Laser-induced breakdown spectroscopy (LIBS) is a potential alternative to wet chemical methods for total soil phosphorus determination, but matrix effects related to physical and chemical sample properties need to be further understood. The aim of this study was to explore matrix effects linked to particle size distribution and chemical form of phosphorus on LIBS response and the ability of LIBS to predict total phosphorus in a range of different soil types. Univariate calibration curves were developed by spiking the soils with increasing doses of phosphorus, and limits of detection for LIBS determined phosphorous (P) (LIBS-P) were calculated. Different particle size distributions in otherwise identical soils were obtained by four milling treatments and effects of chemical form of phosphorus were examined by spiking soils with identical amounts of phosphorus in different chemical compounds. The LIBS-P response showed a high correlation (R2 > 0.99) with total phosphorus for all soils. Yet, the sensitivity of LIBS differed significantly among soils, as the slope of the calibration curves increased with increasing sand content, resulting in estimated limits of detection of 10 mg kg−1 for the sandiest and 122 mg · kg−1 for the most clayey soils. These limits indicate that quantitative evaluation of total phosphorus in sandy and loamy sandy soils by LIBS is feasible, since they are lower than typical total phosphorus concentrations in soil. A given milling treatment created different particle size distributions depending on soil type, and consequently different LIBS-P results. Thus, procedures that specify the required degree of homogenization of soil samples prior to analysis are needed. Sieving after milling could be an option, but that should be tested. The soils spiked with Fe(III) phosphate, potassium phosphate and phytic acid had similar LIBS-P, except for soils with hydroxyapatite, which resulted in markedly lower response. These results suggested that matrix effects related to the chemical nature of phosphorus would be minor for non-calcareous soils in humid regions, where apatites comprise only a small fraction of total phosphorus. Strategies to overcome matrix effects related to particle size and content of apatite-phosphorus by combining multivariate models and soil type groupings should be further investigated.
