Dehydrogenation of methanol on Cu2O(100) and (111) Zahra Besharat, Joakim Halldin Stenlid, Markus Soldemo, Kess Marks, Anneli Önsten, Magnus Johnson, Henrik Öström, Jonas Weissenrieder, Tore Brinck, and Mats Göthelid Citation: The Journal of Chemical Physics 146, 244702 (2017); doi: 10.1063/1.4989472 View online: http://dx.doi.org/10.1063/1.4989472 View Table of Contents: http://aip.scitation.org/toc/jcp/146/24 Published by the American Institute of Physics THE JOURNAL OF CHEMICAL PHYSICS 146, 244702 (2017) Dehydrogenation of methanol on Cu2 O(100) and (111) Zahra Besharat,1,2 Joakim Halldin Stenlid,3 Markus Soldemo,1 Kess Marks,4 Anneli Önsten,1 Magnus Johnson,2 Henrik Öström,4 Jonas Weissenrieder,1 Tore Brinck,3 and Mats Göthelid1,a) 1 Material Physics, KTH Royal Institute of Technology, SCI, S-164 40 Kista, Sweden of Surface and Corrosion Science, Department of Chemistry, KTH Royal Institute of Technology, Stockholm S-100 44, Sweden 3 Applied Physical Chemistry, School of Chemical Science and Engineering, KTH Royal Institute of Technology, S-100 44 Stockholm, Sweden 4 Department of Physics, Stockholm University, S-106 91 Stockholm, Sweden 2 Division (Received 20 April 2017; accepted 1 June 2017; published online 26 June 2017) Adsorption and desorption of methanol on the (111) and (100) surfaces of Cu2 O have been studied using high-resolution photoelectron spectroscopy in the temperature range 120–620 K, in combination with density functional theory calculations and sum frequency generation spectroscopy. The bare (100) surface exhibits a (3,0; 1,1) reconstruction but restructures during the adsorption process into a Cu-dimer geometry stabilized by methoxy and hydrogen binding in Cu-bridge sites. During the restructuring process, oxygen atoms from the bulk that can host hydrogen appear on the surface. Heating transforms methoxy to formaldehyde, but further dehydrogenation √ √ is limited by the stability of the surface and the limited access to surface oxygen. The ( 3 × 3)R30➦-reconstructed (111) surface is based on ordered surface oxygen and copper ions and vacancies, which offers a palette of adsorption and reaction sites. Already at 140 K, a mixed layer of methoxy, formaldehyde, and CHx Oy is formed. Heating to room temperature leaves OCH and CHx . Thus both CH-bond breaking and COscission are active on this surface at low temperature. The higher ability to dehydrogenate methanol on (111) compared to (100) is explained by the multitude of adsorption sites and, in particular, the availability of surface oxygen. Published by AIP Publishing. [http://dx.doi.org/10.1063/1.4989472] INTRODUCTION Alcohols derived from renewable sources are considered an environmentally friendly alternative to fuels derived from petrochemical resources.1–3 Hydrogen gas, for use in fuel cells, can be produced by dehydrogenation or oxidation of alcohols, and methanol is one very potent candidate.4,5 Catalysts for methanol dehydrogenation based on noble metals are effective,6 but also expensive and should be replaced by cheaper and more abundant materials. Cu-based catalysts have been studied in detail.7–10 This shows that Cu, along with Co,11 has the potential to replace noble metals in oxidation reactions.12–14 Cu2 O is the thermodynamically most stable phase of the O/Cu(111) system under relevant conditions,15 and it is also one of the most abundant corrosion products of copper.16,17 In order to achieve optimal performance of the Cu-based catalysts, a detailed understanding of the catalytic processes is necessary. A recent theoretical study of methanol decomposition and reactions on Cu(111) by Zuo et al. found that the first step in methanol chemisorption is OH bond scission leading to methoxy binding in a fcc-hollow site and hydrogen in another fcc hollow site,10 which is in agreement with previous work by Greely and Mavrikakis.9 The formation of methoxy is indeed the most common first reaction step at low coverage a) Author to whom correspondence should be addressed: gothelid@kth.se. Telephone: +46 (0)8 7904154. 0021-9606/2017/146(24)/244702/10/$30.00 on metals.6,18–24 With oxygen, hydroxyl, or water present on a Cu surface, the methanol reaction path is modified,10,25–28 typically facilitating a larger degree of methoxy formation. Further dehydrogenation of methoxy leads either to methyl and oxygen or formaldehyde and hydrogen, depending on the detailed adsorption structure and adsorption site and molecular bond configuration in the transition state.10,24 Under realistic reaction conditions, Cu is likely to be fully or partly oxidized, which make studies on oxides, an important step towards functional catalytic systems. Cox and Schulz29 used temperature programmed desorption (TPD) and photoelectron spectroscopy (PES) to compare product desorption patterns from the Cu2 O(111) and (100) surfaces following methanol adsorption; they noted structural dependence with formaldehyde being the preferred specie from the (100) surface, whereas further dehydrogenation to carbon monoxide is favored on (111). Hydrogen desorption was observed in a wide temperature range from both surfaces. On both surfaces, they suggested methoxy binding to the surface, and at higher doses, they observed physisorbed methanol,29 a conclusion also supported by Jones and Solomon.30 Zhang et al.31 studied methoxy formation from methanol on an ideal unreconstructed Cu2 O(111) surface using density functional theory (DFT) and the influence of pre-adsorbed oxygen. They found that methanol, methoxy, and hydrogen all bind preferentially to under-coordinated surface Cu ions (called CuCUS ).31 Pre-adsorbed oxygen reacts with hydrogen liberated from the methoxy formation, which also binds to 146, 244702-1 Published by AIP Publishing. 244702-2 Besharat et al. CuCUS . It thus appears that Cu ions play an active role, in agreement with previous conclusions from methanol synthesis.8,32 Note, however, that the scientific community has not yet reached consensus regarding the surface structure of the (111) facet of Cu2 O, especially regarding the presence or absence of surface CuCUS atoms. Structure models including oxygen and√copper √ vacancies have been proposed for the prevailing ( 3 × 3)R30➦ reconstruction phase of the Cu2 O(111) surface, as well as for the (1 × 1) surface,33,34 and their influence on adsorption and surface reactions has been addressed.35,36 Although √ the √ Cu2 O(111) surface is typically dominated by the ( 3 × 3)R30➦ structure, there are always (1 × 1) minority areas on the surface,33 which may play an important role in the surface reactivity. Furthermore, the structure of Cu2 O(100) has also been discussed and different models have been proposed,34,37,38 most recently a (3,0; 1,1) reconstruction.38 In this study, we use photoelectron spectroscopy (PES), DFT, and sum frequency generation (SFG) spectroscopy to study methanol adsorption and decomposition on Cu2 O(111) and (100) surfaces under √ UHV √ conditions. The clean Cu2 O(111) surface has a ( 3 × 3)R30➦-reconstruction, and the Cu-terminated Cu2 O(100) surface exhibits a (3,0; 1,1) reconstruction. The (100) surface changes reconstruction during adsorption of methanol into a Cu-dimer structure with methoxy and hydrogen in Cu-bridge site. The stability of this structure rationalizes the lack of further reactivity. Dehydrogenation is more effective on the (111) surface with its more open structure including oxygen, possibly, co-existing copper vacancies, and under-coordinated surface atoms. EXPERIMENTAL DETAILS Cu2 O crystals cut along the (111) and (100) planes were purchased from Surface Preparation Laboratory (The Netherlands). The preparation method used for surface cleaning was repeated cycles of argon ion sputtering and annealing in O2 (2 × 10☞ 6 mbar) at 870 K for 15 min, followed by annealing in UHV at 780 K. The sample heating was performed by running a current through a thin Ta-foil, to which the samples were mounted. The temperature was measured with a K-type thermocouple in close contact with the sample. The purity of the sample was checked by PES and the samples were observed to contain a small amount of K (<0.05 at. %). Methanol (99.9% spectroscopic grade) was introduced through a precision leak valve at pressures around 1 × 10☞ 8 mbar. Before being introduced to the chamber, methanol was purified by repeated freeze-pump-thaw cycles. The PES measurements were performed at beamline I311 at MAX-lab in Lund.39 Photons in the energy range 50-1500 eV were produced in an undulator and selected by a modified Zeiss SX-700 plane grating monochromator. A Scienta SES200 electron analyzer was used to record photoelectron spectra. The base pressure in the analysis chamber was 1 × 10☞ 10 mbar or better. In connection to this chamber was a preparation chamber equipped with low-energy electron diffraction (LEED) optics, a sputter gun, and leak valves. Due to problems with sample charging at low temperatures, especially for Cu2 O(100), the C1s binding energy scale is J. Chem. Phys. 146, 244702 (2017) referenced to K2p at 293.1 eV, measured from a non-charging sample. Due to the very low K-contamination level, this method is estimated to have an error of ±0.1 eV. The binding energy scale for O1s is given relative to the bulk O1s component. Thus, the possible shifts of O1s due to, e.g., band bending could not be studied. Numerical curve fitting was done with the Fit-XPS software, using Voigt functions and a linear background; fitting parameters used are supplied in the figure caption. All spectra were normalized to the background intensity on the low-binding energy side of the core level, where no other treatment was done. For the SFG experiments, the Cu2 O(111) surface was cleaned by argon ion sputtering for 15 min, annealing in O2 at 870 K for 20 min followed by annealing without O2 at 780 K for 20 min. After cleaning, the sample was cooled to 107 K and exposed to 4.8 L of methanol. All measurements were done at 107 K. COMPUTATIONAL DETAILS Computational methods All DFT + U calculations were performed with the Vienna Ab initio Simulation Package (VASP).40–44 The PBE45 functional was used with Grimme’s D3(BJ)46,47 dispersion. Based on a previous benchmark,48 a U-j value of 3.6 eV was chosen for the Hubbard (+U)49 Cu d-state corrections. Standard PBE PAW potentials were employed to represent the cores, whereas the valence electrons (Cu 3d 10 4s1 , O 2s2 2p4 , C 2s2 2p2 , and H 1s1 ) were treated by a plane-wave basis set with a 400 eV cutoff. The Brillouin zone was sampled with a Γ-centred 4 × 4 × 1 k-point mesh using the tetrahedron method with Blöchl corrections.50 Spin-polarization was allowed throughout. For the structural optimizations, forces were relaxed to less than 0.03 eV/Å. Molecular methanol (MeOH) was studied in a cubic (15 × 15 × 15) Å3 cell using a single Γ-point. O1s core-level shifts were computed using the final state approximation51,52 as specified in Ref. 38. For the C1s shifts, relative shifts for surface methanol compared to surface methoxy (MeO) were calculated. Adsorption energies were obtained by
1 (1) Eslab/MeOH − Eslab − nEMeOH , n where n = 1 for asymmetric slabs and n = 2 for symmetric slabs. E slab/MeOH denotes the electronic energy of a surface with molecularly or dissociatively adsorbed methanol, where E slab is the electronic energy of the clean surface and E MeOH gives the electronic energy of methanol in gas phase. Ead = Surface models In Fig. 1, we present a selected set of surface structures used for the (100) and (111) surfaces of Cu2 O, which are previously discussed in Refs. 32 and 33 for (111) and in Ref. 38 for (100). Figure 1(a) is the unreconstructed Cu-terminated (100) surface, Fig. 1(b) shows the ridge-dimer reconstructed c(2 × 2) surface, Fig. 1(c) shows the c(2 × 2) surface with a surface O atom moved from the lattice to a surface adatom position, and Fig. 1(d) depicts the (3,0; 1,1) reconstruction for the (100) surface. The (3,0; 1,1) reconstruction is the experimentally 244702-3 Besharat et al. J. Chem. Phys. 146, 244702 (2017) FIG. 1. Structure models for the (100) and (111) surfaces of Cu2 O:√(a) Cu √ terminated (100)-(1 × 1), (b) ridge-dimer (100)-c(2 × 2),√(c) O√ad -(100)-c(2 × 2), (d) (100)-(3,0; 1,1), (e) non-polar Cu2 O terminated (111)-(1 × 1), (f) ( 3 × 3)R30➦ with 1/3 ML OCUS vacancies (model A), and (g) ( 3 × 3)R30➦ with both 1/3 ML OCUS vacancies and 1 ML CuCUS vacancies (model B). Coloring: CuBulk (light beige), OBulk (red). Surface atoms are indicated as follows: for (100), CuCS (dark beige) and OSurf (deep red); for (111), CuCUS (brown) and CuCS (dark beige) as well as OCUS (deep red) and OCS (purple). most common structure observed in LEED after the surface treatment used in this paper. Figure 1(e) is the unreconstructed (111) surface, which was also used by Zhang et al.31 Note that upon full relaxation, this surface reconstructs and forms CuCUS -CuCS dimers at a distance of 2.43-2.48 Å (compared √ to √ the unrelaxed distance of 3.05 Å). The reconstructed ( 3 × 3)R30➦ surface with 1/3 monolayer (ML) oxygen vacancies is displayed in Fig. 1(f) (depicted √ √ model A in Ref. 33), whereas Fig. 1(g) shows the ( 3 × 3)R30➦ structure with all CuCUS atoms missing in addition to the 1/3 ML oxygen vacancies (model B in Ref. 33). The latter surface structure has been proposed to best match experimental STM images under oxygen-lean conditions.53 Not included in Fig. 1 is the (1 × 1) surface structure of (111) with 1 ML CuCUS vacancies. For the adsorption studies, a ML is defined as one adsorbate per surface Cu site on the ideal bulk terminated surface [0.107 adsorbates/Å2 for (100) and 0.124 adsorbates/Å2 for (111)]. Periodic slab models based on an equilibrium fcc lattice parameter of 4.316 Å (as identified in a previous study38 ) were used to represent the above-mentioned surface structures. The models comprised a periodic structure of six Cu2 O layers separated by a minimum vacuum distance of 20 Å, with the top two layers being allowed to relax. Asymmetric slabs were employed for the (100) surface, using p(2 × 2) supercells to represent the (1 × 1) and c(2 × 2) surface structures. During the optimization of the ideal (1 × 1) reference surface, the atoms were constrained in the lateral dimensions since the unreconstructed (1 × 1) spontaneously adopts the ridge-dimer c(2 × 2) reconstruction. The (3,0; 1,1) reconstruction was represented by the equivalent (2,☞1; 1,1) unit cell enlarged to (2,☞1; 2,2) in order to assure a sufficient distance between the periodic images of the adsorbates. In contrast to the (100) surface, symmetric slabs were used for the (111) surface structures. The ideal √ (1√ × 1) surface was studied by both the p(2 × 2) and ( 3 × 3)R30➦ supercells √ √ yielding essentially identical results. The smaller ( 3 × 3)R30➦ cell was therefore used for the remaining (111) surface structures. DFT results The energetically preferred adsorption sites and respective adsorption energies for methanol, methoxy, hydrogen, and formaldehyde on the different Cu2 O surfaces are presented in Table I. Additional details are presented in Table S1 of the supplementary material. Selected adsorption structures are presented in Fig. 2. In Table II, we present calculated O1s and C1s photoemission core level shifts for the different adsorbates in their respective most√ stable √ position on the Cu2 O(100) c(2 × 2) and Cu2 O(111) ( 3 × 3)R30➦ surfaces. The O1s core level shift is with respect to O1s for bulk oxygen, whereas for C1s we use physisorbed methanol as the reference in our calculations. On the unreconstructed (100) surface methanol, adsorption is accompanied by Cu-dimerization and formation of a c(2 × 2) structure that bears resemblance to a mixture of the unreconstructed (1 × 1) and the ridge-dimer c(2 × 2) reconstructed surfaces in Fig. 1. This behavior is similar to that of water on the same surface.54 Methanol binds with oxygen in a slightly asymmetric bridge site, with the alcohol hydrogen binding to the exposed surface oxygen. The 0.3 eV difference in adsorption energy between the (1 × 1) and c(2 × 2) surfaces reported in Table I is a direct reflection of the ridge-dimer reconstruction energy. Dissociative adsorption is energetically favored over molecular adsorption by approximately 1.55 eV, which leads to methoxy binding with oxygen in a Cu-bridge site and hydrogen in a neighboring Cu-bridge site. The large dissociation energy is of similar magnitude as that of water (∆E diss = ☞1.4 244702-4 Besharat et al. J. Chem. Phys. 146, 244702 (2017) TABLE I. Preferred adsorption sites and energies for methanol (MeOH), methoxy (MeO), and hydrogen (H) on different Cu2 O surface reconstructions. The sites are defined in Fig. 1. Preferred adsorption site Surface (100) Ideal (1 × 1) Ridge dimer c(2 × 2) c(2 × 2)-Oad (3,0; 1,1) (111) Ideal (1 × 1) Ovac CUS (1 × 1) Cuvac CUS (1 × 1)√ √ 1/3 ML Ovac ( 3 × 3)R30➦b CUS √ √ 1/3 ML Ovac +1 ML Cuvac ( 3 × 3)R30➦c CUS CUS Adsorption energy (eV) MeOH MeO H Cu bridge Cu bridge Cu bridge Cu trimer Cu bridge Cu bridge Cu bridge Cu trimer Cu bridge Cu bridge Oad Cu trimer CuCUS CuCUS Cuvac CUS CuCUS Cuvac CUS CuCUS CuCUS /OCUS CuCUS OCS Cuvac /Cu O CS CUS CUS CuCUS OCUS Ovac OCUS CUS Molecular Dissociative ☞1.1 ☞0.8 ☞1.1 ☞0.7 ☞2.6a ☞2.3a ☞2.1 ☞0.9 ☞1.1 ☞0.8 ☞1.2 ☞0.8 ☞0.4 ☞0.2 +0.3 0.0 ☞1.0 ☞1.4 dissociation energy is ☞1.6 eV including sufficient significant figures in the calculations. A. c Model B. The adsorption energy for formaldehyde is ☞0.6 eV. a The b Model to ☞1.5 eV)54–56 rationalized by the stabilizing effect of the hydrogen and methoxy adsorbates on the polar (100) surface. Upon dissociation, the Cu-atoms under hydrogen retain their dimer character, whereas under methoxy the dimer is elongated leading to the surface ridge structure being disrupted. Clearly, surface Cu-atoms are flexible during the adsorption and dissociation process. The surface flexibility is further stressed when looking at adsorption on the (3,0; 1,1) reconstructed surface. There is an energy cost for methanol dissociation, but when the surface atoms are allowed to move, the reconstruction is lifted and the (1 × 1)/c(2 × 2) surface discussed above is formed, which is able to accommodate methoxy and hydrogen. Again, this behavior is analogous to the case of water adsorption.54 In the transient from the (3,0; 1,1) reconstruction towards the (1 × 1)/c(2 × 2) surface, an Oad -c(2 × 2) structure with lattice oxygen shifted towards the surface is found energetically favorable.38 The lattice oxygen ends up in an adatom position on top of the surface [Fig. 1(c)]. Upon dissociative methoxy adsorption onto this surface, the H atom sticks to the surface Oad atom with MeO at Cu bridges. Thus, irrespective of the start structure, methoxy binds to Cu in bridge site, whereas H adsorbs on O if available or to Cu-bridge. On the unreconstructed ideal (111) surface, the undercoordinated CuCUS is the preferred site for methanol, methoxy, and hydrogen in agreement with Zhang et al.31 However, upon co-adsorption of methoxy and hydrogen (e.g., from dissociation of methanol), the hydrogen atom instead prefers to sit on an OCUS site. Dissociation of methanol on the ideal surface is endothermic by approximately 0.45 eV according to our calculations (Table I). This is comparable to the calculated dissociation energy of water (0.3–0.4 eV) on the same surface.55,57,58 When OCUS are removed, OCS becomes the preferred site for hydrogen, whereas methanol and methoxy prefer the same CuCUS sites as on the ideal surface. For dissociative adsorption to be energetically favourable to molecular adsorption (at FIG. 2. This figure shows a selection of the methanol adsorption structures. Included for the (100) surface is a possible initial adsorption state onto the (3,0; 1,1) reconstructed surface (left), which evolves upon methanol dissociation to form the (1 × 1)/c(2 × 2) surface (right) via a Oad c(2 × 2) intermediate structure (middle). On (111) is the molecular methanol adsorption onto the CuCUS site on ideal (1 × 1) surface termination (left) compared√to molecular adsorption to the CuCUS vacancy of the ( 3 √ × 3)R30➦ model B surface with 1/3 ML OCUS and 1 ML CuCUS missing (middle). Dissociative adsorption onto the same surface termination is shown to the right. Coloring: Cu (orange), O (red), C (grey), and H (white). Note that the relative sizes of the atoms are not reflecting their true size ratio. 244702-5 Besharat et al. J. Chem. Phys. 146, 244702 (2017) TABLE II. Calculated O1s and C1s core level shifts for adsorbates in the most stable site on the Cu2 O surfaces. Specie and core level PES shift DFT (eV) (100) O1s MeOH O1s MeO O1s OHa C1s (MeOH–MeO) +1.95 +0.93 +0.65 +0.65b (111) O1s MeOH O1s MeO C1s (MeOH–MeO) +1.90 +1.40 +0.42b a OH by 0.4 eV. The formaldehyde resides at the Cu-trimer of the OCUS vacancy while the additional Had binds to an unoccupied OCUS site. The formation of formaldehyde from methoxy is endothermic by 0.8 eV on this surface. On (100), formaldehyde also prefers a flat geometry, in this case over a Cu-dimer. The formation of formaldehyde from methoxy on the (100) surface is endothermic by only 0.1 eV. However, a limitation for the further dehydrogenation on (100) is the low availability of surface sites (especially oxygen) for hydrogen storage. EXPERIMENTAL RESULTS AND DISCUSSION adsorbed on surface. b MeO at lower binding energy. The (100) surface T = 0 K), both OCUS vacancies and CuCUS vacancies are needed. In this case (i.e., the model B surface), methanol adsorbs in the CuCUS vacancy, methoxy adsorbs on the OCUS vacancy, whereas hydrogen preferentially adsorbs at OCUS . This yields an exothermic dissociation energy of ☞0.6 eV, whereas the dissociation energy with CuCUS present on the surface (i.e., model A) is endothermic by 0.2 eV. Thus, in order to favourably host dissociation products, a surface reconstruction with vacancies Based on the above, the √ is advantageous. √ results from the ( 3 × 3)R30➦ surface with 1/3 ML OCUS and all CuCUS missing (model B) are considered to best correspond to the experimental conditions and results reported for the (111) surface herein. However, the energy difference between molecular and dissociative adsorption of 0.2 eV on the surface with both OCUS and CuCUS vacancies (i.e., model A) is small considering that, e.g., thermal and coverage effects are omitted in the calculations. Hence, our results cannot exclude either surface structure or that both channels are active. Further dehydrogenation of methoxy to formaldehyde (i.e., CH√ 3O + H √ad → CH2 O + 2Had ) was also investigated. On the ( 3 × 3)R30➦ surface with both 1/3 ML OCUS and 1ML CuCUS vacancies (model B), formaldehyde prefers a flat adsorption structure over an O-down adsorption mode Methanol was adsorbed on Cu2 O(100) at 120 K at three doses: 0.25 L, 0.75 L, and 2.8 L (1 L = 10☞6 torr s☞1 ). After the highest dose, the surface was heated to the temperatures indicated in Fig. 3 and then cooled down for measurement. O1s and C1s spectra from each preparation are presented in Fig. 3. Numerical fitting was used to separate different contributions to the spectral development. Selected fitted spectra are shown in Fig. 4. Intensities of each component at different preparations are shown in Fig. 5. O1s has three components on the clean surface: B, S1, and S2. The B-component represents bulk oxygen36 and has been positioned at 0 eV relative binding energy. The O1s components at lower binding energy, S1 and S2, are associated with coordinatively unsaturated oxygen anions at the surface,36 also called CuO-like oxygen.59 Our recent calculations on the Cu2 O(100) (3,0; 1,1) structure essentially confirm this identification; S2 comes from oxygen atoms in the top layer and S1 from oxygen atoms in the second layer.38 The surface is rather open; all atoms contributing to S1 and S2 are at the surface. Adsorption of 2.8 L methanol at 120 K almost entirely removes S1 and S2. The same has been observed for water adsorption and has been interpreted as a fingerprint of the reformation of the (1 × 1) surface structure from the FIG. 3. O1s and C1s spectra recorded from Cu2 O(100) after different methanol doses and heat treatments. The O1s spectra have been aligned to the main peak, whereas the C1s binding energy scale is with respect to the Fermi level. 244702-6 Besharat et al. J. Chem. Phys. 146, 244702 (2017) FIG. 4. Numerical fits of selected O1s and C1s spectra from Fig. 3. For O1s, we use Lorentzian width (WL = 0.35 eV) and variable Gaussian width (WG ) between 0.7 and 1.1 eV. A small asymmetry (α < 0.03) was used to improve the fits. For C1s, WL = 0.35 eV, WG = 0.8 –1.15 eV. The two C1s components represent methanol (C2) and methoxy (C1). reconstructed surface.54 This finding was confirmed in a separate LEED experiment. This adsorbed layer generates two new components on the high binding energy side: A1 at 1.1 eV and A2 at 2.3 eV. A1 appears at a lower dose than A2. The separation between the two peaks is in good agreement with a previous report by Jones et al., who assigned the peaks to methoxy (A1) binding to the surface and methanol (A2) in a second adsorbate layer.30 Methoxy is indeed a very common product following methanol adsorption on oxides and metals,21–23,28–30,60,61 but it has been shown that adsorbed methoxy, formaldehyde, and hydroxyl or the formation of hydrogen binding to surface oxygen all can give similar O1s binding energies.36,60,62 Therefore, we cannot, from the binding energy position of this peak, uniquely identify the adsorbed product. However, the calculated O1s shifts from the thermodynamically most favorable adsorption configurations on the (1 × 1)/c(2 × 2) structure are 0.93 eV and 1.95 eV for methoxy and methanol, in good agreement with our experimental findings. C1s requires two components for a good fit during adsorption: C1 at 286.0 eV and C2 at 286.8 eV. The calculated energy shift between these two species is 0.65 eV, with methoxy at lower energy. Thus, C1 represents methoxy and C2 represents methanol. The intensity development of A2 and C2 during adsorption and desorption is very similar, which suggest that both of them belong to physisorbed methanol. At 155 K, there is still methanol present on the surface; at 190 K, A2 and C2 are almost completely gone from their respective spectra; and at 270 K, there is no methanol contribution, in good agreement with previous results from Cox and Schulz.29 The close relation between C2 and A2 suggests that with the present normalization, the intensity scales for C1s and O1s are comparable. Note that spectra are only normalized to the background of each spectrum; no compensation for photoexcitation cross section or escape depth has been made. At 270 K, A1 is the only O1s component other than B. The A1/B intensity ratio was recently used to determine the coverage of OH on the Cu2 O(100) surface following water adsorption and partial desorption.54 Those data were recorded from the same sample using the same experimental parameters at the same beamline as the present data were collected and are therefore comparable. Using the same method, we estimate that A1 represents 0.47 ML of oxygen species, including methoxy, surface hydroxyls, and perhaps FIG. 5. Intensities of separate components obtained from the numerical fits of O1s and C1s spectra in Fig. 3. The intensity scale has been calibrated according to Ref. 54. 244702-7 Besharat et al. formaldehyde. The scale of the vertical axis in Fig. 5 is based on this calibration. The C1s peak gradually shifts towards lower binding energy during heating. After removing the methanol layer at 270 K, the C1s spectrum is slightly asymmetric with a peak at 285.6 eV. Further heating to 370 K results in a narrower peak at 285.4 eV, and at 520 K, just before the last carbon species desorbs, the binding energy is 284.9 eV. Cox and Schulz29 observed a preference for formaldehyde and hydrogen desorption (in TPD) following methanol adsorption on Cu2 O(100). We suggest that the C1s component at 285.4 eV reflects formation of formaldehyde. The small peak at 284.9 eV is assigned to adsorbed OCH (an intermediate towards CO, which was also observed in the desorption spectra of Cox and Schulz29 ) or CHx .63 The preference for aldehyde in a lying geometry, as suggested by J. Chem. Phys. 146, 244702 (2017) our DFT calculations, allows both CH bond breaking and CO-scission. The intensities of C1 and A1 develop differently from temperature, in fact also during adsorption: from 180 K to 270 K, A1 decreases slower than C1, whereas from 270 K up to 520 K, A1 decreases faster. Core level intensities can change for other reasons than just a change in amount of adsorbed species, for example, photoelectron diffraction effects, but the observed development suggests that A1 has contributions other than from carbon containing species. When going from methanol to methoxy, and further to formaldehyde, hydrogen atoms are released and there will be atomic hydrogen on the surface. During the transformation from (3,0; 1,1) to (1 × 1)/c(2 × 2), our DFT calculations show that an intermediate Oad -c(2 × 2) structure54 may be formed. At this state, the oxygen atoms are pushed outwards and these adatoms can readily react with and FIG. 6. O1s (a) and C1s (b) spectra recorded from Cu2 O(111) after different methanol doses and heat treatments. The O1s spectra have been aligned to the main peak, whereas the C1s binding energy scale is with respect to the Fermi level. Intensities of separate components in the numerical fits of O1s and C1s spectra are presented in (c). (d) shows selected fitted C1s spectra. The four C1s components putatively represent methoxy (C1), methanol (C2), formaldehyde (C3), and multioxidized CHx Oy compounds (C4). 244702-8 Besharat et al. store hydrogen. The difference between C1 and A1 is at most ca. 0.25 ML. This number should not be over-interpreted, but it clearly indicates that O-atoms at the surface play a key role in storing hydrogen as surface OH. Upon full conversion to the thermodynamically more favorable adsorbate covered (1 × 1)/c(2 × 2) structure, the Oad atoms are reincorporated into the lattice. The (111) surface Adsorption and desorption studies of methanol were carried out in the same way on the (111) surface as on the (100) surface. In Fig. 6, we present photoemission spectra from O1s [Fig. 6(a)] and C1s [Fig. 6(b)], their intensity evolutions [Fig. 6(c)], and fits of C1s [Fig. 6(d)]. The start structure √ selected √ was a ( 3 × 3)R30➦. The O1s region contains a main peak from oxygen ions in the bulk (B) and a surface peak (S) at more than 1 eV to lower binding energy, in agreement with previous studies.36 Upon adsorption of methanol at 120 K, a broad peak appears on the high binding energy side and the contribution from the surface peak (S) is reduced but still present in the spectrum at 2.8 L. The clear separation of O1s into two components, as observed on (100), is not observed on (111), neither during adsorption nor desorption. This can be explained by a less homogeneous layer with a broader variation in adsorption sites than the ordered and well-defined layer with methoxy in bridge site on (100). The may also contain, in addi√ surface √ tion to the dominating ( 3 × 3)R30➦ structure, small (1 × 1) patches,33 which will contribute to the inhomogeneity of the surface. Moreover, the difference in our calculated O1s shifts of methanol and methoxy is smaller on the (111) surface than on (100). In order to verify the presence of methanol, we performed sum frequency generation (SFG) on the Cu2 O(111) surface at 107 K. Figure 7 shows the resulting spectra; the spectrum in black is the non-resonant background of the clean Cu2 O(111) and the red spectrum is the measured sum frequency generation signal from the surface after dosing 4.8 l of methanol. The blue curve is a fit of the SFG spectrum, with two clear resonances, at 2834 cm☞ 1 and 2951 cm☞ 1 . They correspond to the symmetric and asymmetric stretch vibrations of the methyl group.28,64,65 The wave numbers correspond better with methanol than with methoxy, which would be shifted by ca. 30 cm☞ 1 to lower frequencies for methoxy compared to methanol.28 Thus at 4.8 L, the surface is covered by a methanol layer with the FIG. 7. SFG spectra from the clean Cu2 O(111) surface (black) and the methanol covered surface (red). The blue curve is a fit of the difference of data with and without methanol data. Two resonances are observed from the symmetric and antisymmetric CH3 stretch. J. Chem. Phys. 146, 244702 (2017) methyl group pointing out from the surface. Although the DFT modeling was carried out at a low surface coverage compared to the expected coverage during the SFG measurements, the optimized structures of both methoxy and methanol correspond well with the picture of a methyl group directed perpendicular to the surface. In Fig. 6(d), numerical fits of C1s are presented. At 0.25 L, C1s comprises a main peak (C1) at 286.0 eV and a shoulder (C3) on the low binding energy side, at 285.4 eV, in fair agreement with the binding energies for methoxy and formaldehyde found on the (100) surface. At 0.75 L, C1s broadens on the high binding energy side with a shoulder (C4) at 287.0 eV. At 2.8 L, we use three components to fit the spectrum: C1 at 286.0 eV, C2 at 286.6, and C4 at 287.2 eV. We suggest that C2 at 286.6 eV stems from a physisorbed layer of methanol, the C1 component at 286.0 eV stems from methoxy, and the low energy C3 peak is due to formaldehyde. The C4 component has previously been observed in other studies and has been suggested to reflect more a oxidized species (CHx Oy ),66 with carbon close to surface oxygen. One can speculate that an OCHx fragment coordinates to surface oxygen. Heating to 155 K desorbs the methanol. At 190 K, there are three peaks: methoxy at 286.0 eV, formaldehyde at 285.4 eV, and the more oxidized specie at 287.1 eV. Upon further heating, to 270 K, C4 is removed from the spectrum. The spectrum still contains two contributions: a weaker peak at 285.4 eV from formaldehyde and a stronger peak at 284.8 eV, which we assigned above to OCH or CHx . At 370 K, C1s is centered at 284.2 eV. A separate measurement on adsorption of CO on Cu2 O(100) revealed C1s intensity above 288 eV binding energy. CO typically binds to surface metal ions with the C-down.35 However, if CO is created through step-wise dehydrogenation from an O-to-Cu σ-bonded methoxy, CO could potentially end up with the O-down. The calculated bond energy for O-down to a surface Cu is, nevertheless, far too low to be observed at this temperature.35 We instead suggest that the C1s component at 284.2 eV represents either CHx or small traces of carbon on the surface that desorbs below 520 K. The total intensity from C1s is slightly higher from (111) than from (100) at 2.8 l. The intensity of O1s follows the same trend. Consequently, we conclude that the surface coverage on (111) is higher than on (100). During heating of (100), O1s and C1s intensities developed differently, tentatively explained by hydrogen binding to surface oxygen and therefore contributing to A1. Although we note that O1s decreases a bit faster at increasing temperature on the (111) surface, there is no clear difference in the intensity development between O1s and C1s. We put forward two explanations for the difference between (100) and (111). First, the stronger C1s signal is in line with CHx on the surface. Second, there is less surface bound hydrogen on (111), which is somewhat counterintuitive since dehydrogenation is more efficient on (111) than on (100). However, hydrogen (and also water) desorbs molecularly and surface oxygen serves as a storage place for atomic hydrogen until another hydrogen is available. The (111) surface structure, with 1/3 ML oxygen vacancies and 2/3 ML OCUS available as well as lattice oxygen 244702-9 Besharat et al. exposed in the CuCUS vacancies, presents a surface rich in hydrogen storage sites on short distance leading to easier H2 formation and therefore swifter dehydrogenation at lower temperatures. CONCLUSIONS Adsorption of methanol and chemical reactions during thermal treatment has been studied using high-resolution photoelectron spectroscopy, density functional theory, and sum frequency generation. In agreement with previous TPD results,29 we find that dehydrogenation is more effective on the (111) surface than on (100). This is explained by a more open (111) surface with a putative mixture of copper vacancies, oxygen vacancies, and unsaturated surface atoms. Unsaturated oxygen (OCUS ) plays a key role in storing hydrogen released from methanol dehydrogenation. The close proximity to the site of dehydrogenation and hydrogen storage on (111) facilitates H2 formation and desorption. The (100) surface restructures from (3,0; 1,1) during the adsorption process into a Cu-dimer terminated surface with methoxy and hydrogen binding in Cu-bridge sites. This structure is comparatively stable and reduces the surface reactivity. SUPPLEMENTAL MATERIAL See supplementary material for geometric details (bond distances and adsorption specification) of the DFT structures. ACKNOWLEDGMENTS The financial support from the Swedish Research Council (VR) and the Swedish Nuclear Fuel and Waste Management Company (SKB) and Knut and Alice Wallenberg (KAW) is acknowledged. J.H.S. gratefully acknowledges the School of Chemical Science and Engineering at KTH for their Excellence stipend. We would also like to thank the helpful staff at MAX-lab. The computations were carried out on recourses provided by the Swedish National Infrastructure for Computing (SNIC) at the National Supercomputer Centre (NSC) in Linköping University, Sweden. 1 C. Okkerse and H. van Bekkum, “From fossil to green,” Green Chem. 1, 107–114 (1999). 2 T. N. Veziroglu, “Hydrogen energy system as a permanent solution to global energy environment problems,” Chem. Ind. 53, 383–393 (1999). 3 M. Momirlan and T. N. Veziroglu, “Current status of hydrogen energy,” Renewable Sustainable Energy Rev. 6, 141–179 (2002). 4 G. W. Huber, J. W. Shabaker, and J. A. Dumesic, “Raney Ni-Sn catalyst for H2 production from biomass-derived hydrocarbons,” Science 300, 2075– 2077 (2003). 5 I. J. Drake, K. L. Fujdala, A. T. Bell, and T. D. Tilley, “Dimethyl carbonate production via the oxidative carbonylation of methanol over Cu/SiO2 catalysts prepared via molecular precursor grafting and chemical vapor deposition approaches,” J. Catal. 230, 14–27 (2005). 6 Z. Jiang, B. Wang, and T. Fang, “A theoretical study on the complete dehydrogenation of methanol on Pd (100) surface,” Appl. Surf. Sci. 364, 613–619 (2016). 7 A. G. Sato, D. P. Volanti, I. C. de Freitas, E. Longo, and J. M. C. Bueno, “Site-selective ethanol conversion over supported copper catalysts,” Catal Commun. 26, 122–126 (2012). 8 A. Y. Rozovskii and G. I. Lin, “Fundamentals of methanol synthesis and decomposition,” Top. Catal. 22, 137–150 (2003). 9 J. Greeley and M. Mavrikakis, “Methanol decomposition on Cu(111): A DFT study,” J. Catal. 208, 291–200 (2002). J. Chem. Phys. 146, 244702 (2017) 10 Z. Zuo, L. Wang, P. Han, and W. Huang, “Insights into the reaction mechanisms of methanol decomposition, methanol oxidation and steam reforming of methanol on Cu(111): A density functional theory study,” Int. J. Hydrogen Energy 39, 1664–1679 (2014). 11 X. Xie, Y. Li, Z.-Q. Liu, M. Haruta, and W. Shen, “Low-temperature oxidation of CO catalysed by Co3 O4 nanorods,” Nature 458, 746–749 (2009). 12 A. E. Baber, X. F. Yang, H. Y. Kim, K. Mudiyanselage, M. Soldemo, J. Weissenrieder, S. D. Senanayake, A. Al-Mahboob, J. T. Sadowski, J. Evans, J. A. Rodriguez, P. Liu, F. M. Hoffmann, J. G. G. Chen, and D. J. Stacchiola, “Stabilization of catalytically active Cu+ surface sites on titanium–copper mixed-oxide films,” Angew. Chem., Int. Ed. 53, 5336–5340 (2014). 13 A. E. Baber, F. Xu, F. Dvorak, K. Mudiyanselage, M. Soldemo, J. Weissenrieder, S. D. Senanayake, J. T. Sadowski, J. A. Rodriguez, V. Matolı́n, M. G. White, and D. J. Stacchiola, “In situ imaging of Cu2 O under reducing conditions: Formation of metallic fronts by mass transfer,” J. Am. Chem. Soc. 135, 16781–16784 (2013). 14 S. Royer and D. Duprez, “Catalytic oxidation of carbon monoxide over transition metal oxides,” ChemCatChem 3, 24–65 (2011). 15 A. Soon, M. Todorova, B. Delley, and C. Stampfl, “Surface oxides of the oxygen–copper system: Precursors to the bulk oxide phase?,” Surf. Sci. 601, 5809–5813 (2007). 16 C. Leygraf and T. Graedel, Atmospheric Corrosion (John Wiley, New York, 2016). 17 J. Kunze, V. Maurice, L. H. Klein, H. H. Strehblow, and P. Marcus, “In situ scanning tunneling microscopy study of the anodic oxidation of Cu(111) in 0.1M NaOH,” J. Phys. Chem. B 105, 4263–4269 (2001). 18 B. R. Sheu, S. Chaturvedi, and D. R. Strongin, “Adsorption and decomposition of methanol on NiAl(110),” J. Phys. Chem. 98, 10258–10268 (1994). 19 J. R. B. Gomes and J. A. N. F. Gomes, “Comparative study of geometry and bonding character for methoxy radical adsorption on noble metals,” J. Mol. Struct.: THEOCHEM 503, 189–200 (2000). 20 S. A. Sardar, J. A. Syed, K. Tanaka, F. P. Netzer, and M. G. Ramsey, “The aluminium-alcohol interface: Methanol on clean Al(111) surface,” Surf. Sci. 519, 218–228 (2002). 21 G. C. Wang and Y. H. Zhou, “Characterization of methoxy adsorption on some transition metals: A first principles density functional theory study,” J. Chem. Phys. 122, 044707 (2005). 22 K. Habermehl-Cwirzen, J. Lahtinen, and P. Hautojärvi, “Methanol on Co(0001): XPS, TDS, WF and LEED results,” Surf. Sci. 598, 128–135 (2005). 23 I. H. Svenum, Ö. Borck, K. Schulte, L. E. Walle, and A. Borg, “Adsorption of methanol on Ni3 Al(111) and NiAl(110): A high resolution PES study,” Surf. Sci. 603, 2370–2377 (2009). 24 L. Xu, D. Mei, and G. Henkelman, “Adaptive kinetic Monte Carlo simulation of methanol decomposition on Cu(100),” J. Chem. Phys. 131, 244520 (2009). 25 K. Mudalige and M. Trenary, “The formation of methoxy from methanol on an oxygen covered Cu(100) surface at temperatures of 90-200 K,” J. Phys. Chem. B 105, 3823–3827 (2001). 26 K. Mudalige and M. Trenary, “Identification of formate from methanol oxidation on Cu(100) with infrared spectroscopy,” Surf. Sci. 504, 208–214 (2002). 27 R. Rydberg, “The oxidation of methanol on Cu(100),” J. Chem. Phys. 82, 567–573 (1985). 28 M. A. Chesters and E. M. McCash, “The adsorption and reaction of methanol on oxidized copper(111) studied by Fourier transform reflection absorption infrared spectroscopy,” Spectrochim. Acta, Part A 43, 1625–1630 (1987). 29 D. F. Cox and K. H. Schulz, “Methanol decomposition on single crystal Cu2 O,” J. Vac. Sci. Technol., A 8, 2599 (1990). 30 P. M. Jones, J. A. May, J. B. Reitz, and E. I. Solomon, “Electron spectroscopic studies of CH3 OH chemisorption on Cu2 O and ZnO single crystal surfaces: Methoxide bonding and reactivity related to methanol synthesis,” J. Am. Chem. Soc. 120, 1506 (1998). 31 R. Zhang, L. Hongyan, L. Lixia, L. Zhong, and W. Baojun, “A DFT study on the formation of CH3 O on Cu2 O(111) surface by CH3 OH decomposition in the absence and presence of oxygen,” Appl. Surf. Sci. 257, 4232–4238 (2011). 32 E. I. Solomon, P. M. Jones, and J. A. May, “Electronic structures of active sites on metal oxide surfaces: Definition of the Cu/ZnO methanol synthesis catalyst by photoelectron spectroscopy,” Chem. Rev. 93, 2623☞2644 (1993). 33 A. Önsten, M. Göthelid, and U. O. Karlsson, “Atomic structure of Cu2 O(111),” Surf. Sci. 603, 257–264 (2009). 244702-10 Besharat et al. J. Chem. Phys. 146, 244702 (2017) 34 K. H. Schulz and D. F. Cox, “Photoemission and low energy electron diffrac- 51 V. Nilsson, M. Van den Bossche, A. Hellman, and H. Grönbeck, “Trends in tion study of clean and oxygen dosed Cu2 O(111) and (100) surfaces,” Phys. Rev. B 43, 1610–1621 (1991). 35 B. Z. Sun, W. K. Chen, J. D. Zheng, and C. H. Lu, “Roles of oxygen vacancy in the adsorption properties of CO and NO on Cu2 O(111) surface: Results of a first principles study,” Appl. Surf. Sci. 255, 3141 (2008). 36 A. Önsten, J. Weissenrieder, D. Stoltz, S. Yu, M. Göthelid, and U. O. Karlsson, “Role of defects in surface chemistry on Cu2 O(111),” J. Phys. Chem. C 117, 19357–19364 (2013). 37 M. M. Islam, B. Diawara, V. Maurice, and P. Marcus, “Surface reconstruction modes of Cu2 O(001) surface: A first principles study,” Surf. Sci. 604, 1516–1523 (2010). 38 M. Soldemo, J. Halldin Stenlid, Z. Besharat, M. Ghadami Yazdi, A. Önsten, C. Leygraf, M. Göthelid, T. Brinck, and J. Weissenrieder, “The surface structure of Cu2 O(100),” J. Phys. Chem. C 120, 4373–4381 (2016). 39 R. Nyholm, J. N. Andersen, U. Johansson, B. N. Jensen, and I. Lindau, “Beamline I311 at MAX-LAB: A VUV/soft x-ray undulator beamline for high resolution electron spectroscopy,” Nucl. Instrum. Methods Phys. Res., Sect. A 467–468(Part 1), 520–524 (2001). 40 G. Kresse and J. Hafner, “Ab initio molecular dynamics for liquid metals,” Phys. Rev. B 47, 558–561 (1993). 41 G. Kresse and J. Hafner, “Ab initio molecular dynamics of open shell transition metals,” Phys. Rev. B 48, 13115–13118 (1993). 42 G. Kresse and J. Hafner, “Ab initio molecular dynamics simulation of the liquid metal amorphous semiconductor transition in germanium,” Phys. Rev. B 49, 14251–14269 (1994). 43 G. Kresse and J. Furthmüller, “Efficient iterative schemes for ab initio total energy calculations using a plane-wave basis set,” Phys. Rev. B 54, 11169–11186 (1996). 44 G. Kresse and J. Furthmüller, “Efficiency of ab-initio total energy calculations for metals and semiconductor using a plane-wave basis set,” Comput. Mater. Sci. 6, 15–50 (1996). 45 J. P. Perdew, K. Burke, and M. Ernzerhof, “Generalized gradient approximation made simple,” Phys. Rev. Lett. 77, 3865–3868 (1996). 46 S. Grimme, J. Anthony, S. Ehrlich, and H. Krieg, “A consistent and accurate ab-initio parametrization of density functional dispersion correction (DFTD) for the 94 elements H-Pu,” J. Chem. Phys. 132, 154104 (2010). 47 S. Grimme, S. Ehrlich, and L. Goerigk, “Effect of the damping function in dispersion corrected density functional theory,” J. Comput. Chem. 32, 1456–1465 (2011). 48 K. Yu and E. Carter, “A communication: Comparing ab initio methods of obtaining effective U-parameters for closed shell materials,” J. Chem. Phys. 140, 121105 (2015). 49 S. L. Dudarev, G. A. Botton, S. Y. Savrasov, C. J. Humphreys, and A. P. Sutton, “Electron energy loss spectra and the structural stability of nickel oxide: An LSDA+U study,” Phys. Rev. B 57, 1505–1509 (1998). 50 P. E. Blöchl, O. Jepsen, and O. K. Andersen, “Improved tetrahedron method for Brillouin zone integration,” Phys. Rev. B 49, 16223–16233 (1994). adsorbate induced core level shifts,” Surf. Sci. 640, 59–64 (2015). Köhler and G. Kresse, “Density functional study of CO on Rh(111),” Phys. Rev. B 70, 165405 (2004). 53 C. Li, F. Wang, S. F. Li, Q. Sun, and Y. Jia, “Stability and electronic properties of the O-terminated Cu2 O(111) surfaces: First-principles investigation,” Phys. Lett. A 374, 2994–2998 (2010). 54 J. H. Stenlid, M. Soldemo, A. J. Johansson, C. Leygraf, M. Göthelid, J. Weissenrieder, and T. Brinck, “Reactivity at the Cu2 O(100):Cu–H2 O interface: A combined DFT and PES study,” Phys. Chem. Chem. Phys. 18, 30570–30584 (2016). 55 J. H. Stenlid, A. J. Johansson, C. Leygraf, and T. Brinck, “Computational analysis of the early stage of cuprous oxide sulphidation: A top-down process,” Corros. Eng., Sci. Technol. (to be published). 56 Y. Li, L. F. Yan, and G. C. Wang, “Adsorption and dissociation of H O 2 on Cu2 O(100): A computational study,” J. Nat. Gas Chem. 20, 155–161 (2011). 57 R. Zhang, J. Li, B. Wang, and L. Ling, “Fundamental studies about the interaction of water with perfect, oxygen-vacancy and pre-covered oxygen Cu2 O(111) surfaces: Thermochemistry, barrier, product,” Appl. Surf. Sci. 279, 260–271 (2013). 58 C. Riplinger and E. A. Carter, “Cooperative effects in water binding to cuprous oxide surfaces,” J. Phys. Chem. C 119, 9311☞9323 (2015). 59 P. Jiang, D. Prendergast, F. Borondics, S. Porsgaard, L. Giovanetti, E. Pach, J. Newberg, H. Bluhm, F. Besenbacher, and M. Salmeron, “Experimental and theoretical investigation of the electronic structure of Cu2 O and CuO thin films on Cu(110) using x-ray photoelectron and absorption spectroscopy,” J. Chem. Phys. 138, 024704 (2013). 60 V. Matolı́n, J. Libra, M. Skoda, N. Tsud, K. C. Prince, and T. Skála, “Methanol adsorption on a CeO2 (111)/Cu(111) thin film model catalyst,” Surf. Sci. 603, 1087 (2009). 61 M. Shen and F. Zaera, “Methanol adsorption on clean and oxygenpredosed V(100) single-crystal surfaces,” J. Phys. Chem. C 112, 1636 (2008). 62 D. R. Mullins, M. D. Robins, and J. Zhou, “Adsorption and reaction of methanol on thin-film cerium oxide,” Surf. Sci. 600, 1547 (2006). 63 R. Larciprete, A. Goldoni, A. Groso, S. Lizzit, and G. Paolucci, “The photochemistry of CH4 adsorbed on Pt(111) studied by high resolution fast XPS,” Surf. Sci. 482-485, 134–140 (2001). 64 J. M. Camplin and E. M. McCash, “A RAIRS study of methoxy and ethoxy on oxidised Cu(100),” Surf. Sci. 360, 229–241 (1996). 65 A. Liu, S. Liu, R. Zhang, and Z. Ren, “Spectral identification of methanol on TiO2 (110) surfaces with sum frequency generation in the C–H stretch region,” J. Phys. Chem. C 119, 23486–23494 (2015). 66 A. Schaefer, W. C. Cartas, R. Rai, M. Shipilin, L. R. Merte, E. Lundgren, and J. F. Weaver, “Methanol adsorption and oxidation on reduced and oxidized TbOx (111) Surfaces,” J. Phys. Chem. C 120, 28617–28629 (2016). 52 L.