A model study of ceria–Pt electrocatalysts: stability, redox properties and hydrogen intercalation †

The electrocatalytic properties of advanced metal-oxide catalysts are often related to a synergistic interplay between multiple active catalyst phases. The structure and chemical nature of these active phases are typically established under reaction conditions, i.e. upon interaction of the catalyst with the electrolyte. Here, we present a fundamental surface science (scanning tunneling microscopy, X-ray photoelectron spectroscopy, and low-energy electron diﬀraction) and electrochemical (cyclic voltammetry) study of CeO 2 (111) nanoislands on Pt(111) in blank alkaline electrolyte (0.1 M KOH) in a potential window between (cid:2) 0.05 and 0.9 V RHE . We observe a size-and preparation-dependent behavior. Large ceria nanoislands prepared at high temperatures exhibit stable redox behavior with Ce 3+ /Ce 4+ electrooxidation/reduction limited to the surface only. In contrast, ceria nanoislands, smaller than B 5 nm prepared at a lower temperature, undergo conversion into a fully hydrated phase with Ce 3+ /Ce 4+ redox transitions, which are extended to the subsurface region. While the formation of adsorbed OH species on Pt depends strongly on the ceria coverage, the formation of adsorbed H ads on Pt is independent of the ceria coverage. We assign this observation to intercalation of H ads at the Pt/ceria interface. The intercalated H ads cannot participate in the hydrogen evolution reaction, resulting in the moderation of this reaction by ceria nanoparticles on Pt.


Introduction
Understanding the catalytic properties of advanced electrocatalysts for energy conversion and storage requires detailed characterization of the catalyst-electrolyte interface under operating conditions.Upon contact with the electrolyte at operating potential, the as-prepared electrocatalysts undergo structural and chemical changes yielding active catalyst phases. 1,2These changes may include the hydration and hydroxylation of the catalyst surfaces, [3][4][5] leaching of electrochemically unstable species, 6,7 or structural transformations. 8,9In specific cases, electrochemical activities of a broad spectrum of the asprepared catalysts can be traced back to a considerably narrower range of electrochemically active phases.A representative example is the formation of transition metal oxyhydroxides on the surface of various transition metal compounds during water electrolysis. 10][13] The complex nature of real high-area electrocatalysts 1,2 and their interaction with electrolytes underlines the importance of model approaches which utilize well-defined model systems.5][16][17][18] The importance of transition metalbased catalysts for alkaline water electrolysis is fueling current interest in model electrocatalytic studies predominantly on Co, Fe, and Ni oxides and oxyhydroxides.Model studies reveal the formation of active hydroxylated phases, [19][20][21] the stability of electrocatalyst surfaces under operating conditions, [22][23][24][25] and activation of electrocatalysts via doping [26][27][28] or bifunctional interactions in the presence of noble metals. 19,29ere, we present a model electrocatalytic study on a specific class of catalytically relevant oxides -rare earth oxides, as represented by cerium oxide (ceria).Apart from its wide application in heterogeneous catalysis, 30,31 ceria finds promising applications in electrocatalysis mainly as a co-catalyst for activation of ammonia, alcohol, and CO electrooxidation, [32][33][34] and for reactions related to electrolysis and fuel cells -the hydrogen evolution reaction (HER), 35 oxygen reduction reaction (ORR), [36][37][38] and oxygen evolution reaction (OER). 39Ceria-based electrocatalysts are generally synthesized in the form of cerium oxide [32][33][34][35][36][37][38] or cerium hydroxide 39 nanoparticles, interfacing nanoparticles of a second catalytically active phasemetal [32][33][34][35][36][37][38] or metal oxyhydroxide. 39The heterogeneous interface between the catalytically active phases is, correspondingly, identified as the source of improvement of ceria-based electrocatalyst properties via charge transfer between a metal and ceria, 32,36,37,39 enhanced water dissociation and OH supply, 33,35 or via formation of specific defects in ceria. 37,38The enhancement of the ceria-based electrocatalyst properties can be codetermined by the size and structure of the ceria nanoparticles. 32,38All the above-mentioned synergistic effects of ceria in ceria-based electrocatalysts must critically depend on the nature of the cerium phases established under electrochemical conditions.These, however, have remained largely unknown so far.
To obtain elementary information on the nature of the cerium phases under electrochemical conditions, we investigate the stability and electrochemical reduction/oxidation properties of CeO 2 (111) nanoislands prepared on Pt(111) in alkaline electrolyte (0.1 M KOH).The properties of the as-prepared and electrochemically treated ceria/Pt model catalysts are quantified using experimental methods of surface science -scanning tunneling microscopy (STM), X-ray photoelectron spectroscopy (XPS), low-energy electron diffraction (LEED), and atomic force microscopy (AFM), in combination with cyclic voltammetry (CV).We use different preparation methods to obtain samples with varying lateral sizes of ceria nanoislands.We collect evidence for hydration of ceria nanoislands smaller than 5 nm, and the corresponding surface or subsurface Ce 3+ /Ce 4+ electrooxidation/reduction of large CeO 2 (111) and small CeO 2 Á 2H 2 O nanoislands, respectively.An evaluation of ceria coverage, in comparison with the charges assigned to Ce 3+ /Ce 4+ transition and to H and OH adsorption/desorption on Pt(111), indicates reversible intercalation of adsorbed H ads at the ceria/Pt interface.The intercalated H ads is unable to participate in the hydrogen evolution reaction (HER).

Results and discussion
Preparation and characterization of the samples by surface science methods Samples of ceria-Pt electrocatalysts were prepared by means of physical vapor deposition in the form of inverse model catalysts consisting of CeO 2 (111) nanoislands supported on Pt(111).The samples were prepared using two methods which yielded distinctly different CeO 2 (111) nanoisland morphologies.The methods differed in thermal treatments during deposition and post-deposition, and are denoted as high-temperature (HT, final annealing temperature 1050 K) and lowtemperature (LT, final annealing temperature 850 K) methods.Detailed description of the experimental procedures is given in the Materials and methods section of the ESI.† Samples with the amounts of deposited CeO 2 of approximately 1, 2, and 3 monolayers (ML) were prepared using both methods.STM images of the as-prepared samples are shown in Fig. 1.In the following discussion, 1 ML represents a basic O-Ce-O repeat unit of CeO 2 (111) containing 7.9 Â 10 14 cm À2 Ce atoms, twice that of O atoms, and featuring a thickness of 3.1 Å. 30,31 HT samples (Fig. 1a-c) exhibit a flat geometry and compact island shapes with straight edges aligned along high-symmetry substrate directions.With increasing amount of deposited CeO 2 (111), the nanoislands extend laterally and coalesce to an almost continuous CeO 2 (111) thin film for 3 ML of deposited This journal is © the Owner Societies 2024 ceria (Fig. 1c).The growth of the ceria nanoislands resembles a layer-by-layer growth and is consistent with previous observations of CeO 2 (111) growth on Pt(111). 40,41The LEED diffraction pattern shows diffraction spots of the clean Pt(111) substrate and unreconstructed CeO 2 (111) (Fig. S1a, ESI †).The LT samples (Fig. 1d-f) exhibit a similarly flat geometry but with distinctly smaller ceria nanoisland sizes compared to HT samples (Fig. 1d).With increasing amount of deposited ceria, coalescence yields open island shapes with rounded edges (Fig. 1e), and coalescence proceeds at a slower rate as compared to the HT samples (Fig. 1f).The round edges of the LT samples are expected to contain an increased amount of kink sites compared to the straight step edges of HT samples.Similar to HT samples, LT samples show a LEED pattern corresponding to CeO 2 (111) on Pt(111) (Fig. S1b, ESI †).The effect of lowering the final annealing temperature on the morphology of LT samples is similar to previous observations of CeO 2 (111) growth on Pt(111). 40,41ased on the STM images, we can quantify the evolution of morphological features of the HT and the LT samples as a function of the amount of deposited ceria.The results are summarized in Fig. S2 and Table S1 (ESI †).For both the HT and the LT samples, the coverage of the CeO 2 (111) layer as determined by STM (y Ce STM ) scales linearly with the amount of deposited ceria, yielding a thickness of the CeO 2 (111) nanoislands (4 AE 1) ML for all samples shown in Fig. 1.A distinct difference between the HT and the LT samples is the different lateral size of the CeO 2 (111) structures, as identified by autocorrelation length obtained from the STM images (Fig. S2a, ESI †).The autocorrelation length of the LT samples is almost constant, reflecting the characteristic width of the CeO 2 (111) structures (B5 nm), much smaller than the characteristic lateral size of HT CeO 2 (111) nanoislands (B20 nm).The HT and the LT samples differ in the length of the metal-oxide boundary.For HT samples, the metal-oxide boundary amounts to 1-3% of the Pt(111) surface sites, while for the LT samples, the value is 5-7% (Fig. S2b, ESI †).
In addition to morphological differences, XPS reveals differences in the chemical state of the as-prepared HT and LT samples.This, particularly the concentration of Ce 3+ species, is determined from the Ce 3d spectrum (Fig. S2c, ESI †).HT and LT samples show Ce 3+ concentrations in the range of 4-8% and 12-16%, respectively.This corresponds, formally, to 1-2% and 3-4% O vacancy concentration relative to stoichiometric CeO 2 for HT and LT samples, respectively.The increased concentration of Ce 3+ in the LT samples can be attributed to an increased concentration of structural defects (step edges, kinks) and oxygen vacancies as compared to the HT samples.

Electrochemical characterization of the samples
After preparation and characterization by surface science methods in an ultra-high vacuum (UHV), the ceria/Pt model catalysts were transferred through air into an electrochemical cell and characterized by CV in alkaline electrolyte (0.1 M KOH electrolyte, purged with Ar).All measurements were performed at potentials ranging from À0.05 to 0.9 V RHE [potential versus the reversible hydrogen electrode (RHE)], and at a scan rate of 400 mV s À1 .An overview of the CV results for the HT and LT samples is presented in Fig. 2 and 3.
In Fig. 2, the CV of clean Pt(111) is presented for comparison.A detailed view of the clean Pt(111) reference as obtained in our experiments is presented in Fig. S3 (ESI †).The CV consists of three distinct regions: The hydrogen region (H desorption/adsorption) between À0.05 and 0.35 V RHE , the double layer region between 0.35 and 0.65 V RHE , and the oxygen region (OH adsorption/desorption) between 0.65 and 0.90 V RHE .The total integrated charges associated with these regions q ref H/Pt = 350 mC cm À2 , q ref DL =125 mC cm À2 , and q ref OH/Pt = 305 mC cm À2 for the hydrogen, double layer, and oxygen region, respectively, are in good agreement with values reported in the literature. 15ig. 2a and b show CVs of HT samples shown in Fig. 1a-c, and CVs of LT samples shown in Fig. 1 d-f, respectively.Both figures illustrate the dependence of the CV shape on the amount of deposited CeO 2 .Compared to the CV of clean Pt(111), CVs of HT samples include a new pair of irreversible peaks positioned in the hydrogen region.CVs of LT samples include a new pair of irreversible peaks as well, however, it is positioned at higher potential (by approx.0.3 V) in the double layer region of clean Pt(111).The features observable on Pt(111), corresponding to H desorption/adsorption, the double layer, and the OH adsorption/desorption, are clearly resolved for all samples.Qualitatively, for both the HT and the LT samples, the areas of the irreversible peaks increase with increasing amount of deposited ceria, while areas of the OH adsorption/desorption on Pt(111) decrease.The area of H desorption/adsorption on Pt(111) does not show apparent changes.
CVs in Fig. 2 represent stabilized CVs obtained upon immersion of the sample in the electrolyte and application of approx.25 cycles.CVs obtained immediately after the immersion of the samples in the electrolyte are shown in Fig. 3. Fig. 3a illustrates that the CVs of the HT samples are stable from the first complete cycle, and remain stable on the time scale of the present experiments (B10 min contact of the samples with the electrolyte, B50 CV cycles).In contrast, distinct changes are observable in the shape of the CVs of the LT samples.Initially, irreversible peaks similar to the peaks of HT samples appear in the H adsorption/desorption region of Pt(111).These peaks, however, are quickly replaced by a pair of irreversible peaks localized at higher potentials.CVs with irreversible peaks in the double layer region stabilize within B20 cycles.

Assignment of the CV peaks
The irreversible peaks in the CVs of both HT and LT samples can be assigned to particular Ce phases as follows: At pH = 13, corresponding to the 0.1 M KOH electrolyte, the solubility of Ce 3+ ions is negligible 42 and electrochemical phase transitions are expected to occur between hydroxylated phases, and, in the presence of oxygen (or traces of oxygen, such as in the Ar-saturated electrolytes), oxidized solid ceria phases. 43n chemical compounds, Ce adopts exclusively Ce 3+ and Ce 4+ oxidation states, 30 and Ce 3+ and Ce 4+ oxidation states are also the only states considered in aqueous electrolytes under potential. 43he observed irreversible peaks in both HT and LT CVs thus correspond to Ce 3+ /Ce 4+ redox transitions.At pH =13, and potentials 40.7 V RHE (highest-potential irreversible peak in our observations, Fig. 2b and d), Ce 4+ containing phases must dominate. 43he observed shift of the irreversible peaks by B0.3 V between HT and LT samples however indicates that the Ce 4+ containing phases in the HT and LT samples are distinctly different.
5][46] In the presence of water, the reduced cubic phase CeO 2Àx can become partly hydroxylated. 31,45,478][49][50] Finally, for highly dispersed CeO 2 powders, the amorphous hydrated phase with a stoichiometry of CeO 2 Á2H 2 O, also denoted as Ce(OH) 4 , has been identified. 45,51,52Given the properties of the known Ce phases, the HT and LT samples at oxidizing potentials 40.7 V RHE are expected to consist of cubic CeO 2 or amorphous CeO 2 Á2H 2 O phases, respectively.
The assignment of the redox transitions on the CeO 2 phase of the HT samples to Ce 3+ /Ce 4+ can be verified by recording a CV of a continuous thin film of CeO 2 (111) on Pt(111) (see Fig. S4, ESI †).The CV of the continuous film does not show the peaks characteristic of clean Pt(111) and is dominated by a pair of irreversible peaks at potentials comparable to those observed on the HT samples (cf.Fig. 2).The CV of the continuous film corresponds well to previous observations on polycrystalline CeO 2 films. 53Assignment of redox transitions observed on the LT samples to the Ce 3+ /Ce 4+ couple of hydrated CeO 2 Á2H 2 O is supported by following indirect evidence: The initial instability of the CV on LT samples can be assigned to a progressive hydration of the as-prepared CeO 2 (111) nanoislands during cycling.Similar as for the case of unsupported ceria nanoparticles, 45,51,52 hydration is observed only for ceria nanoislands with sizes smaller than B5 nm, i.e. for the LT samples.Finally, the mutual shift of the irreversible peaks in the CVs of HT and LT samples corresponds well to phase transitions predicted for ceria phases in water-based electrolytes, where the redox potential of hydrated ceria CeO 2 ÁxH 2 O " Ce(OH) 3 at pH = 13 is found to be B0.3V higher than the one of anhydrous ceria CeO 2 " Ce(OH) 3 . 43

Characterization of samples after electrochemical treatment
After finishing the CV experiments, the samples were taken out of the electrolyte under potential control at Z0.75 V RHE , ensuring that the oxidized character of both the LT and the HT samples is preserved.Samples were further examined via AFM in air and via XPS.AFM images of the samples indicate that the coverage, the flat character, and the compact shape of ceria nanoislands in HT samples are conserved during CV experiments (Fig. S5a-d, ESI †).For the LT samples (Fig. S5e-h, ESI †), the morphology of ceria and the distribution of ceria on the substrate are altered.These observations are in agreement with the electrochemical stability observed for the as-prepared HT samples (Fig. 3a), and with the transformation to hydrated CeO 2 Á2H 2 O observed for the LT samples (Fig. 3b).A more detailed analysis of the AFM images is hampered by high noise level and This journal is © the Owner Societies 2024 sample contamination during the manipulation of samples in air.XPS analysis of Ce 3d spectra on the HT samples after the CV experiments shows no changes with respect to the as-prepared state (Fig. S6a and b, ESI †), indicating a low concentration of structural defects and oxygen vacancies, supporting our assumption that the HT samples are stable under electrochemical conditions.In contrast, we observed reduction of ceria after the CV experiments for the LT samples (increase of Ce 3+ fraction from 15 to 23%, Fig. S6c and d, ESI †).Reduction is observed in spite of several hour exposure of LT samples to air under ambient conditions prior to the XPS measurement.This may reflect an increased disorder or a partly hydroxylated nature of the LT samples after finishing the CV experiments.Both in air, and in the water-based solutions, amorphous nanoparticles of hydrated CeO 2 Á2 H 2 O were previously reported to convert spontaneously to a defect-rich, highly reduced cubic CeO 2Àx phase. 45,51Poor control over water exposure and carbon contamination during the manipulation of model catalyst samples in air did not allow a systematic investigation of the O 1s and C 1s XPS signals after the CV experiments.

Faradaic charges in the CV experiments of the HT and LT samples
We attempted to determine the charges related to the faradaic processes in the CV of the HT and LT samples, in particular the charge corresponding to Ce 3+ /Ce 4+ transition q Ce , the charge corresponding to H adsorption/desorption on Pt(111) q H/Pt , and the charge corresponding to OH adsorption/desorption on Pt(111) q OH/Pt .Our estimate of the charges is based on the following CV features characteristic for both HT and LT samples: (i) The presence of ceria nanoislands blocks the OH adsorption on the Pt(111) surface which results in decreasing q OH/Pt .This effect is clearly visible in HT samples (Fig. 2a), where the OH adsorption/desorption peaks do not overlap with the Ce 3+ / Ce 4+ redox peaks.This effect is apparent in LT samples as well (Fig. 2b), where OH adsorption/desorption and Ce 3+ /Ce 4+ redox peaks are still clearly separated in the cathodic scan (Fig. 2b).In the following discussion, we determine the electrochemically active area of the ceria nanoislands y Ce EC [and the complementary electrochemically active area of clean Pt(111), y Pt EC = 1 À y Ce EC ] from the attenuation of the OH adsorption/desorption peak on the HT and LT samples with respect to the reference CV of clean Pt(111), and the corresponding charge q ref OH/Pt (Fig. S3, ESI †).y Ce EC and y Pt EC are evaluated as fractions of the clean Pt(111) surface.(ii) The presence of ceria nanoislands does not have a pronounced effect on the H adsorption/desorption and q H/Pt in the CV of HT and LT samples, relative to the reference CV of clean Pt.This effect is clearly visible on LT samples (Fig. 2b), where the H adsorption/desorption peaks do not overlap with Ce 3+ /Ce 4+ redox peaks.This effect is apparent in HT samples as well (Fig. 2a), where the Ce 3+ /Ce 4+ redox peaks seem superimposed on the H adsorption/desorption region of clean Pt(111).In the following discussion, we assume q H/Pt to be constant for all HT and LT samples, and equal to q ref H/Pt determined on clean Pt(111) (Fig. S3, ESI †).
(iii) The presence of ceria nanoislands does not influence significantly the double layer charge q DL .In the CVs of HT samples, the double layer regions of HT samples show no or small broadening relative to clean Pt(111) (Fig. 2a, Fig. S7a-c, ESI †).For LT samples, the double layer region is occupied by broad Ce 3+ /Ce 4+ redox peaks (Fig. 2b, Fig. S7d-f, ESI †).For the 1.0 ML LT sample, it is however apparent that the double layer region is not significantly broadened compared to clean Pt(111) either (Fig. S7d, ESI †).Since q DL is the smallest charge contribution, we assume q DL to be constant for all HT and LT samples, and equal to q ref DL determined on clean Pt(111) (Fig. S3, ESI †).
Considering points (i)-(iii), the charge corresponding to Ce 3+ /Ce 4+ redox peaks in the HT and LT samples can be determined as q Ce = q exp tot À q ref DL À q ref H/Pt À (1 À y Ce EC )Áq ref OH/Pt , where q exp tot is the total charge experimentally determined in the CV of HT or LT samples, y Ce EC is the electrochemically active area of the ceria nanoislands determined from the attenuation of the OH peak in the CV of the HT and LT samples, and q ref DL , q ref H/Pt , and q ref OH/Pt are the reference charges determined for a clean Pt(111) surface.Determining q Ce for HT and LT samples is illustrated graphically in Fig. 2c, d  The electrochemically active CeO 2 area y Ce EC for all HT and LT samples is shown in Fig. 4, and compared to the geometrical coverage of CeO 2 nanoislands y Ce STM determined from STM images of the as-prepared samples (Fig. 1, Table S1, ESI †).y Ce EC and y Ce STM are plotted as a function of the amount of deposited CeO 2 .We observe that y Ce STM of both the HT and the LT samples, and y Ce EC of HT samples scale linearly with the amount of CeO 2 .y Ce EC of LT samples is nonlinear with respect to the amount of CeO 2 .y Ce EC of the LT samples is also systematically higher than y Ce EC of HT samples.These observations can be again assigned to the instability and redistribution of ceria nanoislands in LT samples during electrochemical cycling (Fig. 3b and Fig. S5e-h, ESI †), and, potentially, to the increase in volume of the nanoislands due to hydration of CeO 2 to CeO 2 Á 2H 2 O. Another systematic trend can be observed in the y Ce EC of HT samples.In spite of the electrochemical and morphological stability of the HT samples (Fig. 3a and Fig. S5a-d, ESI †), y Ce EC is significantly lower than y Ce STM .In this case, inhomogeneity of the samples represents the most likely explanation.
An evaluation of q Ce and other charge contributions as a function of the amount of the deposited CeO 2 is shown in Fig. 5a for HT samples, and 5b for LT samples.q exp tot in all samples is, within AE10%, equal to q tot on the clean Pt(111).The charge contribution q Ce increases in all samples monotonically with the amount of deposited ceria.The HT samples show a smaller q Ce value as compared to the LT samples.To estimate a bulk or a surface character of the Ce 3+ /Ce 4+ redox transition, we compare the evaluated q Ce for HT and LT samples to the total charge density available for the Ce 3+ /Ce 4+ redox reaction in the bulk and on the surface of the as-prepared samples.The total charge density available for the Ce 3+ /Ce 4+ redox reaction in the bulk of the as-prepared samples is obtained as a product of Ce atom density in 1 ML of CeO 2 (111) (7.9 Â 10 14 cm À2 ), and the amount of deposited Ce, and it is indicated by the dashed line shown in Fig. 5a and b.Given the average thickness of ceria island (4 AE 1) ML for all as-prepared HT and LT samples (Table S1, ESI †), the total charge density available on the surface of the as-prepared samples is estimated as 1/4 of the   1 and 2 as a function of the amount of deposited CeO 2 .Total charge in a CV (q exp tot ), and the faradaic contribution to Ce 3+ /Ce 4+ electrooxidation/reduction (q Ce ) are marked with full symbols, incremental contributions to the total charge from H adsorption/desorption on Pt(111) (q H/Pt ), OH adsorption/desorption on Pt(111) (q OH/Pt ), and the double layer (q DL ) are marked with open symbols.The red crosses indicate the amount of H adsorbed/desorbed on Pt(111), which intercalates at the ceria/Pt interface (H/Ce-Pt).The color shading/hatching of the charge contributions represents a guide to the eyes.The dotted lines represent the total charge in the CV of clean Pt(111) (q Pt tot ), and charge required for a complete oxidation/reduction of all deposited Ce atoms (q Ce-bulk This journal is © the Owner Societies 2024 charge density available in the bulk, and it is indicated by a dotted line shown in Fig. 5a and b.For HT samples, q Ce corresponds well to the charge available at the surface of CeO 2 (111), indicating the surface character of the Ce 3+ /Ce 4+ redox transition.For LT samples, the Ce 3+ /Ce 4+ redox transition involves more Ce atoms than originally available on the surface of the as-prepared LT samples.This indicates that the Ce 3+ /Ce 4+ redox transition in LT samples involves subsurface Ce atoms as well.These subsurface Ce atoms might have been exposed during redistribution of the as-prepared CeO 2 nanoislands in LT samples.

Hydrogen intercalation at the ceria-Pt interface
CVs of ceria nanoislands in both the HT and the LT samples presented in this study differ qualitatively from CVs reported previously for transition metal inverse model catalysts, i.e. nanoislands of transition metal oxyhydroxides on Pt(111). 54,55articularly, in contrast to ceria nanoislands, transition metal oxyhydroxide nanoislands on Pt(111) block both the OH adsorption/desorption and H adsorption/desorption on Pt(111). 54,55This different behavior is also reproduced in our experimental setup.Transition metal ions represent a trace contaminant in commercial ultrapure KOH, and the deposition of transition metal oxyhydroxide nanoparticles can be observed on Pt(111) upon extended electrochemical cycling. 56The CV of Pt(111) upon such extended cycling is displayed in Fig. S8 (ESI †).The CV shows peaks assigned to transition metal oxyhydroxides, and the corresponding attenuation of both OH adsorption/desorption and H adsorption/desorption on Pt(111). 56o explain this somewhat counterintuitive behavior of ceria nanoislands, i.e. our observation that H adsorption/desorption on Pt(111) remains apparently unchanged even in cases when OH adsorption/desorption is significantly attenuated for the HT and LT samples (Fig. 2a, b and Fig. S7, ESI †), we must invoke a fast and reversible reaction removing the adsorbed H atoms from the free Pt(111) surface in the cathodic scan, and replenishing them back on Pt(111) in the anodic scan.We may estimate the amount of adsorbed H atoms that must take part in this reaction as y Ce EC Áq ref H/Pt .This amount is indicated in Fig. 5a and b using red crosses.We propose that the adsorbed H atoms are removed from Pt(111) by surface diffusion and intercalated at the Pt/ceria interface.Such non-faradaic process involves no charge transfer between H atoms and the working electrode, and, consequently, no additional features in the CV.Intercalated H atoms still interact with the Pt(111) substrate, which limits the charge of H adsorbed on Pt(111) in HT and LT samples to 2/3 of the density of Pt(111) surface atoms, 15 as is the case for clean Pt(111).
We note that the intercalated atomic hydrogen may eventually further react with ceria.8][59] Such non-faradaic reactions may be the cause for the irreversible nature and asymmetry of Ce 3+ /Ce 4+ redox peaks in the CV curves of the HT and LT samples (Fig. 2).Identification and analysis of the involved ceria phases are however beyond the scope of this manuscript.

Suppressing the onset of the HER reaction
In spite of the fact that due to H intercalation, ceria/Pt model electrocatalysts can accommodate an amount of adsorbed H which is comparable to that of a clean Pt(111) surface, the presence of ceria nanoislands suppresses the onset of the hydrogen evolution reaction for the HT and the LT samples.The onset of the HER in the CVs shown in Fig. 2 is summarized in Fig. S9 (ESI †).For all HT and LT samples, the onset of the HER is suppressed in comparison with the clean Pt(111) surface.In LT samples, suppression of the HER onset gets stronger with increasing ceria coverage.In the first approximation, the suppressed onset of the HER observed on our model ceria/Pt samples can be interpreted as a geometrical effect with the HER proceeding only on the fraction of Pt(111) exposed to the electrolyte, and with no participation of H intercalated at the ceria surface.In the case of a continuous ceria layer on Pt(111), and at potentials investigated in this study (Fig. S4, ESI †), no HER is observed at all.

Conclusions
We identified structure-property relationships and bifunctional electrocatalytic mechanisms in a rare-earth oxide-based model electrocatalyst, i.e.CeO 2 (111)/Pt(111).We used a model electrocatalytic approach with the catalysts prepared and characterized by experimental techniques of surface science (physical vapor deposition, STM, AFM, XPS, LEED).The structural and chemical properties of the model catalysts were then correlated to electrochemical properties investigated via cyclic voltammetry in alkaline electrolyte (CV, 0.1 M KOH).

Stability and redox properties
Flat and compact CeO 2 (111) nanoislands with lateral sizes of approx.20 nm or larger, and straight edges are stable during cycling at potentials between À0.05 and 0.9 V RHE .The CV peaks assigned to the electrooxidation/electroreduction of surface Ce atoms are irreversible and appear at approx.0.15/0.30V RHE (cathodic/anodic scan).Similar peaks are identified for a continuous CeO 2 (111) layer on Pt(111).The peaks are assigned to Large vs. small CeO 2 (111) nanoislands were prepared using different preparation methods, but their stability/instability in CV is interpreted as a predominantly structural effect.The small CeO 2 (111) nanoislands include a substantial fraction of undercoordinated surface sites (oxygen vacancies, kink sites at rounded edges) as compared to the larger and compact nanoislands.Upon potential cycling, these undercoordinated sites react with H 2 O to yield a hydrated CeO 2 Á2H 2 O phase.Both CeO 2 and CeO 2 Á2H 2 O nanoislands, which appear at oxidizing potentials (above 0.3 and 0.7 V RHE , respectively), are reduced at lower potentials (0.15 and 0.45 V RHE , respectively), resulting in the creation of Ce 3+ containing inhomogeneous and partly hydroxylated phases that can be denoted as cerium oxyhydroxide.

Outlook
Our investigations illustrate the power of model electrocatalyst studies in identifying the conditions and consequences of physicochemical and electrochemical processes relevant to metal-oxide electrocatalysis.The redox properties of the ceria/ Pt(111) model system in the electrochemical environment are complex and involve the formation of different cerium phases which can be partly controlled via preparation procedures and the applied potential.The present identification of the cerium phases established under electrochemical conditions may contribute to understanding and knowledge-based improvement of technologically favorable properties of ceria-based electrocatalysts.

Fig. 2
Fig. 2 Dependence of the stable CVs on the amount of deposited CeO 2 for HT samples shown in Fig. 1a-c (a), and for LT samples shown in Fig. 1d-f (b).For reference, the CV of clean Pt(111) is shown.Determination of the faradaic contribution to Ce 3+ /Ce 4+ electrooxidation/reduction for HT (c) and LT (d) samples.The red shaded area represents a difference between the experimental CV of a ceria/Pt sample (solid line) and scaled reference data for clean Pt(111).The red ticks indicate the positions of Ce 3+ /Ce 4+ redox peaks.CVs in 0.1 M KOH, at a scan rate of 400 mV s À1 .
, and Fig. S7 (ESI †).A reference CV of clean Pt(111) and the corresponding scaling procedure to determine the attenuation of the OH peak are illustrated in Fig. S3 (ESI †).

Fig. 3
Fig. 3 Stability of the CVs upon immersion of the samples into the electrolyte.The representative HT sample in (a), and the LT sample in (b).In (b), the dotted arrows indicate disappearing peaks, and the full arrows indicate growing peaks.CVs in 0.1 M KOH, at a scan rate of 400 mV s À1 .

Fig. 4
Fig. 4 Electrochemically active area of ceria nanoislands y Ce EC in HT and LT samples (full symbols) in comparison with the coverage of CeO 2 (111) nanoislands y Ce STM estimated from STM images of the as-prepared samples shown in Fig. 1 and Table S1 (ESI †) (open symbols).The error bars represent the AE10% uncertainty in the amount of deposited CeO 2 , and the upper and the lower estimates of CeO 2 (111) STM coverage.

Fig. 5
Fig.5Assignment of charge contributions in the CVs of HT and LT samples shown in Fig.1and 2 as a function of the amount of deposited CeO 2 .Total charge in a CV (q exp tot ), and the faradaic contribution to Ce 3+ /Ce 4+ electrooxidation/reduction (q Ce ) are marked with full symbols, incremental contributions to the total charge from H adsorption/desorption on Pt(111) (q H/Pt ), OH adsorption/desorption on Pt(111) (q OH/Pt ), and the double layer (q DL ) are marked with open symbols.The red crosses indicate the amount of H adsorbed/desorbed on Pt(111), which intercalates at the ceria/Pt interface (H/Ce-Pt).The color shading/hatching of the charge contributions represents a guide to the eyes.The dotted lines represent the total charge in the CV of clean Pt(111) (q Pt tot ), and charge required for a complete oxidation/reduction of all deposited Ce atoms (q Ce-bulk

CeO 2 "
Ce(OH) 3 transition.Flat and open CeO 2 (111) nanoislands with lateral sizes below approx.5 nm, and with rounded edges are initially unstable.The CV peaks assigned to the electrooxidation/ electroreduction of surface and subsurface Ce atoms are irreversible and stabilize at higher potentials of approx.0.45/0.70V RHE (cathodic/anodic scan).This observation is interpreted as a phase change of ceria nanoislands from cubic CeO 2 to This journal is © the Owner 2024 Phys.Chem.Chem.Phys., 2024, 1630-1639 | 1637 amorphous ceria hydrate CeO 2 Á2H 2 O, and the shifted irreversible peaks assigned to the CeO 2 Á2H 2 O " Ce(OH) 3 transition.
Quantitative analysis of the charge in the CVs for the transformation of CeO 2 (111) and CeO 2 Á2H 2 O nanoislands on Pt(111) reveals a bifunctional mechanism in which a fraction of H atoms adsorbed on Pt(111) undergoes surface diffusion and intercalation at the Pt/ceria interface.This non-faradaic mechanism is reversible and allows the accommodation of H atoms on the whole Pt(111) surface in spite of the fact that part of the Pt is occupied by CeO 2 (111) or CeO 2 Á2H 2 O.This behavior is contrary to what is observed in the case of nanoislands of transition metal oxyhydroxides on Pt(111), which tend to block the adsorption/desorption of H. H accommodated at the Pt/ceria interface likely contributes to establishing cerium oxyhydroxide phases at reducing potentials, but it cannot participate in the HER on ceria/Pt model catalysts.