Quantitative Sulfur Analysis using Stand-off Laser-Induced Breakdown Spectroscopy

The laser-induced breakdown spectrometer (LIBS) in the ChemCam instrument on Mars Science Laboratory has the capability to produce robust, quantitative analyses not only for major elements, but also for a large range of light elements and trace elements that are of great interest to geochemists. However, sulfur presents a particular challenge because it reacts easily with oxygen in the plasma and because the brightest S emission lines lie outside ChemCam's spectral range. This work was undertaken within the context of our larger effort to identify and compensate for matrix effects, which are chemical properties of the material that influence the ratio of a given emission line to the abundance of the element producing that line. Samples for this study include two suites of rocks: a suite of 12 samples that are mixtures of sulfate minerals and host rocks, generally with high S contents (0.1-26.0 wt% S), and a large suite of 118 igneous rocks from varying parageneses with S contents in the 0-2 wt% range. These compositions provide several different types of matrices to challenge our calibration procedures. Samples were analyzed under ChemCam-like conditions: a Nd:YAG laser producing 17 mJ per 10ns pulse was directed onto samples positioned 5-9 m away from the laser and tele­scope. The samples were placed in a vacuum chamber filled with 7 Torr CO2 to replicate the Martian surface pressure as the atmospheric pressure influences the LIBS plasma. Some of the LIBS plasma emission is collected with a telescope and transmitted through a 1 m, 300 um, 0.22NA optical fiber connected to a commercial Ocean Optics spectrometer. We are testing and comparing three different strategies to evaluate sulfur contents. 1) We have calculated regression lines comparing the intensity at each channel to the S content. This analysis shows that there are dozens of S emission lines in the ChemCam wavelength range that are suitable for use in quantitative analysis, even in the presence of Fe. 2) Partial least-squares analyses of these data show that S can be predicted with better than 10% accuracy, even when present at levels <0.15 wt%. 3) When peaks in the spectra are fit, the resultant peak areas can be regressed against concentration using step-wise multiple regression analysis to determine which subset of S lines gives the most accurate concentrations. All three methods of calibration show that excellent S analyses can be produced under Mars conditions at stand-off distances of up to 9 m.