Int J App Pharm, Vol 10, Issue 5, 2018, 54-59Original Article


APPLICATION OF FTIR SPECTRA COMBINED WITH CHEMOMETRICS FOR ANALYSIS OF CANDLENUT OIL ADULTERATION

FITRI YULIANI1, SUGENG RIYANTO1, ABDUL ROHMAN1,2*

1Faculty of Pharmacy, Gadjah Mada University, Yogyakarta 55281 Indonesia, 2Research Center of Halal Products, Gadjah Mada University, Yogyakarta 55281 Indonesia
Email: abdulkimfar@gmail.com

Received: 09 Jul 2018, Revised and Accepted: 26 Jul 2018

ABSTRACT

Objective: The aim of this study was to use FTIR spectroscopy in combination with chemometrics techniques for quantification and classification of candlenut oil (CnO) from oil adulterants, namely sunflower oil (SFO), soybean oil (SyO), and corn oil (CO).

Methods: The spectra of all samples were scanned using Fourier Transform Infrared (FTIR) Spectrophotometer using attenuated total reflectance (ATR) as sampling technique at mid infrared region (4000-650 cm-1). Multivariate calibrations of principle component regression (PCR) and partial least square regression (PLSR) were used for quantitative models to predict the levels of CnO in the binary mixtures with SFO, SyO, and CO.

Results: The results showed that CnO in SFO was best quantified using PCR at wavenumbers region of 3100-2800 cm-1. Quantitative analysis of CnO in SyO was carried out using PLSR with normal spectra mode using combined wavenumbers of 1765-1625 and 839-663 cm-1, while CnO in CO was analyzed quantitatively using normal spectra at wavenumbers of 970-857 cm-1. The coefficient of determination (R2) obtained were>0.99 with low values of root mean square error of calibration (RMSEC) and root mean square error of prediction (RMSEP). The results of discriminant analysis revealed that authentic CnO can be discriminated from CnO adulterated with SFO, SyO and CO using selected wavenumbers.

Conclusion: FTIR spectroscopy combined with chemometrics could be used as rapid and reliable method for authentication of candlenut oil (CnO) adulterated with other oils.

Keywords: Infrared spectroscopy, Chemometrics, Patchouli oil, Adulteration

© 2018 The Authors. Published by Innovare Academic Sciences Pvt Ltd. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/)
DOI: http://dx.doi.org/10.22159/ijap.2018v10i5.28372

INTRODUCTION

Aleurites moluccana L., widely known as “kemiri” in Indonesia, “buah keras” in Malaysia, “candlenut” in English and “Kukui nut” in Hawaii is a flowering tree with up to 20 m in height, belongs to the family of Euphorbiaceae [1]. This plant is native to the region of Indo-Malaysia and become one of the world’s great domesticated multipurpose trees. Candlenut oil (CnO) obtained from the seeds was taken into account as excellent material in the industries of cosmetics, soap, and pharmaceuticals [2]. It is estimated that the annual production of CnO is approximately 3200 kg/ha [3].

The price of candlenut oil (CnO) in the fats and oils industry, especially in Indonesia, is 10 times higher than that of other common edible oils like sunflower oil (SFO), soybean oil (SyO) and corn oil (CO), therefore, some unethical seller try to blend CnO either unintentionally or intentionally with cheaper oils to get economical profits [4]. Therefore, analytical methods capable of detecting the adulteration practice of high priced oils like CnO must be developed. Several methods based on physico-chemical and molecular biological properties have been reported for authentication of fats and oils. Such methods are differential scanning calorimetry [5], electronic nose [6], high-performance liquid chromatography with several detectors [7, 8], gas chromatography [9], capillary electrophoresis [10], nuclear magnetic resonance [11] and real-time polymerase chain reaction [12]. These methods are complex in instrumentation and need skilled analyst. For this reason, rapid quantitative analysis based on spectroscopic methods such as Fourier transform spectroscopy (FTIR) should be continuously developed.

FTIR spectrophotometry measures the functional group vibrations as a consequence of electromagnetic radiation with sample to generate a spectrum with fingerprint properties [12]. This techniques offer qualitative and quantitative analyses of targeted samples including pharmaceutical products [13, 14].

FTIR spectroscopy, especially in combination with chemometrics techniques, is one of the vibrational spectroscopic methods commonly used for rapid authentication analysis [16]. FTIR spectroscopy coupled with chemometrics have been used for authentication of extra virgin olive oil [17, 18], black seed oil [19], and avocado oil [20, 21], sesame oil adulterated with corn oil [22], and authentication of virgin coconut oil [23]. However, using literature searching there are no publication reports regarding the application of FTIR spectroscopy for authentication of candlenut. Therefore, the objective of this study was to use FTIR spectroscopy in combination with chemometrics for authentication of candlenut oil (CnO) with soybean oil (SyO), sunflower oil (SFO) and corn oil (CO).

MATERIALS AND METHODS

Materials

The seeds of A. moluccana was cut into small to fasten drying process and to increase the yield of oil. A-250 gram of cut seeds was dried into conventional oven at 90 °C for 90 min. The seeds were pressed such a way that its oil was extracted. The candlenut oil obtained was then filtered and centrifuged to separate oil from the impurities. Soybean oil, sunflower oil, corn oil, coconut oil, argon oil, peanut oil, rice bran oil, extra virgin olive oil, palm oil and sesame oil were purchased from supermarket around Yogyakarta, Indonesia. All chemicals and reagents used were of pro-analytical grade purchased from E. Merck (Darmstadt, Germany).

Analysis of fatty acid composition

Fatty acid composition of candlenut oil and oil adulterants (sunflower, soybean and corn oils) were analyzed using gas chromatography with flame ionization detector (GC-FID). Before being injected into GC-FID, the oil samples were subjected to derivatization using sodium methoxide and BF3. The analytical procedure and GC-FID condition for analysis of fatty acid composition can be seen in Che Man et al. [24]. The identification of fatty acid was performed based on retention times of corresponding fatty acids present in samples with those in standards of fatty acid methyl esters from Sigma (Aldrich, USA). Quantification of each fatty acids was carried using internal normalization technique as in Nurrulhidayah et al. [22].

Quantitative analysis of candlenut oil using FTIR spectroscopy

Quantitative analysis of candlenut oil (CnO) in binary mixture with soybean oil (SyO), sunflower oil (SFO) and corn oil (CO) was facilitated with two multivariate calibrations, namely principle component regression (PCR) and partial least square regression (PLSR) by developing prediction models, either in calibration or validation. The calibration samples were prepared by mixing CnO with SFO, SyO and CO in the concentration range of 0-100%. For validation model, another independent samples with similar composition were also prepared. All samples (calibration and validation) were scanned using FTIR spectrophotometer.

Classification of CnO and CnO adulterated with other oils (SFO, SyO and CO)

Qualitative analysis of CnO adulterated with other edible oils (SFO, SyO and CO) was carried out by making a classification between CnO and CnO adulterated with SFO, SyO and CO using discriminant analysis (DA). During DA, a training sets of pure CnO and CnO adulterated with SFO, SyO and CO at concentration range of 1-50% was prepared. All CnO and CnO mixed with SFO, SyO and CO were assigned as “authentic” and “adulterated”, respectively.

FTIR spectra measurement

The measurement of FTIR spectra of samples were done using FTIR spectrophotometer (Nicolet 6700 from Thermo Nicolet Corp., Madison, WI) equipped with detector of deuterated triglycine sulphate (DTGS). This instrument was connected to computer operating systems which include OMNIC operating system (Version 7.0 Thermo Nicolet) software. The sampling compartment was horizontal Attenuated Total Reflectance kit (HATR, Smart ARK, Thermo Electron Corp.) composed of ZnSe crystal. FTIR spectra were scanned at wavenumbers of 4000–650 cm-1, using 32 scans with 4 cm-1 resolution, as absorbance mode.

Data analyses

TQ AnalystTM software version 6 from Thermo electron Corporation (Madison, WI, USA) was used during data analysis, including multivariate calibrations and discriminant analysis. FTIR spectral regions where the variations existed were chosen for analysis.

RESULTS AND DISCUSSION

For selection of edible oils which are potential as adulterants in CnO, principal component analysis (PCA) was used. PCA is one of chemometrics techniques commonly used for the classification of oils with similar chemical properties. PCA was performed using absorbance values at 11 wavenumbers of 3009, 2923, 2853, 1743, 1657, 1463, 1377, 1241, 1159, 1097, 915, and 721 cm-1. Fig. 1 revealed PCA score plot of CnO and other vegetable oils, representing the projection of evaluated samples defined by PC 1 (the first principle component) and PC 2 (the second principle component). PC1 and PC2 showed the most variation among variables used during PCA. Based on PC1 and PC2 scores, soybean oil (SyO), sunflower oil (SFO) and corn oil (CO) have the closer distance to CnO, indicating the close similarity between CnO with SFO, SyO and CO in terms of FTIR spectra, therefore, SFO, SyO and CO were selected as oil adulterants in CnO.

Fig. 1: The PCA sore plots of first principle component (PC1) and second component (PC2) of candle oil (CnO) and other edible oils. SyO = soybean oil; SFO = sunflower oil; CO = corn oil; CoO = coconut oil; ArO = argon oil; PeO = peanut oil; RBO = rice bran oil; OO = olive oil; PaO = palm oil; and SeO = sesame oil

In order to assure that the used oil adulterants were not mixed with other oils, fatty acid composition was used for its characterization. Stearic (C18:0), linoleic (C18:2) and oleic acids were main fatty acids composed CnO accounting of 42.28, 26.32, and 23.12%, respectively. In addition, linoleic, oleic and palmitic acids were dominant fatty acids composed SFO accounting of 67.80, 19.05, and 6.27%, respectively. Linoleic and oleic also main fatty acids composed corn oils, i.e. 59.08 and 25.86%, respectively. SyO was characterized by linolenic acid accounting of 8.05%. These fatty acids are in agreement with those appeared in Norulaini et al. [25] and Che Man et al. [23]. This indicated that the used oils were not added with other oils.

Fig. 2 revealed FTIR spectra candlenut oil, sunflower oil, corn oil and soybean oil scanned using FTIR spectroscopy at mid infrared region (4000-650 cm-1) which were characteristic to triacilglyceride of infrared absoption. The interpretation of each peakas can be seen in Lerma-Garcia et al. [15] and Rohman [13]. The main functional groups were: (a) cis CH=CH stretching vibration appeared in wavenumbers (1/λ) of 3008 cm-1, (b)–CH3 stretching vibration in 1/λ of 2953 cm-1, (c)–CH3 stretching vibration in 1/λ of 2953 cm-1, (d) very intense absorption of carbonyl (C=O) stretching vibration at 1/λ of 1741 cm-1, (e) 1460 cm-1 corresponding to CH2 bending, (f) 1378 cm-1 due to CH3 bending, while (g) at 1235 cm-1, (h) at 1159 cm-1, (i) at 1118 cm-1, (j) at 1098 cm-1 were corresponding to C-O stretching. All these functional groups are present in triglyceridic compounds.

Fig. 2: FTIR spectra of candlenut oil, sunflower oil, corn oil and soybean oil scanned using FTIR spectroscopy at mid infrared region (4000-650 cm-1)

Table 1: The performance of multivariate calibrations along with wavenumbers and FTIR spectral treatments for analysis of candlenut oil (CnO) in the binary mixtures with sunflower oil, soybean oil, and corn oil

Oil adulterans Multivariate calibration Wave-numbers (cm-1) Treatments* Factor Calibration
R2 RMSEC Validation
Sunflower oil PCR 4000-650 Normal 10 0.9984 1.72 R2
Soybean oil PLS 3100-2800 Der. 1 10 0.9985 1.71 0.9982
Der. 2 10 0.9910 4.12 0.9986
Normal 10 0.9990 1.52 0.9904

3100-2800

1500-1000

 Der. 1 10 0.9978 2.78 0.9985
Der. 2 10 0.9961 2.99 0.9992
Normal 10 0.9987 1.54 0.9966

1500-1000

4000-650

 Der. 1 10 0.9986 1.65 0.9975
Der. 2 10 0.9978 2.04 0.9977
Normal 6 0.9989 1.47 0.9981

PLS

PCR

4000-650

3100-2800

 Der. 1 3 0.9995 0.99 0.9976
Der. 2 7 0.9999 0.06 0.9984
Normal 5 0.9989 1.58 0.9926

3100-2800

1500-1000

 Der. 1 3 0.9987 1.71 0.9983
Der. 2 2 0.9964 2.87 0.9987
Normal 6 0.9989 1.47 0.9960

1500-1000

1765-1625; 839-663

 Der. 1 2 0.9974 2.20 0.9976
Der. 2 6 0.9999 0.499 0.9980
Normal 10 0.9994 1.05 0.9978
Corn oil PLS 956-899 Der. 1 10 0.9958 2.77 0.9989
Der. 2 10 0.9705 7.29 0.9970
Normal 10 0.9981 1.86 0.9435

956-899

1765-1625; 839-663

 Der. 1 10 0.9960 2.70 0.9981
Der. 2 10 0.8514 15.9 0.9963
Normal 5 0.9996 0.867 0.8424

PLS

PCR

1765-1625; 839-663

956-899

 Der. 1 5 0.9994 1.02 0.9990
Der. 2 7 0.9986 1.59 0.9968
Normal 2 0.9978 1.99 0.9845

956-899

970-857

 Der. 1 1 0.9953 2.92 0.9979
Der. 2 9 0.9911 4.03 0.9966
Normal 10 0.9968 2.11 0.9682
PLS 970-838; 1125-1111 Der. 1 10 0.9867 4.25 0.9883
Der. 2 10 0.7636 16.9 0.9839
Normal 10 0.9959 2.32 0.6569

970-838; 1125-1111

970-857

 Der. 1 10 0.9899 3.63 0.9946
Der. 2 10 0.9766 5.50 0.9913
Normal 3 0.9959 2.38 0.9692
PLS

970-857

970-838; 1125-1111

 Der. 1 9 0.9992 1.04 0.9958
Der. 2 3 0.9483 8.31 0.9820
Normal 4 0.9939 2.82 0.7683
970-838; 1125-1111 Der. 1 5 0.9975 1.81 0.9927
Der. 2 7 0.9985 1.38 0.9909

*treatments marked with bold are selected for the modelling calibration and validation

Quantitative analysis of candlenut oil

Quantitative analysis of candlenut oil (CnO) in binary mixture with SFO, SyO and CO was facilitated with multivariate calibrations of principle component regression (PCR) and partial least square regression (PLSR). Both regressions are based on inverse regression in which concentrations in y-axis are modelled as function of combination of variables of absorbances called as principle components in x-axis (Miller and Miller, 2010). Using these multivariate calibrations, several statistical parameters namely coefficient determination (R2), root mean square error of calibration (RMSEC) and root mean square error of prediction (RMSEP) were used for evaluation of calibration and validation models. In this study, the models was optimized by selecting the best FTIR spectral regions and spectral treatments (derivatization) capable of providing the highest R2 value and the lowest RMSEC and RMSEP.

Table 1 compiled the performance of PCR and PLSR for quantification of CnO in SFO, CO and SyO in terms of R2, RMSEC, and RMSEP values using several wavenumbers region and spectral treatments (normal, first derivate and second derivate). Based on highest R2 either in calibration and validation samples and lowest RMSEC and RMSEP values obtained, CnO in SFO was quantified using PCR at wavenumbers region of 3100-2800 cm-1 using FTIR normal spectra With R2 calibration of 0.9990, R2 validation of 0.997, RMSEC of 1.52% and RMSEP of 1.56%. In addition, CnO in SyO was analyzed using PLSR with normal spectra in the combined wavenumbers of 1765-1625 and 839-663 cm-1. Finally, the normal spectra at wavenumbers region of 970-857 was selected for quantification of CnO in binary mixture with CO.

Fig. 3: The calibration model describing the relationship between actual values and FTIR predicted values of candlenut oil in not sunflower oil (top), soybean oil (middle), and corn oil (down)

Fig. 4: The calibration model describing the relationship between actual values and FTIR predicted values of candlenut oil in not sunflower oil (top), soybean oil (middle), and corn oil (down)

Fig. 3 and fig. 4 revealed the relationship between actual values and FTIR predicted values of CnO in SFO, SyO and CO along with its equations in calibration and validation models, respectively. The residual point (the difference between actual and predicted) was around zero, indicating that no systematic errors were observed during modelling, therefore the calibration and validation models developed were reliable for quantification of CnO in the mixture mixtures with SFO, SyO and CO.

Discriminant analysis

In order to make classification between authentic CnO and CnO adulterated with other vegetable oils (SFO, SyO and CO), discriminant analysis (DA) was used. DA is one of supervised pattern recognition technique based on linear discriminant function, which is a linear combination of the original measured variables [26]. During classification using DA, the absorbances at selected wavenumbers regions was chosen such a way that they provide clear separation between two-classes of samples. Fig. 5 revealed the Coomans plot for discrimination between CnO and CnO adulterated with SFO (using absorbance at wavenumbers of 3100-2800 cm-1), SyO (at 1765-1625 cm-1 and 839-663 cm-1), and with CO (at wavenumbers of 1125-1111 cm-1). The accuracy levels of 100% were obtained meaning that misclassification did not occur. Sometimes, if the studied oils have close similarity between authentic and adulterated oils, misclassification may occur, and then the model of DA must be re-optimized.

Fig. 5: The Coomans plot for the classification of candlenut oil (□) and candlenut oil adulterated with other oils (∆), namely sunflower oil (top), soybean oil (middle), and corn oil (down)

CONCLUSION

The combination of FTIR spectroscopy with chemometrics of multivariate calibrations (PCR and PLSR) and discriminant analysis using absorbance values as variables at optimized wavenumbers regions are powerful analytical technique for authentication analysis of candlenut oil (CnO) from sunflower oil, soybean oil, and corn oil. Discriminant analysis allow rapid classification between CnO and CnO adulterated with other vegetable oils with accuracy levels of 100%. The developed method was reliable, rapid, and ease in sample preparation, which make it as rapid tools for quality control of edible oils.

ACKNOWLEDGEMENT

The Ministry of Research, Technology and Higher Education, Republic Indonesia are acknowledged for partial financial assistance during this study via Hibah Penelitian Unggulan Perguruan Tinggi (PUPT) 2017.

AUTHORS CONTRIBUTIONS

FY and AR performed research activities and prepared manuscript. SR and AR designed research, analyzed data, and made critical thinking on manuscript.

CONFLICTS OF INTERESTS

All authors have none to declare

REFERENCES

  1. Krisnawati H, Kallio M, Kanninen M. Aleurites moluccana (L.) Willd: Ecology, silviculture and productivity. CIFOR, Bogor; 2011.

  2. Elevitch CR, Manner HI. Traditional tree initiative: species profiles for Pacific islands agroforestry-ecological, economic, and cultural renewal. Agroforestry. Available from: http://www.agroforestry.net/tti/Aleurites-kukui.pdf. [Last accessed on 06 May 2006]

  3. Subroto E, Widjojokusumo E, Veriansyah B, Tjandrawinata RR. Supercritical CO2 extraction of candlenut oil: process optimization using taguchi orthogonal array and physicochemical properties of the oil. J Food Sci Technol 2017;54:1286-92.

  4. Shukla AK, Dixit AK, Singh RP. Detection of adulteration in edible oils. J Oleo Sci 2005;6:317–24.

  5. Indriyani L, Riyanto S, Rohman A. Physico-chemical characterization of avocado (Persea americana Mill.) oil from three indonesian avocado cultivars. Res J Med Plant 2016;10:67-78.

  6. Mansor TST, Che Man YB, Rohman A. Application of fast chromatography and fourier transform infrared spectroscopy for analysis of lard adulteration in virgin coconut oil. Food Anal Methods 2011;4:365-72.

  7. Bajoub A, Medina Rodriguez S, Gomez Romero M, Ajal E, Bagur Gonzalez MG, Fernandez Gutierrez A, et al. Assessing the varietal origin of extra-virgin olive oil using liquid chromatography fingerprints of phenolic compound, data fusion and chemometrics. Food Chem 2017;215:245–55.

  8. Flores G, Ruiz Del Castillo ML, Herraiz M, Blanch GP. Study of the adulteration of olive oil with hazelnut oil by on-line coupled high performance liquid chromatographic and gas chromatographic analysis of filbertone. Food Chem 2006;97:742–9.

  9. Jafari M, Kadivar M, Keramat J. Detection of adulteration in iranian olive oils using instrumental (GC, NMR, DSC) methods. J Am Oil Chem Soc 2009;86:103–10.

  10. Godoy Caballeroa MP, Culzoni MJ, Galeano Diaz T, Acedo Valenzuela MI. Novel combination of non-aqueous capillary electrophoresis and multivariate curve resolution-alternating least squares to determine phenolic acids in virgin olive oil. Anal Chim Acta 2013;763:11–9.

  11. Nurrulhidayah AF, Rohman A, Arief Salleh R, Amin I, Shuhaimi M, Farahwahida MY, et al. Authentication of butter from lard using high-resolution of NMR spectroscopy for halal authentication. Int J Food Prop 2017;20:2147–56.

  12. Kamble VV, Gaikwad NB. Fourier transform infrared spectroscopy spectroscopic studies in Embelia ribes Burm. F.: A vulnerable medicinal plant. Asian J Pharm Clin Res 2016;9:41-7.

  13. Nugrahani I, Dillen N. Rapid assay development of diclofenac sodium coated tablet assay using FTIR compared to HPLC method. Int J Appl Pharm 2018;10:43-50.

  14. Nugrahani I, Khalida FN. Green method for acetaminophen and ibuprofen simultaneous assay in the combination tablet using FTIR. Int J Appl Pharm 2018;10:77-83.

  15. Ramos Gomez S, Busto MD, Albillos SM, Ortega N. Novel qPCR systems for olive (Olea europaea L.) authentication in oils and food. Food Chem 2016;194:447–54.

  16. Rohman A. The use of infrared spectroscopy in combination with chemometrics for quality control and authentication of edible fats and oils: a review. Appl Spectros Rev 2017;52:589-604.

  17. Rohman A, Che Man YB, Mohd Yusof F. The use of FTIR spectroscopy and chemometrics for rapid authentication of extra virgin olive oil. J Am Oil Chem Soc 2014;91:207–13.

  18. Lerma Garcıa MJ, Ramis Ramos G, Herrero Martınez JM, Simo-Alfonso EF. Authentication of extra virgin olive oils by fourier-transform infrared spectroscopy. Food Chem 2010;118:78–83.

  19. Rohman A, Wibowo D, Sudjadi, Lukitaningsih L, Rosman AS. Use of fourier transform infrared spectroscopy in combination with partial least square for authentication of black seed oil. Int J Food Prop 2015;18:775–84.

  20. Lumakso FA, Rohman A, Handoy M, Riyanto S, Mohd Yusuf F. Detection and quantification of soybean and corn oils as adulterants in avocado oil using Fourier transform mid infrared (FT-MIR) spectroscopy aided with multivariate calibration. J Teknologi 2015;77:251–60.

  21. Rohman A, Windarsih A, Riyanto S, Sudjad, Ahmad SAS, Rosman AS, et al. Fourier transform infrared spectroscopy combined with multivariate calibrations for the authentication of avocado oil. Int J Food Prop 2016;19:680–7.

  22. Nurrulhidayah AF, Che Man YB, Rohman A. FTIR spectroscopy combined with chemometric for analysis of sesame oil adulterated with corn oil. Int J Food Prop 2014;17:1275–82.

  23. Manaf MA, Che Man YB, Hamid NSA, Ismail A, Syahariza ZA. Analysis of adulteration of virgin coconut oil by palm kernel olein using fourier transform infrared spectroscopy. J Food Lipids 2007;14:111–21.

  24. Che Man YB, Marina AM, Rohman A, Al-Kahtani HA, Norazura O. A fourier transform infrared spectroscopy method for analysis of palm oil adulterated with lard in pre-fried french fries. Int J Food Prop 2014;17:354–62.

  25. Norulaini N, Budi RS, Omar A, Zaidul MLS, Omar MAK. Major chemical constituents of candlenut oil extract using supercritical carbon dioxide. Malaysian J Pharm Sci 2004;2:61-72.

  26. Che Man YB, Rohman A, Mansor TST. Differentiation of lard from other edible oils by means of fourier transform infrared spectroscopy and chemometrics. J Am Oil Chem Soc 2011;88:187–92.