Development of a Portable NIR Spectrometer for Detecting Pesticide Residues

Abstract

The problem of pesticide residues found in fruits and vegetables that exceed the standard is something that all sectors are interested in solving. The main reason is that farmers, consumers, and relevant authorities do not know the real-time value of the residues. The detection of the pesticide residues is not immediately known since it must have been carried out at the central laboratory, where the received result will also take so long time. To solve this problem, our research team has designed and developed a portable near-infrared (NIR) spectrometer. The developed NIR spectrometer is designed to not only detect the reflected intensity of the residues in the wavelength range from 410 [nm] to 940 [nm] using the AS7265x chipset, but also collect and analyze the normalized spectral signal using the microprocessor ESP32-WROVER-B for detecting each type of the four pesticide residues: Carbendazim, Cypermethrin, Diazinon, and Imidacloprid. From experimental results on forty pesticide residues on basil leaves and chili from the local market in Phitsanulok province, it was conclusively demonstrated that the NIR spectrometer correctly identifies a tested type of the four pesticide residues on the twenty-eight basil leaves and twenty chili, and has more stable, consistent and accurate performance for detecting the pesticide type of the forty residues than the thin-layer chromatography method utilized at the central laboratory. Furthermore, the developed NIR spectrometer exhibits remarkable versatility and the best performance of detecting each type of the four pesticide residues on the twenty-eight basil leaves and twenty-eight chili or the total fifty-six samples as well as a test run repeated 100 times per sample and at seven concentration levels. At the pesticide concentration levels of 1, 2, 3, 4 and 5 mg/l, the Accuracy, Precision and Recall values were perfect at 1.00 and standard deviation of zero in all cases. Also, the Accuracy value was greater than 0.98 and both the Precision and Recall values were greater than 0.97 with an overall standard deviation of less than 0.013 when detecting the two pesticide residue types at concentration levels of 0.05 and 0.1 mg/l. Overall, the results showed that the proposed NIR spectrometer  correctly detects pesticide residues in the concentration range from 1 [mg/l] to 5 [mg/l]. As well, the total cost of the tests with the portable NIR spectrometer was about 4,395 Baht. This cost is very reasonable particularly when the price of the proposed portable NIR spectrometer is nearly half that of devices with identical specifications that are sold on the commercial market.