Review of “pyisotopomer: A Python package for obtaining intramolecular isotope ratio differences from mass spectrometric analysis of nitrous oxide isotopocules” by Kelly et al.

Summary

The accurate determination of intramolecular ‘site preference’ values for 15N/14N in the nitrous oxide molecule by isotope ratio mass spectrometry requires careful consideration of molecular rearrangement that can occur during ionization or ‘scrambling’. This phenomenon has been well acknowledged in the field of N2O isotope analysis, but the approaches for determining and/or applying corrections for this scrambling have varied and have not been systematically evaluated or adopted. This manuscript takes one of the previous approaches for evaluating scrambling (using two scrambling coefficients, k and g) and carefully evaluates how these parameters should be reliably determined, how their variations influence calculated and reported isotope ratios and how this determination may be limited given existing differences in available reference materials. Finally, the manuscript presents results from a two-lab intercomparison exercise that includes field samples collected from a lacustrine water column. The workflow and Python scripts presented are intended to be adopted by the community as a best practice and should help to normalize variations in how different labs generate such data.

Overall, I found this paper to be dense – but readable – and valuable to the field in multiple ways. I have no problem with it being published as is. I see in the revision history that two previous reviews have weighed heavily into the evolution of the arguments laid out in this paper. The effort that has gone into now demonstrating the validity of these approaches and the inclusion of the analytical algebraic solution is clear and valuable. In the history of papers working out analytical systematics of N2O isotopes – this paper should stand as a strong milestone for a long while.

Major comments
The topic presented is highly complex and specialized. Nevertheless, the manuscript is very well written – with a high attention to detail. I appreciated the careful wording in most places that adequately explains the logic behind each of the approaches and the derivations of the formulae underlying these approaches.

For example, I appreciate the detail explaining the linearity correction (‘dummy variable method’) approach. Notably, the situation can arise wherein an analytical blank (of constant composition) will impact the observed ‘size linearity’ relationship (perceived slope) depending on the composition of the reference material (or sample). Although this is briefly acknowledged in the conclusion (L637), it might be useful to point this out in section 3.2.1 and explain that the slopes estimated here are only calculable above sample sizes for which the relative size of the blank is negligible. Secondly, I am also curious whether this approach has been used by others in other isotope correction regimes, as it seems applicable to other isotopic workflows as well.

I also appreciated the practical exercise of determining how compositionally distinct two reference materials are required to be (in terms of SP) to be relevant for determining scrambling coefficients. This paper makes a strong case for dissemination of reference materials having a very wide range of SP values. For example, the recent development of USGS 51 and USGS 52, which have SP values differing by ~28‰, may not be appropriate for determination of N2O from environments where SP values range much more widely. Are the reference materials used in this study available to other research groups? It might be useful to point out where these materials can be obtained.

I also greatly appreciated the distinction between the importance of absolute values of k and g versus the absolute difference between them (g - k). Although I had come to this realization in my own work (wrestling with N2O isotope analyses), I had never evaluated these dynamics in such detail. The Monte Carlo analyses clearly demonstrate the point.

The evidence presented for the importance of evaluating scrambling on a regular basis was also eye-opening and should serve as an important benchmark for steps to be taken by labs performing such measurements on a regular basis.

Given the thorough exploration of scrambling parameters in two distinct labs, and the relatively large differences in scrambling coefficients of these labs (and the differences between the two coefficients) – are there any conclusions to be drawn about factors that control scrambling (source parameters, such as accelerating voltages, box current, trap current, electron volts, etc.)? I wonder similarly whether any conclusions can be drawn about optimization of fragmentation and scrambling in the ion source from these datasets?

Minor comments:

L87-90: consider numbering this list as in 1)… 2)… and 3)…

L92: implementing ‘any’ of the above isotopomer calibrations?

Eq 11 – missing parentheses in denominator

L388: Missing Reference; also line 513; also line 518

L401: A shift in average g and k values is noted for Lab 2 between Aug and November 2020. However, these averages do not appear to be statistically different (overlapping at the 1 stdev level). Thus, I wonder what the relevance is of calling them ‘shifted.’

L414: Please explain in words exactly what ‘d’ is. How should this parameter be conceptualized. I think this will help the reader follow equations 22 and 23.

L520: missing period.

L580: four?

L627-628: should this be nM instead of nmol?

Finally, the title of the paper introduces a software package (pyisotopomer) – yet we learn very little about the actual software in the paper. Rather, the paper walks the reader through a careful analysis of how scrambling can and should be determined and demonstrates the sensitivity of N2O isotopocule analysis to these scrambling parameters. I might suggest considering a title that reflects the content of the paper more directly?