Exposed facet-controlled N2 electroreduction on distinct Pt3Fe nanostructures of nanocubes, nanorods and nanowires

Abstract Understanding the correlation between exposed surfaces and performances of controlled nanocatalysts can aid effective strategies to enhance electrocatalysis, but this is as yet unexplored for the nitrogen reduction reaction (NRR). Here, we first report controlled synthesis of well-defined Pt3Fe nanocrystals with tunable morphologies (nanocube, nanorod and nanowire) as ideal model electrocatalysts for investigating the NRR on different exposed facets. The detailed electrocatalytic studies reveal that the Pt3Fe nanocrystals exhibit shape-dependent NRR electrocatalysis. The optimized Pt3Fe nanowires bounded with high-index facets exhibit excellent selectivity (no N2H4 is detected), high activity with NH3 yield of 18.3 μg h−1 mg−1cat (0.52 μg h−1 cm−2ECSA; ECSA: electrochemical active surface area) and Faraday efficiency of 7.3% at −0.05 V versus reversible hydrogen electrode, outperforming the {200} facet-enclosed Pt3Fe nanocubes and {111} facet-enclosed Pt3Fe nanorods. They also show good stability with negligible activity change after five cycles. Density functional theory calculations reveal that, with high-indexed facet engineering, the Fe-3d band is an efficient d-d coupling correlation center for boosting the Pt 5d-electronic exchange and transfer activities towards the NRR.


INTRODUCTION
Ammonia (NH 3 ), an essential composition for fertilizer feedstock, industrial and chemical precursors [1], not only plays a vital role in the development of ammonia fuel cells but is also a promising candidate for the hydrogen economy because of its high energy density and large hydrogen capacity [2]. However, from a thermodynamic point of view, the high bond energy of the triple bond in nitrogen (N 2 ) makes it an intricate multi-step reaction to convert N 2 to NH 3 [3]. Currently, the Haber-Bosch method is most widely applied because efficient synthesis of NH 3 from N 2 and H 2 has reached industrial-scale yields [4]. However, this method requires conditions of high temperature of 400-500 • C and pressure of 200-250 bar [5]. It also consumes 1-2% of the global annual energy supply and is responsible for >1% of global CO 2 emissions [6]. Therefore, a more effective process for NH 3 synthesis under mild conditions is highly desirable.
Recently, tremendous efforts for converting N 2 to NH 3 have been made in pursuit of efficient and sustainable catalysis with use of biocatalytic, photocatalytic and electrocatalytic methods [7][8][9][10][11][12][13][14]. One possible strategy for NH 3 synthesis is electroreduction of N 2 to NH 3 in which the N 2 reduction reaction (NRR) process can be operated by renewable electricity energy, and operated at mild temperature and pressure [15][16][17][18]. There have been several investigations on NRR with noble-metal catalysts (Ru [19], Au [20], Pd [21] and Rh [22]), as summarized in Table S1. More recently, several efforts have been devoted to optimizing electrocatalysts to enhance NRR, including size regulation [23], crystal engineering [24], ion incorporation [25], introduction of defect sites [26,27] and component regulation [28]. However, to the best of our knowledge, surface structure regulation, as one of the most effective strategies to precisely tune catalytic properties, has not yet been demonstrated for NRR. C The Author(s) 2020. Published by Oxford University Press on behalf of China Science Publishing & Media Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted reuse, distribution, and reproduction in any medium, provided the original work is properly cited.
Previous research revealed that the NRR performance is gravely limited by linear scaling of the two vital intermediate energetics between * N 2 H and * NH 2 ( * indicates the adsorption site). Further theoretical analyses disclose that Fe and Pt atoms can effectively address the energetics of * N 2 H and * NH 2 , respectively, synergistically providing efficient active sites to enhance NRR activity [29,30]. Herein, we report a facile method for selectively growing Pt 3 Fe nanocubes (NCs), Pt 3 Fe nanorods (NRs) and Pt 3 Fe nanowires (NWs) (Supplementary Fig. 1

RESULTS AND DISCUSSION
A simple wet-chemical method is adopted to control the synthesis of Pt 3 Fe nanocrystals. Taking Pt 3 Fe NWs synthesis as an example, potassium tetrachloroplatinate (II) (K 2 PtCl 4 ) and iron nonacarbonyl (Fe 2 (CO) 9 ) are chosen as metal precursors, ribose is used as the reducing agent, cetyltrimethyl ammonium chloride (CTAC) and oleylamine (OAm) are applied as the surfactant and solvent, respectively. Uniform Pt 3 Fe NWs with average diameter of 15 nm and zigzag border along the whole length are obtained ( Fig. 1a and Supplementary  Fig. 2), characterized by high-angle annular darkfield scanning transmission electron microscopy (HAADF-STEM) and TEM. The X-ray diffraction (XRD) pattern (Fig. 1b) Fig. 3) [31]. The crystal structure model of the Pt 3 Fe NWs represents a primitive cubic structure. It is composed of a periodic square matrix of Fe and Pt, which are located at the corner and face center of each unit cell, respectively. The HAADF-STEM image and elemental mappings show the elemental distributions of Pt and Fe (Fig. 1c). The Pt (green), Fe (red) and mixed images indicate that all elements are evenly distributed on the Pt 3 Fe NWs, confirming the alloyed structure ( Supplementary Fig. 4). The selected area electron diffraction (SAED) image (Fig. 1d) further indicates that the Pt 3 Fe NWs have good crystallinity and fcc structure. Spherical aberration correction HRTEM images reveal the distinct lattice fringes. The measured value of lattice spacing is 0.195 nm. In addition, a high-index facet of {311} can be readily observed ( Fig. 1e and Supplementary Fig. 5), which is reported to exhibit much higher catalytic performance compared to most common facets ({200} and {111}) because of the high density of atomic steps and ledges [32][33][34]. Notably, when ribose is replaced by maltose, while maintaining the other synthetic parameters unchanged, Pt 3 Fe NCs with average diameter of 5 nm are obtained ( Fig. 1f and Supplementary Fig. 6). Meanwhile, Pt 3 Fe NRs with average diameter of 3 nm are realized by merely changing the precursor of Pt ( Fig. 1g; Supplementary Figs 7 and 8). Detailed X-ray photoelectron spectroscopy (XPS) is carried out to determine the electronic properties of surface Pt atoms. Compared with commercial Pt/C, the Pt 0 4f 7/2 binding energy of Pt 3 Fe nanocrystals has a negative shift from 71.2 to 70.6 eV, indicating that the charge is transferred from Fe to Pt, because of the lower electronegativity of Fe ( Supplementary Fig. 9a) [35]. We can see that the Pt 4f spectra show two distinct peaks, which are assigned to the Pt 4f 7/2 and Pt 4f 5/2 orbit levels. Each peak can be further split into two doublets, which are assigned to Pt 0 and Pt 2+ chemical valence states. It is clearly shown that the majority of the Pt of Pt 3 Fe is in mainly metallic state. The Pt(II)/Pt(0) ratio is summarized in Table S2. The XPS spectra of Fe 2p reveal two distinct peaks at 710.3 and 724.1 eV, which are assigned to the characteristic peaks of Fe 2p 3/2 and Fe 2p 1/2 orbit levels, respectively. The results indicate that the majority of the Fe is mainly in an oxidized state ( Supplementary Fig. 9b) [36]. The lattice spacings of Pt 3 Fig. 17). During the NRR progress, the Pt 3 Fe NWs/C exhibited negligible decay in current density under different applied potential, indicating the good corrosion resistance ability (Fig. 2a). The Pt NCs/C ( Supplementary Fig. 18) had very poor NRR activity compared with Pt 3 Fe NCs/C, indicating that introduction of Fe is essential for the NRR. In addition, the NRR activity of Pt 3 Fe NRs/C exhibited much improvement compared with Pt 3 Fe NCs/C. The average NH 3 yield and Faraday efficiency (FE) of the Pt 3 Fe NWs/C under different applied potentials is given in Fig. 2b and c. The NRR was performed at 0 V, the measured yield of NH 3 was 10.
indicating that it consumed less energy to effectively convert N 2 into NH 3 . The NRR performance of Pt 3 Fe NWs/C was much better than that of the Fe 3 O 4 catalysts ( Supplementary Fig. 19). The maximum yield of NH 3 was further quantitatively determined by an indophenol blue method [37]. As observed, the calculation value of NH 3 yield was close to the quantitative result from the Nessler reagent method ( Supplementary Figs 20 and 21). The FE linearly decreased when it reached a more negative potential, because of the existence of a competition reaction between NRR and hydrogen evolution reaction (HER) [38]. We compared the NRR activity with different reaction temperatures at −0.05 V to assess the apparent activation energy and investigate the different exposed facets effect of Pt 3 Fe nanocrystals. As observed, the NH 3 yield enhanced with increasing reaction temperature due to faster mass transfer rate of reactants ( Supplementary Fig. 22). The estimated apparent activation energies were 11.0, 16.9 and 27.3 kJ mol −1 for Pt 3 Fe NWs/C, Pt 3 Fe NRs/C and Pt 3 Fe NCs/C, respectively (Fig. 2d). That is, the Pt 3 Fe NWs/C bounded with high-index facets could significantly decrease the apparent activation energy and hence enhance the NRR activity. Significantly, the NH 3 yield and FE of Pt 3 Fe NWs/C were higher than those of Pt 3 Fe NCs/C and Pt 3 Fe NRs/C (Fig. 2e), indicating that different surface structures of Pt 3 Fe nanocrystals indeed have essential influence on the NRR activity and the high-index facets of Pt 3 Fe nanocrystals play a vital role in the improvement of NRR activity.
We then carefully examined the N source of the produced NH 3 . There were no distinguishable peaks in the N 1s region, indicating that no nitrogenous species existed on the surface of synthetic Natl Sci Rev, 2021, Vol. 8, nwaa088 catalysts ( Supplementary Fig. 23). Replacing N 2 by Ar, while maintaining other experimental parameters unchanged, no NH 3 was detected ( Supplementary Fig. 24). The same result was observed when Pt 3 Fe NWs/C was replaced by carbon powder loaded on carbon paper (Supplementary Fig. 25). The amount of NH 3 increased linearly to the electrocatalysis time in N 2 -saturated electrolyte, which indicated that the produced NH 3 came from the NRR process ( Supplementary  Fig. 26). Considering the small amounts of NH 3 and NO x in atmosphere and feeding gas, we kept high-purity Ar and N 2 continuously flowing into the KOH electrolyte without applied potential.  Fig. 30). In addition, by varying the N 2 flow rate, the current density exhibited small change. The tiny fluctuation of FE and NH 3 yield implied that N 2 diffusion was not the ratedetermining step (Supplementary Fig. 31) [39].
The intrinsic reason for higher catalytic activity of Pt 3 Fe NWs/C was investigated. The Tafel slope is a vital parameter to evaluate the HER mechanism [40,41]. The higher value of 163.8 mV dec −1 gained from Pt 3 Fe NWs/C suggested its sluggish HER kinetics, which may in turn enhance the NRR performance ( Supplementary Fig. 32), further confirmed by theoretical calculation (Supplementary Fig. 33). After that, surface valance spectra was created to investigate the relationship between electronic effect and the binding strength of adsorbates. As observed, the d-band center shifts upwards from Pt NCs/C (−3.69) to Pt 3 Fe NCs/C (−3.31) when introducing Fe to Pt NCs/C. For transition metals, the lower binding energy of d-band center would cause weaker bonding between the adsorbates and metal surface [42][43][44]. Accordingly, the Pt 3 Fe NWs/C bounded with high-index facets could enable strong bonding with N 2 and may boost NRR activity (Supplementary Fig. 34).
We used DFT calculations to further interpret the NRR activity differences of Pt 3 Fe (311), (111) and (200). The bonding and antibonding orbitals near the Fermi level (E F ) exhibit an electron-rich distribution on the Pt 3 Fe (311) surface (Fig. 3a), while (111) and (200) surfaces present less electronlocalizing on the surface (Fig. 3b and c). The preeminent d-electron exchange and transfer activities (d-EXTA) on the high indexed surface have been demonstrated through projected partial density of states (PDOSs) analysis (Fig. 3d-f). The edge of dominant peak of Pt-5d band reflects a direct determination of 5d-EXTA for efficient N 2 fixation. Taking this trend, on the (311) surface, the Fe-3d orbital clearly merges at the E F without an evident gap between the e g and t 2g components. The energetic interval between Pt-5d and Fe-3d is nearly 0.95 eV, which cost less energetic barrier for transferring delectrons above E F to N 2 2p orbital (Fig. 3d). The (111) and (200) present nearly two-times higher to freely cross E F (Fig. 3e and f). The comparison of energetic trends reveals that the NRR pathway on the (311) is the most energetically beneficial with an energetic barrier of merely 0. 16 (Fig. 3g). In the distal reaction pathway, the N atom furthest away from the catalyst surface preferentially undergoes hydrogenation, to form the first NH 3 molecule. The NH 3 molecule is released after the N≡N triple bond breaks. The remaining N atom continues under hydrogenation to generate another NH 3 molecule. According to the above results, the reaction path of our work could be a distal reaction pathway. We further reason that the better activity of NRR on the (311) is attributed to the excellent energetic preference of N 2 fixation, while the underperformance of (200) for NRR is ascribed to overbinding of adsorbing H for efficient N-hydrogenation (Fig. 3h). The energetic trend is consistent with analysis of electronic activities from both experimental and theoretical perspectives. A chronoamperometric test was first conducted at −0.05 V in N 2 -saturated KOH electrolyte to evaluate the durability of Pt 3 Fe NWs/C. The current density (Fig. 4a) exhibited negligible decay after 30 h electrolysis. The stability of Pt 3 Fe NWs/C was also evaluated by successive cycle electrolysis at −0.05 V. After five successive cycles, the total current density exhibited no evident fluctuation ( Supplementary Fig. 35). The NH 3 yield and FE of Pt 3 Fe NWs/C were measured after each cycle, with no obvious changes (Fig. 4b). After stability testing, TEM image, elemental mapping ( Supplementary  Fig. 36) and SEM-EDS ( Supplementary Fig. 37) of the Pt 3 Fe NWs/C confirmed that its structure and composition were largely maintained, demonstrating that the Pt 3 Fe NWs/C was stable enough for NRR.

CONCLUSION
In summary, we have demonstrated facile synthesis of Pt 3 Fe nanocrystals with tunable morphologies (NC, NR and NW) to evaluate the NRR performance on different exposed facets. The detailed studies show that the Pt 3 Fe nanocrystals exhibit shape-dependent electrocatalytic activity towards NRR. Notably, the Pt 3 Fe NWs bounded with high-index facets exhibit much improvement in NH 3 yield (18.3 μg h −1 mg −1 cat , 0.52 μg h −1 cm −2 ECSA ), FE (7.3%) as well as selectivity (no N 2 H 4 is detected) at −0.05 V under ambient conditions, which are much better than that those of {200} facet-enclosed Pt 3 Fe NCs and {111} facet-enclosed Pt 3 Fe NRs. The Pt 3 Fe NWs also show durable electrochemical stability with no obvious activity decay after five successive electrolysis cycles. DFT calculation reveals that strong d-d coupling between Pt and Fe sites bridges the electron transfer for prominent NRR. This work provides the first example of the fundamental correlation between exposed surfaces and NRR performances of distinct nanocrystals.

Preparation of Pt 3 Fe NWs
In preparation of monodisperse Pt 3 Fe NWs, K 2 PtCl 4 (10.4 mg), Fe 2 (CO) 9 (4.6 mg), ribose (45 mg), CTAC (32 mg) and OAm (5 mL) were added to a reaction bottle (volume: 35 mL), which was capped and ultrasonicated for 1 h. The reaction bottle was heated from room temperature to 180 • C within 0.5 h and maintained at 180 • C for 5 h in an oil bath. After cooling to room temperature, the obtained products were collected by centrifugation and washed three times with a cyclohexane/ethanol mixture.

Preparation of Pt 3 Fe NRs
The synthesis of monodisperse Pt 3 Fe NRs was similar to that of Pt 3 Fe NWs, except Pt(acac) 2 (9.8 mg) was used as precursor and the amount of ribose was 90 mg at the beginning.

Preparation of Pt 3 Fe NCs
The synthesis of monodisperse Pt 3 Fe NCs was similar to that of Pt 3 Fe NWs, except maltose (108 mg) was used as reductant at the beginning.

Preparation of Pt NCs
In preparation of monodisperse Pt NCs, Pt(acac) 2 (10 mg), PVP (200 mg), formaldehyde solution (40%, 2.5 mL) and benzyl alcohol (10 mL) were added to a reaction bottle. After sonication for 0.5 h, the homogeneous solution was transferred to a 20 mL Teflon-lined stainless autoclave and then heated at 150 • C for 10 h. The product was collected via centrifugation and further washed with an ethanol-acetone mixture.

Preparation of Fe 3 O 4 NPs
In preparation of monodisperse Fe 3 O 4 NPs, Fe 2 (CO) 9 (9.2 mg), ribose (45 mg), OAm (4.8 mL) and OAC (0.2 mL) were added to a reaction bottle, which was capped and then ultrasonicated for 1 h. The reaction bottle was heated from room temperature to 200 • C within 0.5 h and maintained at 200 • C for 5 h in an oil bath. After cooling to room temperature, the obtained products were collected by centrifugation and washed three times with a cyclohexane/ethanol mixture.

Catalytic measurements
Firstly, different morphology of Pt 3 Fe nanocrystals, Fe 3 O 4 NPs, commercial carbon, cyclohexane (1 mL) and ethanol (8 mL) were added to a reaction bottle. The Pt NCs, commercial carbon, ethanol (1 mL) and acetone (8 mL) were also added to a reaction bottle. After sonication for 1 h, the homogeneous solution was centrifuged and dried naturally to obtain a powder. Then, the powder was annealed at 150 • C for 1 h under ambient atmosphere. To prepare catalysts, the above powder (5 mg), Nafion solution (6 μL, 5 wt%) and absolute isopropyl alcohol (500 μL) were mixed and sonicated for 0.5 h to form homogeneous ink, which was then dropped (10 μL) evenly on carbon paper with geometric area of 1 × 1 cm 2 . The carbon paper was dried under ambient conditions.
The electrochemical tests were performed in a gas-tight two-chamber electrolytic cell separated by Nafion 115 membrane. Before electrochemical NRR measurements, the Nafion 115 membrane was first pretreated in 80 • C H 2 O 2 (5%) aqueous solution for 1 h and then washed in 80 • C ultrapure water for another 1 h. The electrochemical experiments were conducted on a CHI660E electrochemical analyzer (CHI Instruments) using a three-electrode configuration (working electrode of synthetic catalysts, reference electrode of saturated calomel electrode and counter electrode of carbon rod). The mentioned potentials have been converted to RHE. The cyclic voltammetry (CV) tests were carried out in 0.1 M HClO 4 solution with a scan rate of 50 mV s −1 under ambient conditions. The ECSA was obtained by integrating the hydrogen adsorption charge (Q H ) between −0.25 V and 0.1 V on the CV curves. The value of adsorbed single-layer hydrogen (q H ) on the Pt surface is 210 μC cm −2 and the formula is ECSA = Q H / (q H × m) ( Supplementary Fig. 38).
Potentiostatic tests were carried out in electrochemical NRR. Before the measurement, highly pure N 2 gas was flowed continuously into the cathode electrolytic cell with a proper position for 0.5 h.

Calculation of NH 3 yield, Faraday efficiency and apparent energy
In NRR tests, FE is defined as the amount of charge for the synthesis of NH 3 divided by the total charge through the electrodes during the electrolysis process. The NH 3 yield was determined by a colorimetric method using Nessler's reagent. The formation of each NH 3 molecule requires three electrons, hence the FE of NH 3 can be determined by the below formula: The NH 3 yield was determined by the below formula: where Q: quantity of electric charge; F: Faraday constant, 96485 C mol −1 ; V: the KOH electrolyte volume; C NH 3 : the calculated NH 3 concentration; t: the reduction time; A: the ECSA of the catalysts. According to the Arrhenius equation, the apparent energy calculations process is below: where K: NH 3 yield rate at temperature T; R: Molar gas constant (J mol −1 K −1 ); E a : Apparent energy (kJ mol −1 ), T: Absolute reaction temperature (K). We plot lnK against 1000/T, and the slope is k, so E a = -kR.

Determination of ammonia yield
The yield of NH 3 was determined via a colorimetric method using Nessler's reagent.

Determination of hydrazine hydrate
The hydrazine hydrate was determined by a colorimetric method using Watt-Chrisp reagent. Para (dimethylamino) benzaldehyde (5.99 g), concentrated HCl (30 mL) and absolute ethanol (300 mL) were mixed as color reagents. A calibration curve was obtained as follows: first, different reference solutions were prepared, by pipetting hydrazine hydratenitrogen 0.1 M HCl solution into colorimetric tubes. These were made up to 5 mL with diluted hydrochloric acid electrolyte (pH = 1), then 5 mL was pipetted above the color reagent and this was left to stand for 0.5 h for color development under ambient conditions. The absorbance of the color solution was measured at 455 nm with a 10 mm glass cuvette, and the yields of hydrazine were determined by the standard calibration curve using a mixture of 5 mL remaining solution and 5 mL color reagent. The calibration curve (y = 1.208x-0.088, R 2 = 0.977) was obtained with hydrazine monohydrate solutions of different known concentrations, and exhibited a good linear relationship between absorbance and N 2 H 4 ·H 2 O concentration according to three independent calibration tests.

Determination of NO x
The concentration of NO x was measured with a colorimetric method using N-(-1-naphthyl) ethylenediamine dihydrochloride as color reagent. A mixture of 0.5 g sulfanilic acid, 5 mg n-(1-naphthyl)-ethylenediamine dihydrochloride, 90 mL H 2 O and 5 mL acetic acid was stirred to form a homogeneous solution. This was transferred to a 100 mL volumetric flask to obtain the color agent. A mixture of 1 mL electrolyte and 4 mL color agent was left for 0.5 h in the dark. The absorbance of the solutions was measured at 540 nm from the UV-vis absorption spectrum. The calibration curve was obtained by using different known concentrations of potassium nitrite solution in 0.1 M KOH.

N isotope labeling experiment
An 15 N isotopic labeling experiment was carried out to verify the source of produced ammonia. After 15 N 2 electroreduction at −0.05 V in KOH electrolyte for 10 h, the obtained product was qualitatively determined by 1 H nuclear magnetic resonance (NMR, Agilent 600 MHz). In detail, 30 mL of the electrolytic solution was moved out and then acidized to pH ∼ 3. The solution was concentrated to 2 mL at 80 • C. Afterwards, 0.9 mL of concentrated solution and 0.1 mL D 2 O containing 100 ppm dimethyl sulphoxide (99.99%) as an internal standard were mixed for 1 H NMR test.

DFT calculations
The DFT+U calculations were carried out with CASTEP code [45]. In this framework, we use rotationally invariant (Anisimov type) functional DFT+U [46]. The Hubbard U parameter is selfconsistently determined for the pseudized C-2p, Fe-3d and Pt-5d orbital by the new linear response way [47][48][49][50][51][52][53]. The geometry optimization used the Broyden-Fletcher-Goldfarb-Shannon (BFGS) algorithm in all DFT+U calculations. The PBE functional was chosen for PBE+U calculations with a kinetic cutoff energy of 750 eV, with the valence electron states expressed in a plane-wave basis set. The ensemble DFT (EDFT) method of Marzari et al. [54] was used for convergence on the transition metal contained compounds.
The 2 × 2 × 1 supercell of Pt 3 Fe (311) surface model was chosen with 112 atoms (i.e. Pt 84 Fe 28 ) and seven layers thick. The 2 × 2 × 1 supercell of Pt 3 Fe (111) surface model was chosen with 128 atoms (i.e. Pt 96 Fe 32 ) and eight layers thick. The Pt 3 Fe (200) surface model was chosen with 144 atoms (i.e. Pt 108 Fe 36 ) and eight layers thick. The vacuum thickness was set to be 10Å. The reciprocal space integration was performed using the mesh of 2 × 2 × 1 [55] with Gamma-center-off, which was self-consistently selected for total energy minimization. With these special k-points, the total energy converged to less than 5.0 × 10 −7 eV per atom. The Hellmann-Feynman forces on the atom converged to less than 0.001 eV/Å.
The Fe and Pt norm-conserving pseudopotentials were generated using the OPIUM code in the Kleinman-Bylander projector form [56], the nonlinear partial core correction [57] and a scalar relativistic averaging scheme [58] are used to treat the spin-orbital coupling effect. For this treatment, we actually similarly chose a non-linear core correction technique for correcting the valence-core charge density overlapping in such heavy fermions elements, the details of this method are given in a previous work [47,48]. In particular, we treated the (3d, 4s, 4p) and (5d, 6s, 6p) states as the valence states of both Fe and Pt atoms. The RRKJ method was chosen for optimization of the pseudopotentials [59].
For all of the electronic state calculations, we used self-consistent determination for the U correction on the localized d orbitals to correct the onsite Coulomb energy of the electron spurious selfenergy. By that method, the Hubbard U parameters on the half-filled shell of 3d 6 orbitals of Fe are selfconsistently determined to be U d = 6.03 eV, and U d = 5.26 eV for Pt-5d 9 within the Pt 3 Fe system.