Silicon Crystals Based Ftir-Atr for the Rapid Monitoring of Phosphorus in Water Coupled with the Algorithms of Deconvolution and Partial Least Squares Regression

Abstract

Conventional laboratory methods for phosphorus (P) analysis are time-consuming and laborious, which limited the application of rapid P monitoring in water body. In this study, silicon crystal was applied to construct the systems of Fourier-transform mid-infrared attenuated total reflectance spectroscopy (FTIR-ATR) and was firstly applied in P determination in water using the typical P-O vibration with the typical wavenumber range of 1200-800 cm–1. FTIR-ATR spectra were obtained using Si crystal (Si-ATR) and conventional ZnSe crystal (ZnSe-ATR). The algorithms of deconvolution and partial least squares regression (PLSR) were involved to establish models for non-destructive and rapid measurement of P. The results showed that the typical absorption band of P was obtained from Si-ATR aligned with the conventional ZnSe-ATR. For the ZnSe-ATR, the relation coefficient (R2) of PLSR models was more than 0.924 for different P content ranges, and the residual prediction deviation (RPD) was more than 3.02; while for the Si-ATR, the R2 was more than 0.883, and the RPD was more than 2.13. There was no significant difference between the predicted and real values, and there was no significant difference between Si and ZnSe crystal; the spectral pretreatment using deconvolution algorithm could effectively remove the interference of sulfate to reach an excellent prediction through extracting the absorption of P. Furthermore, a novel P sensor was developed using Si-ATR combining the algorithms of deconvolution and partial least squares regression, which achieved the rapid P monitoring in water.

Keywords: phosphate, silicon crystal, Fourier transform mid-Infrared attenuated total reflectance spectroscopy, deconvolution, partial least squares regression