J. Dairy Sci. 84:24022412 American Dairy Science Association, 2001.

Thermal and Structural Behavior of Anhydrous Milk Fat. 2. Crystalline Forms Obtained by Slow Cooling
C. Lopez,*, F. Lavigne,*, P. Lesieur, G. Keller,* and M. Ollivon*
` *Equipe Physico-Chimie des Systemes Polyphases, UMR 8612 du CNRS, 5 rue J. B. Clement, 92296 Chatenay-Malabry, France Arilait Recherches, 42 rue de Chateaudun, 75314 Paris cedex 09, France Laboratoire pour lUtilisation du Rayonnement Electromagnetique, Bat. 209D, Universite Paris-Sud, 91898 Orsay, France

ABSTRACT The crystallization behavior of milk fat has been examined on slow cooling at 0.1C/min from 50 to 15C, to determine the variations of triacylglycerol organizations as a function of temperature. The experiments have been conducted with an instrument allowing coupled X-ray diffraction (XRD) at both small and wide angles and high-sensitivity differential scanning calorimetry (DSC) recordings from the same sample by taking advantage of the high-energy ux of a synchrotron. On slow cooling, milk fat triacylglycerols sequentially crystallize in four different lamellar structures with double-chain length of 41.5, 48.3, and 39.2 A and a triple-chain length of 62.2 A stackings. Simultaneous wide-angle XRD has shown that initial nucleation occurs in a packing of type at about 24C. For temperature < 13C, triacylglycerols crystallize in an hexagonal subcell of type, leading to the coexistence of the + polymorphic forms, which is recorded until 15C. Thermal analysis allowed to correlate the formation of the different crystalline species monitored by XRDT (XRD as a function of temperature) to the exothermal events recorded simultaneously by differential scanning calorimetry. The evolution of the species formed during crystallization was also monitored on heating at 2C/min. The absence of polymorphic evolution on heating, as well as the high nal melting point observed, about 40 to 41C, conrmed that cooling at 0.1C/min leads to quasi equilibrium. (Key words: triacylglycerols, polymorphism, differential scanning calorimetry, X-ray diffraction) Abbreviation key: AMF = anhydrous milk fat, DSC = differential scanning calorimetry, 2L = double-chain length, 3L = triple-chain length, SAXD = small-angle Xray diffraction, T = temperature, TG = triacylglycerols, WAXD = wide-angle X-ray diffraction, XRD = X-ray

diffraction, XRDT = X-ray diffraction as a function of temperature. INTRODUCTION Natural fats are mainly composed of triacylglycerols (TG), which determine their physical and thermal properties. Milk fat is one of the most complex fats found in nature. TG represent 97 to 98% of the total lipid content. The other components are diacylglycerols, monoacylglycerols, free fatty acids, free sterols, and phospholipids. Milk fat contains a large variety of fatty acids with different chain lengths, degrees of saturation, and distribution onto glycerol (Jensen and Newburg, 1995). Due to its fatty acid composition, the melting range of anhydrous milk fat (AMF) is broad and spans from about 40 to 40C. Furthermore, TG are characterized by a complex thermal behavior in relation with their polymorphism of monotropic type. Polymorphism is a general feature of TG that results from the different possibilities of lateral packing of the fatty acid hydrocarbon chains and of the longitudinal stacking of molecules in lamella. These two levels of organization are easily identiable from the short and long spacings observed by X-ray diffraction (XRD) at wide (WAXD) and small (SAXD) angles, respectively. For TG, long spacings are commonly double- or triplechain lengths (2L or 3L). The short spacings are characteristic for the type of polymorph. The TG nomenclature summarized by Larsson (1966) is now widely accepted. In brief, the three main types of polymorphic crystal forms in TG are the alpha (), beta prime (), and beta () forms, in the order of their increasing stability. They have been related to different subcells that have been described in detail (Garti and Sato, 1988; Larsson, 1972; Ollivon and Perron, 1992; Small, 1986). Regarding thermal properties in the literature, milk fat is usually separated into three main fractions with the following different melting points: low melting point fraction, middle melting point fraction, and high melting point fraction (Timms, 1980a). It is often assumed that these main fractions would crystallize separately and, thus, would behave more or less as independent

Received March 20, 2001. Accepted June 6, 2001. Corresponding author: M. Ollivon; e-mail: michel.ollivon@cep. u-psud.fr.

2402

ANHYDROUS MILK FAT POLYMORPHISM

2403

pseudocomponents (Marangoni and Lencki, 1998; Timms, 1980a). This view was supported by the occurrence of three main endothermic peaks in a heating thermogram recorded by differential scanning calorimetry (DSC) whose enthalpies and temperatures vary depending on the thermal treatments applied to AMF (Deffense, 1993; Timms, 1980a,b). The crystallization and melting of milk fat are complicated by the occurrence of transitions between several polymorphic crystal forms, which depend strongly on heating or cooling rates and on the entire thermal history of the sample (Ollivon and Perron, 1985). Recently, access to X-ray high-ux sources (synchrotron) has permitted the study of the thermal behavior of pure TG by XRD as a function of temperature (XRDT) (Arishima et al., 1991; Kellens et al., 1990; Lavigne and Ollivon, 1993; Sato et al., 1989). More recently, the polymorphism and phase transitions displayed by complex TG mixtures also have been characterized by XRDT and compared with DSC recordings (Keller et al., 1996; Loisel et al., 1998; ten Grotenhuis et al., 2000). Lavigne (1995) used these techniques to study the thermal behavior of AMF and its fractions. Chemical TG analysis of each fraction allowed the complex thermal proles recorded by DSC to be decomposed into elementary phase transitions and to identify the TG involved in each of them. A model of the molecular organization of the phase of complex fat mixtures was proposed (Lavigne, 1995). Recently, Lopez et al. (2000) studied the unstable species formed by quenching of cream from 50 to 8C and their evolution to more stable varieties either as a function of time or by heating of the sample. This demonstrates that the combinations of these two techniques of thermal analysis is especially adapted for the characterization of coexisting organizations that result from the phase separations frequently occurring in complex fat mixtures. Following the characterization of the most unstable species (Lopez et al., 2001a), the purpose of this second article of a series is to study the crystalline varieties formed by TG of AMF when submitted to very slow cooling rate. An equivalent thermal treatment is applied in parallel to fat globules of cream, to determine the inuence of dispersion onto TG crystallization (Lopez et al., 2001b). Combined and coupled X-ray diffraction as a function of temperature and high-sensitivity DSC techniques were used for this study at a cooling rate of dT/dt = 0.1C/min, between 50 and 15C. The evolution of these structures is also monitored during a subsequent heating at 2C/min. MATERIALS AND METHODS Milk Fat AMF was extracted from fresh concentrated creams (fat content 60% (wt/vol) obtained directly from the

dairy plant (Lactalis, France). They correspond to milk collected in spring 1998 in the south of Normandy, France. Cold extraction of AMF was performed as previously described (Lopez et al., 2001a). Combined XRDT and DSC Measurements A technique allowing simultaneous time-resolved synchrotron XRDT and high-sensitivity DSC to be performed in the same apparatus from the same sample (Keller et al., 1998; Lopez et al., 2001a) was used for the experiments. The XRDT data is recorded simultaneously at small and wide angles through two one-dimensional position-sensitive proportional detectors on D22 bench (wavelength = 1.5498 A) of the DCI synchrotron of L.U.R.E. (Laboratoire pour lutilisation du rayonne ment electromagnetique, Orsay, France) at sample-todetector distances of about 177.4 and 30 cm, respectively. Crystalline form of high-purity tristearin was used as reference for WAXD data calibration, as previously described (Lopez et al., 2001a), and silver behenate was used as reference for SAXD data (Blanton et al., 2000). DSC enthalpies and temperature were determined after lauric acid melting calibration, as described previously (Keller et al., 1998). All XRD patterns were recorded by transmission through glass capillaries (GLAS; Muller, Berlin, Germany). Samples were prepared by loading about 20 l of the melted AMF in the bottom of these thin Lindeman glass capillaries by using a syringe and a thin Teon capillary. This technique allows preparation of capillary samples not wall-polluted by deposited materials. The sample in the capillary was heated to 50C and held at this temperature for 10 min to ensure that all nuclei were eliminated. A hot sample of AMF was cooled from 50 to 15C at a cooling rate of 0.1C/min. This rate (6C/h) has been chosen as slow enough to approach equilibrium and then favor the growth of stable polymorphs, as well as a good quality of XRD recordings, but also high enough to still permit calorimetric signal detection (Keller et al., 1998). The still acceptable signal/noise ratio of the DSC recording, considering the very slow rate used, guarantee that such conditions are fullled. XRD patterns and DSC signal were recorded during cooling with a time frame of 300 s and each 20 s, respectively, by using a single computer for thermal control and data acquisition. After crystallization, the sample of AMF was heated at 2C/min, from 15 to 50C, with an XRD time frame of 60 s and DSC signal recorded each 3 s. Both cooling and heating XRD patterns presented below are raw recordings (no subtraction and no smoothing) except for the domain, 0.09 < q < 0.10 A1,
Journal of Dairy Science Vol. 84, No. 11, 2001

2404

LOPEZ ET AL.

in which a small peak has been subtracted. This peak was present in the whole temperature range considered. The subtraction of this peak, which is obviously not perfect due to the presence of a phase transition in the range studied, was made by tting it at high and low temperature and calculating all intermediate positions by interpolation. We found that the peak subtracted originated from solvent extraction and was not present on patterns recorded from AMF obtained by centrifugation, then without solvent. This peak, ob served at 0.098 A1 (64.1 A), is likely due to phospholipid bilayers extracted by the solvent mixture from milk fat globule membrane. The origin of this peak justies its removal in a study of TG crystallization. Wavelets observed around q 0.13 A1 and q 0.16 1 are generated by a small-angle position-sensitive A gas detector. Each X-ray diffraction pattern was analyzed by using Peakt software (JANDEL Scientic, Germany). X-ray diffraction peaks were tted by gaussian-lorentzian (sum) equation as described elsewhere (Lopez et al., 2001b), to determine position of the maximum, maximal intensity and area under each peak. DSC Experiments Thermal analyses were conducted by DSC, using a DSC-7 (Perkin-Elmer, St. Quentin en Yvelines, France) equipped with Intracooler II and running under Pyris software. Samples of AMF were loaded in aluminum pans of 50 l (pan, part no. B014-3021, and cover, part no. B014-3004) hermetically sealed. An empty, hermetically sealed aluminum pan was used as reference. Calibration was made with lauric acid (m.p. 43.7C, Hm = 35.7 KJ/mol) (Grabielle-Madelmont and Perron, 1983). Crystallization behavior was monitored with the temperature scanning program set from 60 to 15C at 0.1C/min, and melting behavior was monitored from 15 to 60C at 2C/min. RESULTS AND DISCUSSION Milk fat was very slowly cooled in the glass capillaries of the coupled XRDT/DSC apparatus, from 50 to 15C, with a cooling rate at |dT/dt| = 0.1C/min, to study the formation of stable crystalline forms of TG. After crystallization, the sample was analyzed on heating from 15 to 50C, at dT/dt = 2C/min. XRDT recordings were performed with the synchrotron beam simultaneously at small and wide angles during the 11 h of the experiments. The analysis of the intensity evolution of X-ray diffraction lines as a function of temperature and DSC signal recordings allowed us to characterize and delimit the domains of existence of the crystalline varieties formed.
Journal of Dairy Science Vol. 84, No. 11, 2001

Slow Cooling of AMF The plots of the XRD patterns, recorded as a function of time at both small and wide angles during cooling of AMF at |dT/dt| = 0.1C/min, versus temperature (T) as three-dimensional viewing in the 30 > T > 15C range, are presented in Figure 1. The SAXD patterns (Figure 1a) show the progressive development of several diffraction lines corresponding to crystallization of TG in milk fat as a function of temperature. The occurrence of crystalline varieties during cooling is summarized in Table 1. For T > 24C, only scattering signal from liquid crystalline organization of the AMF is observed, with broad peak maximum centered at q = 0.27 A1. The evolution of maximal intensity of each diffraction line recorded during cooling was plotted as a function of temperature in Figure 2a. On cooling at 0.1C/min, crystallization of milk fat TG starts by the appearance, at about 23.7C, of a diffrac tion line at q = 0.15 A1 (41.5 A). This rst crystalline form corresponds to a lamellar structure with a longitudinal organization of molecules in a double chain-length stacking (2L). Next to this major diffraction line, which increases in intensity on cooling, a weak XRD bump, corresponding to the formation of a second 2L variety (2L2), is also recorded at q = 0.13 A1 (48.3 A). For T 12.6C, two additional diffraction lines, centered at about q = 0.10 and 0.195 A1 (62.2 and 32.3 A), are observed. The simultaneous increase in intensity versus T of these two lines during cooling suggests that they belong to the same crystalline variety, the peak at 32.3 A corresponding to the second order of the peak at 62.2 A (Figure 2a). The crystalline form, with a thick ness of 62.2 A, likely corresponds to a lamellar structure with a triple chain-length organization (3L) of the TG molecules. From 5.5C, a diffraction line corresponding to a new 2L stacking (2L3) of milk fat TG molecules is recorded at q = 0.16 A1 (39.2 A). The increase in intensity of this line as a function of temperature is correlated with a decrease in intensity of the peak at 41.5 A. We concluded that this new bilayered structure (2L3) corresponds either to the progressive tilt of the molecules constitutive of the rst crystalline variety (2L1) formed during the cooling of milk fat or it incorporates part of these molecules into a new structure or both. XRDT patterns recorded simultaneously at wide angles (Figure 1b) can be divided in three domains. A signal of X-ray scattering centered at q = 1.38 A1 (4.55 A) is recorded from 50 to about 24C, meaning that all the TG are in a liquid state, according to Larsson (1972). The recording of XRD at wide angles allows identication of the packing of the alkyl chains of acylglycerols in characteristic subcells. From about 23C, two diffrac tion lines, centered at q = 1.46 A1 (4.3 A) and q = 1.6

ANHYDROUS MILK FAT POLYMORPHISM

2405

Figure 1. Three-dimensional plots of small- (a) and wide- (b) angle X-ray diffraction patterns recorded simultaneously with differential scanning calorimetry (DSC) as a function of time during cooling of anhydrous milk fat at 0.1C/min from 50 to 15C. The types of longitudinal, double (2L) or triple (3L) chain lengths, as well as their associated long spacings, and lateral (b) packings (, , and ) are indicated on the gure. Raw data are shown at both small and wide angles except in the 0.09 < q < 0.10 A1 range (see Materials and Methods).

A1 (3.88 A), correspond to a rst organization of the lateral packing of the alkyl chains of TG in milk fat, corresponding to an orthorhombic packing of the chains, also called form. These results show that on cooling at 0.1C/min, nucleation of AMF occurs in the form, with a liquid transition. At about 13C, the occur rence of a second diffraction line at q = 1.49 A1 (4.22 ) indicates the formation of a new crystalline arrangeA

ment of TG in AMF. This line corresponds to the formation of a hexagonal packing of the chains, also called form, which is the less table of the possibilities of crystalline arrangements of the TG. We deduced that the occurrence of this hexagonal packing results from a liquid transition. Because of the polymorphism of monotropic type exhibited by TG, the transi tion is forbidden. A weak line at q = 1.37 A1 (4.58

Table 1. Initial temperature of crystallization and characteristics of the different crystalline species formed during cooling of anhydrous milk fat at 0.1C/min. Crystalline organization Long spacings and type () Short spacings and type () Temperature (C) 23.7 41.5 A (2L1) s 4.3 A () 3.88 A 22.7 48.3 A (2L2) w 4.3 A () 3.88 A 12.6 62.2 A (3L(001)) 32.3 A (3L(002)) 4.3 A () 3.88 A 4.22 A () 5.5 39.2 A (2L3) 4.3 A () 3.88 A 4.2 A ()

2L = double-chain length; 3L = triple-chain length; s = strong; w = weak. Journal of Dairy Science Vol. 84, No. 11, 2001

2406

LOPEZ ET AL.

A) is also observed. This line might correspond to the formation of traces of species. This is in agreement with the observation of a small amount of crystals in AMF after long storage (deMan, 1961; Timms, 1979; Woodrow and deMan, 1968). The evolution of maximal intensity of the main peaks recorded at wide angles as a function of temperature is shown Figure 2b.

Figure 2. Evolution of intensities of X-ray diffraction (XRD) peaks (shown in Figure 1) as a function of temperature and differential scanning calorimetry (DSC) curve. Evolution of the long and short spacings of milk fat as determined by (a) small- and (b) wide-angle XRD recordings during cooling of anhydrous milk fat at 0.1C/min from 40 to 15C. = 41.5 A; = 48.3 A; + = 62.2 A(3L001); L = 32.3 A(3L002); = 39.2 A; = 4.3 A; = 3.88 A; N = 4.2 A(). (c) DSC crystallization curve recorded simultaneously with XRD. Journal of Dairy Science Vol. 84, No. 11, 2001

The simultaneous recording of small- and wide-angle XRD patterns on the same sample of AMF during cooling and the study of the variations in intensity of the peaks as a function of temperature allow relation of the lateral packings to the longitudinal organizations of TG. To compare the development of the different lines recorded and make the proper attribution of small- and wide-angle peaks to crystalline varieties, variations of their intensities between 40 and 15C were plotted on the same graph (Figure 2). The -type organization of acylglycerol chains recorded at wide angles (Figure 2b) is correlated with the formation of the line at 41.5 A (2L1) recorded at small angles (Figure 2a). These results show that from the beginning of the crystallization process to about 13C, the lamellar crystalline variety formed in milk fat corresponds to a stable form with a bilayered stacking. Below this temperature during the slow cooling of AMF, the form recorded at wide angles is related to the formation of the 3L (62.2 A) structure. The coexistence of the + polymorphic forms is recorded until the end of the experiments at 15C. Similar results were found by Lavigne (1995) by using separate DSC and XRDT techniques on cooling at 0.1C/min. He observed two exotherms at 17 and 25C and two diffraction peaks at small angles attrib uted to the crystallization of two species, 2L (41 A) ). Indeed, the low resolution at small angles and 3L (62 A of the XRDT setup used (sample-to-detector distance of about 300 mm) did not allow the characterization of the four and species described above. Thermal analysis was recorded simultaneously with XRD from the same sample of AMF during its cooling, thanks to XRDT/DSC coupling at 0.1C/min (Figure 2c). The DSC crystallization curve shows three exotherms that can be correlated with the formation of the lamellar species analyzed above (Figure 2a and b). The crystal types are noted, in Figure 2c, below the exothermic peaks to which they are attributed. The initial crystallization temperature measured is 24C. The rst two exotherms correspond to major processes of crystallization with respect to the energies released. The broad shape and asymmetry of the rst crystallization exotherm lead us to think that it might correspond to the crystallization of both 2L1 (41.5 A) and 2L2 (48.3 A) species. The second and third exotherms are correlated with the formation of the lamellar 3L (62.2 A) and ) structures, respectively. It is likely that 2L (39.2 A at the end of the experiments all the TG are not crystallized, since the melting range of AMF spans from about 40 to 40C. For technical reasons of sensitivity, it was not possible to record crystallization of AMF at a cooling rate slower than 0.1C/min. Then, it was not possible to determine whether the crystallizations obtained at this

ANHYDROUS MILK FAT POLYMORPHISM

2407

rate are slow enough to more or less correspond to ultimate stabilization of the system. This point is addressed elsewhere (Lopez et al., 2001b). As it is well known that the crystallization process is never at equilibrium and difcult to reproduce exactly, we decided to get multiple recordings of milk fat crystallization to assess this point. Thermal analysis was conducted with a DSC-7 (Perkin Elmer) during cooling of AMF from 60 to 15C at |dT/dt| = 0.1C/min on three replicates (Figure 3). Three exothermal events were always observed with an initial crystallization temperature onset = 24.1 0.4C. From the analysis of this gure, we can deduce that 1) the rst exotherm recorded on cooling is, in fact, composed of at least two overlapped peaks, 2) the third exotherm largely uctuates in temperature and energy, and 3) the two main exotherms are always recorded at the same temperature and with comparable enthalpies. Although the existence of three, and likely four, species is clearly established by DSC recording, the large uctuations observed in the determination of peak areas at the very low scanning rate (0.1C/min) prevented us from giving precise enthalpy determinations. X-ray diffraction peaks correspond to crystalline species formed by AMF TG. However, the maximum intensity of the diffraction lines is strongly inuenced by the effect of crystal size and crystal perfection. Crystalline defects cause an increase in peak width and a proportional decrease in maximum intensity. At 11C, the middle-height width of the peak at 41.5 A (q = 0.0095 1) is half as important than that of the peak at 62.2 A A (q = 0.0190 A1) (Figure 1a). As the study of the width of the diffraction lines allows determination of

Figure 4. Estimation versus temperature of total solid-phase content of anhydrous milk fat and its decomposition into crystalline species. All phase contents, deduced from X-ray diffraction peak areas recorded at small angles during cooling at 0.1C/min (Figure 1a), are expressed as percentages of the total fat.

Figure 3. Crystallization curves of anhydrous milk fat recorded at 0.1C/min with a DSC-7 (Perkin Elmer). The three curves correspond to triplicate recordings.

structural information on the crystalline varieties, the sharper diffraction line of the 2L variety can be attributed to 1) improved crystal perfection, 2) larger crystal sizes, and 3) restricted TG composition (Kellens et al., 1990). The crystallization of the 3L (62.2 A) variety during cooling is correlated with an increase of the Xray scattering signal recorded at very small angles (q 0.03 A1). This X-ray signature can be attributed to the formation of crystals with smaller sizes compared with those formed by the 2L variety. Because the total scattered intensity is not affected by the effect of crystal perfection and crystal size, the crystallization process can therefore be followed more quantitatively from the integral of diffraction peaks (Kellens et al., 1990). The areas under the peaks recorded in an XRD pattern were related to the quantities of the solid phases at a dened temperature during cooling of AMF by using Peakt. The areas under each peak corresponding to a crystalline species were determined and summed. The quantity of liquid TG associated with each XRD pattern was estimated from the intensity of the broad peak recorded at q = 0.27 A1, by comparison to TG in their liquid state (50C). Areas related to crystalline varieties correspond to the quantity of each solid phase formed. The percentage of each solid phase and the liquid phase are plotted as a function of temperature (Figure 4). We arbitrarily estimated a 5% liquid phase at 15C, since this temperature does not correspond to completion of crystallization.
Journal of Dairy Science Vol. 84, No. 11, 2001

2408 Melting of AMF

LOPEZ ET AL.

After crystallization upon cooling at |dT/dt| = 0.1C/ min, the AMF sample was heated at dT/dt = 2C/min, from 15 to 50C, to follow the evolution of the crystalline species formed during slow cooling. The plots of the XRD patterns, recorded as a function of time at small and wide angles, versus temperature as a three-dimensional viewing, are presented in Figure 5a and b, respectively. The small-angle plot shows successively, as a function of temperature, the decrease in intensity of the diffraction lines characteristic of the 3L variety and of the 2L species until their nal melting. XRDT data recorded at wide angles show the coexistence of the + forms from 15C to the melting of the unstable form, and the melting of the orthorhombic packing of the alkyl chains ( form) until the nal melting point of AMF. The intensity variations of the long and short spacings recorded during heating of AMF were plotted as functions of temperature, in Figure 6a and b, respectively. The simultaneous decrease in intensity of the

Figure 6. Simultaneous X-ray diffraction (XRD) as a function of temperature and differential scanning calorimetry (DSC) recordings observed during heating of anhydrous milk fat at 2C/min from 15 to 50C. Evolution of the long and short spacings of milk fat as determined by (a) small-angle and (b) wide-angle XRD. + = 62.2 A = 48.3 A; = 41.5 A; N = (3L001); L = 32.3 A (3L002); = 39.2 A; 4.2 A(); = 4.3 A; = 3.88 A. (c) Corresponding DSC recording. Figure 5. Three-dimensional plots of small- and wide- (insert) angle X-ray diffraction data corresponding to structural evolution of anhydrous milk fat as a function or temperature during heating of the sample at 2C/min from 15 to 50C. Raw data are shown at both small and wide angles except in the 0.09 < q < 0.10 A1 range (see Materials and Methods). Journal of Dairy Science Vol. 84, No. 11, 2001

ANHYDROUS MILK FAT POLYMORPHISM

2409

lines recorded at small angles, at 62.2 and 32.3 A (Fig recorded at wide angles ure 6a) with the line at 4.2 A (Figure 6b), corresponds to the progressive melting of the 3L structure until the nal melting point of this crystalline variety at 13C. For T > 13C, the acylglycerol chains are only packed in the form. SAXD patterns show the progressive melting of the 2L (39.2 A) structure until its nal melting point at 19.5C, and the melting of the 2L (48.3 A) variety until its disappearance at about 24C. For T > 18C, the decrease in intensity of the 41.5 A line corresponds to the melting ) variety until the nal melting point of the 2L (41.5 A of AMF TG at about 40C. The melting of this structure is accompanied by an increase of its lamellar thickness from 41.5 A (21.7C) to 42.3 A (38.8C). The long spacings recorded at small angles do not show a signicant evolution during heating of AMF, meaning that they correspond to independent species. The absence of structural evolution, since no polymorphic transition is observed, during heating at 2C/min, shows that cooling of AMF at 0.1C/min leads to the formation of stable species. The DSC curve, recorded simultaneously with SAXD and WAXD during the heating at 2C/min of the AMF sample, is presented in Figure 6c. The vertical dashed lines drawn delimit the domains of existence of the crystalline varieties observed by XRDT. Several endothermic peaks are observed before the nal melting of AMF occurring at 40.5C (peak offset). From 15 to about 0C, calorimeter equilibration overlaps the DSC signal and the likely progressive melting of the less stable species. From 0 to 24C, two different endotherms can be dened. The rst one is correlated with the melting of the 3L structure, and the second corres ponds to the melting of both the 2L (39.2 A) and 2L (48.3 ) varieties. These overlapped endothermic events A correspond to melting of middle-melting point crystals. The last endothermic peak, which spans from about 24C to the nal melting point of AMF at 40.5C, is related to the melting of the high melting point 2L (41.5 A) crystals. In a previous study, Lavigne (1995) found that the 3L (62 A) species, attributed to both the low- and mediummelting fractions, melted at about 18C; whereas the melting of the 2L (41 A) species, corresponding to the high-melting fraction, spreads from about 25 to 39C. Origin of TG Segregation This study shows that a TG molecular segregation occurs in milk fat on slow cooling at 0.1C/min. The rst crystals formed by high-melting molecules show double chain-length organization mainly characterized by a 41.5 A long spacing (a weak 48.3 A species is also

observed). Previous studies showed that such bilayer conformations are constituted by TG with three acylglycerol chains rather similar in length (about 16 carbons) and unsaturation (mainly saturated and monounsaturated TG), resulting from the tendency for equal hydrocarbon chains to be in the same chain layer (Larsson, 1994). The TG, constituted by an acyl chain that differs much from the other two in length (by four carbons or more) or in unsaturation, tend to form triple chain-length structures (Larsson, 1994; Small, 1986). This differing chain generally segregates into a specic layer, whereas the remaining two chains are located in another layer. The 3L (62.2 A) structure formed on cooling should correspond to the crystallization of TG, with a short acyl chain, number of carbons 10, segregated in a separate layer. In the same experimental conditions, Lavigne (1995) showed that the rst crystals formed are constituted by trisaturated TG as tripalmitin (PPP), which crystallize under a stable 2L stacking. The second crystalline variety is composed of TG containing an unsaturated acyl chain or a short-chain acyl chain, as it is the case for BuPP and PPO (P = palmitic, O = Oleic, Bu = butyric acids). These results show that slow cooling of AMF prevents the formation of unstable mixed crystals in which all TG would have been assembled, regardless of their nature, and that would further evolve as a function of time or temperature. Coupled time-resolved synchrotron XRDT at both small and wide angles and DSC allowed identication with a precision that was not attained before, to our knowledge, of the crystalline structures formed during slow cooling of AMF. Furthermore, this technique has shown that each exothermal event recorded by DSC corresponds to the formation of a different crystal type recorded by XRD. Indeed, the presence of sharp exotherms and endotherms in DSC recordings must be related to the formation of well-dened crystalline species. On the other hand, the observations in this study of a rather limited number of well-dened peaks at small angles, and of their independent thermal behaviors, conrm the existence of such characteristic crystalline species in milk fat. It is surprising that such complex TG mixtures adopt so few different crystalline structures. However, the ability of dry fat fractionation to separate such crystal species into so-called olein and stearin fractions, after slow cooling at rates comparable with that used in this study, conrms that they correspond to different TG compositions (Deffense, 1993; Kaylegian and Lindsay, 1994; Lavigne, 1995). Indeed, all these data show that AMF crystallizes and melts in several independent steps corresponding to the phase separation of several groups of TG. Each fraction acts as an independent solid solution (Timms, 1980a; Tverdokhleb and Vergelesov, 1974). The accuracy attained
Journal of Dairy Science Vol. 84, No. 11, 2001

2410

LOPEZ ET AL.

in long-spacing determinations by using synchrotron SAXD, <1 A, already allows the precise characterization of such phases. Then, it also might allow, in the future, the determination of the mean type of TG involved in each peculiar phase, as done above in a preliminary tentative attribution of 2L and 3L crystals. Concerning data analysis, the choice of separating the different contributions of each crystalline species by using Peakt software was deliberately made to solve the complex overlapping of diffraction peaks. Despite the high-resolution power of D22 bench (about 1024 channels of 0.2 mm, at a sample-to-detector distance of 1.8 m), several diffraction peaks overlap especially in patterns recorded at low temperature. In this respect, the asymmetrical broadening of the peak at 41.5 A, observed at low temperature, has been analyzed. First, it should be observed that this broadening only starts below about 5C. This corresponds approximately to the domain in which the second order of the 3L form (3L002) develops. As this variety shows the second diffracting power of the SAXD patterns and this peak is close to that of the 2L1 (41.5 A), the overlapping of both peaks might have been at the origin of this asymmetric broad shoulder. Peakt analysis has shown that this was not the case, since the difference in amplitude is too large in the vicinity of the 2L peak. Then we analyzed this broadening as the presence of crystalline de fects in the 2L diffraction peak at 41.5 A. The progressive incorporation of TG with odd, shorter, or ramied chains in the main structure might explain such a broadening and especially its observation at low temperature (Guinier, 1964). On the other hand, Peakt analysis allowed separation of a gaussian peak centered at 39.2 A, the contribution of which also ts the asymmetrical broadening. This interpretation of the X-ray data was preferred, since no intensity evolution of the peak at 41.5 A is observed in the range of 5 to 15C. Indeed, cocrystallization of TG, introducing defects in the structure, would have been accompanied by an increase of the whole peak intensity, especially if epitaxy is suspected to induce such a crystal growth. The other interpretation was not ruled out and further experiments at a larger sample-to-detector distance, and using brighter synchrotron, is under way. Comparison with Slow Cooling of Milk Fat Globules of Cream The crystalline varieties formed by TG during slow cooling of AMF are different from those formed by the same TG within milk fat globules during a similar cooling process (Lopez et al., 2001b). In brief, on cooling at 0.15C/min, TG dispersed in fat globules of cream (fat content > 60%) form two double chain-length organizaJournal of Dairy Science Vol. 84, No. 11, 2001

Figure 7. Comparison of the crystallization curves recorded by differential scanning calorimetry during cooling of anhydrous milk fat at (a) 0.10C/min and b) 0.15C/min, simultaneously, with Xray diffraction.

tions (46.5 and 40 A) and two triple chain-length stack ings (71.3 and 65 A). Furthermore, nucleation occurs in the form before formation of a packing of type. Figure 7 shows the crystallization curves recorded by DSC during cooling of AMF at 0.1 and 0.15C/min. Both recordings show similar thermal events. Reducing the cooling rate by 0.05C/min does not induce any change in the thermal behavior of AMF. We concluded that the different structural behaviors of TG in AMF (bulk) and cream (emulsion) are therefore related to their dispersion state. The main differences observed on crystallization at a slow cooling rate between AMF and cream are summarized in Table 2. Both thermal and structural observations suggest that the slow cooling of AMF leads to the formation of a stable species, whereas that of cream yields a mixture of stable and unstable varieties. This can be explained by the dispersion of nucleation centers within the emulsion droplet population, resulting in a lack of nucleation in most of the fat globules. The cooling rate considered is not slow enough to allow nucleation in stable varieties then unstable species form within fat globules. The apparent nal result of this kinetic trap is that the dispersion state maintains the formation of unstable varieties and delays the transition toward a more stable species. CONCLUSION Understanding the crystallization behavior of milk fat is of great importance with regard to its economical

ANHYDROUS MILK FAT POLYMORPHISM Table 2. Comparison of thermal and structural behaviors of anhydrous milk fat and creama observed at slow cooling rates (Rc 0.15C/min). Anhydrous milk fat Varieties observed on cooling Types of chain stackingb , then + 2L (41.5 A) very strong 2L (48.3 A) weak 3L (62.2 A) strong 2L (39.2 A) medium melts rst, then melts 24.1 0.4C 2 majors followed by a minor one No polymorphic transition Cream , then + 2L (46.5 A) medium 2L (40 A) medium 3L (71.3 A) strong 3L (65 A) weak melts rst, then melts 20 20.5C 1 major Unstable varieties transform into 2L (40 A) with increase of LS to 41.6 A 39C

2411

Transitions observed on heating Thermal behavior on cooling:onset Number of thermal events Thermal behavior on heating

Melting offset temperature (C)

a b

40.6 0.5C

From Lopez et al. (2001b). In order of occurrence.

impact, especially in high-fat-containing products. The experiments reported in this study give deeper insight into the crystallization of milk fat. Coupling of timeresolved synchrotron X-ray diffraction, recorded at both small and wide angles simultaneously, to DSC allowed characterization of the molecular organization of TG formed as a function of time and temperature. By cooling milk fat at a very low cooling rate, 0.1C/min, its phase behavior close to thermodynamic equilibrium was investigated. In these conditions, it was found that the milk fat TG crystallize in two main different lateral packing varieties, and , likely with small traces of , leading to four polymorphic forms displaying 2L and 3L stackings. Although only few signicant evolutions are observed in the lateral packing at wide angles versus T, only the crystallization monitoring at small angles by using X-ray synchrotron revealed the existence of these four species and allowed their precise characterization. The absence of polymorphic evolution observed on a following heating at 2C/min conrms that the cooling rate is slow enough to get a stable species and that the system is close to equilibrium. Comparison with the behavior of cream shows that it is not the case when fat is dispersed within globules. Indeed, the crystallization of unstable forms appears favored in dispersed fat, whereas, in fact, it is the crystallization of stable varieties that does not happen because of a lack of nucleation centers. Coupling of small- and wide-angle time-resolved synchrotron X-ray diffraction with DSC, which allows the study of TG behavior as a function of temperature, is necessary for the clarication of the complex polymorphism of milk fat. In a previous study (Lopez et al., 2000), fast quenching (>1000C/min) provided the most unstable species

of AMF, the time- and temperature-dependent release process of which has been studied by the same combination of techniques. In the next article, we will study the structural and thermal behavior of milk fat on cooling at intermediate cooling rates (Lopez et al., in preparation). ACKNOWLEDGMENTS The authors thank Arilait Recherches (Paris, France) and ANRT (Paris, France) for supporting this research. They also thank M.-H. Quemener (Lata, France) and M. Rabache (Lactalis, France), for supplying cream samples from which anhydrous milk fat was extracted, as well as all the members of the Steering Committee for Fat Programs of Arilait Recherches. REFERENCES
Arishima, T., N. Sagi, H. Mori, and K. Sato. 1991. Polymorphism of POS. I. Occurrence and polymorphic transformation. J. Am. Oil Chem. Soc. 68:710715. Blanton, T. N., C. L. Barnes, and M. Lelental. 2000. Preparation of silver behenate coatings to provide low- to mid-angle diffraction calibration. J. Appl. Cryst. 33:172173. Deffense, E. 1993. Milk fat fractionation today: A review. J. Am. Oil Chem. Soc. 70:11931201. deMan, J. M. 1961. Physical properties of milk fat. II. Some factors inuencing crystallization. J. Dairy Sci. 28:117122. Garti, N., and K. Sato. 1988. Crystallization and polymorphism of fats and fatty acids. Surfactant Science series. Vol. 31. Marcel Dekker, New York, NY. Grabielle-Madelmont, C., and R. Perron. 1983. Calorimetric studies on phospholipid-water systems. I. DL-Dipalmitoylphosphatidylcholine (DPPC)-water system. J. Colloid Interface Sci. 95:483 493. Guinier, A. 1964. Theorie et pratique de la radiocristallographie. 3rd ed. Dunod, Paris, France. Jensen, R. G., and D. S. Newburg. 1995. Bovine milk lipids. Page 542 in Handbook of Milk Composition. R. G. Jensen, ed. Academic Press, New York, NY. Kaylegian, K. E., and R. C. Lindsay. 1994. Handbook of Milk Fat Fractionation Technology and Application. AOCS Press, Champaign, IL. Journal of Dairy Science Vol. 84, No. 11, 2001

2412

LOPEZ ET AL. Lopez, C., P. Lesieur, G. Keller, and M. Ollivon. 2000. Thermal and structural behavior of milk fat. 1. Unstable species of cream. J. Colloid Interface Sci. 229:6271. Lopez, C., P. Lesieur, C. Bourgaux, G. Keller, and M. Ollivon. 2001b. Thermal and structural behavior of milk fat 2. Crystalline forms obtained by slow cooling of cream. J. Colloid Interface Sci. 240:150161. Marangoni, A. G., and R. W. Lencki. 1998. Ternary phase behavior of milk fat fractions. J. Agric. Food Chem. 46:38793884. Ollivon, M., and R. Perron. 1985. Fat Science, Part A. J. Hollo, ed. Elsevier, New York, NY. Ollivon, M., and R. Perron. 1992. Manuel des Corps Gras. A. Karleskind and J. P. Wolff, eds. Lavoisier, Paris, France. Sato, K., T. Arishima, Z. H. Wang, K. Ojima, N. Sagi, and H. Mori. 1989. Polymorphism of POP and POS. I. Occurrence and polymorphic transformation. J. Am. Oil Chem. Soc. 66:664674. Small, D. M. 1986. Handbook of Lipid Research. The Physical Chemistry of Lipids: From Alkanes to Phospholipids. Plenum Press, New York, NY. ten Grotenhuis, E., G. A. van Aken, K. F. van Malssen, and H. Schenk. 2000. Polymorphism of milk fat studied by differential scanning calorimetry and real-time X-ray powder diffraction. J. Am. Oil Chem. Soc. 76:10311039. Timms, R. E. 1979. The physical properties of blends of milk fat with beef tallow and beef tallow fractions. Aust. J. Dairy Technol. 34:6065. Timms, R. E. 1980a. The phase behavior and polymorphism of milk fat, milk fat fractions, and fully hardened milk fat. Aust. J. Dairy Technol. 35:4753. Timms, R. E. 1980b. The phase behavior of mixtures of cocoa butter and milk fat. Lebensm.-Wiss. Technol. 13:6165. Tverdokhleb, G. V., and V. M. Vergelesov. 1974. Brief communication. Page 211 in XIX Int. Dairy Congr. 1E. Woodrow, I. L., and J. M. deMan. 1968. Polymorphism in milk fat shown by X-ray diffraction and infrared spectroscopy. J. Dairy Sci. 51:9961000.

Kellens, M., W. Meeussen, C. Rikel, and H. Reynaers. 1990. Time resolved X-ray diffraction studies of the polymorphic behaviour of tripalmitin using synchrotron radiation. Chem. Phys. Lipids 52:7998. Keller, G., F. Lavigne, L. Forte, K. Andrieux, M. Dahim, C. Loisel, M. Ollivon, C. Bourgaux, and P. Lesieur. 1998. DSC and X-ray diffraction coupling. Specications and applications. J. Therm. Anal. 51:783791. Keller, G., F. Lavigne, C. Loisel, M. Ollivon, and C. Bourgaux. 1996. Investigation of the complex thermal behavior of fats: Combined DSC and X-ray diffraction techniques. J. Therm. Anal. 47:15451565. Larsson, K. 1966. Alternation of melting points in homologous series of long-chain compounds. J. Am. Oil Chem. Soc. 43:559562. Larsson, K. 1972. Molecular arrangement in glycerides. Fette Seifen Anstrichm. 74:136142. Larsson, K. 1994. Page 7 in LipidsMolecular Organization, Physical Functions and Technical Applications. The Oily Press, Bridgewater, UK. Lavigne, F. 1995. Polymorphisme et transitions de phases des triglyc erides. Applications aux proprietes thermiques et structurales de ` ` la matiere grasse laitiere anhydre et de ses fractions. PhD. thesis. Univ. Paris. VII, Paris XI et E.N.S.I.A., Paris, France. Lavigne, F., and M. Ollivon. 1993. Differential scanning calorimetry and dynamic X-ray diffraction studies of polymorphic transitions of 2-oleo-dipalmitin. J. Mediteran. Calor. Anal. Therm. (Corte Sept. 93, AFCAT) 24:237241. Loisel, C., G. Keller, G. Lecq, C. Bourgaux, and M. Ollivon. 1998. Phase transition and polymorphism of cocoa butter. J. Am. Oil Chem. Soc. 75:425439. Lopez, C., F. Lavigne, P. Lesieur, C. Bourgaux, and M. Ollivon. 2001a. Thermal and structural behavior of milk fat. 1. Unstable species of anhydrous milk fat. J. Dairy Sci. 84:756766.

Journal of Dairy Science Vol. 84, No. 11, 2001