Development of smartphone-controlled and machine-learning-powered integrated equipment for automated detection of bioavailable heavy metals in soils

Abstract

Rapid and accurate on-site detection of crop-absorbable cadmium (Cd) and lead (Pb) in soils is important for food security and human health. The automated soil sample pretreatment method, including the ultrasonic extraction of weakly acid-soluble heavy metals, suction-filtration, and UV photolysis, was proposed to achieve the high-efficiency preparation from soil sample to extract solution. Bismuth-film-modified glass carbon electrode combined with the homemade potentiostat was fabricated to implement the square-wave anodic stripping voltammetry (SWASV) measurements of soil extracts. The peak-information-acquisition algorithm was designed to automatically obtain peak heights and widths of Zn2+, Cd2+, Pb2+, Bi3+, and Cu2+ stripping currents, and then which were used as input variables for establishing machine-learning models to enhance the detection accuracy of SWASV to Cd2+ and Pb2+ under the coexistence of multiple heavy metal ions. Eventually, the smartphone-controlled integrated-automated detection equipment was developed and successfully applied to the automatic pretreatment of soil samples and the determination of weakly acid-soluble Cd2+ and Pb2+ in real soil samples. The detection speed was 75 min/sample, and the detection results were close to the standard method (BCR-ICP-MS). This equipment can provide powerful technical support for on-site rapid and accurate determination of crop-absorbable heavy metals in soils.