An Exploratory Study of Differentiating Indian Kerosene Based on ATR-FTIR Spectroscopy and Chemometric Methods

Despite the widespread attention given to discrimination of fuel, differentiating Indian kerosene by the scheme origin has yet to be reported. The scheme origin predominantly governs the selling price of Indian kerosene. Subsidized kerosene, which is from the Public Distribution Service (PDS), is often illegally sold as a nonPDS scheme at a higher price. The current study explores the differences between PDS and nonPDS Indian kerosene based on their infrared fingerprints acquired by attenuated total reflectance-Fourier transform infrared (ATR-FTIR) spectroscopy. A total of 19 kerosene samples, composing PDS and nonPDS schemes, were purchased from various geographical locations in India. The global IR region was first truncated into three sub-IR regions containing various clustered peaks. Then, the baseline of the sub-IR regions and the global counterpart were corrected via the asymmetric least squares algorithm, followed by normalization to sum. By this, 12 sub-datasets were prepared for discriminating the kerosene samples by the scheme. The spatial distribution of the samples was first examined using the scores plot of principal component analysis. Then, predictive modeling coupled with random resampling was performed to determine the discriminatory capability of the IR dataset. Results showed that the inter-geographical location variation was much more pronounced than the inter-scheme one. Predictive modeling performed on kerosene samples from Delhi alone achieved an accuracy of approximately 86% for testing samples via the K-nearest neighbor algorithm in discriminating the samples by scheme. In conclusion, ATR-FTIR spectroscopy coupled with predictive modeling could be applied to determine the scheme of unknown kerosene.