Novel rhenium carbides at 200 GPa

Laser heating of rhenium in a diamond anvil cell to 3000 K at about 200 GPa results in formation of two previously unknown rhenium carbides, hexagonal WC-type structured ReC and orthorhombic TiSi2-type structured ReC2. The Re-C slid solution formed at multimegabar pressure has the carbon content of ca. 20 at%. Unexpectedly long C-C distances (ca. 1.76-1.85 A) in 'graphene-like' carbon nets in the structure of ReC2 cannot be explained by a simple covalent bonding between carbon atoms and suggest that at very high pressures the mechanism of interaction between carbon atoms in inorganic compounds may be different from that considered so far.

Transition-metal carbides belong to a large group of industrially important materials.
The rhenium-carbon system provides a striking example of the pressure effect on elements reactivity and the binary phase diagram. At ambient pressure, rhenium does not form stoichiometric carbides; carbon dissolves into rhenium up to 28.45 at% at the eutectic temperature (2778 K). [8] However, even very moderate pressure, just above 6 GPa (and high temperatures) was reported to promote formation of a Re-C compound. [9,10] Its correct chemical composition (Re2C) and crystal structure of anti-MoS2 type (hexagonal primitive, hP, space group P63/mmc) were established relatively recently on the basis of X-ray powder diffraction, Raman spectroscopy data and DFT calculations. [11,12] No other stoichiometric carbides apart of hP-Re2C have been observed at pressures up to ~70 GPa and temperatures ~4000 K. [3] The hP-Re2C was found to be isostructural with Re2N. [11] This analogy and the recently observed very complex and unexpected behavior of the Re-N system, featuring numerous nitrogen-rich compounds, [5,13] stimulated the study of potential reactions between Re and C at multimegabar pressures. Awareness of these reactions is also of a primary interest for the development of the methodology of ultra-high pressure high temperature experiments, in which Re gaskets are commonly used. The range of currently achievable static pressures has been extended to ~1000 GPa due to implementation of double-stage diamond anvil cells (dsDAC) and to ~600 GPa with toroidal type anvils (tDAC). [14][15][16][17] In order to achieve such extreme pressures, the linear size of samples and sample chambers should be drastically decreased. At pressures above ~150 GPa, a pressure chamber's diameter (made, as a rule, of Re) is usually smaller than 50 μm, and in dsDACs above 300 GPa it is less than 10 μm. Meanwhile, the size of a laser beam in typical laser heating (LH) setups used in DAC experiments varies from 15 to 50 μm at FWHM. [18,19] As a result, irradiation of, at least the edge of a Re-gasket, by the laser beam during laser heating becomes unavoidable and may lead to a chemical reaction between Re and carbon of the diamond anvils. Therefore, correct interpretation of the results of experiments with laser-heating at ultra-high static pressures requires knowledge about possible products of rhenium and carbon interaction.
Here, we report on the in situ study of Re-C compounds formed due to chemical interactions between diamond anvils and the rhenium gasket after pulsed laser heating in DACs at about 200 GPa. The structures of all of the synthesized rhenium carbides, Re2C, ReC2, ReC, and ReC0.2 were solved and refined using single-crystal X-ray diffraction (SCXRD) providing direct and unequivocal data to judge on both the atomic arrangement and chemical composition of the crystalline matter.
Two DACs have been prepared, dsDAC for Experiment №1 and conventional assembly for Experiment №2. Secondary anvils for Experiment №1 were made from nano-crystalline diamond (NCD) spheres of about 15-20 µm in diameter. The dsDAC in Experiment №1 was compressed up to ~415 GPa (according to the diamond Raman shift on the secondary anvil), [20] then the sample (Re flake) was laser heated until the first bright flash of light that held for less than a second (temperature was not measured). After the first heating attempt pressure dropped down to ~200 GPa probably due to failure of the secondary anvils. However, the DAC remained intact and rhenium in the central area was again carefully laser heated in a pulsed mode (1 µs pulses, 25 kHz repetition rate, and maximum temperature of about 3000 K). [21] Pressure did not change upon the repeated laser heating (Supporting Information, Figure S1) and the DAC with the temperature-quenched material was investigated using powder and single crystal X-ray diffraction (for details see the Supporting Information). A 2D diffraction map collected across the sample chamber revealed the presence of not only Re, but four additional phases ( Figure 1). All of them have been identified and are described in detail below.  Figure S3). This corresponds to pressures of ~230-240 GPa according to equation of state (EOS) from Dubrovinsky et al., [14] or ~190-195 GPa, according to Anzellini et al. [22] (Supporting Information, Table S1).
As written above, in this experiment the pressure, as determined from the Raman shift of the diamond anvil (Supporting Information, Figure S1), was of about 200 GPa, whereas the Re EOS gave up to 240 GPa. These values do not match the pressure of 172(13) GPa, determined for the given unit cell volume of hP-Re2C according to the EOS reported by Juarez-Arellano et al. [3] The EOS of Re2C in ref. [3] was determined on the basis of powder XRD from a sample in hard pressure transmitting medium that led to a significant uncertainty in the bulk modulus: K=405(30) GPa (K´=4.6). [3] Comparison of our data with literature motivated us to make an independent measurement to establish the EOS of hP-Re2C on the basis of SCXRD. In the experiment described in Supporting Information, Experimental Procedures, the P-V data were obtained up to 50 GPa from a single crystal of hP-Re2C pressurized in a soft (Ne) pressure transmitting medium (Supporting Information, Table S3, Figure S5). The parameters of the 3 rd order Birch-Murnaghan equation of state of hP-Re2C (V0=69.18(4) Å 3 /unit cell, K=375 (15) GPa, K´=5.0(1)) we obtained only slightly (within uncertainties) differ from the values reported by Juarez-Arellano et al. [3] According to our EOS hP-Re2C synthesized in Experiment №1 was under the pressure of 180 (7) Table S2) determined from SCXRD is in a good agreement with model described by Friedrich et al. [11] It is characterized by the stacking sequence AABB of layers of Re atoms  Table S2, Figure S4b). The structure solution and refinement revealed that hP-ReC belongs to the WC structure type (P-6m2, №187) with characteristic c/a ratio (~0.94).
Like in Re2C, carbon atoms are located in trigonal prisms formed by Re atoms with the Re-C distances equal to 2.000(2) Å (Figure 2f).
Remarkably, the same orthorhombic phase (a=3.2880(9) Å, b=4.2088(9) Å, c=5.5645(8) Å, and V= 77.00(3) Å 3 ) was found in Experiment №2, in which iron oxide (FeO) in a Ne pressure medium was compressed to 219(5) GPa and pulsed-laser heated up to ~2500 K. [19] The relatively large beam (of about 25 µm at FWHM) irradiated the Re gasket used in Experiment №2 and ReC2 was found at the border of the pressure chamber. Carbon-rhenium layers stack along b-axis with a translation (½, 0, ½) (Figure 2c). Whereas the Re-C layers are obviously helpful to give a clear geometrical presentation of the oF-ReC2 structure, the structure is not "layered", as the shortest Re-Re distances between the "layers" (~2.72 Å) are similar to those in hP-Re2C (~2.5-2.6 Å) and hP-ReC (~2.55-2.70 Å) at the same pressure. However, a strong diffuse scattering, especially evident in the reciprocal space, suggests a disorder in the stacking of Re-C "layers" along the b-axis -(Supporting Information, Figure S6a).

For the last two cases, ab initio simulations suggest a significant covalent interaction between
Si atoms in zig-zag chains geometrically similar to the C-C chains in our oF-ReC2 carbide. [27,28] At the same time, the shortest C-C distances in hP-ReC and hP-Re2C at 180(7) GPa (in this study, Experiment №1) are 2.5510(9) Å and 2.5860(9) Å, correspondingly, and these values agree with those known for transition metal carbides containing isolated carbon atoms (~2.8 Å and above at ambient pressure). Thus, the nature of chemical bonding in oF-ReC2 is probably different from that in other rhenium carbides -hP-ReC and hP-Re2C.
The phase map built on the basis of powder XRD data ( Figure 1) shows that several carbides (Re2C, ReC, and ReC2) formed within the pressure chamber, and one more phase has been synthesized at its periphery, at the border with the Re gasket. The diffraction patterns are mostly characterized by powder rings, but in a few points in the map it was possible to collect singlecrystal data sets (Supporting Information, Figure S4d Table S2). This may be interpreted as Re at about 100 GPa, according to the EOS of Anzellini et al. [22] or at about 115 GPa, according to the EOS of Dubrovinsky et al. [14] Indeed, single-crystal data suggest that rhenium atoms form hexagonal closed packing as expected for Re. However, the electron density, localized in the octahedral voids of the hcp-Re, is arranged like in the B8 (NiAs)-type structure (Figure 2d) and suggests this phase to be a Re-C interstitial solid solution based on the B8-type structure (Figure 2d, h). Single-crystal diffraction data at hands are not sufficient to refine the carbon atoms occupancies, especially in the case of such a huge difference in X-ray scattering factors of Re and C. Fortunately, we noticed that the unit cell volumes per atom for RexCy compounds (Re, [14] hP-Re2C, hP-ReC, oF-ReC2, and C (diamond) [29] ) at 180 (7) GPa, if plotted as a function of carbon content, all appear along the common straight line (Figure 3b). Considering the volume of the B8 rhenium-carbon solid solution, we have estimated its composition as ReC0.2, which shows that multimegabar pressures do no increase carbon solubility in rhenium, compared to previously observed compositions. [8]  A comparison of our data for Re-C compounds and recent reports on the Re-N system demonstrates obvious crystal-chemical similarities in the two systems for compositions Re:C≥1:1. Two carbides, Re2C and ReC, are isostructural to Re2N and ReN. They are built up of CRe6 or NRe6 trigonal prisms, and even Re-C and Re-N interatomic distances at the same pressures are very similar. [4,5] Contrary, the structures of ReC2 and ReN2 [5] have nothing in common and there are no signs of formation of rhenium polycarbides with more than two carbon atoms per a formula unit, unlike to the rhenium-nitrogen system. [13] Possible explanation of such differences may be a tendency of nitrogen to form di-nitrogen and polynitrogen anions (based on N4 units [13,[30][31][32] ) at high-pressures and high-temperatures (HPHT), whereas rhenium carbides do not reveal formation of polycarbon anions at such conditions.
There is a number of experimental works and theoretical predictions that suggest progressive polymerization of carbon atoms at high pressures. [33][34][35][36][37][38] In all cases polymerization results in formation of short (~1.6 Å or shorter) C-C bonds for single bonded carbon atoms. For hexagonal carbides hP-Re2C or hP-ReC at ~200 GPa, the C-C distances are longer than 2.5 Å, which suggest the absence of chemical bonding between carbon atoms in the crystal structure.
In case of rhenium dicarbide, oF-ReC2, we observed formation of graphene-like' carbon layers, in which the C-C distances (~1.76-1.85 Å) are much longer than expected for sp 2 -or sp 3bonded carbon atoms in inorganic compounds; still, in alkanes the C-C bond length as long as 1.704 Å has been detected. [39] It suggests that at very high pressures a mechanism of interaction between carbon atoms in inorganic compounds may be different from that considered so far.
To summarize, we have extended the knowledge about the chemical interaction between rhenium and carbon to ~200 GPa. Two novel carbon-rich rhenium carbides -WC-type structured hP-ReC and TiSi2-type structured oF-ReC2were synthesized (hitherto only Re2C was known in the Re-C system). The fact that rhenium and carbon can produce numerous compounds at HPHT conditions should be taken into account upon planning of experiments in LHDACs at multimegabar pressures.