Accuracy of plagioclase compositions from laboratory and Mars spacecraft thermal emission spectra

Plagioclase, among the most abundant minerals in terrestrial and martian volcanic rocks, exhibits a range of compositions that reflect changing conditions of lava during crystallizaton. Average plagioclase compositions of most volcanic rocks are often more sofic than the median of the compositional range, due to the volumetric sub ordinance of calcic plagioclase phenocrysts to more sodic plagioclase in the groundmass. Thermal emission spectrometers (TES, THEMIS, and Mini-TES) onboard Mars spacecraft (Mars Global Surveyor, Mars Odyssey, Mars Exploration Rovers respectively) now provide a means of determining average plagioclase compositions of martian rocks directly.

This study demonstrates that spectrally-modeled average plagioclase compositions in terrestrial basalts, andesites, and dacites correspond approximately to measured values (estimated from the weighted average of electron microprobe analyses of phenocrysts and groundmass grains, and from calculated normative plagioclase compositions). Linear deconvolution of the thermal emission spectra of volcanic rocks can generally model plagioclase compositions to within +10/-6 An (Ca/(Ca+Na)) and +13/-14 An of weighted average and normative plagioclase compositions respectively.

Analyses of spectra from two-component plagioclase sand mixtures (whose plagioclase components varied by volume and composition) have provided additional insight into the role of plagioclase zoning (TES spectra mostly reflect spectral contributions from sand-size particles, which could have been derived by comminution of larger zoned grains). Deconvolutions involving different plagioclase endmember sets produce modeled average plagioclase compositions that closely represent measured values. Considering the variability in types of observed plagioclase zoning patterns in terrestrial volcanic rocks, normal zoning (calcic cores and sodic rims) cannot be assumed and thus does not account for the apparent sodic bias in spectrally modeled compositions. The role of zoning is minimal when compared to the influence of more abundant sodic plagioclase in groundmass. Sand mixture deconvolutions also predict that martian spacecraft TES, THEMIS, and Mini-TES instruments (with variable spatial and spectral resolutions) can model plagioclase compositions to within +3/-8 An, +11/-6 An, and +3/-8 An (respectively) of measured values in plagioclase mixtures not involving albite. The presence of other phases apparently increases uncertainties, as observed in rock spectra. Nevertheless, this study provides an increased level of confidence in accurately determining plagioclase compositions from Mars thermal emission spectra.