Developing A Novel Onsite Detection Technology By Using Chemometrical Analysis Of Hand-Held Near-Infrared Sensor Technique For Assessing Coffee Quality

Abstract

This study aims to apply handheld infrared spectral technique to develop a predictive model for the identification of adulterants and rapid estimation of the quality of coffee products. For the classification of coffee varieties, the novel potable NIR spectrometer combined with multivariant qualitative algorithms gave 98.8%, 99.72% and 99.22% identification for raw, roasted and roasted powdered coffee, respectively. Also, in the classification of Africa coffee types, it gave 99.76%, 99.78% and 99.88% identification for raw, roasted and roasted powdered coffee, respectively. Qualitatively, FD-LDA performed better with 97.78% and 100% in both calibration and prediction sets in the determination of coffee husk in roasted coffee powder. Quantitatively, in the detection of coffee adulteration, SPAPLS model had the best results with R=0.9711 and 0.9897 in both calibration and prediction sets respectively. The novel handheld spectroscopy could be employed for the discrimination of coffee varieties and African robusta coffee in three forms (raw, roasted and powder) and quantification of coffee husk in coffee. With 10% occurrence frequencies, two fungi, Aspergillus niger and flavus were found in commercially sold powder coffee in some of the major markets in Ghana with acrylamide levels below the benchmark threshold (400ug/kg) set by the European Commission. The proximate analysis conducted on the commercially sold coffee powder revealed high moisture and ash attributed to a substantial amount of impurities in the coffee samples. Furthermore, minerals namely nitrogen, phosphorus, potassium and magnesium were found in the coffee powders.