Hydrothermal 15 N 15 N abundances constrain the origins of mantle nitrogen.

Nitrogen is the main constituent of the Earth's atmosphere, but its provenance in the Earth's mantle remains uncertain. The relative contribution of primordial nitrogen inherited during the Earth's accretion versus that subducted from the Earth's surface is unclear ^1-6 . Here we show that the mantle may have retained remnants of such primordial nitrogen. We use the rare ¹⁵ N ¹⁵ N isotopologue of N ₂ as a new tracer of air contamination in volcanic gas effusions. By constraining air contamination in gases from Iceland, Eifel (Germany) and Yellowstone (USA), we derive estimates of mantle δ ¹⁵ N (the fractional difference in ¹⁵ N/ ¹⁴ N from air), N ₂ / ³⁶ Ar and N ₂ / ³ He. Our results show that negative δ ¹⁵ N values observed in gases, previously regarded as indicating a mantle origin for nitrogen ^7-10 , in fact represent dominantly air-derived N ₂ that experienced ¹⁵ N/ ¹⁴ N fractionation in hydrothermal systems. Using two-component mixing models to correct for this effect, the ¹⁵ N ¹⁵ N data allow extrapolations that characterize mantle endmember δ ¹⁵ N, N ₂ / ³⁶ Ar and N ₂ / ³ He values. We show that the Eifel region has slightly increased δ ¹⁵ N and N ₂ / ³⁶ Ar values relative to estimates for the convective mantle provided by mid-ocean-ridge basalts ¹¹ , consistent with subducted nitrogen being added to the mantle source. In contrast, we find that whereas the Yellowstone plume has δ ¹⁵ N values substantially greater than that of the convective mantle, resembling surface components ^12-15 , its N ₂ / ³⁶ Ar and N ₂ / ³ He ratios are indistinguishable from those of the convective mantle. This observation raises the possibility that the plume hosts a primordial component. We provide a test of the subduction hypothesis with a two-box model, describing the evolution of mantle and surface nitrogen through geological time. We show that the effect of subduction on the deep nitrogen cycle may be less important than has been suggested by previous investigations. We propose instead that high mid-ocean-ridge basalt and plume δ ¹⁵ N values may both be dominantly primordial features.