A perfectly stoichiometric and flat CeO2(111) surface on a bulk-like ceria film

In surface science and model catalysis, cerium oxide (ceria) is mostly grown as an ultra-thin film on a metal substrate in the ultra-high vacuum to understand fundamental mechanisms involved in diverse surface chemistry processes. However, such ultra-thin films do not have the contribution of a bulk ceria underneath, which is currently discussed to have a high impact on in particular surface redox processes. Here, we present a fully oxidized ceria thick film (180 nm) with a perfectly stoichiometric CeO2(111) surface exhibiting exceptionally large, atomically flat terraces. The film is well-suited for ceria model studies as well as a perfect substitute for CeO2 bulk material.


SAMPLE PREPARATION
The ceria lms are annealed in air in a cylindrical type T7 furnace based on the Chevenard-Joumier system (ADAMEL, Paris, France). The oven has two ends left open to guarantee a constant ow of fresh air during annealing. The sample being almost entirely enclosed by the heating oven walls ensures a homogeneous heating and a precise temperature measurement by a K-type thermocouple placed next to the sample. However, due to the air ow inside the furnace the surface temperature is lower than measured explaining that the temperatures reported here are slightly higher than the ones reported previously [1]. Due to the large thermal mass of the furnace, the target temperature is reached after about 10 hours ( Fig. S1(a)), so that heating and cooling of the sample is always very slow. At a temperature of 700 K, the heating rate is below 3 K/min and decreases at higher temperatures to values between 2 to 0.5 K/min. The cooling rate is below 5 K/min and decreases afterwards.
After annealing in air, the sample is xed by thin tantalum wires spot welded to a tantalum plate compatible with the sample transfer system of the UHV NC-AFM (Omicron NanoTechnology, Taunusstein, Germany). XPS and NC-AFM experiments are conducted in two separate ultra-high vacuum (UHV) chambers allowing for further annealing the samples either in UHV or in an oxygen atmosphere (10 −6 to 2 × 10 −5 mbar oxygen partial pressure). For annealing in the NC-AFM chamber, a home-made cylindrical furnace is used as described earlier [2]. The furnace heats the sample mainly by radiative heat transfer and a K-type thermocouple placed inside the furnace without direct contact to the sample is used for precise temperature measurements.
For XPS measurements, similarly prepared samples are transferred into the XPS chamber. For annealing, the sample is heated by a combined radiant and electron bombardment system (Meca 2000, Vinci-Technologies, Nanterre, France) in the XPS chamber. The temperature is measured with a pyrometer (IMPAC IGA 5 MB20, LumaSense Technologies, Santa Clara, USA) through a quartz UHV window.
Typical values for the speed of heating and cooling in the NC-AFM and XPS chambers are 10 to 20 K/min ( Fig. S1(b)). For the oxidation experiments, each vacuum chamber is back-lled with molecular oxygen (NC- Figure S1. Typical temperature vs. time curves representing the heating and cooling parts during the annealing of the ceria lms. (a) Annealing of the sample in air with a furnace with a heating rate below 3 K/min for T ≈ 700 K and 0.5 to 2 K/min for T > 700 K. The cooling rate is below 5 K/min for T ≈ 800 K. (b) Annealing of a sample in the UHV furnace of the NC-AFM chamber with heating and cooling rates of 18 and 11 K/min, respectively. AFM: 99,995 % purity, Linde MINICAN, Munich, Germany, XPS: 99.998% purity, Air Liquide, Paris, France) via a leakage valve while constantly monitoring the pressure.

AFM AND NC-AFM MEASUREMENTS
The air-annealed samples are imaged in the tapping mode either with a XE 100 AFM (Park Scientic, Suwon, Korea) or with a Nanoscope Multimode III (Bruker, Billerica MA, USA). AFM cantilevers are type NSC15 (µMash, Soa, Bulgaria) having a resonance frequency of 325 kHz and a stiness of 46 N/m. NC-AFM experiments are conducted with a RT-AFM/STM system (Omicron NanoTechnology, Taunusstein, Germany) in a UHV chamber maintaining a base pressure in the low 10 −10 mbar range. Conducting silicon cantilevers (NanoWorld AG, Neuchatel, Switzerland) are used, which have either a frequency of 70 kHz (type: PPP-QFMR) or between 280 and 320 kHz (type: PPP-QNCHR). The instrument is operated in the frequency modulation mode with a peak-to-peak oscillation amplitude of around 10 nm where frequency demodulation is accomplished by an EasyPLL demodulator (NanoSurf, Liestal, Switzerland). All images shown here represented the surface topography as the tip-sample interaction strength is kept constant during imaging by a feedback loop adjusting the tip-sample distance accordingly. Simultaneously to NC-AFM measurements, Kelvin probe force microscopy (KPFM) is performed in the frequency modulation mode [3], where a DC and an AC voltage (U AC between 0.5 and 1.0 V peak-to-peak amplitude) with a frequency of f AC = 470 Hz are applied between tip and surface. On insulating surfaces, the distribution of charges and dipoles on the surface is measured [4,5]. All topography NC-AFM images shown in this work are acquired with the KPFM compensation being active what is important for extracting reliable height information from topography images [6].
XPS MEASUREMENTS X-Ray photoelectron spectroscopy is accomplished with Mg K α radiation (1253.6 eV) delivered by an unmonochromatized X-ray source (PSP Vacuum Technology, Maccleseld, UK) operated at 200 W. Photoelectron spectra are recorded with a hemispherical analyzer Resolve 120 with 5-channel detection (PSP Vacuum Technology, Maccleseld, UK), using a pass energy of 50 eV for the survey spectrum and 20 eV for detail spectra. All XPS spectra are taken with the sample kept at room temperature.
To determine the contribution of contaminants like hydroxyls and carbon as well as the degree of oxidation, the C1s, Ce3d and O1s peaks of the XPS spectra taken after dierent lm preparation steps are analyzed. After removing the inelastic background, each spectrum is normalized by its own integrated intensity to remove experimental variations due to slight changes in the samplesource distance that could vary from experiment to experiment.
The C1s spectrum (Fig. 3(c)) is normalized to the emission in the entire Ce3d area. The Ce3d multiplet ( Fig. 3(a)) is tted using 3 doublets representing the Ce 4+ state (shaded area in red) and two doublets for Ce 3+ (shaded area in blue), as described in Refs. [7,8].
The relative contribution of the Ce 3+ doublet is calculated as 100 * (Ce 3+ / (Ce 3+ + Ce 4+ )) providing a measure for the reduction state [9]. Note that some of the ts do not perfectly match the experimental spectra, which is due to the background subtraction that was dicult to accomplish. Furthermore, we have decided to keep the multiplet parameters (shape, rel. intensities, positions, widths, etc.) as given by Refs. [7,8]. We estimate the error to be around ±5% on the Ce 3+ concentration.
The O1s spectrum (Fig. 3(b)) is normalized to the corresponding Ce3d integrated intensities, which we assume to be constant over the thermal treatments -the O1s normalized intensity should reect the changes in the oxygen surface stoichiometry with respect to the cerium concentration. The O1s spectrum is tted with four Gaussian functions centered at 530.2 eV (peak 1), 531.4 eV (peak 2), 532.6 eV (peak 3) and 533.3 eV (peak 4). We assign peaks 1 and 2 to oxygen bound to Ce 4+ (O-Ce 4+ ) and Ce 3+ (O-Ce 3+ ), respectively, representing structural oxygen of CeO 2 and Ce 2 O 3 [10,11]. Peak 3 centered at 532.6 eV is assigned to hydroxyls or H 2 O molecules bound to Ce 3+ [12,13] whereas peak 4 is assigned to SiO 2 [14].