Effect of Fractional Crystallyzation on Composition and Thermal Behavior of Coconut Oil

Abstract

This study was aimed to fractionate coconut oil into its high- and low-melting fractions and determine the compositional and thermal property changes. A sample of coconut oil was dissolved in acetone, and allowed to crystallize isothermally at a desired temperature to separate into the low- and high-melting components. The isolated fractions were compared to the original sample with respect to fatty acid and triacylglycerol compositions as well as thermal behavior. There were considerable deviations in the fatty acid and triacylglycerol compositions of the two components with respect to those of the original sample. As a consequence, the overall melting behaviors of the two components differed considerably from that of the original sample. In the high-melting fraction, the onset and endset were shifted toward the higher temperature region with a concurrent reduction in its melting range, and vice-versa, the onset and endset of the low-melting fraction had shifted toward the low-temperature region with an increase in its melting range. The reduction of the melting range of the high-melting fraction could make it a specialty fat for applications in confectionery.

Keywords:

• Coconut oil
• Confectionery fat
• DSC
• Fractional crystallization
• Thermal analysis

INTRODUCTION

Natural fats and oils are complex mixtures of triacylglycerols (TAG) with different fatty acid composition. They may differ in their physico-chemical properties depending on the type of fatty acid attached to their TAG molecules. Generally, TAG molecules esterified with more saturated fatty acids display higher melting temperature than those that are esterified with mono and diunsaturated fatty acids.[1] Hence, a naturally occurring fat with a mixture of these two extreme groups of TAG molecules may have a tendency to undergo phase separation to yield solid and liquid components under certain specific conditions.[2] Palm oil fractionation into two distinctly different fat derivatives could have become possible because of this reason. As the major TAG molecules of coconut oil are closely similar in their melting or crystallization characteristics, isolation of two fat derivatives with totally different TAG molecular groups may be difficult to achieve. Despite this, some effort have been put on the solvent fractional crystallization of coconut oil using acetone to recover components enriched with either long-chain saturated fatty acids or medium-chain fatty acids.[3, 4] These efforts have already shown that compositional changes could be possible in coconut oil because of the crystallization, but its impact on the physical properties of the oil fractions has not been highlighted. Information on the physical property changes in the fat derivatives isolated from fractionation would be beneficial to determine their potential uses in product formulations. Hence, the objective of this study was to investigate the thermal behavior changes of coconut oil fractions in relation to fatty acid and TAG compositions using an advanced thermal analysis system, such as differential scanning calorimetry (DSC).

MATERIALS AND METHODS

Materials

Pure samples of coconut oil were received from Adamjee Lukmanjee & Sons Ltd., Colombo, Sri Lanka. All chemicals used in this study were of analytical grade, unless otherwise specified.

Fractionation Processes

A sample of coconut oil was melted at 60°C and mixed with acetone in 1:2 (w/v) ratio. The solution was boiled at 60°C until it became uniformly dissolved allowing it to crystallize isothermally at 10°C for 6 h according to the procedure reported by Siahaan and co-workers.[4] The precipitated fat was filtered off to give a high melting fat fraction. After removing the precipitate, the mother-liquor was evaporated under reduced pressure to yield a liquid called low-melting fraction.

Determination of Iodine Value (IV) of Oil Samples

IV of the oil samples was determined according to the American Oil Chemists’ Society (AOCS) method Cd Id–92.[5]

Determination of Fatty Acid Composition

Fatty acid methyl esters were prepared by reacting oil with BF₃/methanol mixture at 100°C. Upon completion of the reaction, esters were extracted into hexane and a portion from the top hexane layer was injected on an Agilent 6890N gas chromatograph (Agilent Technologies, Singapore) equipped with a polar capillary column RTX-5 (0.32 mm internal diameter, 30 m length, and 0.25 μm film thickness; Restex Corp., Bellefonte, PA, USA) and a flame ionization detector. Split injection was conducted with a split ratio of 58:1, nitrogen was used as a carrier gas at a flow-rate of 1.00 mL/min. The temperature of the column was 50°C (for 1 min), and was programmed to increase to 200°C at 8°C/min. The temperatures of the injector and detector were maintained at 200°C. Each sample was chromatographed two times, and the data were reported as percentage areas.[6]

TAG Analysis by HPLC

The system used was a Waters Model 510 liquid chromatograph equipped with a differential refractometer Model 410 as the detector (Waters Associates, Milford, MA, USA). The analysis of TAG was performed on a Merck Lichrosphere RP-18 column packed with a particle size of 5 μm (12.5 cm × 4 mm i.d.; Merck, Darmstadt, Germany). The mobile phase was a mixture of acetone-acetonitrile (63.5:36.5) and the flow rate was 1 ml/min at 30°C. The injector volume was 10 μl of 5% (w/w) oil in chloroform. Each sample was chromatographed two times, and the data were reported as percentage areas.[6] The TAG peak identification of the samples was based on the retention times of some TAG standards available and the earlier report of Tan and Che Man.[7]

Thermal Analysis by DSC

For thermal analysis, a Mettler-Toledo DSC, Model 823 (Mettler-Toledo GmbH, Zurich, Switzerland), equipped with the STAR^e thermal analysis system was used. Nitrogen (99.9% purity) was used as the purge gas at a rate of ˜20 mL/min. An oil sample weighing approximately 6–8 mg was placed in a standard DSC aluminum pan and then hermetically sealed. An empty and hermetically sealed DSC aluminum pan was used as a reference. In order to obtain melting profiles, the oil samples were subjected to the following temperature program: –70°C isotherm for 1 min, heated at 5°C/min to 70°C.[6]

Statistical Analysis

All experiments were carried out using three replicates and the results were expressed as mean values ± standard deviation. Data were statistically analyzed by one-way analysis of variance (ANOVA) using the MINITAB (version 14; Minitab Inc., PA, USA) statistical package at 0.05 probability level.

RESULTS AND DISCUSSION

Fatty Acid (FA) Composition

The FA compositions of coconut oil and its fractions are compared as shown in Table 1. Coconut oil consists of 91.90% saturated fatty acids (SFA) and 8.12% unsaturated fatty acid (USFA). The major fatty acids of the unfractionated sample are lauric (46.03%) and myristic (19.03%) acids, while caproic (C6:0), caprylic (C8:0), capric (C10:0), palmitic (C16:0), stearic (C18:0), oleic (C18:1), and linoleic (C18:2) occur as minor fatty acids. This is in accordance with the previous findings reported in the literature.[7, 8] Among these fatty acids, the proportion of the medium chain fatty acids (MCFA), such as C6:0, C8:0, C10:0, and C12:0, is about 61.57%. The relative proportion of MCFA in coconut oil is often found to exceed those of the same found in other vegetable oils. Coconut oil having greater content of shorter-chain fatty acids such as caprylic (C8:0) and capric (C10:0) acids as compared to palm oil and pam kernel oil is another interesting property.[9]

Table 1 Fatty acid composition (peak area %) and iodine value of coconut oil and its fractions. ¹

	Sample
Fatty acid	Coconut oil	High-melting fraction	Low-melting fraction
Caproic (C6:0)	0.73 ± 0.00^a	0.25 ± 0.01^b	0.79 ± 0.01^c
Caprylic (C8:0)	8.81 ± 0.01^a	3.38 ± 0.03^b	9.41 ± 0.04^c
Capric (C10:0)	6.00 ± 0.00^a	3.51 ± 0.05^b	6.27 ± 0.03^c
Lauric (C12:0)	46.03 ± 0.01^a	43.10 ± 0.03^b	46.66 ± 0.05^c
Myristic (C14:0)	19.03 ± 0.00^a	28.91 ± 0.01^b	17.98 ± 0.02^c
Palmitic (C16:0)	8.60 ± 0.01^a	13.74 ± 0.01^b	7.95 ± 0.04^c
Stearic (C18:0)	2.70 ± 0.00^a	3.86 ± 0.00^b	2.56 ± 0.01^c
Oleic (C18:1)	6.42 ± 0.00^a	2.62 ± 0.02^b	6.65 ± 0.05^c
Linoleic (C18:2)	1.70 ± 0.01^a	0.65 ± 0.01^b	1.76 ± 0.01^c
∑MCFA	61.57 ± 0.00^a	50.23 ± 0.06^b	63.12 ± 0.13^c
∑SFA	91.90 ± 0.01^a	96.74 ± 0.04^b	91.60 ± 0.06^c
∑USFA	8.12 ± 0.01^a	3.27 ± 0.04^b	8.41 ± 0.06^c
IV (Wij's)	11.91 ± 0.59^a	7.76 ± 0.28^b	13.73 ± 0.77^a
^1Results in the above table represent the means and standard deviation of three replicates. Mean values within each row followed by different superscripts (a, b, c) are significantly different (P < 0.05).

The high- and low-melting fractions isolated through fractionation were found to have fatty acid profiles, which are different from that of the original sample. In the high melting fraction, there is an overall increase in the proportion of SFA (p < 0.05) with a concurrent reduction in the proportion of the USFA. Fatty acids, namely, C14:0, C16:0, and C18:0, experienced increments (p < 0.05), while fatty acids C6:0, C8:0, C10:0, C12:0, C18:1, and C18:2 experienced decreases. Consequently, the proportion of MCFA in the high-melting fraction would have dropped by about 10%. Unlike the high-melting fraction, the changes in the distribution of individual fatty acids of the low-melting fraction are only marginal with respect to those of the original sample. This could be clearly evident from a comparison of SFA (%) and USFA (%) between the original sample and the low-melting fraction. There were slight increases in the proportions of fatty acids, namely, C6:0, C8:0, C10:0, C12:0, C18:1, and C18:2 with the accompanying reductions in the proportions of C14:0, C16:0, and C18:0.

TAG Compositional Changes

Comparative TAG profiles of coconut oil and its fractions are presented in Fig. 1 The major TAG molecular species of coconut oil (Chromatogram A) are CCLa, CLaLa, LaLaLa, and LaLaM (where C for capric, La for lauric, M for myristic) and most of the remaining peaks of the chromatogram represent TAG molecular species, which occur in less than the 10% level. These are also the TAG molecules popularly known as medium chain TAG (MCTAG), mainly attributed to the nutritional significance and functional properties of coconut oil.[10] On the basis of degree of unsaturation, the TAG molecular species of coconut oil could be sub-categorized into trisaturated (SSS, 90.94%), monounsaturated (SSU, 7.71%), diunsaturated (SUU, 0.88%), and triunsaturated (UUU, 0.48%) TAG (Table 2). As evidenced from Fig. 1, the TAG profiles of the high- and low-melting fractions obtained after fractionation has seen considerable deviations from that of the native sample. In the high melting fraction (Chromatogram B), TAG molecular species, such as CCLa, CLaLa, LaLaLa, and LaLaO, have undergone decreases while TAG species, such as LaLaM, LaMM, LaMP, and LaPP+MMO, experienced increments (where O is for oleic, P is for palmitic). Clearly, TAG molecular species esterified with short and medium chain fatty acids are on the decline, while those esterified with long chain fatty acids are on the increase. As a consequence, there has been a significant (p < 0.05) increase in the percentage of SSS with concurrent drop (p < 0.05) in the proportions of SSU and SUU (Table 2). The relative increases in the proportions of SSS TAG molecules in the high-melting fraction would have lead to the increase of long chain saturated fatty acids, such as myristic, palmitic, and stearic, as shown in Table 1. In accordance with the change in the degree of unsaturation, there has been a decline in iodine value of high melting fraction (Table 1).

Table 2 Triacyl glycerol (TAG) composition of coconut oil and its fractions. ¹

	Sample
TAG	Coconut oil	High-melting fraction	Low-melting fraction
CCLa	15.08 ± 0.16^a	5.02 ± 0.04^b	15.36 ± 0.01^a
CLaLa	19.45 ± 0.15^a	7.60 ± 0.03^b	19.54 ± 0.01^a
LaLaLa	22.59 ± 0.11^a	14.37 ± 0.03^b	22.69 ± 0.01^a
LaLaM	16.48 ± 0.04^a	22.33 ± 0.01^b	15.98 ± 0.02^c
LaLaO	2.84 ± 0.00^a	1.15 ± 0.02^b	3.74 ± 0.02^c
LaMM	9.74 ± 0.01^a	21.86 ± 0.05^b	8.60 ± 0.02^c
LaLaP	0.28 ± 0.01^a	0.18 ± 0.01^b	0.67 ± 0.01^c
LaMO	2.27 ± 0.00^a	1.09 ± 0.01^b	2.87 ± 0.01^c
LaMP	5.08 ± 0.01^a	15.00 ± 0.05^b	4.21 ± 0.00^c
LaOO	0.54 ± 0.00^a	0.09 ± 0.00^b	0.73 ± 0.00^c
LaPO	1.42 ± 0.01^a	0.61 ± 0.01^b	1.67 ± 0.00^c
LaPP+MMO	1.63 ± 0.00^a	6.96 ± 0.04^b	1.38 ± 0.02^c
MMP	0.63 ± 0.13^a	0.15 ± 0.01^b	0.51 ± 0.01^a
MPO+POL	0.78 ± 0.26^a	0.36 ± 0.00^b	0.88 ± 0.01^a
OOO	0.48 ± 0.01^a	2.26 ± 0.01^b	0.29 ± 0.00^c
POO	0.34 ± 0.01^a	0.08 ± 0.00^b	0.41 ± 0.00^c
PPO	0.39 ± 0.02^a	0.26 ± 0.00^b	0.46 ± 0.00^c
PPP	ND	0.62 ± 0.01^a	ND
∑SSS	90.94 ± 0.34^a	94.10 ± 0.04^b	88.94 ± 0.00^c
∑SSU	7.71 ± 0.30^a	3.47 ± 0.04^b	9.63 ± 0.01^c
∑SUU	0.88 ± 0.01^a	0.17 ± 0.00^b	1.14 ± 0.00^c
∑UUU	0.48 ± 0.01^a	2.27 ± 0.01^b	0.29 ± 0.00^c
^1 Results in the above table represent the means and standard deviation of triplicate analysis. Mean values within each row followed by different superscripts (a, b, c) are significantly different (P < 0.05).
Abbreviations: SSS: trisaturated; SSU: disaturated; SUU: diunsaturated; UUU: triunsaturated; C: capric; La: lauric; M: myristic; P: palmitic; S: stearic; O: oleic; L: linoleic.

As shown in Fig. 1, the TAG profile of the low-melting fraction is represented by chromatogram c. Since it looked apparently similar to the original sample (chromatogram a), the changes in the proportional distribution of TAG molecules have to be examined as per the data presented in Table 2. With respect to the original sample, increments in the major TAG molecular species, such as CCLa, CLaLa, and LaLaLa, could be observed though they are not significant (p > 0.05). However, a significant (p < 0.05) increase in the proportion of LaLaO with significant (p < 0.05) decreases in the proportions of LaLaM, LaMM, and LaMP, which could be easily noticeable. In contrast to the expectation, the changes in the proportions of oleic containing TAG molecular species, such as OOO, and POO are not remarkable. According to the data presented in Table 2, there was an overall increase in the proportion of SSU with a concurrent decrease in the proportion of SSS. In accordance with these changes, there have been some increases in the proportions of C8:0, C10:0, and C12:0 fatty acids, as well as the total sum of unsaturated fatty acids (Table 1). The increase in the degree of unsaturation would be the cause for the increase of iodine value in the low-melting fraction (Table 1).

Thermal Profile Changes

The melting curves of coconut oil and its fraction are compared as shown in Fig. 2 Under the DSC specified conditions, coconut oil displayed a major melting transition at 12.5°C with a small shoulder peak at 23.4°C (Curve-B). The onset of melting starts at 9.05°C, which continues to the completion point at 25.4°C, which is known as endset. For the unfractionated sample, the endset temperature is almost comparable to the slip melting point (SMP) as the reported value of SMP for most of pure coconut oil is found to be within the range of 24.5–25.5°C.[11] As shown in Fig. 2, the melting behaviors of the high and low melting fractions obtained are represented by curves A and C, respectively. Although the profiles of the two curves looked apparently similar to that of the original sample, there were significant differences with regard to temperatures of onset, endset, and peak maxima (Table 3). While the onset, endset, and peak maxima of the high-melting fraction were found to have shifted to the higher temperature region, those of the low-melting fraction were found to shift to the lower temperature region. This changing trend in DSC parameters of the fractions could be attributed to their changing TAG composition as shown in Table 2. It is because of the fact that the thermodynamic parameters of oils and fats are closely related to their TAG composition.[12] In the case of high-melting fraction, the peak-maxima of the major thermal transition originally appeared at 23.42°C, which was now displaced by at least 7°C further up. Likewise, the onset and endset temperatures of this fraction were also found to have shifted almost in similar magnitudes. As already pointed out, in the high-melting fraction, the saturated fatty acid content has gone up by 4.84% because of the increase of SSS TAG molecular species (LaLaM, LaMM, LaMP, and LaPP) by 3.16%. These could be attributed to the observed changes in the DSC parameters of the high-melting fraction. Naturally, a change in the degree of unsaturation of lipids by any means would result in a change of melting temperatures.[1] The same explanation would be equally attributable to the changes noticed in the DSC parameters of the low-melting fraction (Table 3). According to the data presented in Table 2, there was an increase in the degree of unsaturation of TAG molecules in the low-melting fraction, which might lead to the decrease in the melting temperatures. As a consequence, the DSC parameters of the low-melting fraction were found to shift towards the lower temperature region.

Table 3 DSC thermal characteristics of coconut oil and its fractions. ¹

	Thermal transition parameters (°C)
Sample	Onset	Peak maxima	Endset	Endset–Onset
High-melting fraction (Curve-A)	18.58 ± 0.71^a	31.20 ± 0.11^a	33.00 ± 0.01^a	14.42
Coconut oil (Curve-B)	9.05 ± 0.11^b	23.42 ± 0.18^b	25.37 ± 0.03^b	16.32
Low-melting fraction (Curve-C)	6.12 ± 0.28^c	21.78 ± 0.47^c	23.96 ± 0.04^c	17.84
^1Results in the above table represent the means and standard deviation of triplicate analysis. Mean values within each column followed by different superscripts (a, b, c) are significantly different (P < 0.05).

The rapid-melt-down behavior of coconut oil is a well appreciated property for its inclusion in confectionery products.[9] As this occurs below the physiological temperature of the human body, it gives a definite advantage for coconut oil with regard to its significance in human nutrition.[10] According to the data presented in Table 3, the gap between onset and endset temperatures of the original sample is 16 degrees, which is, in fact, a narrow range when compared to the DSC melting profiles of most of the animal fats. According to past studies, the DSC melting ranges displayed by lard, beef, and goat fat were wider.[13] This difference could be directly attributed to the nature of TAG molecules present in coconut oil as compared to those of the animal fats. For instance, in coconut oil, CCLa, CLaLa, LaLaLa, and LaLaM are the most dominant TAG molecular species, which are mainly esterified with lauric and capric acids. Because of the similarity in fatty acid substitutions, the melting temperature differences among these TAG molecular species would be expected to be small.[14] According to the data presented in Table 3, the DSC parameters of the high- and low-melting components recovered from the fractionation were found to change considerably. With respect to the original sample, the melting range of high-melting fraction tended to reduce, while the melting range of the low-melting fraction has slightly increased. Thus, the high-melting component recovered from the fractionation would display a melting range even sharper than the original sample. According to past studies, fat components having sharp-melting ranges are useful as specialty fats, which could leave a clean, cool, non-greasy sensation on the palate.[9] They would be greatly beneficial for formulation of toffees, chocolate-type coatings, biscuit sprays, non dairy ice cream, as well as filling creamers.[3]

CONCLUSION

Coconut oil could be fractionated into low-and high-melting fractions using acetone as solvent under certain specific conditions. The high-melting component was found to contain a significantly higher proportion of SSS TAG molecules while the low-melting component had a lower proportion of trisaturated TAG molecules. The proportion of MCTAG in high-melting components tended to decline, while the proportion of the same in low-melting components was slightly increased. Although the DSC melting curves of the two fractions were similar in shape, they displayed changes in their melting ranges due to changes in DSC parameters, such as onset, endset, and peak maxima.

ACKNOWLEDGMENT

The principal author gratefully acknowledges a research grant (02-01-10-0889 RU) received under the Research University Grants Scheme of the Universiti Putra Malaysia.