Oxygen isotope (δ18O, Δ'17O) insights into continental mantle evolution since the Archean

Abstract

Oxygen isotopic ratios are largely homogenous in the bulk of Earth's mantle but are strongly fractionated near the Earth's surface, thus these are robust indicators of recycling of surface materials to the mantle. Here we document a subtle but significant ~0.2‰ temporal decrease in δ¹⁸O in the shallowest continental lithospheric mantle since the Archean, no change in Δ'¹⁷O is observed. Younger samples document a decrease and greater heterogeneity of δ¹⁸O due to the development and progression of plate tectonics and subduction. We posit that δ¹⁸O in the oldest Archean samples provides the best δ¹⁸O estimate for the Earth of 5.37‰ for olivine and 5.57‰ for bulk peridotite, values that are comparable to lunar rocks as the moon did not have plate tectonics. Given the large volume of the continental lithospheric mantle, even small decreases in its δ¹⁸O may explain the increasing δ¹⁸O of the continental crust since oxygen is progressively redistributed by fluids between these reservoirs via high-δ¹⁸O sediment accretion and low-δ¹⁸O mantle in subduction zones.

Fig. 1. Evolution of oxygen isotopes in lithospheric continental mantle (CLM).

Temporal trends of δ¹⁸O olivine (a, b), orthopyroxene and bulk (b), and CLM corrected for melt depletion (b, c) in studied mantle nodules. Bulk is computed based on modal mineral abundances and δ¹⁸O values of olivine and orthopyroxene (T Supplementary Table 1). Zircon from the Archean Kaapvaal cratonic mantle is from ref. , and zircon and basalts from the moon are from refs. ^,, lunar mantle estimate is based on zircon, olivine, and basalts (see the text). Notice decreasing trends (95% conf. interval error envelope in line fit) and inbox diagrams.

Fig. 2. Locality-averaged δ¹⁸O values for the studied samples of olivine (Ol), bulk, and reconstructed original undepleted peridotite (see Fig. 3).

Line fit statistics indicate decreasing trends.

Fig. 3. Computed isotope effects of batch (continuous lines) and fractional (dashed lines) peridotite melting from 0 to 45‰ on the equilibrium isotope values of melt and the residual mineral assemblage.

a Isotope effects on minerals and the residual bulk. Starting composition at low-T has 56% olivine, 26% Opx, 15% Cpx, and 3% of Spinel. Isotope fractionation factors are from refs. ^,. For melt, we assumed A = 1.3. Isotope effects are computed against phase change and proportions (% melt) trends observed in ref. in their peridotite melting experiments, also modeled in ref. in pMELTS. Fractional melting assumed the removal of equilibrium fractional melt shown on the diagram. Flux melting trends are computed using Rayleigh melting by water fluxing and up to 5% fractional melt removal with shown effect on bulk peridotite and residual olivine δ¹⁸O values. In this case, it is assumed that water fluxing generates silica-rich andesitic hydrous melt/fluid with δ¹⁸O = 6.1–7.3‰ in equilibrium with the ambient assemblage. The upper bound of the flux melting triangle is computed by assuming decreasing δ¹⁸O from 7.3‰ to 6.1‰, while the lower bound assumes that the hydrous melts maintain a constant high δ¹⁸O = 7.3‰. Notice that melt depletion results in lowering bulk δ¹⁸O, but it also results in higher modal olivine and the δ¹⁸O_Ol values which are higher (closer to the bulk), while δ¹⁸O_Opx is also higher; Δ¹⁸O_Opx-Ol are constant. b effect of melt depletion along the melting path shown in (a) for δ¹⁸O_Ol or δ¹⁸O_bulk. Notice the nearly linear change. Upon melt extraction, δ¹⁸O_dunite < δ¹⁸O_harzburgite < δ¹⁸O_lherzolite. cooling will not affect δ¹⁸O_bulk (blue horizontal lines) but will increase Δ¹⁸O_{mineral-mineral}. fractionation as is shown. c Thin sections of samples showing enriched and depleted harzburgites that lost melt, and the final residual assemblage is olivine-richer. The horizontal width is 3 cm.

Fig. 4. Triple oxygen isotope analyses of Archean and post-Archean olivines and orthopyroxenes showing overlap with San Carlos olivine.

Expected trends to higher-Δ′¹⁷O and lower and higher δ¹⁸O subduction fluids, derived from subducting slabs are shown. The δ¹⁸O in orthopyroxenes is higher than equilibrium olivine by 0.5–0.7‰ and nearly identical in Δ′¹⁷O at mantle temperatures (Fig. 3, ref. ).

Fig. 5. A cartoon that explains the cooling of the Earth and rehydration of the mantle by low-δ¹⁸O fluids in spreading centers^,, plate bending zones, and subduction zones^,, generating progressively lower δ¹⁸O (and overall heterogeneous) continental lithospheric mantle (darker blue) after initiation of plate tectonics.

Δ′¹⁷O is not significantly modified by these processes (Fig. 4). Color code: yellow: low-δ¹⁸O peridotites, red: high-δ¹⁸O sediments, basalts, and eclogites, green: primitive mantle plumes, light blue: oceans. a Early Earth and moon regimes: degassed mantle, plume tectonics, rudimentary subduction around plumes, and intense hot mantle convection. The δ¹⁸O of the original peridotites is 5.57 ± 0.07‰, and Bulk Silicate Earth=Bulk Silicate Moon. b Modern plate tectonics regime: rehydration of mantle, plate accretion, and imbrication, lower-δ¹⁸O peridotites in the subcontinental lithospheric mantle samples by studied xenolith suites.

Fig. 6. Co-evolution of the continental lithospheric mantle (CLM, this work) and the Earth’s crust as expressed by three proxies: granites, bulk shales, and zircons.

Note change in δ¹⁸O scale. Inset shows mass balance of CLM and crust during growth and evolution; 1:10 and 1:20 are assumed crust/CLM proportions based on the modern average lithospheric thickness of 150 km. BSE is estimated in this work and corresponds to the starting point of crustal growth after the Giant Impact. Unidirectional temporal loss of low-δ¹⁸O to the large reservoir of CLM helps to explain the accumulation of heavy δ¹⁸O in the crust and potentially hydrosphere (e.g., 57–60).