Crystallographic Orientation Dependence of Surface Segregation and Alloying on PdCu Catalysts for CO2 Hydrogenation

The influence of the crystallographic orientation on surface segregation and alloy formation in model PdCu methanol synthesis catalysts was investigated in situ using near-ambient pressure X-ray photoelectron spectroscopy under CO2 hydrogenation conditions. Combined with scanning tunneling microscopy and density functional theory calculations, the study showed that submonolayers of Pd undergo spontaneous alloy formation on Cu(110) and Cu(100) surfaces in vacuum, whereas they do not form an alloy on Cu(111). Upon heating in H2, inward diffusion of Pd into the Cu lattice is favored, facilitating alloying on all Cu surfaces. Under CO2 hydrogenation reaction conditions, the alloying trend becomes stronger, promoted by the reaction intermediate HCOO*, especially on Pd/Cu(110). This work demonstrates that surface alloying may be a key factor in the enhancement of the catalytic activity of PdCu catalysts as compared to their monometallic counterparts. Furthermore, it sheds light on the hydrogen activation mechanism during catalytic hydrogenation on copper-based catalysts.


Experimental details
Pd nanoislands were grown by electron beam evaporation on clean Cu single crystals of different crystallographic orientations (Pi-Kem Ltd., purity: 99.999) inside the preparation chamber of a NAP-XPS ultrahigh vacuum (UHV) system. The preparation chamber is equipped with an ion source for sputter cleaning, an electron beam heater for sample annealing and a single pocket electron beam evaporator for the growth of metallic thin films on substrates. The electron beam evaporator was loaded with a Pd rod (Goodfellow GmbH, Ø 2 mm, purity: 99.95 %). The ion source was connected through a leak-valve to an Ar lecture bottle (Linde GmbH, purity 5.0). Cu(100), Cu(110), and Cu(111) single crystals were loaded on flat stainless-steel sample holders.
The crystals were treated following a protocol that involved three consecutive cycles of Ar sputtering (P = 3.0 · 10 −5 mbar, E = 2.5 keV, duration: 45 min per cycle) and subsequent annealing in vacuum at 520 °C (45 min during the first and second cycles; 90 min during the third cycle).
The absence of contamination on the Cu single crystals after the sputtering and annealing treatment was systematically verified by XPS. Survey, C 1s and Pd 3d spectra of a Cu(110) single crystal after sputtering and annealing are exemplarily shown in Figure S1.
After the surface treatment, Pd was deposited onto the Cu single crystals by electron beam evaporation. The deposition rate of Pd was controlled with a quartz microbalance connected to the preparation chamber and amounted to approximately 1 monolayer (ML)/220 s = 0.27 ML/min when the Pd flux current was kept constant at 74.0 nA. The chosen evaporation time for all samples was 52 s, resulting in a nominal Pd coverage of 0.2 MLs. Samples exposed to shorter (15 s) and longer (3 min) Pd evaporation, thus exhibiting lower and higher Pd coverage, respectively, were also investigated for comparison, in order to examine a possible effect of the Pd coverage on the alloying and segregation behavior.

S3
A second set of Pd/Cu samples was prepared following the same procedure as described above in a separate, identically equipped preparation chamber to be used for morphological characterization in an STM characterization setup (Aarhus 150, Specs GmbH). Identical Cu crystals, substrate cleaning and Pd evaporation protocols as for the XPS samples were employed.
The analysis chamber of the NAP-XPS setup is equipped with a differentially pumped PHOIBOS 150 Near Ambient Pressure Hemispherical Energy Analyzer and a monochromatized Al Kα X-ray source (1486.7 eV). XPS spectra of the Pd-covered Cu crystals were obtained in UHV at room temperature (after Pd evaporation), in pure H2 (0.6 mbar) and under CO2 hydrogenation conditions (volume ratio CO2:H2 : 37:63, total pressure: 0.6 mbar). The pressure was regulated by dosing the constituent gases through separate leak valves into the main analysis chamber. The sample temperature during the NAP-XPS measurements was kept constant at 270 °C with an infrared heater (halogen lamp) located on the sample manipulator. After each measurement, the sample was cooled down in the gas and the chamber was evacuated when the sample reached room temperature. For each one of the three different crystallographic orientations of Cu, the XPS study was repeated three times for the CO2 hydrogenation conditions as well as for the pure H2 environment, each time on a freshly prepared sample to ensure the reproducibility of the results.

Details of the XPS analysis and binding energy calibration
For all Pd spectra, the metallic Cu 2p3/2 peak was used as binding energy reference. While some of the Cu atoms are also expected to be present in the alloyed state, it was assumed that the main contribution to the Cu XPS signal is from the unalloyed component, especially considering the small Pd coverage that was discovered for these samples with STM (see Fig. 1 in the main text).
Therefore, all Cu 2p3/2 peaks were aligned at 932.6 eV and used as energy reference for all other peaks.

S4
In agreement with the available literature, Cu was fitted with a doublet for the 2p3/2 and 2p1/2 regions, with the single peak of the Cu 2p1/2 region at 932.6 eV. 1 The fits were constrained with respect to the area ratio (1:2) and the binding energy offset in virtue of the spin-orbit splitting (19.8 eV for metallic Cu in Cu 2p spectra). The two metallic Cu peaks were fitted with Gaussian-Lorentzian line shapes (ratio 0.9) and a Shirley background. For the Pd 3d spectra, a metallic doublet with an asymmetric line-shape was used in the fittings. 2 The area ratio (2:3) and the doublet energy splitting of 5.26 eV were constrained in the fitting. In order to account for initial alloying and final state effects, the binding energy of the Pd 3d5/2 was allowed to vary by ± 0.4 eV compared to the literature value. The asymmetry parameter, the full width at half maximum (FWHM) and the ratio of the Gaussian and Lorentzian portion of the line shapes were also allowed to vary. The fits were adjusted by minimizing the residual of the fits with respect to the measured spectra.
It should be noted that both the standard deviation error of the measurements (0.01 eV) and the nominal energy resolution of the spectrometer at the Al Kα line (about 0.008 eV) are at least one order of magnitude smaller than the reported binding energy shifts. Therefore, the conclusions reported in the main text regarding the alloying trends are beyond experimental uncertainties.

Effect of the Pd surface coverage
In the main text, the Pd 3d5/2 energy shift is used as an indicator for the alloying of Pd and Cu after segregation of the materials. In order to confirm this assumption, additional tests with different Pd coverages were performed.

Computational details
All ab initio calculations were performed with the revised Perdew-Burke-Ernzerhof (RPBE) The surface segregation energy, SE, is defined as the energy difference when moving the Pd monolayer from the bulk to the surface ( Figure S9) in vacuum or when an adsorbate is present (the adsorbate is assumed at the most stable adsorption site for each system). In this work, the surface segregation energy is calculated using equation (1), where Esurface and Enth-layer correspond to the total RPBE energy of the slab with the Pd monolayer on the top and the nth surface layer, respectively. The value of n is chosen so that the energy difference between Enth-layer and E(n-1)th-layer is less than 0.05 eV.
The calculated most stable configurations for Pd/Cu systems with different H coverages are displayed in Figure S10. The hydrogen coverage is calculated as the ratio between the total number of H adatoms and the number of atoms in one atomic layer. Using this definition, the range of H concentration considered is 11% -100%, with the lower limit corresponding to one H atom and the upper limit to 9 H (1 ML H) atoms in the 3×3 cell.