Organic contamination detection for isotopic analysis of water by laser spectroscopy.

Rationale: Hydrogen and oxygen stable isotope ratios (δ H, δ O, and δ O values) are commonly used tracers of water. These ratios can be measured by isotope ratio infrared spectroscopy (IRIS). However, IRIS approaches are prone to errors induced by organic compounds present in plant, soil, and natural water samples. A novel approach using O-excess values has shown promise for flagging spectrally contaminated plant samples during IRIS analysis. A systematic assessment of this flagging system is needed to prove it useful.

Methods: Errors induced by methanol and ethanol water mixtures on measured IRIS and isotope ratio mass spectrometry (IRMS) results were evaluated. For IRIS analyses both liquid- and vapour-mode (via direct vapour equilibration) methods are used. The δ H, δ O, and δ O values were measured and compared with known reference values to determine the errors induced by methanol and ethanol contamination. In addition, the O-excess contamination detection approach was tested. This is a post-processing detection tool for both liquid and vapour IRIS triple-isotope analyses, utilizing calculated O-excess values to flag contaminated samples.

Results: Organic contamination induced significant errors in IRIS results, not seen in IRMS results. Methanol caused larger errors than ethanol. Results from vapour-IRIS analyses had larger errors than those from liquid-IRIS analyses. The O-excess approach identified methanol driven error in liquid- and vapour-mode IRIS samples at levels where isotope results became unacceptably erroneous. For ethanol contaminated samples, a mix of erroneous and correct flagging occurred with the O-excess method. Our results indicate that methanol is the more problematic contaminant for data corruption. The O-excess method was therefore useful for data quality control.

Conclusions: Organic contamination caused significant errors in IRIS stable isotope results. These errors were larger during vapour analyses than during liquid IRIS analyses, and larger for methanol than ethanol contamination. The O-excess method is highly sensitive for detecting narrowband (methanol) contamination error in vapour and liquid analysis modes in IRIS.