On-Line Determination of Hydrogen Peroxide in Industrial Samples by Near-Infrared Spectroscopy

Kazem Kargosha, Seyyed Hamid Ahmadi and Abolghasem Najafi

Chemistry and Chemical Engineering Research Centre of Iran, PO Box 14335-186, Tehran, Iran.

The multivariate analysis of near-IR spectroscopy of hydrogen peroxide in the spectral region of 8000-4000 cm-1, provides an accurate, noninvasive and fast method for it's quantitation in batch (stopped-flow) and flow modes. Two multivariate calibration approaches were employed for data reduction, classical least squares (CLS) and partial least squares (PLS). The calibrations were carried out on 11 standard samples characterized by a wide range of H2O2 concentrations (0-9M). The two calibration methods gave good prediction results, but validation evaluations showed that PLS has the better prediction precision and accuracy in both methods of stopped-flow and flow modes. Standard error of estimation (SEE) and correlation coefficient (R) for CLS and PLS methods in stopped-flow mode were found to be 0.082M, 0.9996 and 0.007M, 0.9999 respectively. The results of analysing industrial hydrogen peroxide samples in flow mode, also showed good agreement with standard methods. The relative standard deviation was found to be 1.0%, in flow mode, for commercial concentrated sample containing 5.34 M of H2O2 (n=8).

spectroscopy; hydrogen peroxide; correlation coefficient