aldose is converted into another aldose (with opposite configuration of C-2) and the corresponding ketose, and a ketose is converted into the corresponding two aldoses. Therefore, by isomerization, D-glucose, Dmannose, and D-fructose can be interconverted (Fig. 5). Isomerization can be catalyzed by either a base or an enzyme. 4.1.2 Monosaccharide Ring Forms Carbonyl groups of aldehydes are reactive and easily undergo nucleophilic attach by the oxygen atom of a hydroxyl group to produce a hemiacetal. The hydroxyl group of a hemiacetal can react further (by condensation) with a hydroxyl group of an alcohol to produce an acetal (Fig. 6). The carbonyl group of a ketone reacts similarly. Hemiacetal formation can occur within the same aldose or ketose sugar molecule wherein the carbonyl function reacts with one of its own properly positioned hydroxyl groups, as illustrated in Figure 7 with D-glucose laid coiled on its side. The resulting sixmembered sugar ring is called a pyranose ring. Notice that for the oxygen atom of the hydroxyl group at C-5 to react to form the ring, C-5 must rotate to bring its oxygen atom upward. This rotation brings the hydroxymethyl group (C-6) to a position above the ring. The representation of the D-glucopyranose ring used in Figure 7 is termed a Haworth projection. Sugars occur less frequently in five-membered (furanose) rings (Fig. 8). FIGURE 5 Interrelationship of D-g lucose, D-mannose, and D-fructose via isomerization. Pag e 163 FIGURE 6 Formation of an acetal by reaction of an aldehyde with methanol. To avoid clutter in writing the ring structures, common conventions are adopted wherein ring carbon atoms are indicated by angles in the ring and hydrogen atoms attached to carbon atoms are eliminated altogether. A mixture of chiral forms is indicated by a wavy line (Fig. 9). When the carbon atom of the carbonyl group is involved in ring formation, leading to hemiacetal (pyranose or furanose ring) development, it becomes chiral. With D-sugars, the configuration that has the hydroxyl group located below the ring is the alpha form. For example, therefore, a-D-glucopyranose is D-glucose in the pyranose (six-membered) ring from with the configuration of the new chiral carbon atom, C-1, termed the anomeric carbon atom, alpha. When the newly formed hydroxyl group at C-1 is above the ring, it is in the beta position, and the structure is named b-D-glucopyranose. This designation holds for all D-sugars. For sugars in the L-series, the opposite is true—that is, the anomeric hydroxyl group is up in the alpha anomer and down in the beta anomer. * (See, for example, Fig. 8.) This is so because, for example, a-D-glucopyranose and a-L-glucopyranose are mirror images of one another. However, pyranose rings are not flat with the attached groups sticking straight up and straight down as the Haworth structure suggests. Rather, they occur in a variety of shapes (conformations), most commonly in one of two chair conformations, so called because they are shaped somewhat like a chair. In this shape, one bond on each carbon atom does project either up or down from the ring; these are called axial bonds or axial positions. The other bond not involved in ring formation is either up or down with respect to the axial bonds but with respect to the ring projects out around the perimeter in what is called an equatorial position (Fig. 10). Using b-D-glucopyranose as an example, C-2, C-3, C-5, and the ring oxygen atom remain in a plane, but C-4 is raised slightly above the plane and C-1 is positioned slightly below the plane as in Figures 10 and 11. This conformation is designated 4C1. The notation C indicates that the ring is chair-shaped; the superscript and subscript numbers indicate that C-4 is above the plane of the ring and C-1 below the plane. There are two chair forms. The second, 1C4, has FIGURE 7 Formation of a pyranose hemiacetal ring from D-g lucose. *The a and b ring forms of a sug ar are known as anomers. The two anomers comprise an anomeric pair. Pag e 164 FIGURE 8 L-Arabinose in the furanose ring from and a-L-config uration. all the axial and equatorial groups reversed. The six-membered ring distorts the normal carbon and oxygen atom bond angles less than do rings of other sizes. The strain is further lessened when the bulky hydroxyl groups are separated maximally from each other by the ring conformation that arranges the greatest number of them in equatorial rather than axial positions. The equatorial position is energetically favored, and rotation of carbon atoms takes place on their connecting bonds to swivel the bulky groups to equatorial positions in so far as possible. As noted, b-D-glucopytanose has all its hydroxyl groups in the equatorial arrangement, but each is bent either slightly above or slightly below the true equatorial position. In b-D-glucopyranose, the hydroxyl grouse, all of which are in an equatorial position, alternate in an up-and-down arrangement, with that at C-1 positioned slightly up, that on C-2 slightly down, and continuing with an up-and-down arrangement. The bulky hydroxymethyl group, C-6 in hexoses, is almost always in a sterically free equatorial position. If D-glucopyranose were in a FIGURE 9 D-Glucopyranose as a mixture of two chiral forms. Pag e 165 FIGURE 10 A pyranose ring showing the equatorial (solid line) and axial (dashed line) bond positions. 1C4 conformation, all the bulky groups would be axial; so very little D-glucopyranose exists in the 1C4 conformation, a much higher energy form. Six-membered sugar rings are than quite stable if bulky groups, such as hydroxyl groups and the hydroxymethyl group, are in equatorial positions. Thus, b-D-glucopyranose dissolves in water to give a rapidly equilibrating mixture containing the open chain form and its five, six-, and seven-membered ring forms. At room temperature, the six-membered (pyranose) ring forms predominate, followed by the five-membered (furanose) ring forms and only a trace of the seven-membered ring forms. The anomeric arrangement of each ring may be alpha or beta. The open-chain, aldehydo form constitutes only about 0.003% of the total forms (Fig. 12). 4.1.3 Glycosides The hemiacetal form of sugars can react with an alcohol to produce a full acetal, called a glycoside. The acetal linkage at the anomeric carbon atom is indicated by the -ide ending. In the case of D-glucose reacting with methanol, the product is mainly methyl a-D-glucopyranoside, with less methyl b-D-glucopyranoside (Fig. 13). The two anomeric forms of the five-membered-ring furanosides are also formed, but being higher energy structures, they reorganize into more stable forms and are present at equilibrium in comparatively low quantities. The methyl group in this case, and any other group bonded to a sugar to make a glycoside, is termed an aglycon. Glycosides undergo hydrolysis to yield a reducing sugar (see Sec. 4.1.4.1) and a hydroxylated compound in the presence of warm or hot aqueous acid. 4.1.4 Monosaccharide Reactions All carbohydrate molecules have hydroxyl groups available for reaction. Simple monosaccharide and most other low-molecularweight carbohydrate molecules also have carbonyl groups available for reaction. Formation of pyranose and furanose rings (cyclic hemiacetals) and of glycosides (acetals) of Monosaccharides has already been presented. FIGURE 11 b-D Glucopyranose in the 4C1 conformation. All bulky g roups are in equatorial positions, and all hydrog en atoms are in axial positions. Pag e 166 FIGURE 12 Interconversion of the acyclic and cyclic forms of D-g lucose. 4.1.4.1 Oxidation to Aldonic Acids and Aldonolactones Aldoses are readily oxidized to aldonic acids. The reaction is commonly used for quantitative determination of sugars. One of the earliest methods for quantitative measurement of sugars employed Fehling solution. Fehling solution is an alkaline solution of copper(II) that oxidizes an aldose to an aldonate and in the process is reduced to copper(I), which precipitates as brick-red Cu2O. Variations, the Nelson-Somogyi and Benedict reagents, are still used for determining amounts of reducing sugars in foods and other biological materials. FIGURE 13 Methyl a-D-g lucopyranoside (left) and methyl b-D-g lucopyranoside (rig ht). Pag e 167 (1) Because, in the process of oxidizing the aldehyde group of an aldose to the salt of a carboxylic acid group, the oxidizing agent is reduced, aldoses are called reducing sugars. Ketoses are also termed reducing sugars because, under the alkaline conditions of the Fehling test, ketoses are isomerized to aldoses. Benedict reagent, which is not alkaline, will react with aldoses but not with ketoses. A simple and specific method for quantitative oxidation of D-glucose to D-gluconic acid uses the enzyme glucose oxidase, with the initial product being the 1,5,-lactone (an intramolecular ester) of the acid (Fig. 14). The reaction is commonly employed to measure the amount of D-glucose in foods and other biological materials, including the D-glucose level of blood. D-Gluconic acid is a natural constituent of fruit juices and honey. The reaction given in Figure 14 is also used for the manufacture of commercial D-gluconic acid and its lactone. D-Glucono-deltalactone (GDL), D-glucono-1,5-lactone according to systematic nomenclature, hydrolyzes to completion in water in about 3 hr at room temperature, effecting a decrease in pH. Its slow hydrolysis, slow acidification, and mild taste set GDL apart from other food acidulants. It is used in meats and dairy products, but particularly in baked goods as a component of chemical leavening agents for preleavened products. 4.1.4.2 Reduction of Carbonyl Groups [9] Hydrogenation is the addition of hydrogen to a double bond. When applied to carbohydrates, it most often entails addition of hydrogen to the double bond between the oxygen atom and the carbon atom of the carbonyl group of an aldose or ketose. Hydrogenation of D-glucose is easily accomplished with hydrogen gas under pressure in the presence of Raney nickel. The product is D-glucitol, commonly known as sorbitol, where the -itol suffix denotes a sugar alcohol (an alditol) (Fig. 15). Alditols are also known as polyhydroxy alcohols and as polyols. Because it is derived from a hexose, D-glucitol (sorbitol) is specifically a hexitol. It is found widely distributed throughout the plant world, ranging from algae to higher orders where it is found in fruits and berries, but the amounts present are generally small. It is 50% as sweet as sucrose, is sold both as a syrup and as crystals, and is used as a general humectant. D-Mannitol can be obtained by hydrogenation of D-mannose. Commercially, it is obtained along with sorbitol from hydrogenolysis of sucrose. It develops from hydrogenation of the D-fructose component of sucrose and from isomerization of Dglucose, which can be controlled FIGURE 14 Oxidation of D-g lucose catalyzed by g lucose oxidase. Pag e 168 FIGURE 15 Reduction of D-g lucose. by the alkalinity of the solution undergoing catalytic hydrogenation (Fig. 16). D-Mannitol, unlike sorbitol, is not a humectant. Rather, it crystallizes easily and is only moderately soluble. It has been used as a nonsticky coating on candies. It is 65% as sweet as sucrose and is used in sugar-free chocolates, pressed mints, cough drops, and hard and soft candies. Xylitol (Fig. 17) is produced from hydrogenation of D-xylose obtained from hemicelluloses, especially from birch trees. Its crystals have an endothermic heat of solution and give a cool feel when placed in the mouth. It is used in dry hard candies and in sugarless chewing gum. It is about 70% as sweet as sucrose. When xylitol is used in place of sucrose, there is a reduction in dental caries because xylitol is not metabolized by the microflora of the mouth to produce plaques. 4.1.4.3 Uronic Acids The terminal carbon atom (at the other end of the carbon chain from the aldehyde group) of a monosaccharide unit of an oligoor polysaccharide may occur in an oxidized (carboxylic acid) form. Such an aldohexose with C-6 in the form of a carboxylic acid group is called a uronic acid. When the chiral carbon atoms of a uronic acid are in the same configuration as they are in Dgalactose, for example, the compound is D-galacturonic acid (Fig. 18), the principal component of pectin (see Sec. 4.10). FIGURE 16 Reduction of D-fructose. Pag e 169 FIGURE 17 Xylitol. 4.1.4.4 Hydroxyl Group Esters The hydroxyl groups of carbohydrates, like the hydroxyl groups of simple alcohols, form esters with organic and some inorganic acids. Reaction of hydroxyl groups with a carboxylic acid anhydride or chloride (an acyl chloride) in the presence of a suitable base produces an ester. (2) Acetates, succinate half-esters, and other carboxylic acid esters of carbohydrates occur in nature. They are found especially as components of polysaccharides. Sugar phosphates are common metabolic intermediates (Fig. 19). Monoesters of phosphoric acid are also found as constituents of polysaccharides. For example, potato starch contains a small percentage of phosphate ester groups. Corn starch contains even less. In producing modified food starch, corn starch is often derivatized with one or the other or both of mono- and distarch ester groups (see Sec. 4.4.9). Other esters of starch, most notably the acetate, succinate and substituted succinate half-esters, and distarch adipates, are in the class of modified food starches (see Sec. 4.4.9). Sucrose (see Sec. 4.2.3) fatty acid esters are produced commercially as emulsifiers. Members of the family of red seaweed polysaccharides, which includes the carrageenans (see Sec. 4.8), contain sulfate groups (half-esters of sulfuric acid, R-OSO3¯). 4.1.4.5 Hydroxyl Group Ethers The hydroxyl groups of carbohydrates, like the hydroxyl groups of simple alcohols, can from ethers as well as esters. Ethers of carbohydrates are not as common in nature as are esters. However, polysaccharides are etherified commercially to modify their properties and make them more useful. Examples are the production of methyl (-O-CH3), sodium carboxyFIGURE 18 D-Galacturonic acid. Pag e 170 FIGURE 19 Examples of sug ar phosphate metabolic intermediates. methyl (-O-CH2-CO2¯Na+), and hydroxypropyl (-O-CH2-CHOH-CH3) ethers of cellulose and hydroxypropyl ethers of starch, all of which are approved for food use. A special type of ether, an internal ether formed between carbon atoms 3 and 6 of a D-galactosyl unit, is found in the red seaweed polysaccharides, specifically agar, furcellaran, kappa-carrageenan, and iota-carrageenan (see Sec. 4.8) (Fig. 20). Such an internal ether is known FIGURE 20 A 3,6-anhydro-a-D-g alactopyranosyl unit found in red seaweed polysaccharides. Pag e 171 as a 3,6-anhydro ring, whose name derives from the fact that the elements of water (HOH) are removed during its formation. A series of nonionic surfactants based on sorbitol (D-glucitol) is used in foods as water-in-oil emulsifiers and as defoamers. They are produced by esterification of sorbitol with fatty acids. Cyclic dehydration accompanies esterification (primarily at a primary hydroxyl group, i.e., C-1 or C-6) so that the carbohydrate (hydrophilic) portion is not only sorbitol but also its mono- and dianhydrides (cyclic ethers). The products are known as sorbitan esters (Fig. 21). The product called sorbitan monostearate is actually a mixture of partial stearic (C18) and palmitic (C16) acid esters of sorbitol (D-glucitol), 1,5-anydro-D-glucitol (1,5- sorbitan), 1,4-anhydro-D-glucitol (1,4-sorbitan), both internal (cyclic) ethers, and 1,4:3,6-dianhydro-D-glucitol (isosorbide), an internal dicyclic ether. Sorbitan fatty acid esters, such as sorbitan monostearate, sorbitan monolaturate, and sorbitan monooleate, are sometimes modified by reaction with ethylene oxide to produce so-called ethoxylated sorbitan esters, also nonionic detergents approved by the FDA for food use. 4.1.4.6 Nonenzymic Browning [10,12,30,59] Under some conditions, reducing sugars produce brown colors that are desirable and important in some foods. Other brown colors obtained upon heating or during long-term storage of foods containing reducing sugars are undesirable. Common browning of foods on heating or on storage is usually due to a chemical reaction between reducing sugars, mainly D-glucose, and a free amino acid or a free amino group of an amino acid that is part of a protein chain. This reaction is called the Maillard reaction. It is also called nonenzymic browning to differentiate it from the often rapid, enzyme-catalyzed browning commonly observed in freshly cut fruits and vegetables, such as apples and potatoes. When aldoses or ketoses are heated in solution with amines, a variety of reactions ensue, producing numerous compounds, some of which are flavors, aromas, and dark-colored polymeric materials, but both reactants, disappear only slowly. The flavors, aromas, and colors may be either desirable or undersirable. They may be produced by frying, roasting, baking, or storage. The reducing sugar reacts reversibly with the amine to produce a glycosylamine, as illustrates with D-glucose (Fig. 22). This undergoes a reaction called the Amadori rearrangement to give, in the case of D-glucose, a derivative of 1-amino-1-deoxy-Dfructose. Reaction continues, espectially at pH 5 or lower, to give an intermediate that dehydrates. Eventually a furan derivative is formed; that from a hexose is 5-hydroxymethyl-2-furaldehyde (HMF) (Fig. 23). Under less acidic conditions (higher than pH5), the reactive cyclic compounds (HMF and others) polymerize quickly to a dark-colored, insoluble material containing nitrogen. Maillard browning products, including soluble and insoluble polymers, are found where reducing sugars and amino acids, proteins, and/or other nitrogen-containing compounds are heated together, such as in soy sauce and bread crusts. Maillard reaction products are important contributors to the flavor of milk chocolate. The Maillard reaction is also important in the production of caramels, toffees, and fudges, during which reducing sugars also react with milk FIGURE 21 Anhydro-D-g lucitols (sorbitans). Pag e 172 FIGURE 22 Products of reaction of D-g lucose with an amine (RNH 2). proteins. D-Glucose undergoes the browning reaction faster than does D-fructose. Application of heat is generally required for nonenzymic browning. While Maillard reactions are useful, they also have a negative side. Reaction of reducing sugars with amino acids destroys the amino acid. This is of particular importance with L-lysine, an essential amino acid whose e-amino group can react when the amino acid is part of a protein molecule. Also, a relationship has been found between formation of mutagenic compounds and cooking of protein-rich foods. Mutagenic heterocyclic amines have been isolated from broiled and fried meat and fish, and from beef extracts. Heating of carbohydrates, in particular sucrose (see Sec. 4.2.3) and reducing sugars, without nitrogen-containing compounds effects a complex group of reactions termed caramelization. Reaction is facilitated by small amounts of acids and certain salts. Mostly thermolysis causes dehydration of the sugar molecule with introduction of double bonds or formation of anhydro rings. Introduction of double bonds leads to unsaturated rings such as furans. Conjugated double bonds absorb light and produce color. Often unsaturated rings will condense to polymers yielding useful colors. Catalysts increase the reaction rate and are often used to direct the reaction to specific types of caramel colors, solubilities, and acidities. Brown caramel color made by heating a sucrose (see Sec. 4.2.3) solution with ammonium bisulfite is used in cola soft drinks, other acidic beverages, baked goods, syrups, candies, pet

Leave a Reply

Your email address will not be published.