Near-infrared (NIR) spectroscopy has been evaluated for monitoring the acid-catalyzed hydrolysis (thinning) of starch. In practice, the extent of starch hydrolysis is measured in fluidity units, which correspond to a physical property of the hydrolyzed starch material. NIR spectra of samples taken periodically during a series of starch-thinning reactions were used to predict fluidity. The standard error of prediction (SEP) was 1.06 mL with the use of partial least-squares (PLS) regression in conjunction with digital Fourier filtering. This SEP was significantly better than that reported before with a univariate calibration model based on the integrated area of the 4400 cm-1 (2272 nm) absorption band for carbohydrates. The improved SEP meets the industry demands for real-time monitoring. Although these calibration models were developed from samples prepared in the laboratory, no spectroscopic differences were apparent between spectra collected from these laboratory samples and spectra collected from samples taken directly from plant starch slurries during actual thinning reactions. This similarity in spectral features, and hence chemical matrix, supports the potential of NIR spectroscopy for on-line monitoring of industrial starch-thinning processes.
Hoeil Chung and Mark A. Arnold, "Near-Infrared Spectroscopy for Monitoring Starch Hydrolysis," Appl. Spectrosc. 54, 277-283 (2000)