Self-absorption model in quantitative laser induced breakdown spectroscopy measurements on soils and sediments

Abstract

Application of laser induced breakdown spectroscopy (LIBS) in the quantitative analysis of elemental composition of soils with different origins and Antarctic marine sediments has been considered. The analytical method followed includes the usual plasma modeling at local thermal equilibrium (LTE) based on average temperature and electron density values, as well as spectra normalization, introduced in order to reduce the effects related both to the substrate optical and thermal properties and to the influence of laser parameters on quantitative data. The computational algorithm takes into account only atomic species and their first ionization states, which is sufficient at the plasma temperature measured in the experiments. Calibration curves are finally generated for each element of interest measured on certified samples with different provenience and matrix composition. In this paper a model is developed which takes into account the effects responsible for non-linearities in the relationship between line intensity and elemental concentration. The model properly includes line re-absorption and contributions from space regions with different plasma densities. Its application permits us to obtain the correlation coefficients between the LIBS measured and certified concentration of each element analyzed. These coefficients, specific for a given experimental layout and atomic lines data base, are successively applied in analytical LIBS measurements allowing for the direct determination of a single element concentration in any sample, regardless of its unknown matrix composition. The LIBS method presented here was tested on a priori unknown samples, and gave uncertainties in concentration varying from 15 to 40% over a large concentration range covering several orders of magnitude. The measuring error depends on element type, on the concentration value and also on the number of certified samples used for the initial calibration. The present results are already significant for some field application, such as on-board marine sediment analysis where a significant matrix variation with layer depth is common.