Implementation Of LIBS Echelle Spectrometer And Research On Data Processing Method

As one of the most promising techniques of atomic emission spectroscopy,laser-induced breakdown spectroscopy(LIBS)features in situ and nearly non-destructive detection as well as simple preparation of samples.However,there are still a lot of problems existing.In one aspect,LIBS adopts pulsed laser as the inducing source,therefore the plasma lasts for a short time with time resolved spectroscopy.In the other aspect,in the detection of complex samples,especially for metal elements,atomic spectral lines occupy a wide wavelength range with high density.All these facts present challenges for the spectrometers and the analyzing method.Aiming to apply LIBS technique in the detection of metal elements in soil samples,this work studies the design and optimization of LIBS echelle spectrometer and LIBS data processing method.We propose novel methods of implementation of a time-resolved LIBS echelle spectrometer,broadband astigmatism correction and LIBS variable selection.The main contents of this paper include the following aspects:1.Aiming at the problem of LIBS collection,a solution of the implementation of a time-resolved LIBS echelle spectrometer is proposed.According to the theory of echelle diffraction and prism dispersion,a compact cross dispersion optical system with wide wavelength range and high resolution is designed.Then,the topological structure of time resolved ICCD imaging system is established based on the FPGA and ARM platform.Finally,an optimized spectral reduction algorithm combining the least square method and the polynomial fitting method is propoed based on the imaging model and the centroid extraction algorithm.2.To optimize the imaging quality,we propose an astigmatism-correction method using off-the-shelf cylindrical lens.The astigmatic model of the echelle spectrometer is established.And a cylindrical lens is introduced in the optical design,of which the position parameters are calculated depending on the model and the wavelength range.Therefore the astigmatism could be corrected overall in the wide wavelength range.The simulation result indicates that the imaging quality is improved obviously after correction,and the RMS radius of the spot reduces by nearly 10 times.3.Usually,traditional variable selection method cannot be applied to LIBS.To solve this problem,a fast variable selection method combining interval partial least squares(iPLS)and modified iterative predictor weighting-partial least squares(mIPW-PLS)is proposed in this work.Firstly,the whole spectrum is divided into several intervals.Then the correction factors are calculated based on the regression coefficients of iPLS and PLS on the whole spectrum.Finally,mIPW-PLS is utilized to select variables rapidly and reduce the dimension of the spectral data.4.Considering conventional variable selection methods work with low efficiency,we propose an automatic variable selection method based on full spectrum correction.On the basis of mIPW-PLS,correction fators are established with the correlation coefficient and regression coefficient of PLS on the whole spectrum.Meanwhile a selection criterion is put forward based on the number of variables and RMSECV values to select variables automatically.Compared with classical methods GA and SPA,FSC-mIPW-PLS is quite superior on the aspect of computational time,which is just 1/100 of GA and 1/600 of SPA.5.We finally develop the LIBS echelle spectrometer and establish the LIBS experiment system.On one hand,we implement the calibration experiment using a mercury lamp.The result demonstrates that the LIBS echelle spectrometer meets the requirements of the project.On the other hand,we carry out the quantitative analysis of some metal elements in soil samples.The result presents that the proposed variable selection method could improve the prediction ability and the calculation efficiency significantly.