Oxygen isotope analysis of carbonates in the calcite-dolomite-magnesite solid-solution by high-temperature pyrolysis:...

Accurate and efficient measurement of the oxygen isotope composition of carbonates (delta(C) (18)O) based on the mass spectrometric analysis of CO(2) produced by reacting carbonate samples with H(3)PO(4) is compromised by: (1) uncertainties associated with fractionation factors (alpha(CO)(2)C) used to correct measured oxygen isotope values of CO(2)(delta(CO(2)(18)O) to delta(C) (18)O; and (2) the slow reaction rates of many carbonates of geological and environmental interest with H(3)PO(4). In contrast, determination of delta(C) (18)O from analysis of CO produced by high-temperature (>1400 degrees C) pyrolytic reduction, using an elemental analyser coupled to continuous-flow isotope-ratio mass spectrometry (TC/EA CF-IRMS), offers a potentially efficient alternative that measures the isotopic composition of total carbonate oxygen and should, therefore, theoretically be free of fractionation effects. The utility of the TC/EA CF-IRMS technique was tested by analysis of carbonates in the calcite-dolomite-magnesite solid-solution and comparing the results with delta(C) (18)O measured by conventional thermal decomposition/fluorination (TDF) on the same materials. Initial results show that CO yields are dependent on both the chemical composition of the carbonate and the specific pyrolysis conditions. Low gas yields (<100% of predicted yield) are associated with positive (>+0.2 per thousand) deviations in delta(C(TC/EA) (18)O compared with delta(C(TDF) (18)O. At a pyrolysis temperature of 1420 degrees C the difference between delta(C) (18)O measured by TC/EA CF-IRMS and TDF (Delta(C(TC/EA,TDF) (18)O) was found to be negatively correlated with gas yield (r = -0.785) and this suggests that delta(C) (18)O values (with an estimated combined standard uncertainty of +/-0.38 per thousand) could be derived by applying a yield-dependent correction. Increasing the pyrolysis temperature to 1500 degrees C also resulted in a statistically significant correlation with gas yield (r = -0.601), indicating that delta(C) (18)O values (with an estimated uncertainty of +/-0.43 per thousand) could again be corrected using a yield-dependent procedure. Despite significant uncertainty associated with TC/EA CF-IRMS analysis, the magnitude of the uncertainty is similar to that associated with the application of poorly defined values of alpha(CO)(2), (C) used to derive delta(C) (18)O from delta(CO(2) (18)O measured by the H(3)PO(4) method for most common carbonate phases. Consequently, TC/EA CF-IRMS could provide a rapid alternative for the analysis of these phases without any effective deterioration in relative accuracy, while analytical precision could be improved by increasing the number of replicate analyses for both calibration standards and samples. Although automated gas preparation techniques based on the H(3)PO(4) method (ISOCARB, Kiel device, Gas-Bench systems) have the potential to measure delta(CO)(2) (18)O efficiently for specific, slowly reacting phases (e.g. dolomite), problems associated with poorly defined alpha(CO)(2), (C) remain. The application of the Principle of Identical Treatment is not a solution to the analysis of these phases because it assumes that a single fractionation factor may be defined for each phase within a solid-solution regardless of its precise chemical composition. This assumption has yet to be tested adequately.