Rationale

Water stable isotope ratios (δ²H and δ¹⁸O values) are widely used tracers in environmental studies; hence, accurate and precise assays are required for providing sound scientific information. We tested the analytical performance of 235 international laboratories conducting water isotope analyses using dual-inlet and continuous-flow isotope ratio mass spectrometers and laser spectrometers through a water isotope inter-comparison test.

Methods

Eight test water samples were distributed by the IAEA to international stable isotope laboratories. These consisted of a core set of five samples spanning the common δ-range of natural waters, and three optional samples (highly depleted, enriched, and saline). The fifth core sample contained unrevealed trace methanol to assess analyst vigilance to the impact of organic contamination on water isotopic measurements made by all instrument technologies.

Results

For the core and optional samples ~73 % of laboratories gave acceptable results within 0.2 ‰ and 1.5 ‰ of the reference values for δ¹⁸O and δ²H, respectively; ~27 % produced unacceptable results. Top performance for δ¹⁸O values was dominated by dual-inlet IRMS laboratories; top performance for δ²H values was led by laser spectrometer laboratories. Continuous-flow instruments yielded comparatively intermediate results. Trace methanol contamination of water resulted in extreme outlier δ-values for laser instruments, but also affected reactor-based continuous-flow IRMS systems; however, dual-inlet IRMS δ-values were unaffected.

Conclusions

Analysis of the laboratory results and their metadata suggested inaccurate or imprecise performance stemmed mainly from skill- and knowledge-based errors including: calculation mistakes, inappropriate or compromised laboratory calibration standards, poorly performing instrumentation, lack of vigilance to contamination, or inattention to unreasonable isotopic outcomes. To counteract common errors, we recommend that laboratories include 1–2 'known' control standards in all autoruns; laser laboratories should screen each autorun for spectral contamination; and all laboratories should evaluate whether derived d-excess values are realistic when both isotope ratios are measured. Combined, these data evaluation strategies should immediately inform the laboratory about fundamental mistakes or compromised samples.