- Email: [email protected]
The use of Fourier transform mid infrared (FT-MIR) spectroscopy for detection and quantification of adulteration in virgin coconut oil
Food Chemistry 129 (2011) 583–588
Contents lists available at ScienceDirect
Food Chemistry journal homepage: www.elsevier.com/locate/foodchem
Analytical Methods
The use of Fourier transform mid infrared (FT-MIR) spectroscopy for detection and quantification of adulteration in virgin coconut oil Abdul Rohman a,c, Yaakob B. Che Man b,d,⇑ a
Halal Research Group, Gadjah Mada University, Yogyakarta 55281, Indonesia Halal Products Research Institute, Universiti Putra Malaysia, 43400 UPM, Serdang, Selangor, Malaysia c Department of Pharmaceutical Chemistry, Faculty of Pharmacy, Gadjah Mada University, Yogyakarta 55281, Indonesia d Department of Food Technology, Faculty of Food Science and Technology, Universiti Putra Malaysia, 43400 UPM, Serdang, Selangor, Malaysia b
a r t i c l e
i n f o
Article history: Received 2 March 2010 Received in revised form 27 February 2011 Accepted 22 April 2011 Available online 29 April 2011 Keywords: FT-MIR spectroscopy Virgin coconut oil Corn oil Sunflower oil Adulteration
a b s t r a c t Currently, the authentication of virgin coconut oil (VCO) has become very important due to the possible adulteration of VCO with cheaper plant oils such as corn (CO) and sunflower (SFO) oils. Methods involving Fourier transform mid infrared (FT-MIR) spectroscopy combined with chemometrics techniques (partial least square (PLS) and discriminant analysis (DA)) were developed for quantification and classification of CO and SFO in VCO. MIR spectra of oil samples were recorded at frequency regions of 4000–650 cm1 on horizontal attenuated total reflectance (HATR) attachment of FTIR. DA can successfully classify VCO and that adulterated with CO and SFO using 10 principal components. Furthermore, PLS model correlates the actual and FTIR estimated values of oil adulterants (CO and SFO) with coefficient of determination (R2) of 0.999. Ó 2011 Elsevier Ltd. All rights reserved.
1. Introduction Since very early time, fats and oils have been liable to adulteration, either intentionally or accidentally (Rossell, King, & Downes, 1983). The detection of adulteration is an attractive issue for researchers, because food producers do not wish to be subjected to unfair competition from devious processors who would get economical profit (Gallardo-Velázquez, Osorio-Revilla, Zuñiga-de Loa, & Rivera-Espinoza, 2009). The detection of adulteration is more difficult, especially when the adulterant has similar chemical composition to that of the original oil (Anklam & Bantaglia, 2001). Virgin coconut oil (VCO) is obtained from the flesh coconut in which the oil extraction does not involve the use of thermal or chemical treatments (Nik Norulaini et al., 2009). VCO is an emerging functional food oils due to its ability to posses several biological activities such as antiviral and antimicrobial (Marina, Che Man, & Ismail, 2009). In the market, it is estimated that the price of VCO is approximately 10–20 times higher than that of common plant oils like corn, palm, and sunflower oils. Therefore, VCO is a target to adulteration practise with the low price plant oils. Several analytical methods have been developed for detection and quantification of adulterants in fats and oils such as ⇑ Corresponding author at: Halal Products Research Institute, Universiti Putra Malaysia, 43400 UPM, Serdang, Selangor, Malaysia. Tel.: +60 3 89430405; fax: +60 3 89439745. E-mail address: [email protected] (Y.B. Che Man). 0308-8146/$ - see front matter Ó 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.foodchem.2011.04.070
differential scanning calorimetry (DSC) (Chiavaro, Vittadini, Rodriguez-Estrada, Cerretani, & Bendini, 2008); spectroscopic based-methods (Lerma-García, Ramis-Ramos, Herrero-Martínez, & Simó-Alfonso, 2010), and wet chemical methods. Quantification of adulterants in fats and oils by chromatographic method was reviewed by Cserhati, Forgacs, Deyl, and Miksik (2005) and by Aparicio and Aparicio-Ruiz (2000). For the analysis of VCO adulteration, Marina, Che Man, and Amin (2010) have developed electronic nose based on surface acoustic wave in combination with the chemometrics of principal component analysis to monitor the adulteration of VCO with palm kernel oil. Another technique used was DSC for monitoring the presence of palm kernel oil and soybean oil (Marina, Che Man, Nazimah, & Amin, 2009). Some of these methods are impractical and too laborious. Therefore, rapid and accurate analytical methods must be developed in order to detect and to quantify the oil adulterants. Over the last 3 decades, mid infrared (MIR) spectroscopy combined with chemometric methods have been used in numerous analytical applications (Sinelli, Cerretani, Di Egidio, Bendini, & Casiraghi, 2009). MIR spectroscopy has been identified as an ideal analytical method for authenticity studies of edible fats and oils (Reid, O’Donnell, & Downey, 2006) due to its capability to serve as ‘‘fingerprint’’ technique, meaning that there are no two samples with the same FTIR spectra, either in the number of peaks or in the maximum peak intensities (Pavia, Lampman, & Kriz, 2001). The methods allow sensitive, fast, and reliable technique, ease in sample presentation, and can be used for monitoring the quality
584
A. Rohman, Y.B. Che Man / Food Chemistry 129 (2011) 583–588
aspects of fats and oils at spectral region of 4000–650 cm1 (Roggo et al., 2007; Wilson & Tapp, 1999). FT-MIR spectroscopic methods were developed for detection and quantification of oil adulterants such as sunflower, corn, soybean and hazelnut oils in extra virgin olive oil (EVOO) using chemometrics of multiple linear regression and linear discriminant analysis (Lerma-García et al., 2010), sunflower and corn oils in EVOO with aid of principal component analysis and PLS-discriminant analysis (Gurdeniz & Ozen, 2009), soybean oil in camellia oil (Wang, Lee, Wang, & He, 2006), hazelmut oil in refined olive oil (Baeten et al., 2005) and animal fats in cod-liver oil (Rohman & Che Man, 2009a). Our group has used FT-MIR spectroscopy combined with chemometrics of PLS and DA for quantification and classification of palm kernel oil (Manaf, Che Man, Hamid, Ismail, & Syahariza, 2007) and palm oil (Rohman & Che Man, 2009b) as oil adulterants in VCO. The present study highlights the application of FT-MIR spectroscopy for detecting and quantifying corn oil (CO) and sunflower oil (SFO) as oil adulterants in VCO.
2. Materials and methods The samples of virgin coconut oil (VCO), sunflower oil (SFO), and corn oil (CO) were purchased from the local market in Jogjakarta, Indonesia. The used plant oils were coming from the mixture of three different brands with similar fatty acid (FA) composition. The FA profiles of these oils are in accordance with those listed in Codex Alimentarius (2003). Some factors contributed to the slight FTIR spectral variation of VCO, such as origin of plants, year of coconut plantation, maturity, etc. For this reason, the samples of VCO were mixed to compensate this variation. 2.1. Classification The classification of VCO and VCO blended with adulterants (CO and SFO) was carried out using discriminant analysis (DA). DA is one of the supervised pattern recognition techniques which start with a number of samples whose group membership is known. These samples are sometimes called the learning or training samples (Miller & Miller, 2005). VCO and adulterants were blended in order to obtain a series of training sets of pure VCO (50% VCO in chloroform) and VCO containing 1–50% of adulterants in chloroform. VCO samples mixed with adulterants were marked as ‘‘adulterated’’, while a series of pure VCO was assigned with ‘‘VCO’’. Both classes were classified using DA based on their FT-MIR spectra.
Table 1 The composition percentage for calibration and validation sets used in the binary mixtures of CO and SFO with VCO. Samples
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
CO in VCO
SFO in VCO
Calibration
Validation
Calibration
Validation
CO
VCO
CO
VCO
SFO
VCO
SFO
VCO
2.0 3.0 5.0 6.0 8.0 9.0 10.0 12.5 15.0 17.5 22.5 25.0 27.5 30.0 32.5 35.0 50.0 60.0 70.0 100.0
98.0 97.0 95.0 94.0 92.0 91.0 90.0 87.5 85.0 82.5 77.5 75.0 72.5 70.0 67.5 65.0 50.0 40.0 30.0 0.0
0.0 1.5 2.5 7.5 8.0 10.0 14.0 15.0 20.0 22.5 25.0 30.0 32.5 37.5 47.5 50.0 55.0 60.0 65.0 80.0
100.0 98.5 97.5 92.5 92.0 90.0 86.0 85.0 80.0 77.5 75.0 70.0 67.5 62.5 52.5 50.0 45.0 40.0 35.0 20.0
0.0 1.5 3.0 4.0 5.0 6.0 7.5 10.0 12.5 15.0 17.5 20.0 22.5 25.0 30.0 35.0 40.0 50.0 75.0 100.0
100.0 98.5 97.0 96.0 95.0 94.0 92.5 90.0 87.5 85.0 82.5 80.0 77.5 75.0 70.0 65.0 60.0 50.0 25.0 0.0
0.0 2.0 3.0 4.0 5.0 6.0 7.5 10.0 12.5 15.0 17.5 20.0 22.5 25.0 30.0 35.0 37.5 40.0 45.0 50.0
100.0 98.0 97.0 96.0 95.0 94.0 92.5 90.0 87.5 85.0 82.5 80.0 77.5 75.0 70.0 65.0 62.5 60.0 55.0 50.0
cleaned blank HATR crystal before the measurement of each oil sample replication. The sample spectra were collected in triplicate and displayed as the average spectra. At the end of every scan, the surface of HATR crystal was cleaned with hexane twice and dried with special soft tissue, cleaned with acetone, and finally dried with soft tissue following the collection of each spectrum. 2.4. Chemometrics The chemometrics analyses were performed using the software TQ Analyst™ version 6 (Thermo electron Corporation, Madison, WI). Classification and quantification of adulterants (CO and SFO) in VCO were carried out using discriminant analysis (DA) and partial least square (PLS), respectively. Frequency regions for PLS and DA were automatically selected by the software and were confirmed by investigating peaks where variations were observed. PLS calibration model was cross-validated using ‘‘leave-one-out’’ technique. This model was further used to predict the level of CO and SFO in independent samples in order to evaluate its predictive capability.
2.2. Quantification 3. Results and discussion For quantitative analysis, CO and SFO were mixed as binary mixture with VCO (each comprises 20 samples for calibration and 20 samples for validation). The concentration of each oils used in both calibration and validation is presented in Table 1. Each sample was subjected for FT-MIR analysis. 2.3. FT-MIR analysis FT-MIR spectra of samples were obtained using Nicolet 6700 FTIR spectrometer (Thermo Nicolet Corp., Madison, WI) with HATR crystal of ZnSe 45° equipped with deuterated triglycine sulphate (DTGS) as detector, potassium bromide (KBr) as beam splitter and controlled with the Omnic Software (Version 7.0 Thermo Nicolet). The measurements were directly carried out by putting oil samples on HATR surface at controlled room temperature (20 °C) in MIR region of 4000–650 cm1, by accumulating 32 scans with the resolution of 4 cm1. These spectra were subtracted from reference spectrum of air, acquired by collecting a spectrum from the
Chemically, fats and oils are glycerol esterified with fatty acids. Some of the fats and oils might have quite similar composition; consequently, it is often difficult to detect adulteration of fats and oils physically (Christy, Kasemsumran, Du, & Ozaki, 2004). However, because of its capability as a fingerprint technique, MIR spectroscopy allows one to differentiate authentic oils and those adulterated with others by observing the spectra changes due to the adulteration (Yap, Chan, & Lim, 2007). Fig. 1 exhibits MIR spectra of VCO, CO, and SFO at frequency region of 4000–650 cm1. The assignment of functional groups responsible for IR absorption is as follows: 3008 cm1 (trans @C–H stretch), 2954 (–CH3 asymmetrical stretch), 2922 and 2853 (symmetrical and asymmetrical stretching of –CH2), 1743 (–C@O stretch), 1654 (cis –C@C stretch), 1463 (–CH2 bending), 1417 (cis @C–H bending), 1377 (–CH3 bending), 1237 (–C–O stretch), 1160 (–C–O stretch; –CH2 bending), 1120 (–C–O stretch), 1098 (–C–O stretch), 1032 (–C–O stretch), 965 (trans-CH@CH– bending
A. Rohman, Y.B. Che Man / Food Chemistry 129 (2011) 583–588
585
Fig. 1. MIR spectra of corn oil (CO), sunflower oil (SFO), and virgin coconut oil (VCO) at MIR region of 4000–650 cm1.
out of plane), 871 (@CH2 wagging), and 722 cm1 (cis-CH@CH– bending out of plane) (Guillen & Cabo, 1997; Lerma-García et al., 2010). Taking into account the spectrum of VCO, CO, and SFO, it can be seen that spectra of CO and SFO revealed some differences to VCO, especially in region around 3008 cm1 and at fingerprint region (1500–650 cm1). There is no band at 3008 cm1 for VCO, and the otherwise was observed for CO and SFO. Furthermore, at spectral regions of 1120–1098 cm1, VCO has one peak; meanwhile CO and SFO reveal two peaks. These differences can be exploited for detection and quantification of CO and SFO as adulterants in VCO. 3.1. Classification DA was used to make the classification between pure VCO and that adulterated with CO and SFO. DA can be exploited to determine the class of VCO to that adulterated with CO and SFO by calculating the distance from each class centre using the Mahalanobis distance units. After the classification model is obtained, the class of unknown samples to that of the definite classes can be predicted (Ballabio & Todeschini, 2009). The classification of VCO and that adulterated with CO was carried out using spectral regions at combined frequencies of 3028– 2983, 2947–1887, and 1685–868 cm1, meanwhile frequencies at 3030–2980 and 1300–1000 cm1 was exploited for classification of VCO adulterated with SFO. The selection of this frequency region was based on its capability to provide the least or no misclassification between two classes (pure VCO and adulterated VCO). The Coomans plot for the classification between pure VCO and VCO adulterated with CO and SFO using 10 principal components is shown in Fig. 2A and B, respectively. It is clear from Fig. 2 that both classes (pure VCO and VCO mixed with CO and SFO) are well separated. DA accurately classifies 100% of all samples according to its classes, meaning that no samples were classified into the wrong group (Tay, Singh, Krishnan, & Gore, 2002). 3.2. Quantification The quantification of CO and SFO as adulterants in VCO was carried out using PLS algorithm. The spectral regions used for PLS calibration models are 858–705, 943–863, 1392–983, and
3027–2983 cm1 for quantitative analysis of CO in VCO, and at 1685–686, 2946–1887, and 3027–2983 cm1 for quantification of SFO in VCO. The selection of these frequency regions was based on the optimisation processes in which they offer the highest values of R2 and the lowest values of error, either in calibration or in prediction models. The five principal components (factors) were sufficient for describing PLS model for both adulterants. Fig. 3 shows the PLS calibration model which correlates the actual and estimated values of CO and SFO (%v/v) obtained from FT-MIR spectra at the specified regions. The difference between the actual and the observed concentration of adulterants is relatively small with coefficient of determination (R2) values are 0.999 for both adulterants (CO and SFO). Root mean square error of calibration (RMSEC) was used to evaluate the error in calibration model. RMSEC value was calculated as follows:
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi Pn 2 i¼1 ðactual calculatedÞ RMSEC ¼ Nf 1 ðParadkar and Irudayaraj; 2002Þ The term ‘‘actual’’ refers to the known or true concentration of selected standards. Meanwhile the ‘‘calculated’’ or ‘‘predicted’’ refers to a value computed by the model using spectral data; where N is the number of samples used in the calibration sets; and f is number of factors used in the calibration model. The low value RMSEC indicates the good performance of PLS model. The RMSEC values of CO and SFO in VCO obtained are 0.866% and 0.374% (v/v), respectively. In order to asses the prediction ability of the developed model, PLS calibration model was used to predict the levels of independent CO and SFO in VCO samples as validation or prediction data sets. The evaluation of the goodness of fit in the validation is performed by calculating the root mean square error of prediction (RMSEP) and R2. RMSEP is calculated using the following equation:
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi Pm 2 i¼1 ðactual calculatedÞ RMSEP ¼ M1 ðParadkar and Irudayaraj; 2002Þ where M is the number of samples used in the prediction sets. The values of R2 are 0.998 for both adulterants (Fig. 4); meanwhile the RMSEP values are 0.994% and 1.060% (v/v), respectively for analysis of CO and SFO as adulterants in VCO samples. The high
586
A. Rohman, Y.B. Che Man / Food Chemistry 129 (2011) 583–588
Fig. 2. The Coomans plot of VCO and adulterants: (h) VCO; (4) VCO containing adulterants. (A) VCO adulterated with CO. (B) VCO adulterated with SFO.
Fig. 3. PLS calibration model for the relationship between actual and estimated concentrations of CO and SFO in VCO. (A) CO in VCO. (B) SFO in VCO.
values of R2 and low values of RMSEP indicate the success of PLS regression model. Table 2 compiled PLS performance in terms of R2, RMSEC, RMSEP, and the number of principal components for
quantification of CO and SFO in VCO. The scatter plot for the relationship between actual and estimated concentration of CO and SFO in validation model is shown in Fig 4.
587
A. Rohman, Y.B. Che Man / Food Chemistry 129 (2011) 583–588
Fig. 4. The relationship between actual and estimated concentrations of CO and SFO in VCO in validation model. (A) CO in VCO. (B) SFO in VCO.
Table 2 PLS performance for analysis of corn and sunflower oils as oil adulterants in VCO. Adulterants
Principal components
Corn oil Sunflower oil
5 5
R2 Calibration
Prediction
0.999 0.999
0.998 0.996
The developed PLS model was further evaluated by cross-validation using ‘‘leave one out’’ technique. In this technique, one of the calibration samples is removed. Subsequently, the removed sample was predicted with the fashioned model using the residual samples and the procedure was repeated until each sample was excluded once from the model (Gurdeniz, Tokatli, & Ozen, 2007). The values of root mean square error of cross validation (RMSECV) obtained are relatively low, i.e. 1.68% and 1.32% (v/v), respectively for CO and SFO. Based on this result, it can be stated that PLS appears to have a reasonable ability to estimate the percentage of CO and SFO as oil adulterants in VCO samples.
RMSEC (%v/v)
RMSEP (%v/v)
0.866 0.374
0.99 1.06
classify VCO and that adulterated with adulterants using 10 principal components. Acknowledgments The first author acknowledges to The Ministry of National Education, Republic of Indonesia for its scholarship to pursue Ph.D. programme in Halal Products Research Institute, Universiti Putra Malaysia (UPM). References
4. Conclusions It can be concluded that FT-MIR spectroscopy using HATR accessory in combination with chemometrics can be used to detect and to quantify the adulteration of virgin coconut oil with corn and sunflower oils. The level of adulterants was successfully determined with the aid of PLS calibration model. DA can correctly
Anklam, E., & Bantaglia, R. (2001). Food analysis and consumer protection. Trends in Food Science and Technology, 12, 71–102. Aparicio, R., & Aparicio-Ruız, R. (2000). Review: Authentication of vegetable oils by chromatographic techniques. Journal of Chromatography A, 881, 93–104. Baeten, V., Pierna, J. A. F., Dardenne, P., Meurens, M., García-González, D. L., & Aparicio-Ruiz, R. (2005). Detection of the presence of hazelnut oil in olive oil by FT-Raman and FT-MIR spectroscopy. Journal of Agricultural and Food Chemistry, 53, 6201–6206.
588
A. Rohman, Y.B. Che Man / Food Chemistry 129 (2011) 583–588
Ballabio, D., & Todeschini, R. (2009). Multivariate classification for qualitative analysis. In D.-W. Sun (Ed.), Infrared spectroscopy for food quality: Analysis and control (pp. 83–104). London: Elsevier. Chiavaro, E., Vittadini, E., Rodriguez-Estrada, M. T., Cerretani, L., & Bendini, A. (2008). Diffrential scanning calorimeter application to the detection of refined hazelnut oil in extra virgin olive oil. Food Chemistry, 110, 248–256. Christy, A. A., Kasemsumran, S., Du, Y., & Ozaki, K. (2004). The detection and quantification of adulteration in olive oil by near-infrared spectroscopy and chemometrics. Analytical Sciences, 20, 935–940. Cserhati, T., Forgacs, E., Deyl, Z., & Miksik, I. (2005). Chromatography in authenticity and traceability tests of vegetable oils and dairy products: A review. Biomedical Chromatography, 19, 183–190. Codex Alimentarius Commision (2003). Amended. Codex Standard for Named vegetable Oils. Codex Stan 210. Gallardo-Velázquez, T., Osorio-Revilla, G., Zuñiga-de Loa, M., & Rivera-Espinoza, Y. (2009). Application of FTIR-HATR spectroscopy and multivariate analysis to the quantification of adulterants in Mexican honeys. Food Research International, 42, 313–318. Guillen, M. D., & Cabo, N. (1997). Characterization of edible oils and lard by Fourier transform infrared spectroscopy. Relationships between composition and frequency of concrete bands in the fingerprint region. Journal of the American Oil Chemists Society, 74, 1281–1286. Gurdeniz, G., Tokatli, F., & Ozen, B. (2007). Differentiation of mixtures of monovarietal olive oils by mid-infrared spectroscopy and chemometrics. European Journal of Lipid Science and Technology, 109, 1194–1202. Gurdeniz, G., & Ozen, B. (2009). Detection of adulteration of extra-virgin olive oil by chemometric analysis of mid-infrared spectral data. Food Chemistry, 116, 519–525. Lerma-García, M. J., Ramis-Ramos, G., Herrero-Martínez, J. M., & Simó-Alfonso, E. F. (2010). Authentication of extra virgin olive oils by Fourier-transform infrared spectroscopy. Food Chemistry, 118, 78–83. Manaf, M. A., Che Man, Y. B., Hamid, N. S. A., Ismail, A., & Syahariza, Z. A. (2007). Analysis of adulteration of virgin coconut oil by palm kernel olein using Fourier transform infrared spectroscopy. Journal of Food Lipids, 14, 111–121. Marina, A. M., Che Man, Y. B., Nazimah, S. A. H., & Amin, I. (2009a). Monitoring the adulteration of virgin coconut oil by selected vegetable oils using differential scanning calorimetry. Journal of Food Lipids, 16, 50–61. Marina, A. M., Che Man, Y. B., & Ismail, A. (2009b). Virgin coconut oil: Emerging functional food oil. Trends in Food Science and Technology, 20, 481–487. Marina, A. M., Che Man, Y. B., & Amin, I. (2010). Use of the SAW sensor electronic nose for detecting the adulteration of virgin coconut oil with RBD palm kernel olein. Journal of the American Oil Chemists Society, 87, 263–270.
Miller, J. N., & Miller, J. C. (2005). Statistics and chemometrics for analytical chemistry (5th Edition). Edinburgh Gate Harlow: Pearson Education Limited. Nik Norulaini, N. A., Setianto, W. B., Zaidul, I. S. M., Nawi, A. H., Azizi, C. Y. M., & Mohd Omar, A. K. (2009). Effects of supercritical carbon dioxide extraction parameters on virgin coconut oil yield and medium-chain triglyceride content. Food Chemistry, 116, 193–197. Paradkar, M. M., & Irudayaraj, J. (2002). A rapid FTIR spectroscopic method for estimation of caffeine in soft drinks and total methylxanthines in tea and coffee. Journal of Food Science, 657, 2507–2511. Pavia, D. L., Lampman, G. M., & Kriz, G. S. (2001). Introduction to spectroscopy (3rd ed.). Australia: Thomson Learning. Reid, L. M., O’Donnell, C. P., & Downey, G. (2006). Recent technological advances for the determination of food authenticity. Trends in Food Science and Technology, 17, 344–353. Roggo, Y., Chalus, P., Maurer, L., Lema-Martinez, C., Edmond, A., & Jent, N. (2007). A review of near infrared spectroscopy and chemometrics in pharmaceutical technologies. Journal of Pharmaceutical and Biomedical Analysis, 44, 683–700. Rohman, A., & Che Man, Y. B. (2009a). Analysis of cod-liver oil adulteration using Fourier transform infrared (FTIR) spectroscopy. Journal of the American Oil Chemists Society, 86, 1149–1153. Rohman, A., & Che Man, Y. B. (2009b). Monitoring of virgin coconut oil (VCO) adulteration with palm oil using Fourier transform infrared spectroscopy. Journal of Food Lipids, 16, 618–628. Rossell, J. B., King, B., & Downes, M. J. (1983). Detection of adulteration. Journal of the American Oil Chemists Society, 60, 333–339. Sinelli, N., Cerretani, L., Di Egidio, V., Bendini, A., & Casiraghi, E. (2009). Application of near (NIR) infrared and mid (MIR) infrared spectroscopy as a rapid tool to classify extra virgin olive oil on the basis of fruity attribute intensity. Food Research International. doi:10.1016/j.foodres.2009.10.008. Tay, A., Singh, R. K., Krishnan, S. S., & Gore, J. P. (2002). Authentication of olive oil adulterated with vegetable oils using Fourier transform infrared spectroscopy. LWT – Food Science and Technology, 35, 99–103. Wang, L., Lee, F. S. C., Wang, X., & He, Y. (2006). Feasibility study of quantifying and discriminating soybean oil adulteration in camellia oils by attenuated total reflectance MIR and fibre optic diffuse reflectance NIR. Food Chemistry, 95, 529–536. Wilson, R. H., & Tapp, H. S. (1999). Mid-infrared spectroscopy for food analysis: Recent new applications and relevant developments in sample presentation methods. Trends in Analytical Chemistry, 18, 85–93. Yap, K. Y.-L., Chan, S. Y., & Lim, C. S. (2007). Infrared-based protocol for the identification and categorisation of ginseng and its products. Food Research International, 40, 643–652.
Contents lists available at ScienceDirect
Food Chemistry journal homepage: www.elsevier.com/locate/foodchem
Analytical Methods
The use of Fourier transform mid infrared (FT-MIR) spectroscopy for detection and quantification of adulteration in virgin coconut oil Abdul Rohman a,c, Yaakob B. Che Man b,d,⇑ a
Halal Research Group, Gadjah Mada University, Yogyakarta 55281, Indonesia Halal Products Research Institute, Universiti Putra Malaysia, 43400 UPM, Serdang, Selangor, Malaysia c Department of Pharmaceutical Chemistry, Faculty of Pharmacy, Gadjah Mada University, Yogyakarta 55281, Indonesia d Department of Food Technology, Faculty of Food Science and Technology, Universiti Putra Malaysia, 43400 UPM, Serdang, Selangor, Malaysia b
a r t i c l e
i n f o
Article history: Received 2 March 2010 Received in revised form 27 February 2011 Accepted 22 April 2011 Available online 29 April 2011 Keywords: FT-MIR spectroscopy Virgin coconut oil Corn oil Sunflower oil Adulteration
a b s t r a c t Currently, the authentication of virgin coconut oil (VCO) has become very important due to the possible adulteration of VCO with cheaper plant oils such as corn (CO) and sunflower (SFO) oils. Methods involving Fourier transform mid infrared (FT-MIR) spectroscopy combined with chemometrics techniques (partial least square (PLS) and discriminant analysis (DA)) were developed for quantification and classification of CO and SFO in VCO. MIR spectra of oil samples were recorded at frequency regions of 4000–650 cm1 on horizontal attenuated total reflectance (HATR) attachment of FTIR. DA can successfully classify VCO and that adulterated with CO and SFO using 10 principal components. Furthermore, PLS model correlates the actual and FTIR estimated values of oil adulterants (CO and SFO) with coefficient of determination (R2) of 0.999. Ó 2011 Elsevier Ltd. All rights reserved.
1. Introduction Since very early time, fats and oils have been liable to adulteration, either intentionally or accidentally (Rossell, King, & Downes, 1983). The detection of adulteration is an attractive issue for researchers, because food producers do not wish to be subjected to unfair competition from devious processors who would get economical profit (Gallardo-Velázquez, Osorio-Revilla, Zuñiga-de Loa, & Rivera-Espinoza, 2009). The detection of adulteration is more difficult, especially when the adulterant has similar chemical composition to that of the original oil (Anklam & Bantaglia, 2001). Virgin coconut oil (VCO) is obtained from the flesh coconut in which the oil extraction does not involve the use of thermal or chemical treatments (Nik Norulaini et al., 2009). VCO is an emerging functional food oils due to its ability to posses several biological activities such as antiviral and antimicrobial (Marina, Che Man, & Ismail, 2009). In the market, it is estimated that the price of VCO is approximately 10–20 times higher than that of common plant oils like corn, palm, and sunflower oils. Therefore, VCO is a target to adulteration practise with the low price plant oils. Several analytical methods have been developed for detection and quantification of adulterants in fats and oils such as ⇑ Corresponding author at: Halal Products Research Institute, Universiti Putra Malaysia, 43400 UPM, Serdang, Selangor, Malaysia. Tel.: +60 3 89430405; fax: +60 3 89439745. E-mail address: [email protected] (Y.B. Che Man). 0308-8146/$ - see front matter Ó 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.foodchem.2011.04.070
differential scanning calorimetry (DSC) (Chiavaro, Vittadini, Rodriguez-Estrada, Cerretani, & Bendini, 2008); spectroscopic based-methods (Lerma-García, Ramis-Ramos, Herrero-Martínez, & Simó-Alfonso, 2010), and wet chemical methods. Quantification of adulterants in fats and oils by chromatographic method was reviewed by Cserhati, Forgacs, Deyl, and Miksik (2005) and by Aparicio and Aparicio-Ruiz (2000). For the analysis of VCO adulteration, Marina, Che Man, and Amin (2010) have developed electronic nose based on surface acoustic wave in combination with the chemometrics of principal component analysis to monitor the adulteration of VCO with palm kernel oil. Another technique used was DSC for monitoring the presence of palm kernel oil and soybean oil (Marina, Che Man, Nazimah, & Amin, 2009). Some of these methods are impractical and too laborious. Therefore, rapid and accurate analytical methods must be developed in order to detect and to quantify the oil adulterants. Over the last 3 decades, mid infrared (MIR) spectroscopy combined with chemometric methods have been used in numerous analytical applications (Sinelli, Cerretani, Di Egidio, Bendini, & Casiraghi, 2009). MIR spectroscopy has been identified as an ideal analytical method for authenticity studies of edible fats and oils (Reid, O’Donnell, & Downey, 2006) due to its capability to serve as ‘‘fingerprint’’ technique, meaning that there are no two samples with the same FTIR spectra, either in the number of peaks or in the maximum peak intensities (Pavia, Lampman, & Kriz, 2001). The methods allow sensitive, fast, and reliable technique, ease in sample presentation, and can be used for monitoring the quality
584
A. Rohman, Y.B. Che Man / Food Chemistry 129 (2011) 583–588
aspects of fats and oils at spectral region of 4000–650 cm1 (Roggo et al., 2007; Wilson & Tapp, 1999). FT-MIR spectroscopic methods were developed for detection and quantification of oil adulterants such as sunflower, corn, soybean and hazelnut oils in extra virgin olive oil (EVOO) using chemometrics of multiple linear regression and linear discriminant analysis (Lerma-García et al., 2010), sunflower and corn oils in EVOO with aid of principal component analysis and PLS-discriminant analysis (Gurdeniz & Ozen, 2009), soybean oil in camellia oil (Wang, Lee, Wang, & He, 2006), hazelmut oil in refined olive oil (Baeten et al., 2005) and animal fats in cod-liver oil (Rohman & Che Man, 2009a). Our group has used FT-MIR spectroscopy combined with chemometrics of PLS and DA for quantification and classification of palm kernel oil (Manaf, Che Man, Hamid, Ismail, & Syahariza, 2007) and palm oil (Rohman & Che Man, 2009b) as oil adulterants in VCO. The present study highlights the application of FT-MIR spectroscopy for detecting and quantifying corn oil (CO) and sunflower oil (SFO) as oil adulterants in VCO.
2. Materials and methods The samples of virgin coconut oil (VCO), sunflower oil (SFO), and corn oil (CO) were purchased from the local market in Jogjakarta, Indonesia. The used plant oils were coming from the mixture of three different brands with similar fatty acid (FA) composition. The FA profiles of these oils are in accordance with those listed in Codex Alimentarius (2003). Some factors contributed to the slight FTIR spectral variation of VCO, such as origin of plants, year of coconut plantation, maturity, etc. For this reason, the samples of VCO were mixed to compensate this variation. 2.1. Classification The classification of VCO and VCO blended with adulterants (CO and SFO) was carried out using discriminant analysis (DA). DA is one of the supervised pattern recognition techniques which start with a number of samples whose group membership is known. These samples are sometimes called the learning or training samples (Miller & Miller, 2005). VCO and adulterants were blended in order to obtain a series of training sets of pure VCO (50% VCO in chloroform) and VCO containing 1–50% of adulterants in chloroform. VCO samples mixed with adulterants were marked as ‘‘adulterated’’, while a series of pure VCO was assigned with ‘‘VCO’’. Both classes were classified using DA based on their FT-MIR spectra.
Table 1 The composition percentage for calibration and validation sets used in the binary mixtures of CO and SFO with VCO. Samples
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
CO in VCO
SFO in VCO
Calibration
Validation
Calibration
Validation
CO
VCO
CO
VCO
SFO
VCO
SFO
VCO
2.0 3.0 5.0 6.0 8.0 9.0 10.0 12.5 15.0 17.5 22.5 25.0 27.5 30.0 32.5 35.0 50.0 60.0 70.0 100.0
98.0 97.0 95.0 94.0 92.0 91.0 90.0 87.5 85.0 82.5 77.5 75.0 72.5 70.0 67.5 65.0 50.0 40.0 30.0 0.0
0.0 1.5 2.5 7.5 8.0 10.0 14.0 15.0 20.0 22.5 25.0 30.0 32.5 37.5 47.5 50.0 55.0 60.0 65.0 80.0
100.0 98.5 97.5 92.5 92.0 90.0 86.0 85.0 80.0 77.5 75.0 70.0 67.5 62.5 52.5 50.0 45.0 40.0 35.0 20.0
0.0 1.5 3.0 4.0 5.0 6.0 7.5 10.0 12.5 15.0 17.5 20.0 22.5 25.0 30.0 35.0 40.0 50.0 75.0 100.0
100.0 98.5 97.0 96.0 95.0 94.0 92.5 90.0 87.5 85.0 82.5 80.0 77.5 75.0 70.0 65.0 60.0 50.0 25.0 0.0
0.0 2.0 3.0 4.0 5.0 6.0 7.5 10.0 12.5 15.0 17.5 20.0 22.5 25.0 30.0 35.0 37.5 40.0 45.0 50.0
100.0 98.0 97.0 96.0 95.0 94.0 92.5 90.0 87.5 85.0 82.5 80.0 77.5 75.0 70.0 65.0 62.5 60.0 55.0 50.0
cleaned blank HATR crystal before the measurement of each oil sample replication. The sample spectra were collected in triplicate and displayed as the average spectra. At the end of every scan, the surface of HATR crystal was cleaned with hexane twice and dried with special soft tissue, cleaned with acetone, and finally dried with soft tissue following the collection of each spectrum. 2.4. Chemometrics The chemometrics analyses were performed using the software TQ Analyst™ version 6 (Thermo electron Corporation, Madison, WI). Classification and quantification of adulterants (CO and SFO) in VCO were carried out using discriminant analysis (DA) and partial least square (PLS), respectively. Frequency regions for PLS and DA were automatically selected by the software and were confirmed by investigating peaks where variations were observed. PLS calibration model was cross-validated using ‘‘leave-one-out’’ technique. This model was further used to predict the level of CO and SFO in independent samples in order to evaluate its predictive capability.
2.2. Quantification 3. Results and discussion For quantitative analysis, CO and SFO were mixed as binary mixture with VCO (each comprises 20 samples for calibration and 20 samples for validation). The concentration of each oils used in both calibration and validation is presented in Table 1. Each sample was subjected for FT-MIR analysis. 2.3. FT-MIR analysis FT-MIR spectra of samples were obtained using Nicolet 6700 FTIR spectrometer (Thermo Nicolet Corp., Madison, WI) with HATR crystal of ZnSe 45° equipped with deuterated triglycine sulphate (DTGS) as detector, potassium bromide (KBr) as beam splitter and controlled with the Omnic Software (Version 7.0 Thermo Nicolet). The measurements were directly carried out by putting oil samples on HATR surface at controlled room temperature (20 °C) in MIR region of 4000–650 cm1, by accumulating 32 scans with the resolution of 4 cm1. These spectra were subtracted from reference spectrum of air, acquired by collecting a spectrum from the
Chemically, fats and oils are glycerol esterified with fatty acids. Some of the fats and oils might have quite similar composition; consequently, it is often difficult to detect adulteration of fats and oils physically (Christy, Kasemsumran, Du, & Ozaki, 2004). However, because of its capability as a fingerprint technique, MIR spectroscopy allows one to differentiate authentic oils and those adulterated with others by observing the spectra changes due to the adulteration (Yap, Chan, & Lim, 2007). Fig. 1 exhibits MIR spectra of VCO, CO, and SFO at frequency region of 4000–650 cm1. The assignment of functional groups responsible for IR absorption is as follows: 3008 cm1 (trans @C–H stretch), 2954 (–CH3 asymmetrical stretch), 2922 and 2853 (symmetrical and asymmetrical stretching of –CH2), 1743 (–C@O stretch), 1654 (cis –C@C stretch), 1463 (–CH2 bending), 1417 (cis @C–H bending), 1377 (–CH3 bending), 1237 (–C–O stretch), 1160 (–C–O stretch; –CH2 bending), 1120 (–C–O stretch), 1098 (–C–O stretch), 1032 (–C–O stretch), 965 (trans-CH@CH– bending
A. Rohman, Y.B. Che Man / Food Chemistry 129 (2011) 583–588
585
Fig. 1. MIR spectra of corn oil (CO), sunflower oil (SFO), and virgin coconut oil (VCO) at MIR region of 4000–650 cm1.
out of plane), 871 (@CH2 wagging), and 722 cm1 (cis-CH@CH– bending out of plane) (Guillen & Cabo, 1997; Lerma-García et al., 2010). Taking into account the spectrum of VCO, CO, and SFO, it can be seen that spectra of CO and SFO revealed some differences to VCO, especially in region around 3008 cm1 and at fingerprint region (1500–650 cm1). There is no band at 3008 cm1 for VCO, and the otherwise was observed for CO and SFO. Furthermore, at spectral regions of 1120–1098 cm1, VCO has one peak; meanwhile CO and SFO reveal two peaks. These differences can be exploited for detection and quantification of CO and SFO as adulterants in VCO. 3.1. Classification DA was used to make the classification between pure VCO and that adulterated with CO and SFO. DA can be exploited to determine the class of VCO to that adulterated with CO and SFO by calculating the distance from each class centre using the Mahalanobis distance units. After the classification model is obtained, the class of unknown samples to that of the definite classes can be predicted (Ballabio & Todeschini, 2009). The classification of VCO and that adulterated with CO was carried out using spectral regions at combined frequencies of 3028– 2983, 2947–1887, and 1685–868 cm1, meanwhile frequencies at 3030–2980 and 1300–1000 cm1 was exploited for classification of VCO adulterated with SFO. The selection of this frequency region was based on its capability to provide the least or no misclassification between two classes (pure VCO and adulterated VCO). The Coomans plot for the classification between pure VCO and VCO adulterated with CO and SFO using 10 principal components is shown in Fig. 2A and B, respectively. It is clear from Fig. 2 that both classes (pure VCO and VCO mixed with CO and SFO) are well separated. DA accurately classifies 100% of all samples according to its classes, meaning that no samples were classified into the wrong group (Tay, Singh, Krishnan, & Gore, 2002). 3.2. Quantification The quantification of CO and SFO as adulterants in VCO was carried out using PLS algorithm. The spectral regions used for PLS calibration models are 858–705, 943–863, 1392–983, and
3027–2983 cm1 for quantitative analysis of CO in VCO, and at 1685–686, 2946–1887, and 3027–2983 cm1 for quantification of SFO in VCO. The selection of these frequency regions was based on the optimisation processes in which they offer the highest values of R2 and the lowest values of error, either in calibration or in prediction models. The five principal components (factors) were sufficient for describing PLS model for both adulterants. Fig. 3 shows the PLS calibration model which correlates the actual and estimated values of CO and SFO (%v/v) obtained from FT-MIR spectra at the specified regions. The difference between the actual and the observed concentration of adulterants is relatively small with coefficient of determination (R2) values are 0.999 for both adulterants (CO and SFO). Root mean square error of calibration (RMSEC) was used to evaluate the error in calibration model. RMSEC value was calculated as follows:
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi Pn 2 i¼1 ðactual calculatedÞ RMSEC ¼ Nf 1 ðParadkar and Irudayaraj; 2002Þ The term ‘‘actual’’ refers to the known or true concentration of selected standards. Meanwhile the ‘‘calculated’’ or ‘‘predicted’’ refers to a value computed by the model using spectral data; where N is the number of samples used in the calibration sets; and f is number of factors used in the calibration model. The low value RMSEC indicates the good performance of PLS model. The RMSEC values of CO and SFO in VCO obtained are 0.866% and 0.374% (v/v), respectively. In order to asses the prediction ability of the developed model, PLS calibration model was used to predict the levels of independent CO and SFO in VCO samples as validation or prediction data sets. The evaluation of the goodness of fit in the validation is performed by calculating the root mean square error of prediction (RMSEP) and R2. RMSEP is calculated using the following equation:
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi Pm 2 i¼1 ðactual calculatedÞ RMSEP ¼ M1 ðParadkar and Irudayaraj; 2002Þ where M is the number of samples used in the prediction sets. The values of R2 are 0.998 for both adulterants (Fig. 4); meanwhile the RMSEP values are 0.994% and 1.060% (v/v), respectively for analysis of CO and SFO as adulterants in VCO samples. The high
586
A. Rohman, Y.B. Che Man / Food Chemistry 129 (2011) 583–588
Fig. 2. The Coomans plot of VCO and adulterants: (h) VCO; (4) VCO containing adulterants. (A) VCO adulterated with CO. (B) VCO adulterated with SFO.
Fig. 3. PLS calibration model for the relationship between actual and estimated concentrations of CO and SFO in VCO. (A) CO in VCO. (B) SFO in VCO.
values of R2 and low values of RMSEP indicate the success of PLS regression model. Table 2 compiled PLS performance in terms of R2, RMSEC, RMSEP, and the number of principal components for
quantification of CO and SFO in VCO. The scatter plot for the relationship between actual and estimated concentration of CO and SFO in validation model is shown in Fig 4.
587
A. Rohman, Y.B. Che Man / Food Chemistry 129 (2011) 583–588
Fig. 4. The relationship between actual and estimated concentrations of CO and SFO in VCO in validation model. (A) CO in VCO. (B) SFO in VCO.
Table 2 PLS performance for analysis of corn and sunflower oils as oil adulterants in VCO. Adulterants
Principal components
Corn oil Sunflower oil
5 5
R2 Calibration
Prediction
0.999 0.999
0.998 0.996
The developed PLS model was further evaluated by cross-validation using ‘‘leave one out’’ technique. In this technique, one of the calibration samples is removed. Subsequently, the removed sample was predicted with the fashioned model using the residual samples and the procedure was repeated until each sample was excluded once from the model (Gurdeniz, Tokatli, & Ozen, 2007). The values of root mean square error of cross validation (RMSECV) obtained are relatively low, i.e. 1.68% and 1.32% (v/v), respectively for CO and SFO. Based on this result, it can be stated that PLS appears to have a reasonable ability to estimate the percentage of CO and SFO as oil adulterants in VCO samples.
RMSEC (%v/v)
RMSEP (%v/v)
0.866 0.374
0.99 1.06
classify VCO and that adulterated with adulterants using 10 principal components. Acknowledgments The first author acknowledges to The Ministry of National Education, Republic of Indonesia for its scholarship to pursue Ph.D. programme in Halal Products Research Institute, Universiti Putra Malaysia (UPM). References
4. Conclusions It can be concluded that FT-MIR spectroscopy using HATR accessory in combination with chemometrics can be used to detect and to quantify the adulteration of virgin coconut oil with corn and sunflower oils. The level of adulterants was successfully determined with the aid of PLS calibration model. DA can correctly
Anklam, E., & Bantaglia, R. (2001). Food analysis and consumer protection. Trends in Food Science and Technology, 12, 71–102. Aparicio, R., & Aparicio-Ruız, R. (2000). Review: Authentication of vegetable oils by chromatographic techniques. Journal of Chromatography A, 881, 93–104. Baeten, V., Pierna, J. A. F., Dardenne, P., Meurens, M., García-González, D. L., & Aparicio-Ruiz, R. (2005). Detection of the presence of hazelnut oil in olive oil by FT-Raman and FT-MIR spectroscopy. Journal of Agricultural and Food Chemistry, 53, 6201–6206.
588
A. Rohman, Y.B. Che Man / Food Chemistry 129 (2011) 583–588
Ballabio, D., & Todeschini, R. (2009). Multivariate classification for qualitative analysis. In D.-W. Sun (Ed.), Infrared spectroscopy for food quality: Analysis and control (pp. 83–104). London: Elsevier. Chiavaro, E., Vittadini, E., Rodriguez-Estrada, M. T., Cerretani, L., & Bendini, A. (2008). Diffrential scanning calorimeter application to the detection of refined hazelnut oil in extra virgin olive oil. Food Chemistry, 110, 248–256. Christy, A. A., Kasemsumran, S., Du, Y., & Ozaki, K. (2004). The detection and quantification of adulteration in olive oil by near-infrared spectroscopy and chemometrics. Analytical Sciences, 20, 935–940. Cserhati, T., Forgacs, E., Deyl, Z., & Miksik, I. (2005). Chromatography in authenticity and traceability tests of vegetable oils and dairy products: A review. Biomedical Chromatography, 19, 183–190. Codex Alimentarius Commision (2003). Amended. Codex Standard for Named vegetable Oils. Codex Stan 210. Gallardo-Velázquez, T., Osorio-Revilla, G., Zuñiga-de Loa, M., & Rivera-Espinoza, Y. (2009). Application of FTIR-HATR spectroscopy and multivariate analysis to the quantification of adulterants in Mexican honeys. Food Research International, 42, 313–318. Guillen, M. D., & Cabo, N. (1997). Characterization of edible oils and lard by Fourier transform infrared spectroscopy. Relationships between composition and frequency of concrete bands in the fingerprint region. Journal of the American Oil Chemists Society, 74, 1281–1286. Gurdeniz, G., Tokatli, F., & Ozen, B. (2007). Differentiation of mixtures of monovarietal olive oils by mid-infrared spectroscopy and chemometrics. European Journal of Lipid Science and Technology, 109, 1194–1202. Gurdeniz, G., & Ozen, B. (2009). Detection of adulteration of extra-virgin olive oil by chemometric analysis of mid-infrared spectral data. Food Chemistry, 116, 519–525. Lerma-García, M. J., Ramis-Ramos, G., Herrero-Martínez, J. M., & Simó-Alfonso, E. F. (2010). Authentication of extra virgin olive oils by Fourier-transform infrared spectroscopy. Food Chemistry, 118, 78–83. Manaf, M. A., Che Man, Y. B., Hamid, N. S. A., Ismail, A., & Syahariza, Z. A. (2007). Analysis of adulteration of virgin coconut oil by palm kernel olein using Fourier transform infrared spectroscopy. Journal of Food Lipids, 14, 111–121. Marina, A. M., Che Man, Y. B., Nazimah, S. A. H., & Amin, I. (2009a). Monitoring the adulteration of virgin coconut oil by selected vegetable oils using differential scanning calorimetry. Journal of Food Lipids, 16, 50–61. Marina, A. M., Che Man, Y. B., & Ismail, A. (2009b). Virgin coconut oil: Emerging functional food oil. Trends in Food Science and Technology, 20, 481–487. Marina, A. M., Che Man, Y. B., & Amin, I. (2010). Use of the SAW sensor electronic nose for detecting the adulteration of virgin coconut oil with RBD palm kernel olein. Journal of the American Oil Chemists Society, 87, 263–270.
Miller, J. N., & Miller, J. C. (2005). Statistics and chemometrics for analytical chemistry (5th Edition). Edinburgh Gate Harlow: Pearson Education Limited. Nik Norulaini, N. A., Setianto, W. B., Zaidul, I. S. M., Nawi, A. H., Azizi, C. Y. M., & Mohd Omar, A. K. (2009). Effects of supercritical carbon dioxide extraction parameters on virgin coconut oil yield and medium-chain triglyceride content. Food Chemistry, 116, 193–197. Paradkar, M. M., & Irudayaraj, J. (2002). A rapid FTIR spectroscopic method for estimation of caffeine in soft drinks and total methylxanthines in tea and coffee. Journal of Food Science, 657, 2507–2511. Pavia, D. L., Lampman, G. M., & Kriz, G. S. (2001). Introduction to spectroscopy (3rd ed.). Australia: Thomson Learning. Reid, L. M., O’Donnell, C. P., & Downey, G. (2006). Recent technological advances for the determination of food authenticity. Trends in Food Science and Technology, 17, 344–353. Roggo, Y., Chalus, P., Maurer, L., Lema-Martinez, C., Edmond, A., & Jent, N. (2007). A review of near infrared spectroscopy and chemometrics in pharmaceutical technologies. Journal of Pharmaceutical and Biomedical Analysis, 44, 683–700. Rohman, A., & Che Man, Y. B. (2009a). Analysis of cod-liver oil adulteration using Fourier transform infrared (FTIR) spectroscopy. Journal of the American Oil Chemists Society, 86, 1149–1153. Rohman, A., & Che Man, Y. B. (2009b). Monitoring of virgin coconut oil (VCO) adulteration with palm oil using Fourier transform infrared spectroscopy. Journal of Food Lipids, 16, 618–628. Rossell, J. B., King, B., & Downes, M. J. (1983). Detection of adulteration. Journal of the American Oil Chemists Society, 60, 333–339. Sinelli, N., Cerretani, L., Di Egidio, V., Bendini, A., & Casiraghi, E. (2009). Application of near (NIR) infrared and mid (MIR) infrared spectroscopy as a rapid tool to classify extra virgin olive oil on the basis of fruity attribute intensity. Food Research International. doi:10.1016/j.foodres.2009.10.008. Tay, A., Singh, R. K., Krishnan, S. S., & Gore, J. P. (2002). Authentication of olive oil adulterated with vegetable oils using Fourier transform infrared spectroscopy. LWT – Food Science and Technology, 35, 99–103. Wang, L., Lee, F. S. C., Wang, X., & He, Y. (2006). Feasibility study of quantifying and discriminating soybean oil adulteration in camellia oils by attenuated total reflectance MIR and fibre optic diffuse reflectance NIR. Food Chemistry, 95, 529–536. Wilson, R. H., & Tapp, H. S. (1999). Mid-infrared spectroscopy for food analysis: Recent new applications and relevant developments in sample presentation methods. Trends in Analytical Chemistry, 18, 85–93. Yap, K. Y.-L., Chan, S. Y., & Lim, C. S. (2007). Infrared-based protocol for the identification and categorisation of ginseng and its products. Food Research International, 40, 643–652.