Whole Rock Major Element Analysis in the Field Using Laser-Induced Breakdown Spectroscopy

Minimal sample preparation was sufficient to obtain accurate predictions for major element concentrations in 24 lava samples with varied compositions (45 – 80% SiO₂) using Laser Induced Breakdown Spectroscopy (LIBS) and multivariate analysis. Sample preparation consisted of cutting the samples in order to produce a flat surface for laser shots and removing unwanted loose particles and fingerprints with water and ethanol. The LIBS data collection process used an Ocean Optics 2500 LIBS system with a Nd-YAG laser (1064 nm wavelength) to ablate a small amount of material from the sample surface, forming a plasma whose emissions were used to generate a spectrum for the sample by diffracting with a series of seven Echelle gratings in Czerny-Turner design. In order to approximate the homogeneity of powders used in XRF analysis, the data set for each sample consists of spectra from 600 random locations on the sample surface. These were then reduced to six averages of 100 shots each to account for the shot to shot variability inherent in the LIBS technique. A Partial Least Squares (PLS) regression of this LIBS data using multivariate analysis and the program Unscrambler predicted whole rock major element concentrations with reasonable accuracy. Predicted concentrations were compared to those obtained using XRF to produce R² values of 0.89 for SiO₂, 0.77 for TiO₂, 0.74 for Al₂O₃, 0.86 for Fe₂O₃, 0.78 for MnO, 0.74 for MgO, 0.89 for CaO, 0.84 for Na₂O, 0.88 for K₂O, and 0.71 for P₂O₅ (all with intercepts near zero). These results indicate that this method of minimal sample preparation, coupled with short data collection and processing times is ideal for the determination of whole rock major element analysis in the field.