Composition, evolution and structure of the deep mantle, Reidar G Trønnes, University of Oslo

INVITATION

#ESRFscience Live Online Seminars

"Composition, evolution and structure of the deep mantle"

Wednesday, March 23rd, 10:00 a.m.

Please click here to watch the replay of the webinar

Composition, evolution and structure of the deep mantle
ESRF-GeoBridge seminar, March 23, 2022
R.G. Trønnes1,2 and C.E. Mohn2
Natural History Museum1 and Centre for Earth Evolution and Dynamics2, University of Oslo
Several recent developments require an updated view of the phase relations, composition and structure of the lower mantle. A lower solidus temperatures for peridotite in the lowermost mantle [1] and a lower melting curve for iron at outer core pressures [2] indicate that our estimated core-mantle boundary (CMB) temperature should be reduced to about 3600 K. Such a low CMB temperature, combined with a dp/dT-slope of the bridgmanite to post-bridgmanite (bm-pbm) transition in peridotite [1] and MgSiO3 [3] of less than 8 MPa/K, imply that the double-crossing scenario for the bm-pbm transition along ambient to cool geotherms [4], is untenable. Pbm is stable from the CMB up to an elevation of 200-400 km above the CMB, even in the hot LLSVPs (large low S-wave velocity provinces) [3,5].
Another important realisation is the need to develop an improved compositional model for the convecting mantle. A conservative estimate of the volume of subducted oceanic lithosphere during the 3-0 Ga period [e.g. 6] is 123 % of the total mantle volume and 156 % of the convecting mantle volume [5]. This is corroborated by Pb model ages of about 2 Ga for the MORB source [7,8], a range of 1.5-2.5 Ga for different OIB sources [9] and by Cook Island basalts with olivine sulphide inclusions, characterised by S-isotopes indicative of Archean subduction and recycling of hydrothermally altered oceanic crust [10]. Because peridotite in the convecting mantle must have been through at least one, and to a large extent two or more, melting episodes, the common use of a pyrolitic compositional model for the convecting mantle is unfortunate. A present ambient mantle potential temperature, Tp, of 1388 C [11] and a reasonable secular cooling rate in the 3-0 Ga period [12,13], imply an accumulation of thick sections of strongly depleted mantle lithosphere and thick layers of recycled oceanic crust (ROC), which are picritic, rather than basaltic. With a current Tp of 1388 C, the primary melt is a primitive basalt with 12 wt% MgO, but in the 0.5-3.0 Ga, time period the ROC would be picritic with 14-21 wt% MgO [5]. Even if the thickness ratio of ROC to the combined ROC + depleted mantle section increases from 10 to 21% with increasing age from 0 to 3 Ga, the lithosphere might thicken beyond the thickness of the most recently depleted mantle section as a function of cooling away from the ridges [14]. This, combined with the partial segregation of some of the dense ROC-material onto the LLSVP base layers [15] and the preferential ROC remelting in the upwelling asthenosphere under the ridges, might have kept the ROC proportion in the convecting mantle relatively constant, e.g. near 10 %.
Strong petrological evidence indicates that highly viscous Hadean refractory domains (HRD) constitute major parts of the lower mantle (LM). Such bridgmanitic domains, dominated by the MgSiO3 endmember, would have formed by early crystallisation of the main magma ocean (MO) and by extensive partial melting in deep Haden plumes originating from the ceiling of the basal magma ocean (BMO). Protocore-MO and core-BMO chemical exchange of FeO (to the core) and SiO2 (to the MO-BMO) would partially buffer the MO-BMO composition at relatively high Si/(Mg+Fe) and Mg/Fe ratios, promoting extensive bm-crystallisation [e.g. 16,17]. The high viscosity and neutral buoyancy of the bridgmanitic material in the solid mid-LM, would make it susceptible to convective aggregation into bm-enriched ancient mantle structures (BEAMS) with sectional dimensions of 1000-2000 km [5, 17-19]. Based on seismic tomography sections [21,22], the estimated the volume proportions of the convecting mantle, BEAMS, LLSVP base layers (assuming 300 km average thickness) and sub-continental lithospheric mantle (assuming 100 km thickness and a 40% continental area fraction) are approximately 83, 13, 1 and 2%, respectively [5].
[1] Kuwayama et al. 2021, GRL 49, e2021GL096219. [2] Sinmyo et al. 2019, EPSL 510, 45. [3] Mohn, in prep. [4] Hernlund et al. 2005, Nat. 434, 882. [5] Trønnes & Mohn, in prep. [6] Shirey & Richardson, 2011, Sci. 333, 434. [7] Hart 1984, Nat. 309, 753. [8] Jackson et al. 2020, PNAS 117, 30993. [9] Andersen et al. 2015, Nat. 517, 536. [10] Cabral et al. 2013, Nat. 496, 490. [11] Bao et al. 2022, Sci. 375, 57. [12] Herzberg et al. 2010, EPSL 293, 79. [13] Ganne & Feng 2017, GGG 18, 872. [14] Parsons & Sclater 1977, JGR 82, 803. [15] Torsvik et al. 2016, Can. L. Earth Sci. 53, 1073. [16] Laneuville et al. 2018, PEPI 276, 86. [17] Trønnes et al. 2019, Tectonophys. 760, 165. [18] Manga 1996a, GRL 23, 403; 1996b, J. Fluid Mech. 8, 1732. [19] Ballmer et al. 2017, Nat. Geosci. 10, 236. [20] Becker & Boshi 2002, GGG 3, 1003. [21] van der Meer et al. 2010, Nat. Geosci. 3, 36; 2018, Tectonophys. 35, 309.