Applications of infrared spectroscopy to agricultural and food products

Infrared (IR) spectroscopy has been widely used for component analysis because it is simple, fast, accurate and environmentally friendly. These studies were conducted to investigate the possibility of using IR spectroscopy as an alternative to the conventional methods for determining degree of degradation of frying oils, and also for determining blend level of mixtures of biodiesel and petrodiesel. In addition, the potential of using a single universal calibration to measure degree of degradation of frying oils or blend level of blended biodiesels, respectively, was evaluated in the studies. The first study shows that increased absorption due to a longer path length (4 mm) slightly improved the results for frying oils in the shorter wavelength region (700–100 nm), but the results were not as good as the results obtained by models using a 2 mm path length and longer wavelength region from a previous study. In the second study, near-infrared (NIR) methods were successfully developed for measuring total polar material (TPM) and free fatty acid (FFA) contents in frying oils used to fry various foods. Additionally, a single universal calibration model could be used to determine quality of frying oils used to fry various foods. With the partial least-squares (PLS) models, the best correlations (r) between the NIR predicted data and reference method data for TPMs and FFAs in frying oils were 0.995 and 0.981, respectively. The third study investigated the possibility of using mid-infrared (MIR) spectroscopy for determining degree of degradation of frying oils used to fry various foods. The results show that MIR spectroscopy, coupled with a single reflection attenuated total reflection (ATR) cell, was not as good as NIR transmission spectroscopy for predicting TPM and FFA contents in frying oils used to fry various foods. With the PLS models, the r values between MIR predicated data and reference method data for TPMs and FFAs in frying oils were 0.975 and 0.892, respectively. The fourth study shows that NIR transmission spectroscopy was successful for estimating the blend levels of blended biodiesels. Also, blend levels of fuels from different sources could be determined by a single universal calibration model. With the PLS models using longer wavelengths, the r and root-mean-square of the differences (rmsd) values were 1 and 0.018%, respectively