Determination of sugars in aqueous mixtures using mid-infrared spectroscopy

The combination of mid-infrared (MIR) spectroscopy and multivariate statistics for determination of glucose, fructose, and sucrose in aqueous mixtures was investigated. Spectra of different samples were recorded in the MIR range between 400 and 4000 nm. Aqueous mixtures of 10, 20, and 40% total sugars with different combinations of glucose, fructose and sucrose were used for calibration and validation studies. Spectral wavenumber region from 700 to 1530 cm^-1 was selected for Principal Component Regression (PCR) and Partial Least Square regression (PLS) to develop calibration models for sugar content determination (%, w/v) in liquid foods. Results from the PLS-1 (first derivative) models were slightly better than those obtained with other used calibration models. Merging of the 10, 20, and 40% calibration data into single calibration set enabled the prediction (average standard error of calibration, SEC = 0.34) of individual sugars in the entire range accurately. The squared correlation (R²) values for the calibrated models for glucose, fructose and sucrose were 0.997, 0.998, and 0.997, respectively, between the predicted values and actual values. The calibrations were successfully applied to predict sugar contents in complex mixtures and commercial beverages. Results indicated that fourier transform infrared (FTIR) spectroscopy can be used for rapid detection of sugars in complex mixtures.