The Accuracy of Derived Plagioclase Compositions from Multi-component, Multi-phase Sand Mixtures: Implications for Determining Martian Plagioclase Compositions

Introduction. Plagioclase feldspar has been identified as the dominant mineral phase in the martian crust [1-2] through analysis of Mars Global Surveyor Thermal Emission Spectrometer (TES) data. Martian meteorites , samples of isolated terrain on Mars, contain plagioclase in abundances secondary only to pyroxene. Terrestrial basalts and andesites contain primary plagioclase that is calcic-intermediate (An 50-75) in composition. Sodic plagioclase (>An 30) occurs as a primary phase in felsic and alkali-rich volcanic rocks and as an alteration product in metamorphosed and hydrother-mally-altered rocks. Primary plagioclase compositions vary with the extent and type of crystallization processes and with a variety of emplacement mechanisms. In low-temperature (<200˚C) aqueous environments, plagioclase often converts to clay minerals through hydrolysis. At hydrothermal temperatures, calcic pla-gioclase converts to more sodic compositions (<An 30) through a process known as albitization [3-5]. Our ability to accurately derive feldspar compositions from thermal infrared (TIR) emission spectra of martian surface materials affects our ability to interpret evidence for primary igneous and alteration (low vs. high T) processes that may have been at work on Mars. A previous study [6] demonstrated that average plagioclase compositions of two-component, coarse plagioclase sand mixtures can be derived to within 4 An (1σ standard deviation) at laboratory, TES, and Miniature-Thermal Emission Spectrometer (Mini-TES) spectral resolutions and 6 An (1 σ) at and Thermal Emission Imaging System (THEMIS) resolution, with absolute variations ranging up to 12 An. Additional work with multi-component (multi-solid solution), coarse sand plagioclase mixtures yields similar results, with plagioclase compositions derived to within 4-6 An (1 σ) of measured values at lab, TES, THEMIS, and Mini-TES resolutions with variations of up to 15 An [7]. Both of the above studies used controlled mixtures with known average plagioclase compositions for comparison with derived values. The accuracy is comparable in natural samples. In terrestrial volcanic rocks, compositions have been modeled within 10-15 An [6,8] of average [8], weighted average [6], and CIPW Norm plagioclase compositions [6] at laboratory and TES resolutions. In natural samples, with multiple phases, variable grain size, and complex solid solutional zoning, determination of true average mineral compositions by geo-chemical analysis is problematic [6]. Thus, a compari