Unveiling the Mechanisms Ruling the Efficient Hydrogen Evolution Reaction with Mitrofanovite Pt3Te4

By means of electrocatalytic tests, surface-science techniques and density functional theory, we unveil the physicochemical mechanisms ruling the electrocatalytic activity of recently discovered mitrofanovite (Pt3Te4) mineral. Mitrofanovite represents a very promising electrocatalyst candidate for energy-related applications, with a reduction of costs by 47% compared to pure Pt and superior robustness to CO poisoning. We show that Pt3Te4 is a weak topological metal with the invariant, exhibiting electrical conductivity (∼4 × 106 S/m) comparable with pure Pt. In hydrogen evolution reaction (HER), the electrode based on bulk Pt3Te4 shows a very small overpotential of 46 mV at 10 mA cm–2 and a Tafel slope of 36–49 mV dec–1 associated with the Volmer–Heyrovsky mechanism. The outstanding ambient stability of Pt3Te4 also provides durability of the electrode and long-term stability of its efficient catalytic performances.

. Survey XPS spectra of as-cleaved (black curve) and air-exposed (red-curve) Pt3Te4. The C-1s core level in both cases is shown in panel (b).

S2. Valence band
The valence band was measured for pristine Pt3Te4 and for the same surface exposed to most common ambient gases and air. Valence-band states survive with negligible modifications even in air-exposed surface. Figure S2. Valence band of pristine Pt3Te4 and the same surface after exposure to O2, H2O and air. The photon energy is 900 eV. Experiments were carried out at Elettra synchrotron, Trieste.

S3 Details of the DFT calculations:
The electronic band structure and the phonon spectrum were calculated using the Vienna Ab-initio Simulation Package (VASP). 1- 3 We used projector augmented wave (PAW) pseudopotentials with a plane wave basis. The kinetic energy cut-off of the plane wave basis was set to 400 eV. In order to perform the Brillouin zone integration, a Γ-centered 12x12x8 k-grid is used. The exchange-correlation part of the potential was treated using the generalized gradient approximation (GGA) scheme. 4 We relaxed the atomic positions of the experimental structure until the force on each atom became vanishingly small. Phonon calculations were performed within the "frozen-phonon" method with the help of PHONOPY software. 5

S4. Theoretical insights on infrared-and Raman-active phonons
The phonon frequencies (in cm -1 ) at the Γ-point of bulk Pt3Te4 are presented in Table S1. The irreducible representations and the optical activity of each of these phonon modes are also tabulated. In the primitive cell of Pt3Te4, there are a total of seven atoms, leading to 21 distinct phonon bands. For symmetry, 9 (including three pairs of degenerate modes) of such 21 phonon bands contribute to the Raman activity.

S5. Core levels in post-mortem electrode
In the post-mortem Pt3Te4-based electrode traces of residual solvents are evident, as indicated by the appearance of N-1s and S-2p core levels, absent in the as-prepared electrode. Figure S3. Survey XPS spectra for as-prepared and post-mortem electrodes.

S6. Synchrotron X-ray powder diffraction
The in-house XRD system is insufficient for phase identification and structure analysis due to the malleability of Pt3Te4 crystals. To reduce the broadening of peaks and preferential orientation, a synchrotron x-ray powder diffraction (SXRD) experiment was conducted. The SXRD patterns were collected from 100 to 480 K with the MYTHEN detector with 15 keV beam at beam line 09A, Taiwan Photon Source, National Synchrotron Radiation Research Center (NSRRC) in Hsinchu, Taiwan. The single crystal was pulverized and packed in a 0.1 mm borosilicate capillary to minimize the absorption effect. The capillary was kept spinning during data collection for powder averaging. As shown in Fig. S4a, all diffraction peaks match well with the Pt3Te4 structure (ICSD # 41372).

S7. Transport measurements
For the Hall measurements, the Pt3Te4 crystal was mechanically exfoliated onto a Si/SiO2 substrate with a 300nm oxide layer. The sample was then patterned by electron-beam lithography and metallized by electron beam deposition with Ti/Au 5/100 nm at background pressure of ~2 X 10 7 Torr. See the inset of Fig. R3 (a) for the topographic map of the device. Hall measurements were taken at a temperature of 1.6 K and at magnetic field of 0-7 T. The measured temperature dependence of the longitudinal resistivity, and the B dependence of both the longitudinal and the Hall resistivity is shown in Fig. R3. We find the temperature coefficient of resistance to be almost constant ~0.5%/K [see the red line in panel a)]. The calculated electronic band structure of Pt3Te4 clearly shows the presence of both electron and hole pockets at the Fermi energy. Thus, the obtained Hall resistivity data is fitted by a two carrier (electrons and holes) model for the Hall conductivity, given by Remarkably, we find that in spite of the presence of both the carriers, the Hall resistivity varies linearly with the magnetic field and it has a negative slope. The negative slope confirms the dominating charge carriers to be the electrons. We estimate the density of electrons to be ~2·10 21 cm -3 . The mobility of the electrons is found to be 196.7 . Figure S5. Temperature dependence Hall mobility and carrier concentration for Pt3Te4 measured at H = 6 T.

S8. Transmission electron microscopy (TEM)
TEM and HR-TEM images were obtained with FEI Tecnai G2 F30 at accelerating voltage of 300 kV. The bulk Pt3Te4 was grounded into powder and dissolved in ethanol for sample preparation.

S9. Scanning electron microscopy (SEM)
SEM experiments were performed with FEI Quanta 250 FEG at accelerating voltage of 10 kV. From the SEM images, the layered structure of pristine Pt3Te4 can be easily identified. After HER test, there is no obvious change can be found for Pt3Te4, and it remains the layered structure very well.

S10. CO poisoning: an assessment by HREELS
The adsorption of CO can be nicely followed by the HREELS technique, that is especially sensitive to CO adsorption (see our review for CO adsorption on catalytic surfaces 6 for more details), due to the high oscillating dipole. We carried out the experiment in specular conditions in order to maximize the sensitivity to dipole oscillations. These experiments were carried out with an Delta0.5 HREELS spectrometer by Specs GmbH, with a primary electron beam energy of 4 eV. Explicitly, we dosed CO onto (i) Pt3Te4 (001)

S11. Determination of the thickness of the oxide skin
Assuming an ideal situation in which a uniform layer of TeO2 is formed, we estimated the surface oxide thickness, d (Å) from Te-3d spectrum. The intensity ratio between oxidized species Io and pristine surface Im of Pt3Te4, using the following equation [8][9] : where λ and λ are the effective attenuation lengths (EALs) of TeO2 and Pt3Te4, respectively.

S12. Characterization of the sample after electrochemical treatment
To correlate the observed modifications in HER activity upon electrocatalytic treatment with changes in physicochemical and electronic properties, mitrofanovite modified by electrochemical treatments was characterized by means of XPS investigations (Fig. S11). Specifically, the oxidation treatment implies the emergence of Te(0) (Te-3d5/2 peak at 573.3 eV) and TeO2 components (38% of the total spectral area with Te-3d5/2 peak at 575.5 eV). In the reduced sample, the TeO2 component is reduced by 70%. The analysis of the Pt-4f reveals that the oxidation treatment also introduces PtO2 species (14% of the total spectra area, Pt-4f7/2 at BE=74.0 eV). Corresponding microscopical data are reported in Fig. S12. It is important to point out that XPS data clarify that no Te dissolution is induced by electrochemical treatments, as evidenced by the quantitative analysis of the Pt/Te ratio. Figure S11. Core-level spectra of Te-3d and Pt-4f for Pt3Te4 after oxidized treatment (blue curve) and after reduced treatment (brown curve). The photon energy is 1486 eV (Al Kα) and the spectra are normalized to the maximum.

S13. Band structure
To understand the topology of this system, we computed all four Z2 topological invariants. As Pt3Te4 preserves inversion symmetry, we used the parity-based method developed by Fu-Kane 11 to calculate the topological invariant. By calculating the parity eigenvalues of all the filled bands, we find the Z2 index to be (0;111). Since all the weak Z2 indices are 1, while the strong index is 0, Pt3Te4 can be described as a weak topological metal. Figure S13 reports the effects of spin-orbit coupling (SOC) on the band structure. The Brillouin zone is reported in Figure S14.

S15. Optimized atomic structures of the adopted supercell
In Fig. S16 the optimized atomic structures of the adopted supercell is shown.

S16. Comparison with previous papers on mitrofanovite
In the in Ref. 12 , authors reported results of the calculations for PtTe2 and Pt2Te2 monolayers, which constitute the sub-units of Pt3Te4. Note that in these calculations, the contributions from interlayer interactions are not considered at all. It is evident that mitrofonovite Pt3Te4 is not the simple addition of PtTe2 and Pt2Te2 units not interacting between them. Therefore, results reported in discussed paper are actually inappropriate for understanding catalytic properties of mitrofanovite, which are obviously ruled by the surface properties of bulk crystals. In addition, in the paper in Ref. 12 calculations for the monolayer PtTe2 and Pt2Te2 have been performed without optimization of lattice parameters, which instead is essential for the description of chemical properties of flexible membranes. Thus, values reported in Ref. 12 are irrelevant also with free-standing monolayers. Therefore, we can conclude that unfortunately results in Ref. 12 are not reliable. Note that the above-mentioned technical aspects regarding the modelling of layered systems are known about one decade. Since pioneering experimental reports about oxidation of mono-, bi-and multilayers of graphene at different temperatures 13 and special pathways of functionalization of graphene on SiC 14 , taking into account of the presence of sublayers and flexibility of membranes is essential for building of proper model (see for review Ref. 15 ). Turning from graphene to other layered systems also demonstrate importance of adopting proper models for DFT-based calculations. The most spectacular examples of this issue are InSe and GaSe, for which monolayers are chemically stable [16][17] in contrast to bulk materials (see our recent work 18 ).
Concerning HER modelling, obtained results in our case (1.06 eV/H + for PtTe2-termination and 0.65 eV/H+ for Pt2Te2-termination) are smaller than those calculated for monolayers (1.20 and 0.89 eV/H + , respectively) 12 . The difference in the values calculated for slab and monolayers demonstrates the significant influence of interlayer non-covalent interactions on chemical properties. Therefore, adopting a theoretical model based on monolayers as in Ref. 12 is improper for the description of catalytic properties of the mitrofanovite surface. Note that even the calculations in Ref. 12 , performed for monolayers without optimization of lattice parameters, also indicate decreasing of the free energy of this step to +0.17 eV/H + for PtTe2 and +0.29 eV/H + for Pt2Te2 12 . In the presence of the vacancies, the effect of sublayers on energetics of adsorption is more dramatic than in the case of monolayer (see discussion above), corresponding to the appearance of additional structural degrees of freedom of the top layer after defects formation. Concerning band structure in Ref. 12 , we note that the authors consider the band structure and formation energy of monolayers of PtTe2 and Pt2Te2 [see Fig. 5 (a)-(c) in Ref. 12 ) to describe the electronic and stability properties of bulk Pt3Te4. Even if we consider super-simplified model of the outermost (at the surface) unit cell of Pt3Te4 -it has multiple layers of PtTe2 and Pt2Te2 which have significant interlayer interactions between them. Extrapolating the electronic properties of monolayers for describing the bulk properties can lead to wrong conclusions such as Pt3Te4 being an insulator (see Fig. 5(a) in Ref. 2 ), while actually it is a metal with a large carrier concentration. This also leads to wrong estimation of the formation energy.
Concerning experimental data, in Figure 4a of the paper in Ref. 12 , in correspondence of an overpotential of 0 V vs RHE, both the current densities for Pt3T4 nanocrystal and Pt/C samples are clearly larger than 0, which is quite puzzling. This result indicates that the reference electrode they used may not be calibrated correctly, and thus the HER data they reported are not really solid.

S17. Nyquist plots
The Nyquist plots of the pristine and the electrochemically treated Pt3Te4 were tested in 0.1 M KCl solution containing 5mM [Fe(CN)6 ] 3-/ 4− (SI, Fig. S17). The charge transfer resistance (Rct) of the pristine, oxidized and reduced Pt3Te4 samples are fitted to be 102.9, 84.2 and 76.5 Ω, respectively. This result indicates the slightly faster charge transfer kinetics on the oxidized and reduced Pt3Te4 than on the pristine Pt3Te4, which agrees with the cyclic voltammetry test.