Advertisement

Topics in Catalysis

, Volume 61, Issue 5–6, pp 419–427 | Cite as

Stabilization of Oxidized Copper Nanoclusters in Confined Spaces

  • Nusnin Akter
  • Mengen Wang
  • Jian-Qiang Zhong
  • Zongyuan Liu
  • Taejin Kim
  • Deyu Lu
  • J. Anibal Boscoboinik
  • Dario J. Stacchiola
Original Paper

Abstract

Copper is an important industrial catalyst. The ability to manipulate the oxidation state of copper clusters in a controlled way is critical to understanding structure–reactivity relations of copper catalysts at the molecular level. Experimentally, cupric oxide surfaces or even small domains can only be stabilized at elevated temperatures and in the presence of oxygen, as copper can be easily reduced under reaction conditions. Herein bilayer silica films grown on a metallic substrate are used to trap diluted copper oxide clusters. By combining in situ experiments with first principles calculations, it is found that the confined space created by the silica film leads to an increase in the energy barrier for Cu diffusion. Dispersed copper atoms trapped by the silica film can be easily oxidized by surface oxygen chemisorbed on the metallic substrate, which results in the formation and stabilization of Cu2+ cations.

Graphical Abstract

Keywords

Copper Nanocluster Silica In situ 

1 Introduction

Copper is used extensively as a catalyst in industrial processes such as the purification of hydrogen through the water–gas shift reaction (WGS) [1, 2] and the synthesis of methanol from syngas [3, 4]. Cu cations stabilized on mixed oxides or zeolites have been shown to be highly active for automotive emission controls [5, 6] and small clusters of copper oxide in zeolites have been shown to be promising catalysts for directly converting methane into methanol. A central challenge to understand fundamental aspects related to the chemistry of copper catalysts is that in the presence of small concentrations of oxidants Cu can be easily oxidized [7] and, in most cases, be present in more than one oxidation state [8, 9]. For this reason, it is important to prepare well-defined model systems for fundamental studies on structure–reactivity relationships. Metallic copper surfaces are readily available and homogenous cuprous oxide (Cu2O) films can be grown [10, 11], but cupric oxide (CuO) surfaces [12] or even small domains [8] can only be stabilized in the presence of oxygen. The synthesis of disperse small Cu nanoclusters with different degrees of oxidation, can help to explore reactivity changes as a function of the size of catalysts.

Two-dimensional (2D) nanoporous ultrathin silicate films have been recently grown on metal substrates, in which the bilayer films interact weakly with the support [13]. These thin films can act as molecular sieves, allowing the diffusion of single Pd atoms through the film onto the interface with the underlying metal substrate while blocking Au atoms [14, 15, 16]. More recently the trapping of noble gas atoms in the nano cages of the silicate film has been demonstrated [17]. The presence of the film has been shown to passivate the oxidation of the underlying metal substrate, due to changes on the chemistry at the confined interface [18], where the thickness of the space at the silicate/metal interface can be tuned by the type and concentration of adsorbed molecules on the Ru(0001) surface [19].

In this study, small amounts of diluted copper atoms are deposited onto 2D silica films. By combining experimental and computational results, it is shown that a high concentration of oxygen at silica/Ru(0001) interfaces stabilizes clusters with Cu2+ sites at room temperature by limiting its mobility on the surface. Furthermore, the highly oxidized copper clusters remain stable under ultra-high vacuum conditions without the need of an oxygen atmosphere.

2 Experimental and Computational Methods

The Ru(0001) single crystal was held tightly using a Ta loop embedded in a groove machined around the crystal edge. The Ta loop was attached directly to the manipulator feed-through which was used for both mechanical support and heating/cooling. The sample temperature was measured by a K-type thermocouple attached to the top edge of the crystal. The Ru(0001) single crystal surface was cleaned with several cycles of Ar+ sputtering and annealing at 1200 K. It was then exposed to 3 × 10−6 Torr O2 at 1200 K in order to form a (2 × 2)-3O/Ru(0001) surface. The silica bilayer film, from now on referred to as 2D-SiO2, was grown on the (2 × 2)-3O/Ru(0001) surface as described in detail elsewhere [20]. Briefly, the 3O(2 × 2)–Ru(0001) phase was prepared by oxidizing the clean surface in 3 × 10−6 Torr of O2 at 1200 K for 5 min, cooling to 500 K and then evacuating the chamber. To grow the 2D-SiO2 film, Si was thermally evaporated onto the (2 × 2)-3O/Ru(0001) surface at room temperature under 2 × 10−7 Torr O2, followed by the oxidation at 1200 K in 3 × 10−6 Torr O2 for 10 min and then slowly cooled down under the same O2 pressure. Cu was deposited at room temperature in UHV condition by heating a Cu wire loop or by using a SPECS EBE-4 evaporator and a quartz crystal microbalance to control the amount of metal deposited. The amount of Cu deposited on the silica film was measured by Auger or XPS spectroscopy. High-purity CO (GTS-WELCO, 99.999%) and O2 (GTS-WELCO, 99.9999%) gases were further purified in liquid nitrogen traps prior to being dosed into the high-pressure cell. Infrared reflection absorption spectroscopy (IRRAS) experiments were performed in a combined UHV surface analysis chamber and elevated-pressure reactor/IRRAS cell system. The surface analysis chamber is equipped with Auger electron spectroscopy (AES). The elevated-pressure cell is coupled to a commercial Fourier transform infrared (FT-IR) spectrometer (Bruker, IFS 66v/S) for IRRAS experiments. The sample was exposed to CO by backfilling the IRRAS cell via a precision leak valve. IRRA spectra were collected at 4 cm−1 resolution using a grazing angle of approximately 85° to the surface normal. AP-XPS measurements were carried out at the Coherent Soft X-ray Scattering and Spectroscopy beamline (CSX-2) of the National Synchrotron Light Source II (NSLS-II). The main chamber (base pressure 2 × 10−9 Torr) of the end-station was equipped with a differently pumped hemispherical analyzer (Specs Phoibos 150 NAP), which was offset by 70° from the incident synchrotron light. The sample surface normal is 20° off the axis of the electron analyzer.

DFT calculations were performed using the Vienna Ab initio simulation package (VASP) [21, 22]. The consistent exchange van der Waals density functional (vdW-DF-cx) [23, 24] was used to describe the non-local vdW interactions in the silica/Ru(0001) heterojunction. The substrate of the silica/Cu/Ru(0001) system was modeled by five layers of Ru in a 2 × 2 supercell (a = 5.392 Å and b = 9.339 Å) and a 4 × 2 supercell based on different Cu concentrations. In the surface normal direction, a super cell size of c = 27 Å was used to ensure the vacuum region to be at least 10 Å thick. The kinetic energy cutoff of 800 eV was used, together with an 8 × 4 × 1 and a 4 × 4 × 1 k-point grid for the Brillouin zone sampling of the 2 × 2 and the 4 × 2 supercell, respectively. The silica film and the top two layers of the substrate were relaxed during the structure optimization until forces were smaller than 0.02 eV/Å. Harmonic vibrational frequencies of CO were calculated using the finite displacement method as implemented in VASP. All degrees of freedom of CO molecules were included with a step size of 0.015 Å.

3 Results and Discussion

Figure 1 shows Auger spectra from the sequential steps followed to deposit a small amount of Cu on an oxygen-rich 2D silica film. The green line in Fig. 1 presents the spectrum for the Ru(0001) substrate, showing the main Ru features in the energy range of 100–300 eV [25]. The black spectrum corresponds to the (2 × 2)-3O/Ru(0001) surface prepared as described in the experimental section, where the presence of O is shown at 508 eV [26]. The blue spectra corresponds to the as-deposited silica film on Ru(0001), containing fully oxidized silicon Auger peaks at 63 and 79 eV [25], attenuated Ru peaks from the substrate and the feature related to oxygen at the interface and from the SiO2 film. The bottom (red) spectrum corresponds to the sample after the Cu deposition on the silica film. To highlight the signal from the small amount of copper, the inset in Fig. 1 is used to zoom in the region at 700–1000 eV before (blue spectrum) and after (red spectrum) the Cu deposition at room temperature on the silica film. From the Auger spectroscopy, we measured the Ru/SiO2 ratio before and after the Cu deposition and we did not notice any significant difference. Based on the intensity of the observed features and using the corresponding Auger sensitivity factors of Si and Cu, the coverage of Cu atoms is calculated to be ~ 1% ML.

Fig. 1

Auger spectroscopy of (from top to bottom) Ru(0001) (green), oxygen pre-covered Ru(0001) (black), silica/O/Ru(0001) (blue) and Cu/silica/O/Ru(0001) (red). In the bottom inserted box is the zoom-in spectra of the region showing the presence of ~ 1% ML of copper

The oxidation state of Cu was investigated by in situ IRRAS, using CO as a probe molecule (Fig. 2). All spectra are taken under 1 Torr of CO. The vibrational frequency of CO is very sensitive to the adsorption site. Of particular interest in our case is the adsorption of CO on Cun+. The intense signal of CO allows differentiating oxidation states and changes in the local environment. Moreover, the permeability of CO through the nanoporous silica film and adsorption on Ru(0001) are well documented in the literature [27, 28]. The top (green) spectrum for CO adsorption on the clean Ru(0001) sample shows the two broad rotational bands from gas phase CO, and a sharp strong feature at 2067 cm−1. This is in agreement with previously reported CO adsorption spectra on clean Ru(0001) by Emmez et al. [27], where at 100 K and 10−8 mTorr of CO a signal appears first at 1995 cm−1, and then continuously shifts to 2060 cm−1 upon increasing the CO coverage until saturation is reached. The next (black) spectrum in Fig. 2 corresponds to a (2 × 2)-3O/Ru(0001) surface exposed to CO. A small peak is evident at ~ 2070 cm−1. The low CO population on Ru sites is related to the oxygen layer that blocks the adsorption of CO. This is well established in the literature, where it is reported that UHV annealing (1100 or 1275 K) results in oxygen desorption and the sites available for CO adsorption are restored [17]. The next (blue) spectrum in Fig. 2 corresponds to a Cu/(2 × 2)-3O/Ru(0001) surface exposed to CO. The Cu coverage is approximately ~ 1%, as estimated by AES. The IRRA spectrum shows a small feature at ~ 2111 cm−1. This feature is hard to distinguish since it overlaps with the rotational band due to absorption from gas phase CO, but its presence is evident when comparing with the spectrum under the same conditions but without Cu present (black spectrum in Fig. 2). This feature can thus be assigned to CO adsorbed on Cu sites.

Fig. 2

In situ IRRAS spectroscopy data under a CO pressure of 1 Torr. From top to bottom: (green) clean Ru(0001) substrate; (Black) oxygen terminated (2 × 2)-3O/Ru(0001); (blue) Cu deposited on oxygen terminated (2 × 2)-3O/Ru(0001); (red) Cu deposited on the 2D-SiO2 thin film grown on oxygen terminated (2 × 2)-3O/Ru(0001); (purple) the difference between the black and red spectrum. Lines indicating the frequencies associated in the literature with adsorption of CO on Cu2+ and Cu+ sites are included for reference

In order to aid the assignment, we include in Fig. 2 (doted lines) the peak position of features assigned to CO adsorption on Cu+ and Cu2+ sites during in situ IRRAS CO + O2 (in the presence of 30 mTorr CO with 15 mTorr O2) experiments on Cu(111) [8]. According to Xu [8] and Hollins and Pritchard [29] the peak at 2115 cm−1 can be assigned to CO adsorbed on disordered Cu2O (Cu+) surfaces, while peaks at ~ 2105 cm−1 can be assigned to CO adsorbed on under-coordinated Cu0 sites [30]. Based on these assignments, we can associate the 2111 cm−1 peak to the presence of small copper clusters with exposed Cu+ sites, although under-coordinated Cu0 cannot be fully ruled out. In addition to the aforementioned feature corresponding to CO adsorption on Cu sites, a sharp peak is observed at 2067 cm−1 (blue spectrum), coincident with the peak for CO adsorption on metallic Ru(0001) sites (green spectrum). This indicates that the deposition of Cu increases the number of exposed metallic Ru sites as compared to the (2 × 2)-3O/Ru(0001) surface (black spectrum). This implies that the deposited Cu partially reduces the (2 × 2)-3O/Ru(0001) surface leaving patches of bare Ru(0001) for CO adsorption. That oxygen “stolen” from the ruthenium surface explains the presence of partially oxidized copper clusters containing Cu+ sites. Note that the small amount of Cu on the surface would not completely account for the relatively large amount of CO adsorbed on Ru(0001) sites. This can be explained by considering that once some of the bare ruthenium has become available for CO adsorption, CO can adsorb and further reduce adjacent Ru sites.

The deposition of Cu on 2D silica films grown over a (2 × 2)-3O/Ru(0001) surface was ultimately studied. The presence of a bilayer 2D silica structure was confirmed by a characteristic phonon vibration at 1300 cm−1 in the IRRA spectrum (see Supp. Info S1). Note that the presence of this phonon using vibrational spectroscopy is currently the best way to unequivocally verify the presence of the bilayer structure. Several other possible silica structures such as 1D stripes, [31] monolayer and 3D amorphous films have very different signature peaks in the vibrational spectra [32]. The second from the bottom (red) spectrum in Fig. 2 corresponds to the system Cu/2D-SiO2/(2 × 2)-3O/Ru(0001) under 1 Torr of CO. It has been shown that CO can adsorb at the oxygen-free 2D-SiO2/Ru(0001) interface, but the addition of oxygen to the interface blocks CO adsorption. A broad weak feature is observed at ~ 2067 cm−1, indicating the presence of a small number of Ru sites available for CO adsorption when Cu is deposited onto the 2D silica/(2 × 2)-3O/Ru(0001) film. Clear changes are observed in the spectra on both sides of the deep in the spectrum associated with the rotation bands from gas phase CO at 2140 cm−1. By subtracting the spectrum related to the exposure of an O-Ru(0001) surface (second from top, black) to the same 1 Torr of CO, the gas phase contribution to the spectrum can be removed, resulting in a weak feature centered at 2145 cm−1 (bottom, purple). Based on assignments from the literature discussed above, where Cu+ and Cu2+ sites can be identified with CO IR features at 2115 and 2148 cm−1 respectively, associated with features at, this new feature at 2145 cm−1 can be associated with the formation of highly oxidized Cu clusters (Cu2+). It is important to notice that this oxidized copper clusters are stable under UHV, and required elevated CO pressures to be observed due to their weak interaction with the probe molecule. IRRA spectra as a function of increasing CO pressure is included in the supplemental information (see Supp. Info. S2). Figure S2 also shows the reversible and weak nature of the CO adsorption on oxidized Cu clusters, where removing the gas phase CO leads to a spectrum containing only a small feature associated with CO adsorption on Ru sites.

In order to determine the oxidation states of Cu of O/Ru(0001) surface, further IRRAS experiments were carried out under CO oxidation conditions for the Cu/(2 × 2)-3O/Ru(0001) system without a 2D-SiO2 film (Fig. 3). First, under 15mTorr of O2 (Fig. 3a), no peaks are observed as expected. This is followed by the addition of 30 mTorr of CO to the 15 mTorr of O2 where a peak at 2109 cm−1 appears (Fig. 3b). This peak remains constant for at least 30 min under reaction conditions (Fig. 3c), and can be associated to the presence of copper clusters containing Cu+ and Cu0 sites. No peak is observed at ~ 2067 cm−1, due to the blocking of Ru sites by the adsorption of oxygen from gas phase O2. More importantly, no formation of Cu2+ is detected as evident by the absence any changes or features ~ 2148 cm−1. The spectrum in Fig. 3d was taken after evacuation, showing no indication of CO remaining on the surface.

Fig. 3

AP-IRRAS results for exposure of CO: O2 = 2:1 ratio on the Cu deposited O2 modified Ru surface: (a) 15 mTorr of O2, (b) 15 mTorr of O2 and 30 mTorr of CO, (c) after 30 min, and (d) UHV (after evacuation of O2 and CO)

A similar experiment following CO oxidation by IRRAS was carried out after depositing Cu on the 2D-SiO2 film, namely the Cu/2D-SiO2/(2 × 2)-3O/Ru(0001) system. Figure 4a shows a UHV spectrum taken for reference before introducing gases to the system. Figure 4b was taken after exposing the surface to 10 mTorr of O2. The spectrum in Fig. 4c was taken after adding 20 mTorr of CO to the 10 mTorr of O2. A very weak feature at 2115 cm−1 could indicate the presence of Cu clusters with Cu+ sites. The clearest change in the spectrum is associated to the appearance of a peak at ~ 2148 cm−1. As described above, this can be associated to the formation of clusters with fully oxidized Cu2+ sites. Spectra 4D–4H show the evolution with time where both features become slightly more intense and are stabilized after 20 min. Spectrum 4I was taken after evacuation showing only a very small amount of CO adsorbed Ru(0001), evident by a small peak at 2076 cm−1. Spectrum 4J was taken under 50 mTorr CO and included for comparison, where there is no O2 gas present but Cu2+ is still present as evident by the 2148 cm−1 feature in the spectrum. Note that the formation of Cu2+ has been reported in literature [8], but only under the presence of O2 in addition to CO. In the study presented here, Cu2+ sites remain stable even in the absence of O2 in the Cu/2D-SiO2/(2 × 2)-3O/Ru(0001) system.

Fig. 4

AP-IRRAS results for exposure of CO: O2 = 2:1 ratio on Cu deposited Silica film: (a) UHV, (b) 10 mTorr of O2, (c) 10 mTorr of O2 and 20 mTorr of CO, (d) following time in after 4 min, (e) after 8 min, (f) after 12 min, (g) after 16 min, (h) after 20 min, (I) UHV (after evacuation of O2 and CO), and (J) 50 mTorr of CO. Lines indicating the frequencies associated in the literature with adsorption of CO on Cu2+ and Cu+ sites are included for reference

Figure 5 shows XPS core level spectra taken for 2D-SiO2/(2 × 2)-3O/Ru(0001) and Cu/2D-SiO2/(2 × 2)-3O/Ru(0001), using a photon energy of 650 eV. Figure 5a–c show O 1s, Si 2p and Ru 3d spectra respectively. Spectra under UHV conditions for 2D-SiO2/(2 × 2)-3O/Ru(0001) are shown in black and for Cu/2D-SiO2/(2 × 2)-3O/Ru(0001) in red. The blue spectra correspond to the Cu/2D-SiO2/(2 × 2)-3O/Ru(0001) surface after exposure to 1 Torr of CO. The O 1s from 2D-SiO2/(2 × 2)-3O/Ru(0001) shows a main peak at 531.7 eV corresponding to the Si–O–Si linkages in 2D-SiO2, and a shoulder peak around 529.80 eV from the (2 × 2)-3O chemisorbed layer [33]. After the Cu deposition the O 1s peak from the Si–O–Si moieties shifts to higher binding energy by ~ 0.48 eV, while the position of the O 1s peak from chemisorbed oxygen does not change. This indicates that copper reacts with some of the chemisorbed oxygen, reducing the effective coverage of O at the interface and thus inducing a shift in the O 1s peak from the 2D-SiO2. The origin of this shift has been described before by Wang et al. [34], and it is further addressed below. The Cu/2D-SiO2/(2 × 2)-3O/Ru(0001) surface was subsequently exposed to 1 Torr of CO. The blue spectra in Fig. 5a–c were taken after exposure to CO. The presence of CO increased the shift in the O 1s peak by ~ 0.64 eV compared to the 2D-SiO2/(2 × 2)-3O/Ru(0001) surface. Likewise, the Si 2p core level (Fig. 5b) shows the same shifts as the O 1s from 2D-SiO2 [34]. However, in the case described by Wang et al., the shifts are related to desorption of chemisorbed oxygen and thus there is an associated decrease in the intensity of the 529.8 eV peak. In our case, copper “steals” some of the chemisorbed oxygen from Ru(0001), inducing the shifts, but the oxygen does not desorb, so the integrated area of the ~ 529.8 eV peak does not decrease, but its binding energy shifts slightly upwards. This is in agreement with the formation of CuO and Cu2O clusters, which have O 1s binding energies at 529.9 and 530.8 eV respectively [35]. It is also consistent with the IRRAS data showed above where CO adsorption experiments suggested the presence of copper clusters with both Cu+ and Cu2+ sites. Note that in addition to the effect of copper “stealing” oxygen from Ru(0001), the presence of positively charged copper on the surface would have a similar effect as oxygen removal, and the actual observed shifts could be interpreted as a combination of these two effects. The Ru 3d region, which overlaps with C 1s, is shown in Fig. 5c. Ru 3d5/2 at 280.4 eV and Ru 3d3/2 at 284.1 eV peaks are observed. The Ru 3d5/2 peak in the black curve, from 2D-SiO2/(2 × 2)-3O/Ru(0001), can be deconvoluted into two peaks, where the lower binding energy peak is due to the surface core level shift (SCLS) of the Ru 3d states of the surface atoms relative to the bulk Ru atoms [28]. No clear difference is observed in the Ru 3d5/2 peak upon deposition of copper (Fig. 5c), while the presence of CO results in the disappearance of the SCLS peak.

Fig. 5

XPS (hv = 650 eV) O 1s (a), Si 2p (b) and Ru 3d (c) spectra for 2D-SiO2/Ru(0001) (black), Cu/2D-SiO2/Ru(0001) (red) and CO/Cu/2D-SiO2/Ru(0001) (blue)

Based on the IRRAS experiments comparing CO adsorption on Cu/O/Ru(0001) and Cu/2D-SiO2/O/Ru(0001), Cu atoms trapped in the later system have overall higher oxidation state (Cu2+) than Cu in Cu/O/Ru(0001) (Cu+). In order to explain how the presence of the silica film on Ru(0001) affects the oxidation state of Cu atoms, we performed density functional theory (DFT) studies for Cu/O/Ru(0001) with and without the silica film. We quantified oxidation states of Cu atoms using Bader charge analysis [36]. Although the Bader charge analysis tends to underestimate the oxidation state due to the charge-self regulation mechanism [37], it gives us the qualitative trend to compare with the experimental results.

It has been well established that the Cu atoms tend to cluster on the O/Ru(0001) surface [38], which could lead to lower oxidation states of Cu atoms. Figure 6a, b shows the atomic model with a Cu coverage of Θ = 1.5 (Cu3_cluster/4O/Ru) and Θ = 0.5 (Cu1/4O/Ru), and Fig. 6c corresponds to Cu1/(SiO2)8/4O/Ru with Θ = 0.5. It is important to note that Cu atoms can easily diffuse through the SiO2 film and get trapped at the SiO2/Ru(0001) interface. Θ is defined as the number of Cu atoms in the 2 × 1 supercell, which is equivalent to the number of Cu atoms per silica cage in Fig. 6c. When Cu clusters on the O/Ru(0001) surface (Cu3_cluster/4O/Ru), each Cu atom loses 0.16 e, which is the Bader charge difference (Δq) between the adsorbed Cu atom and the isolated Cu atom. However, if Cu atoms are dispersed (Cu1/4O/Ru), Δq increases to 0.35, as they lose more electrons to chemisorbed O atoms (ORu) and Ru substrate in Cu1/4O/Ru. This is a clear indication that dispersed Cu atoms can be more easily oxidized on the O/Ru(0001) surface than small Cu clusters.

Fig. 6

Side (left) and top (right) view of relaxed structures of a Cu3_cluster/4O/Ru, b Cu1/4O/Ru and c Cu1/(SiO2)8/4O/Ru. Black rectangles indicate the unit cell. Color code: Cu (purple), Ru (white), Si (yellow), O in silica films (red), and O chemisorbed on Ru(0001) (pink). Δq represents the electrons transferred from the Cu atom to the substrate after the adsorption

The silica film deposited on O/Ru(0001) introduces a spatial confinement at the interface, with a spacing of ~ 3.9 angstrom between the Ru substrate and the silica bilayer. The silica film has a mild effect on the oxidation state of Cu, as the Δq increases slightly from 0.35 in Cu1/4O/Ru to 0.41 in the Cu1/(SiO2)8/4O/Ru. Compared with Cu1/4O/Ru, the excess 0.06 electrons are transferred from Cu to ORu in Cu1/(SiO2)8/4O/Ru. Therefore, the adsorption of copper on the silica films makes Cu slightly more oxidized than on the Cu1/(SiO2)8/4O/Ru surface. The charge transfer from Cu atoms to the substrate is further confirmed by the change of the dipole moment of the system. Upon the adsorption of Cu atoms, the dipole moment of Cu1/(SiO2)8/4O/Ru becomes larger than that of (SiO2)8/4O/Ru by 0.48 eÅ. The increase of the dipole moment lowers the work function and increases the core-level binding energies of the silica film [34], which is consistent with the XPS measurements described above where the O 1s binding energy increases by 0.48 eV upon the adsorption of Cu atoms (Fig. 5a).

The effects of the ORu distribution on the oxidation states of Cu atoms were further studied. Two atomic models with Θ = 0.25 are shown in Fig. 7 to represent a possible Cu coverage after the silica film deposition. Both systems have the same amount of O coverage, which correspond to 0.5 ML in the experiment. However, the O distributions are different in these two systems. In Cu1/(SiO2)16/8Ocluster/Ru (Fig. 7a), there are O rich regions and O poor regions, which correspond to the p(1 × 1) and p(2 × 2) O coverage, respectively. Cu atoms were deposited on the O-rich regions in Cu1/(SiO2)16/8Ocluster/Ru. On the other hand, in Cu1/(SiO2)16/8Ouniform/Ru (Fig. 7b) Cu atoms were deposited on a surface with a uniform p(2 × 1) ORu coverage [39]. We found a strong dependence of the oxidation state of Cu on the number of ORu surrounding it, as Δq = 0.84 in Cu1/(SiO2)16/8Ocluster/Ru is nearly the twice of Δq = 0.46 in Cu1/(SiO2)16/8Ouniform/Ru.

Fig. 7

Side (left) and top (right) view of relaxed structures of a Cu1/(SiO2)16/8Ocluster/Ru and b Cu1/(SiO2)16/8Ouniform/Ru. Black rectangles indicate the unit cell. Color code: Cu (purple), Ru (white), Si (yellow), O in silica films (red), and O chemisorbed on Ru(0001) (pink). Δq represents the electrons transferred from one Cu atom to the substrate after the adsorption

Based on the above analysis, both Cu and ORu distributions have a strong impact on the oxidation state of the surface Cu, while the spatial confinement from the silica bilayer alone has a marginal impact. A likely explanation of the increased oxidation state for Cu with the presence of the silica bilayer is that the silica film creates a higher diffusion barrier for Cu atoms. Cu atoms may not easily pass across the edges of nano-cages at the interface. Consequently, this high diffusion barrier would prevent Cu atoms from clustering and make the distribution of Cu atoms more dispersed than on the system without the silica film.

To validate this model, we compare the Cu diffusion pathway and energy barriers on the 4O/Ru surface and at the (SiO2)8/4O/Ru interface using the climbing image nudged elastic band (CI-NEB) method [40]. The diffusion of a Cu atom from an hcp hollow site to the nearest hcp hollow sites is studied. The minimum energy path (MEP) is shown in Fig. 8. On the 4O/Ru surface, the fcc site of the Ru surface is a metastable state, while the bridge site is the transition state. The energy barrier (ΔE) for Cu diffusion is 0.20 eV. The diffusion of Cu atoms on 4O/Ru(0001) is similar to that of Pd and Pt atoms on the Ru(0001) surface, where similar MEP was reported and ΔE for Pd and Pt were 0.16 and 0.25 eV, respectively [41]. In contrast, Cu atoms need to cross the edges of the nano-cages at the (SiO2)8/4O/Ru interface to diffuse. Due to the repulsion from the O and Si atoms of the silica film, the preferred diffusion pathway does not go through the fcc site. Instead, Cu atoms move under the bond center of the O and Si atoms, which becomes the transition state. As a result, ΔE for Cu diffusion increases to 0.29 eV at the (SiO2)8/4O/Ru interface. The increase in ΔE and the change in the transition state support our argument that silica bilayer prevents surface Cu atoms from clustering, which results in a higher oxidation states of Cu atoms at the (SiO2)8/4O/Ru interface. Note that this analysis is done for a single Cu atom. If the Cu atoms react with oxygen, as expected, the diffusion barrier would be even large due to chemical and steric constraints.

Fig. 8

The minimum energy path for the diffusion of a Cu atom on the 4O/Ru surface (black) and at the (SiO2)8/4O/Ru interface (red). ΔE (in eV) represents the energy barrier, i.e. the energy difference from the initial state to the current transition or metastable state. Color code: Cu (purple), Ru (white), Si (yellow), O in silica films (red), and O chemisorbed on Ru(0001) (pink)

In order to confirm the interpretation of Cu oxidations states determined by CO vibrational frequencies from IRRAS, we did DFT calculations on the vibrational frequencies of CO adsorbed on the oxygen modified Ru(0001) surface (Δq = 0.16 e in Fig. 6a) and the silica pre-covered Ru(0001) surface (Δq = 0.84 e in Fig. 7b). A benchmark calculation was performed for a free CO gas molecule. The calculated bond length (d CO ) of a free CO gas molecule is 1.143 Å and the vibrational frequency (ν) is 2116 cm−1, which is consistent with d CO  = 1.141 Å and ν = 2106 cm−1 calculated from vdW-DF2 functional [42]. Compared with the experimental anharmonic vibrational frequency of ν = 2143 cm−1 [43], calculated ν from vdW-cx tends to be softer by 1.26%. Therefore, all the calculated vibrational frequencies for CO on Cu atoms are scaled by 1.26%. The structures with CO adsorbed on Cu3_cluster/4O/Ru and Cu1/(SiO2)16/8Ocluster/Ru are shown in Figure S3 A and Figure S3 B. The CO coverage (ΘCO) in CO/Cu3_cluster/4O/Ru is 1/3, which is defined as the number of adsorbed CO atoms per Cu atom. When CO molecules are adsorbed on Cu3_cluster/4O/Ru, ν decreases to 2099 cm−1, which is consistent with the 2109 cm−1 peak from Fig. 3. This is because of the electron back-donation from Cu to the CO 2π* anti-bonding orbitals [44], which increases the CO bond length to 1.151 Å. The frequency of ν = 2099 cm−1 on Cu3_cluster/4O/Ru is close to the experimental vibrational frequency of 2100 cm−1 for CO on under coordinated surface sites from Cu(211) [8]. It further supports our argument that the oxidation state of Cu atoms on Ru(0001) surface is a combination of metallic Cu and partially oxidized Cu1+ with a calculated Δq of 0.16 e. In CO/Cu1/(SiO2)16/8Ocluster/Ru with ΘCO = 1, due to the decrease on the Cu cluster size and the increase in the surrounding O atoms in Cu1/(SiO2)16/8Ocluster/Ru, the effective Cu oxidation increases towards Cu2+, with a corresponding to Δq of 0.84 e in the DFT calculation. Consequently, the electron back-donation from Cu to the CO 2π* orbital decreases, which decreases the CO bond length to 1.150 Å and increases the vibrational frequency to 2134 cm−1 in CO/Cu1/(SiO2)16/8Ocluster/Ru, close to the 2145 cm−1 peak observed in Fig. 4 that shows the stabilization of oxidized Cu2+ under the silica pre-covered Ru(0001) surface. Compared with Cu3_cluster/4O/Ru, the increase in ν is 35 cm−1, which is in agreement with the increase of 39 cm−1 from IRRAS.

4 Conclusions

Deposition of copper on a 2D-SiO2 film leads to the formation and stabilization of small highly oxidized Cu clusters. Without the confinement introduced by the SiO2/Ru(0001) film to adsorbed Cu atoms, Cu adsorbed on oxygen terminated Ru(0001) agglomerates and becomes only partially oxidized. We performed DFT calculations to compare the charge state of Cu atoms with and without silica films. The increased oxidation states of Cu atoms in the presence of the deposited silica films can be explained by a decrease on surface mobility. The silica bilayer prevents surface Cu atoms from clustering, and thus makes them more disperse. These disperse Cu atoms can be fully oxidized by chemisorbed oxygen and the resulting oxide clusters do not decompose under vacuum conditions.

Notes

Acknowledgements

This research used resources of the Center for Functional Nanomaterials and beamline 23-ID-2 of NSLS-II, which are U.S. DOE Office of Science Facilities, at Brookhaven National Laboratory under Contract No. DE-SC0012704. This research used resources of the National Energy Research Scientific Computing Center, a DOE Office of Science User Facility supported by the Office of Science of the U.S. Department of Energy under Contract No. DE-AC02-05CH11231.

Supplementary material

11244_2017_879_MOESM1_ESM.docx (5.3 mb)
Supplementary material 1 (DOCX 5398 KB)

References

  1. 1.
    Rodriguez JA, Liu P, Wang X, Wen W, Hanson J, Hrbek J, Perez M, Evans J (2009) Catal Today 143(1–2):45–50CrossRefGoogle Scholar
  2. 2.
    Gokhale AA, Dumesic JA, Mavrikakis M (2008) J Am Chem Soc 130(4):1402–1414CrossRefGoogle Scholar
  3. 3.
    Klier K (1982) Adv Catal 31:243–313Google Scholar
  4. 4.
    Wang W, Wang SP, Ma XB, Gong JL (2011) Chem Soc Rev 40(7):3703–3727CrossRefGoogle Scholar
  5. 5.
    Paolucci C, Khurana I, Parekh AA, Li SC, Shih AJ, Li H, Di Iorio JR, Albarracin-Caballero JD, Yezerets A, Miller JT, Delgass WN, Ribeiro FH, Schneider WF, Gounder R (2017) Science 357(6354):898CrossRefGoogle Scholar
  6. 6.
    Baber AE, Yang XF, Kim HY, Mudiyanselage K, Soldemo M, Weissenrieder J, Senanayake SD, Al-Mahboob A, Sadowski JT, Evans J, Rodriguez JA, Liu P, Hoffmann FM, Chen JGG, Stacchiola DJ (2014) Angew Chem Int Ed 53(21):5336–5340CrossRefGoogle Scholar
  7. 7.
    Eren B, Heine C, Bluhm H, Somorjai GA, Salmeron M (2015) J Am Chem Soc 137(34):11186–11190CrossRefGoogle Scholar
  8. 8.
    Xu F, Mudiyanselage K, Baber AE, Soldemo M, Weissenrieder J, White MG, Stacchiola DJ (2014) J Phys Chem C 118(29):15902–15909CrossRefGoogle Scholar
  9. 9.
    Royer S, Duprez D (2011) ChemCatChem 3(1):24–65CrossRefGoogle Scholar
  10. 10.
    Stacchiola DJ (2015) Acc Chem Res 48(7):2151–2158CrossRefGoogle Scholar
  11. 11.
    Therrien AJ, Zhang RQ, Lucci FR, Marcinkowski MD, Hensley A, McEwen JS, Sykes ECH (2016) J Phys Chem C 120(20):10879–10886CrossRefGoogle Scholar
  12. 12.
    Jiang P, Prendergast D, Borondics F, Porsgaard S, Giovanetti L, Pach E, Newberg J, Bluhm H, Besenbacher F, Salmeron M (2013) J Chem Phys 138(2):024704CrossRefGoogle Scholar
  13. 13.
    Boscoboinik JA, Yu X, Yang B, Fischer FD, Włodarczyk R, Sierka M, Shaikhutdinov S, Sauer J, Freund HJ (2012) Angew Chem Int Ed 51(24):6005–6008CrossRefGoogle Scholar
  14. 14.
    Büchner C, Lichtenstein L, Stuckenholz S, Heyde M, Ringleb F, Sterrer M, Kaden WE, Giordano L, Pacchioni G, Freund H-J (2014) J Phys Chem C 118(36):20959–20969CrossRefGoogle Scholar
  15. 15.
    Kaden WE, Büchner C, Lichtenstein L, Stuckenholz S, Ringleb F, Heyde M, Sterrer M, Freund H-J, Giordano L, Pacchioni G (2014) Phys Rev B 89(11):115436CrossRefGoogle Scholar
  16. 16.
    Baron M, Stacchiola D, Ulrich S, Nilius N, Shaikhutdinov S, Freund HJ, Martinez U, Giordano L, Pacchioni G (2008) J Phys Chem C 112(9):3405–3409CrossRefGoogle Scholar
  17. 17.
    Zhong J-Q, Wang M, Akter N, Kestell JD, Boscoboinik AM, Kim T, Stacchiola DJ, Lu D, Boscoboinik JA (2017) Nat Commun 8:16118CrossRefGoogle Scholar
  18. 18.
    Zhong J-Q, Kestell J, Waluyo I, Wilkins S, Mazzoli C, Barbour A, Kaznatcheev K, Shete M, Tsapatsis M, Boscoboinik JA (2016) J Phys Chem C 120(15):8240–8245CrossRefGoogle Scholar
  19. 19.
    Schlexer P, Pacchioni G, Włodarczyk R, Sauer J (2016) Surf Sci 648:2–9CrossRefGoogle Scholar
  20. 20.
    Yang B, Kaden WE, Yu X, Boscoboinik JA, Martynova Y, Lichtenstein L, Heyde M, Sterrer M, Włodarczyk R, Sierka M (2012) Phys Chem Chem Phys 14(32):11344–11351CrossRefGoogle Scholar
  21. 21.
    Kresse G, Furthmüller J (1996) Phys Rev B 54(16):11169CrossRefGoogle Scholar
  22. 22.
    Kresse G, Furthmüller J (1996) Comput Mater Sci 6(1):15–50CrossRefGoogle Scholar
  23. 23.
    Berland K, Hyldgaard P (2014) Phys Rev B 89(3):035412CrossRefGoogle Scholar
  24. 24.
    Bjorkman T (2014) J Chem Phys 141(7):074708CrossRefGoogle Scholar
  25. 25.
    Grant J, Haas T (1970) Surf Sci 21(1):76–85CrossRefGoogle Scholar
  26. 26.
    Büchner C, Wang Z-J, Burson KM, Willinger M-G, Heyde M, Schlögl R, Freund H-J (2016) ACS Nano 10(8):7982–7989CrossRefGoogle Scholar
  27. 27.
    Emmez E, Yang B, Shaikhutdinov S, Freund H-J (2014) J Phys Chem C 118(50):29034–29042CrossRefGoogle Scholar
  28. 28.
    Starr DE, Bluhm H (2013) Surf Sci 608:241–248CrossRefGoogle Scholar
  29. 29.
    Hollins P, Pritchard J (1983) Surf Sci 134(1):91–108CrossRefGoogle Scholar
  30. 30.
    Xu X, Vesecky S, He JW, Goodman D (1993) J Vac Sci Technol A 11(4):1930–1935CrossRefGoogle Scholar
  31. 31.
    Kaya S, Baron M, Stacchiola D, Weissenrieder J, Shaikhutdinov S, Todorova TK, Sierka M, Sauer J, Freund HJ (2007) Surf Sci 601(21):4849–4861CrossRefGoogle Scholar
  32. 32.
    Yang B, Kaden WE, Yu X, Boscoboinik JA, Martynova Y, Lichtenstein L, Heyde M, Sterrer M, Wlodarczyk R, Sierka M, Sauer J, Shaikhutdinov S, Freund H-J (2012) Phys Chem Chem Phys 14(32):11344–11351CrossRefGoogle Scholar
  33. 33.
    Włodarczyk R, Sierka M, Sauer J, Löffler D, Uhlrich JJ, Yu X, Yang B, Groot IMN, Shaikhutdinov S, Freund HJ (2012) Phys Rev B 85(8):085403CrossRefGoogle Scholar
  34. 34.
    Wang M, Zhong J-Q, Kestell J, Waluyo I, Stacchiola DJ, Boscoboinik JA, Lu D (2017) Top Catal 60(6–7):481–491Google Scholar
  35. 35.
    Tahir D, Tougaard S (2012) J Phys Condens Matter 24(17):175002CrossRefGoogle Scholar
  36. 36.
    Henkelman G, Arnaldsson A, Jónsson H (2006) Comput Mater Sci 36(3):354–360CrossRefGoogle Scholar
  37. 37.
    Raebiger H, Lany S, Zunger A (2008) Nature 453(7196):763CrossRefGoogle Scholar
  38. 38.
    Wolter H, Meinel K, Ammer C, Wandelt K, Neddermeyer H (1999) J Phys: Condens Matter 11(1):19Google Scholar
  39. 39.
    Pfnür H, Held G, Lindroos M, Menzel D (1989) Surf Sci 220(1):43–58CrossRefGoogle Scholar
  40. 40.
    Henkelman G, Uberuaga BP, Jónsson H (2000) J Chem Phys 113(22):9901–9904CrossRefGoogle Scholar
  41. 41.
    Lu Y, Sun Q, Jia Y, He P (2008) Surf Sci 602(14):2502–2507CrossRefGoogle Scholar
  42. 42.
    Ma X, Genest A, Spanu L, Rösch N (2015) Comput Theor Chem 1069:147–154CrossRefGoogle Scholar
  43. 43.
    Lide DR (2000) CRC handbook of chemistry and physics. CRC Press, Boca RatonGoogle Scholar
  44. 44.
    Kresse G, Gil A, Sautet P (2003) Phys Rev B 68(7):073401CrossRefGoogle Scholar

Copyright information

© Springer Science+Business Media, LLC, part of Springer Nature 2017

Authors and Affiliations

  1. 1.Center for Functional NanomaterialsBrookhaven National LaboratoryUptonUSA
  2. 2.Materials Science and Chemical Engineering DepartmentStony Brook UniversityStony BrookUSA
  3. 3.Chemistry DepartmentBrookhaven National LaboratoryUptonUSA

Personalised recommendations