Место издания:Université Claude Bernard Lyon 1 Lyon
Первая страница:P23
Последняя страница:P23
Аннотация:Laser-induced breakdown spectroscopy is a powerful technique for the qualitative and quantitative direct elemental analysis of a variety of objects due to its rapidity, easy maintenance, and possibility of the remote analysis under different ambient conditions. Each of the elements has its own emission spectrum and can be identified and quantitatively analyzed. However, spectral interferences are still the issues in spectral analysis of such complex objects like high alloy steels, soils, ores and rocks because of a huge number of emission transitions of d elements like iron, chromium, titanium and others. We should carefully identify whether the peak observed in a spectrum contains an analytical signal to prevent misinterpretation of experimental data. The contribution of emission of each possible element within narrow spectral region of the peak into the integral intensity will depend on the content of the elements in a sample, fundamental parameters of the expected transitions, known from databases, experimental conditions in laser plasma and instrumental resolution. Since the excitation temperature and electron density in laser plasma are significantly varied in a time after laser pulse, and the plasma parameters depend on the sample matrix, the evaluation of the contributions is a challenging task.
We suggested spectra modeling for the evaluation of the possibility of quantitative analysis of a certain element in a sample with the use of a certain emission line. Despite the plasma inhomogeneity we can estimate the contributions of the lines of interest into the signal under its spectral interfering for the different conditions under LTE, choose the best ones and calculate the best achievable theoretical limit of detection in the homogeneous plasma under known concentrations of matrix elements. The approach helps to choose the best analytical lines and experimental conditions (“hot” or ”cold”, “dense” or “tenuous” plasma) which can be achieved by setting of an appropriate delay of spectra registration or ablation scheme (single or double pulse), or to evidence the impossibility of quantitative analysis. Also by correlation of experimental spectrum with modeled ones we can estimate the excitation conditions and verify what of the lines are really observed in the spectrum instead of what do we want to observe in the certain peak wavelength region.
We have implemented the homogeneous plasma rod model with a length l (for the rough accounting of self absorption) for the choosing of lines for carbon analysis in low alloy steels under atmospheric conditions (>200 nm) and silicon in high alloy steels. It was shown that the worsening of the spectral resolution with the increasing of the slit width improves the analytical results for carbon determination due to specific relationships between Stark width of C I 833.51 line, interfering iron lines and instrumental width. We have also predicted the limit of detection of lead, yttrium and lanthanum in soils and rocks and chlorine in plants from model spectra for the known level of experimental noises. The estimated values were within one order of magnitude with experimental values.