Depressed 660-km seismic discontinuity beneath cold subduction zones caused by akimotoite-bridgmanite phase transition

The 660-km seismic discontinuity (D660) is the boundary between the Earth’s lower mantle and transition zone and is commonly interpreted as the dissociation of (Mg,Fe)₂SiO₄ ringwoodite to (Mg,Fe)SiO₃ bridgmanite plus (Mg,Fe)O ferropericlase (post-spinel transition). Prominent features of D660 are significant depressions to 750 km and multiplicity beneath cold subduction zones. Previous high-pressure experiments provided negative but gentle Clapeyron slopes (−1.3 to −0.5 MPa/K) of the post-spinel transition. Thus, the post-spinel transition cannot interpret the D660 depression. Therefore, another phase transition with a steep negative slope is required, and the akimotoite−bridgmanite transition in (Mg,Fe)SiO₃ is one candidate.

In the current study, we determined the boundaries of the post-spinel (RBP) and akimotoite−bridgmanite (AB) phase transitions in the MgO-SiO₂ system over a temperature range of 1250–2085 K using advanced multi-anvil techniques with in situ X-ray diffraction. We judged a stable phase assemblage by observing relative increase/decrease in the ratio of coexisting high- and low-pressure assemblages at spontaneously and gradually decreasing pressure and a constant temperature from diffraction intensities. Since this strategy is strictly based on the principle of phase equilibrium, it excludes problems in determining phase stability caused by sluggish kinetics and surface energy.

We found that the RBP boundary has a slightly concave curve, whereas the AB boundary has a steep convex curve. The RBP boundary is located at pressures of 23.2–23.7 GPa in the temperature range of 1250–2040 K. Its slope varies from −0.1 MPa/K at temperatures less than 1700 K to −0.9 MPa/K at 2000 K with an averaged value of −0.5 MPa/K. The slope of the AB boundary gradually changes from −8.1 MPa/K at low temperatures up to 1300 K to −3.2 MPa/K above 1600 K. Based on these findings, we predict that, beneath cold subduction zones, ringwoodite should first dissociate into akimotoite plus periclase, and then akimotoite transforms to bridgmanite with increasing depth; these successive transitions cause the multiple D660. Moreover, the steep negative boundary of the AB transition should result in cold-slab stagnation due to significant upward buoyancy. Our predictions are supported by the seismological observations beneath cold (e.g., Tonga, Izu-Bonin) subduction zones.