Impact of Catalysis-Relevant Oxidation and Annealing Treatments on Nanostructured GaRh Alloys

In this study, we examine the surface-derived electronic and chemical structures of nanostructured GaRh alloys as a model system for supported catalytically active liquid metal solutions (SCALMS), a novel catalyst candidate for dehydrogenation reactions that are important for the petrochemical and hydrogen energy industry. It is reported that under ambient conditions, SCALMS tends to form a gallium oxide shell, which can be removed by an activation treatment at elevated temperatures and hydrogen flow to enhance the catalytic reactivity. We prepared a 7 at. % Rh containing the GaRh sample and interrogated the evolution of the surface chemical and electronic structure by photoelectron spectroscopy (complemented by scanning electron microscopy) upon performing surface oxidation and (activation treatment mimicking) annealing treatments in ultrahigh vacuum conditions. The initially pronounced Rh 4d and Fermi level-derived states in the valence band spectra disappear upon oxidation (due to formation of a GaOx shell) but reemerge upon annealing, especially for temperatures of 600 °C and above, i.e., when the GaOx shell is efficiently being removed and the Ga matrix is expected to be liquid. At the same temperature, new spectroscopic features at both the high and low binding energy sides of the Rh 3d5/2 spectra are observed, which we attribute to new GaRh species with depleted and enriched Rh contents, respectively. A liquefied and GaOx-free surface is also expected for GaRh SCALMS at reaction conditions, and thus the revealed high-temperature properties of the GaRh alloy provide insights about respective catalysts at work.


XPS ANALYSIS
The total energy resolution of the XPS and UPS setup is determined via a fitting of the measured Fermi-edge (EF) of a clean gold film by the following fit function with correction of temperature (T = 300 K) employing the Boltzmann constant (kb•T = 25 meV): σtotal is the total Gaussian broadening including instrumental and thermal (kb•T) broadening, Ef denotes the energy of Fermi-edge.a, b, c, d are dependent variables in the fit function.

Quantification of XPS data
All XPS data were fitted and quantified by Winspec (LISE, Université de Paix, Namur).The XPS peaks of metallic components (GaRh IMCs, isolated Rh atoms, metallic Ga matrix) are fitted by an asymmetric (Doniach-Sunjic) profile; the XPS peaks ascribed to GaOx are fitted by Voigt profiles.The fitting includes a Shirley background.In the Ga 2p3/2 and 3d fitting, a GaOx peak with different peak width is used (Fig. S5, S6, S9 and S14) due to presence of oxygen vacancies 1 and several GaOx species causing a peak broadening.The fits shown in Fig. S9 and S14 are conducted with the following constraints to obtain a reasonable fitting result for quantitative analysis.The peak width and position and intensity ratio of spin-orbit split doublet peaks are coupled in all fitting routines and the spectrum is fitted sequentially with fixed peak shape and width.The derived area of the studied core level peaks is corrected by the transmission function of the used analyzer: = * (0.61 + 0.00021 * ( − ) I and I0 denote the calibrated peak area and original peak area, respectively.Ex denotes the excitation energy (Mg Kα = 1253.56eV), BE denotes to the binding energy of core level peak.

S4
Additionally, the derived peak areas are corrected for by IMFP and photoionization cross sections.3][4] Photoionization cross sections () are calculated from the tabulated value from Trzhaskovskaya, Nefedov, and Yarzhemski, 5- 6 considering the geometry of the XPS setup.The quantitative analyses of certain core level peaks or elements in this study are then processed by following equation: IA and IB denote the peak area of core level A and B. A and B are the respective photoionization cross sections of core level A and B. IMFPA and IMFPB denote to inelastic mean free path of photoelectrons of core level A and B, respectively.

Determination of Rh concentration and GaOx/Ga ratio
The fit results of Ga 3d, Ga 2p3/2, and Rh 3d5/2 core level peaks shown in Fig. S5, S6, S7, S9, S10 have been used to determine the Rh concentration of the GaRh sample before and after all oxidation steps.Note that for this consideration always the total peak area is used.4]8 The Rh concentration is then calculated by following equation: The results are shown in Table S3-S5.Note that the Rh content derived from the Ga 3d and Rh 3d5/2 spectra is less error prone (in case the sample has a significant chemical structure profile) as the corresponding photoelectrons have very similar kinetic energies and thus comparable IMFPs.The Ga 2p3/2 photoelectrons have a significantly lower kinetic energy and thus lower IMFP.Hence, the Rh content derived from the Ga 2p3/2 and Rh 3d5/2 lines -due to their S5 different information depths and different background profile-has a larger experimental uncertainty.The fitting shown in Fig. S5, S6, S9 are utilized to derive the intensity ratio of Ga and GaOx which is used for GaOx film thickness calculation elaborated in section 1.3.

Determination of GaOx film thickness
In this study, a simple overlayer model is utilized to get insights on the GaOx layer formation on top of metallic Ga, assuming a mechanism of homogeneous, closed packed oxide film growth. 9The following equation can be used to calculate the GaOx film thickness, D: λi, GaOx and λi,Ga are the IMFP values in GaOx and metallic Ga, respectively for core level i (calculated using the TPP2-M formula [2][3][4] with the density and electron configuration of stoichiometric Ga2O3 as the absorbing layer -we considered this the best approximation available due to the lack of reliable parameters for GaOx ).3][4] Ii,GaOx and Ii,Ga are the intensities (i.e., areas) of the GaOx and Ga peak contributions, respectively, derived for the core level i (obtained from XPS data fitting, see Fig. S5, S6, S9).N(Ga)GaOx and N(Ga)Ga are the atomic densities of Ga in Ga2O3 (0.038 Atoms per cubic Å) and Ga (0.053 Atoms per cubic Å), respectively. 10It is noted that the formula for D assumes a uniform bi-layer system with a closed capping oxide layer, and thus the discrepancies in the oxide thicknesses calculated using Ga 2p3/2 (Table S1) and using Ga 3d (Table S2) are presumably related to any deviation from this assumption (e.g., a different sample topography and/or an incomplete coverage of the Ga by the oxide capping layer), which will cause an underestimation of the derived layer thickness.As the oxide coverage of Ga increases, the results of the two calculations converge.This approach allows the GaOx contribution to the spectra to be monitored, which is sufficient for the analysis contained here, but leads to imperfect fitting of the measured spectra.This is particularly apparent around 1117.7 eV in the 60-and 240-min oxidized sample, where an indication for a distinct spectral feature can be observed, which may be attributed to intrinsic (i.e., not n-type) Ga2O3.This contribution is fitted by a peak with broader FWHM which is tentatively attributed to a less ordered material (compared to the metallic GaRh alloy) resulting in varying bond lengths and bond angles -all of which causing BE variations that may increase the FWHM.S1.GaOx/metallic Ga ratio of the respective spectral contributions to the Ga 2p3/2 core level and calculated GaOx film thickness for the nanostructured 7 at% Rh containing GaRh alloy samples oxidized in 1×10 -6 mbar O2 for different times.

S12
Table S2.GaOx/metallic Ga ratio of the respective spectral contributions to the Ga 3d core level and calculated GaOx film thickness for the nanostructured 7 at% Rh containing GaRh alloy samples oxidized in 1×10 -6 mbar O2 for different times.

Figure S4 .S9Figure S5 .S10Figure S6 .
Figure S4.Analysis of the nanoparticle size distribution of the (initially oxidized) GaRh sample containing 7% Rh on SiOx/Si support.The statistical analysis is based on 200 particles depicted from the SEM image shown in Fig. S3a.

Figure S7 .
Figure S7.Fits of the Rh 3d5/2 XPS data collected with Mg Kαexcitation for a 7 at% Rh containing nanostructured GaRh sample oxidized in 1×10 -6 mbar O2 for (a) 0 min, (b) 10 min, (c) 30 min, (d) 60 min, and (e) 240 min.The Ga-Rh peak is fitted by an asymmetric (Doniach-Sunjic) profile and O-Rh peak is fitted by Voigt profile.The minor Rh feature (green line) is ascribed to Rh atoms located in/close to the GaOx layer (presumably forming Rh-O bonds).

Figure S8 . 1 S13Figure S9 .S14Figure S10 .S15Figure S11 .
Figure S8.He II-UPS spectra of a 7 at% Rh containing nanostructured GaRh alloy compared to that of a pure (Rh-free) Ga reference sample (from Ref. 1) oxidized in 1×10 -6 mbar O2 for different times (10-240 min).The Ga spectrum is scaled to have a similar intensity of Fermi edge.The position of the Rh 4d derived spectral feature is indicated by the vertical dashed line.The spectral intensity around 3 eV in the Ga spectra is attributed to oxygen vacancies.1

Figure S13 .S18Figure S14 .S19Figure S15 .
Figure S13.Mg Kα-XPS (a) survey spectra and (b) Si 2p spectra of the Si support before and after the removal of the native SiOx by Ar + -ion sputtering.