On polarization resolved laser induced breakdown spectroscopy combined with support-vector regression to improve the accuracy of soil heavy-metal (Cd) detection

Elemental Cd is the most toxic heavy-metal pollution element. If the Cd content in the soil is too high, it accumulates on crops and endanger human health through the food chain. Therefore, the detection of Cd content in soil has great significance. In this study, polarization resolved laser induced breakdown spectroscopy (PRLIBS) was combined with the support-vector regression (SVR) algorithm for Cd determination in soil. The soil samples were compressed, the PRLIBS data on the surface of soil samples were collected as input data, and the calibration-curve models were established for the quantitative analysis. A combination of the internal-standard method and SVR algorithm was used. Results showed that under the PRLIBS conditions, the quantitative-analysis parameters obtained by SVR were better than those of the internal-standard method, and the fitting coefficient (R-2) increased to 0.9946. The root-mean-square (RMS) error calibration decreased by 11.8783 mu g/g, and the RMS error prediction decreased by 11.8906 mu g/g. These results showed that PRLIBS combined with SVR was suitable for Cd detection in soils and with good detection effect.

Automated summary

In this study, polarization resolved laser induced breakdown spectroscopy (PRLIBS) was combined with the support-vector regression (SVR) algorithm for Cd determination in soil. Soil samples were compressed and PRLIBS data on the surface of soil samples were collected as input data to establish calibration-curve models for quantitative analysis. The results showed that PRLIBS combined with SVR was suitable for Cd detection in soils with good detection effect.