Prediction of heavy metal Cd and stress on minerals in rice by analysis of LIBS spectra

To predict the nutrition and safety of agricultural products by laser-induced breakdown spectroscopy (LIBS), heavy metal Cd in rice was selected as an analytical target. Mature rice grain samples from 40 growing geographical areas around Poyang Lake were picked on-site and processed by grinding to obtain the edible rice. The content of Cd in rice samples was determined by graphite furnace atomic absorption spectrometry, and the rice pellets were detected by LIBS. The risk intake was estimated by the target hazard quotient and Chinese National Standard. Moreover, the samples were classified as clean, slight, and severe ones according to evaluation. The content of Cd was predicted by analyzing LIBS spectra coupled with the partial least square (PLS) model. The correlation coefficients (R2) reached 0.9036 and 0.9771 for the training and prediction sets, respectively, and the root mean square errors were 0.0487 and 0.027, respectively. It denotes that the PLS model has a higher prediction ability especially after LIBS spectra were processed by smoothing and multiplicative scatter correction. For the clean, slight, and severe rice samples, the LIBS intensity ratio between minerals Mg, K, Na, Si, and Mn to Ca was compared. The ratio was decreased in all samples as Cd stress increased. Correlation analysis results show that Mn displayed a highly significant negative correlation with Cd stress, while Mg, K, and Na displayed a significant negative correlation with Cd stress. The relationship between Si and Cd did not reach a significant level. This work indicated that it was feasible to use LIBS combined with a suitable data process to predict Cd content and the effect of Cd stress on minerals in rice. It is promising to evaluate the nutrition and safety of food products by analyzing LIBS spectra.