No abstract

No abstract

A new procedure for determining free fatty acids (FFA) in olive oil based on spectroscopic Fourier transform infrared-attenuated total reflectance spectroscopy measurements is proposed. The range of FFA contents of samples was extended by adding oleic acid to several virgin and pure olive oils, from 0.1 to 2.1%. Calibration models were constructed using partial least-squares regression (PLSR). Two wavenumber ranges (1775-1689 cm −1 and 1480-1050 cm −1 ) and several pretreatments [first and second derivative; standard normal variate (SNV)] were tested. To obtain good results, splitting of the calibration range into two concentration intervals (0.1 to 0.5% and 0.5 to 2.1%) was needed. The use of SNV as a pretreatment allows one to analyze samples of different origins. The best results were those obtained in the 1775-1689 cm −1 range, using 3 PLSR components. In both concentration ranges, at a confidence interval of α = 0.05, no significant differences between the reference values and the calculated values were observed. Reliability of the calibration vs. stressed oil samples was tested, obtaining satisfactory results. The developed method was rapid, with a total analysis time of 5 min; it is environment-friendly, and it is applicable to samples of different categories (extra virgin, virgin, pure, and pomace oil).Paper no. J8932 in JAOCS 76, 611-616 (May 1999). FIG. 3. Absolute errors for calibration (•) and prediction samples (• •) obtained in the absorbance mode and wavenumber range 1775-1689 cm −1 .

No abstract

Near-infrared diffuse reflectance spectroscopy (NIRS) with a fiber-optic probe was used for the determination of the active compound in a commercial pharmaceutical preparation. In order to reduce the strong scatter in the spectra and prevent scatter-induced changes in measurements from prevailing over concentration-induced changes, several data preprocessing methods were tested: normalization, derivatives, multiplicative scatter correction, standard normal variate, and detrending. The effectiveness for reducing the scattering of each data preprocessing was assessed, and the best results were obtained with the use of the second derivative. The effect of the treatments on the quantitation of the active compound by partial least-squares regression (PLSR) was studied, similar results being obtained in all cases, with a relative standard error of prediction lower than 1.55%.