Application To Triglyceride Crystallization

Turning at last to foods, the rates at which oils crystallize and undergo polymorphic transitions are important in determining texture and physical stability in a variety of food products (Larsson, 1982). Edible oils consist almost entirely of triglycerides, with small amounts of mono- and di-glycerides and other minor components (Swern, 1979, Chapter 1). Hence attention has been concentrated on crystallization of triglycerides (TG's).

Before continuing this section, it may be useful to the reader to review a few basics. Thus the "diffraction angle", denoted "29", is defined as the angle between the line of the incident beam, continued through the sample, and the line of any given diffracted (or "scattered") ray from the sample to the X-ray detector. This is the angle that appears in the following figures.

By determining the diffraction angle, and knowing the X-ray wavelength, it is a simple matter to calculate the "Bragg spacing", or simply "spacing"; spacings are given in A in the figures. These spacings, which measure packing at the scale of molecules and above, are diagnostic of specific structures, e.g. crystal forms of fat or starch . Finally, it may help to note that, in the reciprocal fashion of diffraction science, smaller angles of diffraction correspond to larger spacings in the sample, and of course, larger angles of diffraction correspond to smaller spacings.

4.1. Classification of lipid crystals

XRD has given useful insights into food structures and dynamics, at the molecular and supramolecular levels. Thus XRD of fats demonstrated sixty years ago that crystal polymorphism is the basis for the intriguing phenomenon of multiple melting points of fats (Clarkson and Malkin, 1934), which are observed even when these compounds are painstakingly purified (Duffy, 1853). Even today, XRD is a primary method for classifying TG crystals (Fig. 1).

Although there have been disagreements in the past regarding the classification scheme (Clarkson and Malkin, 1934; Bailey et al., 1945; Lutton, 1945; Malkin, 1952), there appears now to be general agreement (Chapman, 1962). Three major forms recognized are: alpha; beta prime (with two sub-forms [Simpson and Hagemann, 1982]); and beta (two sub-forms also are indicated [Kellens et al., 1990]). A less common variant is the sub-alpha form (Jackson & Lutton, 1950; Fig. 1); structurally, the sub-alpha form resembles a rather disordered beta prime (Eads et al., 1992). The same crystal forms seen in purified TG's also are seen in commercial oils. In general, the different crystal forms are related by monotropic transitions, i.e. transitions that cannot be reversed directly (Hernqvist, 1990); the reversible transition between alpha and sub-alpha is an exception to this rule.

4.2. Kinetics of lipid crystallization

As noted, lipid crystallization is key to processing foods and food ingredients such as margarine, spreads, shortening, peanut butter and chocolate, and it plays an important role in cream cheese and ice cream making.

The sequence of events occurring when margarine emulsion is cooled — rapid crystallization to a less stable form, followed by a slower transition to more stable crystals (Fig. 2) — has been characterized using a conventional laboratory X-ray generator. However, the recording of such a process by XRD with fine time resolution has had to await the development of synchrotron radiation methods.

The first synchrotron radiation experiments on fat crystallization were reported in 1990 by a group based in Belgium and using a synchrotron source in Germany

Lipid Crystallization

DIFFRACTION ANGLE, 2 6

Figure 1. Wide-angle diffraction patterns from major crystal forms of TP. (a) Alpha form.

(b) Beta prime form, ß '2 variant (Simpson and Hagemann, 1982). (c) Beta form, (d) Sub-alpha form, with its ß '-like combination of a peak and shoulder; the shift to slightly smaller spacings is due to the low temperature (-130°C). Reproduced from Eads et al. (1992) by permission of the American Oil Chemists Society.

Figure 1. Wide-angle diffraction patterns from major crystal forms of TP. (a) Alpha form.

(b) Beta prime form, ß '2 variant (Simpson and Hagemann, 1982). (c) Beta form, (d) Sub-alpha form, with its ß '-like combination of a peak and shoulder; the shift to slightly smaller spacings is due to the low temperature (-130°C). Reproduced from Eads et al. (1992) by permission of the American Oil Chemists Society.

Figure 2.

DSC showing crystallization events in a margarine oil. Crystals formed during primary and secondary phases were identified by XRD using a conventional X-ray generator.

(Kellens et al., 1990). Exposure times were 5-10 seconds. Working on a "simple" TG, tripalmitin (TP; three CI6:0 chains), this group documented the transition from alpha to beta TP as the alpha form is heated at 5°C/min (Fig. 3). They also suggested, tentatively, that beta prime TP appears very briefly between the alpha and beta forms. A pair of heavy dashes in Figure 3, one at 20 1/2° and one at 23 1/2°, do indeed suggest the brief existence of an intermediate form just as the alpha TP crystals melt at about 50°C.

A year later, the same group reported the formation of beta TP crystals directly when alpha TP crystals were heated at 1.25°C/min. This time, the beta prime intermediate was excluded (Kellens et al., 1991).

Working at NSLS, Malcolm Capel and myself recorded crystallization data from trilaurin (TL), TP and tristearin (TS). We studied very rapid crystallization events: (1) formation of alpha crystals from the melt (see below), and also (2) the transition from alpha to beta during heating (Fig. 4). In the latter case, an intermediate form, identified as beta prime, is seen beyond any shadow of doubt.

In order to form alpha crystals, the molten trilaurin (TL) was cooled rapidly (~40°C/min). The crystals were then heated at a similar rate. The series of 1/2 second XRD patterns recorded during heating was then summarized in a pseudo-color plot; a color version of this plot has appeared on the cover of INFORM (Blaurock, 1993). The black and white version in Figure 4 shows the single X-ray reflection expected for alpha, near the bottom of the figure, and the three major reflections expected for beta crystals, near the top of the figure (refer Fig. 1). Figure 4 also demonstrates the existence of intermediate crystals - which clearly are neither alpha nor beta. The intermediate form is identified as beta prime TL (Blaurock et al., 1992).

The intermediate beta prime crystals in Figure 4 were expected on the basis of DSC curves which showed two exothermic events upon rapidly heating alpha TL (Blaurock et al., 1992). In conformity with the monotropic transition scheme (Hernqvist, 1990), the two events were interpreted as follows: (1) a peak for the formation of beta prime crystals; and (2) a second peak with onset about 10 seconds later, interpreted as the transition from beta prime to beta TL. Hence the brief existence of the the intermediate crystals in Figure 4 was anticipated by DSC observations (see also Lutton, 1946). In the case of TP, a similar pair of DSC exotherms has been published by Kellens et al. (1990; 1991), which are consistent with the beta prime TP intermediate suggested by these authors (Kellens et al., 1990).

At the same time, there is a definite surprise in Figure 4: the beta prime TL peaks are considerably broader than usually observed (cf. Fig. 1). In the reciprocal manner of XRD, broader peaks indicate smaller structures (Guinier, 1963). In fact, the broad peaks near the middle of Figure 4 indicate crystallites less than 50A in diameter! The miniscule size of these beta prime crystallites may explain why they are so short lived, i. e. the surface/volume ratio is extraordinarily high, and there is correspondingly a large surface energy. As noted above, DSC heating curves demonstrate that beta prime TL crystals formed in this way are quite unstable while those formed directly from the melt are more stable. Presumably this is so because the latter are larger, as is indicated by the narrower diffraction peaks in Figure 1(b).

Diffraction angle (20)

Figure 3. Kinetics of alpha to beta transition in TP. The series was recorded at the rate of one pattern every 5 sec as the sample was heated at 5°C/min. At the bottom, alpha TP is represented by a single peak at an angle of 21 1/2°. The transition to beta TP at about 50°C, with three main peaks, appears abrupt. Reproduced by permission of Elsevier Scientific Publishers; c 1990.

Diffraction angle (20)

Figure 3. Kinetics of alpha to beta transition in TP. The series was recorded at the rate of one pattern every 5 sec as the sample was heated at 5°C/min. At the bottom, alpha TP is represented by a single peak at an angle of 21 1/2°. The transition to beta TP at about 50°C, with three main peaks, appears abrupt. Reproduced by permission of Elsevier Scientific Publishers; c 1990.

Figure 4. Kinetics of alpha to beta transition in TL. The series was recorded at the rate of one pattern every 1/2 sec as the sample was heated at about 40°C/min. The transition from alpha (bottom) to beta (top) is mediated by an intermediate form that lasts 9 sec. This form is identified as tiny beta prime crystals (see text). Reproduced by permission of the American Oil Chemists Society; c 1993.

Figure 4. Kinetics of alpha to beta transition in TL. The series was recorded at the rate of one pattern every 1/2 sec as the sample was heated at about 40°C/min. The transition from alpha (bottom) to beta (top) is mediated by an intermediate form that lasts 9 sec. This form is identified as tiny beta prime crystals (see text). Reproduced by permission of the American Oil Chemists Society; c 1993.

The kinetics of the formation of alpha crystals also have been recorded (Blaurock et al., 1992). This work is the most severe test of the technique since alpha crystals form more rapidly than the other forms. Indeed, owing to the monotropic relations between forms (Hernqvist, 1990), alpha crystals would not be seen at any time if they did not form more rapidly than the more stable forms.

As described above, diffraction patterns were recorded at the rate of one every 1/2 second in our study. In analyzing short exposures like that shown in Figure 5, we took advantage of the fact that the total intensity in a diffraction peak is defined with good statistical accuracy when the number of counts in the entire peak is of the order of one thousand. If the peak is covered by a large number of recording points (electronic "channels"), say 40, then the number of counts per channel may be small, i.e., just 25 counts on average. This means that the peak shape in Figure 5 is not so precisely defined, but the total intensity is known with a 3% statistical uncertainty (Poisson statistics). As to peak shapes and peak positions, we relied on published data, as well as our own longer XRD exposures (Fig. 1), showing the alpha, beta prime and beta diffraction peaks (Bailey et al., 1945; Hoerr, 1964; Hernqvist and Larsson, 1982).

An example of the formation of alpha TP crystals from the melt is shown in Figure 6. Exposures were taken at the rate of one every 1/2 second, with about 1 millisecond between exposures required to store the pattern before starting the next exposure. The 'S'-shaped curve in Figure 6 is typical of crystallization events (Avrami, 1939).

The series of integrated intensities in Figure 6 have been analyzed in terms of the kinetics theory put forward by Melvin AvramL(1939). At the onset of ^rystallization, the data are fit well by a curve of the format . A curve of the form t fits the data significantly less well. A curve of the form t fits the data well, but this would require an early start-up of crystallization, for which there is as yet no supporting evidence (Blaurock et al., 1992).

These results can be used to test possible kinetic schemes such as those put forward by Avrami (1939). The data in Figure 6 are consistent with two, alternative kinetic schemes, both of which are simple but unexpected. One interpretation is that (a) nucleation of alpha crystals is a random event with equal probability at all times, as is found for many other crystals (Avrami, 1939), and (b) the accumulation of crystal mass is proportional to the square of the time, rather than the cubic dependence seen for crystals in general. A second possible interpretation is that crystal mass does, in fact, accumulate as the cube of the time, as for other crystals (Avrami, 1939), but that nucleation of alp^a crystals is not a random evejjit. Either of these interpretations would account for the t fit. It should be noted that t kinetics - as found for other kinds of crystals (Avrami, 1939) - seem unlikely, but they have not been ruled out decisively (Blaurock et al, 1992).

In closing this section, it may be mentioned that XRD data are commonly divided into two regimes, namely the small- or low-angle region; and the wide-angle region. The two regions are distinguished on the basis of the different technologies typically used to record the XRD pattern. The patterns discussed in this chapter are all wide-angle patterns, but useful, complementary information can be obtained by recording the small-angle patterns as well.

0 0

Post a comment