Analytical Methods
Multivariate statistical analysis treatment of DSC thermal properties for animal fat adulteration
Omar Dahimi
a, Alina Abdul Rahim
a,⇑, S.M. Abdulkarim
b,⇑, Mohd Sukri Hassan
a, Shazamawati B.T. Zam Hashari
a, A. Siti Mashitoh
a, Sami Saadi
b,⇑aInstitute of Halal Research and Management (IHRAM), Universiti Sains Islam Malaysia, 71800 Nilai, Negeri Sembilan, Malaysia
bDepartment of Food Science, Universiti Putra Malaysia, 43400 UPM, Serdang, Malaysia
a r t i c l e i n f o
Article history:
Received 24 May 2013
Received in revised form 2 February 2014 Accepted 18 February 2014
Available online 28 February 2014
Keywords:
Adulteration Lard Beef tallow Chicken fat DSC PCA
a b s t r a c t
The adulteration of edible fats is a kind of fraud that impairs the physical and chemical features of the original lipid materials. It has been detected in various food, pharmaceutical and cosmeceutical products.
Differential scanning calorimetry (DSC) is the robust thermo-analytical machine that permits to finger- print the primary crystallisation of triacylglycerols (TAGs) molecules and their transition behaviours.
The aims of this study was to assess the cross-contamination caused by lard concentration of 0.5–5%
in the mixture systems containing beef tallow (BT) and chicken fat (CF) separately. TAGs species of pure and adulterated lipids in relation to their crystallisation and melting parameters were studied using prin- cipal components analysis (PCA). The results showed that by using the heating profiles the discrimination of LD from BT and CF was very clear even at low dose of less than 1%. Same observation was depicted from the crystallisation profiles of BT adulterated by LD doses ranging from 0.1% to 1% and from 2% to 5%, respectively. Furthermore, CF adulterated with LD did not exhibit clear changes on its crystallisation pro- files. Consequently, DSC coupled with PCA is one of the techniques that might use to monitor and differ- entiate the minimum adulteration levels caused by LD in different animal fats.
Ó2014 Published by Elsevier Ltd.
1. Introduction
Nowadays food quality control has attracted much attention by private and governmental companies in Malaysia. The prevention of the adulterant materials in particular, fats and oils are one of the topics of high concern, with potential of adulteration both on deliberate or accidental cross-contamination. It necessitates the use of robust analytical machines capable to differentiate these small adulterant materials, especially when they exist at low dose in the food matrix. This issue prompted the scientist in Malaysia to probe the real causes leading to this type of fraud both in food and cosmeceutical products. Differential scanning calorimetry (DSC) is a thermal analysis technique used to study the thermal behaviours of formulated pharmaceutical and food systems. It has gained much application in edible oils and fats crystallisation, that directly influenced by their chemical structure and composition of tri, di-,
and mono-acylglycerols, fatty acid chain lengths, single and double hydrogen (H2) bonds, symmetrical and asymmetrical triacylglyce- rols (TAGs). For instance, most TAGs have been found to exist in three forms including alpha (
a
), beta prime (b0) and beta (b) form, which, in this order, display increasing thermodynamic stability, melting point, heat of fusion and melting dilatation.DSC has some advantages over other classical detection meth- ods, as it is rapid and does not require excess sample preparation or solvent utilization and therefore, it is an environmental friendly technique. The applications of DSC in the analysis and character- ization of oils and fats for the determination of solid fat content, crystallisation and melting profiles, enthalpy of transitions and polymorphic forms, is well known and reviewed (Marangoni &
Narine, 2002; Narine & Marangoni, 2002; Sato, 1996; Sato & Ueno, 2001; Sato, Ueno, & Yano, 1999; Tang & Marangoni, 2007). The detection of adulteration of edible oils, fats, and fat-based products necessitate the use of DSC that can help to fingerprint their thermal behaviours including melting and crystallisation (Tan & Che Man, 2002). Moreover, several works had extensively studied DSC appli- cation on olive oils, also in the sector of adulteration (Chiavaro, Vittadini, Rodrigeuez-Estrada, Cerretani, & Bendini, 2008; Chiavaro et al., 2009). DSC was able to monitor the presence of pig and
http://dx.doi.org/10.1016/j.foodchem.2014.02.087 0308-8146/Ó2014 Published by Elsevier Ltd.
⇑Corresponding authors. Tel.: +60 6 798 8236/6 798 8237; fax: +60 6 799 8699 (A.A. Rahim). Tel.: +60 3 8946 8537; fax: +60 3 8942 3552 (S.M. Abdulkarim). Tel.:
+60 142331473 (S. Saadi).
E-mail addresses: [email protected] (A.A. Rahim), [email protected] (S.M. Abdulkarim),[email protected],[email protected](S. Saadi).
Contents lists available atScienceDirect
Food Chemistry
j o u r n a l h o m e p a g e : w w w . e l s e v i e r . c o m / l o c a t e / f o o d c h e m
buffalo body fat in cow and buffalo ghee (Lambelet & Ganguli, 1983), butter in tallow and margarine (Aktas & Kaya, 2001), li- pase-catalyzed interesterified lard in refined-bleached-deodorised palm oil (Marikkar, Lai, Ghazali, & Che Man, 2002), animal fat in butter (Coni, Di Pasquale, Coppolelli, & Bocca, 1994), and canolla oil (Marikkar, Ghazali, Che Man, & Lai, 2002). Principal components analysis (PCA) has been successfully applied to indicate the simi- larities and differences that exist in between different food compo- nents and their chemical compositions such as screening of adulterated honey (Subari, Saleh, Shakaff, & Zakaria, 2012); milk adulteration (Ntakatsane, Liu, & Zhou, 2013; Santos, Pereira-Filho,
& Rodriguez-Saona, 2013), adulteration of sesame and peanut oils (Zhao et al., 2013); olive oil adulteration (Dourtoglou et al., 2003) and discrimination of commercial cheeses using fatty acid and phytosterol contents as markers (Kim et al., 2014). Evaluation of the relationship between thermal properties and chemical com- pounds resulting from hydrolysis and lipid auto-oxidation of fatty acids and triacylglycerols during the storage time of virgine olive oil (Chiavaro et al., 2013). Correlation between thermal properties of olive oil during refining process steps and their molecular spe- cies formed during lipids oxidation, polymerization, and hydrolysis by means of principal compounds analysis (Caponio et al., 2013).
DSC coupled with PCA was used also in the measurement of ther- mal stress and selection of appropriate heating conditions applied on extra-virgine olive oil in food industry (Maggio, Cerretani, Barnaba, & Chiavaro, 2012).
In view of these, the main objective of this study was to differ- entiate lard from chicken and beef fats and their mixtures using principal components analysis (PCA). TAGs species of pure and adulterated lipids in relation to their crystallisation and melting parameters were determined.
2. Materials and methods 2.1. Sample and reagents
The analytical solvents used for the extraction of fats and the solvents used during HPLC analysis were methanol 99.9%; acetone 99.9%;n-hexane 99.0%, chloroform 99.8% and sodium methoxide solution 1%. All of these solvents and TAGs standards were pur- chased from Sigma- Aldrich, USA. Adipose tissues of pig (lard), beef tallow, and chicken fats were obtained from local slaughtered houses in Selangor, Malaysia.
2.2. Samples preparation
The extraction of fats from the samples was carried out based on the reported protocol ofBligh–Dyer method (1959)with slight modifications. An amount of 350 mg of sample was weighed and dissolved by adding a mixture of methanol (4 ml), chloroform (2 ml) and water (0.4 ml) and then homogenized and vortexed for 30 s. After that, the sample was washed again with 2 ml of chloroform and 2 ml of water and the resulting extract were added; vortexed for 30 s and centrifuged at 5000gfor 10 min, at room temperature. Using a Pasteur pipette, the upper layer (methanol/water) was then removed and the lower layer (methanol/chloroform) was transferred into a clear tube. Finally, the organic solvents were removed using a rotary evaporator under reduced pressure, at 60°C. The extracted animal fat including, lard (LD), beef fat (BF), chicken fat (CF) and a series of 14 samples containing 0.5–10% w/w of lard were adulterated with BF, and CF separately and then were analysed. Samples containing lard were assigned as adulterated; while pure beef fat, pure chicken fat and pure lard were designated as beef tallow (BT), chicken fat (CF) and lard (LD).
2.3. Analysis of triacylglycerols (TAGs) composition by HPLC
The determination of TAGs composition of pure samples and their blends was carried out according to the procedures described by Saadi et al. (2011). TAGs composition was obtained by dissolving 0.1 g of the sample in 1 ml of HPLC grade acetone and 20
l
l aliquots was injected into the high performance liquid chromatography (Shimadzu HPLC, Japan) equipped with a Borwin software coupled by a refractive index detector model (Shimadzu RI-1530 RI detector, Japan). A Purospher Star RP-C18 column (2504.4 mm with a particle size of 5l
m, Merck, Darmstadt, Germany) was fitted into HPLC and the TAGs species were eluted from the column using a mixture of acetone/acetonitrile (70:30 v/v) at a flow rate of 1 ml/min. The identification of TAGs species was done with referring to their TAGs standards during a running time of 25 min.2.4. Thermal analysis of fat samples by DSC
A Mettler Toledo DSC was used for analyzing the thermal characteristics of the fat samples. Nitrogen (99.99% purity) was the purge gas used and flowed at 20 ml/min. The instrument was calibrated with indium (DHf= 28.45 J/g) and dodecane (DHf= 216.73 J/g) (Tan & Che Man, 2002). Sample of 8–10 mg was weighed into aluminium pans covered and fixed on the sam- ple platform. Prior to sample analysis, the baseline was obtained with an empty hermetically sealed aluminium pan. All samples were subjected to the following temperature programs, where the samples were cooled from 50 to 70°C, held for 5 min and heated from70 to 50°C at a rate of 5°C/min. The melting and crystallisation parameters concerning each sample were obtained using Mettler Toledo DSC STAReSW 9.20 software. All samples were carried out in triplicate and average of three measurements was used for data analysis.
2.5. Statistical analysis
In the principal component analysis (PCA), the starting point was the construction of a data matrix X that consists ofkvariable andnobjects (Brereton, 1990). The DSC results concerning cooling and heating profile parameters including Ton, DH and Tend were considered as variables. Pure CF, BT, and LD and their blends were the objects. The data matrix X, its standardised version and corre- lation matrix R, were calculated using the Microsoft Excel software version 2007. PCA was performed by using an Unscrambler soft- ware (X10) from Camo, USA. After further calculation two new matrices including principal component scores (P) and principal component loadings (W) were obtained. The matrix P reflects main relations among objects and makes a possible a classification of adulterated lipid materials, whereas matrix W illustrates main relations among variables and enables their selection. Principal components (PC) were determined by considering eigenvalues and associated eigenvectors (Wesołowski & Erecin´ska, 1998).
3. Results and discussion
3.1. Changes in TAGs species of pure and contaminated-mixed edible fats
TAGs species of the samples were detected and identified based on the retention time of their compatible TAGs standards.Table 1 shows the composition of TAGs of the pure samples (BT, LD, and CF) and their mixtures (BT + LD and CF + LD). The results showed that almost all TAGs types were exist in the samples except for two types of TAGs that were present only in BT [e.g.1,2-dipalmi- toyl-3-stearoyl-sn-glycerol (PPS) and 1,2-distearoyl-3-oleoyl-sn-
glycerol (SOS)]; where their relative amounts were found to be 8.84% and 0.98%, respectively. In addition, three prominent TAGs species were detected in all pure samples for example PPO, PPP and PLO in BT and they represent 30.81%, 14.97% and 14.29%, respectively from the total TAGs species detected. On the other hand, CF revealed that three principal TAGs species PPL, PPO and PPP with relative composition of 29.32%, 19.51% and 11.75%, respectively (Table 1) were found to be the dominating TAGs spe- cies for this type of lipid material. Meanwhile, the TAGs species PPL, PPO and PPP in LD were found to be similar to those in CF with relative contents of 25.57%, 23.73% and 13.14%, respectively.
Furthermore, LD had two additional TAGs species; the first TAG marker is POS and its content was in double amount than that exist in BT (10.77% versus 5.30%, respectively), whereas, the second TAG marker is OOO, and its relative content was approximately three times more than that exhibited by BT (7.81% versus 2.96%, respectively).
3.2. Crystallisation regime of pure edible lipid materials
DSC is a suitable technique to characterise such phase transi- tions as crystallisation and melting regimes of lipids that require the intake or release of thermal enthalpy. All DSC thermograms ob- tained upon cooling and heating of fat samples are reported in Fig. 1(a), which indicated typical DSC cooling curves of the pure samples BT, CF, and LD. In general, all samples showed two main well-defined exothermic events. In the LD sample, the first peak (minor) was appeared at low temperature(45°C) and the major peak with a shoulder was detected at higher temperature(2°C).
CF and BT also showed almost similar cooling profiles. In CF the major exothermic peak appeared at(41°C), and the second one was depicted at5°C. Meanwhile, in BT sample, a clear shift- ing in exothermic event was observed where the first peak was at (40°C), and the second peak was shown at(2°C). This cool- ing behaviour of the pure samples was attributed to the amount of saturated and unsaturated TAGs exist in the samples (Dahimi, Hassan, Rahim, Abdulkarim, & Mashitoh, 2014). In this case, the crystallisation profile of the sample containing high-unsaturated TAGs will start to shift to the lower temperature region due to the percentage of unsaturated TAGs of LD, CF, and BT of 89%, 85%
and 74%, respectively (SeeTable 2).
3.3. Melting regime of pure edible lipid materials
Heating thermograms profiles of the samples inFig. 1(b) helped to distinguish two major peaks in all samples. In the LD sample, the first major peak was appeared at low temperature(50°C) and the other one with shoulders was observed at higher temperature 2°C. In part, CF sample exhibited two small endothermic events where the first melting peak was at(46°C), and the end peak was at(8°C). BT also showed different melting profiles where the first major endothermic peak with a shoulder appeared at (44°C), and the second one was monitored at higher tempera- ture4°C. These changes in the melting profiles were possibly attributed to the contents of saturated TAGs including middle (SUS/SSU) and high melting (SSS) TAGs constituted in the samples, where it was observed that the percentage of saturated TAGs of BT was the highest (23%) flowed by CF (13%) and ended by LD (9%).
3.4. DSC crystallisation curves in monitoring the small doses of lard adulterants
Fig. 1(c) and (d) showed the DSC major exothermic peaks of BT and CF adulterated with different doses of LD. It was well observed fromFig. 1(c) that contaminated BT by LD even at low doses their exothermic peaks corresponding to TAG of triunsaturated (UUU) Table1 TAGsresultsofthepureandtheadulteratedsamples. TAGsMMMPLLOOLMPLPLOPPLOOOOOPPPOPPPOOSPOSPPSSOS BT0.70±0.070.70±0.073.34±0.023.34±0.0314.29±0.278.34±0.192.96±0.244.09±0.0730.81±0.1914.97±0.702.32±0.155.30±0.228.84±0.180.97±0.03 CF1.67±0.031.67±0.023.38±0.033.38±0.0410.25±0.1429.32±0.0911.90±0.016.77±0.04919.51±0.1511.75±0.112.62±0.067.36±0.510.0000.000 LD1.22±0.111.22±0.112.50±0.122.50±0.128.24±0.3325.57±0.197.81±0.154.69±0.5923.73±1.5113.14±0.212.80±0.0510.77±0.100.0000.000 BT10.79±0.140.79±0.013.32±0.613.32±0.2313.33±0.528.36±0.372.72±0.854.10±0.7431.46±1.2114.81±1.913.46±0.895.16±0.338.30±0.461.18±0.02 BT20.70±0.090.70±0.083.32±0.563.32±0.5514.01±0.137.85±0.202.39±0.603.34±0.1830.89±2.4014.25±1.193.74±0.245.86±0.1710.27±0.371.42±0.2 BT30.70±0.010.67±0.013.25±0.323.25±0.3215.60±0.398.03±0.553.06±0.123.91±0.7430.89±0.5315.14±0.173.55±0.355.18±0.599.73±0.291.38±0.13 BT40.74±0.080.74±0.083.16±0.243.16±0.2315.53±0.588.06±0.513.01±0.144.39±0.1930.32±0.4415.79±0.382.46±0.264.91±0.188.62±0.511.40±0.14 BT50.67±0.090.67±0.093.38±0.423.38±0.4114.80±1.077.76±0.562.57±0.704.12±0.6630.87±0.2915.70±0.612.86±0.785.15±0.419.67±0.361.44±0.15 BT60.70±0.090.70±0.093.47±0.493.47±0.4915.88±0.517.72±0.542.97±0.204.07±0.7230.02±0.9015.90±0.303.24±0.705.20±0.279.56±0.521.70±0.11 CF11.93±0.013.47±0.043.47±0.033.47±0.0310.54±0.1427.99±0.6911.23±0.056.62±0.0217.81±0.1910.35±0.292.11±0.022.34±0.060.0000.000 CF22.08±0.032.07±0.033.43±0.093.43±0.1010.44±0.2728.07±0.1511.01±0.136.47±0.0117.77±0.1210.59±0.322.14±0.042.37±0.020.0000.000 CF31.91±0.051.91±0.043.30±0.033.30±0.0510.21±0.6627.32±1.3210.77±0.636.40±0.0717.36±0.8010.51±0.261.72±0.072.30±0.030.0000.000 CF42.08±0.013.39±0.053.39±0.053.38±0.1010.67±0.0327.57±0.7611.13±0.186.44±0.0317.80±0.0610.31±0.012.15±0.022.58±0.030.0000.000 CF52.05±0.032.04±0.023.42±0.273.40±0.2610.43±0.9727.84±1.2810.82±0.826.32±0.4817.57±1.5010.20±0.932.15±0.052.67±0.210.0000.000 CF61.92±0.041.92±0.043.43±0.043.42±0.0810.41±0.1328.11±0.1910.97±0.226.42±0.0317.98±0.4610.55±0.122.13±0.012.77±0.010.0000.000 Abbreviations:⁄DominantTAGspecies;BTispurebeeftallow;CFispurechickenfat;LDispurelard;BT,(b)LD,and(c)CF;where:M,myristic;P,palmitic;O,oleic;L,linoleic;S,stearic. BT1:99.5%BT+0.5%LDCF1:99.5%CF+0.5%LD. BT2:99%BT+1%LDCF2:99%CF+1%LD. BT3:98%BT+2%LDCF3:98%CF+2%LD. BT4:97%BT+3%LDCF4:97%CF+3%LD. BT5:96%BT+4%LDCF5:96%CF+4%LD. BT6:95%BT+5%LDCF6:95%CF+5%LD.
and diunsaturated (SUU/USU/UUS) started to shift to the higher temperature region of the DSC cooling thermograms. In addition, Fig. 1(c) showed also that the height of exothermic peaks increased gradually with the increasing of the LD doses, especially when a LD dose is more than 1%. These exothermic peaks have been found to be shifted continuously toward higher temperatures as well. For instance, the first peak was shifted from(40°C) when LD dose was 0.5% to(37°C) when LD dose was 5%, while the end of the second peak was shifted from8°C to12°C when LD dose was 0.5% and 5%, respectively. This peaks movement can be a very strong indicator on the contamination levels associated to the
capability of small doses of lard to change the crystallisation regime of mixed CF and BT edible lipid materials. On the other side, DSC cooling thermograms of CF contaminated with LD inFig. 1(d) revealed an opposite shifting in the exothermic events toward low- er temperatures region of the DSC cooling thermograms, and this behaviour was observed only when the dose of LD exceeded 2%.
For instance, the first peak was detected at (–41°C) and the end of the second peak was appeared at3°C when the LD dose equal 0.5%, while they start shifting to (42°C) and 1°C, respectively when the LD dose increased to 5%. This peaks movement and the narrowing of transition was due to the TAG Fig. 1.(a and b) DSC cooling and heating thermograms of the pure samples including LD, BT, and CF, respectively. (c and d) DSC cooling thermograms of BT and CF adulterated with different dose of LD, respectively. (e and f) DSC melting thermograms of BT and CF adulterated with different dose of LD, respectively.
Table 2
Cooling and heating thermograms concerning pure and treated samples.
Crystallisation parameters Melting parameters
Samples Code Onset Enthalpy End set Onset Enthalpy End set
BT 1 40.140 39.123 8.613 44.505 69.100 36.500
LD 2 45.410 76.065 15.895 50.390 56.940 30.390
CF 3 41.033 22.625 16.315 46.140 69.753 32.650
S1 4 40.580 36.870 8.395 52.200 31.571 28.720
S2 5 40.350 31.820 10.070 48.260 32.160 29.080
S3 6 40.180 30.170 12.147 45.370 38.716 30.325
S4 7 38.020 29.340 12.180 42.810 28.457 32.585
S5 8 37.850 26.220 12.310 40.050 27.808 32.760
S6 9 37.160 20.450 12.460 38.210 19.850 32.920
S1⁄ 10 41.420 20.900 3.395 42.260 16.390 31.240
S2⁄ 11 41.530 21.440 3.270 46.240 17.310 29.420
S3⁄ 12 41.680 21.820 3.147 47.180 18.075 28.980
S4⁄ 13 41.850 22.740 3.080 48.760 19.093 28.170
S5⁄ 14 41.980 23.770 2.460 50.420 16.620 28.050
S6⁄ 15 42.160 25.060 1.930 50.600 19.820 27.435
Abbreviations:BT is pure beef tallow; CF is pure chicken fat; LD is pure lard.
S1/BT 1: 99.5% BT + 0.5% LD.
S2/BT 2: 99% BT + 1% LD.
S3/BT 3: 98% BT + 2% LD.
S4/BT 4: 97% BT + 3% LD.
S5/BT 5: 96% BT + 4% LD.
S6/BT 6: 95% BT + 5% LD.
S1⁄/CF 1: 99.5% CF + 0.5% LD.
S2⁄/CF 2: 99% CF + 1% LD.
S3⁄/CF 3: 98% CF + 2% LD.
S4⁄/CF 4: 97% CF + 3% LD.
S5⁄/CF 5: 96% CF + 4% LD.
S6⁄/CF 6: 95% CF + 5% LD.
distribution behaviours, and their saturated/unsaturated fatty acid (SFA/USFA) ratios constituted CF and LD mixed systems.
Moreover, the exothermic peaks differ in the liberated energy (enthalpy) for crystal formation (aggregation) (Saadi et al., 2012a). Thus, in the mixture of BT and LD, the enthalpy decreased proportionally with the increasing of LD dose in BT. For instance, the DHcrystallization changed from 36.87 (J/g) when the LD dose equals to 0.5% to 20.45 (J/g) when LD dose equals to 5%. Mean- while, it was vice versa in the mixture of CF and LD, where the DHcrystallizationwas changed gradually with the increment of LD con- centrations. In this case, theDHcrystallizationwas increased from 20.9 (J/g) when the LD dose was 0.5% to 25.45 (J/g) when the LD dose was 5%. These crystallisation behaviour results might correspond to the polymorphic transition of TAGs crystals that is induced un- der the compaction of single and double H2bonds (Galeb et al., 2012). In addition, the crystallisation rate is directly proportional with high melting TAGs such as PPO, PPS, and SOS contents (Saadi et al., 2012b).
3.5. DSC melting curves in monitoring the small doses of lard adulterants
According toFredrick, Foubert, Van De Sype, and Dewettinck (2008) the melting profiles give an indication of the amount of crystallised fat and the occurrence of polymorphic transitions.
Thus, it has been considered as a very useful tool in determining melting points and various polymorphic forms associated with fat crystals (Lovegren, Gray, & Feuge, 1976). The DSC melting pro- file of BT and CF mixed with LD are shown inFig. 1(e) and (f). From Fig. 1(e) in BT mixed with LD, it is very easy to characterise two major endotherms peaks. The first event was detected at (44°C), which was attributed to the low melting TAGs such as POO, PLO and OOO contents; while the second event appeared at (6°C), which was due to the high melting TAGs such as PPO, PPP and PPS. As shown inFig. 1(e) the addition of LD into BT re- vealed numerous effects on both endothermic peaks. The first ef- fect of LD was the changing in the peaks area in dose dependant manner. The second effect was depicted from the position of those peaks, where they have shifted slightly once the concentration of lard was increased. For instance, the onset temperature moved from(42°C) to(38°C) when the LD was 0.5% and 5%, respec- tively. Meanwhile the end set temperature also moved from 26°C to30°C when LD dose was increased from 0.5% and 5%, respectively. Moreover, the third effect of LD addition into BT was remarked from the changes in the enthalpy levels. The interac- tion between energy and TAGs during melting regime caused sig- nificant changes in the DHmelting. As a result, the DHmelting has been found to be changed with LD in dose dependant manner.
For example, it decreased gradually from 69.10 (J/g) to 19.86 (J/g) when LD dose varied from 0% to 5%, respectively. However, in Fig. 1(f), opposite effects were observed in the melting curve of CF and LD mixtures including the area of both endothermic peaks, the position of those peaks, and their enthalpy values. For instance, the onset temperature decreased from (42°C) to (50°C) when LD was increasing from 0.5% to 5%, respectively. Meanwhile the end set temperature also dropped from31°C to27°C when LD was 0.5% and 5%, respectively. In addition, the level of the DHmeltinghas been found to be increased from 16.38 (J/g) to 19.82 (J/g) when LD dose rose from 0% to 5%, respectively.
3.6. Chemometric results interpretation: Principal components analysis (PCA)
In order to differentiate and to classify LD from BT, CF and their different blending, the multivariate data matrix of 15 samples (i.e., the pure BF, LA, and CF as well as their mixtures) targeting DSC
parameters including onset temperature, enthalpy, and end set temperature was subjected to principal component analysis (PCA). Scores and loadings matrices were generated using 4 princi- pal components (PCs). An eigen-value of about 99% was achieved using four PCs where PC1 accounted for 95% of the variation, while PC2 described 4% of the variation; therefore, the remaining two (61% total) did not explain significant variability in the data. On the other hand, PCA results of cooling thermograms inFig. 2(a) do not give conclusive information about the separation and the distribution of the samples. For instance, in the scores plot the samples were displayed only in two groups. The first group (A) (1, 2 and 3) which referred to the pure samples including BT, LD, and CF, respectively. It was located in the positive side of theX- axis. The distribution of this group helped to observe that LD sam- ple located far from BT and CF samples. The second group B (4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14 and 15) referred to all adulterated samples with LD (BT + LD, and CF + LD). It was located in the negative side of theX-axis. In addition,Fig. 2(b) showed the loading plot for the different DSC parameters of cooling thermograms including onset temperature, enthalpy and end set temperature of the pure samples (BT, LD and CF) as well as mixtures by the first (PC1) and second (PC2) components. The PCA results of loadings plot de- scribed the different distribution of variables. It was well shown the non-homogenized location of the three parameters where the onset (number 1) and the end set temperatures (number 2) were located in the negative side of the figure. At the same time, the en- thalpy was located at the positive side. This distribution due to physical characteristics of these measured parameters. However, PCA results for the melting thermograms in Fig. 3(a), showed Fig. 2.(a and b) Principal component analysis (PCA) scores plot of PC1 versus PC2 showing the cooling and heating thermograms results of the pure and the adulterated samples, where: 1 refers to beef tallow (BT); 2 lard (LD); 3 chicken fat (CF); 4 BT + 0.5% LD; 5 BT + 1% LD; 6 BT + 2% LD; 7 BT + 3% LD; 8 BT + 4% LD; 9 BT + 5% LD; 10 CF + 0.5% LD; 11 CF + 1% LD; 12 CF + 2% LD; 13 CF + 3% LD; 14 CF + 4%
LD; and 15 CF + 5% LD.
different results compared with the cooling thermograms. It (Fig. 3 (a)) demonstrated the scores plot of the different concentrations of the samples (BT + LD, and CF + LD) as well as the pure samples BT, CF and LD. The PCA results of the scores plot described the different distribution of variables (concentrations) in three main groups. The first groupA(1, 2 and 3) that referred to pure BT and pure LD and pure CF respectively are very isolated from the other two groups and their distance is convergent; and it was located under of the X-axis. Meanwhile, the second groupB(4, 5, 6, 7 and 8) referred to BT + LD and the third groupC(9, 10, 11, 12, 13, 14 and 15) which referred to CF + LD was located up ofX-axis. This grouping of sam- ples was because of the distinction in their physico-chemical char- acteristics; and it is known that TAGs play extremely important role in this differentiation (Wood et al., 2008). Therefore,Fig. 3(a) helped to differentiate BT and CF from LA, and it aided to distin- guish very low adulterated concentration (0.5%) of LD in BT and CF. In addition,Fig. 3(b) showed the loadings plot for the different DSC parameters of melting thermograms including onset tempera- ture, enthalpy, and end set temperature of the pure samples (BT, LD and CF) as well as mixtures. The PCA result of the loadings plot of heating thermograms appeared similar to that one from cooling results, where it described the different distribution of variables.
The allocation of the three parameters was very separately. For in- stance, the onset (number 1) and the end set temperatures (num- ber 2) were located in the negative side of the figure. At the same time, the enthalpy was located at the positive side. This distribu- tion was due to physical characteristics of these measured parameters.
4. Conclusion
This study has shown the ability of DSC coupled with PCA in discriminating LD adulteration even at low mixed dose of less than 0.5%. Therefore, the crystallisation and melting profiles provided clear evidence on the cross-contamination caused in between BF–LD and CF–LD. According toSaadi et al. (2012a)the cooling curves are more interpretable, whereas the melting profiles is
more complex due to the TAGs melting kinetics and their transi- tion protocols. This study also followed similar trends, where cool- ing thermograms provides better definition of different animal profiles adulterated with LD.
References
Aktas, N., & Kaya, M. (2001). Detection of beef body fat and margarine in butterfat by differential scanning calorimetry.Journal of Thermal Analysis and Calorimetry, 66, 795–801.
Bligh, E. G., & Dyer, W. J. (1959). A rapid method for total lipid extraction and purification.Canadian Journal of Biochemistry and Physiology, 37, 911–917.
Brereton, R. G. (1990).Chemometrics, applications of mathematics and statistics to laboratory systems. London: Ellis Horwood.
Caponio, F., Chiavaro, E., Paradiso, V. M., Paciulli, M., Summo, C., Cerretani, L., et al.
(2013). Chemical and thermal evaluation of olive oil refining at different oxidative levels. European Journal of lipid Science and Technology, 115(10), 1146–1154.
Chiavaro, E., Barnaba, C., Vittadini, E., Teres, M., Estrada, R., Cerretani, L., et al.
(2009). Microwave heating of different commercial categories of olive oil: Part II. Effect on thermal properties.Food Chemistry, 115, 1393–1400.
Chiavaro, E., Cerretani, L., Paradiso, V. M., Summo, C., Paciulli, M., Toschi, T. G., et al.
(2013). Thermal and chemical evaluation of naturally auto-oxidised virgin olive oils: A correlation study.Journal of the Science of Food and Agriculture, 93(12), 2909–2916.
Chiavaro, E., Vittadini, E., Rodrigeuez-Estrada, M. T., Cerretani, L., & Bendini, A.
(2008). Differential scanning calorimetry: A potential tool for discrimination of olive oil commercial categories.Food Chemistry, 110, 248–256.
Coni, E., Di Pasquale, M., Coppolelli, P., & Bocca, P. (1994). Detection of animal fats in butter by differential scanning calorimetry: A pilot study. Journal of the American Oil Chemists’ Society, 71, 807–810.
Dahimi, O., Hassan, M. S., Rahim, A. A., Abdulkarim, S. M., & Mashitoh, S. A. (2014).
Differentiation of lard from other edible fats by gas chromatography–flame ionisation detector (GC–FID) and chemometrics.Journal of Food Pharmaceutical Science, 2, 27–31.
Dourtoglou, V. G., Dourtoglou, T. H., Antonopoulos, A., Stefanou, E., Lalas, S., &
Poulos, C. (2003). Detection of olive oil adulteration using principal component analysis applied on total and regio FA content.JAOCS, 80(3), 203–208.
Fredrick, E., Foubert, I., Van De Sype, J., & Dewettinck, K. (2008). Influence of monoglycerides on the crystallization behavior of palm oil.Crystal Growth &
Design, 8(6), 1833–1839.
Galeb, H. A., Salimon, J., Eid, E. E. M., Nacer, N. E., Saari, N., & Saadi, S. (2012). The impact of single and double hydrogen (H2) bonds on crystallization and melting regimes of Ajwa and Barni lipids.Food Research International, 48, 657–666.
Kim, N. S., Lee, J. H., Han, K. M., Kim, J. W., Cho, S., & Kim, J. (2014). Discrimination of commercial cheeses from fatty acid profiles and phytosterol contents obtained by GC and PCA.Food Chemistry, 143, 40–47.
Lambelet, P., & Ganguli, N. C. (1983). Detection of pig and buffalo body fat in cow and buffalo ghees by differential scanning calorimetry.Journal of the American Oil Chemists’ Society, 60, 1005–1008.
Lovegren, N. V., Gray, M. S., & Feuge, R. O. (1976). Polymorphic changes in mixtures of confectionery fats.Journal of the American Oil Chemists’ Society, 53, 83–88.
Maggio, R. M., Cerretani, L., Barnaba, C., & Chiavaro, E. (2012). Application of differential scanning calorimetry–chemometric coupled procedure to the evaluation of thermo-oxidation on extra virgin olive oil. Food Biophysics, 7, 114–123.
Marangoni, A. G., & Narine, S. S. (2002). Identifying key structural indicators of mechanical strength in networks of fat crystals.Food Research International, 35, 957–969.
Marikkar, J. M. N., Ghazali, H. M., Che Man, Y. B., & Lai, O. M. (2002). The use of cooling and heating thermograms for monitoring of tallow, lard and chicken fat adulterations in canola oil.Food Research International, 35, 1007–1014.
Marikkar, J. M. N., Lai, O. M., Ghazali, H. M., & Che Man, Y. B. (2002). Compositional and thermal analysis of RBD palm oil adulterated with lipase-catalyzed interesterified lard.Food Chemistry, 76, 249–258.
Narine, S. S., & Marangoni, A. G. (2002). Structure and mechanical properties of fat crystal networks.Advances in Food and Nutrition Research, 44, 33–145.
Ntakatsane, M. P., Liu, X. M., & Zhou, P. (2013). Short communication: Rapid detection of milk fat adulteration with vegetable oil by fluorescence spectroscopy.Journal of Dairy Science, 96, 2130–2136.
Saadi, S., Ariffin, A. A., Ghazali, H. M., Miskandar, M. S., Abdulkarim, S. M., & Boo, H.
C. (2011). Effect of blending and emulsification on thermal behavior, solid fat content and microstructure properties of palm oil based margarine fats.Journal of Food Science, 76(1), 21–30.
Saadi, S., Ariffin, A. A., Ghazali, H. M., Miskandar, M. S., Boo, H. C., & Abdulkarim, S.
M. (2012a). Application of differential scanning calorimetry (DSC), HPLC and pNMR for interpretation primary crystallization caused by combined low and high melting TAGs.Food Chemistry, 132, 603–612.
Saadi, S., Ariffin, A. A., Ghazali, H. M., Miskandar, M. S., Boo, H. C., & Abdulkarim, S.
M. (2012b). Crystallisation regime of w/o emulsion [e.g. multipurpose margarine] models during storage.Food Chemistry, 133(4), 1485–1493.
Santos, P. M., Pereira-Filho, E. R., & Rodriguez-Saona, L. E. (2013). Rapid detection and quantification of milk adulteration using infrared microspectroscopy and chemometrics analysis.Food Chemistry, 138(2013), 19–24.
Fig. 3.(a and b) The loadings plot of cooling and heating thermograms for the first two principal components (PCs) for DSC parameters of the studied samples where 1 refers to onset temperature, 2 for enthalpy, and 3 for end set temperature.
Sato, K. (1996). Polymorphism of pure triacylglycerols and natural fats. In F. Padley (Ed.),Advances in applied lipid research(pp. 213–268). New York: JAI Press.
Sato, K., & Ueno, S. (2001). Molecular interactions and phase behavior of polymorphic fats. In N. Garti & K. Sato (Eds.),Crystallization processes in fats and lipid systems(pp. 177–209). New York: Marcel Dekker.
Sato, K., Ueno, S., & Yano, J. (1999). Molecular interactions and kinetic properties of fats.Progress in Lipid Research, 38, 91–116.
Subari, N., Saleh, J. M., Shakaff, A. Y. M., & Zakaria, A. (2012). A hybrid sensing approach for pure and adulterated honey classification. Sensors, 12, 14022–14040.http://dx.doi.org/10.3390/s121014022.
Tan, C. P., & Che Man, Y. B. (2002). Comparative differential scanning calorimetric analysis of vegetable oils: I. Effect of heating rate variation. Phytochemical Analysis, 13, 129–141.
Tang, D., & Marangoni, A. G. (2007). Modeling the rheological properties and structure of colloidal fat crystal networks.Trends in Food Science and Technology, 18, 474–483.
Wesołowski, M., & Erecin´ska, J. (1998). Thermal analysis in quality assessment of rapeseed oils.Thermochimica Acta, 323, 137–143.
Wood, J. D., Enser, M., Fisher, A. V., Nute, G. R., Sheard, P. R., Richardson, R. I., et al.
(2008). Fat deposition, fatty acid composition and meat quality: A review.Meat Science, 78, 343–358.
Zhao, F., Liu, J., Wang, X., Li, P., Zhang, W., & Zhang, Q. (2013). Detection of adulteration of sesame and peanut oils via volatiles by GC T GC–TOF/MS coupled with principal components analysis and cluster analysis.Europeans Journal of Lipid Science Technology, 115, 337–347.
