Citation

Wei-Jia Wang, Yan Wang, Qian Xu, Huan-Xin Ju, Tao Wang, Zhi-Jie Tao, Shan-Wei Hu, Jun-Fa Zhu. Interaction of cobalt with ceria thin films and its influence on supported Au nanoparticles[J]. Chin. Chem. Lett., 2017, 28(8): 1760-1766  

Wei-Jia Wang, Yan Wang, Qian Xu, Huan-Xin Ju, Tao Wang, Zhi-Jie Tao, Shan-Wei Hu, Jun-Fa Zhu    

Ceria, a key component for heterogeneous catalysts, is an active support for catalytic metals or multicomponent alloys and efficiently contributes to achieve desired chemical reactions [1, 2]. Ceria-supported noble metal catalysts like Rh, Pd, and Pt have attracted intensive research interests owing to their excellent activities in promoting various reactions including low-temperature CO oxidation, water-gas-shift reactions (WGS) and hydrogen production from ethanol [3-5]. These applications benefit from the unique oxygen storage capacity of ceria as well as the strong metal-support interaction (SMSI) between metals and the ceria supports [1, 6-8]. However, the high price and limited reserves of these noble metals remain as major concerns, catalysts that are cheaper and harder to denature are attractive industrial alternatives. Co-based catalyst emerges as a less expensive but promising alternative for certain reactions, for example, as a mixed oxide, in CO oxidation [9] and for steam reforming of ethanol (SRE) [10, 11]. Supported cobalt catalyst represents high ethanol conversion and selectivity to H₂ and CO₂ at relatively low temperatures [12, 13]. It has been also extensively studied for Fischer-Tropsch synthesis (FTS) because of its high activity and selectivity for linear hydrocarbons [14, 15]. Cobalt is also a doping agent for ceria, which is demonstrated its positive role in facilitating bulk oxygen diffusion in ceria [16-20].

Ceria-supported Au is an active catalyst in many applications including WGS, NO reduction, as well as CO oxidation [21-24]. Interestingly, ceria-supported gold modified with cobalt used as catalysts exhibited unique properties compared to individual Co and Au counterparts due to the synergistic effects between the two metals as well as the interaction between the metals and oxide supports [25]. The presence of Co could improve the catalytic activity of Au-based catalyst. For example, gold supported on Co₃O₄-modified ceria has been reported enhanced catalytic activities compared to conventional Au/ceria catalysts in the CO oxidation reaction [26-28]. Gold incorporation at Co₃O₄/CeO₂ catalysts have been identified as promising catalysts for WGS, which have been observed to increase the capacity to release lattice oxygen from the catalysts to oxidize CO at low temperatures in oxygen-enhanced WGS reaction (OWGS) [29]. In addition, gold catalysts supported on ceria doped with cobalt oxide could serve as an excellent catalyst in complete benzene oxidation and obtain very high and stable catalytic activity [30, 31]. Despite the importance of Co modification to Au the catalysts on ceria, however, the atomic level understanding of the influence of Co on Au catalyst is still unknown. The Co-Au bimetallic clusters on CeO_x(111) (1.5 < x ≤ 2) present a model system for understanding the nature of metal-metal interactions and metal-support interactions.

We report here the results of a systematic investigation of the interactions between Co and ceria thin film surfaces and the thermal stability of pure and bimetallic Co-Au particles on ceria films using techniques including scanning tunneling microscopy (STM), synchrotron radiation photoemission spectroscopy (SRPES) and X-ray photoelectron spectroscopy (XPS). In particular, the effect of Co on the sintering behavior of Au nanoparticles is examined. The bimetallic particles are prepared by sequential deposition of Au on Co pre-deposited ceria surfaces. Both fully oxidized CeO₂(111) and partially reduced CeO_1.82(111) thin films grown on Cu(111) are used as model of ceria supports in order to examine the effect of the degree of ceria reduction on the sintering behavior of Co-Au.

To study the interaction between Co and ceria, we first used XPS and SRPES to investigate the electronic structure evolution upon deposition of Co on ceria at 300 K at different coverages, as shown in Fig. 1a. Changes in the ceria oxidation state as a function of Co coverage on CeO₂(111) surface are monitored by the Ce 3d XPS spectra (Fig. 1b). The satellite peaks (u''', u'', u, v''', v'', v) are associated with the tetravalent Ce cations [32]. The presence of Ce³⁺ cations in the films results in four additional peaks indicated as u', u⁰, v', and v⁰ [32]. The labels u and v refer to the 3d_3/2 and 3d_5/2 spin-orbit components, respectively. The deposition of Co at 300 K induces a well detectable growth of the reduced component, which is best visible as an intensity enhancement in the valleys at ~885.0 and ~903.5 eV (Fig. 1b). By peak fitting the spectra shown in Fig. 1b, the contribution of the Ce³⁺ states can be derived at different Co coverages. As can be seen in Fig. 1c, the intensity of Ce³⁺ increases sharply below 0.3 monolayer (ML), followed by a slower increase up to θ_Co ~ 0.68 ML. This clearly indicates that an electron transfer from Co to ceria occurs upon Co deposition, especially at low coverages. Fig. 1d shows the valence band spectra for Co deposited on CeO₂(111) at different Co coverages. The main feature of the spectrum from the clean substrate (bottom curve) is the broad peak ranging from 3 to 8 eV which can be ascribed to O 2p structure [33, 34]. With increasing Co coverage, this broad peak shifts toward higher binding energy side due to the coupling of Co states. In addition, a peak located at ~1.5 eV below the Fermi level, which is attributed to Ce 4f, gradually increases in intensity [34]. This peak is only related to the reduced state Ce³⁺. The appearance of this state as well as its gradual increase in intensity indicates that the CeO₂ surface is reduced upon Co deposition, in agreement with the results from Ce 3d spectra. Moreover, the density of state [35] near E_F, which is attributed to the DOS near Fermi edge E_F of metallic Co [36], gradually increases with increasing Co coverage.

Figure 1

Download: JPG larger image

Fig. 1. (a) Ball model of Co deposited on CeO2(111) surface, red, ivory and blue balls represent oxygen, cerium, and cobalt, respectively; (b) Ce 3d XPS spectra (hv = 1486.6 eV) of 2 nm thick CeO2(111) layer before and after the deposition of different amounts of Co at 300 K; (c) the intensity ratio of Ce3+ 3d peak to that of total Ce 3d peak calculated from the fitting of spectra presented in (b) as a function of Co coverage; (d) valence band photoemission spectra of Co on CeO2(111) at different Co coverages taken at a photon energy of 170 eV; (e) Co 2p XPS spectra (hv = 1486.6 eV) for different amounts of Co deposited on CeO2(111) surface; (f) Change of work function versus Co coverage on CeO2. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

XPS spectra of the Co 2p aftereach deposition stepareplotted in Fig. 1e. According to the literature [37-40], for metallic cobalt (Co⁰), an asymmetric Co 2p_3/2 peak is observed at 778.0-778.5eV, while for Co²⁺ the Co 2p_3/2 peak locates at 780-781eV with a strong satellite at 786-787eV. The Co 2p_3/2 peak for Co³⁺ also locates at 780-781eV but with no satellite [37-40]. As can be clearly seen in the Fig. 1e, in the initial stage (θ_Co=0.15 ML) of Co deposition on CeO₂(111), the appearance of a Co 2p_3/2 peak at 781.0eV with a strong satellite at 786.1eV can be assigned to Co²⁺. The shoulder at 778.6eV is attributed to metallic Co. Moreover, at 0.15 ML, the doublet separation between Co 2p_3/2 peak at 781.0eV and Co 2p_1/2 peak at 796.9eV is 15.9eV, which fits well with previous results on Co²⁺ [7]. For metallic Co, the doublet separation between Co 2p_3/2 peak at 778.6eV and Co 2p_1/2 peak at 793.7eV is 15.1eV, similar to the value reported previously for metallic Co (15eV) [7]. As the Co coverage increases, both the Co²⁺ and the metallic Co peaks gain intensity, but the latter becomes dominant in the Co 2p region at higher Co coverages (Fig. 1e). The oxidation of Co upon deposition is in good agreement with the aforementioned Ce 3d result. Fig. 1f shows the relative work function changes of CeO₂(111) upon Co deposition at 300K. After deposited 0.15 ML Co, the work function decreases by 0.35eV due to the electron transfer from Co to ceria. Further deposition of Co induces the increase of work function. This can be attributed to the appearance of metallic Co on the surface. Finally at high Co coverages (>2.5 ML), the absolute work function reaches 5.75eV, in close agreement with the value reported for metallic Co (5.76eV) [41]. Overall, the work function measurements further confirm that at low Co coverages the Co is partially oxidized, in consistent with the result from the Co 2p observations.

Fig. 2a-c shows the STM results of three different coverages of Co (0.21, 0.27, and 0.30 ML) deposited on CeO₂ at 300K. The corresponding particle diameters, heights, and densities are plotted in Fig. 2e-f. As can be seen, Co grows randomly on the CeO₂ surface for the whole coverage range investigated. A representative STM image for the low coverage (θ_Co=0.21 ML) is shown in Fig. 2a: the Co clusters appear as bright spots on top of the long-range modulated ceria film and are uniformly distributed on the surfaces. The apparent height and average diameter of the Co clusters amount to be about 0.14 ±0.06nm and 0.99 ±0.21nm, as inferred from Fig. 2d and e, respectively. When increasing the Co coverage to 0.27 ML, the island density increases but the size of Co clusters keep almost the same, which is about 1.00 ± 0.20nm in diameter and 0.14 ±0.05nm in height (Fig. 2b). Further increasing the Co coverage to 0.30 ML (Fig. 2c) leads to the average size of Co clusters increasing to 1.07 ±0.26nm in diameter and 0.18 ±0.07nm in height. On the other hand, Co clusters exhibit a monotonic increase in cluster density during the whole deposition (Fig. 2f). On the basis of these observations, we conclude that Co follows a 2D-like growth mode on CeO₂(111) surface at the submonolayer coverage.

Figure 2

	Download: JPG larger image
Fig. 2. STM images of Co deposited on CeO2 with three different coverages at 300K: (a) 0.21 ML, (b) 0.27 ML, (c) 0.30 ML. All images are 100nm ×100nm. (d) Mean diameter, (e) mean height, and (f) density of Co particles on CeO2 surface at different Co coverages

Detailed STM studies allow to directly extract the information of thermally induced morphological changes of the Co nanoparticles on CeO₂(111). Shown in Fig. 3 is the STM results for 0.3 ML Co deposited on CeO₂(111) at 300K followed by annealing to 800K in three steps. As seen in the Fig. 3a, at 300K relatively small Co particles with an average width of about 1.07 ± 0.26nm and a height of 0.18 ±0.07nm are formed on the ceria surface. Upon heating the surface to 500K, larger clusters with an average diameter of ~1.52 ±0.42nm and a height of ~0.26 ± 0.07nm emerge accompanied by the decrease of cluster density (Fig. 3b). Further annealing to 600K, the number of clusters on the surface continues to decrease while the cluster size keeps almost the same, which is about 1.44 ± 0.39 nm in diameter and 0.23 ± 0.07 nm in height (Fig. 3c). Eventually, as shown in Fig. 3d, upon heating to 800 K, only a few relatively small Co particles can be found on the surface. Heating to high temperatures also results in a strong intensity loss of Co 2p XPS spectra (Fig. S2 in Supporting information). For Co/CeO₂(111) surfaces, the diffusion of Co²⁺ into the subsurface of CeO₂(111) coexisting with the agglomeration of Co⁰ upon heating to higher temperatures has been reported in the previous study of Vari et al. [7]. Co could even diffuse through the ceria layer deep into the Cu crystal [7]. However, the possibility of formation of CoCu alloy can be ruled out [42].

Figure 3

	Download: JPG larger image
Fig. 3. STM images of 0.3 ML Co deposited on CeO2(111) at 300 K (a) and after annealing to 500 (b), 600 (c) and 800 K (d), respectively. All images are 100 nm × 100 nm. Plots of mean particle height and diameter (e) and particle density (f) of 0.3 ML Co deposited on CeO2(111) as a function of annealing temperatures

In order to explore the influence of pre-deposited Co on the stability of Au nanoparticles on ceria, we prepared Co-Au bimetallic clusters on ceria by depositing more mobile Au atoms (Θ_Au = 0.5 ML) on less mobile Co (Θ_Co = 0.8 ML) pre-covered CeO₂(111). Compositional changes in the bimetallic clusters are investigated by XPS after subsequently step-by-step annealed to high temperatures in UHV (annealing time = 5 min/step). At each annealing temperature, all XPS signal intensities are normalized to that at room temperature so that changes in intensity can be observed on the same scale. For comparison, studies of the thermal stability of pure Au nanoparticles on ceria were performed first. In addition, we also studied the influence of ceria stoichiometry on the Au thermal stability by depositing 0.5 ML Au on 2 nm thick CeO₂(111) and CeO_1.82(111) surfaces at 300 K followed by annealing to different temperatures in UHV for 5 min. After the surface was cooled down to room temperature, the Au 4f XPS spectra were collected. Shown in Fig. 4a are the Au 4f peak intensities normalized to the value at 300 K as a function of temperature. As can be seen, with increasing annealing temperature, the Au 4f intensity for Au on both CeO₂(111) and CeO_1.82(111) decreases at a slower rate below 600 K but after that drops rapidly. However, in general the Au 4f intensity for the Au/CeO₂ system decreases more rapidly than what is observed for the Au/CeO_1.82 system at the same temperature. This is attributed to the oxygen vacancies on the reduced CeO_2-x surface which can stabilize the Au particles through a stronger interaction [43]. When comparing the normalized Au 4f intensity of the pure Au clusters on ceria with those of bimetallic Co-Au clusters on the same ceria surface, one can see that below 700 K the decrease of Au 4f intensity from the Co-Au/CeO₂ surface follows almost identically with that of pure Au on CeO₂ at each annealing temperature (Fig. 4b). However, above 700 K, the Au 4f intensity of the pure Au cluster drops more rapidly than that of the bimetallic cluster. This implies that the predeposited Co on CeO₂ suppresses the sintering of post-deposited Au particles at high temperatures in comparison with that of pure Au particles. Interestingly, at the same annealing temperature on CeO_1.82(111), the Au 4f intensity of the bimetallic Co-Au clusters drops more quickly than that of the pure Au clusters in the whole temperature range (Fig. 4c). A possible explanation is that the inward diffusion of Co is more hindered on the partially reduced ceria thin film (Fig. S2), making Au more easily encapsulated in the presence of surface Co or diffusing into the bulk of bimetallic clusters upon annealing, similar to the case of Pt-Co bimetallic clusters on the titania support [44].

Figure 4

	Download: JPG larger image
Fig. 4. Au 4f intensities as a function of annealing temperature for (a) pure Au on CeO2 and CeO1.82; (b) pure Au, and Co-Au clusters on CeO2; (c) pure Au, and Co-Au clusters on CeO1.82. At each temperature, the samples were annealed for 5 min and then cooled down to 300 K for collecting the spectra. All the intensities were normalized to those at 300 K

The XPS results presented above suggest that the addition of Co to the Au/CeO₂ surface significantly affects the thermal stability of Au at elevated temperatures in comparison with that of pure Au particles. In order to further prove the information gained from XPS, an STM study during step-wise annealing of the sample should provide more straightforward structural/morphological information. The deposition of 0.25 ML Co on CeO₂ results in clusters with average diameter and heights of 1.17 ± 0.23 and 0.17 ± 0.07 nm and a cluster density of 2.7 ×10¹³/cm² (Fig. 5a). Subsequent deposition of 0.18 ML Au on this Co pre-covered CeO₂ surface leads to only a small (~4%) increase in cluster density up to 2.8 × 10¹³/cm² (Fig. 5b), indicating that the majority of Au atoms nucleate at existing Co sites. This is consistent with the observed increase in average cluster diameter (1.23 ± 0.26 nm) and height (0.22 ± 0.08 nm) as the incoming Au atoms are incorporated into Co clusters to form bimetallic clusters. In the absence of Co clusters, the growth of 0.18 ML Au on CeO₂ (Fig. S3) results in much larger clusters (diameter: 1.13 ± 0.36 nm, height: 0.35 ± 0.11 nm) with a lower cluster density (1.0 ×10¹³/cm²), providing further evidence that Au nucleates at the existing Co clusters in Fig. 5b.Upon heating, the particles aggregate to form larger ones at the expense of the particle density. The particle density decreases by 66% when the surface was annealed to 500 K (Fig. 5c). The mean particle sizes are 1.26 ± 0.30 nm wide and 0.36 ± 0.11 nm high. Shown in Fig. 5d is the STM image taken after the surface was further heated to 600 K. Obviously, the increase in the particle size but decrease in the particle density was observed. Further annealing to 800 K, the particle density decreases by 89% (Fig. 5e). The particles have an average diameter and height of 1.65 ± 0.36 and 0.47 ± 0.16 nm, respectively. For comparison, same amount of pure Au deposited on CeO₂(111) at 300 K followed by heating to 800 K is shown in Fig. 5f. Clearly, the pure Au particles undergo more severe sintering on CeO₂(111) than the Co-Au particles, indicating that the Co pre-deposited on ceria surface can improve the thermal stability of gold particles on the ceria surface, in good agreement with those XPS results. Based on the STM results and the aforementioned XPS results, a schematic representation of sintering behavior of Au on ceria and Co pre-deposited ceria surfaces is shown in Fig. S5.

Figure 5

	Download: JPG larger image
Fig. 5. STM images of (a) 0.25 ML Co deposited on CeO2(111) at room temperature; (b) 0.18 ML Au on CeO2(111) pre-deposited with 0.25 ML Co; and subsequently annealed to (c) 500 K; (d) 600 K; (e) 800 K; (f) 0.18 ML Au deposited on CeO2(111) at 300 K following by annealing to 800 K. All images are 100 nm × 100 nm and recorded at room temperature

The interaction of Co with ultrathin ceria thin films and its influence on the thermal stability of Au particles on ceria were investigated at the atomic level. After small amount of cobalt is deposited on CeO₂ at 300 K, Co is oxidized to Co²⁺ and simultaneously the ceria substrate is partially reduced. With increasing the Co coverage, XPS results suggest an increasing fraction of metallic Co⁰. At high temperatures, diffusion of Co²⁺ ions into the CeO₂ thin film is found, accompanied with the agglomeration of metallic Co0 into larger particles. Annealing of Au particles on CeO₂(111) leads to significant sintering of Au small particles into large ones. Oxygen defects on ceria can enhance the thermal stability of Au particles on CeO_1.82(111). The Co-Au bimetallic clusters were prepared on both stoichiometric CeO₂(111) and reduced CeO_1.82(111) by deposition of Au on Co pre-covered surfaces. Comparing with the bare ceria surface, on the stoichiometric CeO₂(111) surface, the existing of Co can improve the thermal stability of Au particles at high temperature, while on the reduced CeO_1.82(111) surface, the thermal stability of Au in Co-Au particles is even lower than that for the pure Au particles on CeO_1.82.

All of the experiments were performed in two separate ultrahigh vacuum (UHV) systems. STM measurements were performed in a system that comprises three chambers with base pressures all below 8×10^-11Torr. The SRPES and XPS measurements were performed at the Catalysis and Surface Science Endstation at the BL11U beamline in the National Synchrotron Radiation Laboratory. The beamline is connected to an undulator and equipped with two gratings that offer soft X-rays from 20 to 600eV with a resolution (E/ΔE) better than 104 at 29eV. Detailed descriptions of these two systems could be found in our previous works [45, 46]. The valence-band spectra were measured by using synchrotronradiationlight with photon energyof 170eV. A sample bias of -10V was applied in order to observe the secondary electron cutoff.

The Cu(111) single crystal (Mateck, 8mm diameter, 2mm thickness) was cleaned using several cycles of Ar⁺ ion sputtering and annealing until no impurity could be detected by XPS, and LEED gave a sharp (1×1) pattern with low background. The 2nm thick well-ordered CeO₂(111) thin films were grown on the Cu(111) substrate at gradually increased substrate temperature by depositing Ce from a multi-pocket electron beam evaporator in the 2×10^-7mbar of oxygen environment followed by annealing at 1000K for 5min. Our previousstudy has provedthat theceria films grown by this recipe are fully oxidized and well-ordered, exhibiting large and flat terraces with few surface defects [47]. The partially reduced ceria can be obtained by decreasing the oxygen pressure during the Ce evaporation.

Au was deposited from a gold foil (Alfa Aesar, 99.999%) using a knudsen-cell evaporator, while Co was deposited from a homemade source using Co wire (Alfa Aesar, 99.994%). The Au and Co flux were calibrated with a quartz crystal microbalance (QCM) before each deposition. A coverage of one monolayer (ML) of Co or Au is defined as the packing density of the Co(0001) (1.83 ×10¹⁵ atoms/cm²) or Au(111) (1.40 ×10¹⁵atoms/cm²) surface, respectively [25]. The deposition rates were 0.2 ML/min for Co and 0.2ML/min for Au.

All the STM images were collected using an etched tungsten tip at room temperature in a constant current mode (0.01-0.05nA, 3-4V). During the STM experiments, great efforts were made to minimize the tip convolution effect. The STM images were processed using the WSXM program [48]. Particle size distribution (PSD) analysis was achieved by commercial software SPIP.

The authors gratefully acknowledge the National Natural Science Foundation of China (Nos. 21473178 and 21403205) and National Basic Research Program of China (No. 2013CB834605) for the financial support of this work.

Supplementary data associatedwith this article can be found, in the online version, at10.1016/j.cclet.2017.04.012.