Seismological Expression of the Iron Spin Crossover in Ferropericlase in the Earth’s Lower Mantle

Abstract

The two most abundant minerals on Earth which together make up over 90% of the Earth’s lower mantle are (Mg,Fe)O-ferropericlase (Fp) and (Mg,Fe)SiO3-bridgmanite (Bm). Iron in Fp undergoes a high-spin to low-spin (HS-LS) crossover that influences density, viscosity, elasticity, thermal conductivity, and elemental partitioning, however, the predicted effects of this transition are not apparent in global 1D seismic velocity profiles. This discrepancy suggests that the predictions are inaccurate, seismic resolution is insufficient to resolve the effects, or a substantial portion of the mid-lower mantle is relatively SiO2-rich (hence Fp poor) compared to the shallow mantle. The melt-depleted mantle lithosphere of subducted oceanic slabs that sink into the lower mantle contains 22% Fp, and thus offers the best opportunity to prospect for a spin change in Fp. Here we reveal a loss in the abundance of fast seismic velocity anomalies in compressional (P-wave) tomography models at 1,400-2,000 km depth that is opposite to the trend in shear (S-wave) models. This can be explained by the decreasing temperature sensitivity of P-velocity expected for the mixed spin state of iron in Fp at corresponding pressures7. We also observe a similar but subtle signal for seismically slow regions below 1,800 km, consistent with a pressure increase and broadening of the Fp spin transition at higher temperatures. Seismic wave raypath distribution is similar for both P- and S-waves in this depth range, therefore this signature cannot be attributed to substantial differences in data coverage. Our identification of the spin transition signal in seismically fast and slow regions indicates that the spin crossover can identify the presence of Fp in the lower mantle. The absence of a Fp spin crossover signal in global seismic profiles supports the notion that the lower mantle is chemically heterogeneous at large scales and contains SiO2-rich regions that suppress the average signature of these pressure-induced electron spin pairing transitions.

Broad Elastic Softening of (Mg,Fe)O Ferropericlase Across the Iron Spin Crossover and a Mixed‐Spin Lower Mantle

“…A central observation of Shephard et al. (2021), discussed also by Cammarano et al. (2010), was that the P‐velocity profile calculated for a pyrolitic mantle composition deviates substantially from the seismic Preliminary Reference Earth Model PREM below about 1,600 km.…”

“…The reflection of these iso‐depth property changes in ferropericlase in geophysical properties will depend on the property. The unique seismic signature of the spin crossover ‐ most importantly the elevated V S / V P ratio (Marquardt et al., 2009; Shephard et al., 2021; Wu & Wentzcovitch, 2014) ‐ is expected to affect seismic observables in all regions of the mantle where ferropericlase is in mixed‐spin state, particularly the green shaded region in Figure 4b. However, transport properties, such as viscosity (Deng & Lee, 2017; Marquardt & Miyagi, 2015; Saha et al., 2013; Wentzcovitch et al., 2009), or electrical conductivity (Lin et al., 2013), will only be strongly affected if an interconnected network of ferropericlase develops, possibly as a result of mantle flow processes (Deng & Lee, 2017; Girard et al., 2016; Marquardt & Miyagi, 2015; Thielmann et al., 2020; Yamazaki et al., 2009).…”

“…Recent attempts to detect an expression of the iron spin crossover in tomographic models derived from seismic observations (Shephard et al., 2021) relied on theoretical calculations to predict the spin crossover at mantle temperatures. The employed calculations, however, were based on models where the distances between iron atoms were maximized, and assumed ideal mixing of high‐ and low‐spin iron (Wu et al., 2013), similar to other studies (Muir & Brodholt, 2015; Tsuchiya et al., 2006; Wentzcovitch et al., 2009; Wu & Wentzcovitch, 2014), which produce a sharp crossover at 300 K. In contrast, previous experimental work showed that at ambient pressure ferropericlase samples quenched from high temperature (∼1000 K) largely exhibit random ordering (Waychunas et al., 1994), suggesting that maximizing the distance between iron atoms might not well describe natural systems.…”

Coupled dynamics and evolution of primordial and recycled heterogeneity in Earth's lower mantle

“…However, the employed rheology was simplified, and the models did not produce Earthlike tectonic behavior. Moreover, none of the models explic-itly predicted the formation of thermochemical piles in the lowermost mantle, which are perhaps the most evident heterogeneities in Earth's lower mantle (e.g., Solomatov and Stevenson, 1993;Christensen and Hofmann, 1994;Li and Romanowicz, 1996).…”

Catastrophic slab loss in southwestern Pangea preserved in the mantle and igneous record

“…To build positive wave speed vote maps we used 26 global models (P-Wave and S-wave) with most of them differing in data selection and parametrization, and regularization of the inversion [GyPSuM-S 111 ; DETOX P2 and P3 112 ; HMSL-P06 and S06 113 ; PRI-P05 and -S05 114 ; SPani-P and –S 115 ; GAP-P4 116 ; LLNL_G3Dv3 117 ; Hosseini2016 118 ; SEISGLOB1 119 ; MITP08 120 ; UU-P07 45 ; TX2019Slab-P and S 121 ; S362ANI + M 122 ; S20RTS 123 ; S40RTS 124 ; SAVANI 125 ; SAW642ANb 126 ; SEMUCB-WM1 127 ; SEMum 128 ; TX2011 129 ; TX2015 130 ; see model details in Table S1 ]. To build lower mantle vote maps we implemented the standard deviation threshold following Shephard et al 131 . Also, we built high-velocity votemaps applying a zero threshold (Supplementary Fig.…”