A new analytical method for measuring hydrogen isotopes using GC-IRMS: applications to hydrous minerals

A new analytical method is proposed for measuring the deuterium to hydrogen ratio (D/H) of non-stoichiometric water in hydrous minerals via pyrolysis facilitated gas-chromatography - isotope ratio mass spectrometry (GC-IRMS). Previously published analytical methods have reported a poorly understood nonlinear dependence of D/H on sample size, for which any accurate correction is difficult. This sample size effect been variously attributed to kinetic isotope fractionation within the mass spectrometer and peripheral instruments, ion source linearity issues, and an unstable H_3^+-factor or incorrect H_3^+-factor calculations. The cause of the sample size effect is here identified by examinations of individual chromatograms as well as bulk data from chromatographic peaks. It is here determined that it is primarily an artifact of the calculations employed by the manufacturer's computer program, used to both monitor the functions of the mass spectrometer and to collect data. Ancillary causes of the sample size effect include a combination of persistent background interferences and chromatographic separation of the isotopologues of molecular hydrogen. Previously published methods are evaluated in light of these findings. A new method of H_3^+-factor and D/H calculation is proposed which makes portions of the Isodat software as well as other published calculation methods unnecessary. Using this new method, D/H is measured in non-stoichiometric water in chert from the Cretaceous Edwards Group, Texas, as well as the Precambrian Kromberg Formation, South Africa, to assess hydrological conditions as well as to estimate the maximum average surface temperature during precipitation of the chert. Data from Cretaceous chert are consistent with previously published data and interpretations, based upon conventional analyses of large samples. Data from Precambrian chert are consistent with maximum average surface temperatures approaching 65°C during the Archean, instead of the much lower temperatures derived from erroneous methods of sample preparation and analysis. D/H is likewise measured in non-stoichiometric water in silicified basalt from the Precambrian Hooggenoeg Complex, South Africa. Data are shown to be consistent with D/H of the Archean ocean similar to present day values.