Unpredicted but It Exists: Trigonal Sc2Ru with a Significant Metal–Metal Charge Transfer

The Sc2Ru compound, obtained by high-temperature synthesis, was found to crystallize in a new trigonal hP45 structure type [space group P3̅m1; a = 9.3583(9) Å and c = 11.285(1) Å]: Ru@Sc8 cubes, Ru@Sc12 icosahedra, and uncommon Ru@Sc10 sphenocoronae are the building blocks of a unique motif tiling the whole crystal space. According to density functional theory studies, Sc2Ru is a metallic compound characterized by multicenter interactions: a significant charge transfer occurs from Sc to Ru, indicating an unexpectedly strong ionic character of the interactions between the two transition metals. Energy calculations support our experimental results in terms of stability of this compound, contributing to the recurrent discussion on the limits of the high-throughput first-principles calculations for metallic materials design.

A comparison between experimental data and high-throughput first principles calculation results is illustrated in Figure S1 for the Sc-Ru system. In general, there is a good correlation between them except the cases indicated by red numbers for which, according to high-throughput data: 1) the Sc5Ru3 is unstable at low temperatures; 2) the experimentally obtained Sc2Ru (hP96-Ti2Ni) is far away from the decomposition line: there are several 2:1 stoichiometry structural models (including the ground state tI6-Mo2Si) that possess noticeably lower formation energies; 3) the Sc57Ru13 is slightly unstable at low temperatures. Figure S1. Sc-Ru convex-hull taken from http://www.aflow.org/aflow-chull implemented within the opensource, ab initio framework http://www.aflow.org. Red markers indicate discrepancies between experimental and high-throughput data.

Synthesis and SEM-EDXS characterisation
Stoichiometric amounts of Ru and Sc, with nominal purities > 99.9 mass% were used for synthesis of an alloy of ca. 0.8 g. The sample was arc melted several times under Ar atmosphere on a watercooled copper hearth with a tungsten electrode; weight losses were less than 1 mass%.
Subsequently, the obtained alloy was put in an alumina crucible, sealed in a silica tube under Ar atmosphere and heat treated at 1000 °C for 7 days. Annealing was followed by water quenching.
In order to obtain smooth surfaces suitable for microscopic analysis, some specimen fragments were embedded in a phenolic hot mounting resin with carbon filler, employing an automatic hot compression mounting press Opal 410 (ATM GmbH, Germany). The grinding procedure with abrasive SiC papers (grain size from 600 to 1200) was followed by diamond pastes (6, 3 and 1 μm) polishing. Microstructure examination (see Figure S2) and phase composition analyses were

X-ray studies and crystal structure solution
The single crystal selection was conducted on the mechanically fragmented alloy. X-ray diffraction measurements were performed by a Bruker Kappa APEXII CCD area-detector diffractometer, using Mo Kα ( = 0.71073 Å) radiation. Datasets were collected operating in -scan mode over the reciprocal space up to  30°, with an exposure time of 30 seconds per frame. The crystal-todetector distance was fixed to 5 cm. Data integration, Lorentz polarization, and semiempirical absorption corrections were applied to all data by using the SAINT and SADABS softwares. 1 Crystal structure refinement was carried out by full-matrix least-squares methods on |F| 2 using the SHELXL program 2 as implemented in WinGX 3 . The indexation of collected intensity data yielded to a trigonal symmetry unit cell with a=9.35 Å and c=11.29 Å (precession images of h0l and hk0 zones are shown in Fig. S3). a) b) Figure S3. Reconstructed intensity profiles for h0l (a) and hk0 (b) zones of the Sc2Ru diffraction pattern.
Since no systematic absences were found, the following space-groups were suggested by XPREP 1 : P3 (143), P-3 (147), P321 (150), P3m1 (156), P-3m1 (164). The lowest combined figure of merit was associated to the last one. Indeed, the crystal structure was quickly solved in this space-group by means of the charge-flipping algorithm implemented in JANA2006 4 figuring out the almost complete structural model with 45 atoms per cell (5 Ru and 7 Sc Wyckoff sites). Site occupancy factors (SOF) and possible statistical mixture were carefully checked for each species in separate refinement cycles, but no significant deviation from an ordered scenario was detected, suggesting Sc2Ru to be a stoichiometric compound. This is also in perfect agreement with the measured EDXS composition. A further refinement including anisotropic thermal displacement parameters resulted in acceptable residuals and flat difference Fourier map (Table S1). The generated CIF file has been deposited at Cambridge Crystallographic Data Centre (code-2071469). Details of the data collection and refinement are listed in Table S1. Interatomic distances for Sc2Ru (hP45) -in Table S2 and constructed histograms are shown in Figure S4.   The X-ray powder diffraction pattern was collected using a Philips X'Pert MPD (Cu Kα radiation, scanning step ca. 0.015°) diffractometer; indexing was performed by Powder Cell software 5 (see Figure S5).

Computational details and results
DFT total energy calculations have been performed for different Sc2Ru structural models, relaxed at the equilibrium geometry (in P1 space group) using both the plane wave pseudopotential software QUANTUM-ESPRESSO (QE) 6 and the all-electron, full-potential FHI-aims package 7,8 . In both cases, the PBE functional for the exchange and correlation energy was applied. Details on k-point mesh chosen to sample the Brillouin zone are reported in Table S3. The QE calculations were conducted with the recommended projector-augmented wave (PAW) 9 pseudopotentials, available at the PSLI library 10 . The semicore 3s and 3p states for Sc and the 4s and 4p for Ru were treated as valence electrons. The plane-wave and density cut-off were set to 56 Ry and 540 Ry, respectively. The orbital occupancies at the Fermi level were treated with a Gaussian smearing of 0.01 Ry. In order to make a comparative analysis of our results with those available at the AFLOW repository, the QE obtained energies have been discussed in the main text for consistency since both QE and VASP are DFT codes based on pseudopotentials. The FHI-aims code was also then used to check the reproducibility of the results. For this purpose, predefined default "tight" basis sets for Sc and Ru were selected including scalar-relativistic effect according to the ZORA approximation. A Gaussian smearing of 0.01 eV was set.
The formation enthalpies (fH) were evaluated using the calculated total-energies per atom (E), according to the following formula: ∆ (Sc) + .
These values are listed in Table S4 and shown in Figure S6, together with those available at the AFLOW library calculated with VASP code.

S-8
It has to be noted that, though the values of fH slightly differ, the energy differences between the already reported structural models obtained by G.L.W. Hart et al. 11 are very well reproduced with the computational procedures applied in this work.  S-10