Characterization of the analytical performance of δ 15 N and δ 18 O measurements by the silver nitrate method in the framework of nitrate source apportioning

Nitrate pollution represents one of the most important issues for ground and surface water quality and source identification is essential for developing effective mitigation practices. Nitrate isotopic fingerprinting can be utilized to identify the sources of nitrate pollution in aquifers. However, it is crucial to assess the performances (precision and accuracy) of the analytical procedure applied to measure the δ N and δ O values of nitrates from field samples to correctly apply this tool. Nitrates were extracted from a large number of KNO samples using the AgNO method, and the δ N and δ O values of these nitrate extracts were measured by isotope ratio mass spectrometry. The availability of this dataset, comprising 693 unprocessed quality control (QC) KNO samples and 618 processed samples, allowed us to rigorously quantify the performance of the procedures employed. A salt doping experiment was also performed from which the effects of contaminants on the performance of the method could be ascertained. The overall instrumental reproducibility for the analysis of unprocessed QC samples was 0.5‰ and 2‰ for δ N and δ O values, respectively, and a strict dependence on signal amplitude was observed. No isotope fractionation was reported for reference samples that were processed according to the "identical treatment" principle (ITP) but normalized by unprocessed reference materials. A significant increase in the standard deviation (SD) was, however, observed compared with that for unprocessed samples. The SD of the processed QC samples allowed us to quantify the reproducibility of the entire procedure as 0.6‰ and 1.0‰ for δ N and δ O values, respectively. This was comparable with the system reproducibility when normalization using processed reference materials was applied according to the ITP. Normalization with processed standards is essential to achieve high-precision measurements of the δ N and δ O values of nitrates extracted from unknown samples. This procedure allowed good accuracy to be guaranteed, and precision levels comparable with the observed instrumental performance to be achieved. A salt doping experiment showed a significant influence of the SO content on the δ N values.