Electronic Supporting Information – Structural dynamics in Ni-Fe catalysts during CO2 methanation – role of iron oxide clusters

Bimetallic Ni – Fe catalysts show great potential for CO 2 methanation concerning activity, selectivity and long-term stability even under transient reaction conditions as required for Power-to-X applications. Various contrary suggestions on the role of iron in this system on CO 2 activation have been proposed, hence, its actual task remained still unclear. In this study, we used X-ray absorption spectroscopy (XAS) combined with X-ray diffraction (XRD), XAS in combination with modulation excitation spectroscopy (MES) and density functional theory (DFT) to shed detailed light on the role of iron in a bimetallic Ni – Fe based CO 2 methanation catalyst. During catalyst activation we observed a synergistic effect between nickel and iron that led to higher fractions of reduced nickel compared to a monometallic Ni-based catalyst. By XAS – XRD combined with DFT, we found formation of FeO x clusters on top of the metal particles. Modulation excitation coupled XAS data complemented with DFT calculations provided evidence of a Fe 0 ⇌ Fe 2+ ⇌ Fe 3+ redox mechanism at the interface of these FeO x clusters. This may promote CO 2 dissociation. This is the first time that this highly dynamic role of iron has been experimentally confirmed in bimetallic Ni – Fe based catalysts with respect to CO 2 activation during the methanation reaction and may also be at the origin of better performance of other CO 2 -hydrogenation catalysts. The insight into the structural surface changes reported in this study show the dynamics of the Fe – Ni system and allow the development of realistic surface models as basis for CO 2 activation and possible intermediates in these bimetallic systems.

1 Table of Content   1 Table of

Rietveld refinement of synchrotron-based powder XRD
FullProf software package was used for refinement of PXRD data. [1] An instrumental resolution file was first obtained by profile fitting of a LaB6 NIST 640b standard to correct for peak broadening based on the instrument. The nano-crystalline nature of the used γ-Al2O3 makes it difficult to refine this phase well, which can be especially seen in deviations of the measured and calculated patterns at 17.5 ° or 20.5 ° in all refinements shown in Figure Table S 2 -Table S 7 while the error for the γ-Al2O3 phase is quite high. Additionally, the values do not match the expected ones of 17 wt.% for Ni3.0Fe and 83 wt.% for γ-Al2O3 or as determined by elemental analysis. Reasons for this are that we are only considering the metal to be Ni and not the contribution of Fe, the γ-Al2O3 phase is not described well potentially overestimating the amount of γ-Al2O3 and only crystalline parts of the metal nanoparticles account to the phase fraction. Therefore, the obtained fractions should be rather considered qualitatively and not quantitatively.

Calculation of alloy composition using Vegard´s law
The calculation of the alloy composition was performed using Vegard´s law [2] , i.e. the linear correlation of the lattice parameter a to the respective fraction of Ni (fcc, a=3.520 Å) and Ni3Fe (fcc, a=3.553 Å) [3][4][5][6][7][8][9] . The lattice parameter of the alloy was determined based on the shifted Ni reflections obtained in the refined XRD results (cf. sections 2.2 and 3.2.2).

Catalyst activation and Ni3Fe alloy formation
Figure S 2: Continuously recorded XRD data of (a) Ni/γ-Al2O3 and (b) Ni-Fe/γ-Al2O3 during H2-TPR at 6 bar. (c) Fe K-edge XANES spectra of the Ni-Fe/γ-Al2O3 catalyst as prepared in comparison to references and (d)k 2 -weighted FT-transformed EXAFS spectra obtained at the Fe K-edge of the Ni-Fe/γ-Al2O3 catalyst before and after H2-TPR in comparison to a Ni and Fe metal foil.

Active state of the Ni-Fe catalyst under steady state conditions
To verify the formation of FeOx, the Fe K-edge XANES spectra were compared to their state during H2- formation of an Fe/FeO rich surface was found (see next section and paper).

EXAFS fitting details
At the Ni K-edge EXAFS spectra were fitted in the range of R = 1.0-3.2 Å and k = 2.0-11.0 Å -1 for the as prepared catalyst and in range of R = 1.0-5.0 Å and k = 2.7-12.5 Å -1 for catalyst after TPR . The amplitude reduction factor (S0 2 ), as determined from Ni foil, was fixed at 0.84 and one energy shift parameter (E0) was defined for all scattering paths. Scattering paths Ni-O, and Ni-Ni obtained from reference models were used and parameters N, ΔR and σ 2 were fitted.
At the Fe K-edge EXAFS spectra were fitted in the range of R = 1.0-5.0 Å and k = 2.7-10.5 Å -1 . The amplitude reduction factor (S0 2 ), as determined from Fe foil, was fixed at 0.82 and one energy shift parameter (E0) was defined for all scattering paths. Scattering paths Fe-Fe (Fe bulk) and Fe-Fe (Ni-Fe alloy) obtained from reference models were used and parameters N, ΔR and σ 2 were fitted.

Simulation of XANES spectra using FEFF9 code
For XANES simulations at Ni K-and Fe K-edges in FEFF9, the ab initio self-consistent real-space Green's function (RSGF) approach was used including inelastic losses, core-hole effects, vibrational amplitudes, etc. The polarization dependence, core-hole effects, and local field corrections were based on self-consistent, spherical muffin-tin scattering potentials. In the present ab-initio calculations, the Hedin-Lundqvist potential was chosen and XANES, Absolute, SCF (self-consistent field), and FMS    Table S1). Structural parameters a of the 17 wt.% Ni-Fe/γ-Al2O3 catalyst determined of EXAFS spectra t the Fe K-edge spectra, S0 2 =0.82, data fit is given in Figure 2 and Figure Figure

Development of a structural model
To develop the structural model, it is important to understand how the surface composition changes. The number and composition of nickel and iron atoms in the particle and on the surface was estimated (Table   S 8). The error margin in the obtained numbers of atoms is quite high due to the error of ± 0.9 in the particle size of 3.9. However, the numbers give a good trend how surface and particle composition change.
Based on the estimation, the amount of iron at the surface strongly increases with temperature. The iron on the surface is nearly fully oxidized as soon as CO2 methanation conditions were applied.

Steady state of Ni/Al2O3 and Ni-Fe/Al2O3 before and after the MES experiment
The overall catalyst performance (Table S 9) reflected the trends observed in previous studies with comparison between Ni and Ni-Fe catalysts. [11][12][13]

Ni K-edge XANES spectra of Ni/Al2O3 and Ni-Fe/Al2O3 before the MES experiment
The normalized XANES spectra at the Ni K-edge after reaching the steady state of the catalyst during methanation of CO2 at various temperatures are shown in Figure S 16 for Ni and Ni-Fe, respectively.

Fe K-edge XANES spectra of Ni/Al2O3 and Ni-Fe/Al2O3 before the MES experiment
The changes occurring during MES can depend on the steady state of the catalyst before MES. Figure   S 17a shows the normalized XANES spectra at Fe K-edge for the steady state of the catalyst before MES experiment at different temperatures.  Hence, at higher temperatures Fe is more oxidized in its steady state before MES.

Time-resolved XANES spectra of the monometallic Ni/Al2O3 catalyst
Time-resolved XANES spectra obtained from 30 periods averaged into one period (a total of 24 spectra) and the corresponding phase-resolved spectra obtained during the experiment performed at 450 °C are given in Figure S 18.

Phase-resolved Ni K-edge MES spectra of Ni/Al2O3 and Ni-Fe/Al2O3 catalyst
First, the monometallic Ni/Al2O3 catalyst was investigated. No distinct changes were visible in the characteristic features of the conventional time-resolved Ni K-edge XANES spectra (Figure S 18b).  One representative curve of the changes after the demodulation of the Ni/Al2O3 catalyst at 450 °C and a phase angle of 300° recorded at the Ni K-edge is displayed in Figure S 21.

S23
In order to interpret the features of the demodulated spectrum, XANES spectra for DFT calculated structures of Ni surface with different species, i.e., Ni-C, Ni-CO, Ni-HCOO and Ni-HOCOO have been simulated using FEFF9 as shown in Figure S 21a along with simulated Ni metal spectrum. Their corresponding difference spectra with the simulated Ni metal spectrum were calculated and compared to our experimental demodulated spectrum (Figure S 18b, black curve). Four features were observed in the experimental spectrum, and denoted as P, S, A and B, respectively. Feature P at ~ 8334 eV (preedge region) correlated well to the difference spectra of Ni-NiC and Ni-NiCO, although shifted to slightly higher energy. The shoulder S at around 8345 eV was found in all references. However, the shoulder feature was not intense in the experimental data indicating the contribution of Ni-HCOO/Ni-HOCOO species, as the shoulder was less intense for these species compared to Ni-C / Ni-CO. The formation of Ni-HCOO/Ni-HOCOO species would be in a good agreement to DRIFTS studies [14][15][16] on Ni based catalysts. After the white line region, feature A at 8355 eV in the post edge region further substantiated the formation of Ni-C / Ni-CO species, as already suggested by feature P. However, feature B at 8368 eV correlated rather more to Ni-C than Ni-CO. Hence, the dissociative pathway of CO2 activation seems to be preferred on the monometallic Ni/γ-Al2O3 catalyst at 450 °C. As feature S did decrease, although more Ni-C species were formed at higher temperatures, Ni-HCOO / Ni-HOCOO species must be present and formed with an increasing ratio. Hence, we can conclude that in the investigated temperature regime all mentioned species were present during the methanation of 22a), where we determined a lower catalytic activity (see Table S

Defining U value for GGA+U method
We applied the GGA+U method in order to better describe delocalized iron d orbitals and obtain more accurate energies. Based on the value of the FeO heat of formation compared to metallic Fe [17] , we have applied different U values to find the best fit for the experimental data and found a U of 2.7 eV to be optimal.

Oxygen adsorption on Ni4Fe(111) alloy
To address the segregation of Fe in the Ni3Fe alloy during the CO2 hydrogenation process we constructed a Ni4Fe(111) slab with the dimension 5x2 in x and y direction. We also take this slab to study FeO ontop of the Ni-Fe alloy. The orthogonal orientation of this unit cell provides close to perfect lattice match in the y direction (4.366 Å ) with the FeO lattice parameter (4.363 Å).

Carbon monoxide adsorption on Ni3Fe(111) alloy and Ni top layer /Ni3Fe(111)
We tested the effect of Fe on the CO adsorption energy using a stoichiometric Ni-Fe alloy (Ni3Fe(111)) as well as Fe in the sublayers (Ni top layer /Ni3Fe(111)). Our calculations indicate that this does not affect the CO or H binding strength (see Table A). However, the presence of oxygen at the surface will reduce the CO binding energy through repulsive adsorbate-adsorbate interactions (see Table A)." The reason behind the increase of the size of the nanowire is to look not only at the "flat" single layer nanowires (FeO)2-4 but few layers thick (FeO)6 as well, so that different parts of the particle could be analyzed. Charges are given in Table S 11.

Phase diagram
For the construction of the final phase diagram, we have used all structures given in Section 3.5.2 and 3.5.3 of SI and we have calculated oxygen binding energies in following way: Number of (Febulk ) is y-2 because Ni4Fe(111)* and Ni4Fe(111) differ in two sublayer Fe atoms.

Oxygen hydrogenation
To compare the activity of the additional oxygen at the interface O+(FeO)6/Ni4Fe(111)* we calculate the hydrogenation of this oxygen to form H2O. The results are given in Table S 13: Energy of the first and second hydrogenation steps of interface oxygen and stoichiometric oxygen.  Table S 14 the structures are shown in Figure S 28.