Abstract

Soybean with the scientific name of Glycine max (L.) Merr. is nutritious vegetable food sources in Indonesia, and it may be processed into various kinds of food products. Soybeans are a rich source of isoflavones, rather than nuts and meat products. The research was aimed at quantifying daidzein (DN) and genistein (GN) in various soybean varieties applying Fourier transform Infrared (FTIR) spectroscopy coupled with two multivariate calibrations [principal component regression (PCR) and partial least square (PLS)]. Before being used as variables during quantification processes, some pretreatments of FTIR spectra including the selection of wavenumbers and spectral derivative (high-order derivatives) were carried out to get the best prediction models for the correlation between the actual values of DN and GN. The actual values of DN and GN were determined with High-performance liquid chromatography (HPLC). The results revealed that PLS provides a better modeling than PCR for the relationship between actual values and FTIR spectroscopy-predicted values. The first derivative spectra at wavenumbers of 3,600–2,800 and 1,500–780 using PLS regression were preferred for the quantification of DN and GN in soybean. PLS calibration model yielded R2 for the relationship between HPLC-based actual values (x-axis) and FTIR-predicted values (y-axis) of DN and GN, which were 0.9947 and 0.9900, respectively. FTIR spectroscopy method-combined PLS regression provides fast and acceptable results as indicated by high accuracy and precision during quantification of DN and GN in soybean.

Key words: Daidzein, FTIR, genistein, PLS, PCR, soybean