Performance of induction module cavity ring-down spectroscopy (IM-CRDS) for measuring δ18O and δ2H values of soil, stem, and leaf waters

RATIONALE: Induction module cavity ring-down spectroscopy (IM-CRDS) has been proposed as a rapid and cost-effective alternative to cryogenic vacuum distillation (CVD) and isotope ratio mass spectrometry (IRMS) for the measurement of delta O-18 and delta H-2 values in matrix-bound waters. In the current study, we characterized the performance of IM-CRDS relative to CVD and IRMS and investigated the mechanisms responsible for differences between the methods. METHODS: We collected a set of 75 soil, stem, and leaf water samples, and measured the d18O and d2H values of each sample with four techniques: CVD and IRMS, CVD and CRDS, CVD and IM-CRDS, and IM-CRDS alone. We then calculated the isotopic errors for each of the three CRDS methods relative to CVD and IRMS, and analyzed the relationships among these errors and suites of diagnostic spectral parameters that are indicative of organic contamination. RESULTS: The IM-CRDS technique accurately assessed the delta O-18 and delta H-2 values of pure waters, but exhibited progressively increasing errors for soil waters, stem waters, and leaf waters. For soils, the errors were attributable to subsampling of isotopically heterogeneous source material, whereas for stems and leaves, they were attributable to spectral interference. Unexpectedly, the magnitude of spectral interference was higher for the solid samples analyzed directly via IM-CRDS than for those originally extracted via CVD and then analyzed by IM-CRDS. CONCLUSIONS: There are many types of matrix-bound water samples for which IM-CRDS measurements include significant errors from spectral interference. As a result, spectral analysis and validation should be incorporated into IM-CRDS post-processing procedures. In the future, IM-CRDS performance could be improved through: (i) identification of the compounds that cause spectral interference, and either (ii) modification of the combustion step to completely oxidize these compounds to CO2, and/or (iii) incorporation of corrections for these compounds into the spectral fitting models used by the CRDS analyzers. Copyright (C) 2016 John Wiley & Sons, Ltd.