Sm on CeO2(111): A Case for Ceria Modification via Strong Metal–Ceria Interaction
- 144 Downloads
The growth, electronic structure and stability of Sm on ordered CeO2(111) thin films grown on Cu(111) were investigated by means of X-ray photoelectron spectroscopy (XPS), low energy electron diffraction, and scanning tunneling microscopy (STM). Metallic samarium was deposited on the CeO2(111) surface by thermal evaporation under ultrahigh vacuum conditions at room temperature. The XPS data suggest that metallic Sm is oxidized to Sm3+ upon the deposition of Sm on CeO2, accompanied by the reduction of Ce4+ to Ce3+. With increasing the Sm coverage, the concentration of Ce3+ increases monotonically. After depositing 6 ML of Sm, only Ce3+ is observed within the detection depth of XPS. The STM results indicate that Sm exhibits a two-dimensional growth on the CeO2(111) surface at low coverages. Annealing to higher temperatures leads to the agglomeration of Sm particles and concurrent diffusion of Sm into the ceria film. These results illustrate that Sm can modify both the electronic and structural properties of ceria.
KeywordsCeria Samarium Model catalyst XPS STM
Metal–oxide system has been drawing intense attention in various fields, such as electrochemistry, materials, and catalysis. Specifically, oxide-supported metal catalysts have been widely used in numerous important catalytic reactions, such as CO oxidation, NO reduction, low-temperature methanol synthesis, and propene oxidation [1, 2, 3, 4]. In these reactions, it has been found that the morphologies, chemical and electronic structures, and thermal stabilities of the oxide-supported metal catalysts strongly influence their catalytic properties [5, 6, 7]. For example, compared with amorphous ceria (CeO2) support, nanocrystalline CeO2 supported gold catalyst exhibited higher catalytic performance in CO oxidation . Whereas, for ceria-supported metal catalyst, Gao et al. found that Pt0 was more active than Pt2+ on ceria in CO oxidation in excess O2, while Pt2+ was the reactive site in preferential oxidation of CO in a H2-rich gas . Thus the local surface structure of these catalysts has to be understood to establish a precise relationship between the catalyst structure and catalytic performance, which is a prerequisite for the rational design of more effective catalysts. However, metal/oxide catalysts prepared in practical systems are rather complex. Moreover, most metal oxides have poor conductivity, which notably hinders the application of electron/ion detection-based techniques in modern surface science. Such difficulties prevent the clear identification of active sites and interactions at metal–oxide interfaces for an oxide-supported metal catalyst. The use of model catalysts prepared on well-defined single crystal substrates under ultrahigh vacuum (UHV) conditions helps overcome the complexity of real catalysts and allows the use of modern spectroscopic and microscopic techniques in surface science to characterize them at a molecular or atomic level [8, 9].
CeO2 is one of the most important and widely used oxides in a variety of catalytic reactions, either as the support for active metal nanoparticles or as an active component itself, such as ethanol oxidation and low-temperature water–gas shift reaction [10, 11, 12]. To gain fundamental insights into the electronic structure and properties of ceria-related catalysts, a few nanometers thick well-ordered thin films of CeO2(111) grown on Cu(111) [13, 14, 15], Ru(0001) , and Pt(111)  single crystals were prepared and served as substrates for metal deposition to study the metal–ceria interaction. Depending on the metal property itself, the interaction of the deposited metal with ceria can be either weak or relatively strong, which results in the changes of either morphology or chemical state of ceria or both. It has been found that the late transition metals, such as Pt [18, 19], Pd [20, 21], Ni [22, 23] and Ag [24, 25], usually exhibit relatively weak interactions with ceria. Upon deposition, these metals usually form small islands on the CeO2 surface, and simultaneously reduce the ceria substrate. Whilst the deposition of more active metals like Zr [26, 27], Al , Sn [29, 30], Gd , and Mn  on ceria films can seriously reduce the ceria substrate, forming mixed metal–O–Ce interfaces or mixed ceria-based oxides.
Samarium (Sm), a rare-earth metal, can enter ceria to form a solid solution . Sm-doped ceria (SDC) shows excellent catalytic performance in many important reactions, such as benzyl alcohol oxidation and allylic oxidation of cyclohexene [34, 35]. To elucidate the origin of improved catalytic properties of SDC with a unique structure, a wide range of theoretical and experimental work has been carried out [36, 37, 38]. However, the conclusions are still a matter of debate. For instance, Xie et al. found that the presence of Sm in CeO2(111) resulted in the appearance of the oxygen weakly bound with the surrounding Sm, which reacted with CO easily . However, Kuntaiah et al. reported that Sm-incorporation could improve the CO oxidation efficiency of CeO2 owing to the large number of oxygen vacancies and MO8 complex defects and enhanced bulk oxygen mobility . Despite of these studies, mechanism of CO oxidation over Sm/CeO2 surface is far from well understood. Moreover, the relationship between the Sm–ceria interface structure and their catalytic performance such as thermal stability remain uncertain, which motivates the present study.
In this work, we focus on the fundamental understanding of the interaction between Sm and CeO2 and the thermal stability of Sm on CeO2. The well-ordered CeO2(111) thin films grown on a Cu(111) substrate were prepared as the model ceria surface. The growth, electronic properties and thermal stability of Sm on CeO2(111) were investigated by X-ray photoelectron spectroscopy (XPS) and scanning tunneling microscopy (STM), which is beneficial to design more effective Sm-doped ceria catalysts.
2 Experimental Section
The experiments were performed in two separate ultrahigh vacuum (UHV) systems. Low energy electron diffraction (LEED) and STM measurements were performed in a three-chamber system, which has been described previously , at base pressures all below 1 × 10−8 Pa. The XPS measurements were performed on the Catalysis and Surface Science Endstation in the National Synchrotron Radiation Laboratory, which consists of four UHV chambers with base pressures all in the 10−9 Pa range, as described in detail elsewhere .
The Cu(111) single crystal (8 mm diameter and 2 mm thickness) sample, purchased from MaTeck GmbH, Germany, was used as the substrate for oxide growth. The Cu(111) surface was cleaned by repeated cycles of Ar+ sputtering and then annealing in UHV at 720 K until no contaminants were detected by XPS and a sharp (1 × 1) LEED pattern was observed. The CeO2(111) films were epitaxially grown on the Cu(111) surface, adopting the procedure from Matolínova and co-workers , by depositing Ce from a water-cooling e-beam evaporator in oxygen atmosphere of 4 × 10−5 Pa at gradually increased substrate temperature. The substrate temperature increased from room temperature to 750 K at a rate of 1 K/s. After deposition, the sample was annealed in oxygen at 850 K for 10 min to achieve a better ordering. The oxygen was pumped out after the substrate cooling down to 400 K. One monolayer (ML) of the ceria film is defined as an O–Ce–O stack normal to the (111) plane of CeO2 with a thickness of 0.31 nm [16, 42]. In the present study, the CeO2(111) film was controlled to be about 3 nm thick.
Sm was evaporated onto the ceria films from a homemade evaporation source at 300 K. The deposition rate of Sm was kept at 0.04 ML/min, measured by a quartz crystal microbalance (QCM) and calibrated with STM measurements of the island volume of metallic Sm. One ML of Sm is defined as 7.9 × 1014 atoms/cm2, which is the number of oxygen atoms per unit area in the topmost atomic layer of the CeO2(111) surface . The monolayer thickness of Sm is calculated to be 0.27 nm. Chemically etched tungsten tip was used for imaging the surfaces. All the STM images were collected with a constant current mode (sample bias: 2–3 V; tunneling current: 0.01–0.05 nA) at room temperature. The STM images were processed using the WSXM program . Particle size distribution analysis was achieved by commercial software SPIP. Monochromatic Al Kα (1486.6 eV) was used to probe the Sm 3d5/2 and Ce 3d core levels. The photoelectron spectra were measured at 0° with respect to the surface normal. In order to enhance the surface sensitivity, some of Sm 3d5/2 spectra were acquired at 60° with respect to the surface normal.
3 Results and Discussion
3.1 Sm Growth on the CeO2(111) Film at Room Temperature
3.1.1 XPS Studies
The chemical state of Sm deposited on the CeO2(111) film, as well as the interaction between Sm and ceria, can be investigated by monitoring the Sm 3d and Ce 3d core-level spectra during Sm growth on ceria. Figure 1a depicts a series of Sm 3d5/2 spectra as a function of Sm coverage on the CeO2(111) surface at room temperature. For reference, a Sm 3d5/2 spectrum from a pure Sm2O3 film which was prepared by depositing Sm in oxygen environment on CeO2(111) is also shown in Fig. 1b. As can be seen, at the lowest coverage (0.10 ML), only a single peak at 1083.8 eV is observed. The binding energy of this peak is consistent with that of Sm2O3 shown in Fig. 1b, suggesting that Sm is oxidized immediately to Sm3+ upon deposition on CeO2. With increasing Sm coverage, the intensity of this peak increases monotonically and, on the other hand, a shoulder at the low binding energy side develops. After the coverage increases to 6.00 ML, the shoulder emerges as a peak exhibiting a binding energy of 1074.2 eV. For comparison, the spectrum from a bulk-like thick Sm film (θSm > 50 ML) is shown as the top trace. Both the shape and the peak position of this spectrum are in accordance with the bulk metallic Sm reported in a previous study . For the pure metallic Sm, it exists two different 4f configurations, namely, 4f6 and 4f5, which relate to metallic divalent [Sm(II)] and trivalent [Sm(III)] Sm, respectively . Moreover, it has been reported that Sm(II) and Sm(III) coexist in metallic Sm, where Sm(II) is attributed to surface Sm atoms and Sm(III) appears in the bulk [45, 46]. Therefore, the peaks locating at the binding energies of 1073.9 and 1081.6 eV observed on the bulk-like Sm film can be attributed to metallic Sm(II) and Sm(III), respectively [41, 45, 47]. At 6.00 ML, the binding energy of the shoulder is close to that of metallic Sm(II). Thus, we attribute this peak to metallic Sm(II), suggesting the presence of metallic Sm. Further increasing the Sm coverage, the peaks attributed to metallic Sm(II) and Sm(III) markedly increase, suggesting that at high coverages the deposited Sm mainly appears metallically.
Since Sm is immediately oxidized after evaporating onto the CeO2 surface, the CeO2 substrate is expected to be reduced. Towards this aim, the Ce 3d spectra acquired after each step of Sm deposited onto CeO2 are collected (Fig. 2a). Illustrated as the bottom spectrum in Fig. 2a, six peaks corresponding to three pairs of spin–orbit split doublets associated with Ce4+ are observed for pure CeO2 [48, 49], which are labeled as u, u″, u′′′, v, v″, and v′′′ locating at 901.1, 907.7, 917.0, 882.6, 889.3 and 898.6 eV, respectively. The u and v refer to the 3d3/2 and 3d5/2, respectively. After depositing 0.10 ML Sm on CeO2, four extra peaks labeled as u0, u′, v0, and v′ with binding energy of 899.8, 904.9, 881.4 and 886.5 eV, respectively, start to show up, which are typical peaks for Ce3+ states . This finding suggests that partial reduction of CeO2 occurs upon Sm deposition. A further increase in Sm coverage leads to a continuous increase in peak intensities related to Ce3+ ions, indicating a further reduction of ceria. CeO2 gets fully reduced to Ce2O3 when the Sm coverage reaches 6.00 ML, as indicated by the disappearance of the characteristic peaks related to Ce4+ ions. Above 6.00 ML, the whole Ce 3d spectrum decreases in intensity due to the deposition of metallic Sm layers on top of the film. The percentage of Ce3+ can be determined by spectral deconvolution, which typically consists of Gaussian–Lorentzian fitting after subtracting a Shirley background [30, 50]. The evolution of Ce3+/(Ce4+ + Ce3+) after each Sm deposition step is shown in Fig. 2b. It can be seen clearly that with increasing the Sm coverage, the percentage of Ce3+ increases. After 6.00 ML Sm deposition, the percentage of Ce3+ increases from 4.5% (0.00 ML) to nearly 100%, indicating that the CeO2 surface is almost fully reduced to Ce2O3 within the detection limit. Accordingly, thereafter the deposited Sm exhibits aforementioned metallic features. These results may be expected from the bulk energetics of oxide formation. Based on the oxide formation energy, △Hf298, the reaction 2Sm + 6CeO2 → Sm2O3 + 3Ce2O3 is thermodynamically favorable at room temperature by 678 kJ/mol . Thus, the charge transfer from Sm atoms to CeO2 is assumed to occur when Sm is deposited on CeO2 via the formation of the Sm–O bond. Similar phenomena have been observed in the Sm/Al2O3/Ni3Al(111) system  and other active metals supported on the CeO2(111) surface, such as Zr [26, 27], Ti , and Sn [29, 53].
3.1.2 STM studies
To further investigate the growth of Sm on the CeO2(111) surface, the morphological changes of the CeO2 film after Sm deposition at different coverages was studied by STM. Shown in Fig. 3a is the STM image of the 3 nm thick CeO2(111) grown on the Cu(111) surface. As seen, the CeO2 film has large and flat terraces that completely cover the copper substrate. The measured step height between the two neighboring terraces is 0.30 nm, which agrees with the spacing of O–Ce–O trilayers in the bulk CeO2(111) structure (0.31 nm) . As shown in the atomically resolved STM image (inset of Fig. 3a), the surface of stoichiometric CeO2 presents a well-ordered atomic structure with almost no defects. The distance between the two adjacent Ce atoms is measured to be 0.39 nm, which matches well with the expected surface spacing of 0.38 nm . The(111) surface termination of pure CeO2 is confirmed by LEED (Fig. 3g), which exhibits a sharp p(1.5 × 1.5) pattern.
Considering that in real catalysts Sm mostly acts as an additive to ceria, here only low Sm coverages deposited on the CeO2(111) surface at 300 K were examined by STM, which are shown in Fig. 3b–f. At the lowest coverage of 0.01 ML (Fig. 3b), the deposited Sm exhibits uniformly distributed islands that are 1.5 ± 0.2 nm in width and 0.15 ± 0.04 nm in height with an island density of 3.7 × 1012 cm−2. No growth preference at the step edges is observed, suggesting a strong interaction between the Sm islands and the CeO2 support. This is in contrast to the results of Ag and Au on CeO2(111), but similar to the case of Zr on CeO2(111) [24, 26, 27, 55]. Increasing the Sm coverage to 0.03, 0.05, and 0.10 ML (Fig. 3c–e) causes the continuing increase in island density. However, the island size keeps nearly the same, which is about 1.6 nm wide and 0.15 nm high. As the aspect ratio (height/diameter) is only 0.09, Sm tends to wet the CeO2 surface which also demonstrates a strong interaction between Sm and the CeO2 support. Moreover, LEED results suggest that after deposition of 0.20 ML Sm, the background intensity increases except for the weakening of original spots due to CeO2(111) (Fig. 3h). Further increasing the Sm coverage to 2.00 ML, the CeO2(111) spots were significantly weakened (Fig. 3i), suggesting that deposition of Sm onto the CeO2 surface leads to a decrease in the ordering of CeO2. The measured island sizes and densities from STM images as a function of Sm coverage are also investigated, which are plotted in Fig. 4. Between 0.01 and 0.10 ML, the size of deposited Sm islands keeps almost the same (Fig. 4a). While the island density linearly increases (Fig. 4b), which can be attributed to the sufficient exposed ceria surface for the incoming Sm atoms to nucleate at low coverages. As the coverage increases from 0.10 to 0.20 ML, the island diameter increases to 2.0 ± 0.4 nm, and the height of Sm islands exhibits no change. The island density of the deposited Sm increases less sharply because the incoming Sm atoms join the existing Sm nanoparticles, as evidenced by the increase in the average diameter. Based on these observations, it is concluded that Sm exhibits a trend for two-dimensional growth on the CeO2(111) surface at low coverages.
Considering that the deposited Sm is oxidized to Sm3+ at these coverages, the observed Sm islands may be assigned to Sm3+ nanoparticles. However, the mean height of islands is 0.15 nm, which is less than that of a complete O–Sm–O trilayer (0.29 nm) . Since Sm2O3(111) is accumulated by the repeating structure element (O–Sm–O trilayers), the islands at these coverages is proposed as a Sm termination, indicating the formation of Ce–O–Sm mixed oxide layer at the Sm–ceria interface. Such phenomenon is similar to the previously studied cases of Zr [26, 27], Al , Sn [29, 30], Ga , and Mn  interaction of CeO2, which showed the formation of metal–O–Ce linkages when these metals were deposited on the CeO2(111) surface.
3.2 Thermal stability of 0.1 ML Sm on CeO2(111) surface
To investigate the stability of Sm on ceria upon heating, we deposited 0.1 ML Sm on the CeO2(111) surface at 300 K followed by annealing to different temperatures. Figure 5 presents the STM results showing the morphological changes of Sm/CeO2 upon annealing. At 300 K, the deposited Sm forms islands that are uniformly distributed on the CeO2 surface (Fig. 5a) with average height lower than 0.2 nm, as indicated in the line scan profile (inset of Fig. 5a). Upon heating, the average size of Sm islands increases, whereas its density decreases. At 500 K, the average size of Sm islands significantly increases to 2.0 ± 0.4 nm in width and 0.19 ± 0.04 nm in height, while the island density decreases to 8.6 × 1012 cm−2 due to aggregation, as shown in Fig. 5b. Heating the surface to 650 K leads to the formation of even larger islands with an average diameter of 2.4 ± 0.4 nm and an average height of 0.25 ± 0.04 nm, accompanied with the continuing decrease in island density (3.9 × 1012 cm−2) (Fig. 5c). Further heating the Sm/CeO2 to 800 K causes the island density decreasing to 0.8 × 1012 cm−2, resulting in a large increase in uncovered ceria area. A line scan profile along one of the Sm islands displayed in the inset of Fig. 5d suggests that the height of the island is measured to be about 0.28 nm, which is consistent with the thickness of O–Sm–O trilayers along the (111) direction of the bixbyite structure of bulk c-Sm2O3 (0.29 nm) . Thus, a single O–Sm–O layer on the ceria surface is formed. As mentioned above, the average height of the Sm islands with a Sm termination is only 0.15 nm when Sm is deposited on the CeO2 surface at room temperature. The top oxygen in the single O–Sm–O layer should come from the lattice oxygen of the ceria below when annealing of Sm/ceria to 800 K. Such O2− diffusion from CeO2 to the deposited metal was also observed in Ce–CeO2(111) system, where the Ce2O3 was formed after annealing Ce/CeO2 to a higher temperature in vacuum . In addition, a range of theoretical and experimental work has proved that incorporation of Sm enhances the oxygen ion migration in ceria [36, 37, 58, 59]. Thus, the variation of Sm3+ islands thickness with the increased temperature is attributed to oxygen ions diffusion between the CeO2 substrate and the Sm3+ islands. This is further supported by the observation of Ce 3d spectra, as shown below. It should be noted that at 800 K, only ~ 25% of total deposited Sm left on the surface, suggesting that ~ 75% of Sm either diffuse into the substrate or desorb from the surface after annealing to 800 K. Due to the strong interaction between Sm3+ and the CeO2 support, desorption of Sm3+ from the CeO2(111) surface is unlikely. Therefore, the decrease in total Sm3+ amount during annealing can be most probably attributed to the diffusion of Sm into the CeO2(111) subsurface, which is further confirmed by the XPS results shown below.
To further understand the thermal behavior of Sm overlayers on the CeO2 surface, the XPS spectra of Sm 3d5/2 upon annealing, collected at two different angles with respect to the surface normal (θ), were examined. The results are shown in Fig. 6a and b. As can be seen, the binding energies of Sm 3d5/2 shift gradually to a lower side upon heating. The maximum shift reaches 0.5 eV after the temperature increases to 900 K. This shift can be attributed to the size increase and structure change of Sm3+ islands as indicated by our STM images (Fig. 5). Moreover, the intensities of Sm 3d5/2 peaks decrease upon heating to 900 K. The decrease rate is different between the Sm 3d5/2 peak intensities collected at 0° (with respect to the surface normal) and those collected at 60°. To better illustrate this difference, the integrated Sm 3d5/2 peak intensities of 0.1 ML Sm deposited on CeO2(111) collected at θ = 0° and 60° as a function of temperature are plotted in Fig. 6c. All the Sm 3d5/2 peak intensities were normalized to their own initial values. At the detection angle of 0°, the intensity of Sm 3d5/2 keeps almost constant at temperatures below 700 K, after which it starts to decrease. At 900 K, the intensity of Sm 3d5/2 decreases only ~ 5%. In contrast, when the detection angle is alternated to 60°, the Sm 3d5/2 peak intensity decrease even can be observed between 300 and 700 K. At 700 K, the Sm 3d5/2 intensity decreases by 7%. However, when the annealing temperature is higher than 700 K, a sharp decrease is clearly seen. The decrease of Sm 3d5/2 intensity reaches 17 and 34% for 800 and 900 K, respectively. Given the fact that the XPS spectra collected at 60° are more surface sensitive than those at 0° (the detection depth of the spectrum collected at 60° is only half of that collected at 0°), the difference of the decrease rates in the Sm 3d5/2 peak intensities at these two different detection angles can be clearly ascribed to the diffusion of Sm into ceria and oxygen diffusion from ceria to the top of Sm. This diffusion is also accompanied by the agglomeration of Sm2O3 into larger islands on the CeO2 surface upon annealing, as observed by STM images (Fig. 5).
The Ce 3d spectra of 0.1 ML Sm deposited on CeO2 at 300 K followed by annealing are shown in Fig. 7a. For comparison, the spectrum from clean CeO2 is also shown in the bottom in Fig. 7a. As can be seen, the characteristic peaks due to Ce4+ get attenuated upon Sm deposition, while those associated with Ce3+ develop. To quantitatively analyze the evolution of Ce chemical states upon annealing, we deconvoluted the spectra and calculated the ratio of Ce3+ at different temperatures. Figure 7b displays the evolution of Ce3+ fraction after stepwise annealing the 0.1 ML Sm on CeO2 from 300 to 900 K. As shown in Fig. 7b, the Ce3+ percentage increases from 0.034 to 0.111 after Sm deposition, which is attributed to the partial reduction of CeO2 upon Sm deposition. With increasing temperature between 300 and 600 K, the amount of Ce3+ progressively decreases. It reaches a minimum of 0.055 upon heating the surface to 600 K. Similar observation has been found for Pt/CeO2  and W/CeO2 , where they attributed this to either the migration of Ce3+ into deeper layers  or the filling of surface oxygen vacancies by bulk oxygen . In fact, the increased height of the islands of bright protrusions upon heating (Fig. 5) strongly confirms the conclusion that oxygen moves toward the surface from the bulk. Further heating the surface to temperatures above 600 K leads to the slight increase of Ce3+ fraction. The amount of Ce3+ increases to 0.078 at 900 K, which suggests the reduction of ceria. In our previous study, we found that the pure CeO2(111) film grown on Cu(111) was reduced at 900 K; below 900 K it was difficult to reduce the CeO2(111) surface . However, the presence of Sm was reported to cause the distortions of geometric and electronic structures of CeO2 surface, which facilitated the formation of oxygen vacancies [39, 62]. Therefore, it is easier to reduce ceria at lower temperatures due to the Sm presence, which increases the oxygen storage/release capacity of ceria.
The growth, electronic properties, and thermal stability of Sm on the CeO2(111) thin films grown on Cu(111) were studied by XPS and STM. The results show that oxidation/reduction processes occur at the Sm–CeO2 interface when Sm is deposited on the CeO2 surface. Sm is oxidized to Sm3+, accompanied with Ce4+ reduced to Ce3+. At low coverages, Sm particles are two-dimensional and uniformly distributed on the CeO2 surface. Annealing to higher temperatures induces Sm agglomeration, as well as Sm diffusion into the CeO2. Our study demonstrates that Sm can modify both the electronic and structural properties of ceria, which enhances the oxygen storage/release capacity of ceria.
The authors gratefully acknowledge the National Natural Science Foundation of China (Nos. U1732272, 21473178 and 21403205), National Key R&D Program of China (No. 2017YFA0403402), China Postdoctoral Science Foundation (BH2310000032), and Chinese Universities Scientific Fund (WK2310000068) for the financial support of this work.
- 46.Yang A-B (1989) Retrospective theses and dissertations. 9254. http://lib.dr.iastate.edu/rtd/9254. Accessed 1989
- 50.Skála T, Šutara F, Škoda M, Prince KC, Matolín V (2009) J Phys: Condens Matter 21(5):055005Google Scholar
- 61.Skála T, Tsud N, Prince KC, Matolín V (2011) J Phys: Condens Matter 23(21):215001Google Scholar