Towards in-situ chemical classification of seafloor deposits: Application of neural networks to underwater laser-induced breakdown spectroscopy

Laser-induced breakdown spectroscopy (LIBS) is a form of chemical analysis that can determine the elemental composition of targets. LIBS has been used for underwater exploration, at over 1000 m depth in active vent fields, and it can potentially form the basis of an efficient screening method prior to detailed sampling or boring surveys. In this study, a method for in-situ chemical classification of hydrothermal deposits, based on metallic element compositions using Cu-Pb-Zn ternary diagrams, is developed that uses Artificial Neural Networks (ANNs). ANNs have the advantage of being able to describe non-linear properties, which is relevant when modeling the spectra fluctuations associated with underwater LIBS. We analyzed the effect of database size and constitution on the classification results with spectra that were both simulated and measured in the laboratory. In our future work, we develop classifiers that use other elements, and extend this method for multi-elemental quantitative analysis.

[1]  Robert L. Tokar,et al.  Pre-flight calibration and initial data processing for the ChemCam laser-induced breakdown spectroscopy instrument on the Mars Science Laboratory rover , 2013 .

[2]  S. Mahadevan,et al.  A study of machine learning regression methods for major elemental analysis of rocks using laser-induced breakdown spectroscopy , 2015 .

[3]  Aaas News,et al.  Book Reviews , 1893, Buffalo Medical and Surgical Journal.

[4]  T. Takahashi,et al.  Development and field testing of laser-induced breakdown spectroscopy for in situ multi-element analysis at sea , 2012, 2012 Oceans.

[5]  Blair Thornton,et al.  Development of a deep-sea laser-induced breakdown spectrometer for in situ multi-element chemical analysis , 2015 .

[6]  Steven J. Spencer,et al.  Comparison of principal components regression, partial least squares regression, multi-block partial least squares regression, and serial partial least squares regression algorithms for the analysis of Fe in iron ore using LIBS , 2012 .

[7]  Kazuo Ishii,et al.  Support vector machine based classification of seafloor rock types measured underwater using Laser Induced Breakdown Spectroscopy , 2016, OCEANS 2016 - Shanghai.

[8]  Y. Ogata,et al.  On-site quantitative elemental analysis of metal ions in aqueous solutions by underwater laser-induced breakdown spectroscopy combined with electrodeposition under controlled potential. , 2015, Analytical chemistry.

[9]  Blair Thornton,et al.  Calibration-free analysis of immersed brass alloys using long-ns-duration pulse laser-induced breakdown spectroscopy with and without correction for nonstoichiometric ablation , 2015 .

[10]  D. L. Death,et al.  Multi-element and mineralogical analysis of mineral ores using laser induced breakdown spectroscopy and chemometric analysis , 2009 .

[11]  Ross R. Large,et al.  Australian volcanic-hosted massive sulfide deposits; features, styles, and genetic models , 1992 .

[12]  Tamaki Ura,et al.  Spectroscopy and imaging of laser-induced plasmas for chemical analysis of bulk aqueous solutions at high pressures , 2011, OCEANS'11 MTS/IEEE KONA.

[13]  G. G. Stokes.  "J." , 1890, The New Yale Book of Quotations.

[14]  W. C. Martin,et al.  Handbook of Basic Atomic Spectroscopic Data , 2005 .

[15]  Grgoire Montavon,et al.  Neural Networks: Tricks of the Trade , 2012, Lecture Notes in Computer Science.

[16]  Blair Thornton,et al.  Temperature based segmentation for spectral data of laser-induced plasmas for quantitative compositional analysis of brass alloys submerged in water , 2016 .

[17]  Lionel Canioni,et al.  Artificial neural network for on-site quantitative analysis of soils using laser induced breakdown spectroscopy , 2013 .

[18]  Peter D. Wentzell,et al.  Comparison of principal components regression and partial least squares regression through generic simulations of complex mixtures , 2003 .