Impact of sample preparation on bitumen content measurement using laser-induced breakdown spectroscopy

The impact of sample preparation on bitumen content measurement using LIBS was investigated by collecting spectra from wet and dry tailings. A multivariate data analysis model was developed using optimal wavelength selection for bitumen content classification and prediction in tailings. Wet tailings can be classified into three classes (low, medium, and high bitumen) with 12.1 % error, while dry tailings have a classification error of 6.1 %. Quantitative analysis showed a bitumen content prediction error of 4.7 % for wet tailings and 8.9 % for dry tailings. Wet tailings showed a 1.8–2.5 times improvement in the limit of detection range compared to dry tailings. Plasma density and crater size measurements revealed that plasma density fluctuation was 2.7 times lower in wet tailings due to consistent crater formation from laser-tailings interaction. The lower plasma density fluctuation indicates a stable mass ablation for wet samples, which is attributed as the primary reason for significant LIBS performance improvement on wet tailings. [Display omitted] •Optimization of multivariate calibration model for improved prediction accuracy of the bitumen measurement in using LIBS.•Bitumen measurement accuracy significantly improved in wet tailings compared to dry tailings.•Plasma density and laser-produced craters are substantially consistent for wet tailings.•Use of compact laser and developed calibration model are promising for the future advancement of portable LIBS.