authentication analysis of red fruit (pandanus conoideus lam) oil using ftir spectroscopy in...

Research Article

Received: 24 August 2010; Revised: 18 November 2010; Accepted: 19 November 2010 Published online in Wiley Online Library: 24 March 2011

(wileyonlinelibrary.com) DOI 10.1002/pca.1304

462

Authentication Analysis of Red Fruit (PandanusConoideus Lam) Oil Using FTIR Spectroscopy inCombination with ChemometricsAbdul Rohman,a,b Yaakob B. Che Mana* and Sugeng Riyantob

ABSTRACT:Introduction – Red fruit (Pandanus conoideus Lam) is endemic plant of Papua, Indonesia and Papua New Guinea. The price ofits oil (red fruit oil, RFO) is 10–15 times higher than that of common vegetable oils; consequently, RFO is subjected toadulteration with lower price oils. Among common vegetable oils, canola oil (CaO) and rice bran oil (RBO) have similar fattyacid profiles to RFO as indicated by the score plot of principal component analysis; therefore, CaO and RBO are potentialadulterants in RFO.Objective – To develop FTIR spectroscopy in combination with chemometrics of partial least square regression (PLSR) anddiscriminant analysis (DA) for authentication of RFO from CaO and RBO.Results – The presence of CaO in RFO was better determined at frequency regions of 1200–1050 cm−1; meanwhile, thecombined frequency ranges of 1207–1078 and 1747–1600 cm−1 were exploited for quantitative analysis of RBO withacceptable values of coefficient of determination (R2) and errors in calibration, prediction and during cross validation. DAbased on Mahalanobis distance was able to discriminate between RFO and RFO adulterated with CaO and RBO.Conclusion – FTIR spectroscopy combined with PLSR and DA can be successfully used for quantification and classification ofoil adulterants in RFO. The developed method is rapid and environmentally friendly and sample preparation is easy.Copyright © 2011 John Wiley & Sons, Ltd.

Keywords: FTIR spectroscopy; partial least square regression; discriminant analysis; authentication

* Correspondence to: Y. B. Che Man, Halal Products Research Institute,Universiti Putra Malaysia, 43400 UPM, Serdang, Selangor, Malaysia. E-mail:[email protected]

a Halal Products Research Institute, Universiti Putra Malaysia, 43400 UPM,Serdang, Selangor, Malaysia

b Department of Pharmaceutical Chemistry, Faculty of Pharmacy, GadjahMada University, Yogyakarta, 55281, Indonesia

IntroductionRed fruit or scientifically named with Pandanus conoideus Lam isan indigenous plant from Papua Province, Indonesia and PapuaNew Guinea. Red fruit is believed to treat several degenerativediseases such as cancer, arteriosclerosis, rheumatoid arthritisand stroke (Budi and Paimin, 2004; Mun'im et al., 2006). Rohmanet al. (2010a) have studied the antioxidant activities of red fruitextracts and its fractions in vitro. The price of red fruit oil (RFO) isapproximately 10–15 times higher than that of commonvegetable oils like palm, corn and soybean oils. This fact canattract unscrupulous market players to adulterate RFO. Theadulteration practice usually involves the substitution or dilutionof high‐price edible oils with the lower price ones (Dourtoglouet al., 2003). Therefore, the authentication of RFO must beaddressed in order to assure its quality and safety.

Our group have explored Fourier transform infrared (FTIR)spectroscopy for authentication of high value edible oils, namelyextra virgin olive oil from palm oil and lard (Rohman andChe Man, 2010; Rohman et al., 2010b), and virgin coconut oilfrom palm kernel oil and palm oil (Manaf et al., 2007; Rohmanand Che Man, 2009). In addition, FTIR spectroscopy was alsoused for monitoring oil parameters such as peroxide value(Setiowaty et al., 2000) and other secondary oxidation products(Mirghani et al., 2002).

This study is the first report on the application of FTIRspectroscopy for authentication of RFO. Partial least squareregression (PLSR) was used to facilitate the quantification of oil

Phytochem. Anal. 2011, 22, 462–467 Copyright © 2011 John

adulterants (CaO and RBO) in RFO. Furthermore, discriminantanalysis (DA) was optimised to classify RFO and that adulteratedwith CaO and RBO.

Materials and Methods

Materials

Red fruit was taken from Papua, Indonesia. Botanical identificationof this fruit was carried out in Department of Biological Pharmacy,Faculty of Pharmacy, Gadjah Mada University, Yogyakarta,Indonesia with supervision by Professor Dr Subagus Wahyuono.The common vegetable oils, including canola (CaO) and rice bran(RBO) oils, were purchased from the local market in Selangor,Malaysia. All chemical and reagents used were of analytical grade.Fatty acid (FA) compositions of RFO and other vegetable oils weredetermined using gas chromatography with flame ionisation

Wiley & Sons, Ltd.

Figure 1. PCA classification of RFO and other vegetable oils. (a) Score plot; (b) loading plot.

Authentication of red fruit oil using FTIR spectroscopy

detector as fatty acid methyl ester (FAME) derivates, according toRohman and Che Man (2009). The separation of FAME was carriedusing capillary column of DB‐5 (0.25mm internal diameter, 30mlength and 0.2μm film thickness) from Restex Corp. (BellefontePA, USA). The initial temperature for column oven was 120 °C(held for 1min), increased into 180 °C (8 °C/min), ramped to 240 °C(10 °C/min), and held at 240 °C for 5min. The temperatures ofdetector and injector were at 240 °C. The flow‐rate of helium ascarrier gas was set at 6.8mL/min. Quantification of FAs wasperformed using the technique of internal normalisation.

46

RFO preparation

Oil from red fruit was obtained using solvent extraction. Briefly,fruits were cut into small pieces using a cutter and subsequentlysubjected to commercial blender containing ethanol (one part offruit was added with one part of ethanol). The ethanolic extract


obtained was further macerated with methanol (1:3 v/v) for4 days. The extract was evaporated at 70 °C and partitioned threetimes using hexane (1:1 extract:hexane, v/v). The hexane extractscontaining RFO were evaporated at 60 °C. The oil obtained wasfurther used for FTIR spectral analysis and fatty acid composition.

Preparation of calibration and prediction samples

Calibration and prediction samples were prepared separately bymixing RFO with CaO and RBO in the concentration ranges of0.5–50.0 (v/v). These sample mixtures were further subjected toFTIR analysis.

Preparation of samples for discriminant analysis

The DA study was done by mixing CaO or RBO with RFO in theconcentration range of 0.5–50.0% (v/v) in chloroform. Pure RFO

Wiley & Sons, Ltd. wileyonlinelibrary.com/journal/pca

3

Table 1. Fatty acid compositions of RFO, CaO and RBOa

Fatty acid RFO CaO RBO

C12:0 0.06 ± 0.00 0.02 ± 0.00 0.01 ± 0.00C14:0 0.07 ± 0.00 0.06 ± 0.00 0.36 ± 0.03C14:1 0.15 ± 0.00 0.10 ± 0.01 0.08 ± 0.01C16:0 20.05 ± 0.09 4.39 ± 0.12 18.95 ± 0.35C16:1 0.15 ± 0.01 0.22 ± 0.02 0.19 ± 0.02C18:0 0.18 ± 0.01 1.19 ± 0.08 1.97 ± 0.20C18:1 68.80 ± 1.29 60.83 ± 0.32 41.57 ± 0.86C18:2 8.49 ± 0.02 19.46 ± 0.22 32.17 ± 0.67C18:3n6 0.17 ± 0.00 8.54 ± 0.17 0.78 ± 0.04C20:0 0.13 ± 0.00 0.08 ± 0.01 1.34 ± 0.09aObtained from at least three replications. The value after “±”is the standard deviation (SD) of measurements.

A. Rohman et al.

464

was assigned as unadulterated; meanwhile RFO containing CaOor RBO was marked as adulterated. Both classes (unadulteratedand adulterated samples) were classified and discriminatedusing DA on the basis of their FTIR spectra.

Measurement of FTIR spectra

FTIR spectra of red fruit oil were measured using FTIRspectrometer Nicolet 6700 (Thermo Nicolet Corp., Madison, WI,USA) equipped with deuterated triglycine sulphate as a detectorand potassium bromide (KBr)/germanium as beam splitter, andconnected to the software of the OMNIC operating system(version 7.0 Thermo Nicolet). The sampling compartment wasSmart Attenuated Total Reflectance kit (Smart ARK, ThermoElectron Corp.) with a dimension of 10 × 60mm. The rest of theprocedure was as described in Rohman and Che Man (2010).

Statistical analysis

Principal component analysis for classification of RFO and otheroils using fatty acid profiles as variable matrix was performed

Figure 2. FTIR spectra of red fruit oil, canola oil and ri

Copyright © 2011 Johnwileyonlinelibrary.com/journal/pca

with the aid of The Unscrambler software from Camo, USA. Inaddition, PLSR and DA were performed using TQ AnalystTM

software from Thermoelectron Corporation included with theFTIR spectrophotometer. The leave‐one‐out cross‐validationprocedure was used to verify the calibration model.

Results and DiscussionMultivariate analysis of principal component analysis (PCA)based on the correlation matrix to elaborate the differentiationbetween RFO and other oils containing different levels of fattyacids (FAs) was carried out. Figure 1(a) exhibited the score plotof PCA for classification of RFO and other common vegetableoils using FA profiles as variables. As shown in Fig. 1, the firsttwo principal components (PCs) described 84 and 13% of thetotal variation, respectively. Therefore, it can be stated thatvariation of 97% could be described by first two PCs. Theremaining PCs accounted for a very small proportion of the totalvariability and were probably unimportant.

RFO was located on the negative side in PC1 and PC2.Similarly, CaO also exhibited the same mode as RFO. RBO hadnegative side on PC1 and positive side on PC2. Both CaO andRBO were closer to RFO, meaning that both oils have moresimilar FA profiles to RFO than other oils. Therefore, both oils arepotential oil adulterants in RFO. Using a loading plot as shown inFig. 1b, it can be seen that oleic acid (C18:1) has more influenceon RFO classification than other fatty acids. The FAs compositionof RFO, CaO and RBO is compiled in Table 1.

Quantification of CaO and RBO in RFO using FTIRspectroscopy and PLSR

FTIR spectroscopy can be used as potential means forquantitative analysis of edible oils, because FTIR spectra aretaken into account as a “fingerprinting tool”, meaning that thereare no two oil samples with the same FTIR spectra, either in thenumber of peaks/shoulders or in the intensities (absorbances) ofthe maximum peak (Guillen and Cabo, 1997). Figure 2 exhibitsthe FTIR spectra of red fruit oil, canola oil and rice bran oil in

ce bran oil scanned at frequency of 4000–650 cm−1.

Phytochem. Anal. 2011, 22, 462–467Wiley & Sons, Ltd.

Figure 3. PLS calibration model for the relationship between actual and FTIR measured values. (a) CaO and (b) RBO.

Table 2. The PRESS and RMSECV values obtained duringcross validation

Factor PRESS RMSECV

0 3029.851 15.889861 350.3785 5.403542 147.187 3.502233 342.5442 5.342794 62.53543 2.282825 90.20225 2.741696 89.82717 2.735987 87.74183 2.704048 88.47105 2.715259 88.47268 2.7152810 88.52979 2.71615


46

absorbance mode scanned at region range of 4000–650 cm−1.The FTIR spectra of RFO and oil adulterants (CaO and RBO) appearquite similar. Using detailed investigation, the FTIR spectra ofRFO exhibited some differences compared with other vegetableoils. In the region 1800–1700 cm−1, RFO revealed two sharppeaks at 1744 and 1710 cm−1, while CaO and RBO exhibited onepeak at 1744 cm−1. These peaks are attributed to carbonyl (C =O)stretching vibration. In addition, some frequency regions alsorevealed different peak intensities, especially at 1118, 1095 and942 cm−1. The functional groups responsible for absorption ofthe common edible fats and oils have been reported elsewhere(Guillen and Cabo, 1997; Lerma‐García et al., 2010). Thesefrequency regions were further optimised in order to quantifythe level of adulterants (CaO and RBO) in RFO with the aid ofPLSR.

Using the optimisation process in terms of frequency regionoffering the high correlation between actual value and FTIRmeasured values of CaO and RBO in RFO and a low level oferrors, the frequency region of 1200–1050 cm−1 was used forquantification of CaO. Figure 3(a) shows the PLSR model forcorrelation between actual and measured values of CaO in RFO.The coefficient of determination (R2) value obtained was high,i.e. 0.999, with an intercept of −0.081. The error during


calibration as expressed as root mean square error of calibration(RMSEC) was 0.812% (v/v).The PLSR model was further used to predict the level of CaO

using laboratory prepared samples. The R2 value for correlationbetween actual and measured values of CaO in prediction


5

Table 3. Analytical figure of merit using PLSR for quantifica-tion of RBO in RFO

Parameter Value

Number of PCs 2R2calibration 0.996R2prediction 0.991R2cross validation 0.990RMSEC 0.957% (v/v)RMSEP 0.914% (v/v)RMSECV 1.91% (v/v)

A. Rohman et al.

466

models is 0.996 with a root mean square error of prediction(RMSEP) of 1.05% (v/v). In order to validate the performance ofthe PLSR model, leave‐one‐out‐technique was used. The rootmean square error of cross validation (RMSECV) obtained was2.28% with an R2 of 0.990. The number of principal components(PCs) suggested to be used in PLSR model, as obtained duringcross validation, was 4 because this number offered theminimum level of predicted residual error sum square (PRESS).Table 2 shows the relationship between PRESS values and

Figure 4. The Coomans plot for classification of

Copyright © 2011 Johnwileyonlinelibrary.com/journal/pca

RMSECV values using different numbers of PCs or factors duringcross‐validation for quantification of CaO in RFO.

The presence of RBO in RFO was better determined at thecombined frequencies of two region ranges (1207–1078 and1747–1600 cm−1). The equation obtained for correlation be-tween actual and measured values of RBO in RFO is shown inFig. 3(b). The analytical figures of merit of PLSR are compiled inTable 3. These results show that FTIR spectroscopy coupled withPLSR can offer an accurate and reliable method for quantifica-tion of CaO and RBO in RFO.

Discriminant analysis

DA is a chemometric technique with supervised patternrecognition type which allocates a new object of unknown groupto the correct group (Miller andMiller, 2005). Figure 4 exhibits theCoomans plot for discrimination of RFO and RFO adulteratedwithCaO and RBO in the concentration of 1.0–50% (v/v) of CaO (a) andRBO (b) based on the Mahalanobis distance.

As shown in Fig. 4, DA using frequency regions of 1200–1050 cm−1 can discriminate pure RFO samples and thoseadulterated with CaO and RBO with no misclassificationreported during the study. Compared with RBO, RFO has acloser Mahalanobis disctance to CaO. This supports that CaO is

RFO and RFO containing CaO (a) and RBO (b).

Phytochem. Anal. 2011, 22, 462–467Wiley & Sons, Ltd.


more similar to RFO than RBO in the fatty acid profile and FTIRspectra. This result suggests that FTIR spectroscopy combinedwith DA is an effective means of discriminating oil samples.

Acknowledgements

This study was jointly supported by The National EducationMinistry through the Research Competitive Grant XVII awarded toProfessor Sugeng Riyanto and by Ministry of Science, Technologyand Innovation,Malaysia awarded to Professor Yaakob B. CheManthrough Science Fund grant no. 05‐01‐04‐SF0285.

ReferencesBudi M, Paimin FR. 2004. Red Fruit (Pandanus conoideus Lam). Jakarta:

Penebar Swadaya; 3–26.Dourtoglou VG, Dourtoglou Th, Antonopoulos A, Stefanou E, Lalas S,

Poulos C. 2003. Detection of olive oil adulteration using principalcomponent analysis on total and region FA content. J Am Oil ChemSoc 80: 203–208.

Guillen MD, Cabo N. 1997. Characterisation of edible oils and lard byFourier transform infrared spectroscopy. Relationships betweencomposition and frequency of concrete bands in the fingerprintregion. J Am Oil Chem Soc 74: 1281–1286.

Lerma‐García MJ, Ramis‐Ramos G, Herrero‐Martínez JM, Simó‐Alfonso EF.2010. Authentication of extra virgin olive oils by Fourier‐transforminfrared spectroscopy. Food Chem 118: 78–83.


Manaf MA, Che Man YB, Hamid NSA, Ismail A, Syahariza ZA. 2007. Analysisof adulteration of virgin coconut oil by palm kernel olein using Fouriertransform infrared spectroscopy. J Food Lipids 14: 111–121.

Miller JN, Miller JC. 2005. Statistics and Chemometrics for AnalyticalChemistry, 5th edn. Harlow: Pearson Education.

Mirghani MES, Che Man YB, Jinap S, Baharin BS, Bakar J. 2002. Rapidmethod for determining malondialdehyde as secondary oxidationproduct in palm olein system by Fourier transform infraredspectroscopy. Phytochem Anal 13: 195–201.

Mun'im A, Andrajati R, Susilowati H. 2006. Tumorigenesis inhibition ofwater extract of red fruit (Pandanus conoideus Lam.) on Sprague–Dawley rat female induced by 7,12‐dimetilbenz(a)antrasen (DMBA).Indon J Pharm Sci 3: 153–161.

Rohman A, Che Man YB. 2009. Monitoring of virgin coconut oil (VCO)adulteration with palm oil using Fourier transform infrared (FTIR)spectroscopy. J Food Lipids 16: 618–628.

Rohman A, Che Man YB. 2010. Fourier transform infrared (FTIR)spectroscopy for analysis of extra virgin olive oil adulterated withpalm oil. Food Res Int 43: 886–892.

Rohman A, Riyanto S, Yuniarti N, Saputra WR, Utami R, Mulatsih, W.2010a. Antioxidant activity, total phenolic, and total flavonoid ofextracts and fractions of red fruit (Pandanus conoideus Lam). Int FoodRes J 17: 97–106.

Rohman A, Che Man YB, Ismail A, Puziah H. 2010b. FTIR spectroscopycombined with chemometrics for analysis of lard adulteration insome vegetable oils. Cyta J Food; doi: 10.1080/19476331003774639.

Setiowaty G, Che Man, YB, Jinap S, Moh MH. 2000. Quantitativedetermination of peroxide value in thermally oxidized palm oleinby Fourier transform infrared spectroscopy. Phytochem Anal 11:74–78.


467

authentication analysis of red fruit (pandanus conoideus lam) oil using ftir spectroscopy in...

Documents