Interface atom mobility and charge transfer effects on CuO and Cu2O formation on Cu3Pd(111) and Cu3Pt(111)

We bombarded \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\mbox{$\text{Cu}_{3}\text{Pd}(111)$}$$\end{document}Cu3Pd(111) and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\mbox{$\text{Cu}_{3}\text{Pt}(111)$}$$\end{document}Cu3Pt(111) with a 2.3 eV hyperthermal oxygen molecular beam (HOMB) source, and characterized the corresponding (oxide) surfaces with synchrotron-radiation X-ray photoemission spectroscopy (SR-XPS). At \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$300\,\text{K}$$\end{document}300K, CuO forms on both \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\mbox{$\text{Cu}_{3}\text{Pd}(111)$}$$\end{document}Cu3Pd(111) and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\mbox{$\text{Cu}_{3}\text{Pt}(111)$}$$\end{document}Cu3Pt(111). When we increase the surface temperature to \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$500\,\text{K}$$\end{document}500K, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\mbox{$\text{Cu}_{2}\text{O}$}$$\end{document}Cu2O also forms on \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\mbox{$\text{Cu}_{3}\text{Pd}(111)$}$$\end{document}Cu3Pd(111), but not on \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\mbox{$\text{Cu}_{3}\text{Pt}(111)$}$$\end{document}Cu3Pt(111). For comparison, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\mbox{$\text{Cu}_{2}\text{O}$}$$\end{document}Cu2O forms even at \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$300\,\text{K}$$\end{document}300K on Cu(111). On \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\mbox{$\text{Cu}_{3}\text{Au}(111)$}$$\end{document}Cu3Au(111), \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\mbox{$\text{Cu}_{2}\text{O}$}$$\end{document}Cu2O forms only after \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$500\,\text{K}$$\end{document}500K, and no oxides can be found at \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$300\,\text{K}$$\end{document}300K. We ascribe this difference in Cu oxide formation to the mobility of the interfacial species (Cu/Pd/Pt) and charge transfer between the surface Cu oxides and subsurface species (Cu/Pd/Pt).

LEED patterns. The clean Cu 3 Pd(111) exhibits a ( 2 × 2 ) LEED pattern with some spot splittings (cf.,   Cu-L 3 M 4,5 M 4,5 and Cu-2p spectra. In Fig. 2, we show the Cu-L 3 M 4,5 M 4,5 AES (Auger electron spectroscopy) and Cu-2p XPS spectra of the corresponding Cu oxides formed on Cu 3 Pd(111) and Cu 3 Pt(111) . For reference, note the Cu 2 O features appearing at ca. 917 eV kinetic energy (cf., bulk Cu 2 O peak, Fig. 2a). From Fig. 2a, we find prominent Cu 2 O formation only for Cu 3 Pd(111) oxidized at T S = 500 K . On the other hand, in Fig. 2b, we find characteristics of CuO (cf., shoulder at ca. 936 eV and satellite peaks between ca. 939 eV and 946 eV 29 ) for both Cu 3 Pd(111) and Cu 3 Pt(111) oxidized at T S = 300 K . We also see that these CuO features persist at T S = 500 K for Cu 3 Pt(111) , but disappear for Cu 3 Pd(111) . On Cu 3 Pt(111) , AES spectra analyses indicate CuO formation only, both at T S = 300 K and 500 K . On Cu 3 Pd(111) , AES spectra analyses indicate CuO formation at T S = 300 K and Cu 2 O formation at T S = 500 K . (Pd-3d and Pt-4f XPS spectra analyses also indicate that only   Table 2, we show a summary of oxides formed on the Cu and Cu-based alloy surfaces studied. In . This suggests the reduction of CuO into Cu 2 O at high T S . However, a lingering CuO peak remains on Cu 3 Pt(111) , even after the annealing at T S = 600 K , which we no longer see on Cu 3 Pd(111) . We found that annealing at T S = 650 K completes the reduction from CuO to Cu 2 O on the Cu 3 Pt(111) . Therefore, we conclude more stable CuO formation on Cu 3 Pt(111) as compared to Cu 3 Pd(111) . Similar effect can be observed for Cu(410) oxidation at lower T S . After 2.2 eV HOMB irradiation, CuO forms on Cu(410) at T S = 100 K 30 . Annealing at T S = 273 K reduces CuO into Cu 2 O . Alloying increases the corresponding transition temperature from CuO to Cu 2 O . CuO persists even at T S = 300 K.  33 ). However, the uptake curves in Fig. 1 show different tendency of reactivity. We see O-coverages lower than what we would expect from the d-band for both Cu 3 Pd(111) and Cu 3 Pt(111).

Discussions
Now, let us consider what the uptake curves taken at T S = 300 K and 500 K tell us (cf., Fig. 1). Here, we discuss the early stage of oxidation where only the dissociative adsorption of O 2 occurs on Cu(111). Note that on Cu (111)     The protective layer. The difference in efficiency of oxide formation between Cu and Cu alloy surfaces can be ascribed to the resulting protective layer of Pd (Pt) layer formed at the interface between the bulk and surface Cu oxide (cf., e.g., Cu 3 Au(111) 20 ). As mentioned in S.4 in supplementary information, only Cu oxidation occurs on Cu 3 Pd(111) and Cu 3 Pt(111) . Previous studies also show that only Cu oxide forms on Cu-Pt alloy 28 and Cu deposited on Pt(111) 35 . The selective Cu oxidation results in Pd-and Pt-rich interface layers. We see a steep O uptake curve, coming from Cu 2 O formation 34 , for Cu(111) at T S = 300 K (cf., Fig. 1 (left-panel), region above ca. 10 17 molecules cm −2 ). On Cu(111), Cu 2 O formation occurs due to collision induced absorption (CIA) 36 .
On Cu 3 Au , the inert Au interface layer prevents O atoms from diffusing further into the bulk by CIA process. As a result, Cu oxide hardly forms at 300 K [17][18][19][20] . Here, we find that the interface Pd-and Pt-layers also prevent O diffusion further into the bulk to realize the CIA process. It also prevents Cu diffusion from the bulk to the surface CuO.
Mobility/diffusion. At T S = 500 K , we expect that the increased temperature would enhance atom diffusion, allowing for Cu oxide formation further into the bulk. On Cu 3 Au(111) , Cu 2 O forms at 500 K 19 . Cu 3 Pd(111) also has a steeper O uptake curve at 500 K than at 300 K (cf., above 10 17 molecules cm −2 in Fig. 1). However, Cu 3 Pt(111) remains relatively inactive, even at 500 K . This is because of the presence of less mobile Pt at the interface. In Cu, Pd has an activation/diffusion barriers of ca. 0.88 eV 37 , Au has 1.1 eV 38 , and Pt has 1.51 eV 39 . This is consistent with the reactivity observed above ca. 10 17 molecules cm −2 at 500 K , i.e., the region where mainly CuO forms. The interface Pt suppresses Cu oxide growth into the bulk even at 723 K 28 . Cu diffusion through the Pd (Pt) interface would also be unlikely considering the high Cu diffusion barriers (ca. 2.5 eV in Pd, 2.75 eV in Pt, and 2.0 eV in Au) 40 . As expected, these differences in diffusion barriers affect the kind of Cu oxides formed on the Cu alloy surfaces. At 100 K , metastable CuO forms on Cu(410) 30 . The low temperature suppresses O diffusion from the surface to the bulk (and also Cu diffusion from the bulk to the surface), while collision induced absorption (CIA) allows for a continuous supply of O atoms to the surface. We can expect similar effects on the Cu-Pd and Cu-Pt alloy surfaces.
At 300 K , the presence of Pd or Pt at the corresponding interfaces suppresses the diffusion of O and Cu, while CIA allows for a continuous supply of O atoms on the surface. Similarly, on Cu 3 Au(111) at 300 K, O atoms adsorbed on surface cannot diffuse into bulk due to the interface Au 17,18 . As a result, no Cu oxides form on Cu 3 Au(111) at 300 K.
At 500 K , enhanced diffusion allows oxidation further into the bulk of Cu 3 Au(111) and Cu 3 Pd(111) , and we find growth of the thermodynamically more stable Cu 2 O . On the other hand, CuO persists on Cu 3 Pt at 500 K because of the higher Pt diffusion barrier. As shown in Fig. 3, Cu 2 O forms after annealing (after the HOMB irradiation at 300 K ). Higher T S enables O diffusion further into the bulk and further Cu supply to the surface CuO. This occurs at T S = 650 K on Cu 3 Pt , and T S = 600 K on Cu 3 Pd . The less diffusive Pt present at the interface prevents further Cu 2 O formation as compared to Pd. At T S = 723 K , Cu 2 O islands grow on Cu-Pt alloy. However, the oxide does not grow deeper into bulk even at high temperature because of less diffusivity of Pt 28 .
Charge distribution. In Fig. 4, for 0.5 ML-O adsorbed on Cu 3 Pt(111) , we see that in the early stage of oxidation, Cu segregates to the surface and oxidized to form CuO. The charge distribution also shows that the more electronegative Pt competes with O for the Cu electrons. This, together with the mobility arguments presented earlier, accounts for why Cu 2 O easily forms on Cu(111) and not on Cu-alloys. Note that this could also consistently explain previous reports for that electron transfer from the metal substrate (Au, Ni, Mo, Cu, V) to the metal oxide resulted in Mo 6+ reduction to Mo 4+ and/or Mo 5+ near the interface 41

Summary and conclusions
In conclusion, we studied the oxidation of Cu 3 Pd(111) and Cu 3 Pt(111) , using 2.3 eV hyperthermal oxygen molecular beam (HOMB) source, and synchrotron-radiation X-ray photoemission spectroscopy (SR-XPS) for surface characterization. We determined the Pd-and Pt-layer profiles of Cu 3 Pd(111) and Cu 3 Pt(111) from the corresponding Pd-3d and Pt-4f spectra. At 300 K , we found mainly (only) the presence of CuO on both Cu 3 Pd(111) and Cu 3 Pt(111) . At 500 K , we found Cu 2 O on Cu 3 Pd(111) , and only CuO on Cu 3 Pt(111) . For comparison, at 300 K , Cu 2 O forms on Cu(111), and no oxides form on Cu 3 Au(111) . The early formation of Cu oxides on Cu 3 Pd(111) and Cu 3 Pt(111) results in hindered reactivity (susceptibility) to further oxidation into the bulk (resulting in the formation of Cu 2 O ) as compared to Cu. Cu oxide formation depends on the Cu alloy component and temperature. We ascribe this difference/preference of Cu oxide species to the mobility of the interfacial Cu/Pd/Pt, and the charge transfer between the initial (pre-oxidized) surface (Cu) and subsurface (Cu, Pd, or Pt) species. The presence of Cu 2 O and metastable CuO at the Pd and Pt interface could play an important role in catalytic reactions. We showed that we can control the oxidation state of the surface metal oxide by alloying, which in turn would allow us to control the catalytic reactivity of the oxides. built at BL23SU in SPring-8 [17][18][19][20]43,44 , with the base pressure of < 2 × 10 −8 Pa. Briefly, our surface reaction analysis chamber has an electron energy analyzer (OMICRON EA125-5MCD) and a Mg/Al-Kα twin-anode x-ray source (OMICRON DAR400). We also have a quadrupole mass spectrometer, for monitoring the molecular beam, located opposite to the HOMB (hyperthermal oxygen molecular beam) source. We purchased Cu 3 Pd(111) and Cu 3 Pt(111) samples from SPL and MaTeck, respectively. We cleaned the Cu 3 Pd(111) and Cu 3 Pt(111) samples by repeated sputtering with Ar + and annealing for 20 min ( Cu 3 Pd(111) : 1.0 keV, 723 K , Cu 3 Pt(111) : 0.5 keV, 773 K ), until the impurities were no longer detectable by SR-XPS (synchrotron-radiation X-ray photoemission spectroscopy). We generated a HOMB by the free expansion of mixed gas of O 2 , He and/or Ar from a nozzle with a small orifice. The translational energy of HOMB, E SG can be expressed as: where S (= 1.557) is a factor that is expressed by using the Mach number, R (= 8.617 × 10 −5 eV · K −1 ) is gas constant, T 0 is the nozzle temperature, m SG is the mass of the reactant gas ( O 2 ) and m s is the reduced mass of the mixed gas (He and/or Ar). By changing the gas mixing ratios at the nozzle and nozzle temperature T 0 , we can control the kinetic energy of the incident HOMB, E SG . The detailed explanation for HOMB generation is shown in Refs. 45,46 . We set the nozzle temperature to 1400 K , obtaining a 2.3 eV HOMB. We irradiate the sample surface with a HOMB (along the surface normal) at T S = 300 and 500 K . The pressure during the HOMB irradiation is about 1 × 10 −5 Pa. The oxidation by the scattered O 2 which causes the pressure increase is not important because the percentage of O 2 in the gas is only 1% for the 2.3 eV HOMB, and the oxidation by thermal O 2 is less reactive than by HOMB as shown in Fig. 1. After each irradiation, we then obtained the corresponding high-resolution SR-XPS spectra at T S = 300 K , at detection angles of θ = 0 • and 70 • from the surface normal, using a monochromatic SR beam with a photon energy of 1100 eV. We performed the HOMB irradiations at T S = 300 and 500 K.
Theoretical calculations. We performed density functional theory (DFT)-based total energy calculations as implemented in the Vienna Ab Initio Simulation Package 47,48 , within the generalized gradient approximation (GGA) 49 , using plane waves (600 eV cutoff energy) and the projector augmented wave method 50 . To model Cu 3 Pd(111) and Cu 3 Pt(111) , we used slabs. Each slab has seven fcc(111) layers, separated by ca. 1.50 nm ( Cu 3 Pd ) and 0.7 nm ( Cu 3 Pt ) vacuum, repeated in a supercell geometry (shown in Fig. S.2). We also applied dipole corrections. Each layer in the slab contains 4 atoms, so that the composition (of Pd or Pt) can be varied in steps of 25% . Convergence tests for k-point meshes and cutoff energy values were performed (cf., Table. S.1). We have chosen sufficiently large supercells so as to avoid interaction between adsorbates in the neighboring super-(1) E SG =S 2 · R · T 0 · m SG m s ,