The low primordial heavy noble gas and 244Pu-derived Xe contents of Earth's convecting mantle

•We use dynamic mass spectrometry to measure Ar-Kr-Xe in upper mantle-derived gases.•Eifel and Mt. Etna mantle sources are essentially devoid of accretionary Kr-Xe.•Both sites show lower fractions of 244Pu-derived Xe than upper mantle estimates.•Requires extensive volatile loss followed by pervasive regassing via subduction.•Points to mantle source heterogeneities or revision of upper mantle Xe composition. Clues to unraveling the origin and history of terrestrial volatiles lie in the noble gas record of Earth's mantle. However, the low abundance of heavy noble gases (Ar-Kr-Xe) in mantle-derived rocks presents a major analytical challenge that limits our understanding of mantle volatile evolution. Here, we employ a new technique of ultrahigh precision dynamic mass spectrometry to measure Ar-Kr-Xe isotopes in mantle-derived gas collected from Mt. Etna (Italy) and Eifel (Germany), which both tap depleted convecting mantle reservoirs. We find that the fractions of primordial Kr-Xe from accretionary sources (≤ 7 % of non-radiogenic, non-fissiogenic isotopes) and 244Pu-derived 136Xe (≤ 9.8 ± 9.3 % of total fissiogenic Xe) are both markedly lower than previously estimated. For Mt. Etna, we find an apparent lack of detectable primordial Xe, which could reflect an additional contribution from recycled atmospheric volatiles from nearby subduction. In addition, slight excesses of 238U-derived fissiogenic Xe relative to the upper mantle composition may reflect the contribution of a crustal component related to the occurrence of a HIMU (“high μ” where μ = 238U/204Pb)-type source in Mt. Etna volcanic products. The low primordial heavy noble gas and 244Pu-derived Xe contents of Earth's convecting mantle, as derived from these new data, requires extensive volatile loss during terrestrial accretion, followed by long-term degassing and pervasive overprinting of primordial heavy noble gases by subduction recycling. In addition, we suggest that quantitative incompatible element (including Pu, U) extraction to the Hadean crust and subsequent reintroduction of U via subduction could have contributed to lowering the ultimate fraction of 244Pu-derived 136Xe in the upper mantle. The differences observed between this study and other upper mantle Xe studies may reflect mantle source heterogeneities (e.g. due to the heterogeneous overprinting of mantle volatiles by subduction) but could also result from analytical inconsistencies and/or subsurface isotope fractionation in natural sy