Ceria-based materials have found wide applications in many catalytic processes [1], such as vehicle emission purification [2], water-gas shift reactions [3], CO oxidation [4] etc. Their excellent performance primarily stems from the unique Ce 4f orbital with strong electron localization characteristics, typically exhibiting the dynamic Ce⁴⁺/Ce³⁺ transition [5-8]. Accordingly, the process of oxygen vacancy (O_v) formation has aroused great interest, anticipating that the geometric/electronic structure and/or thermostability of O_v may directly affect the performance of CeO₂ catalysts [9, 10]. For example, Li et al. [11] found that the subsurface O_v is 0.18 eV more stable than the surface one at CeO₂(111), and they also proposed a preferable two-step O_v diffusion mechanism in contrast to the conventional surface hopping mechanism. Nolan et al. [12] and Fabris et al. [13] investigated O_v formation on the three low index CeO₂ surfaces and found that the formation energies follow the order of (111) > (110) > (100). Nevertheless, most of previous theoretical studies employed pure ceria models for simulating physical and/or chemical properties, while the significant role of the interface structure in composite catalysts remains largely elusive.

Being different from the conventional metal/oxide catalysts, the reverse type of 'oxide supported on metal (i.e. oxide/metal)' composites have recently emerged as novel catalytic materials with distinct activities [14-16]. In particular, thin ceria films deposited on metal substrates, such as Pt(111) [17], Rh(111) [18, 19], Cu(111) [20], and Au(111) [21], were successfully synthesized and characterized in experiments. Using the low-energy electron diffraction (LEED) technique, Hardacre et al. [22] observed a (4 × 4) geometry pattern at ceria/Pt composites, which corresponds to a 3:4 lattice matching between CeO₂(111) and Pt(111). Chan et al. [19] carefully characterized the structure of ceria films on the Rh(111) surface using the LEED, scanning tunneling microscopy (STM), and X-ray photoemission spectroscopy (XPS). They found that the ceria/Rh composites are composed of several atomic layers of CeO₂(111) on the Rh(111) substrate, and the deposited CeO₂ shows a O:Ce ratio of 1.96:1 with slightly contracted lattice parameter relative to the bulk ceria. There are also STM measurements showing that ultrathin ceria films at metal substrate often yield small islands with a thickness about one or two O-Ce-O layers [23, 24]. Theoretically, Spiel et al. [17] constructed three types of ceria/Pt composites composed of monolayer CeO₂(111) on the Pt(111) substrate, and found that aligning surface oxygens of CeO₂(111) right on the top sites of interface Pt atoms could enhance the stability of CeO₂(111)/Pt(111) composites. Chan and coworkers [15] investigated the thermostabilities of defective CeO₂/Rh(111) composites, and found that O_v prefers to form at the interface rather than on the ceria surface. Nevertheless, fine structural resolution of the reverse oxide/metal catalysts at the atomic level is still very rare, hindering the progress in the clarification and regulation of the structure-property relationship.

From the catalytic perspective, CO oxidation was widely considered as a model reaction to probe the reactivity of CeO₂-based catalysts [25-27]. As both the geometric and electronic structures of ceria thin films are certainly affected by the metal support, it was anticipated that the reverse ceria/metal catalysts might show distinct catalytic properties compared with the pure ceria materials [26-29]. Recently, by growing well-defined monolayer ceria islands on the Pt(111) surface and monitoring Ce oxidation state, Suchorski et al. [27] observed enhanced CO oxidation activity on the ceria/Pt(111) composites with respect to the free-standing Pt(111) or CeO₂(111), and they attributed such enhancement to the optimized electronic structures and accelerated oxygen supply brought by ceria deposition. Besides the ceria/Pt composites, Palomino et al. [30] and Mullins et al. [31] prepared a series of oxide/Au(111) reverse catalysts (e.g. MgO, CeO₂, TiO₂, Fe₂O₃), among which they consistently found enhanced oxide-metal interactions and CO oxidation activities relative to the conventional metal/oxide structures. Obviously, the unique catalytic performance of reverse catalysts has been evidenced in many experiments [32, 33], though the fundamental understandings on the origins of the promotion effect and catalytic activities are currently very limited, largely owing to the difficulty in determining reliable interface structures.

In this work, we conducted density functional theory (DFT) calculations corrected by on-site Coulomb interaction (for Ce) to systematically examine the structures and stabilities of various CeO₂(111)/Pt(111) reverse catalysts. We first focused on the identification of the optimal structure of monolayer CeO₂(111) on the Pt(111) support, and then moved to investigate the changes of the thermodynamic, geometric, and electronic properties at the interface as the ceria thickness increases. It was found that the significant impact of metal substrate on ceria slabs, particularly for interface oxygens, can be directly captured by a simple parameter of the O_v formation energy. After clarifying the physical properties of the CeO₂/Pt(111) composites, we used the model reaction of CO oxidation to probe the unique catalytic features of the reverse catalysts, through comparative studies of the monolayer CeO₂(111) with and without the Pt(111) support as well as the regular CeO₂(111) surface. Our results underline the significant role of the moderate ceria-Pt interaction at the interface that endows the CeO₂/Pt reverse catalysts both good thermostability and high catalytic activity.

Spin-polarized density functional theory calculations were performed by using the Vienna Ab-initio Simulation Package (VASP) [34, 35]. The exchange-correlation interaction was treated by the Perdew-Wang 1991 (PW91) functional on the basis of the generalized gradient approximation (GGA) [36]. The Kohn-Sham wavefunctions were expanded in a plane wave basis with an energy cutoff of 400 eV, utilizing the projector augmented wave (PAW) method [37, 38]. The Pt 5p, 5d, and 6s, the Ce 4f, 5s, 5p, 5d, 6s, and the 2s, 2p orbitals of carbon and oxygen atoms were treated as valence states. Throughout the DFT calculations [39], we applied an on-site Coulomb interaction correction, i.e. the DFT + U method, with a U value of 5 eV for the Ce 4f orbital, consistent with the previous reports [40-42]. The transition states (TSs) of reactions were located using a constrained optimization scheme [43], and all the calculations for structural optimization and TS search were converged until the forces on each ion were less than 0.05 eV/Å. For the STM simulation, we adopted the Tersoff-Hamann approximation, using a constant-current mode with bias voltage between -2 V and +2 V, to simulate the tunneling current proportional to the local density of states [44]. In addition, we checked the influence of the Van der Waals [45] interaction on both surface reactions and the O_v formation (Table S1), and the negligible energy difference indicates that DFT + U is competent to capture the catalytic trends in CO oxidation on ceria/Pt composites.

To construct the composite models, we adopted the 3:4 lattice-matching pattern observed in experiments, and built a p(3 × 3) CeO₂(111) surface cell on the top of a four-layer p(4 × 4) Pt(111) substrate. Considering the lattice parameters of 3.87 Å for ceria and 2.84 Å for Pt, such pattern can cause only small lattice contraction of 2.15% for CeO₂(111). The Brillouin-zone integration was treated using a 1 × 1 × 1 k-points mesh. During all calculations, the bottom two layers of Pt substrate were kept fixed, while the other Pt and ceria layers were allowed to relax. We altogether considered three types of CeO₂/Pt(111) composites with different thickness of ceria. As one can see from Fig. 1(a‒c), the monolayer (ML), bilayer (BL), and trilayer (TL) CeO₂(111) slabs were deposited on the Pt(111) surface with a large vacuum gap (> 10 Å) separating periodic cells, denoted as the ML, BL, and TL CeO₂/Pt(111) models, respectively. For calculations of the regular CeO₂(111) surface, we used a TL CeO₂(111) surface model with the bottom two layers being fixed during calculations, as frequently suggested by previous reports [46].

In this work, the oxygen vacancy formation energy (E_Ov) was defined as

where E_(O2) is the energy of a gaseous O₂ molecule, and E_slab and E_vac are energies of the catalyst without and with one oxygen vacancy, respectively.

The adsorption energy of species X (E_ad(X)) was calculated according to

where E_total represents the energy of the interacting system, and E_X is the energy of the adsorbate X species in the gas phase. We can also calculate the adsorption energy of ceria at Pt substrate, E_ad(ceria), as follows

where E_Pt and E_ceria are energies of the separated Pt and ceria slabs, respectively.

Moreover, we also calculated the normalized adsorption strength of ceria on the substrate with interface adhesion energies (E_adh) according to E_adh = E_ad(ceria)/A, where A is the interface area, and defined the deposition energy of ceria, E_depos(ceria), as the adsorption energy of the outmost ceria layer on the underneath CeO₂ slabs in the BL and TL CeO₂/Pt(111) composites. Obviously, positive E_ad(X), E_depos(ceria) or E_adh corresponds to the exothermic adsorption/adhesion process, and the more positive the E_ad(X), E_depos(ceria) or E_adh is, the stronger the adsorbate X or ceria slab binds to the surface.

As the structure of the oxide/metal interface largely determines the thermostability of the corresponding composite catalysts, we first systematically calculated the various adhesion geometries of ML CeO₂(111) on the Pt(111) support. In total, 9 types of ML CeO₂/Pt(111) structures with different in-plane positions of ceria relative to the Pt substrate are presented in Fig. 2, from which we showed how these structures were constructed by moving the CeO₂(111) layer along the x and y axes to the different positions. From the calculated energetics, we found that the interaction between ceria and Pt is generally not strong, and the relative shift of the two materials only causes small impact on the E_ad(ceria), showing a narrow energy window of 3.20‒3.50 eV (Table 1). Even for the calculated ML CeO₂/Pt(111) structure with the lowest energy (Fig. 2(a)), the calculated E_adh of 0.43 J﹞m^‒2 is just comparable to that of the van der Waals junctions in MoS₂ nanosheets (0.36‒0.47 J﹞m^‒2) [47]. A simple reason for the weak ceria-Pt interaction is that Pt is generally not an oxophilic metal (for example, the E_ad(O₂) was calculated to be as low as 0.76 eV on the close-packed Pt(111) surface).

In Fig. 3(a), we present the interface structure of the most stable ML CeO₂/Pt(111) composite from the side view. One can see that the interface oxygens (O_i) located on (or near) the top of substrate Pt atoms can slightly pull the corresponding Pt out of the plane, forming in total 7 O_i-Pt bonds. However, for the less stable structures (Fig. 2(b)‒(i)), the mismatched bonding pattern with slightly longer distances of O_i-Pt bonds (Table 1) can be easily recognized. For example, we determined that the average O_i-Pt distance decreases by 0.02 Å for the least stable composite (Fig. 2(i)) compared to the most stable one (Fig. 2(a)), though they involve the same number of O_i-Pt bond.

With the same approach, we also systematically calculated the possible structures of BL and TL CeO₂/Pt(111) composites, and the determined most stable ones are shown in Fig. 1. It needs to mention that Chan et al. [19] have conducted STM measurement of the BL CeO₂/Rh(111) composite, which helped them resolve the atomic configurations of the exposed CeO₂(111) surface. In the current work, we also performed STM simulation on the most stable BL CeO₂/Pt(111) structure (Fig. 3(b)). The ordered bright spots representing surface (or near-surface) Ce cations can be easily recognized, and each surface cell is marked by the dashed line in red. It should be note that, Chan et al. [19] characterized a 5:7 lattice-matching pattern between CeO₂(111) and Rh(111) from their experiments. In the current work, we adopted the 3:4 pattern observed in experiments for the CeO₂/Pt(111) composites (Fig. 3(b)), and four bright spots along each side of the surface cell can be recognized. Besides, relative lower resolution contrast and absence of one bright spot within each surface cell in the simulated STM image might be largely caused by the rugged surface configuration of the supported CeO₂(111).

To determine the effects of ceria thickness on the interface properties, we compared the interface structures of the ML, BL, and TL CeO₂/Pt(111) composites with the most and least stable structures (see Table 1). It is evidenced that, for each type of CeO₂/Pt(111) composites, the E_adh decreases as the l_Oi-Pt increases and/or the N_Oi-Pt decreases (see more samples in Table S2). From their most stable interface structures, we determined the same number of 7 O_i-Pt bonds with the average distances of 2.13, 2.17, 2.16 Å for the ML, BL and TL CeO₂/Pt(111) composites, respectively. On the other hand, their least stable interface structures exhibit either identical N_Oi-Pt but elongated l_Oi-Pt (e.g. ML and TL composites), or identical l_Oi-Pt but fewer N_Oi-Pt (e.g. the BL composite). Most importantly, we determined that the ceria-Pt interaction strength (E_ad(ceria) or E_adh) largely decreases as the thickness of the ceria slab increases. Accordingly, the BL and TL CeO₂/Pt(111) structures both give longer l_Oi-Pt distances than the ML CeO₂/Pt(111). Such changes can be attributed to the gradually improved bonding strength within ceria slabs as evidenced by the calculated E_depos(ceria) in Table 1. In general, our results regarding the interaction energies and interface structures show that the physical properties of the reverse CeO₂/Pt(111) composites basically get converged when ≥ 2 ceria layers are deposited on the Pt substrate.

Having obtained the optimal composite structures, we then performed electronic structure and Bader charge analyses to investigate the electronic properties of the ML, BL and TL CeO₂/Pt(111) composites. From the density plots of calculated charge differences (Fig. 4), one can see that drastic charge redistributions occur at the interface regions for all the three types of composites. Specifically, one oxide layer and two metal layers around the interface were found to be significantly affected, in similar manners to the previous results regarding the metal/TiO₂ systems [48]. For the thicker ceria slabs, the top CeO₂ layers gradually become less disturbed as they are located away from the interface (Fig. 4(b), and (c)). Accordingly, calculated Bader charges in Table 2 show that each O_i atom is positively charged by ~0.07 |e| and interface Pt is positively charged by ~0.04 |e| for the CeO₂/Pt (111) composites with respect to the free-standing ceria slabs and Pt(111) surface; while the charges of the surface oxygens (O_s) on the top layer of CeO₂/Pt (111) remain basically unchanged (≤ 0.01 |e|). Overall, we determined that the ceria slab only transfers 0.138, 0.031 |e| to the Pt substrate for the ML, BL CeO₂/Pt (111) composites, respectively, whereas Pt donates as low as 0.003 |e| to ceria for the TL composite (last row in Table 2). These results show that the interfacial interactions are mainly contributed by the electrostatic interactions between the top Pt layer and the bottom ceria layer at the interface rather than the covalent hybridization. Such feature was also evidenced by the significant electronic polarization at the interface (Fig. 4) and redistribution within the individual ceria or Pt slabs (Table S3). In addition, considering the accumulation of positive charges at O_i and the negligible electronic changes at O_s, distinct differences in these two types of O can be further expected in catalytic reactions.

Considering that the excellent oxygen storage/release capacity of ceria is closely related to the O_v formation, [5, 6] we then systematically calculated and compared the E_Ov at different sites (O_i and O_s) of CeO₂/Pt(111) composites and on the CeO₂(111) surface. Table 3 shows that, for the free-standing CeO₂(111) slabs, the O_v formation energy at surface O_s site (E_Ov-s) was computed to be 2.99 and 3.02 eV on the BL and TL surfaces respectively, being ~0.8 eV larger than that on the ML slab (E_Ov-s = 2.21 eV). These results again indicate that the ML CeO₂(111) is quite active and the BL and TL CeO₂(111) are less active and may give reduced adhesion strength with the Pt substrate (Table 1).

In addition, the calculated E_Ov-s for all the three types of CeO₂/Pt(111) composites were found to be ~0.3 eV lower than those in the corresponding free-standing ceria slabs, and such small energy differences are also consistent with the similar calculated Bader charges on O_s (Table 2). Interestingly, we found that O_v formation turns to be markedly promoted by 1.0‒1.5 eV at the O_i site (E_Ov-i) at the interface with the Pt substrate, being 0.92, 1.28 and 1.26 eV for the ML, BL, and TL composites, respectively. The easy removal of the O_i relative to the O_s is probably due to the proper structural accommodation to the as-formed O_v-i as provided by the Pt substrate. Besides, these E_ov results again demonstrated the converged physicochemical properties of the reverse CeO₂/Pt(111) composites when more than two layers of ceria slabs are deposited.

It needs to be particularly noted for the ML CeO₂/Pt(111) composite that both the E_Ov-i and E_Ov-s are much smaller than that of bulk oxygens in ceria (E_Ov = 3.05 eV) [49] or other oxides such as TiO₂ (E_Ov = 3.16 eV) [50]. After carefully examining the composite structures, we found that ceria slabs in the ML CeO₂/Pt(111) is more flexible than that in the BL and TL composite. As illustrated in Fig. 5, the vertical Ce-O_i and Ce-O_s distances (along the z direction; Table S4) in the ML CeO₂/Pt(111) can change dynamically to better stabilize the composite, forming different extents of compressed Ce-O_s interactions depending on the absence or presence of O_v at the interface (see vertical distances in Fig. 5(b)‒(c) and Ce-O bond length in Table S5). The dynamic distance changes were also observed in the BL and TL CeO₂/Pt(111) composites, but their corresponding extents are relatively suppressed (Tables S4 and S5), owing to the increased structural rigidness in ceria as the slab thickness increases.

Particularly for the Ce-O_s bond compress induced by the O_v-i at the ML CeO₂/Pt (111) composite, we calculated the charge differential density of ML CeO₂/Pt(111) composite without (Fig. S1(a)) and with (Fig. S1(b)) a O_v-i. It was found that the presence of O_v-i causes slightly stronger electronic interaction as evidenced by the broader differential density, which accordingly induces structural changes such as bond compression or extension. After comparing the calculated bader charges of the two structures (Table S3), we revealed the following features: (i) the interface first oxygen (O1) layer generally holds similar charge quantity before and after the O_v-i formation, whereas the first cerium (Ce1) layer becomes less positive in the presence of O_v-i, which therefore leads to the elongated O1-Ce1 distance (0.926 Å vs. 0.856 Å; Fig. 5) owing to the reduced electrostatic interaction; (ii) the upper oxygen layer (O2) carries more negative charges (1.222 vs. 1.189 |e|) after O_v-i formation, and taking into account the already reduced O1-Ce1 interaction, it is reasonable that the Ce1 would bind the O2 layer stronger with compressed Ce‒O distance; (iii) the overall charge transfer between ceria and Pt substrate remains fairly small (0.273 |e|) after O_v-i formation, which again signifies the intensive electrostatic interactions rather than the covalent hybridization between the ceria slab (containing O_v at the interface) and Pt substrate.

Overall, these geometric and electronic features regarding the compressed Ce‒O_s (or extended Ce‒O_i) distances and stronger structural relaxation in the ML CeO₂/Pt(111) could reasonably account for the lower O_v formation energies. Given that O_v formation is widely regarded as an important step in ceria catalysis (e.g. CO oxidation [51]), it is therefore expected that the ML CeO₂/Pt(111) composite might show excellent catalytic activities.

Regarding the catalytic performance, we took CO oxidation as a model reaction to study the reactivities of the reverse CeO₂/Pt(111) catalysts. We first calculated CO adsorption at the ML, BL, TL composites and compared to that on the regular CeO₂(111) surface (using a TL model). It was found that ML CeO₂/Pt(111) gives the largest E_ad(CO) of 0.20 eV, being twice the value on the CeO₂(111), and the adsorption strength decreases as the ceria slab becomes thicker. These results agree well with experimental observations that the CO adsorption capacity (including both the amount and the strength of CO adsorption) decreases as more ceria are deposited on the metal substrate [26], and also indicate that the ML CeO₂/Pt(111) might be more active than other composites. In addition, considering the weak adsorption of CO, one may also anticipate that CO oxidation via the Langmuir-Hinshelwood (LH) mechanism [52] is not likely to occur on the CeO₂/Pt(111) composites. Therefore, we mainly focused on the Mars-van-Krevelen (MvK) mechanism [53] as widely validated for ceria based catalysts [54, 55], and compared the calculated energetics of CO oxidation at ML CeO₂(111) and ML CeO₂/Pt(111) with those on the regular CeO₂(111) surface.

The calculated reaction pathways of CO oxidation (toward CO₂ production) via the MvK mechanism on the unsupported ML and regular CeO₂(111) surfaces are shown in Fig. 6. This process starts with a gas-phase CO molecule (IS) which then forms weak interaction with the surface (IM1) and combines with an O_s to form the OC-O_s bond. At the coupling TS, the OC-O_s distance was calculated to be ~1.50 Å while the O-C-O_s angle is ~120º. We determined a coupling barrier of 0.61 eV on the regular CeO₂(111) surface, but a low barrier of 0.33 eV on the ML CeO₂(111). After the coupling, the linear CO₂ (IM2) occurs with an exothermic energy of 0.73 and 1.38 eV on the regular and ML CeO₂(111), respectively. Obviously, the lower barrier and larger energy release on the ML CeO₂(111) relative to the regular surface can be rationalized by their different E_Ov-s values (2.21 vs. 3.02 eV; Table 3). For the last step of CO₂ desorption, the as-formed CO₂ can leave both surfaces quite easily by overcoming its adsorption energies of ~0.38 eV, leaving an O_v on the surface. Note that previous reports have suggested that the surface O_v at ceria surfaces can be readily recovered [56, 57] and consequently the overall reactivity of CO oxidation can be largely determined by the OC···O coupling. Therefore, judging from the above calculated energetics, we expect that the ultrathin ML CeO₂(111) surface can show much higher CO oxidation activity than bulk ceria materials.

For CO oxidation on the ML CeO₂/Pt(111) composite (Fig. 7), the initial reaction steps are very similar to those on CeO₂(111) surfaces, leading to the formation of surface CO₂ (IM2), though lower coupling barrier (0.45 vs. 0.61 eV) and larger energy release (1.69 vs. 0.73 eV) were also determined. These results are again consistent with the vast E_ov difference between the two catalysts (1.92 vs. 3.02 eV; Table 3). In contrast, CO oxidation on the ML composite could form two types of O_v defects at the O_s or O_i site (FS and FS' in Fig. 7) after CO₂ desorption. The former O_v-s pathway shows no obvious difference to that on free-standing ceria surfaces, releasing gaseous CO₂ directly with a small endothermic energy of 0.36 eV; while the latter O_v-i pathway involves the migration of an interface O_i to fill the surface O_v-s (IM3), which, however, turns to be thermodynamically more preferable by ~1.0 eV than IM2.

In addition, though the result of much lower E_Ov-i than E_Ov-s was also evidenced in the BL and TL composites (Table 3), O_i migration requires multi-steps of oxygen relay within such thicker ceria slabs, thus being a kinetically much slower process. In this regard, we calculated the energy barrier of O_v migration in the ML CeO₂/Pt(111) composite using the climbing image nudged elastic band (CI-NEB) method [58]. We determined an energy barrier of 0.92 eV, and the dynamic picture of O_v migration was presented in Fig. S2. Therefore, our results indicate that the additional CO oxidation pathway involving O_v-i formation is an unique feature of the ML CeO₂/Pt(111) or perhaps other ML ceria/metal composites. In fact, this speculation was supported by experimental observation of O_v-i formation in the ML CeO₂/Rh(111) catalyst under flowing CO atmosphere [18].

Finally, though an higher OC···O coupling barrier of 0.45 eV was determined on the ML CeO₂/Pt(111) than that on the ML CeO₂(111) film (0.33 eV), the calculated results still indicate that the moderate ceria-Pt interaction at the interface can not only greatly stabilize the composite structure, realizing the synthetization of ultrathin ceria/metal reverse catalysts in experiments [59, 60], and it may also largely preserve the superior reactivity of the ML ceria slab. Therefore, the ML ceria/Pt composite reaches a good balance between the thermostability and the reactivity, manifesting itself superior catalysts for catalyzing CO oxidation.

In this work, we performed DFT + U calculations to systematically investigate the geometric, electronic, and catalytic properties of the reverse type of CeO₂/Pt(111) composites. The following main conclusions can be obtained.

The ceria-Pt interaction strength is closely relevant with the distance and number of O_i-Pt bonds formed at the interface, and ML CeO₂/Pt(111) shows larger adhesion interaction (0.43 J﹞m^‒2) than the BL or TL composites. Increased thickness of the ceria slab in the CeO₂/Pt(111) composites is highly favored for their thermostability, though their physical properties basically get converged when ≥ 2 ceria layers being deposited on the Pt substrate.

The contact between the ceria slab and the Pt substrate can significantly affect the electron distribution of the one ceria layer and two metal layers around the interface, and lowers the O_v formation energy by ~0.3 eV at the exposed ceria surface whereas as high as 1.3‒1.8 eV for the interface O_i as compared with the free-standing CeO₂(111) slabs.

The key step of OC···O coupling in CO oxidation gives the reaction barriers of 0.61, 0.33, and 0.45 eV on the regular and ML CeO₂(111) surfaces and ML CeO₂/Pt(111) composite, respectively, indicating an intrinsic activity trend of ML CeO₂(111) > ML CeO₂/Pt(111) > regular CeO₂(111). Moderate ceria-Pt interaction at the interface not only greatly stabilizes the composite structure, but also successfully preserves the superior reactivity of the ML ceria slab. Our results demonstrate the ML CeO₂/Pt(111) composite as a promising type of catalyst with high activity for CO oxidation.

There are no conflicts of interest to declare.