Formation of a 2D Meta-stable Oxide by Differential Oxidation of AgCu Alloys

Metal alloy catalysts can develop complex surface structures when exposed to reactive atmospheres. The structures of the resulting surfaces have intricate relationships with a myriad of factors, such as the affinity of the individual alloying elements to the components of the gas atmosphere and the bond strengths of the multitude of low-energy surface compounds that can be formed. Identifying the atomic structure of such surfaces is a prerequisite for establishing structure–property relationships, as well as for modeling such catalysts in ab initio calculations. Here, we show that an alloy, consisting of an oxophilic metal (Cu) diluted into a noble metal (Ag), forms a meta-stable two-dimensional oxide monolayer, when the alloy is subjected to oxidative reaction conditions. The presence of this oxide is correlated with selectivity in the corresponding test reaction of ethylene epoxidation. In the present study, using a combination of in situ, ex situ, and theoretical methods (NAP-XPS, XPEEM, LEED, and DFT), we determine the structure to be a two-dimensional analogue of Cu2O, resembling a single lattice plane of Cu2O. The overlayer holds a pseudo-epitaxial relationship with the underlying noble metal. Spectroscopic evidence shows that the oxide’s electronic structure is qualitatively distinct from its three-dimensional counterpart, and because of weak electronic coupling with the underlying noble metal, it exhibits metallic properties. These findings provide precise details of this peculiar structure and valuable insights into how alloying can enhance catalytic properties.

S-2   For experiments where some data might be possible, but are not listed here, it was found that the quality of the data was not sufficient for the data to be useful, due to instrument artefacts, noise or inaccuracy. For instance, NEXAFS was measured in XPEEM, but the signal quality was insufficient. Or NEXAFS can be calculated with DFT but the accuracy of the results was not sufficient for comparison with experiment.

S-4
By networking commonly observed data sets among different experiments, we draw a link between atomic structure and function.
In B) we see that valence band connects all experiment types. In C) we have a connection between experiment types 1: in-situ epoxidation and 4: NAP-XPS of crystal. Nodes that have no connection are greyed out. Thus here we see that this combination of data only gives us correlation between spectroscopic fingerprints and functional performance.
In D. we have links between experiment types 2: XPEEM and 3: DFT. This link provides correlation between experimental data and atomic structure. In E we have a connection between experiment types 1: in-situ epoxidation and 4: NAP-XPS of crystal. This link provides a connection between a spectroscopic signature and a correlation with functional performance.

S-5
In F. we have links between experiment types 1: in-situ epoxidation, 2: XPEEM, 3: DFT and 4: NAP-XPS of crystal. These links provide confidence that what was observed in each experiment is the same as the experiment it is linked to.
Panel G. shows a complete connectivity graph, where the colors of the lines represent the type of data or attribute that gives rise to the link. Panel H. shows a complete connectivity graph, where the thickness of the line represents the number of links that exist.
S-6 Figure S4 schematically illustrates the XPS and NEXAFS process in the case of Cu 2 O and the p2 structure. The excitation of a core level electron (i.e. from the Cu 2p level) lead to a formation of a core-hole in both cases. While in XPS the excited electron is removed from the system, in NEXAFS the electron remains in the system, bound as an exciton to the core-hole. Generally, in metallic systems, core-hole screening effects are similar for XPS and NEXAFS processes, since there is a vast availability of valence electrons in the Fermi sea of metallic systems. Consequently, XPS binding energies (Cu 2p) and NEXAFS absorption edge positions (Cu L-edge) are the same in metallic systems. They generally differ in cases where there is a band gap (e.g. Cu 2 O in Figure 13a). 12 In such cases, the NEXAFS process requires the core electron not only to be excited to the Fermi energy (E f ), but beyond by an amount equal to conduction band minimum (CBM) -E f . Furthermore, the core-hole screening in semiconductors is generally more effective in NEXAFS than in XPS because the final state in the NEXAFS process contains one more electron than the final state in the XPS process. The observed difference between the Cu L 3 -edge of the Cu x O y structure and Cu 2 O is not trivial to explain. Both materials have very similar initialstate energies (as calculated by DFT), and both have essentially identical Cu 2p binding energies. DFT also shows that both have very similar charge densities. However, NEXAFS shows that Cu x O y is substantially different from Cu 2 O. The observed difference can be explained if one assumes the Cu x O y structure has no band gap (i.e has a finite density of states at the Fermi level). If this is the case, then the band gap term plays no role in the NEXAFS process, and the electron only needs to be excited from the Cu 2p state to the Fermi level for X-ray absorption to occur at the Cu L-edge.
S4. Schematic illustration of the XPS and NEXAFS process in the case of (a) a semiconductor exhibiting a band gap (Cu 2 O) and (b) a system without bandgap (well coupled p2 structure).  To investigate the relative stability of p2 structures in different unit cells, we compute the formation energy per unit area defined as: where is the total energy of the adsorption system, is the total energy of an isolated ( 2/ ) ( 2) layer of the p2 oxide layer with its optimized lattice parameter, is the total energy of the Ag slab ( ) and A is the surface area. We consider three unit cells, where the p2 overlayer is accommodated in a 12×2, 13×2 or 14×2 Ag(111) surface unit cell. In the 12×2 unit cell, the p2 overlayer, with a stoichiometry Cu 18 O 12 , has a 2×2 periodicity. In the 13×2 unit cell, the same p2 overlayer is stretched along one S-8 direction (by 1/12 ~ 8%) and in the 14×2 unit cell it is stretched even further. Given the large size of these unit cells, HSE06 calculations would be extremely expensive; the total energies reported above were therefore computed using the PBE exchange and correlation functional. Table S.5 shows that the 13×2 unit cell provides the lowest formation energy, in agreement with the experimental findings showing that the p2 structure is stretched by ~8% with respect to a commensurate structure with 2×2 periodicity with respect to the Ag substrate. We also tested the effect of including van der Waals (vdW) interactions, using the exchange dipole model (XDM) of Becke and Johnson 5 . We find that, while significant in magnitude (-21.1, -22.6 and -17.6 meV/Å 2 for the 12x2, 13x2 and 14x2 unit cells, respectively), dispersion interactions do not show a strong variation with the registry of the oxide with the substrate. The 13x2 unit cell provides the lowest formation energy even in this case. S-9 Figure S8 (a) LEEM image (E kin = 6 eV) of AgCu(111) surface after cooling the sample from 300 °C to liquid nitrogen (LN 2 ) temperature in 1 × 10 -5 mbar O 2 . There is no microscopic evidence that the surface oxide has morphologically changed upon cooling. We cooled down the sample in an attempt to stabilize (minimization of thermal drift) and enhance the contrast of the collected LEED pattern in order to perform a full LEED-IV analysis for the oxide structure determination. The survey spectrum at LN 2 temperature of the region shown in the LEEM image (b), shows that the most common contaminants carbon and sulphur, nor any other contaminants, are present on the surface. The absence of sulphur containing species can be seen from the missing S2p states in the survey with binding energies between 165 -167.5 eV 1 , as well as the absence of intensity of S-related states in the valence band (c), which would occur at binding energies between 6.5 -8.5 eV. 1 However, the valence band spectra stack collected at different photon energies in the area shown in the LEEM image (c), indicates the formation of an additional species on the surface upon cooling, with a binding energy of 1.7 eV. As the photon energy is swiped across the Cooper minimum (around 170 eV), one can see that those states scale with the Ag 4d states. We therefore interpret this additional species as adsorbed oxygen on Ag. A strong hybridization of the Ag 4d and O 2s states leads to the observed scaling of the two bands. (d) shows the direct comparison of the VB measured with a photon energy of hν = 170 eV at 350°C (red curve) and at liquid nitrogen LN 2 temperature (both at 5.5 × 10 -5 mbar O 2 ). By comparison of the O 1s spectra measured under the two temperatures, shown in (e), it becomes evident that an additional shoulder on the high binding energy side in the O1s spectrum appears. It is conceivable that, during cooling, additional oxygen atoms adsorb onto free fcc sites in-between the p2 structure, as depicted with the filled red circles in the inset of (f).
The calculated projected density of states of this structure show that, indeed, this oxygen species gives reason to expect a higher XPS intensity in the valence band at a binding energy of about 1.5 eV. However, the calculations also reveal that this structure is not particularly stable. The bond between the S-10 extra O atom and the surface is approximately 0.2 eV less stable as compared to an O-O bond in an O 2 molecule. If the position of the adsorbed oxygen atom is moved closer to a Cu atom of the p2 oxide, the stability increases to a maximum of -0.1 eV, relative to an O 2 molecule. In summary, the identity of this additional mystery oxygen species at low temperature remains unknown for now, as a higher spatial resolution is required to identify the exact adsorption sites. A further complication that arises with the adsorption of this species is that fact that it contributes in an unknown way to the LEED spot intensities, that are of critical interest in a LEED-IV analysis. Hence, we were not able to use LEED-IV to fully describe the observed oxide surface structure of Cu x O y.  S-14  Figure S14 -Ambiguity of non-spatially resolved O1s spectra Figure S14 shows an O1s spectrum measured from AgCu (0.5 % Cu) in a 1:1 mixture of O 2 and C 2 H 4 at 0.5 mbar and 350°C. The peaks binding energies used to fit this spectrum are taken from literature values of all common Cu-O and Ag-O surface species (as well as SiO 2 , a common contaminant). One can see that there are at least 10 possible O1s species within a binding energy range of 3.3 eV, with each peak having a typical full-width-half-max of 1.1 eV. With so many free parameters, the ability to analyze the O1s spectrum from non-spatially resolved measurements is very unreliable, and conclusions are associated with a very high uncertainty.