Pyrohydrolysis-IRMS determination of silicate chlorine stable isotope compositions. Application to oceanic crust and meteorite samples

This contribution describes the optimization of chlorine extraction from silicate samples by pyrohydrolysis prior to the precise determination of Cl stable-isotope compositions ( δ 37Cl) by gas source, dual inlet Isotope Ratio Mass Spectrometry (IRMS) on CH 3Clg. The complete method was checked on three international reference materials for Cl-content and two laboratory glass standards. Whole procedure blanks are lower than 0.5 μmol, corresponding to less than 10 wt.% of most of the sample chloride analysed. In the absence of international chlorine isotope rock, we report here Cl extracted compared to accepted Cl contents and reproducibilities on Cl and δ 37Cl measurements for the standard rocks. After extraction, the Cl contents of the three international references compared within error with the accepted values (mean yield = 94 ± 10%) with reproducibilities better than 12% (1 σ). The laboratory glass standards – andesite SO100DS92 and phonolite S9(2) – were used specifically to test the effect of chloride amount on the measurements. They gave Cl extraction yields of 100 ± 6% (1 σ; n = 15) and 105 ± 8% (1 σ; n = 7), respectively, with δ 37Cl values of − 0.51 ± 0.14‰ and − 0.39 ± 0.17‰ (1 σ). In summary, for silicate samples with Cl contents between 39 and 9042 ppm, the Pyrohydrolysis/HPLC method leads to overall Cl extraction yields of 100 ± 8%, reproducibilities on Cl contents of 7% and on δ 37Cl measurements of ± 0.12‰ (all 1 σ). The method was further applied to ten silicate rocks of various mineralogy and chemistry (meteorite, fresh MORB glasses, altered basalts and serpentinized peridotites) chosen for their large range of Cl contents (70–2156 ppm) and their geological significance. δ 37Cl values range between − 2.33 and − 0.50‰. These strictly negative values contrast with the large range and mainly positive values previously reported for comparable silicate samples and shown here to be affected by analytical problems. Thus we propose a preliminary, revised terrestrial Cl cycle, mainly dominated by negative and zero δ 37Cl values.