Quantitative determination of silica mixtures by multivariate calibration applied to infrared spectroscopy

Abstract

Mixtures of crystalline and amorphous silica cannot be quantified by standard infrared spectroscopic techniques alone. This is due to the severe overlap of the primary spectral SiO 2 band in the most characteristic region, 850-750 cm −1. This problem may be solved with the use of multivariate calibration by partial least squares regression (PLSR). Using this method, the complete relevant infrared spectral profile of the sample may be used for quantitative determinations of silica mixtures. The three silica modifications of quartz, cristobalite and fumed amorphous silica have been simultaneously quantified in mixtures using this technique. The standard errors of prediction are 25, 26 and 24 μg, respectively, for the three silica components in 1 mg dust samples. Multivariate calibration with the PLSR technique thus offers an accurate, rapid and low cost method for the determination of silica mixtures using infrared spectroscopy.