Combining FTIR spectroscopy with multivariate chemoinformatic analysis for rapid detection of adulteration in coconut oil
Coconut oil (CNO) is one of the most widely utilised and valued plant-based products for numerous domestic, cosmetic, and commercial applications since ancient civilizations. Pure oils are frequently blended with lower grade, cheaper alternatives like palm kernel oils (RPKO) to increase profits when demand rises. These adulterated oils, on the one hand reduce the shelf life of the pure coconut oil and on the other might be hazardous and cause health problems. Therefore, it's critical to develop techniques that are rapid and accurate in distinguishing pure and adulterated coconut oils. For oil adulteration analysis, chemoinformatic techniques are commonly coupled with several other analytical methods such as spectroscopy or chromatography. In order to investigate coconut oil adulteration, we employed Fourier transform infrared (FTIR) spectroscopy in conjunction with multivariate statistical analysis to detect and quantify various concentrations of RPKO (1-100 % v/v) in CNO. The entire spectrums were first examined, and it was observed that CNO and RPKO differed primarily in the CH stretching area between 2800 and 3020 cm-1. This variation was also seen to change with increase in concentration of RPKO in CNO. For the chemoinformatic analysis, the offset for the selected spectral region was eliminated, and the spectra was normalised with the intensity of C=O was used as an internal standard. Following that, an multivariant curve resolution (MCR)-ALS calibration plot was created, and it was discovered that the plot was not very consistence with the RPKO concentration. The cis-double bond stretch (3000 and 3020 cm-1) in the chosen area, on the other hand, was shown to be more compatible with the MCR-ALS study. And this cis-double bond stretch was identified as region of interest, and calibration curves were produced to find the best-fitting model. The prediction effectiveness of the model was reported to be close to 1, with the detectable limit predicted up to 1% v/v. Further, various other multivariate statistical methods including (PCR and PLSR) are being explored. The results from these latter methods will be reported during the conference.
File with abstract
Department of Biophysics, Manipal School of Life Sciences, Manipal Academy of Higher Education, Manipal
File with report