Quantitative in situ mapping of elements in deep-sea hydrothermal vents using laser-induced breakdown spectroscopy and multivariate analysis

This study describes a method to quantify the chemical composition of deep-sea hydrothermal deposits in situ using laser-induced breakdown spectroscopy (LIBS). Partial least squares (PLS) regression analysis is applied to spectra obtained using a long laser pulse with a duration of 150 ns. The number of measurements needed to address the spatial heterogeneity of samples is determined through high-resolution mapping of the elemental distribution in rock samples. PLS applied to laboratory measured seawater-submerged samples achieved an average relative error (RE) of 25% for Cu, Pb, and Zn compared to benchmark concentration values in cross-validation and validation studies, where both the benchmark concentration values and LIBS spectral data are made available with this publication. The PLS model was applied to LIBS signals obtained in situ from hydrothermal deposits at 1000 m depth in the ocean. The results show that target inhomogeneity limits the accuracy of the surface LIBS measurements compared to benchmark values from bulk analysis of samples. Making multiple measurements with small position offsets at each location improves the accuracy of estimates compared to an equivalent number of measurements at a single position. Maps of element distribution generated using quantified in situ data demonstrate how chemical survey outputs can be generated by combining LIBS with multivariate analysis. This enables real-time chemical feedback during deep-sea operations and chemical surveys in situations or with platforms where sample recovery is not possible. •In situ quantification of Cu, Pb, and Zn in deep-sea minerals has been demonstrated by LIBS and multivariate analysis.•An average relative error of 25% was achieved for all 3 elements that were quantified for powder submerged bulk pellets.•Rock surface measurements are more susceptible to inhomogeneity than pellets with both XRF and LIBS measurements showing double the variation.•The representation of bulk composition can be improved by moving the measurement location over small distance intervals.