Introduction

Organic chemistry knows more than 10 million compounds and their number increases by 200–300 thousand annually. The richness of carbon chemistry and the existence of the vast number of organic compounds in nature are due to unlimited possibilities of carbon catenation resulting in the formation of molecules, clusters, chains, and rings differing in their length and complexity. Very frequently, simple structural motifs of these carbon units, e.g., C6 rings, are repeating in organic compounds. High pressures (those above 100,000 atmospheres, ~10 GPa) have never been considered to promote the formation of new classes of organic materials. There are, however, indications that high pressure can alter the bonding patterns in carbides, leading to new compounds with unusual structural units and interesting properties.

Carbides of the alkali and alkaline earth metals with a mainly ionic bonding character contain the simplest possible carbon anions1,2,3,4,5,6: isolated carbon atoms in methanides with C4 anions (in Be2C or Al4C3), [C2]2−-dumbbells in acetylides (e.g., in CaC2, SrC2, BaC2, Li2C2, Na2C2), or linear trimers [C3]4− isoelectronic to CO2 in allenides (with Mg2C3 being the only well-documented example)7. In rare earth metal carbides with metallic properties similar carbon units are found: C2 dumbbells (e.g., in YC2, LaC2, La2C3 ≡ La4(C2)3, CeC2, TbC2, YbC2, and LuC2)2,3,4, isolated carbon atoms and (almost) linear C3 units (in Me4C7 with Me = Y, Ho, Er, Tm, Lu)5,6, or, most remarkably, C1–C3 units in Sc3C4 and Ln3C4 (Ln = Ho–Lu)8. However, due to the metallic bonding character, the assignment of ionic charges is not as straightforward as for the salt-like carbides mentioned above. At high pressure, for the binary systems Mg–C9, Ca–C10, Y–C11, La–C12, and Li–C13 ab initio structure search predicts the formation of unusual metal carbides with exotic [C4] and [C5] units, [C6] rings, and graphitic carbon sheets, as well as a number of structural transitions so that studying them under compression might enable to explore novel catenations of carbon atoms.

Calcium carbide has been the focus of high-pressure research during recent years. CaC2 is of particular interest, as it is characterized by a rich polymorphism. Three polymorphs of CaC2 (CaC2-I, CaC2-II, CaC2-III) are known at ambient conditions14. Their structures feature [C2]2− dumbbells common for acetylides. It was proposed that at relatively low pressures, these [C2]2− units start to polymerize, and at about 7 GPa, CaC2 transforms into a metallic phase (space group Cmcm) with 1D zig–zag polymeric carbon chains, as established using Raman spectroscopy and X-ray diffraction (XRD)15. (Note that in ref. 15 at ~10 GPa, the signal of the C≡C stretching vibration is still dominating, and the quality of the resulting diffraction patterns is too low for an unambiguous Rietveld fit). Theory predicts further structural transformations in CaC2 upon pressure increase coupled with the progressive polymerization of carbon atoms. According to calculations by Li et al.16, above 20 GPa, the structure of CaC2 should adopt the \(P\bar{1}\) symmetry and contain infinite carbon strips built of interconnected pairs of fused five-membered rings. This phase was supposed to be thermodynamically stable up to 37 GPa16. Above this pressure, according to the calculation, metallic \(P\bar{1}\,\)-CaC2 should transform into metallic Immm-CaC213,16, in which carbon atoms are polymerized to form infinite quasi-1D ribbons built of fused six-membered rings. The latter prediction is of particular interest, as the suggested carbon catenation in Immm-CaC213,16 resembles deprotonated polyacene-like nanoribbons.

Here, we report the results of high-pressure high-temperature (HPHT) studies of CaC2 in laser-heated diamond anvil cells (LHDACs) performed using synchrotron single-crystal X-ray diffraction (SCXRD) up to ~150 GPa at temperatures up to ~3000 K. The predicted Immm-CaC213,16 high-pressure polymorph of CaC2 (HP-CaC2) containing deprotonated polyacene nanoribbons was synthesized and fully characterized. A novel Ca3C7 compound, never anticipated or observed before, with an orthorhombic crystal structure (space group Pnma) featuring infinite, fully deprotonated para-poly-indenoindene (p-PInIn)-like chains was discovered, and its structure was solved and refined using SCXRD. We demonstrate that compounds with fused six-membered (reminiscent in shape of benzene) carbon rings can be synthesized above 100 GPa. We demonstrated that on decompression in the presence of water, HP-CaC2 forms polycyclic aromatic hydrocarbons (PAHs).

Results and discussion

Several diamond anvil cells with different samples of calcium carbide were prepared as described in “Methods”. Studies cover the pressure range up to ~150 GPa. A summary of all experiments presented in this work, including pressure–temperature (P–T) conditions of the synthesis and the unit cell parameters of the observed phases, is given in Supplementary Table S1.

Upon compression to the target pressures, we performed Raman spectroscopy measurements of the starting material (Supplementary Fig. 1). In agreement with previous reports14,16 at pressures below ~5 GPa we observed spectral signatures of co-existing tetragonal CaC2-I and monoclinic CaC2-II phases; above ~10 GPa only CaC2-I was observed, and at ~25 GPa the Raman spectrum became featureless (suggesting amorphization of CaC2 as proposed in ref. 14). Microphotographs taken during compression reveal changes in the sample’s visual appearance (Supplementary Fig. 1): the crystal, which was initially transparent (3.3 GPa), became translucent at ~10 GPa and then opaque above ~25 GPa. Such changes can be attributed to the progressive band gap closure in CaC2 upon compression that agrees with resistivity measurements performed by Zheng et al.17.

The systematic analysis of the XRD data obtained from different DACs after laser heating at different pressures (Supplementary Table 1) allowed identifying diamond (as a product of paraffin oil decomposition), CaO (in some cells, due to possible contamination of the commercial calcium carbide with the oxide or Ca(OH)2), and two new phases: a high-pressure polymorph of CaC2 (called here HP-CaC2) and a new Ca–C compound Ca3C7. After laser heating, the sample remains dark and non-transparent (Supplementary Fig. 9). The results of different experiments are mutually consistent and reproducible. The structures of the new solids were solved and refined using the best quality SCXRD datasets, obtained at 44(1) GPa and 147(2) GPa for HP-CaC2 and at 38(1) GPa for Ca3C7 (see Supplementary Tables 2, 3, and 5). They are described in detail below.

The structure of the novel HP-CaC2 polymorph has an orthorhombic unit cell (space group Immm, Z = 4) with the lattice parameters a = 2.5616(3) Å, b = 6.0245(17) Å, c = 6.6851(15) Å at 44(1) GPa (Fig. 1a, Supplementary Fig. 2, and Supplementary Table 2). There are four calcium atoms in the unit cell (Fig. 1a), which occupy the Wyckoff site 4 i (0, 0, 0.2039(2)). The Ca–C distances to the ten nearest carbon atoms vary from ~2.44 Å to ~2.46 Å at 44(1) GPa. Carbon atoms occupy two distinct crystallographic positions (Fig. 1a, b): C1 − 4 g (0, 0.3824(14), 0) and C2 − 4 h (0, 0.2422(16), 0.5). They all are polymerized, forming infinite nanoribbons of fused planar six-membered rings (Fig. 1b). These planar ribbons, lying in the ab plane, are oriented along the a-axis. Thus, the structure can be described as an alternation of layers containing carbon nanoribbons and layers of calcium atoms along the c-axis. The described arrangement of carbon atoms gives rise to an exotic form unprecedented in any known carbide poly-C entity: a [C4] unit with a formal charge of 4− (Supplementary Fig. 4). The C–C distances in the hexagons, refined at 44(1) GPa, are equal to ~1.41 Å and ~1.48 Å for the C1–C1 and C1–C2 bonds, respectively. Two angles in the hexagon are equal at ~120.5°, while the other four angles are ~119.8° (Fig. 1b). It has been shown that upon compression, the C6-ring symmetrizes: the two bond lengths, which were different at 44(1) GPa, became identical and equal with ~1.41 Å at 144(2) GPa (Supplementary Table 5).

Fig. 1: Crystal structure of HP-CaC2 at 44(1) GPa.
figure 1

a A ball and stick model with the unit cell outlined; calcium atoms are shown as white spheres, and carbon atoms as red and blue balls for two distinct crystallographic positions, C1 (Wyckoff site 4 g) and C2 (4 h), respectively. b The geometry of a single deprotonated polyacene nanoribbon; the C–C distances and C-C-C angles are labeled. Cross-sections of the calculated electron localization function (ELF) are shown in the planes perpendicular (c) and parallel (d) to the polyacene nanoribbons.

Laser-heating of CaC2 in paraffin oil at pressures of ~38 GPa results in the appearance of new diffraction peaks, which belong neither to HP-CaC2 nor to B1-CaO. Indexing reveals an orthorhombic unit cell with the lattice parameters a = 4.762(4) Å, b = 8.3411(11) Å, and c = 8.8625(13) Å at 38(1) GPa (space group Pnma) (see Supplementary Fig. 3, Supplementary Table 3). The structure solution and refinement evidence a Ca3C7 stoichiometry with four formula units per unit cell. There are two independent crystallographic positions of calcium atoms in the crystal structure of Ca3C7: Ca1 on the Wyckoff site 4 c (0.0517(3), 0.25, 0.33976(5)) and Ca2 on the Wyckoff site 8 d (0.1956(2), 0.5744(4), 0.3533(4)). The carbon atom C1 is located on the Wyckoff position 4 c (0.1913(12), 0.25, 0.6330(2)). The other carbon atoms C2 (0.2158(8), 0.0535(2), 0.1205(2)), C3 (0.1818(8), 0.6127(2), 0.06479(18)) and C4 (0.0172(8), 0.1627(2), 0.04884(18)) occupy the general Wyckoff sites 8d (x,y,z). In the crystal structure of Ca3C7, carbon atoms form infinite non-planar chains made of fused distorted C6 and C5 rings, which are isomorphous to deprotonated para-type poly(indenoindene) (p-PInIn)18 (Fig. 2). The projection on the [011] plane visualizes the alternation of layers of calcium atoms with deprotonated p-PInIn carbon chains (Fig. 2). To the best of our knowledge, such type of polycarbide entity has never been predicted or reported for any of the metal carbides known in inorganic chemistry19,20,21. There are five different C–C distances in the complex [C14] polyanion with the formal charge 12− (Supplementary Fig. 5); at 38(1) GPa, the shortest and the longest C–C bonds are equal to ~1.43 Å and ~1.48 Å, correspondingly. Unlike the C6-rings in the deprotonated polyacene nanoribbons of HP-CaC2, the carbon hexagon in the p-PInIn anion of Ca3C7 exhibits pronounced distortion, where two angles strongly deviate from 120° with ~111.6°. These modest differences underline a subtle but recognizable dissimilarity in their chemical environment, namely the complexity of poly-C species and their coordination by Ca2+ ions.

Fig. 2: Crystal structure of Ca3C7 at 38(1) GPa.
figure 2

a A projection of the Ca3C7 structure along the a-axis, emphasizing 2D chains of carbon atoms aligned along the b-axis. Calcium atoms are shown as white spheres, and carbon atoms as red and blue balls for the two distinct crystallographic positions C1 (4c) and C2 (8d), respectively. Carbon atoms, named C3 (8d) and C4 (8d), are shown as gray balls. b The geometry of a single deprotonated para-poly(indenoindene) (p-PInIn) chain with the C-C distances and C-C-C angles labeled. c, d Cross-sections of the calculated electron localization function (ELF) are shown in the two different planes containing p-PInIn chains.

Theoretical calculations perfectly reproduce experimental structural data (Supplementary Tables 2 and 3). The pressure dependences of the unit cell parameters and volumes of novel calcium carbides were derived from DFT calculations (Fig. 3). The results were described by the Birch–Murnaghan equation of state (EoS) with the parameters V0 = 1 27.17(2) Å3, K0 = 144.5(2) GPa, and K´ = 3.92(1) for HP-CaC2 and V0 = 432.0(2) Å3, K0 = 125(1) GPa, and K´ = 4 (fixed) for Ca3C7. As seen in Fig. 3, both compounds experience an anisotropic contraction upon compression, with the most incompressible directions being along the polyacene-like nanoribbons in HP-CaC2 (a-axis) and deprotonated p-PInIn chains in Ca3C7 (b-axis). The unit cell parameters derived from several independent SCXRD experiments conducted at various pressures (Supplementary Table 1) show a good agreement with the calculated EoSes (Fig. 3).

Fig. 3: Pressure dependence of the lattice parameters and the unit cell volumes for the two Ca–C compounds.
figure 3

a HP-CaC2; b Ca3C7. Solid symbols are experimental data points; dashed curves represent theoretical data; the black dashed lines are the fit of the calculated PV data using the third and second-order Birch–Murnaghan equations of state with the following parameters: V0 = 127.17(2) Å3, K0 = 144.5(2) GPa, and K´ = 3.92 for HP-CaC2; and V0 = 432.0(2) Å3, K0 = 125(1) GPa, K´ = 4 (fixed) for Ca3C7.

Harmonic phonon dispersion calculations reveal that the novel calcium carbides are dynamically stable at their synthesis pressures (Fig. 4). The calculated electron density of states shows that HP-CaC2 is metallic and the main contributions at the Fermi level come from the Ca 3d and C 2p states (Fig. 4). The computed charge density was analyzed in terms of the electron localization function (ELF), which revealed a strong covalent bonding between carbon atoms within deprotonated polyacene nanoribbons, and ionic bonds between Ca1 and C2 atoms (Fig. 1). Bader charge analysis at 40 GPa for Ca1, C1, and C2 atoms yields values of 1.191, −0.378, and −0.813 (Supplementary Table 4), respectively.

Fig. 4: Results of DFT-based calculations for HP-CaC2 at 44(1) and Ca3C7 at 38(1) GPa.
figure 4

Calculated phonon dispersion curves for HP-CaC2 (a) and Ca3C7 (b). Calculated electron density of states for HP-CaC2 (c) and Ca3C7 (d); the Fermi energy level was set at 0 eV.

Calculated electron density of states of Ca3C7 suggests it to be a narrow-band p-type semiconductor (Fig. 4). Computed ELF charge density shows the formation of covalent bonds between carbon atoms in the deprotonated p-PInIn carbon chains, and strong ionic interaction between Ca and C atoms (Fig. 2). Bader charge analysis at 40 GPa for Ca1, Ca2 atoms yields values of 1.227, 1.232; for C1, C2, C3, C4 atoms: −0.718, −0.798, −0.338, −0.351, respectively (Supplementary Table 4).

To explore the thermodynamic stability of HP-CaC2 and Ca3C7 in comparison to other calcium carbides, a convex hull was constructed for the binary Сa–C system at a pressure of ~40 GPa considering the predicted phases16. Both phases observed in the present study lie on the convex hull and therefore are (theoretically) thermodynamically stable at 40 GPa (Fig. 5). Notably, HP-CaC2 also remains thermodynamically stable at 140 GPa (Supplementary Fig. 6). Although an attempt to recover HP-CaC2 and Ca3C7 at ambient conditions was unsuccessful, according to our calculations, the compounds should be dynamically stable even at atmospheric pressure (Supplementary Fig. 7).

Fig. 5: The calculated convex hull for the Ca–C binary system with predicted and observed calcium carbides at 40 GPa.
figure 5

The phases HP-CaC2 and Ca3C7 lie on the convex hull (black dashed line) and are, thus, thermodynamically stable. The phases Ca2C3 (P\(\bar{1}\)), Ca2C3 (C2/c) and CaC2 (P6/mmm) are off the convex hull and are, therefore, unstable.

Polycyclic aromatic hydrocarbons (PAHs) consisting of a finite number of fused aromatic rings give rise to various classes of polymeric compounds, if a PAH is extended into a polymer. For example, a particular class of fully conjugated polymers is represented by an extension of indenofluorene into a polymer, that is poly(indenoindene) (PInIn)22, where the alternation of five- and six-membered rings gives rise to a ladder structure. Although its targeted synthesis in bulk has remained elusive so far, very recently, on-surface synthesis of para-type oligo(indenoindene) (p-OInIn) was reported to be successful through a sequence of thermally activated reactions from an organic precursor18. In the present work, we have demonstrated the inorganic synthesis of the Ca3C7 compound in bulk with the structure featuring deprotonated para-poly-indenoindene-like chains. The HP-CaC2 synthesized in this work also gives a unique example of polyacene-like nanoribbon moieties previously unknown among any organic or metal-organic compounds. Moreover, the synthesis of single crystals of this material with fused carbon rings similar to those in benzene at pressures above 140 GPa demonstrates that, eventually, organic chemistry may be extended to truly extreme conditions.

The fact that carbides (and eventually other classes of chemical compounds) with deprotonated polyacene- and para-poly(indenoindene)-like nanoribbons can exist (and even form) at conditions of planets’ interiors may have important implications for geosciences and astrobiology. Indeed, PAHs were detected at many locations in the Universe23,24,25,26, but their origin (as well as a possible role in the emergence of life) remains controversial. The hydrolysis of carbides in planets’ interiors, on their surfaces or atmospheres may be a source of hydrocarbons. Hydrolysis products of carbides as complex as HP-CaC2 or Ca3C7 we discovered may contain PAHs. In order to test this hypothesis, a mixture of CaC2-I and Ca(OH)2 was loaded into a DAC chamber, compressed to ~40 GPa, and the whole sample was double-sided laser-heated between 2500 K and 3000 K. On decompression at ambient temperature, featureless Raman spectra were observed down to ~1.8 GPa (Supplementary Fig. 8). At this pressure, the DAC was warmed to 85 °C (above the melting point of water) using a hot plate for 72 h. Several of the Raman modes we observed (Supplementary Fig. 8) belong to C–H vibrations of saturated and aromatic hydrocarbons. Moreover, at some spots within the sample, the Raman spectra closely resemble those of pyrene (C16H10) (Supplementary Fig. 8), confirming a possible abiotic route to aromatic compounds through the hydrolysis of high-pressure carbides within planetary interiors.

Methods

Sample preparation

Calcium carbide (CaC2) in the form of pieces of technical grade was purchased from Sigma-Aldrich. Single crystals with an average size of ~0.05 × 0.05 × 0.01 mm3 were preselected using a three-circle Bruker diffractometer (SMART APEX CCD detector, Ag-Kα radiation, wavelength λ = 0.5594 Å) installed in the Bayerisches Geoinstitut (BGI, Bayreuth, Germany). All manipulations with the crystals were conducted in paraffin oil to avoid their contamination by air and/or moisture. Although XRD showed a phase purity of the selected samples (no other phases except CaC2-I13 were detected), we cannot exclude that they might contain some very small, undetectable amount of calcium oxide CaO or calcium hydroxide Ca(OH)2. In order to exclude the existence of CaO and Ca(OH)2 completely, we did additional experiments with a pre-synthesized CaC2 sample of high purity. It was prepared from distilled calcium (70.6 mg, 1.76 mmol, 1 eq.) and a slight surplus of graphite (44.4 mg, 3.70 mmol, 2.1 eq.) heated at 800 °C in high vacuum for 24 h prior to the syntheses. Calcium and graphite were sealed in a purified tantalum ampoule (~0.7 bar argon pressure) and sealed under vacuum in a second ampoule made of quartz to avoid oxidation of the inner tantalum ampoule. These ampoules were heated at 1200 °C for 14 h. The purity of the products was checked by powder X-ray diffraction. It revealed that a mixture of CaC2-I, CaC2-II, and CaC2-III had formed. No reflections indicating CaO or Ca(OH)2 impurities were found. All handling of these air and moisture-sensitive samples was carried out in inert atmospheres (argon), and they were loaded without a pressure-transmitting medium. The preselected single crystals of commercial CaC2 were loaded into BX90 diamond anvil cells27 (DAC) equipped with pairs of Boehler–Almax diamond anvils with culet diameters of 250 µm or 120 µm. The gasket was made of chemically pure Re-foil indented to ~25 μm thickness with a hole in the center of the indent of ~110 or ~60 μm in diameter. Paraffin oil was used as a pressure-transmitting medium. The in-house laser heating was performed in a continuous mode using a double-sided laser-heating setup at the BGI28. The NIR laser beam (λ = 1070 nm) was focused to a ~5 µm spot, with the temperature determined according to the gray body approximation of Planck’s law (Supplementary Fig. 10). As a rule, a sample easily absorbs laser light and its heating requires relatively low power (~5 to ~15 W from each side, depending on pressure). Heating duration varies from ~15 to ~60 s. Two or three spots were heated at each sample, and transformed materials were found in all heated spots. The pressure was determined using the Raman signal from the diamond anvils29 and additionally monitored by the equation of states (EOS) of Re gasket30.

Synchrotron X-ray diffraction studies

High-pressure SCXRD experiments were conducted on the Extreme Conditions Beamline P02.2 at PETRA III, Hamburg, Germany (λ = 0.2885 Å, beam size ~1.5 × 1.7 μm2 at FWHM, PerkinElmer XRD1621 detector), on the 13-IDD beamline at the Advanced Photon Source (APS), Chicago, USA (λ = 0.2952 Å, beam size ~3 × 3 μm2, Pilatus CdTe 1 M detector), and on the Material Science beamline ID11, ESRF, Grenoble, France (λ = 0.2952 Å, beam size ~0.5 × 0.5 μm2, Eiger 4 M CdTe detector). Prior to the collection of the SCXRD datasets, a 2D X-ray mapping was performed over the heated area in order to find the best spots for data acquisition. During the SCXRD measurements, the DACs were rotated about the vertical ω-axis in a range of ±38°, and the diffraction images were recorded with an angular step of Δω  =  0.5°.

XRD data processing

The SCXRD data analysis, including the indexing of diffraction peaks, data integration, frame scaling, and absorption correction, was performed using CrysAlisPro software package31. A crystal of orthoenstatite (Mg1.93,Fe0.06)(Si1.93,Al0.06)O6 (space group Pbca, a = 8.8117(2) Å, b = 5.18320(10) Å, c = 18.2391(3) Å) was used as a calibration standard for refinement of the instrumental parameters of the diffractometer (the sample-to-detector distance, the detector’s origin, offsets of the goniometer angles and rotations of the X-ray beam and the detector around the instrument axis). The structures of novel phases were solved with the SHELXT32 structure solution program using intrinsic phasing and refined with the Jana2006 program33. Crystal structure visualization was made with the VESTA software34.

Raman spectroscopy

The spectra were collected using a LabRam system equipped with a He:Ne laser source (excitation wavelength of 632 nm) and a DilorXY system equipped with an Ar laser source (excitation wavelength of 514 nm). Laser power in the range of 5–15 mW with the He:Ne laser and up to 100 mW for the Ar laser were applied. Raman spectra were collected in the region 1100–2000 cm−1 by means of 5 accumulations for 30 s each. The frequency resolution was 2 cm−1.

DFT calculations

The first-principles calculations were done using the framework of density functional theory (DFT) as implemented in the Vienna Ab initio Simulation Package (VASP)35. To expand the electronic wave function in plane waves, we used the Projector-Augmented-Wave (PAW) method36. The Generalized Gradient Approximation (GGA) function was used for calculating the exchange-correlation energies, as proposed by Perdew–Burke–Ernzerhof (PBE)37. The PAW potentials with the following valence configurations of 3s3p4s for Ca and 2s2p for C were used. Convergence tests with a threshold of 1 meV per atom in energy and 1 meV/Å per atom for forces led to a Monkhorst-Pack38 k-point grid of 16 × 6 × 6 for CaC2 and 9 × 5 × 5 for Ca3C7 and an energy cutoff for the plane wave expansion of 900 eV. The phonon frequencies and phonon band structure calculations were performed in the harmonic approximation with the help of PHONOPY software using the finite displacement method39 for 3 × 3 × 3 (CaC2) and 3 × 2 × 2 (Ca3C7) supercells with respectively adjusted k-points. For Brillouin zone integrations, meshes of 10 × 10 × 10 (CaC2) and 10 × 6 × 6 (Ca3C7) k-points were used within the tetrahedron method. The charge densities obtained from VASP calculations were used for Bader’s charge analysis in order to obtain the total charges associated with each atom and the zero flux surfaces defining the Bader volumes. Equation of state (EoS) and static enthalpy40 calculations were performed via variable-cell structural relaxations between 0–100 GPa for Ca3C7 and 0–150 GPa for HP-CaC2. In our calculations, temperature, configurational entropy, and the entropy contribution due to lattice vibrations were neglected.