Epitaxial growth of ultrathin ZrO2(111) films on Pt(111)

AbstractOrdered epitaxial ZrO2 films were grown on Pt(111) and characterized by low energy electron diffraction (LEED), synchrotron radiation photoemission spectroscopy (SRPES) and X-ray photoelectron spectroscopy (XPS). The films were prepared by vapor deposition of zirconium in an O2 atmosphere followed by annealing under ultra high vacuum. At low coverages, the films grew as discontinuous two-dimentional islands with ordered structures. The size and structure of these islands were dependent on the coverage of ZrO2 films. At coverage <0.5 monolayer (ML), $$
\left( {\sqrt {19} \times \sqrt {19} } \right)
$$ R23.4° and (5×5) structures coexisted on the surface. As the coverage increased, the $$
\left( {\sqrt {19} \times \sqrt {19} } \right)
$$ R23.4° structure developed with increasing degree of long-range order, while the (5×5) structure gradually faded. When the coverage reached >6 ML, a continuous ZrO2(111) film was formed with a (1×1) surface LEED pattern coexisting with a (2×2) pattern. These ordered thin ZrO2 films provide good model surfaces of bulk ZrO2 and can be used for further fundamental studies of the surface chemistry of ZrO2 using modern surface science techniques.

ZrO 2 is an excellent support in the field of catalysis and has been successfully applied in many catalytic reaction systems because of its ideal chemical and mechanical stability [1][2][3]. However, fundamental studies on the roles of ZrO 2 in these reactions using modern surface science techniques are often hampered by charging problems because of the poor electrical conductivity of bulk ZrO 2 ; the band gap of ZrO 2 is approximately 5 eV [4]. In order to overcome this difficulty and further investigate the catalytic properties of ZrO 2 , we adopt the commonly used method of epitaxial growth of an ordered thin oxide film on a metal single crystal surface to prepare ordered ZrO 2 thin films.
The epitaxial growth of thin ZrO 2 films has been reported on Au(111) [5], Cu(111) [6], Ag(100) [7] and Pt(111) [8][9][10][11]. A number of surface science techniques including *Corresponding author (email: jfzhu@ustc.edu.cn) low energy electron diffraction (LEED), scanning tunneling microscope (STM), ion scattering spectroscopy (ISS) and Auger electron spectroscopy (AES) have been used to monitor the growth of ZrO 2 films and characterize their structures. Because of the large crystal lattice mismatch, the quality and structure of ZrO 2 films varied with the substrates and were very sensitive to the substrate temperature. Maurice et al. [8] reported that ordered ZrO 2 (111) films with cubic CaF 2 -like structures were formed on clean Pt(111) by depositing metallic Zr at room temperature in an oxygen atmosphere followed by annealing to 900 K. They used AES and ISS to characterize the ZrO 2 films and found that ZrO 2 growth on Pt(111) followed Frank-van der Merwe mechanism (i.e., layer-by-layer growth). However, following the same preparation procedure, STM analysis revealed that these films had hillock or discontinuous morphology [9]. After increasing the temperature of deposition up to 470 K and post-annealing at 950 K, ordered continuous films with large terraces were obtained when the ZrO 2 thickness was >4 monolayer (ML). However, these films become discontinuous again when annealed above 1000 K.
In this paper, we investigated ordered epitaxial ZrO 2 thin films on Pt(111) surfaces using synchrotron radiation photoemission spectroscopy (SRPES), X-ray photoelectron spectroscopy (XPS) and LEED. Because of the tunability of the photon energies, SRPES can provide very surface sensitive structure information of ZrO 2 on Pt(111). Our results prove that better quality films can be obtained with a slightly higher substrate temperature for deposition than those reported in the literature [8][9][10][11].

Experimental
All the experiments were carried out on the surface physics endstation at beamline U18 in the National Synchrotron Radiation Laboratory (NSRL) in Hefei, China. The endstation equipment has been described in detail previously [12,13]. Briefly, the U18 beamline is connected to a bending magnet and equipped with three gratings that cover photon energies from 10 to 250 eV with a resolving power (E/ΔE) > 1000. The endstation consists of chambers for analysis, preparation, and molecular beam epitaxy (MBE), with base pressures of 5×10 -11 , 2×10 -10 and 1×10 -10 Torr, respectively. The analysis chamber was equipped with a VG ARUPS10 electron energy analyzer for SRPES and XPS, a twin-anode X-ray gun, retractable four-grid optics for low energy electron diffraction (LEED), and an Ar + sputter gun. The preparation chamber was connected with a quick load-lock port and housed a small hemispherical electron energy analyzer, an electronic gun for AES, and an Ar + sputter gun. The MBE chamber included a quartz crystal microbalance (QCM) for monitoring the deposition rates, a high-energy electron gun for reflection high energy electron diffraction (RHEED), and several K-cell evaporation sources.
The Pt(111) substrate (approximately 10 mm diameter × 1 mm thick) was purchased from MaTeck Company, Germany. It was cleaned, until no impurities could be detected by XPS and a sharp Pt(111)-(1×1) LEED pattern was obtained, by several cycles of Ar + -ion sputtering, annealing at 800 K in an O 2 environment to remove carbon, and then flashing to 1300 K to remove oxygen [14]. The sample temperature was monitored by a calibrated infrared pyrometer. High-purity metallic Zr (99.99%) was evaporated at approximately 0.12 ML/min using a simple custom-built electron beam evaporator. The thin ZrO 2 films were prepared by evaporating metallic Zr onto Pt(111) under an oxygen partial pressure of 1×10 -7 Torr at about 550 K, followed by post-annealing under the same conditions for 10 min and further annealing to approximately 1000 K without oxygen for 1 min. The annealing temperature was con-trolled to be no higher than 1000 K to avoid diffusion of Zr into the Pt substrate and the decay of the continuity of the film [8][9][10][11].
Taking both the photoionization cross section and grating efficiency into account, a 165 eV photon energy was used for the Pt 4f core level spectra. At this energy, the emitted electrons from Pt 4f are very surface sensitive. The valence band spectra were excited using 70 eV radiation. Before any quantitative analysis, the synchrotron radiation photoemission spectra were normalized to the photocurrents, which were measured by an Au grid located directly before the endstation. Because of the limitation of the beamline photon energy, the O 1s and Zr 3d spectra were excited using a normal X-ray source (Al Kα radiation). All the binding energies were referenced to the Fermi level of the clean Pt(111). Figure 1 shows a group of thickness-dependent LEED patterns of the ordered epitaxial ZrO 2 films on Pt(111). On the clean Pt(111) surface, a sharp LEED pattern was observed (Figure 1(a)). After deposition of ZrO 2 , the intensities of the original Pt(111) spots decreased and the background intensity increased. In addition, some new spots emerged. At a coverage of 0.4 ML, many new spots were observed around the first-order Pt(111) spots ( Figure 1(b)). The same number of spots around the (0,0) spot of Pt(111) were also observed after rotating the sample so that the (0,0) spot could be seen. Analysis of this LEED pattern indicates that it corresponds to the coexistence of ( ) 19 19 × R23.4° and (5×5) structures (the latter structure was marked with white circles, as seen in the inset of Figure 1(b)). The other missing spots of these two structures are due to absence of long-range order. Meinel et al. [11] observed the (5×5) structure at a ZrO 2 coverage of 0.7 ML by STM, but did not observe it in LEED and also did not observe the ( ) 19

Low energy electron diffraction
(enlargement in white rectangle) shows new spots induced by ZrO 2 around the original Pt(111) spots. Among these new spots, the spots that belong to the (5×5) structure are marked with white circles, while the rest represents the ( ) which is confirmed by SRPES (see below) and also supported by Meinel et al. [9]. Maurice et al. [8] observed a similar structure after annealing a fresh monolayer film, which was prepared at room temperature, at above 1100 K for several minutes with an oxygen partial pressure of 5×10 -7 Torr. They assigned the structure to a for the missing spots due to lack of long-range order. Consequently, we assigned this structure to the ( ) 19 (Figure 1(d)). The same pattern was reported by Meinel et al. [11] after annealing a 5 ML ZrO 2 film in an O 2 atmosphere at 950 K for 1 min. This provides an additional indication that under our preparation condition, the films grow better in terms of 2D dispersion. The ( ) 19 19 × structure is perceivable until the film thickness reaches about 6 ML, in agreement with previous results [11]. When the film is thicker than 6 ML, the surface structure evolves into a non-rotated hexagonal p(1×1) of ZrO 2 (111) structure that coexists with a (2×2) pattern. In the 8.3 ML ZrO 2 (111) film (Figure 1(e)) the spots of the Pt(111) substrate were no longer visible, which indicates that the films are continuous and cover the whole Pt(111) surface. Meinel et al. [9][10][11] and Maurice et al. [8] both observed the same structure with films prepared by different methods. Meinel et al. [9][10][11] obtained this surface structure after annealing a 4 ML ZrO 2 film at 980 K or a 10 ML ZrO 2 film at 950 K in 10 -6 Torr of O 2 , while Maurice et al. [8] achieved it by annealing the fresh films, which were grown at 300 K, with thickness ranging from sub-monolayer to multilayer to temperatures between 900 K and 1100 K in 5×10 -7 Torr of O 2 . According to Meinel et al. [11], charging effects appeared when the film thickness was >7 ML, which results in a significant increase in background intensity in LEED images as observed in Figure 1(e). Considering that in the reciprocal space the nearest distance of the two spots in the Pt(111) LEED pattern corresponds to the distance between neighboring Pt atoms of the Pt crystal in real space (0.28 nm), the real space unit vectors of ZrO 2 (111)-(1×1) films were calculated to be 0.359 nm from Figure 1(e). This result is in close agreement with the value of 0.36 nm reported by Meinel et al. [9][10][11]. Our calculated value is equivalent to the (111) planes of face centered cubic ZrO 2 with a CaF 2 -like structure [9,15]. Because of the approximately 28% lattice misfit between ZrO 2 (111) and Pt(111), the (5×5) superstructure of Pt(111) observed at low coverage corresponds to the (4×4) superstructure of ZrO 2 (111), as described previously [11]. The ( ) 19 19 × superstructure of Pt(111) at medium coverage can be ascribed to the ( ) 2 3 2 3 × superstructure of ZrO 2 (111) [11]. For thick ZrO 2 films, the (2×2) structure of ZrO 2 (111) has been assigned to a ( ) reconstruction of the Pt(111) interface [11,16]. These LEED observations clearly demonstrate that the epitaxial ZrO 2 films are ordered but discontinuous at low coverage and become continuous at high coverage. and O 1s, respectively. We attributed the peak shifts to surface charging effects because of the poor conductivity, as mentioned above. The charging effects could be expected to create shifts of the same magnitude in the spectra. However, this is true only if the surfaces are uniformly exposed to the same flux and energy of incident X-rays or electrons. In the present work, the Pt 4f spectra were acquired with a phonon energy of 165 eV from synchrotron light source, different from the Zr 3d and O 1s spectra which were obtained using Al Kα (hν = 1486.6 eV) irradiation. Moreover, the Pt was directly connected to the ground. Therefore, it is expected that Pt 4f should have little or even no peak shift, while Zr 3d and O 1s should have peak shifts. The fact that Zr 3d has 0.2 eV less binding energy shift than that of O 1s may indicate that for thick ZrO 2 films Zr is probably not fully oxidized. Note that the electron inelastic mean free path (IMAP) of Pt 4f 7/2 electrons with a kinetic energy of ~94 eV passing through ZrO 2 films is about 0.66 nm [18]. In other words, if ZrO 2 grows uniformly as 2D islands up to ~6 ML (1.98 nm), the Pt 4f 7/2 signal should be damped by 95%. However, the intensity of the Pt 4f peak at 6.0 ML remains approximately 10%. This indicates that at high coverages ZrO 2 films may grow discontinuously as 3D islands before the underlying 2D islands completely cover the Pt(111) surface, i.e., as a hillock mode of growth, in line with previous results observed by STM [9][10][11]. Figure 3 shows the valence band spectra of ZrO 2 /Pt(111) with different ZrO 2 thicknesses measured with a photon   [19]. The feature at ~8.0 eV is ascribed to the Zr 4d states [20]. When the thickness of ZrO 2 reached >6.0 ML, the valence bands of Pt could still be seen, but their intensities were considerably attenuated. In contrast, the feature between 4-10 eV appeared similar to that of bulk ZrO 2 . After depositing 8.3 ML ZrO 2 , the valence bands of Pt disappeared completely and the bulk ZrO 2 valence band feature was observed, which is indicative of the formation of bulk ZrO 2 on the surface. At the photon energy of 70 eV, the photoionization section of O 2p is three times larger than that of Zr 4d. Therefore, the valence band of the ZrO 2 films is dominated by the O 2p character, and only a small Zr 4d contribution can be seen, which is similar to other oxides like HfO 2 [21]. With an excitation energy of 70 eV the electrons from the valence band of Pt have kinetic energies that fall into the valley region of the universal curve, and they represent the surface sensitive information. Considering the IMAP of these electrons is only approximately 0.6 nm [18], the observation of a clear valence band for Pt at ZrO 2 thickness >4.5 ML (1.48 nm) again suggests the formation of 3D ZrO 2 islands before they coalesce into 2D uniform films to cover the substrate Pt(111) surface completely. Only at higher coverage does the deposited ZrO 2 become a continuous film on Pt(111). These valence band results agree with the aforementioned Pt 4f and LEED results.

Photoelectron spectroscopy
From these results, it is clear that compared with the film obtained by Meinel et al. [9][10][11] the ZrO 2 film grows better by 2D dispersion under our preparation conditions, especially at low coverage values. Obviously, the preparation conditions affect the growth of ZrO 2 film. Meinel et al. [9] grew ZrO 2 films at substrate temperatures of 300 and 470 K, and found that an increased substrate temperature improved the film quality. In addition, the post-annealing temperature was also critical to the film growth. When the ZrO 2 films grown at 470 K were post-annealed in oxygen environments at 950 K, STM analysis revealed smoother film morphologies than after post-annealing at 900 K [9]. However, as discussed previously [9], increasing the post-annealing temperature to about 1050 K did not yield the desired completely smooth film, but instead induced discontinuity of the films. Based on these investigations, we hypothesized that further increases in the substrate temperature during film growth would improve the film quality. Consistent with this hypothesis, increases in the Pt(111) sample temperature to 550 K resulted in larger 2D ZrO 2 islands. Therefore, the substrate temperature is an important parameter for ZrO 2 film growth on Pt(111). However, when the substrate temperature was increased to be >600 K for the film growth, the ZrO 2 superstructure spots in the LEED images became blurred (results not shown), which indicates the film quality is reduced. This is probably caused by ZrO 2 film agglomeration on Pt(111) because the interaction between the film and substrate is relatively weak, as found in DFT calculations [10,11].
Meinel et al. [10] observed a growing carbon signal in the low ZrO 2 coverage region when the film was discontinuous. They claimed that this growing C contamination was not observed for continuous ZrO 2 film. However, in the present study, we did not observe carbon contamination during the entire process of ZrO 2 growth. This may also be partially responsible for a better quality of the films presented here compared with those of Meinel et al.

Conclusions
Ordered epitaxial ZrO 2 (111) films were grown on Pt(111) with sub-monolayer to multilayer coverage. These films were of better quality and grown using a slightly different preparation method than those previously reported [8][9][10][11]. The surface structures of the ordered ZrO 2 films varied with the film thickness. At low coverages, the films were discontinuous, and the degree of long-range order increased as the film thickness increased. At high coverage, a continuous fcc crystalline structure similar to bulk ZrO 2 formed with the (111) plane parallel to the substrate Pt(111) plane. These ordered thin ZrO 2 films could be used as model surfaces for further studies of the surface properties of bulk ZrO 2 .