Calcium silicate perovskite, CaSiO3, is arguably the most geochemically important phase in the lower mantle, because it concentrates elements that are incompatible in the upper mantle, including the heat-generating elements thorium and uranium, which have half-lives longer than the geologic history of Earth. We report CaSiO3-perovskite as an approved mineral (IMA2020-12a) with the name davemaoite. The natural specimen of davemaoite proves the existence of compositional heterogeneity within the lower mantle. Our observations indicate that davemaoite also hosts potassium in addition to uranium and thorium in its structure. Hence, the regional and global abundances of davemaoite influence the heat budget of the deep mantle, where the mineral is thermodynamically stable.

Uploaded data include X-ray powder diffraction data, a cif-file, infrared spectra, calculated IR-active mode zone center energies, and LA-ICP-MS depth profiles.

Micro-diffraction and -Xray fluorescence maps were collected with a 50degree incidence angle onto the planoparallel specimen while LA-ICP-MS depth profiles, IR transmission- and Raman spectra were collected at 90degree (= normal) incidence.

Details of data collection and analysis are provided in (1, 2).

LA-ICP-MS data:

The excel sheet contains the depth profiles (cps versus analysis time) of the inclusion (analysis #14), of the same diamond (analysis #15), of a diamond blank, and of the standard (GSE-1G). The laser was turned on multiple times at some spots, which can be seen in the 12C signal.

Among all measured elements only Si and P posed specific issues because of a high background level, most likely because the active laser excited Si and P in the sample chamber. We removed the Si and P background by using the ²⁸Si/¹²C and ³¹P/¹²C ratios measured on the diamond anvil. That is, the Si and P signals were normalized by ¹²C and the signal level in the diamond blank was taken as background level.

The uncertainty associated with the particular element Si obviously does not affect the X/Ca ratios obtained from our LA-ICP-MS measurement, where X represent for cations other than Si. The Si content of davemaoite is constrained by charge balance (see supplement in (Tschauner et al. Science 374, 891-894, 2021). Consequently, the large standard deviation for Si only affects the Si content of coexisting iron that is subject to large variation (see table S1 in ref. 2).

X-ray diffraction raw data:

The data are in tif format. The frames are labeled ‘GRR1518’ but they were taken from specimen GRR1507 (1). Frames were collected in a grid scan around the area of the inclusion and in transmission geometry (1).

The 'poni'-file gives the calibration parameters.

Frame # 139 was used for background subtraction, that is: as frame, pixel by pixel, and not after integration.

A zero-intensity low pass was used for noise suppression in the frames after background-frame subtraction and prior integration. Remnant diamond single crystal signal and spurious secondary diamond reflections (Umweg-anregung) were masked prior integration.

In the frames four different types of signal occur: 1) only diamond, 2) a serpentine-group mineral, similar to lizardite at a few GPa compression. Some frames show diffraction of calcite or an isotypic carbonate along with lizardite and in some frames there is signal of alpha-Fe. 3) davemaoite, iron, and wuestite. 4) magnetite or chromite. Signal of magnetite/chromite overlaps with davemaoite + iron + wuestite, but is from a different depth because no micro-Raman- and IR-signal was obtained from magnetite (1). Signal from lizardite +/- carbonate does not overlap with davemaoite. Calcite (or an isotypic carbonate) and lizardite are much deeper inside the diamond than davemaoite because no Raman-signal was detected (1,2). Davemaoite was observed in frames 127-130 and 146-149.

Frames 146 to 148 were added and integrated for assessment of the structure of davemaoite using Le Bail extraction of |F(hkl)| and subsequent reversed Monte Carlo optimization of the perovskite model against the observed |Fhkl| with (1) and without symmetry bias (2).

Alternative phase mixture models are discussed in (2) and are statistically significantly worse.

1) O. Tschauner et al. Discovery of davemaoite, CaSiO₃-perovskite, as a mineral from the lower mantle. Science 374, 891-894 (2021).

2) O. Tschauner et al. Response to the Technical Comment by Michael Walter et al. on Science 374, 891-894 (2021).

This submission contains one Crystallographic Information File (.cif), one data sheet of integrated diffraction patterns and infrared spectra (.xls), a data sheet of LA-ICP-MS data which is explained above, a set of diffraction image frames which is explained above, and the diffraction data calibration file. Detailed explanation is given in the accompanying Readme file (.txt)