Quantification of fermentation levels in cocoa beans using FT-NIR spectroscopy

Abstract. Calibration models to predict cocoa bean's quality parameters, such as fermentation index, pH and total polyphenol content were developed using Fourier transform (FT) near-infrared (NIR) spectroscopy in a nondestructive manner. Cocoa pods were stored up to 21 days as a pretreatment method and the effect on quality parameters were investigated at specified intervals. The beans from the stored pods were fermented by the heap method. Fermentation index of all treatments ranged between 0.54 and 1.24, pH values between 4.26 and 6.13, and total polyphenol content between 6.48 and 15.58 mg/g. The samples were scanned in the NIR region (780 and 2500 nm) for the spectral analysis. Results obtained from conventional quality analysis were correlated with spectra of respective treatment for calibration models development. Partial least square regression technique was used to develop the calibration models. Scanned spectra were preprocessed with vector normalization, multiplicative scatter correction, and first derivative techniques. The calibration models fitted the prediction of fermentation index and total polyphenols well (R² ≥ 0.80), compared to that of pH (R² < 0.80). The proposed method was quick and it took less than a minute when compared to the conventional method (~28 h). The FT-NIR spectroscopy method of analysis produced rapid and accurate quality attributes of fermenting cocoa, and can be employed to similar food products.