The accurate analysis of trace concentrations of hydrogen in NAMs is a long-standing problem, with wide-ranging implications in geology and planetology. SIMS and FTIR are two powerful and complementary analytical tools capable of measuring concentrations down to levels of less than 1 ppm H2O. Both methods, however, are subject to matrix effects and rely on other techniques such as manometry or nuclear reaction analysis (NRA) for quantitative calibration. We compared FTIR and SIMS data for a wide variety of NAMs: olivine, orthopyroxene, clinopyroxene, pyrope and grossular garnet, rutile, zircon, kyanite, andalusite, and sillimanite. Some samples were also characterized using high-resolution FE-SEM to assess the potential contribution of submicrocopic inclusions to the analyses. For SIMS, we use high mass resolution (≥5000 MRP) to measure 16O1H, using 30Si and/or 18O as reference isotopes. We use both primary standards, measured independently using manometry or NRA (e.g., [1]), and secondary standards, measured using polarized FTIR referenced back to calibrations developed on primary standards. Our major focus was on on olivine, for which we collected repeated calibration data with both SIMS and NanoSIMS, bracketing measurements of H diffusion profiles in both natural and experimentally annealed crystals at levels of 5-100 ppm H2O. With both instruments we establish low blanks (≤5 ppm) and high precision (typically less than 5% 2-σ errors in 16O1H/30Si), critical requirements for the low concentration levels being measured. Assessment of over 300 analyses on 11 olivines allows us to evaluate the suitability of different standards, several of which are in use in other laboratories [2,3,4]. Seven olivines, with 0-125 ppm H2O, give highly reproducible results and allow us to establish well-constrained calibration slopes with high correlation coefficients (r2 = 0.98-99), in contrast to previous studies [2,3,4]. However, four kimberlitic megacrysts with 140-243 ppm H2O show 16O1H/30Si ratios that vary by up to a factor of 2 even in sequential analyses (cf. [3,4]). A potential cause of these discrepancies is the presence of sub-micron scale pores (as small as 100 nm), which we have documented by SEM. These pores probably contain liquid H2O and/or condensed hydrous phase precipitates. Although the ionization potential of fluids under high-vacuum conditions is an under-studied problem, these sub-micron features may contribute to the measured 16O1H, resulting in analyses with erratic depth profiles and corresponding high uncertainties (up to 15% 2-σ). However, if we omit analyses with such high uncertainties, the data for all olivines fit well together. Our results imply that the Bell et al. IR calibration [1] can be applied accurately to all olivines with IR bands from ~3400-3650 cm-1, without the need for band-specific IR absorption coefficients (cf. [3]). A five-fold total variation was observed among the calibration line slopes for the different minerals analyzed, confirming the need for mineral-specific calibrations.
[1]Bell et al. (2003), JGR, 108 [2] Tenner et al. (2009), Chem. Geol., 262, 42-56 [3] Kovács, I. et al. (2010), Am. Min., 95, 292-299 [4] O'Leary et al. (2010), EPSL