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 detectionand 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, Indonesiab Halal Products Research Institute, Universiti Putra Malaysia, 43400 UPM, Serdang, Selangor, Malaysiac Department of Pharmaceutical Chemistry, Faculty of Pharmacy, Gadjah Mada University, Yogyakarta 55281, Indonesiad Department of Food Technology, Faculty of Food Science and Technology, Universiti Putra Malaysia, 43400 UPM, Serdang, Selangor, Malaysia

a r t i c l e i n f o a b s t r a c t

Article history:Received 2 March 2010Received in revised form 27 February 2011Accepted 22 April 2011Available online 29 April 2011

Keywords:FT-MIR spectroscopyVirgin coconut oilCorn oilSunflower oilAdulteration

0308-8146/$ - see front matter � 2011 Elsevier Ltd. Adoi:10.1016/j.foodchem.2011.04.070

⇑ Corresponding author at: Halal Products ResearMalaysia, 43400 UPM, Serdang, Selangor, Malaysia. Te89439745.

E-mail address: yaakobcm@gmail.com (Y.B. Che M

Currently, the authentication of virgin coconut oil (VCO) has become very important due to the possibleadulteration of VCO with cheaper plant oils such as corn (CO) and sunflower (SFO) oils. Methods involvingFourier transform mid infrared (FT-MIR) spectroscopy combined with chemometrics techniques (partialleast square (PLS) and discriminant analysis (DA)) were developed for quantification and classification ofCO and SFO in VCO. MIR spectra of oil samples were recorded at frequency regions of 4000–650 cm�1 onhorizontal attenuated total reflectance (HATR) attachment of FTIR. DA can successfully classify VCO andthat adulterated with CO and SFO using 10 principal components. Furthermore, PLS model correlates theactual and FTIR estimated values of oil adulterants (CO and SFO) with coefficient of determination (R2) of0.999.

� 2011 Elsevier Ltd. All rights reserved.

1. Introduction

Since very early time, fats and oils have been liable to adultera-tion, either intentionally or accidentally (Rossell, King, & Downes,1983). The detection of adulteration is an attractive issue forresearchers, because food producers do not wish to be subjectedto unfair competition from devious processors who would get eco-nomical profit (Gallardo-Velázquez, Osorio-Revilla, Zuñiga-de Loa,& Rivera-Espinoza, 2009). The detection of adulteration is more dif-ficult, especially when the adulterant has similar chemical compo-sition to that of the original oil (Anklam & Bantaglia, 2001).

Virgin coconut oil (VCO) is obtained from the flesh coconut inwhich the oil extraction does not involve the use of thermal orchemical treatments (Nik Norulaini et al., 2009). VCO is an emerg-ing functional food oils due to its ability to posses several biologicalactivities such as antiviral and antimicrobial (Marina, Che Man, &Ismail, 2009). In the market, it is estimated that the price of VCOis approximately 10–20 times higher than that of common plantoils like corn, palm, and sunflower oils. Therefore, VCO is a targetto adulteration practise with the low price plant oils.

Several analytical methods have been developed for detectionand quantification of adulterants in fats and oils such as

ll rights reserved.

ch Institute, Universiti Putral.: +60 3 89430405; fax: +60 3

an).

differential scanning calorimetry (DSC) (Chiavaro, Vittadini,Rodriguez-Estrada, Cerretani, & Bendini, 2008); spectroscopicbased-methods (Lerma-García, Ramis-Ramos, Herrero-Martínez,& Simó-Alfonso, 2010), and wet chemical methods. Quantificationof adulterants in fats and oils by chromatographic method wasreviewed by Cserhati, Forgacs, Deyl, and Miksik (2005) and byAparicio and Aparicio-Ruiz (2000). For the analysis of VCO adulte-ration, Marina, Che Man, and Amin (2010) have developed elec-tronic nose based on surface acoustic wave in combination withthe chemometrics of principal component analysis to monitorthe adulteration of VCO with palm kernel oil. Another techniqueused was DSC for monitoring the presence of palm kernel oil andsoybean oil (Marina, Che Man, Nazimah, & Amin, 2009). Some ofthese methods are impractical and too laborious. Therefore, rapidand accurate analytical methods must be developed in order todetect and to quantify the oil adulterants.

Over the last 3 decades, mid infrared (MIR) spectroscopy com-bined with chemometric methods have been used in numerousanalytical applications (Sinelli, Cerretani, Di Egidio, Bendini, &Casiraghi, 2009). MIR spectroscopy has been identified as an idealanalytical method for authenticity studies of edible fats and oils(Reid, O’Donnell, & Downey, 2006) due to its capability to serveas ‘‘fingerprint’’ technique, meaning that there are no two sampleswith the same FTIR spectra, either in the number of peaks or in themaximum peak intensities (Pavia, Lampman, & Kriz, 2001). Themethods allow sensitive, fast, and reliable technique, ease insample presentation, and can be used for monitoring the quality

Table 1The composition percentage for calibration and validation sets used in the binarymixtures of CO and SFO with VCO.

Samples CO in VCO SFO in VCO

Calibration Validation Calibration Validation

CO VCO CO VCO SFO VCO SFO VCO

1 2.0 98.0 0.0 100.0 0.0 100.0 0.0 100.02 3.0 97.0 1.5 98.5 1.5 98.5 2.0 98.03 5.0 95.0 2.5 97.5 3.0 97.0 3.0 97.04 6.0 94.0 7.5 92.5 4.0 96.0 4.0 96.05 8.0 92.0 8.0 92.0 5.0 95.0 5.0 95.06 9.0 91.0 10.0 90.0 6.0 94.0 6.0 94.07 10.0 90.0 14.0 86.0 7.5 92.5 7.5 92.58 12.5 87.5 15.0 85.0 10.0 90.0 10.0 90.09 15.0 85.0 20.0 80.0 12.5 87.5 12.5 87.510 17.5 82.5 22.5 77.5 15.0 85.0 15.0 85.011 22.5 77.5 25.0 75.0 17.5 82.5 17.5 82.512 25.0 75.0 30.0 70.0 20.0 80.0 20.0 80.013 27.5 72.5 32.5 67.5 22.5 77.5 22.5 77.514 30.0 70.0 37.5 62.5 25.0 75.0 25.0 75.0

584 A. Rohman, Y.B. Che Man / Food Chemistry 129 (2011) 583–588

aspects of fats and oils at spectral region of 4000–650 cm�1 (Roggoet al., 2007; Wilson & Tapp, 1999).

FT-MIR spectroscopic methods were developed for detectionand quantification of oil adulterants such as sunflower, corn, soy-bean and hazelnut oils in extra virgin olive oil (EVOO) usingchemometrics of multiple linear regression and linear discriminantanalysis (Lerma-García et al., 2010), sunflower and corn oils inEVOO with aid of principal component analysis and PLS-discrimi-nant analysis (Gurdeniz & Ozen, 2009), soybean oil in camelliaoil (Wang, Lee, Wang, & He, 2006), hazelmut oil in refined oliveoil (Baeten et al., 2005) and animal fats in cod-liver oil (Rohman& Che Man, 2009a). Our group has used FT-MIR spectroscopy com-bined with chemometrics of PLS and DA for quantification andclassification of palm kernel oil (Manaf, Che Man, Hamid, Ismail,& Syahariza, 2007) and palm oil (Rohman & Che Man, 2009b) asoil adulterants in VCO. The present study highlights the applicationof FT-MIR spectroscopy for detecting and quantifying corn oil (CO)and sunflower oil (SFO) as oil adulterants in VCO.

15 32.5 67.5 47.5 52.5 30.0 70.0 30.0 70.016 35.0 65.0 50.0 50.0 35.0 65.0 35.0 65.017 50.0 50.0 55.0 45.0 40.0 60.0 37.5 62.518 60.0 40.0 60.0 40.0 50.0 50.0 40.0 60.019 70.0 30.0 65.0 35.0 75.0 25.0 45.0 55.020 100.0 0.0 80.0 20.0 100.0 0.0 50.0 50.0

2. Materials and methods

The samples of virgin coconut oil (VCO), sunflower oil (SFO),and corn oil (CO) were purchased from the local market inJogjakarta, Indonesia. The used plant oils were coming from themixture of three different brands with similar fatty acid (FA)composition. The FA profiles of these oils are in accordance withthose listed in Codex Alimentarius (2003). Some factors contrib-uted to the slight FTIR spectral variation of VCO, such as origin ofplants, 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 (COand SFO) was carried out using discriminant analysis (DA). DA isone of the supervised pattern recognition techniques which startwith a number of samples whose group membership is known.These samples are sometimes called the learning or training sam-ples (Miller & Miller, 2005). VCO and adulterants were blended inorder to obtain a series of training sets of pure VCO (50% VCO inchloroform) and VCO containing 1–50% of adulterants in chloro-form. VCO samples mixed with adulterants were marked as ‘‘adul-terated’’, while a series of pure VCO was assigned with ‘‘VCO’’. Bothclasses were classified using DA based on their FT-MIR spectra.

2.2. Quantification

For quantitative analysis, CO and SFO were mixed as binarymixture with VCO (each comprises 20 samples for calibrationand 20 samples for validation). The concentration of each oils usedin both calibration and validation is presented in Table 1. Eachsample was subjected for FT-MIR analysis.

2.3. FT-MIR analysis

FT-MIR spectra of samples were obtained using Nicolet 6700FTIR spectrometer (Thermo Nicolet Corp., Madison, WI) with HATRcrystal of ZnSe 45� equipped with deuterated triglycine sulphate(DTGS) as detector, potassium bromide (KBr) as beam splitterand controlled with the Omnic Software (Version 7.0 Thermo Nico-let). The measurements were directly carried out by putting oilsamples on HATR surface at controlled room temperature (20 �C)in MIR region of 4000–650 cm�1, by accumulating 32 scans withthe resolution of 4 cm�1. These spectra were subtracted from refer-ence spectrum of air, acquired by collecting a spectrum from the

cleaned blank HATR crystal before the measurement of each oilsample replication. The sample spectra were collected in triplicateand displayed as the average spectra. At the end of every scan, thesurface of HATR crystal was cleaned with hexane twice and driedwith special soft tissue, cleaned with acetone, and finally driedwith soft tissue following the collection of each spectrum.

2.4. Chemometrics

The chemometrics analyses were performed using the softwareTQ 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 par-tial least square (PLS), respectively. Frequency regions for PLS andDA were automatically selected by the software and were con-firmed 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 COand SFO in independent samples in order to evaluate its predictivecapability.

3. Results and discussion

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 fatsand oils physically (Christy, Kasemsumran, Du, & Ozaki, 2004).However, because of its capability as a fingerprint technique, MIRspectroscopy allows one to differentiate authentic oils and thoseadulterated with others by observing the spectra changes due tothe adulteration (Yap, Chan, & Lim, 2007).

Fig. 1 exhibits MIR spectra of VCO, CO, and SFO at frequency re-gion of 4000–650 cm�1. The assignment of functional groupsresponsible for IR absorption is as follows: 3008 cm�1 (trans @C–Hstretch), 2954 (–CH3 asymmetrical stretch), 2922 and 2853 (sym-metrical and asymmetrical stretching of –CH2), 1743 (–C@Ostretch), 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

Fig. 1. MIR spectra of corn oil (CO), sunflower oil (SFO), and virgin coconut oil (VCO) at MIR region of 4000–650 cm�1.

A. Rohman, Y.B. Che Man / Food Chemistry 129 (2011) 583–588 585

out of plane), 871 (@CH2 wagging), and 722 cm�1 (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 beseen that spectra of CO and SFO revealed some differences to VCO,especially in region around 3008 cm�1 and at fingerprint region(1500–650 cm�1). There is no band at 3008 cm�1 for VCO, andthe otherwise was observed for CO and SFO. Furthermore, at spec-tral regions of 1120–1098 cm�1, VCO has one peak; meanwhile COand SFO reveal two peaks. These differences can be exploited fordetection and quantification of CO and SFO as adulterants in VCO.

3.1. Classification

DA was used to make the classification between pure VCO andthat adulterated with CO and SFO. DA can be exploited to deter-mine the class of VCO to that adulterated with CO and SFO by cal-culating the distance from each class centre using the Mahalanobisdistance units. After the classification model is obtained, the classof unknown samples to that of the definite classes can be predicted(Ballabio & Todeschini, 2009).

The classification of VCO and that adulterated with CO was car-ried out using spectral regions at combined frequencies of 3028–2983, 2947–1887, and 1685–868 cm�1, meanwhile frequencies at3030–2980 and 1300–1000 cm�1 was exploited for classificationof VCO adulterated with SFO. The selection of this frequency regionwas based on its capability to provide the least or no misclassifi-cation between two classes (pure VCO and adulterated VCO). TheCoomans plot for the classification between pure VCO and VCOadulterated with CO and SFO using 10 principal components isshown in Fig. 2A and B, respectively. It is clear from Fig. 2 that bothclasses (pure VCO and VCO mixed with CO and SFO) are well sep-arated. DA accurately classifies 100% of all samples according to itsclasses, meaning that no samples were classified into the wronggroup (Tay, Singh, Krishnan, & Gore, 2002).

3.2. Quantification

The quantification of CO and SFO as adulterants in VCO wascarried out using PLS algorithm. The spectral regions used for PLScalibration models are 858–705, 943–863, 1392–983, and

3027–2983 cm�1 for quantitative analysis of CO in VCO, and at1685–686, 2946–1887, and 3027–2983 cm�1 for quantification ofSFO in VCO. The selection of these frequency regions was basedon the optimisation processes in which they offer the highest val-ues of R2 and the lowest values of error, either in calibration or inprediction models. The five principal components (factors) weresufficient for describing PLS model for both adulterants.

Fig. 3 shows the PLS calibration model which correlates theactual and estimated values of CO and SFO (%v/v) obtained fromFT-MIR spectra at the specified regions. The difference betweenthe actual and the observed concentration of adulterants isrelatively small with coefficient of determination (R2) values are0.999 for both adulterants (CO and SFO). Root mean square errorof calibration (RMSEC) was used to evaluate the error in calibrationmodel. RMSEC value was calculated as follows:

RMSEC ¼

ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiPni¼1ðactual� calculatedÞ2

N � f � 1

s

� ðParadkar and Irudayaraj; 2002Þ

The term ‘‘actual’’ refers to the known or true concentration ofselected standards. Meanwhile the ‘‘calculated’’ or ‘‘predicted’’ refersto a value computed by the model using spectral data; where N isthe number of samples used in the calibration sets; and f is numberof factors used in the calibration model. The low value RMSEC indi-cates the good performance of PLS model. The RMSEC values of COand 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 indepen-dent CO and SFO in VCO samples as validation or prediction datasets. The evaluation of the goodness of fit in the validation is per-formed by calculating the root mean square error of prediction(RMSEP) and R2. RMSEP is calculated using the following equation:

RMSEP ¼

ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiPmi¼1ðactual� calculatedÞ2

M � 1

s

� 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); mean-

while the RMSEP values are 0.994% and 1.060% (v/v), respectivelyfor analysis of CO and SFO as adulterants in VCO samples. The high

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.

586 A. Rohman, Y.B. Che Man / Food Chemistry 129 (2011) 583–588

values of R2 and low values of RMSEP indicate the success of PLSregression model. Table 2 compiled PLS performance in terms ofR2, RMSEC, RMSEP, and the number of principal components for

quantification of CO and SFO in VCO. The scatter plot for the rela-tionship between actual and estimated concentration of CO andSFO in validation model is shown in Fig 4.

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 2PLS performance for analysis of corn and sunflower oils as oil adulterants in VCO.

Adulterants Principal components R2 RMSEC (%v/v) RMSEP (%v/v)

Calibration Prediction

Corn oil 5 0.999 0.998 0.866 0.99Sunflower oil 5 0.999 0.996 0.374 1.06

A. Rohman, Y.B. Che Man / Food Chemistry 129 (2011) 583–588 587

The developed PLS model was further evaluated by cross-vali-dation using ‘‘leave one out’’ technique. In this technique, one ofthe calibration samples is removed. Subsequently, the removedsample was predicted with the fashioned model using the residualsamples and the procedure was repeated until each sample wasexcluded 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), respectivelyfor CO and SFO. Based on this result, it can be stated that PLS ap-pears to have a reasonable ability to estimate the percentage ofCO and SFO as oil adulterants in VCO samples.

4. Conclusions

It can be concluded that FT-MIR spectroscopy using HATRaccessory in combination with chemometrics can be used to detectand to quantify the adulteration of virgin coconut oil with corn andsunflower oils. The level of adulterants was successfullydetermined with the aid of PLS calibration model. DA can correctly

classify VCO and that adulterated with adulterants using 10 princi-pal components.

Acknowledgments

The first author acknowledges to The Ministry of National Edu-cation, Republic of Indonesia for its scholarship to pursue Ph.D.programme in Halal Products Research Institute, Universiti PutraMalaysia (UPM).

References

Anklam, E., & Bantaglia, R. (2001). Food analysis and consumer protection. Trends inFood Science and Technology, 12, 71–102.

Aparicio, R., & Aparicio-Ruız, R. (2000). Review: Authentication of vegetable oils bychromatographic 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 byFT-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 qualitativeanalysis. In D.-W. Sun (Ed.), Infrared spectroscopy for food quality: Analysis andcontrol (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 refinedhazelnut oil in extra virgin olive oil. Food Chemistry, 110, 248–256.

Christy, A. A., Kasemsumran, S., Du, Y., & Ozaki, K. (2004). The detection andquantification of adulteration in olive oil by near-infrared spectroscopy andchemometrics. Analytical Sciences, 20, 935–940.

Cserhati, T., Forgacs, E., Deyl, Z., & Miksik, I. (2005). Chromatography in authenticityand traceability tests of vegetable oils and dairy products: A review. BiomedicalChromatography, 19, 183–190.

Codex Alimentarius Commision (2003). Amended. Codex Standard for Namedvegetable 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 thequantification of adulterants in Mexican honeys. Food Research International, 42,313–318.

Guillen, M. D., & Cabo, N. (1997). Characterization of edible oils and lard by Fouriertransform infrared spectroscopy. Relationships between composition andfrequency of concrete bands in the fingerprint region. Journal of the AmericanOil Chemists Society, 74, 1281–1286.

Gurdeniz, G., Tokatli, F., & Ozen, B. (2007). Differentiation of mixtures ofmonovarietal 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 bychemometric 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 infraredspectroscopy. 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 Fouriertransform infrared spectroscopy. Journal of Food Lipids, 14, 111–121.

Marina, A. M., Che Man, Y. B., Nazimah, S. A. H., & Amin, I. (2009a). Monitoring theadulteration of virgin coconut oil by selected vegetable oils using differentialscanning calorimetry. Journal of Food Lipids, 16, 50–61.

Marina, A. M., Che Man, Y. B., & Ismail, A. (2009b). Virgin coconut oil: Emergingfunctional 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 electronicnose for detecting the adulteration of virgin coconut oil with RBD palm kernelolein. 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 extractionparameters 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 forestimation 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 (3rded.). Australia: Thomson Learning.

Reid, L. M., O’Donnell, C. P., & Downey, G. (2006). Recent technological advances forthe 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). Areview of near infrared spectroscopy and chemometrics in pharmaceuticaltechnologies. Journal of Pharmaceutical and Biomedical Analysis, 44, 683–700.

Rohman, A., & Che Man, Y. B. (2009a). Analysis of cod-liver oil adulteration usingFourier transform infrared (FTIR) spectroscopy. Journal of the American OilChemists 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 theAmerican Oil Chemists Society, 60, 333–339.

Sinelli, N., Cerretani, L., Di Egidio, V., Bendini, A., & Casiraghi, E. (2009). Applicationof near (NIR) infrared and mid (MIR) infrared spectroscopy as a rapid tool toclassify extra virgin olive oil on the basis of fruity attribute intensity. FoodResearch International. doi:10.1016/j.foodres.2009.10.008.

Tay, A., Singh, R. K., Krishnan, S. S., & Gore, J. P. (2002). Authentication of olive oiladulterated 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 anddiscriminating soybean oil adulteration in camellia oils by attenuated totalreflectance 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 presentationmethods. Trends in Analytical Chemistry, 18, 85–93.

Yap, K. Y.-L., Chan, S. Y., & Lim, C. S. (2007). Infrared-based protocol for theidentification and categorisation of ginseng and its products. Food ResearchInternational, 40, 643–652.