Ice cores represent the only direct paleo-atmospheric archive that allow the reconstruction of greenhouse gas concentrations such as N₂O. However, processes in the ice can alter the atmospheric information stored in air bubbles, for example by adding extra N₂O by in situ production. This in situ production of N₂O is especially severe in mineral dust-rich ice core sections corresponding to glacial periods. Understanding the production process and its link to the mineral dust content is key to systematically detecting altered samples and correcting for the in situ contribution. Isotope analysis is particularly useful for characterizing these processes and thus isolating the paleoclimatic signal from archived data. 

We measured the bulk nitrogen and oxygen isotopic composition of N₂O in Antarctic and Greenland ice cores from glacial periods. The isotopic signatures of N₂O produced in situ, calculated using a mass balance approach, differ from that of the atmospheric N₂O. In addition, enrichment or depletion in ¹⁵N and/or ¹⁸O relative to atmospheric values varies with drilling site, snow accumulation rate, and properties of the snow-ice transition. Interestingly, isotopic signatures of nitrate (NO₃^-) exhibit similar dependencies. It is well established that NO₃^- is drastically altered by post-depositional processes in low accumulation areas. Joint isotopic analysis of N₂O and NO₃^- in samples from the EDC and EDML ice cores revealed a correlation between δ¹⁵N values of NO₃^- and in situ N₂O, pointing to NO₃^- as a potential precursor for in situ production. While being linearly correlated, the nitrogen isotopic signature of NO₃^- is twice as enriched as in situ produced N₂O. This suggests that the two N atoms of N₂O originate from two distinct sources and only one is likely derived from nitrate.

We additionally measured the site preference of ¹⁵N in N₂O in ice core samples (SP = δ¹⁵N_α - δ¹⁵N_β, where α is the central and β the terminal N atom in the N₂O molecule). Previous work on SP suggests that SP might be indicative of the N₂O formation pathway provided both N atoms are derived from the same N precursor. The SP signature in Vostok samples ranges from +57 to +242 ‰, and the δ¹⁵N_α values from +92 to +234 ‰, which is comparable to the δ¹⁵N values of NO₃^- at Vostok. Although similar reaction pathways were expected in different ice cores, in situ N₂O from Taylor Glacier samples exhibits very different SP values from -17 to -7 ‰, with δ¹⁵N_α values from -45 to -32 ‰. Given that the difference in δ¹⁵N of NO₃^- is also up to 200 ‰ between these two locations, our findings suggest that the center-position nitrogen (α) of in situ N₂O comes from NO₃^- and the terminal-position nitrogen (β) from another N-bearing compound. Thus, the SP signature seems to reflect not the N₂O formation pathway but the difference in δ¹⁵N of the two nitrogen pools involved in the reaction.

Gaining a thorough understanding of the N₂O production in ice marks a significant advancement towards interpretation of the N₂O record and possibly correction for in situ production.