Physics
Scientific paper
Dec 2010
adsabs.harvard.edu/cgi-bin/nph-data_query?bibcode=2010agufm.p54a..07d&link_type=abstract
American Geophysical Union, Fall Meeting 2010, abstract #P54A-07
Physics
[1060] Geochemistry / Planetary Geochemistry, [1065] Geochemistry / Major And Trace Element Geochemistry, [5410] Planetary Sciences: Solid Surface Planets / Composition, [5494] Planetary Sciences: Solid Surface Planets / Instruments And Techniques
Scientific paper
The ChemCam instrument on Mars Science Laboratory will be the first deployment of laser-induced breakdown spectroscopy (LIBS) for remote geochemical analysis. Successful quantitative analyses of those results will use in-situ calibration targets and laboratory calibrations, and employ sophisticated algorithms for data reduction in order to correct for variations in peak intensities and areas caused by interactions in the plasma that are a function of chemical composition. Such chemical matrix effects influence the ratio of each emission line to the abundance of the element that produces it, and are directly related to the elemental composition of the sample. Advances in statistical analysis of LIBS data that mitigate matrix effects and provide for accurate and precise bulk analysis of major, minor, and trace elements are reported here. Our in-house data set currently includes LIBS spectra of >140 rock powders (igneous, metamorphic, and sedimentary) with highly-varying compositions (as determined by XRF) that were acquired at 7-9 m standoff distance under Mars atmospheric conditions using a laboratory instrument [1]. LIBS spectra were modeled using partial least squares analysis (PLS) to predict elemental compositions. Within the igneous suite, 10 repeat measurements of a single sample demonstrates consistency and precision; calculated 1-σ errors were 1.6 wt.%SiO2, 1.5 wt.% Al2O3, 0.4 wt.% TiO2, 1.2 wt.% Fe2O3T, 1.6 wt.% MgO, 0.02 wt.% MnO, 1.1 wt.% CaO, 0.5 wt.% Na2O, 0.2 wt.% P2O5, and 0.4 wt.% K2O. In the overall suite, predictions of all elements, expressed as root mean square errors (RMSEP), are better than ±2.45 for SiO2, ±1.64 for Al2O3, ±0.38 for TiO2, ±1.50 for Fe2O3T, ±1.88 for MgO, ±0.03 for MnO, ±0.82 for CaO, ±0.55 for K2O, ±0.62 for Na2O, and ±0.24 for P2O5 in units of wt.% oxides. On-going work should reduce these values even further. For elements at low concentrations, multivariate analyses must be interpreted with care because their spectral lines are weak relative to the strong lines from dominating major elements. In such cases, either the weak peaks must be individually fit (as for S; see Dyar et al. [2]) or the PLS predictions will rely on correlations between the low-abundance element and major elements. For example, PLS predicts the concentration of Rb with an RMSEP of ±23 ppm, but this prediction relies heavily on K lines that are better predictors of Rb due to the near-perfect correlation between K and Rb in the igneous rocks that dominate our calibration suite. Because any multivariate calibration based on rocks of a certain type will be heavily biased by such geochemical relationships, our in-house laboratory calibration set is expanding with a wider compositional range (and deliberate rock mixtures to counter such correlations) to produce more even more robust predictions of elements with weaker emission lines. [1] Tucker, J.M. et al. (2010) Chem. Geol., in press. [2] Dyar, M.D. et al. (submitted) Spectrochim. Acta B.
Carmosino M. L.
Clegg Samuel M.
Darby Dyar M.
Humphries Stephen
Tucker Jordan
No associations
LandOfFree
Geochemical Predictions of Elemental Compositions using Remote LIBS under Mars Conditions does not yet have a rating. At this time, there are no reviews or comments for this scientific paper.
If you have personal experience with Geochemical Predictions of Elemental Compositions using Remote LIBS under Mars Conditions, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Geochemical Predictions of Elemental Compositions using Remote LIBS under Mars Conditions will most certainly appreciate the feedback.
Profile ID: LFWR-SCP-O-1499881