H2O in nominally anhydrous mineral inclusions in diamonds and the volatile composition of diamond forming media

• H 2 O content of 118 nominally anhydrous mineral inclusions in diamonds was studied. • Inclusions in diamonds have lower H 2 O contents than the same minerals in xenoliths. • Eclogitic inclusions have a higher H 2 O content than peridotitic inclusions. • The H 2 O content of inclusions is compatible with silicate or carbonatitic melts. • The H 2 O content of inclusions is compatible with fluids with low H 2 O activity. Many studies have investigated the H 2 O content of mantle minerals in xenoliths, which in various cases show evidence of metasomatism, implying that the measured H 2 O contents may reflect these modification processes rather than the H 2 O content in the environment where they were sourced. In this work, the H 2 O content of 118 mineral inclusions in lithospheric diamonds, that are shielded from metasomatic and alteration events by their diamond host, was investigated. Most of the inclusions studied here are clinopyroxene and garnet of eclogitic and peridotitic paragenesis and their H 2 O content was determined by recording Fourier Transform Infrared (FTIR) spectra of each inclusion within their diamond host. Additional analyses conducted on seven inclusions that were extracted from their diamond host show good agreement with the analysis performed while still trapped inside the diamond, confirming the validity of the H 2 O contents determined for the inclusions still trapped by their diamond host. The measured H 2 O contents range from 0 to 550 ppm wt. with eclogitic inclusions showing a higher H 2 O content with respect to their peridotitic counterparts. On average, clinopyroxene has the highest H 2 O content compared to all other inclusion minerals (with average H 2 O content of 110 ppm wt. and 61 ppm wt. for eclogitic and peridotitic samples, respectively). The H 2 O contents of mineral inclusions in diamonds studied here and from the literature are much lower than those reported for cratonic mantle xenoliths. This difference may be due to (i) sampling bias, (ii) metasomatism subsequent to diamond formation, or (iii) lithospheric mantle involved in diamond formation not being representative of the average cratonic lithosphere (i.e., with a different H 2 O activity). Modelled H diffusion in mineral inclusions in diamonds at the pressure and temperature conditions of diamond formation is fast, indicating complete H 2 O re-equilibration with the diamond forming medium. It follows that the diamond forming medium must be a silicate melt with a low H 2 O content, a carbonatitic melt and/or a fluid characterized by a low H 2 O activity.