Research ArticleGEOCHEMISTRY

Iron isotopes trace primordial magma ocean cumulates melting in Earth’s upper mantle

See allHide authors and affiliations

Science Advances  12 Mar 2021:
Vol. 7, no. 11, eabc7394
DOI: 10.1126/sciadv.abc7394
  • Fig. 1 Covariation of measured δ57Fe values with elemental and isotopic tracers.

    (A) δ57Fe vs. MgO, (B) δ57Fe vs. μ182W, (C) δ57Fe vs. Lu/Hf, and (D) δ57Fe vs. Zr/Nd. Large blue circles show the Isua samples analyzed for Fe-isotopes, with 2 SD error bars. ISB, Isua Supracrustal Belt samples studied [Northern Terrane 3.72-Ga suite; (9, 10); PM, primitive mantle; MORB, average mid-ocean ridge basalts; see (21) and Materials and Methods for sources of iron isotope and trace element data, respectively]. The dashed blue lines are regression curves fitted to the data; R2 values are shown for reference. The green model curves in (B) to (D) show mixing calculations between a 50% upper mantle melt of the cumulate-derived lithology and a 30% melt of depleted upper mantle peridotite (model details given in the main text and Materials and Methods). Ticks show mixing proportions in 10% intervals.

  • Fig. 2 Schematic demonstration of how the mantle component sampled by Isua basalts may have been formed.

    (A to C) Show the formation of the initial bridgmanite (Bg) cumulate with a small amount of interstitial calcium perovskite (CaPv) and ferropericlase (Fp). (D to F) Demonstrate the chemical effects after a eutectic melt is extracted from the cumulate before recrystallizing as a new assemblage in the lower mantle. (B), (C), (E), and (F) show the geochemistry of the Isua samples relative to primitive mantle (PM) as a blue ellipse. EM, enriched mantle; DM, depleted mantle.

  • Fig. 3 Covariation of measured δ57Fe values with Mg# and μ142Nd, and δ57Fe values corrected to Mg# of 74 (Table 1; δ57FeMg#74) with μ182W, Lu/Hf, Zr/Nd, and μ142Nd.

    Large blue circles show the Isua samples analyzed for Fe isotopes, with 2 SD error bars. ISB, Isua Supracrustal Belt samples studied (Northern Terrane 3.72-Ga suite; 11); PM, primitive mantle.

  • Fig. 4 Covariation between bulk rock Al/Fe and Mn/Fe.

    Ratios were calculated from published major element oxide data for these samples assuming that all iron is present as Fe2+.

  • Fig. 5 Model results from upper mantle phase equilibria calculations, with efficient Fe metal removal in the lower mantle.

    Results are shown from lithologies intermediate between KLB-1 peridotite (middle), the lower mantle bridgmanite cumulate (right), and the lower mantle melt derived from the cumulate (left). Efficient Fe metal removal in the lower mantle results in an Fe3+-rich upper mantle lithology. (A and B) Phase assemblages as a function of temperature and composition; the shading indicates field variance. The solidus and liquidus are indicated by thick lines. (C and D) Fraction of melt present. (E and F) Deviation of δ57Fe of the liquid phase from the bulk δ57Fe. The persistence of g + liq to very high T (with very high melt fraction) is an artifact of the THERMOCALC model.

  • Fig. 6 Model results from upper mantle phase equilibria calculations with inefficient Fe metal removal in the lower mantle.

    In this case, sufficient Fe metal is retained so that no Fe3+ is present in the upper mantle cumulate and cumulate-derived melt assemblages. (A and B) Phase assemblages as a function of temperature and composition; the shading indicates field variance. The solidus and liquidus are indicated by thick lines. (C and D) Fraction of melt present. (E and F) Deviation of δ57Fe of the liquid phase from the bulk δ57Fe. The Fe 3+ content of KLB-1 is unchanged. The persistence of g + liq to very high T (with very high melt fraction) is an artifact of the THERMOCALC mode.

  • Table 1 Iron isotope and published major, trace element, μ142Nd, and μ182W data for 3.7 North Terrane ISB samples.

    LOI, loss on ignition.

    Sample name00-00700-00800-01000-01200-04200-044
    SiO2 (wt %)55.3052.8651.0449.9250.4451.58
    Al2O314.0714.7115.1414.8115.8013.18
    Fe2O3total11.9011.9512.3512.8810.0313.88
    MgO5.455.926.606.988.678.16
    CaO8.889.1210.4810.0411.649.22
    Na2O2.893.812.603.852.592.38
    K2O0.280.450.550.320.110.33
    TiO20.900.830.880.940.491.03
    MnO0.180.200.180.170.130.19
    P2O50.130.130.150.050.060.04
    LOI1.180.520.470.200.630.29
    Al/Fe*0.890.930.930.871.190.72
    Mn/Fe*0.020.020.020.010.010.02
    Nd10.599.8014.244.333.3811.89
    Zr54.2057.80106.3012.2013.5029.20
    Hf2.352.092.910.820.821.56
    Lu0.420.310.440.210.190.61
    Lu/Hf0.180.150.150.260.230.39
    Zr/Nd5.125.907.462.823.992.46
    μ142Nd8.39.44.516.32.68.2
    ±21.41.54.17.74.8
    μ182W15.05.415.614.321.3
    ±3.17.26.88.55.3
    δ57Fe0.02−0.020.060.130.160.30
    ±2 SD0.030.050.020.030.030.01
    δ56Fe0.02−0.010.040.100.110.21
    ±2 SD0.020.010.020.020.010.03
    n563648
    δ57Fe(replicate dissolution of sample 00-044)0.31
    ±2 SD0.01
    δ56Fe0.23
    ±2 SD0.02
    n2
    δ57Fe(replicate dissolution of sample 00-044)0.30
    ±2 SD0.02
    δ56Fe0.19
    ±2 SD0.03
    n6
    δ57FeMg#74−0.07−0.10−0.010.060.130.24
    ±2 SD0.030.050.020.030.030.01

    *Calculated using standard atomic masses and assuming all Fe present as Fe2+.

    †Calculated by olivine addition [e.g., (31, 62)].

    • Table 2 Partition coefficients used for modeling trace element fractionations between minerals and magmas in the upper and lower mantle.

      The initial concentration of elements in the magma ocean is given by C0 (27). ppmw, parts per million by weight.

      LuHfZrNd
      C0 (ppmw)0.0640.198.260.81
      Bridgmanite*0.731.31.380.016
      Ferropericlase0.110.0760.150.036
      Ca perovskite*111.31.622
      Upper mantle0.120.0350.0330.031

      *Corgne et al. (25).

      †Walter et al. (27).

      ‡Workman and Hart (68).

      • Table 3 The molar partition coefficients used to calculate the quantities of elements on each crystallographic site in bridgmanite (Bg), ferropericlase (Fp), and calcium perovskite (Ca-Pv).

        The molar partition coefficients used to calculate the quantities of elements on each crystallographic site in bridgmanite (Bg), ferropericlase (Fp), and calcium perovskite (Ca-Pv).. See Materials and Methods text for more details. Partition coefficient values are from (1) Walter et. al. (2004) [27] Exp. 62; (2) Liebske et. al. (2005) [26] Exp. H2033; (3) Corgne et. al. (2005) [25] Exp. H2020b.

        ElementBg (A)Bg (B)Fp (A)Ca-Pv (A)Ca-Pv (B)
        Si0Stoichiometric0.00520Stoichiometric
        MgStoichiometric0Stoichiometric0.1130
        Ca0.11100.0122Stoichiometric0
        Fe2+ and Fe3+*0.67101.2520.130
        Fe3+See note*See note*0See note*See note*
        Al0.6510.6510.3620.5530.553
        Na0.12100.520.3730
        Cr0.2610.2611.2420.2130.213

        *It is assumed that Fe3+ and Fe2+ are incorporated in the ratio 6:4, with Fe3+ distributed evenly between the two perovskite sites.

        • Table 4 The major element chemistry of the lithologies calculated here, with KLB-1 for comparison.

          The major element chemistry of the lithologies calculated here, with KLB-1 for comparison.. All quantities are in mole percent. LM, lower mantle.

          LithologySiO2Al2O3CaOMgOFeOFe2O3Na2OCr2O3
          KLB-1 (69)38.491.782.8250.575.690.100.250.11
          Cumulate (no
          Fe0 retained)
          48.312.452.4844.551.200.900.040.08
          Cumulate (most
          Fe0 retained)
          47.882.422.4644.152.970.010.040.08
          LM melt (no Fe0
          retained)
          48.492.3314.7832.680.910.680.060.07
          LM melt (most
          Fe0 retained)
          48.162.3214.6832.462.260.010.060.07
        • Table 5 Parameters used in calculating of the force constants for each mineral site.

          Parameters used in calculating of the force constants for each mineral site.. All other constants are as used in (13). The ionicity is set to 0.36. For each equilibrium calculation, KT is calculated for each mineral using the molar proportions of Fe2+ and Fe3+ on each site, as determined using THERMOCALC.

          MineralSpeciesSiteCation
          coord.
          Bond lengthOxygen
          coord.
          No. in formulaKf
          (N m−1)
          KT
          (N m−1)
          SpinelFe2+M62.154243178
          SpinelFe2+T42.004179221
          SpinelFe3+M62.0254276319
          SpinelFe3+T41.87541145401
          GarnetFe2+M182.2914326147
          GarnetFe3+M262.0244277319
          OlivineFe2+M62.16854142173
          OrthopyroxeneFe2+M162.1353.66148198
          OrthopyroxeneFe3+M162.1183.66173304
          OrthopyroxeneFe2+M262.2283.33146192
          ClinopyroxeneFe2+M162.143.66147197
          ClinopyroxeneFe3+M162.0333.66183344
          ClinopyroxeneFe2+M262.5263.75128117

        Supplementary Materials

        Stay Connected to Science Advances

        Navigate This Article