EGU General Assembly 2020
© Author(s) 2020. This work is distributed under
the Creative Commons Attribution 4.0 License.

Phase relations of Ca-Mg-carbonates and trace element partition coefficients between carbonates and dolomitic melt at 6 and 9 GPa

Melanie Sieber1,2, Franziska D.H. Wilke1, Hans-Josef Reichmann1, and Monika Koch-Müller1
Melanie Sieber et al.
  • 1GeoForschungsZentrum, Section 3.6, Germany (
  • 2University of Potsdam, Institute for Mineralogy, Germany (

Calcite (CaCO3) and magnesite (MgCO3) are among the most common carbonates on Earth. The presence of Ca‑Mg‑carbonates in the mantle affects the melting and phase relations of peridotites and eclogites and (partial) melting of carbonates liberates carbon from the mantle to shallower depths. The onset of melting and the incipient melt composition of carbonated peridotites and carbonated eclogites are influenced by the pure CaCO3‑MgCO3‑system. Thus, a deeper insight into the phase relations and melting behavior of the CaCO3‑MgCO3‑system is crucial to better understand the carbon cycle in the Earth’s mantle.

We performed quenched multi-anvil experiments at 6 and 9 GPa to (a) examine the phase relations of the nominally anhydrous CaCO3‑MgCO3‑system and to (b) establish partition coefficients for Li, Na, K, Sr, Ba, Nb, Y and rare earth elements (REEs) between carbonates and dolomitic melt. We used a rotating multi-anvil press to overcome quenching problems as observed in previous studies. Rotation of the multi‑anvil press is, additionally, indispensable to establish equilibrium between solid carbonates and dolomitic melt.

The melting temperature and phase relations of Ca‑Mg‑carbonates depend on the Mg/Ca‑ratio. For instance, at 6 GPa Ca‑rich carbonates with a molar Mg/(Mg+Ca)‑ratio (XMg) of 0.2 will transform into a dolomitic melt (XMg=0.33‑0.31) and calcite crystals (XMg=0.19‑0.14) at 1350‑1440 ℃. Partial melting of Mg‑rich carbonates (XMg=0.85) will produce a dolomitic melt (XMg=0.5‑0.8) and Ca‑bearing magnesite (XMg=0.89‑0.96) at 1400‑1600 ℃. Trace element distribution into Ca-Mg-carbonates depends on XMg, temperature and seems to follow lattice constraints for divalent cations. Partition coefficients of REEs between magnesite (Ca0.11Mg0.89CO3) and dolomitic melt (Ca0.5Mg0.5CO3) at 6 GPa and 1400 ℃ are uniform scattering marginal between 0.1‑0.2. The partition coefficient of Lu (D=0.1) is unmodified to lower Ca-content in magnesite (Ca0.04Mg0.96CO3) and higher temperature (1600 ℃), but the partition coefficients between such Ca-poor magnesite and dolomitic melt (Ca0.2Mg0.8CO3) decrease continuously from heavy-REEs to light‑REEs from 0.1 to 0.001, respectively.

Our findings have important implications for the cycle of carbon and trace elements in the mantle because Ca‑Mg‑carbonates will (partially) melt at 6 GPa and temperatures above ~1300 ℃ producing a dolomitic melt. Consequently, CO2 will be liberated by partial melting of an upwelling carbonated mantle at a depth of ~200 km considering the thermal structure of the upper mantle. The results also affirm that carbonates are stable in the subducting slab even for hot subduction zone geothermal gradients unless carbonate-bearing lithologies in the slab are infiltrated by aqueous fluids.

How to cite: Sieber, M., Wilke, F. D. H., Reichmann, H.-J., and Koch-Müller, M.: Phase relations of Ca-Mg-carbonates and trace element partition coefficients between carbonates and dolomitic melt at 6 and 9 GPa, EGU General Assembly 2020, Online, 4–8 May 2020, EGU2020-2990,, 2020

Comments on the presentation

AC: Author Comment | CC: Community Comment | Report abuse

Presentation version 1 – uploaded on 28 Apr 2020
  • CC1: Comment on EGU2020-2990, Max Wilke, 04 May 2020

     Nice Data! Interestingly Eu does not display any anomalie. I would have expected at least some Eu2+ in such an experiment. The correlation with ionic radii is also quite unexpected. Even in the case where you observe a variation (magnesite) it is not parabolic. Any ideas?

    • AC1: Reply to CC1, Melanie Sieber, 04 May 2020

      We didn't use any mineral buffer. However, the use of Pt-capsules and graphite heater may have imposed fO2 close to the CCO buffer enhancing the presence of Eu3+ over Eu2+. Thus, no Eu anomaly is observed.

      I agree that no clear parabolic trend can be seen for the incorporation of trace elements into magnesite. From our set of data we cannot evaluate the substitution mechanism for e.g. the incorporation of trivalent REEs, but those may diminish a parabolic trend.