Résumé
In order to evaluate the budget of nitrogen buried to the mantle in subduction environments, we performed quantitative and N isotopic analyses of Alpine metamorphic rocks subducted to different depths, and on their non-metamorphic analogues. The samples investigated represent a complete section of the oceanic lithosphere including from bottom to top, serpentinized peridotites, metagabbros, metabasalts and overlying metasediments. A global annual flux of N subducted by metagabbros has been estimated at about 4.2 (+ or - 2.0) X 10 (super 11) g/yr with an average delta (super 15) N-value of 1.8 + or - 0.8 ppm (1sigma ). This flux is about half that of sedimentary rocks, which indicates that gabbros carry a significant portion of the subducted nitrogen. The net budget between subducted N and that outgassed at volcanic arcs indicates that approximately 80% of the subducted N is not recycled to the surface. On a global scale, the total amount of N buried to the mantle via subduction zones is estimated to be 3 times higher (13.2 X 10 (super 11) g/yr) than that released from the mantle via mid-ocean ridges, arc and intraplate volcanoes and back-arc basins. This implies that N contained in Earth surface reservoirs, mainly in the atmosphere, has been progressively transferred and sequestered into the mantle, with a net flux of approximately 9.6 X 10 (super 11) g/yr. Assuming a constant flux of subducted N over the Earth's history suggests that an amount equivalent to the present atmospheric N may have been sequestered into the silicate Earth over a period of 4 billion years. Considering a present day mantle value of approximately -5ppm, an average delta (super 15) N value of subducted nitrogen of about 3.4 + or -1.4 ppm implies that the secular evolution of the mantle delta (super 15) N value should have increased through time, whereas that of the crust and atmosphere should have decreased.