Fast Determination Of Chromium Content In Rice Based On Laser Induced Breakdown Spectroscopy

Heavy metals pollution has been a great challenge for modern agriculture,because heavy metals could enter into body through food-chain and threaten human' s health.As an important part of agricultural production system,crop is the main supplement of regular food.Hence,it is of crucial importance for precision agriculture to fast detect heavy metals in crops,which might provide an approach for monitoring the soil pollution and ensuring food safety.The traditional approaches for heavy metals detection need amounts of chemical reagents and laboratorial instruments,which cannot meet the demand of fast and in-field measurement.Laser-induced breakdown spectroscopy(LIBS)is one kind of novel atomic emission spectroscopy,which has the advantages of minimal sample preparation,fast analysis speed and multi-element detection capability.Therefore,we mainly focused on the application of LIBS for chromium content detection in rice(Oryza sativa L.).LIBS combined with data preprocessing methods and modeling methods were used for fast diagnosis of chromium stress under different stress time and quantification of chromium content in rice leaves.Moisture content reducing method,532 nm and 1064 nm LIBS and dual pulse LIBS were applied to further improve the calibration performance for chromium content detection.This study might provide a novel and efficient approach for monitoring heavy metals pollution in crops and ensuring food safety,and it also supplies technical support for heavy metals transportation research.The main conclusions were achieved as follows:(1)Multivariate discriminant models for fast diagnosis of chromium stress level of rice were established,and it provided an approach for heavy metal pollution diagnosis,prevention and treatment in crop.For the rice with different chromium stress levels(CK,mild pollution,severe pollution)under different treatment periods(1,3,5,9 and 13 days),the LIBS was used for obtaining the spectra of fresh rice leaves and dried rice leaves,and the discriminant models for diagnosis were established based on full spectrum.In addition,the feature variables associated with chromium stress was selected based on regression coefficient(RC)values of partial least squares(PLS)models,and the discriminant models based on feature variables were established.The results indicated that ? the optimal time for chromium stress diagnosis was after 3 days treatment.And based on the selected feature variables,the variation of concentration of Si,Mg,Ca might be related to chromium stress;? the best result for chromium stress diagnosis was achieved by extreme learning machine(ELM)model.For the model based on full spectrum,the discriminant accuracy for fresh rice leaves and dried rice leaves were 93.62%and 89.53%,respectively.For the model based on feature variables,the discriminant accuracy for fresh rice leaves and dried rice leaves were 89.36%and 90.70%.The discriminant accuracy based on feature variables was similar to that based on feature variables,while the variables used for modeling were greatly reduced;? It was recommended that the rice leaves should be dried before LIBS analysis for chromium stress diagnosis.The discriminant results based on dried rice leaves were less affected by moisture content,which were better and stable.(2)The moisture content influence reducing method for chromium content detection in fresh rice leaves was proposed,and it increased precision and stability of LIBS detection,and provided technological support for field application.In order to clarify the role of moisture content played in LIBS analysis,we firstly investigated the influences of moisture content on signal intensity,stability and plasma parameters(temperature and electron density)using the rice leaves with various moisture contents and tried to find main influence factors on the variations of signal.We reduced the moisture content using quick drying procedure,and two strategies were used to compensate the effect of moisture content and shot-to-shot fluctuation.The results indicated that ? the signal intensity and the signal stability could be greatly reduced by the moisture content in plant materials.The variation of emissions might be mainly caused by the experimental parameters F and the change of analyte concentration;? It has shown that an exponential function based on the intensity of background could be used to correct the actual element concentration in plant materials;? In addition,using the ratio of signal-to-background for univariate analysis and PLSR for multivariate analysis could eliminate the detrimental effects and improve the calibration performance.Best calibration result was achieved by the PLS model that relating the variables in the range of 425-428 nm with corrected reference values,with the correlation coefficient of prediction(RP)of 0.9669 and root mean standard error of predicton(RMSEP)of 4.75 mg/kg.In addition,the best calibration result of PLS outperformed that in univariable calibration,which indicated that multivariable analysis was preferred in complex situation.(3)With the study of influence and mechanism of laser wavelength,the quantification models for detecting chromium content in rice leave pellets were established using 532 nm and 1064 nm LIBS.The proposed method provided an approach for fast and efficient detection of heavy metal in crop,as well as the technological support for study of transportation mechanism of heavy metal in crop.We compared the influence of wavelength on signal(sensitivity and repeatability),plasma features(electron density and temperature)and quantification capability;And we optimized the experimental parameters for both 532 nm and 1064 nm laser wavelengths;With the optimal experimental parameters,we established univariate models for fast detection of chromium content in rice leaves,and further improved the performance with preprocessing methods(background normalization,area normalization,standardized normal variate(SNV)and multiplicative scatter correction(MSC)).The results indicated that ? the predominate mechanism of laser wavelength might differ with the variations of delay time,pulse energy and lens-to-sample distance(LTSD).The effect of wavelength on chromium signal might be the results of transition energy,inverse bremsstrahlung and ablation efficiency;? With the comparison of the signal sensitivity and stability,the experimental parameters including delay time,gate width,pulse energy and LTSD were optimized at 4/4?s,16/16 ?s,90/80 mJ and 98/99 mm for 532/1064 nm excitations,respectively;? In the aspect of quantification capability,532 nm excitation obtained better calibration results than 1064 excitation,with higher values of R and lower RMSEs and LODs.Best calibration result was achieved by relating the SNV normalized signal of Cr I 425.43 nm with reference chromium content,with Rp of 0.9790,RMSEP of 4.62 mg/kg,LOD of 2.72 mg/kg.(4)The quantification models for detecting chromium content in rice leave pellets and dried rice leaves were established using dual pulse signal enhancement LIBS,and the distribution of chroumium content in dried rice leaves was successfully visualized.The proposed method provided an approach for fast detection,prevention and treatment of heavy metal in crop.We studied the influence of inter-pulse delay time and energy ratio of the 1st pulse to 2nd pulse on signal enhancement and found the optimal parameters.For rice leave pellets,the univariate and multivariate models for chromium content detection were established with single pulse LIBS(SPLIBS)and collinear dual pulse LIBS(DPLIBS);For dried rice leaves,the univariate and multivariate models for chromium content detection were established with SPLIBS and reheating DPLIBS.With the combination of LIBS spectra and spatial ablation spots,the chromium content in rice leaves was visualized.The results indicated that ? the optimized inter-pulse delay time for collinear DPLIBS was 1.5 ?s,and the optimized and energy ratio of the 1st to 2nd pulse for collinear DPLIBS was 1:3 with total energy of 80 mJ;The optimized inter-pulse delay time for collinear DPLIBS was 1.5 ?s,and the optimized and energy ratio of the 1st to 2nd pulse for collinear DPLIBS was 6:5 with 1st energy pulse of 60 mJ;? Better prediction results was achieved by SVM models based on feature variables from DPLIBS.For rice leave pellets detection,SVM model based on feature variables obtained the best results,with RP=0.9946,RMSEP=4.85 mg/kg,RPD=9.70.For dried rice leaves,SVM model based on feature variables also obtained the best results,with Rp=0.9585,RMSEP=13.39 mg/kg,RPD=3.59;? The sensitivity of LIBS systems was improved by reheating DPLIBS.The detection limits for SPLIBS and DPLIBS were 10.62 mg/kg and 6.30 mg/kg,respectively;? The distribution of chromium content in rice leaves was achieved with the combination of ablation spots and chromium content predicted by SVM model.This study might provide an approach for the mechanism study of heavy metal transportation.