Calcium silicate, CaSiO
3, occurs in a variety of natural and synthetic polymorphs. The low-pressure polymorph wollastonite is a common metamorphic mineral. Breyite (
1), an intermediate-pressure polymorph (
2), has been found as inclusions in diamond. At the pressures and depth range of Earth’s transition zone (420 to 660 km) and lower mantle (LM; 660 to ~2700 km), CaSiO
3 assumes a perovskite structure. Perovskite-type CaSiO
3, first synthesized by Liu and Ringwood (
3), is a liquidus phase for basaltic and peridotic bulk rock compositions at LM pressures and temperatures and has been experimentally shown to host many elements that are incompatible in upper-mantle minerals (
4–
7). These include rare-earth elements (REEs), large ion lithophile elements (LILEs; K, Sr, and Ba), Ti, U, and Th. In other words, these elements are compatible rather than incompatible in a LM mineral assemblage that contains a few vol % of CaSiO
3-perovskite. The composition and abundance of this phase in the LM are therefore key in constraining the budget and distribution of REEs and LILEs and the elements with abundant radioactive isotopes (K, U, and Th) that make an important contribution to the heat of Earth’s mantle (
8). Through these parameters, CaSiO
3 perovskite provides essential constraints on the fate of recycled crust in deep Earth, thermochemical anomalies, and the existence of a magma ocean at the base of Earth’s mantle. The synthetic perovskite phase of pure CaSiO
3 has been found to assume either cubic or tetragonal symmetry (
9) and belongs to the tausonite (SrTiO
3)–type perovskites that adhere to fundamentally different structural distortion mechanisms (
10) and crystal-chemical constraints than the GdFeO
3-type perovskites such as bridgmanite [MgSiO
3-perovskite (
11)] and the CaTiO
3 mineral, actually named perovskite.
The difficulty of finding CaSiO
3-perovskite in nature stems from its stability at pressures only above 20 GPa (
3,
4) along with a low kinetic barrier for back-conversion into low-pressure polymorphs (
2). This barrier is lower than for bridgmanite, which has been found as a rare occurrence in highly shocked meteorites despite its stability only above 23 GPa (
11,
12). Nestola
et al. (
13) reported the presence of CaSiO
3-perovskite as an inclusion in a diamond from the Cullinan mine, South Africa. The reported phase deviates from synthetic CaSiO
3-perovskite in several ways: (i) its volume at ambient conditions is >20% larger (
9); (ii) it sustains the beam of an electron microscope, whereas any synthetic CaSiO
3-perovskite vitrifies rapidly at ambient conditions; (iii) its cell axis ratios and Raman spectrum are nearly equal to those of CaTiO
3; and (iv) its space group indicates a structural distortion mechanism different from that of synthetic CaSiO
3-perovskite but much closer to that of CaTiO
3 (
7,
8). Nestola
et al. (
13) proposed that this distinctive phase of CaSiO
3 is the result of partial decomposition of a Ti-bearing CaSiO
3-perovskite. The coexistence of CaTiO
3 + CaSiO
3 polymorphs in diamond inclusions may also point to retrograde transformation of stoichiometric Ca-Si-Ti-perovskites (
1) that form in the deep upper mantle (5 to 10 GPa).
The findings by Nestola
et al. (
13) are notable by themselves but differ from the expected high-pressure CaSiO
3-perovskite and have not resulted in the approval of CaSiO
3-perovskite as a mineral. We report the discovery of CaSiO
3-perovskite as a mineral approved by the Commission of New Minerals, Nomenclature and Classification (CNMNC) of the International Mineralogical Association (IMA). The new mineral [IMA2020-012a (
14)] is named “davemaoite” in honor of Dave (Ho-kwang) Mao for his eminent contributions to the field of deep-mantle geophysics and petrology. The type material—inclusions in a diamond from Orapa, Botswana—is deposited in the Natural History Museum Los Angeles (catalog number NHMLA 74541, formerly GRR1507 of the Caltech mineral collection) (
15). Davemaoite coexists with orthorhombic carbonaceous α-iron and wüstite (Fe
0.8Mg
0.2)O at 8 to 9 GPa remnant pressure (
Fig. 1). Separate inclusions of ilmenite, iron, and ice-VII in the same diamond (
16) have remnant pressures of 7 GPa and 8 to 9 GPa, respectively. The x-ray diffraction (XRD) pattern of davemaoite is that of a cubic perovskite (
Fig. 1) (
15), with, at most, contributions of <5 vol % of material with ±2.5% smaller or larger volume, whereas an overall distortion of the lattice can be excluded on the basis of the reflection intensities (
Fig. 1). Cubic ABO
3 perovskites have no internal structural degrees of freedom and comprise only one chemical formula unit. Thus, the identification of this phase is unambiguous even in diffraction patterns with contributions from more than one phase (
Fig. 1).
Davemaoite was identified through the XRD pattern of cubic perovskite at a location in the hosting diamond with a CaKα x-ray fluorescence (XRF) signal far above background. Both, XRF and XRD data were obtained at beamline 34-ID-E at the Advanced Photon Source (
15). We superimposed the CaKα XRF map (
Fig. 2) on a visible light image of the holotype material at the beamline where we subsequently collected the XRF and XRD data. We also made corresponding maps of Fe and Ti (fig. S1). Areas with XRF signal at noise level in
Fig. 2 show no x-ray diffraction besides that of diamond. Right after acquisition of the XRF map, we examined by XRD the inclusions that were found by XRF. XRD and XRF were collected on the inclusions when they were fully entrapped in the doubly polished platelet of the hosting diamond. We focused the x-ray beam to an area of 0.5 μm by 0.5 μm in order to identify inclusions with high spatial resolution. The inclusions of 4 μm by 6 μm and 4 μm by 16 μm areas within the red circle of the Ca XRF map corresponded to the XRD patterns of the cubic perovskite (
Fig. 1). We added frames with perovskite patterns to obtain better signal and powder statistics.
We confirmed the identification of davemaoite by infrared spectroscopy. Cubic ABO
3 perovskites have no Raman-active modes and three infrared (IR)–active modes (
15). We observed two of the three IR-active modes (
Fig. 1, inset), while the third one was below the diffraction limit for objects as small as these inclusions. As expected, we observed no Raman peaks (fig. S2). We calculated mode energies by fitting force constants to match ab initio calculated zone-center phonon energies of the tetragonal structure (
17) and mapped the tetragonal Brillouin zone onto that of the cubic structure. Intensities are based on the calculated phonon density of state.
Subsequently, we used laser-ablation inductively coupled mass spectrometry (LA-ICP-MS) to excavate and analyze the chemical compositions of two of the inclusions with a 100-μm-diameter laser beam. We indicate the ablation area with a red circle (
Fig. 2). We monitored all mass peaks under medium mass resolution (
m/Δ
m = 4000), which allowed us to resolve numerous carbon-related molecular interferences on low-mass isotopes. We hit two inclusions of davemaoite at 5 to 8 μm and 80 to 100 μm depth below the polished surface (
Fig. 2, inset). The time-resolved
44Ca signal of the LA-ICP-MS measurement (
Fig. 2, inset) shows signal clearly above background level. We also collected equivalent profiles for
56Fe,
39K, and
52Cr (fig. S1). In all cases, the signals rose above background at the same depth, consistent with their origin in the same two inclusions. We obtained an average davemaoite composition of (Ca
0.43(1)K
0.20(1)Na
0.06Fe
0.11(1)Al
0.08Mg
0.06Cr
0.04(2))(Si
1.0(2)Al
0.00(1))O
3 (
15). We performed a Rietveld refinement (
Fig. 1) and provide a crystallographic information file (
15).
Next, we describe the hosting diamond and then discuss the pressure-temperature (P-T) conditions of formation of the type davemaoite and its composition. The hosting diamond is different from diamonds that contain potential retrograde products of high-pressure minerals at 0 to 1 GPa remnant pressures (
13,
18). The diamond is of type IaAB with frosted octahedral faces and trigon features (
15). We found davemaoite, iron, wüstite, ilmenite, and ice-VII in the center of the diamond. Our analysis of the N-defect IR bands (fig. S3) (
12) indicates a low average residence temperature (~1500 K) or a short residence time in the mantle, similar to the holo- and cotype diamonds of ice-VII (
18). Short residence time and low average residence temperature are common features of lithospheric diamonds, but in sublithospheric diamonds both parameters act in favor of conserving high remnant pressures and high-pressure minerals by reducing viscoelastic relaxation of the hosting diamond and by preventing retrograde transformations. The bulk modulus of davemaoite depends on its composition (
7,
8) and is unknown for the given composition. However, coexisting wüstite is at a remnant pressure of 8 to 9 GPa (
16). For a single inclusion of wüstite, this remnant pressure would correspond to an entrapment pressure of ~40 GPa if pressure were to evolve along a purely elastic path (
15). However, diamond becomes viscoelastic between 1100 and 1200 K, even at laboratory time scales (
19). To account for this nonelastic behavior of the hosting diamond, we use the method of Wang
et al. (
20), which does not rely on initial assumptions about the entrapment temperature and uses the P-T paths of separate inclusions in the same diamond: wüstite, iron, ice-VII, and ilmenite. Using this approach, we assessed entrapment conditions of 29 ± 5 GPa at 1400 to 1600 K (
15). Because viscoelastic processes are path- and time-dependent, we cannot exclude a higher entrapment pressure or temperature.
We cannot entirely rule out that our chemical analysis is affected by minor contaminants, although we did not observe an XRD signal or a marked XRF signal of any phase other than davemaoite, wüstite, and iron in the excavated region. Furthermore, we note that (i) the low Ti is a result unaffected by potential contamination and (ii) the
39K signal occurs at the same depth as the
44Ca signal of davemaoite. We did not observe by XRD alternative hosts of K and Ca such as liebermannite and harmunite-type (Ca,K,Na)(Al,Si)
2O
4 anywhere in this diamond. Both of these phases, and any phase dominated by K and Ca, give diffraction patterns that are very different from the perovskite-type pattern we observed. Hence, we believe the presence of K and Al in davemaoite is not likely the result of a contaminated analysis but rather indicates coupled substitution of a large and a small cation K,Na + Al,Fe for Ca. Generally, a substitution of K for Ca and Al for Si shifts the material into the stability field of ABO
3 perovskites with a trend toward high crystal symmetry (
21).
Our result indicates that the postspinel phase (Ca,Na,K)(Al,Si)
2O
4 is not required as a host of Ca, alkalis, and Al in at least the upper region of the lower mantle. It is possible that type davemaoite formed retrograde out of postspinel through a reaction (Ca,Na,K)(Al,Fe
3+,Si)
2O
4 + Fe
0 → (Ca,Na,K)(Al,Si)O
3 + FeO, but for this process one expects the presence of remnant postspinel in the paragenesis, which is not observed. In the deep mantle, davemaoite takes on a role similar to that of garnet in the upper mantle. Both minerals have a “garbage can” crystal chemistry that allows them to host many elements that are incompatible in upper-mantle minerals (
6,
7). Our observations are fully consistent with the experimental results that this mineral dissolves LILEs, specifically K (
6,
7). Experimental studies that were based on peridotite and MORB (mid-ocean ridge basalt)–like bulk compositions formed davemaoite with lower K content and higher Ti content than the type material, which is expected for these starting compositions. We argue that the low Ti and high K content of type davemaoite reflects a different, K-rich source composition, possibly resulting from deep-mantle metasomatism, which is also indicated by the presence of ice-VII and by the hosting diamond itself (
16). This point emphasizes the importance of studying natural specimens of high-pressure minerals, because they record a petrologic complexity of deep Earth that may not be assessed in experiments. Depletion of Ti in type davemaoite is a possible result of the presence of phases that strongly partition Ti, such as liuite, FeTiO
3-perovskite. Ilmenite has been observed in the same diamond at similar remnant pressure as the davemaoite-wüstite-iron inclusion, and its P-T path intersects the phase boundary of liuite (
15). Hence, our findings indicate that the source rock composition of type davemaoite deviated from peridotite and, thus, that chemical segregation occurs in the lower mantle, possibly down to 900 km according to our estimate (fig. S4). This variation in rock composition affects heat generation through radioactive decay in the lower mantle where davemaoite scavenges K, as shown here, and U and Th, as experimentally shown (
7).
Acknowledgments
We thank N. Tomioka and an anonymous reviewer for their helpful comments.
Funding: This work was supported by awards NSF-EAR-1838330, -EAR-1942042, and -EAR-1322082; the NSF Cooperative Agreement No. DMR-1644779; and the State of Florida. Use of the Advanced Photon Source and the Advanced Light Source were supported by the US Department of Energy, Basic Energy Sciences, contracts DE-AC02-06CH11357 and DE-AC02-05CH11231, respectively.
Author contributions: O.T., S.H., S.Y., and M.H. participated in design, interpretation, data collection, and analysis of the reported results and in drafting and revising the manuscript. W.L., S.N.G.C., H.A.B., J.T., and G.R.R. participated in data collection and revising the manuscript.
Competing interests: The authors have no competing interests.
Data and materials availability: Additional chemical and crystallographic information about davemaoite is provided in the supplementary materials. Raw data are deposited at Dryad (
23). Crystallographic and chemical information on type davemaoite is deposited in the Inorganic Crystal Structure Database. The type material is deposited with the NHMLA under accession number 74541.