Chiang Mai Journal of Science

Near infrared (NIR) spectroscopy detects the absorbance in the region between 800 and 2500 nm, providing spectral information of overtone and combination bands from the infrared (IR) absorption. The NIR measurement is non-destructive and requires no or less sample preparation which is preferably suitable for detection of agricultural samples such as coffee, rice, soybean and mung bean. However, at present, commercial NIR spectrometers are relatively expensive and many of them require specific setting for the operation such as large space and stability in the experimental conditions. In this research, a low-cost NIR spectrometer was developed. The use of micro-electro-mechanical system (MEMS) compartment allowed the NIR detection to be in a compact size with relatively low instrumental cost. To demonstrate the performance of the developed NIR spectrometer, several agricultural products including coffee, rice, soybean and mung bean were used. The detection parameters affecting to the obtaining NIR spectra were studied. The digital resolution, exposure time, and width of pixels, respectively, of 225 scanning points, 0.635 ms and 8.20 nm were recommended for the studied agricultural samples. The recorded NIR spectra were exploratorily analyzed using principal component analysis (PCA). It was found that the developed NIR spectrometer combined with PCA could successfully characterize the difference among the studied samples as well as their combinations. In addition, the potential tendencies to classify coffee bean according the roasting level and to predict the impurity of adulterated Thai jasmine rice were presented.