Quantitative Analysis Of Heavy Metal Pollutants In Soils Based On LIBS

In situ and rapid analyzing the concentration distribution of heavy metal pollutants in soils has special meaning in controlling the degree of heavy metal polltution and assessing the quality of soils.LIBS provides a great potential method which can quickly,in situ and simultaneously analyze multi-elements,and it shows great advantages of simply analysis and rapid detection.Therefore,it has great values to do some research on the LIBS detection methods for heavy metal pollutants.This paper has done some research work on weakness of soil matrix effects and quantitative analysis methods on heavy metal pollutants in soils,and the main researches are as follows.(1)The analytical spectral lines separation and the highly stable spectral acquisition method are studied.The LIBS spectra of the soil are measured and the main spectral lines of heavy metal elements in soils are obtained.The analytical lines for different elements are also chosen based on the principles.Then,we investigate the effect of the grain size of soils,the hydraulic pressure,and the average times of laser pulse on the LIBS spectra intensity,optimizing the experimental system parameters.Finally,we explore the methods of spectral data pre-processing methods to improving the stability of the measured LIBS spectra.(2)The influences of soil matrix effects on quantitative analysis are analyzed and the correction methods are discussed.The results of univariate quantitative analysis of six heavy metal elements in soils and the limits of detections are analyzed.Using Lorentz fitting and multiple linear regression analysis to correct the influence of soil matrix effects.The soil classification is realized by using principal component analysis to select the quantitative analysis method.The results show that the relative standard deviations of Cr are all less than 6.81%when the matric of predict samples are similar to the calibration samples,and the relative standard deviations of Cr are lager when the matric of predict samples are different to the calibration samples.(3)LIBS technique based on supporting vector machine regression(SVR)is proposed for quantitative analysis of heavy metal elements in soils.In the case of Cr,the feature extraction model and the whole spectra model are founded,the results show that,when the matric of the calibration samples are complex,the single relative standard deviations are all less than 6.13%and the average relative standard deviations is 3.03%under the feature extraction model.All the predicted results under the whole spectra model are worse than the feature extraction model,and the predicting speed is lower.(4)Calibration-Free LIBS(CF-LIBS)is applied to analyze the heavy metals in soils.For Si is the basic element in soils,the concentration ratio of Cr and Si is analyzed by CF-LIBS.The results show that the relative standard deviations of Cr in certified soil samples are all below 6.54%,and for the soil samples collected from field,the relative standard deviations are all below 16.44%.Then the concentration distribution of Cr in soils around a smelter is calculated by CF-LIBS(assuming the concentration of Si is constant in a little area)and the results is compared with those by ICP-OES.