Calcium silicate-diorthocarbonate is a missing link in deep-mantle chemistry of silicates and carbonates

Abstract

Interactions between carbonates and silicates play a central role in many industrial and planetary processes. Although current knowledge is largely limited to silicate–carbonate exchange mechanisms, high-pressure compounds incorporating both carbonate and silicate units have long been anticipated. Here we report the synthesis of the high-pressure silicate–diorthocarbonate CaSiC₂O₇ at 122(2) GPa and 2800(200) K. CaSiC₂O₇ is based on the framework of vertex-sharing SiO₆ octahedra and CO₄ tetrahedra. Ab initio calculations are consistent with the experimental results and indicate dynamical and mechanical stability of this phase over a wide pressure range. These findings reveal an additional pathway for carbonate–silicate interaction in carbonate-rich deep-mantle environments. CaSiC₂O₇ has a density lower than that of major mantle minerals and exhibits comparatively low acoustic velocities, suggesting that silicate-carbonates may contribute to seismic anomalies near the core–mantle boundary.

This study finds that silicates and carbonates can form a single compound deep within Earth’s mantle, revealing chemistry that may help explain unusual seismic signals near the core-mantle boundary.

Introduction

Carbonates and silicates are fundamental to both chemistry and planetary science. They are important technological materials with a wide range of applications, including cement production¹, catalysis², water treatment³, among others. As geomaterials, silicates are the most abundant minerals in the Earth’s crust and mantle, thereby defining its structure and evolution. In turn, carbonates play a key role in regulating the Earth’s climate and habitability, as they are responsible for long-term hosting and recycling of the CO₂ through subduction⁴. In addition, both silicates and carbonates are often discussed as contributors to the Earth’s seismic anomalies^5–8.

Despite the fact that carbon and silicon are both group IV elements, their oxides reveal remarkably different bond characteristics. At ambient conditions, the crystal chemistry of carbonates is driven by sp² hybridization of carbon atomic orbitals, which allows the formation of [CO₃]²⁻ groups with triangular planar geometry. In contrast, silicon adopts sp³ hybridization in most of the silicates, forming four directional bonds in a tetrahedral geometry [SiO₄]⁴⁻. Interaction between carbonates and silicates draws long-lasting attention, driven by interest in CO₂ sequestration in silicate-based cements^9,10, silicate-carbonate global cycle¹¹ and fate of subducting carbonates upon contact with silicate mantle^12–14. In industrial and geological processes, silicate-carbonate exchange reactions (primarily involving (Ca, Mg)SiO₃ and (Ca, Mg)CO₃) are dominant and silicate-carbonate compounds containing both carbon and silicon are rare. Known to date silicate-carbonate minerals^15,16 are based upon CO₃ and SiO₄ polyhedra and usually formed at elevated temperatures (up to ~1300 K) and moderate pressures (up to ~1 GPa).

High pressure enables the formation of phases with properties and stoichiometries that differ from those commonly encountered in conventional solid-state chemistry¹⁷. Pressure-induced densification typically increases atomic coordination (often given as Roman numbers). This trend is well-established for silicates where silicon transitions from four-fold (SiO₄, or Si[IV]) towards six-fold (SiO₆, or Si[VI]) coordination^18–20, with the occasional formation of transient five-fold SiO₅ (Si[V]) state^21–24. Similarly, a change from sp²- to sp³-configuration of carbon^25,26 drives high-pressure chemistry of carbonates^27–29 and results in the formation of CO₄ tetrahedra (C[IV]). This tendency reduces the chemical difference between carbonates and silicates as shown experimentally³⁰ and theoretically³¹ for the Si_xC_1-xO₂ compound at 18–26 GPa and 600–980 K. Silicate-carbonate compounds containing both C[IV] and Si[VI] have long been predicted to form at high pressure²⁵, although, to our knowledge, no experimental observations have been reported. The very recent observation of the calcium silicon double carbonate Ca₂Si(CO₃)₄, containing both SiO₆ and CO₃ groups, provides further evidence for the possibility of such a pressure-induced pathway³².

Herein, we report on the synthesis of CaSi^[VI]C^[IV]₂O₇ in a laser-heated diamond anvil cell. In situ synchrotron-based single-crystal X-Ray diffraction of a polycrystalline sample was employed to solve and refine its structure at 122 GPa. Ab initio density functional theory has been employed to extend the experimental observation, particularly to derive its equation of state and elastic constants. The synthesis of CaSiC₂O₇ provides experimental evidence for silicate-carbonates built from CO₄ tetrahedra and SiO₆ octahedra, as well as for a sp³-dicarbonate containing [C₂O₇]⁶⁻ units. Our results demonstrate that interactions between carbonates and silicates under lowermost-mantle conditions can produce silicate-carbonate phases and suggest that such compounds may contribute to seismic anomalies in this region.

Results

Experimentally determined crystal structure of CaSiC₂O₇

The silicate-carbonate CaSi^[VI]C^[IV]₂O₇ was synthesized from a starting assemblage consisting of CaSiO₃, CaCO₃ and Fe_0.945O (used as the laser absorber; see “Methods” for details) at 2800(200) K in a laser-heated diamond anvil cell, together with other phases listed in Table S1. The synthesis pressure was measured to be 122(2) GPa after quenching, and it could be up to 134 GPa during the heating (see “Methods”).

The CaSiC₂O₇ compound crystalizes in the orthorhombic space group Cmcm (#63) with a = 4.4750(2) Å, b = 6.4691(8) Å, c = 9.013(3) Å, V = 260.9193(16) ų (Table S2). The asymmetric unit of CaSiC₂O₇ contains one calcium, one carbon, one silicon, and three oxygen atoms (Fig. 1). Calcium cations are 12-fold coordinated with formation of cuboctahedra, silicon atoms are octahedrally coordinated and carbon atoms are in tetrahedral coordination (Fig. 2). The Ca-O bond distances slightly vary in the range of 2.221(5) to 2.317(6) Å with an average of 2.26 Å (Fig. 2c). The SiO₆ octahedra also possess a geometry close to a regular one. Two apical bonds Si-O2 and four equatorial bonds Si-O3 are of 1.627(4) Å and 1.635(4) Å, respectively, while O2-Si-O3 and O3-Si-O3 angles are of 88.53(18)° and 90.4(4)° (Fig. 2b). The CO₄ tetrahedra reveal a higher level of distortion. C-O bond distances vary from 1.295(9) to 1.404(8) Å with an average of 1.329 Å (Fig. 2a) while the tetrahedral angles O–C–O are in the range of 106.4(5)–112.6(5)°. CO₄ tetrahedra link through common O1 vertices with the formation of [C₂O₇]⁶⁻ dimers. The SiO₆ octahedra are isolated from each other and share common O2 and O3 vertices with CO₄ tetrahedra, with the formation of the three-dimensional framework that hosts Ca atoms in its voids. It is also noteworthy that layers of Ca and O atoms are forming a cubic close-packing arrangement where the C and Si atoms are filling tetrahedral and octahedral voids, respectively (Fig. S1).

Fig. 1. Polyhedral model of the crystal structure of CaSiC₂O₇ at 122 GPa obtained from single crystal X-Ray diffraction.

CO₄ and SiO₆ polyhedra are given in yellow and blue, respectively. Ca, Si, C and O atoms are given as gray, blue, yellow and red spheres, respectively. The black lines outline the unit cell.

Fig. 2. Structural units of the CaSiC₂O₇.

a C₂O₇ dimer composed of two CO₄ tetrahedra. b SiO₆ octahedron. c CaO₁₂ cuboctahedron. Ca, Si, C and O atoms are given as gray, blue, yellow and red spheres, respectively. O1, O2, and O3 denote crystallographically distinct oxygen sites. Characteristic C–O, Si–O and Ca–O bonds are marked with the corresponding bond lengths in Angstroms.

The topology of CaSiC₂O₇ framework resembles the one of sodium trisilicate, Na₂Si₂^[VI]Si₄^[IV]O₁₄ (a = 8.922(2) Å, b = 4.8490(5) Å, c = 11.567(1) Å, β = 102.64(1)°, space group C2/c, Fig. S2)³³. The sodium trisilicate belongs to the group of high-pressure framework silicates with mixed 4- and six-fold coordination of silicon³⁴. Synthesized at pressures >4 GPa, these frameworks feature relatively dense arrangements of SiO₄ tetrahedra and SiO₆ octahedra and contain large sites suitable for alkali or alkaline-earth cations. Earlier, Oganov et al.²⁵. proposed that these structures can be adapted by high-pressure silicate-carbonates^33,34. Herein, we demonstrate that, at megabar pressures, sp³ carbon can substitute silicon in the tetrahedral site, forming a framework based on CO₄ and SiO₆ polyhedra.

Ab initio calculations for CaSiC₂O₇

To our knowledge, the phase synthesized in this work was not reported in previous theoretical studies. To assess whether the experimentally observed CaSiC₂O₇ phase can be reproduced theoretically, we have performed ab initio simulations (See “Methods”). Full structural relaxation at the experimental unit cell volume results in an excellent agreement between measured and calculated unit-cell parameters, atomic coordinates and interatomic distances (Table S3). Harmonic phonon dispersion calculations reveal that CaSiC₂O₇ is dynamically stable both at the synthesis pressure of 122 GPa and at 1 bar (Fig. S3). The pressure dependence of the theoretical values of the unit cell volume of CaSiC₂O₇ can be described by the third-order Birch-Murnaghan equation of state with the parameters V₀ = 377.5(2) ų, K₀ = 151.2(9) GPa and K’ = 4.41(2) (Fig. 3).

Fig. 3. The pressure dependence of the CaSiC₂O₇ unit cell volume.

Red diamonds represent experimental data upon decompression after synthesis at 122 GPa, while black solid circles are values from the calculations. Pressure measurement errors are presented as horizontal error bars in the experimental data. Unit cell volume errors from structure refinement are smaller than the size of the symbols. The dashed black line shows the fit of the calculated pressure-volume data using the third-order Birch-Murnaghan equation of state (V₀ = 377.5(2) ų, K₀ = 151.2(9) GPa and K’ = 4.41(2)). Source data are provided as a Source Data file.

The elastic constants (C_ij) of CaSiC₂O₇, calculated at 0 K for both at 122 GPa (Table S4) and at ambient pressures (Table S5), indicate that the phase is mechanically stable under these conditions, according to the Born stability criteria for orthorhombic systems³⁵ (Text S1). The bulk and shear moduli were derived from the elastic tensor by the Voigt and Reuss approach³⁶ (Text S2). As expected, the bulk modulus K = 155 GPa calculated from the elastic constants at 1 bar is consistent with the value K₀ = 151.2(9) GPa obtained from the equation-of-state fit, confirming the internal consistency of the calculations.

The calculated electronic density of states shows that CaSiC₂O₇ is a wide-bandgap semiconductor with a band gap (E_g) of 5.6 eV at 122 GPa (Fig. S4).

Discussion

Implications for the high-pressure chemistry of silicates-carbonates

The current study demonstrates the synthesis of the calcium silicate-diorthocarbonate, CaSi^[VI]C^[IV]₂O₇, based on CO₄ and SiO₆ units. Previously, about 50 silicate-carbonates were known at ambient conditions, with three being from the CaO–SiO₂–CO₂ system: spurrite, Ca₅[SiO₄]₂(CO₃)³⁷, tilleyite, Ca₅[Si₂O₇](CO₃)₂³⁸, and galuskinite, Ca₇(SiO₄)₃(CO₃)¹⁵. While being based on CO₃ triangles and SiO₄ tetrahedra at ambient conditions, these compounds were proposed as possible precursors for high-pressure silicate-carbonates hosting C[IV] and/or Si[VI]. However, spurrite, Ca₅[SiO₄]₂(CO₃), and tilleyite, Ca₅[Si₂O₇](CO₃)₂, were found to decompose into their component silicates and carbonates at pressures below 4 GPa at ∼1300 K³⁹. Recent studies at elevated temperatures demonstrated the phase stability of spurrite up to 27 GPa and 700 K, with no phase transitions being observed⁴⁰. Tilleyite was also found to persist up to 7 GPa and 675 K, while at higher pressures it transformed to post-tilleyite phase containing again CO₃ and SiO₄ units⁴¹. Ab initio calculations have also suggested the stability of these calcium silicate-carbonates minerals up to 25–35 GPa⁴². The presence of rigid carbonate units was proposed to enhance the stability of SiO₄ tetrahedra and to prevent their transition to SiO₆ octahedra⁴⁰. While this work was already under consideration, a calcium silicon double carbonate, Ca₂Si(CO₃)₄, based on SiO₆ octahedra and CO₃ flat triangular groups, was obtained by reacting tilleyite with CO₂ at 39.5 GPa and 2600 K³², demonstrating that these transitions can be induced by high temperatures. The synthesis of CaSi^[VI]C^[IV]₂O₇ at the megabar pressure regime completes the long-discussed crystal chemical trend. The phase adapts the structure and stoichiometry of sodium trisilicate, Na₂Si₂^[VI]Si₄^[IV]O₁₄, thus hinting that other high-pressure framework silicates with mixed 4- and six-fold coordination of silicon³⁴ are plausible isomorphs for silicate-carbonates.

The CaSiC₂O₇ synthesized in this work is a sp³-dicarbonate hosting [C₂O₇] units (Fig. 2a). Its structure expands the known structural diversity of carbonates and illustrates the crystal-chemical relationship between high-pressure sp³-carbonates and low-pressure tetrahedral silicates. Typically, the structural diversity of C[IV] carbonates was considered more limited in contrast to Si[IV] silicates, due to the smaller flexibility of inter-tetrahedral C–O–C angles in comparison to Si–O–Si angles^25,43. However, recent advancements in carbonates chemistry allowed to experimentally discover nesocarbonates with isolated CO₄ tetrahedra (Me₂CO₄, Me = Ca, Sr)^44,45, carbonates based on four-membered chains (Me₄C₄O₁₃, Me = Mg/Fe)^46–48 and four-membered pyramidal groups (MeC₂O₅, Me = Ca, Fe, Mn, Cd)^48–51, inocarbonates with infinite chains of vertice-sharing CO₄ tetrahedra (CaCO₃)^52,53, cyclocarbonates based on three-membered rings (Me₃C₃O₉, Me = Fe, Mg, Ca)^27,54–56, and a tectocarbonate with three-dimensional connectivity of tetrahedra (CaC₂O₅)⁵⁷. Even greater structural diversity of carbonates is proposed on the basis of theoretical calculations^7,58, highlighting the need for further experiments to explore the complex chemistry of high-pressure sp³-carbonates.

Implications for the deep carbon storage

The results obtained in the geologically relevant system CaCO₃–CaSiO₃–FeO have several implications for the mineralogy and elasticity of Earth’s lowermost mantle. Firstly, the formation of CaSiC₂O₇ indicates that high-pressure silicate-carbonates may serve as another class of carbon hosts in Earth’s deep interior. Carbonates, predominantly calcite (CaCO₃) and dolomite (Ca, Mg(CO₃)), carry the inorganic carbon into the deeps via subduction and interact with the minerals of the average mantle upon descent. Previous high-pressure-high-temperature studies within the (Ca, Mg, Fe)–Si–C–O system demonstrated that a plethora of phases can be formed depending on the phase assemblages and experimental conditions, including polymorphs of CaCO₃, MgCO₃, CaSiO₃, MgSiO₃, Fe₃C, FeO, CaO, SiO₂, Fe₇C₃, C, MgO, CO₂ (Table S6). These observations allowed to propose melting⁴, decarbonation^13,59,60, silicate-carbonate chemical exchange^12,14,61, redox freezing with diamond or carbide formation^13,62–65, as potential mechanisms that govern the fate of carbonates in the Earth’s deeps. Our results indicate a possible scenario for interactions between carbonates and silicates under the pressure and temperature conditions of the lowermost lower mantle, involving the formation of silicate-carbonate compounds hosting both carbon and silicon (Fig. 4). In the light of the ongoing discussions on the redox heterogeneity of the lower mantle^66–68, presence of carbonates and silicate-carbonates can be envisioned in its locally oxidized domains, that are thought to arise from subducted crustal materials⁶⁹, Fe³⁺ disproportionation reactions^70,71, or core–mantle exchange processes⁷². Persistence of oxidized forms of carbon under these conditions is supported by the occurrence of carbonates as inclusions in deep diamonds^73–75 as well as by experimental evidence of calcium carbonate formation from reactions between calcium silicate and iron carbides⁶¹. Importantly, calcium carbonates were also proposed to be less sensitive than (Mg, Fe)CO₃ to redox breakdown throughout the lower mantle⁶⁴. These observations advocate for the possible presence of carbonates in the deep mantle and their availability for interactions with the abundant lower mantle silicates.

Fig. 4. Schematic diagram of a possible silicate-carbonate formation pathway in the Earth’s mantle.

Solid arrows indicate the increase in coordination number with depth, from SiO₄ to SiO₆ in silicates and CO₃ to CO₄ in carbonates. The region close to the core-mantle boundary (CMB, solid line), where large low-shear-velocity provinces (LLVPs) have been identified, is a possible domain for silicate-carbonate formation through interactions among carbonate-rich subducted slabs, silicates from the surrounding mantle, and iron-rich material from the outer core. The transition zone and D” layer mantle boundaries are marked by dashed lines.

The silicate-diorthocarbonate CaSiC₂O₇ was obtained as a product of the subsolidus reaction between CaCO₃ and CaSiO₃ in the presence of FeO at conditions close to the core-mantle boundary. We note, however, that carbon contamination from the diamond anvils is a common issue in laser-heated diamond anvil cell experiments^76–78, which prevents a quantitative carbon mass balance, leaving the precise formation pathway of CaSiC₂O₇ less tightly constrained than implied by the nominal CaCO₃ + CaSiO₃ + FeO starting assemblage alone. In addition, although iron is not incorporated into CaSiC₂O₇, the formation of Fe³⁺-bearing phases from FeO (Table S1) indicates electron-transfer reactions that may have influenced the local chemical environment during crystallization. Despite these uncertainties in the exact reaction pathway, the nominal starting assemblage remains geophysically relevant, as CaCO₃, CaSiO₃ and FeO have each been proposed as possible components of large low-velocity provinces (LLVPs)^7,29,79–82 that extend thousands of kilometers horizontally and hundreds of kilometers vertically above the core-mantle boundary and are seismologically detected below Africa and Pacific Ocean^83–85. As a representative of the high-pressure silicate-carbonate class, CaSiC₂O₇ is of interest for examining the elastic properties and possible contribution of this group of compounds to the seismic signatures of the lowermost mantle. For this, the elastic moduli and acoustic wave velocities at 122 GPa were derived from the calculated elastic parameters (C_ij) as described in Texts S1 and S2. Their values are compared in Table 1 with the radial Reference Earth model (REM1D)⁸⁶, with minerals of the pyrolitic lower mantle assemblage⁸⁷ and with relevant high-pressure calcium carbonates. Interestingly, CaSiC₂O₇ is closer in elastic properties to carbonates rather than silicates, and together they belong to the class of low-density minerals with low compressional and shear velocities (V_p and V_s, respectively). This is in line with the observation on the bulk moduli of low-pressure silicate-carbonates, composed of CO₃ and SiO₄ units: there is negative correlation between CO₃ content and compressibility⁴². Our results demonstrate that in the case of a reaction between silicates and carbonates within or at the boundaries of LLVPs, the possible formation of silicate-carbonates will reduce the bulk density (by consumption of dense CaSiO₃ and production of low-density carbon-bearing phases) and decrease the velocities. In addition, the particularly low shear modulus (188 GPa) of CaSiC₂O₇ indicates a shear-soft phase whose presence would facilitate plume formation. Deep-mantle plumes are associated with the edges of LLVPs where hotspots appear^88,89, emphasizing the potential importance of silicate-carbonates when formed at these boundaries.

Table 1.

Comparison of elastic properties of CaSiC₂O₇ with rock-forming minerals of the pyrolitic lower mantle and high-pressure carbonates at 122 GPa (~2700 km depth)

	Compressional Wave Velocity / Vp (km/s)	Shear Wave velocity / Vs (km/s)	Density / ρ (g/cm3)	Bulk Modulus / K (GPa)	Shear Modulus / G (GPa)	Reference
REM1D	13.57	7.21	5.47	627	284	86
CaSiC2O7*	12.45	6.01	5.198	555	188	this study
CaC2O5*	13.24	6.98	5.05	560	245	7
post-aragonite CaCO3*	13.34	7.22	5.29	571	275	80
cubic CaSiO3†	13.7	7.16	5.63	677	289	112
low-spin Fe0.1Mg0.9O‡§	13.1	6.97	5.6	600	271	113
Mg-perovskite‡	14.3	7.7	5.45	685	324	114
Mg-postperovskite‡	14.36	7.84	5.49	678	340	6

*at 0 K. †at 1000 K. ‡at 300 K. §experimental data (extrapolated).

The values derive from computational studies, except Fe_0.1Mg_0.9O ferropericlase, where the values are extrapolated from the experimental data at lower pressures.

The experimental confirmation of a silicate-carbonate phase suggests possible intricate chemical exchanges between carbonates and silicates in the heterogeneous lowermost mantle (Fig. 4). The co-existence of CaSiC₂O₇ with other phases (Table S1) indicates that the sample chamber likely remained in a non-equilibrium state, as is typical of laser-heated diamond anvil cell experiments. Consequently, the present experiments do not allow us to assess the thermodynamic stability of CaSiC₂O₇ under lowermost mantle conditions.

Independent theoretical support for the potential stability of this phase has recently been provided by Tsuchiya & Isobe (2026)⁹⁰, whose study –motivated by the public preprint of the present work– predicts that CaSiC₂O₇ is thermodynamically stable over a broad range of lowermost mantle pressures. While these calculations offer an important framework for understanding the possible stability field of CaSiC₂O₇, experimental confirmation under equilibrium conditions is still required.

Further systematic in situ investigations of carbon- and silicon-rich systems across a range of pressures, temperatures and chemical compositions are therefore needed to establish the equilibrium stability of silicate-carbonate phases and to clarify their potential role in the mineralogy, geochemistry, and geophysics of Earth and Earth-like planets.

Methods

Sample preparation

A BX90 diamond anvil cell (DAC) equipped with synthetic diamond anvils with a culet size of 150 μm was prepared⁹¹. The rhenium gasket used was indented from an initial thickness of 220 μm down to 20 μm and laser-drilled to create a circular pressure chamber of 80 μm in diameter. A powdered natural sample of CaSiO₃ wollastonite from Stanisław locality, Kopianec, Poland, containing approximately 5 vol% of CaCO₃ (Fig. S5), was loaded into the pressure chamber. A crystal of ⁵⁷Fe-enriched Fe_0.945O of ~10 × 10 x 5 μm³ was also loaded into the pressure chamber^92,93, sandwiched by layers of CaSiO₃-CaCO₃ mixture at the top and at the bottom. The DAC was then compressed to 122(2) GPa as determined by the pressure-induced shift of the Raman spectrum of the diamonds at the culets⁹⁴. The sample was then heated for a few minutes from both sides at 2800(200) K using the double-sided Nd-YAG laser system at the Nuclear Resonance Beamline (ID18)^95–97 of the European Synchrotron Radiation Facility (ESRF) using the gray-body spectroradiometry method to measure the surface temperature of the heated sample⁹⁸. Due to thermal pressure expected to develop inside the pressure chamber during heating, the synthesis pressure may be as high as 134 GPa, as estimated using the approximation described in⁹⁹ and the thermal expansion and bulk modulus values for CaSiO₃ at 122 GPa¹⁰⁰.

In situ single crystal X-Ray diffraction

In situ X-ray diffraction measurements were performed at the High-Pressure Diffraction Beamline ID15B (ESRF, Grenoble)¹⁰¹. An EIGER2 X CdTe 9 M detector was used, and a monochromatic X-ray beam (λ = 0.41505 Å) was focused on the sample with a spot area of approximately 1.5 × 1.5 μm². An X-ray map of the laser-heated spot was obtained by scanning the sample chamber with a step of 2 μm at ω = 0^o. The resulting map was analysed using the XDI software suite¹⁰². At each position of interest, single-crystal datasets were collected during ω rotations of ±35° of the DAC with a step of 0.5°. The resulting diffraction images were analysed using the CrysAlis^Pro software suite. The Domain Auto Finder (DAFi) program was used to search for reflection groups belonging to the individual single crystal domains¹⁰³. The multigrain analysis allowed to identify several products of chemical reactions between CaSiO₃ and CaCO₃ in the presence of FeO, including the silicate-diorthocarbonate CaSiC₂O₇ described here (Fig. S6). The crystal structure solution and refinement were performed using the Jana2006 software¹⁰⁴. Crystal data and structure refinement details are presented in Table S2. The Crystallographic Information File for CaSiC₂O₇ is deposited as Supporting Information as well as in the Cambridge Structural Database under deposition number CSD 2352267.

The pressure inside the DAC was then released in several steps as shown in Fig. 3. At each step, a new X-ray map of the laser-heated area was obtained, and the silicate-diorthocarbonate phase was located and probed as described before. The crystal structure of CaSiC₂O₇ was solved and refined down to the pressure of 58(2) GPa. The reciprocal space reconstructions at 122 and 58 GPa are shown in Fig. S7 (a) and Fig. S7 (b), respectively. Further decompression down to 34(2) GPa led to a drop of the diffraction intensity from the sample, not allowing to refine the crystal structure anymore.

Given that the decompression inside a non-hydrostatic pressure medium (in this case the CaSiO₃-CaCO₃ mixture) without the annealing of the sample chamber is known to induce residual stresses on the materials within, the decompression data are not considered for the calculation of the equation of state of CaSiC₂O₇, but are only used to check if the phase can survive on decompression and for the comparison of experimental and calculated unit cell volumes.

Computational Details

The first-principles calculations were done using the framework of density functional theory (DFT) as implemented in the Vienna Ab initio Simulation Package (VASP)¹⁰⁵. To expand the electronic wave function in plane waves, we used the Projector-Augmented-Wave (PAW) method^106,107. The Generalized Gradient Approximation (GGA) functional was used for calculating the exchange-correlation energies, as proposed by Perdew–Burke–Ernzerhof (PBE)¹⁰⁸. The PAW potentials with the following valence configurations of 3s²3p⁶4s² for Ca (“Ca_sv”), 3s²3p² for Si (“Si”), 2s²2p² for C (“C”) and 2s²2p⁴ for O (“O”) were used. Convergence tests with a threshold of 2 meV per atom in energy led to a Monkhorst-Pack¹⁰⁹ k-point grid of 11 × 7 × 5 and an energy cut-off for the plane wave expansion of 750 eV. The phonon frequencies and phonon band structure calculations were performed in the harmonic approximation with non-analytical term correction with the help of PHONOPY software¹¹⁰ using the finite displacement method for a 3 × 3 × 2 supercell (made for the primitive cell (PPOSCAR)) with respectively adjusted k-points: 4 × 4 × 2. Equation of state (EoS) calculations were performed via variable-cell structural relaxations between 1 bar and 160 GPa. In the static EoS calculations, the effect of zero-point motion and temperature was neglected. The elastic constants were obtained using the stress–strain method¹¹¹.

Author contributions

G.A. conceived and designed the study. G.A. and X.L. contributed to sample preparation. G.A., I.K., X.L., S.M., M.H., and A. Pakhomova contributed to data collection and beamline operation. A.A. performed the ab initio calculations. A. Pantousas, X.L. and G.A. performed the analysis of the data. A. Pakhomova and L.D. oversaw and coordinated the project. All authors reviewed, approved and contributed to the final version of the manuscript.

Peer review

Peer review information

Nature Communications thanks the anonymous reviewers for their contribution to the peer review of this work. A peer review file is available.

Funding

Calculations were performed using the Festus-cluster of the Bayreuth Center for High-Performance Computing, funded by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) [grant number 523317330]. L.D. discloses support for the research of this work from Bundesministerium für Bildung und Forschung (Federal Ministry of Education and Research) [grant number 05K25WC2].

Data availability

The experiments were performed on the beamlines ID15B and ID18 of the European Synchrotron Radiation Facility (ESRF), Grenoble, France (proposal number ES-1328). The original data is stored at https://doi.org/10.15151/esrf-dc-2445978915 and https://doi.org/10.15151/ESRF-DC-2445943418. Source data are provided with this paper.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at https://doi.org/10.1038/s41467-026-77204-w.