The Research And Application For Quantitative Analysis Of Heavy Metals In Soil Based On XRF

Soil is the basis of the material and energy circulation of the earth's ecosystem and the essential element for human survival and development.Heavy metal pollution in China's soil is becoming increasingly serious,with more than 12 million tons of grain contaminated by heavy metals every year.Soil pollution is related to national food security and people's health.In order to better manage the limited arable land resources and scientifically carry out land use planning,it is necessary to acquire the information of soil heavy metal pollution accurately and quickly.X-ray Fluorescence(XRF)analysis technology is a multi-element analysis method with excellent performance,and is one of the standard analysis methods of the International Organization for Standards(ISO).The energy dispersive XRF has the advantages of simple structure,low power and easy miniaturization,which can be applied to the field investigation of soil pollution.Due to the variety of soil elements,the obvious influence of matrix effect,and the disunity of physical parameters,in situ XRF spectra have complicated background components,overlapping spectral peaks,and water content seriously affecting the accuracy,which limit the application of this technology.Therefore,it is of great significance to develop the technology suitable for soil in situ detection and improve the accuracy of XRF analysis.The spectral background affect the calculation of the net area of the element characteristic peak.The premise of element quantification is accurate estimation of spectral background and deduction.This paper discusses the advantages and disadvantages of penalized least squares and wavelet transform and proposes a background deduction algorithm using asymmetric weighted penalized least squares.The Logistic function is established by the error of the estimated background and the original spectrum,so that the weight of the estimated background and the characteristic peak falls evenly between 0 and 1,which is very suitable for the background estimation of energy dispersion XRF.Compared with the results of other deduction algorithms,the accuracy of the proposed method is better than other methods in estimating the background of low content elements.The performance evaluation results show that the asymmetric weighted penalized least square method can accurately deduct the spectral background and improve the accuracy of XRF quantification.In situ analysis,water in soil has always been considered as one of the main factors affecting the accuracy of detection.The presence of water directly affects the effective atomic number and the total mass absorption coefficient of the sample.The influence of water on the element characteristic peak,element content,mass absorption coefficient and other parameters of the influence mechanism of water was obtained through the derivation of physical theory.The experimental results of water content gradient show that with the increase of water content,soil spectral background increases,Compton and Rayleigh scattering increase,and element characteristic peaks decrease.The physical model established shows that water is a part of the soil matrix,which leads to the decrease of the ray attenuation ability of the sample.The reduction in the effective atomic number is analogous to the "dilution" of the target element.In this study,a correction method of moisture influence was established based on the experimental verification results,and a more accurate correction result was obtained by introducing Compton internal standard method and correction coefficient.The high detection limit of trace elements has always restricted the popularization and application of XRF.To solve this problem,two breakthroughs are needed.Firstly,the background deduction method was optimized to improve the background deduction rate and SNR,and the influence of noise and background count on the characteristic peak was overcome.Secondly,the mechanism of multi-detector combination was explored and the stoichiometric model of data fusion was established by introducing near-infrared spectroscopy.The results show that the fusion quantitative model is more accurate than the single spectral model.The spectral pretreatment and fusion methods were optimized by stoichiometry,and the quantitative models of Cd and Hg were established,and each index met the requirements of the calibration specification.In a word,based on the existing techniques,the background deduction method based on asymmetric weighted penalty least squares is established in this study.An overlapping peak analysis method based on Gaussian mixture model-chaotic particle swarm optimization is proposed,and the related user interface is developed.The influence mechanism of moisture on portable XRF is revealed,and the corresponding correction method is proposed.It further enriches the detection technology of heavy metals in soil by X-ray fluorescence and provides an important technical support for future promotion and demonstration application.