Stability, electronic properties and CO adsorption properties of bimetallic PtAg/Pt(111) surfaces †

Aiming at a better fundamental understanding of the chemistry of bimetallic PtAg/Pt(111) surfaces, we have investigated the stability, electronic properties and CO adsorption properties of bimetallic PtAg surfaces, including pseudomorphic Ag film covered Pt(111) surfaces and Pt x Ag 1 (cid:2) x /Pt(111) monolayer surface alloys, using periodic density functional theory calculations. The data provide detailed insights into the relative stabilities of diﬀerent surface configurations, as indicated by their formation enthalpies and surface energies, and changes in their electronic properties, i.e. , in the projected local densities of states and shifts in the d-band center. The adsorption properties of diﬀerent Pt n ensembles were systematically tested using CO as a probe molecule. In addition to electronic ligand and strain eﬀects, we were particularly interested in the role of diﬀerent adsorption sites and of the local CO ad coverage, given by the number of CO molecules per Pt surface atom in the Pt n ensemble. Diﬀerent from PdAg surfaces, variations in the adsorption energy with adsorption sites and with increasing local coverage are small up to one CO ad per Pt surface atom. Finally, formation of multicarbonyl species with more than one CO ad per Pt surface atom was tested for separated Pt 1 monomers and can be excluded at finite temperatures. General trends and aspects are derived by comparison with comparable data for PdAg bimetallic surfaces. Fundamental insights relevant for applications of bimetallic Pt catalysts, specifically PtAg catalysts, are briefly discussed.


Introduction
2][3][4][5][6] The improved catalytic performance was attributed to an interplay of a number of different effects such as geometric ensemble effects, electronic ligand effects and electronic strain effects, where the first describes the influence of the size/configuration of active surface ensembles. 7,8Electronic ligand effects describe the effect of electronic modifications of the active site by different neighboring surface atoms. 9,10These as well as strain effects, which reflect the effect of lattice distortions as compared to the natural lattice of the respective surface layer on the electronic structure, 11 can lead to changes in the adsorption energy and thus to significant modifications in the reaction kinetics. 12,135][16] Initially, these different effects were mainly derived from changes in the catalytic activity upon varying the concentration of the respective components.More direct studies of these effects, correlating structural and electronic modifications with catalytic properties, became feasible with the advent of modern high-resolution spectroscopies and microscopies and in particular of theoretical methods, in combination with the increasing knowledge of the reproducible preparation of structurally well-defined model surfaces/systems. 17,18s part of an extensive series of combined experimental and theoretical studies on the structure, stability, electronic properties and adsorption behavior of structurally well-defined bimetallic PdAg/Pd(111) [18][19][20][21][22][23] and PtAg/Pt(111) [24][25][26][27][28] surfaces, we here report results of a theoretical study on the stability of different Ag/Pt(111) and PtAg/Pt(111) surfaces, as indicated by the formation enthalpy and surface energies, on their electronic properties, and on the CO adsorption behavior.
The structure, formation and stability of PtAg/Pt(111) surface alloys have been characterized previously using a variety of different techniques, [29][30][31][32] and the results were summarized previously by Jankowski et al. 33 and by Bauer et al. 34 These studies found that deposition of submonolayer amounts of Ag on Pt(111) and subsequent annealing to about 620 K result in the formation of monolayer PtAg surface alloys, where the Ag atoms are confined to the topmost layer 24,29,30,32 with a tendency to phase separation to form two-dimensional, homoatomic ensembles rather than a random distribution. 25Thermal and CO induced intermixing of Pt and Ag in Ag films on Pt(111)  and in Pt film covered Ag(111) surfaces were investigated by Jankowski et al. 33 experimentally and by Hua et al. theoretically. 35hermodynamically, none of these structures are stable because of entropic reasons.It depends on the reaction conditions whether these structures are kinetically stable.For the oxygen reduction reaction, our earlier work demonstrated that they are stable at room temperature. 27,28ost important for the present study is the finding of a pronounced ligand effect, with neighboring Ag surface atoms, leading to strengthening of the Pt-CO bond, due to an increased back-donation from the Pt atom into the 2p* orbital of the adsorbed CO.Counteracting effects resulting from the compressive strain, due to the larger size of the Ag atoms and the pseudomorphic growth of the surface alloy, seemed to be less important.These experimental results, which were based mainly on temperature-programmed desorption (TPD) and infrared spectroscopy (FTIR) results, were also supported by DFT calculations. 25Furthermore, the authors reported distinct changes in the adsorption site on these surface alloys with increasing CO ad coverage, which also depended on the Ag surface concentration and thus on the relative concentration of small ensembles such as Pt 1 monomers, Pt 2 dimers, compact Pt 3 trimers and larger Pt n ensembles. 25A later high-resolution XPS study, which could also distinguish bridge-bonded CO from on-top adsorbed CO based on their slightly different C 1s binding energies, essentially confirmed these findings. 34hese bimetallic surfaces can be considered as model systems for realistic bimetallic systems, which because of their well-defined and simple structure can provide detailed information on the structural and electronic effects on the surface chemistry of bimetallic surfaces that is not easily accessible from realistic systems.][38] In the present study, we are particularly interested in identifying possible highly stable local structures and in correlations between the electronic properties of specific Pt n ensembles and their CO adsorption strength.Considering the CO adsorption behavior we are most interested in changes in the adsorption characteristics with increasing local coverage, i.e., with an increasing number of CO ad per Pt surface atom in Pt n ensembles in bimetallic surfaces.This expands on our earlier report, where we explored the low-coverage CO adsorption behavior on small Pt n ensembles (n = 1-3) in Pt x Ag 1Àx /Pt(111) monolayer surface alloys, with 1 CO molecule per Pt n ensemble. 252 Combining the present data with earlier results, 25,34 we will also try to obtain further information on the role of electronic ligand and strain effects.Finally, aiming also at a more general understanding, we will compare the results obtained here with comparable data reported earlier for a structurally rather similar PdAg/Pd(111) system. 23

Computational details
Plane-wave DFT calculations were performed using version 5.3.3.4 of the VASP code, 43 together with the Perdew-Burke-Ernzerhof (PBE) 44 and the revised-PBE (RPBE) 45 exchangecorrelation functionals.The ionic cores are represented by projector augmented wave (PAW) potentials 46 as constructed by Kresse and Joubert. 47,48The electronic one-particle wave functions are expanded in a plane-wave basis set up to a cutoff energy of 400 eV.This cutoff energy, which corresponds to the maximum cutoff for the four atomic species considered in this study (Pt, Ag, C, and O) and is larger than the default values preset in the code for platinum and silver, was set manually in order to keep it constant for all calculations.Default cutoff energies are already expected to provide convergence better than 1 mRy (B13 meV) in eigenvalues for this kind of basis set.Spin polarization is not considered due to the non-spinpolarized nature of the system.Dipole moment correction is set up in order to account for effects derived from using asymmetric slabs.Scalar relativistic effects are already included from the parametrization at the basis set generation.Convergence criteria for the electronic self-consistency and the ionic relaxation were set to 1 Â 10 À5 eV and 1 Â 10 À4 eV, respectively.A sufficiently large set of k-points was chosen in order to guarantee convergence.
First, the bulk energies (E b ) and bulk lattice parameters (d b ) of Pt and Ag were computed using an fcc unit cell and a 11 Â 11 Â 11 G-centered k-point grid.Values obtained using PBE/PAW for the bulk lattice parameter are 3.97 Å and 4.15 Å for Pt and Ag, respectively.These are in close agreement with the experimental values of 3.9231 Å and 4.0862 Å. 49 50 The bimetallic surfaces are represented by periodic slabs consisting of five monolayers.The vertical height of the threedimensional unit cell was set to an integer multiple of the surface lattice parameter, 10d s , which allows us to have a separation between slabs close to 18 Å in all cases.Geometry optimization of the various surface configurations was carried out keeping the two bottom Pt(111) layers fixed at their corresponding bulk positions, while the three upper layers were allowed to relax.We used a (3 Â 3) surface unit cell and performed geometry optimizations using a 3 Â 3 Â 1 G-centered k-point grid.For the various structures studied here we used the following notation: Ag nL /Pt(111) denotes a structure with n pseudomorphic silver overlayers above the Pt(111) substrate, Pt 1Àx Ag x /Pt(111) denotes a surface alloy in the topmost layer, and Pt 1Àx Ag x /Ag nL /Pt(111) denotes a structure with a surface alloy at the topmost overlayer and n pseudomorphic Ag layers underneath.These structures will later in the paper be abbreviated as Pt 5L (= Pt(111)), Ag 1L /Pt 4L (Ag 1L /Pt(111)), PtAg 1L /Pt 4L (PtAg 1L /Pt(111)) etc.Finally, we would like to add that because of the similarity in these calculations we expect relative energy changes by 0.01 eV to be significant, both for formation enthalpies and adsorption energies.

Stability of silver/platinum bimetallic surfaces
The stability of the bimetallic surfaces was mainly characterized by their formation enthalpies.The formation enthalpies (DH) of the bimetallic surfaces were calculated according to Barabash et al. 51 Starting from the Pt(111) surface, we obtain where E slab is the total energy of the slab, n i denotes the number of atoms of type i in the slab (per unit cell), E i denote the bulk energies (per atom) of these species, and A is the surface area of the unit cell.However, surface energies calculated this way represent an average of the surface energies of the upper and lower side of the slab, which are not necessarily the same in our study.Furthermore, these surface energies also contain contributions from the energy of the internal interface between Pt and Ag (interface energy).To avoid the first problem, we will also provide estimated surface energies of the Ag containing surfaces, which were calculated from the value calculated via eqn (2) by assuming that the surface energy of the bottom side is identical with that of a Pt(111) surface, which seems to be reasonable.Nevertheless, we will base our discussion of the stability of the different surface configurations mainly on the formation enthalpies and use surface energies only for comparison of rather similar surfaces, such as the Ag film covered surfaces.We would also like to note that while both surface energy and formation enthalpy represent measures of the stability, the references are different.The surface energy, often defined as the energy per surface area required to cleave an infinite crystal along a certain orientation, reflects the driving force towards reducing the surface area, e.g., by sintering, and also the tendency of a given component to accumulate at the surface, e.g., by surface segregation.The formation enthalpy indicates the enthalpy gained upon formation of the respective slab, as compared to the bulk energies of the separate components.Hence, it also indicates the tendency for phase separation.For the following it is important to keep in mind that in all calculations the Pt atoms in the bottom most two layers are fixed on their bulk positions and not allowed to relax vertically, providing a structurally adequate model of a Pt(111) substrate.For surface energies, which are determined by the breaking of bonds that are three layers away from the bottom two layers, deviations caused by the limited thickness of the Pt substrate part, will be negligible.For CO adsorption energies, which are mainly determined by the direct neighborhood of the adsorbed molecule, we expect the same.For the enthalpies of formation, there may be small deviations in the absolute values of the formation enthalpies due to the limited thickness of the Pt(111) bulk representation, but the differences in formation enthalpies of different configurations and thus the general trends between different structures are expected to be fully correct.
3.1.1.Ag nL /Pt(111) pseudomorphic overlayers.Table 1 shows both formation enthalpies and surface energies calculated using PBE/PAW and RPBE/PAW.Here we find the most negative formation enthalpies (highest stability) for the Ag film covered surfaces, with minute differences between different Ag film thicknesses.Obviously, the costs for surface and interface formation vary very little with Ag film thickness.For the Ptterminated surfaces, the formation energies are close to zero, i.e., the presence of the Ag underlayer and the formation of the two internal Pt-Ag interfaces has little effect on the energetics of these systems as compared to Pt(111).For the monometallic Ag systems we do not provide formation enthalpies for reasons discussed in Section 3.1.
By and large, these trends are also reflected by the surface energies, with higher surface energies for less negative formation enthalpies.Among the Ag containing surfaces, those with Ag at the surface exhibit the lower surface energy.Hence, as expected, the termination by Ag (plus the formation of the Pt-Ag interface) is more favorable than Pt termination.Note that the estimated surface energies of the top layer of the Ag-terminated surfaces differ only little from that of Ag(111).On the other hand, contributions from interface energies are likely responsible for the high value of the surface energy of Pt 2L /Ag 1L /Pt 2L , which is even higher than that of Pt(111).
Interestingly, these trends differ significantly from those obtained for comparable Ag nL /Pd(111) systems. 23While for Ag nL /Pt(111) surfaces the surface energies/formation enthalpies are approximately identical for films with 1, 2 or 3 pseudomorphic Ag layers, there are significant differences for Ag nL / Pd(111) film surfaces, with a distinct increase of the surface energy/decrease of the formation enthalpy from the first to the second Ag layer film, while for the 3 layer system the changes are small.We do not have a specific explanation for this trend, but we like to note that in spite of the fact that Pt is situated below Pd in the periodic table, there are many differences in the chemical properties of these two elements, for example with respect to their CO adsorption properties.This shall be investigated in more detail in the future.
Finally, we would like to note that all calculations were performed for pseudomorphic overlayer systems, even though experimentally only Ag monolayer films remain pseudomorphic, while for bilayer and thicker films strain relief results in the formation of a unidirectionally expanded (striped) phase or, upon annealing, in a trigonal incommensurate phase, where strain is relieved isotropically. 52,53Despite these structural differences, the data derived here provide detailed insight in trends in the stability of the different configurations.
3.1.2.Pt x Ag 1Àx /Pt(111) surface alloys.The formation enthalpies and the surface energies were evaluated also for Pt x Ag 1Àx / Pt(111) monolayer surface alloys with compact Pt 1 , Pt 2 , Pt 3 , and Pt 4 ensembles in the topmost layer, where these ensembles are separated from each other by Ag surface atoms.The smallest unit cell allowing this is a 3 Â 3 unit cell.The ensembles correspond to the most compact structures, for example a triangle for Pt 3 and a rhombus for Pt 4 .They are the result of substituting 8, 7, 6, or 5 Pt atoms by Ag atoms at the pure Pt(111) surface within the 3 Â 3 unit cell, or equivalently, substituting 1, 2, 3, or 4 Ag atoms by Pt atoms in the pseudomorphic Ag monolayer in the Ag 1L /Pt(111) surface.These different ensembles allow the occupation of different types of CO adsorption sites such as on-top sites, bridge sites, threefold hollow sites, etc.Similar calculations were performed also for monolayer surface alloys supported on pseudomorphic Ag monolayer (PtAg/Ag 1L / Pt 3L ) and bilayer (PtAg/Ag 2L /Pt 2L ) films on the Pt(111) substrate.The results obtained for using PBE/PAW are collected in Table 2, and those for RPBE/PAW in Table S2 in the ESI.† Furthermore, the formation enthalpies are plotted in Fig. 1 as a function of the Ag content in the surface layer, including also higher Pt concentrations.
As shown in Table 2 and in Fig. 1, the formation enthalpy decreases almost linearly (becomes less negative) with increasing Pt content in the topmost layer for PtAg/Pt 4L .Conversely, it increases with increasing Ag content.In Fig. 1, the first (x = 0) and last (x = 1) point represent the values for the Pt(111) and Ag 1L Pt 4L systems, respectively.The approximately linear decay of the mean formation enthalpy can be converted into a differential value, indicating the differential change in system energy upon exchanging another surface atom, which are also listed in Table 2.Here we see a more pronounced decay, reaching the zero level at n = 4. Consequences of these data with respect to the preferential formation of larger ensembles will be discussed later in this section.Finally, the surface energy increases with increasing Pt content in the surface layer, due to the stronger Pt-Pt bonds that have to be broken upon cleaving the crystal.
The PtAg/Ag 1L /Pt 3L surface alloys with a single Ag sublayer are noticeably more stable (more negative DH) than the PtAg/ Pt 4L surface alloys.In Fig. 1, the first/last points correspond to the values of the Pt/Ag 1L /Pt 4L and the Ag 2L /Pt 3L systems, respectively.In this case, we find pronounced deviations from a strictly linear increase in the formation enthalpy, with almost constant values up to n = 4, followed by a slow increase.Actually, it looks like there is a minimum at Y Pt = 3/9.Reasons for this very different behavior, also compared to the PtAg/Ag 2L / Pt 2L system, can only be speculated upon in the moment.Also for the surface energy we find almost constant values for the Pt 1 monomer up to the Pt 4 tetramer (x Z 0.55) (Table 2).
For the surface alloys with two Ag sublayers, we again find an almost linear relationship between the number of exchanged Pt surface atoms, but with a lower slope than for the PtAg/Pt 4L surface alloys.In this case, the systems with x = 0 and x = 1 in Fig. 1 correspond to the Pt 1L /Ag 2L /Pt 2L and Ag 3L /Pt 2L systems in Table 1, respectively.For a pure Ag surface layer (x = 1), all three surface alloys exhibit about the same formation enthalpy, which reflects the similar formation enthalpies of the Ag nL /Pt (5Àn)L systems with n = 1 to n = 3 (see Table 1).Except for the first and second data point, the formation energies are smaller (less negative) than for the PtAg/Ag 1L /Pt 3L systems, and for all data points higher than for the PtAg/Pt 4L surface alloys.The higher (more negative) values than for the PtAg/Pt 4L surface alloys reflect a stronger bonding between Pt in the surface layer and Ag in the sublayer as compared to the mean value of the bulk energies (see eqn (1)).Apparently, this effect is more pronounced for the PtAg/Ag 1L /Pt 3L than for the PtAg/Ag 2L /Pt 2L system, which can be understood based on vertical ligand effects.Due to the stronger Ag-Pt bonding between the second and third layer (from the top) in PtAg/Ag 1L /Pt 3L compared to Ag-Ag bonding in the other system, bonding between the Ag subsurface layer and the surface layer is relatively stronger in the PtAg/Ag 1L /Pt 3L system.The lower slope compared to PtAg/Pt 4L indicates that the energy gain per exchanged surface Pt, relative to the change in bulk energies, decreases less (more negative DH) with increasing Pt content than for the PtAg/Pt 4L system.For PtAg/Ag 1L /Pt 3L , this effect is apparently in between the other two cases, and not linear.Finally, for the surface energy we find a slow increase from the Pt 1 monomer to the Pt 4 tetramer (Table 2).
For all systems considered, the surface alloys are stable with respect to phase separation (negative DH).Most stable is the PtAg/Ag 1L /Pt 3 system, although the differences between the different systems (PtAg/Pt 4L , PtAg/Ag 1L /Pt 3L , PtAg/Ag 2L /Pt 2L ) are small.In a simple picture, one would expect a higher formation enthalpy for the PtAg/Ag 1L /Pt 3L system compared to PtAg/Pt 4L , based on vertical ligand effects and bond order conservation, which is obviously not the case.Again, we expect this to result from contributions from the Pt-Ag interface(s).Using RPBE/ PAW, there are no differences in the general trends (see Table S2, ESI †).There are some differences in individual results, but these are at the limit of the accuracy of the calculations.
These results can be compared with those obtained in a recent in situ microscopy study of alloying and de-alloying of Ag and Pt in Ag films on Pt(111) 33 and of a combined Monte Carlo and DFT study of segregation phenomena in Pt/Ag(111) systems. 35Exploring the growth and alloying/dealloying behavior of Ag on Pt(111) at about 800 K, Jankowski et al. observed a complex dynamic behavior with increasing Ag deposition, confirming alloy formation in the first monolayer and subsequent dealloying at Ag coverages of more than 70% of a monolayer equivalent, by formation of monolayer islands with rather large Pt-rich central areas and Ag-rich rim areas. 33The observation of PtAg monolayer surface alloy formation even at these temperatures underlines that this is not mainly due to kinetic limitations, but (largely) reflects thermodynamics.The large-scale 2D phase separation in the monolayer islands can hardly be compared with the present calculations due to the unknown composition of the original surface layer.Our calculations only covered the case of a surface layer on one or two Ag sublayers (see above), which is likely not the case in those experiments.Comparing the present results of a tendency towards monolayer surface phase separation with those reported by Hua et al., 35 they do not seem to agree, since these authors arrived at high Pt 1 monomer surface concentrations for Pt/Ag(111) systems with low Pt bulk concentrations (o8%).In a more detailed look one has to keep in mind, however, that their simulation results were obtained for rather high temperatures, under dissolution -surface segregation equilibrium conditions, where phase separation is entropically disfavored.Furthermore, the formation of Pt 1 surface monomers was supported by CO adsorption, which, as will be shown in Section 3.3, is strongest on Pt 1 monomers.Considering these aspects, there is no discrepancy to the present results for formation enthalpies and surface energies.
Finally, comparing these trends with those obtained for PdAg/Pd(111) surface alloys, 23 the general trends are very similar, but the effects of the Ag underlayer(s) are much more pronounced for PdAg/Pd(111) than for PtAg/Pt(111), which we tentatively attribute to differences in the interface effects.
In total, these data provide quantitative insights into the relative stabilities of different pseudomorphic surface configurations, starting from Ag film covered surfaces via surfaces terminated by one or two Pt(111) layers on one or two Ag underlayers via Pt x Ag 1Àx /Pt(111) monolayer surface alloys to finally Pt x Ag 1Àx /Pt(111) monolayer surface alloys supported on one or two Ag layers, all of them based on a Pt(111) substrate.They not only confirm the experimentally well-known tendency towards Ag termination of Ag containing surface regions, but also allow us to differentiate between different configurations.Furthermore, they also underline the need for improved calculations of the surface energy, where this is determined free from contributions from the bottom side of the slab.

Electronic properties of silver/platinum bimetallic surfaces
Since it is often used as a measure of the adsorption energy and therefore also of the chemical reactivity of catalyst surfaces, based on the d-band model and on the Sabatier principle, 54,55 we evaluated the local density of states (LDOS) on the Pt atoms in the topmost layer for the three surface alloys with Pt 1 , Pt 2 , and Pt 3 ensembles in the surface layer (Y Pt = 1/9, 2/9, 3/9), respectively, on the pure Pt 4L slab, and with one or two Ag sublayers in between.Fig. 2a shows the local DOS of the d-states projected on the Pt surface atoms in the pure Pt(111) surface and in the three corresponding surface alloys.Since this was already discussed in detail previously, 23 we here only summarize the main findings.While for the pure Pt(111) surface the center of the d-band is located at À2.07 eV, surface alloying with Ag leads to an upshift of the center of the d-band on the Pt atoms in the topmost layer, to À1.49 eV for Pt 1 Ag 8 , to À1.65 eV for Pt 2 Ag 7 and to À1.72 eV for Pt 3 Ag 6 (see Table 3).According to the d-band model, this upshift should correspond to an increase in CO adsorption energy on the Pt ensembles, which fully agrees with experimental findings. 25Increasing the size of the Pt ensembles in the overlayer down-shifts the center of the d-band and brings it closer to that in pure Pt(111).Though not very big, these shifts are significant when compared with shifts reported for structural effects, e.g., for steps 56 or for lattice strain effects. 11The catalytic activity, in turn, would depend on the optimum adsorption strength for a given reaction.As an example, for the oxygen reduction reaction (ORR) the optimum oxygen adsorption strength was reported to be for a slightly weaker bonding than on Pt(111). 57Hence, in that case the Pt 1 Ag 8 /Pt(111) sites would be less active than Pt(111) sites.This behavior is quite different to the trend in PdAg/Pd(111) surface alloys. 23First of all, the shifts are much less   23 This leads to a maximum shift of the d-band center of about 0.2 eV, while for PtAg/Pt(111) it reaches up to 0.6 eV.Thus, Ag induced changes in the activity are expected to be much more pronounced for the PtAg/Pt(111) surface alloys than for PdAg/Pd(111), depending on the optimum adsorption strength and thus on the optimum position of the d-band center.
Next, we also evaluated the changes in the LDOS on the Pt atom in a Pt 1 Ag 8 surface alloy upon varying the number of Ag underlayers from 0 to 2 (see Fig. 2b).The LDOS on a Ag atom in a Ag(111) surface is also included.Replacing the Pt subsurface layer with a Ag layer leads to a strong narrowing of the LDOS at the Pt atom at the surface, and the same effect is also observed for inserting 2 Ag layers.This kind of narrowing can be understood as a result of a weaker coupling of the Pt atom to the Ag subsurface layer as compared to the situation with no Ag layer underneath.Commonly, such effects are described as vertical ligand effects.Together with the narrowing of the LDOS, the center of the d-band shifts from 2.07 eV (Pt(111)) to À1.49 eV (Pt 1 Ag 8 /Pt(111)), À1.51 eV (Pt 1 Ag 8 /Ag 1L /Pt(111)) and À1.56 eV (Pt 1 Ag 8 /Ag 2L /Pt(111)) (see Table 3).
Comparing again with PdAg surface alloys, 23 the effects are again much smaller for these latter surface alloys, similar to the much smaller shifts observed upon varying the Pt or Pd ensemble size.Introducing one or two Ag underlayers for PtAg surface alloys shifts the center of the d-band by 0.02 and 0.07 eV, respectively, towards the Fermi level, while for PdAg surface alloys these shifts were 0.20 and 0.27 eV, respectively.Therefore, we expect also less pronounced changes in the binding energy of adsorbates to the Pd atom in the PdAg alloys with increasing number of Ag underlayers as compared to PtAg surface alloys.Finally, using RPBE/PAW we found the same trends for the shifts of the d-band center, which are listed in Table S3 in the ESI.† The results presented in this section can at least qualitatively be understood in the following picture.Surface alloying a transition metal with a noble metal of larger size, while keeping the lattice constant in the pseudomorphic, topmost layer, leads to a competition between two effects: compressive strain because of the replacement of a smaller transition metal by a larger atom (strain effects) vs. the weaker interaction of the transition metal atoms with neighboring noble metal atoms (lateral ligand effects).In addition, structural effects, e.g., the requirement of a certain minimum ensemble size (ensemble effects), may also play a role. 17In contrast to PdAg/Pd(111) surface alloys, where these effects largely balance each other, and therefore cause only small shifts in the d-band center, ligand effects seem to be more dominant for PtAg/Pt(111) surface alloys.The weaker bonding to neighboring noble metal atoms results in an up-shift of the d-band center from Pt(111) to Pt 1 Ag 8 /Pt(111), followed by a back-shift with increasing number of Pt neighbors.Hence, while we expect strain effects to be of comparable magnitude for Pt and Pd based on their rather similar lattice parameters (3.9231 Å and 3.8898 Å, respectively 49 ), ligand effects seem to be considerably stronger for PtAg/Pt(111) surface alloys than for PdAg/Pd(111) surface alloys.They dominate the modification in the electronic structure and thus also in the adsorption strength.Possible physical reasons for this discrepancy are currently being investigated in more detail.
In total, these data demonstrate subtle differences in the LDOS of the surface Pt atoms in the surface alloys, which we will later compare with the trends in the CO adsorption strength.

CO adsorption on Pt
x Ag 1Àx /Pt(111) surface alloys In this section we will focus on specific aspects of CO adsorption that are relevant for studies of catalytic reactions under relevant reaction conditions, 58,59 but are often overlooked in model studies.This includes CO ad coverage effects, e.g., due to interactions between CO ad molecules adsorbed at higher local coverages, both on small Pt x ensembles and also on larger ensembles.As an extreme case, this also includes the formation of multicarbonyls, where more than 1 CO ad is adsorbed per active (Pt) surface atom.Furthermore, we will explore proximity effects, i.e., interactions between CO ad on ensembles that are separated from each other by at least one Ag atom.In addition to the adsorption energies we also evaluated the adsorption geometry of the CO ad species for additional structural information.Hence, this work expands on our previous study where we focused on low-coverage CO adsorption on PtAg/Pt(111) bimetallic surfaces. 25.3.1.CO ad coverage effects and CO ad -CO ad interactions on small Pt n ensembles.First we will explore the effect of CO ad -CO ad interactions on CO adsorption on small Pt n ensembles with n = 1-3 in PtAg/Pt(111) surface alloys.The CO ad coverage is varied by stepwise addition of 1 CO per 3 Â 3 unit cell, until full coverage with 1 CO ad per Pt surface atom is reached.The resulting configurations are illustrated in Fig. 3, together with the corresponding adsorption energies and tilt angles.The tilt angles refer to the angle of the CO molecule relative to the surface normal, with positive values indicating a tilt away from a neighboring adsorbed CO, due to CO ad -CO ad repulsions, or away from a neighboring Pt surface atom.In addition to the mean adsorption energy per CO ad , E ad , we will also provide values for the differential adsorption energy E ad,diff , as this is decisive for comparison with the experimentally determined activation energy for desorption or with the adsorption energies derived from adsorption isotherms.This is generally defined as , where dE ad is the differential change in total adsorption energy and dy is the differential increase in CO ad coverage.In the present case, where the local CO ad coverage is stepwise increased by 1 CO molecule per step, this changes to where DE ad is the change in total adsorption energy per unit cell and Dn is the increase in CO ad molecules per unit cell, which is typically chosen to be Dn = 1 in the following examples.
This journal is © the Owner Societies 2024 The resulting data show the following trends: First, there is a trend toward increasing adsorption energies with increasing number of Ag neighbors (decreasing Pt ensemble size) in the low coverage limit (1 CO ad per ensemble), even though the increase is rather small.Second, the differences between different adsorption sites are small, very much in contrast to previous findings of distinct ensemble effects for CO adsorption on Pd n ensembles in PdAg/Pd(111) 20,23 and PdAu/Pd(111) surface alloys, [60][61][62] where the most favorable adsorption site changes from on-top for Pd 1 monomers to adsorption on Pd 2 bridge sites for Pd 2 dimers and finally Pd 3 threefold hollow sites on compact Pd 3 trimers.In contrast, for PtAg/Pt(111) adsorption on on-top sites is most favorable for Pt 1 , Pt 2 and Pt 3 ensembles, although the differences are small for the latter two. 25hile these trends were already reported previously, 25 we now furthermore find that there is also little difference in the adsorption energy per CO ad molecule when increasing the CO ad coverage, up to saturation of the respective Pt n ensemble (1 CO ad per Pt surface atom).Using PBE/PAW, going from a single 1 CO ad to two CO ad on a Pt 2 dimer changes the mean CO adsorption energy from À1.98 eV on on-top sites (À1.95 eV on bridge sites) to À1.93 eV, again on on-top sites.This corresponds to a change in the differential CO adsorption energy from À1.98 eV to À1.88 eV (to À1.91 eV if the first CO ad was located on a bridge site), i.e., a change by 0.1 eV or less.The two CO ad are located on top of the two Pt surface atoms, with a slight outwards tilt to reduce repulsions.Occupation of mixed PtAg bridge sites or even PtAg 2 threefold hollow sites is unfavorable.Very similar trends were obtained also when using RPBE/PBA.Also for CO adsorption on compact Pt 3 trimers (Fig. 3), adsorption of a single CO on different sites of the trimer (ontop, bridge or threefold hollow) results in very small differences of the adsorption energy by at most 30 meV.Interestingly, for adsorption on the less symmetric on-top and bridge sites, there is a small tilt off from the vertical configuration towards the other Pt atom(s), indicating that the respective other Pt surface atom(s) have a small effect on the adsorption geometry.Addition of a second and a third CO ad , which both adsorb on on-top sites, leads to changes of the mean CO adsorption energy by about 30 meV per CO, which is very close to the changes observed previously for increasing CO adsorption on a Pt 2 dimer.The differential CO adsorption energies accordingly decay from À1.96 eV (Y CO = 1/3) via À1.84 eV (Y CO = 2/3) to finally 1.81 eV (Y CO = 1).Also in this case, the two CO ad molecules tilt away from each other to reduce CO ad -CO ad repulsions.
The decrease in the adsorption energy of the first CO ad on the same adsorption site with increasing size of the Pt ensemble can in principle be caused by lateral ligand effects or strain effects. 25his will be discussed in more detail at the end of the next section.
3.3.2.CO ad coverage effects and CO ad -CO ad interactions on larger Pt n ensembles.To mimic larger ensembles, we used linear stripes of 3 or 6 Pt atoms per 3 Â 3 unit cell, which represent infinitely long strings or double strings of Pt surface atoms along the h110i direction, separated by 1 or 2 strings of Ag surface atoms (Fig. 4 and 5).These strings were populated with a stepwise increasing number of adsorbed CO molecules.Starting with a single CO ad per unit cell on a single string of Pt atoms (Fig. 4), there is little difference in the adsorption energies between on-top adsorption and adsorption on a bridge site.This closely resembles our findings for CO adsorption on a Pt 2 dimer, where the difference in adsorption energy (on-top or bridge) was equally small.The adsorption energies are slightly lower than on the Pt 2 dimer, in full agreement with a decay in the Ag induced stabilization of the Pt-CO bond discussed in the previous section.Addition of a second CO ad molecule results in a slightly lower (mean) adsorption energy, independent of whether the two molecules are adsorbed on on-top sites or on bridge sites.Again, the effective repulsion is more obvious when using the differential adsorption energy, which decreases from À1.93 eV (À1.95 eV for bridge bonded CO ad ) to À1.83 eV (À1.79 eV for bridge bonded CO ad ).Here we notice also a slight outwards tilt of the CO ad molecules along the h110i direction to reduce the mutual repulsions.Finally, adding a third CO ad results in a more significant decrease in the (mean) adsorption energy to À1.67 eV (À1.66 eV), independent of the adsorption site, which corresponds to a rather strong decrease of the differential adsorption energy from À1.83 eV (À1.79 eV) to À1.25 eV (À1.24 eV).In addition to electronic effects, this decrease in adsorption energy is partly caused by the fact that the CO ad  S4.
molecules can no longer reduce direct repulsions by tilting away from each other along the string direction.Instead, repulsions could be reduced only by zig-zag type outwards tilts, away from the Pt string, as observed, e.g., in the 2 Â 1 structures of CO on Ni(110) and Pd(110). 63,64Unfortunately, within the symmetry and size of a 3 Â 3 cell it was not possible to model these kinds of tilts.
Similar calculations were also performed for a double string of Pt surface atoms (Fig. 5).Starting with a single CO ad per double string, we find a further slight destabilization of the Pt-CO bond as compared to CO adsorption on the single string of Pt atoms, independent of the adsorption site.This fully agrees with the explanation of dominant Ag-induced lateral ligand effects, where replacement of a neighboring Ag surface atom by a Pt atom destabilizes the Pt-CO bond.As before, the differences between different adsorption sites are very small, with adsorption on on-top sites or bridge sites along the row (closest to the Ag atoms) being most favorable.Adding a second CO ad leads to a slight lowering of the mean adsorption energy, by about 0.05 eV when comparing with the equivalent on-top configuration for a single CO ad .Only when adding two more CO ad species do we find a more significant weakening of the Pt-CO bond, with mean bond strengths of À1.74 eV and À1.72 eV, which is somewhere between the bond strengths obtained for 2 and 3 CO ad on the single string of Pt surface atoms (E ad,diff around À1.60 eV).In the configurations with CO ad on nearestneighbor sites, this is accompanied also by a significant tilt of the CO ad away from each other to reduce steric repulsions.Correspondingly, we expect a mean Pt-CO bond strength below À1.65 eV when saturating the double string of Pt atoms by 6 CO ad , and a significantly lower value of E ad,diff .In total, these calculations indicate that even on these larger Pt ensembles coverages up to 1 CO ad per Pt surface atom are possible, as they still allow stabilization of all sites by neighboring Ag surface atoms and lowering of the repulsions by tilting.
In combination with the results reported previously for the low-coverage CO adsorption on Pt x Ag 1Àx /Pt(111), 25 these data demonstrate that ensemble effects do not play a significant role in these systems, as the CO adsorption energy hardly changes for CO adsorption on Pt 1 monomers, Pt 2 dimers or compact Pt 3 trimers.Secondly, furthermore, CO ad coverage effects seem to have little effect.Changes in the CO adsorption energy with increasing coverages, up to a local coverage of Y CO = 1, are very small, at least for small Pt n ensembles.Hence, the effective CO ad -CO ad interactions must be small.This agrees perfectly with previous experimental findings from temperature programmed desorption experiments where peak broadening with increasing coverage was rather small, 25 and also from equilibrium measurements. 34Thirdly, the fact that CO adsorption on these pseudomorphic Pt n ensembles embedded in a matrix of Ag surface atoms is significantly stronger than adsorption on  S4.S4.
This journal is © the Owner Societies 2024 Pt(111), indicates that adsorption on Pt atoms in the surface alloy is dominated by lateral ligand effects rather than strain effects.In the present case of a pseudomorphic overlayer, we expect compressive strain induced by the larger Ag surface atoms, which would lead to a destabilization of the Pt-CO bond. 11Such compression is indeed observed in the calculations, which show a decrease of the Pt-Pt surface bond from 2.81 Å for Pt(111) to 2.71 Å in the Pt 2 dimer in Pt 2 Ag 7 /Pt(111) and in the hcp Pt 3 ensemble in Pt 3 Ag 6 /Pt(111) (2.73 Å in the fcc Pt 3 ensemble).This destabilization is apparently overcompensated by lateral ligand effects, where the weaker Pt-Ag bonding compared to Pt-Pt results in a stabilization of the Pt-CO bond. 65Considering the increase in bond strength when going from CO adsorption on Pt(111) to CO adsorption on individual Pt surface atoms or small ensembles (Pt 2 , Pt 3 ) in Pt x Ag 1Àx / Pt(111) surface alloys, ligand effects must be dominant in this system.In absolute terms, however, these changes are rather small.This also fits well with the trend observed experimentally, where the shifts in the desorption temperature with increasing CO ad coverage are rather small. 25,34inally, we compare these results with similar ones obtained for CO adsorption on Pd x Ag 1Àx /Pd(111) surface alloys. 23First of all, changes in the electronic properties of the Pd surface atoms with increasing Pd n ensemble size are very small, much smaller than for Pt n , as indicated by the small shift in d-band center. 23n the other hand, CO adsorption on PdAg/Pd(111) surface alloys shows much more pronounced shifts in the adsorption energy for increasing CO ad population of the Pd n ensembles, while changes in the calculated CO adsorption energies for the same site are small.The wide range in adsorption energies is reflected also by the considerable width of the CO desorption peaks in TPD measurements. 20The apparent discrepancy between these findings can be explained by pronounced ensemble effects on Pd-containing surfaces, which result in a significant increase in Pd-CO bond strength when changing from on-top adsorption on Pd 1 monomers (À1.37 eV), which dominates at lowest Pd concentrations, to bridge site adsorption on Pd 2 dimers (À1.64 eV), which becomes more dominant upon increasing Pd concentration, to finally adsorption on threefold hollow sites on compact Pt 3 trimers and larger ensembles (À1.97 eV), which is the dominant site at higher Pd surface concentrations. 20,23This spread is in complete contrast to the negligible changes in adsorption energies observed on PtAg/ Pt(111) monolayer surface alloys.Furthermore, for larger ensembles, adsorbate-adsorbate repulsions also play a role, while for small ensembles the displacement of adsorbed CO molecules by post-adsorbing CO molecules on less favorable adsorption sites, e.g., the replacement of a more strongly bound bridge-bonded CO ad on a Pd 2 dimer by two less strongly bound CO ad on on-top sites, can become important.Such effects are much more pronounced for small Pd n ensembles than for small Pt n ensembles (see Fig. 3).8][69] Furthermore, adsorption on Pd(111) also exhibits pronounced repulsive interactions, as indicated by the very broad TPD peaks at high coverages, 66 while this is much less the case for CO desorption on Pt(111). 39,40.3.3.Multicarbonyl formation and proximity effects on Pt x Ag 1Àx /Pt(111) surface alloys.The formation of multicarbonyl species with more than one CO ad per active Pt atom was investigated for a Pt 1 Ag 8 /Pt(111) surface alloy.Such kind of multicarbonyl species has been reported for numerous metals, 70,71 mainly in catalysis studies performed under ambient conditions.They were often proposed to form on undercoordinated sites such as edge or corner sites on metal nanoparticles, where steric repulsion between these species can be reduced, in addition to the generally stronger binding of adsorbates on these sites.65 Repulsion between closely neighbored CO ad species can also be reduced for adsorption on individual active surface atoms or small ensembles of surface atoms embedded in a matrix of inert surface atoms.In the present surface alloys, Pt surface atoms in a Pt 1 Ag 8 /Pt(111) surface alloy would be the most simple example (see Fig. 6).For the first CO ad the on-top site with a vertical CO ad is the most stable adsorption site and configuration.Adding a second CO ad , the most stable configuration involves adsorption on two bridge sites or threefoldhollow sites opposite from each other on both sides of the active Pt atom.For adsorption on threefold-hollow sites this must include one fcc and one hcp site, respectively.In both cases the CO ad are significantly tilted, by about 171 with respect to the surface normal away from each other, to increase the separation between the two molecules, in particular between the two O atoms. UsingPBE/PAW, the adsorption energy (per CO ad molecule) decreases from 2.06 eV to 1.04 eV.The drastic decrease in adsorption energy is even more obvious when using the differential adsorption energy E ad,diff , which for the addition of a second CO ad molecule is only 0.02 eV.With this low value, the formation of multicarbonyl species on these sites can be excluded at finite temperatures, even though these Pt 1 sites are more strongly binding than Pt atoms on Pt(111) and even though the CO ad molecules can reduce the CO ad -CO ad repulsion by a significant tilt.This agrees with the experimental situation in so far as the formation of multicarbonyl species has rarely been proposed for Pt nanoparticle catalysts, 41,42 in contrast, e.g., to multicarbonyl formation on Ru nanoparticles.[71][72][73] Finally, we would like to note that this conclusion is purely based on the calculated adsorption energy, i.e., it does not As a last point we explored proximity effects, i.e., the effective interaction between CO ad molecules adsorbed on two active surface atoms which are not directly neighbored, but also not very distant.This was modelled by CO adsorption on two separated Pt 1 monomers, as illustrated in Fig. 7.Note that this configuration with two next-nearest neighbor Pt surface atoms along the h210i direction represents the one with the closest Pt-Pt possible for non-neighbored surface atoms.Starting with adsorption of a single CO on one of the two Pt surface atoms in Fig. 7 (left), we find that electronic Pt-Pt interactions do not affect the CO adsorption energy on one of these Pt surface atoms.With 2.06 eV it is identical to that obtained for adsorption on a Pt 1 monomer in a Pt 1 Ag 8 /Pt(111) unit cell (Fig. 6).Furthermore, adsorption of a second CO on the second Pt 1 monomer has no measurable effect on the CO adsorption energy. Thonly difference is a very small tilt of the two CO ad , both pointing with the same orientation along the close-packed h210i direction, most likely due to the surrounding non-symmetric atom distribution rather than by mutual repulsions or attractions.Apparently, these kinds of proximity effects are negligible for the present system.This also fits well with the observation of rather small effects (B50 meV per CO ad ) for adsorption of 2 CO on neighboring Pt surface atoms in a Pt 2 dimer (Fig. 3).
Going to higher coverages of non-neighbored Pt surface atoms, which is illustrated for 3 Pt surface atoms per unit cell in Fig. 7, we again find small effects from effective Pt-Pt interactions.This is indicated by the further decrease of the adsorption energy of the first CO to À2.02 eV, while in the previous case (adsorption on two Pt monomers) this remained at À2.06 eV.Most simply, the lower CO adsorption energy can be explained by a lower contribution from stabilizing (longer-range) ligand-plus-strain effects, due to replacement of a larger Ag surface atom by Pt.Additional adsorption of two further CO on the remaining two Pt surface atoms results in an increase of the mean CO adsorption energy to À2.08 eV per CO ad .This corresponds to an increase in the differential adsorption energy to À2.11 eV per CO ad for the 2 nd and 3 rd CO ad .Different from the previous case of two Pt 1 monomers in the 3 Â 3 unit cell there is no measurable tilt of the CO ad .Most simply, the absence of a tilt is caused by the presence of two neighboring CO, whose effects compensate each other.For the increase in adsorption energy, which differs from the previous cases investigated here and which despite its small size is significant, we speculate that this is due to attractive through-bond interactions, which overcompensate repulsive direct interactions.
The results presented in this section first of all indicate that there is no driving force for the formation of multicarbonyl species on these PtAg/Pt(111) surface alloys, which seems to be valid also in more general for Pt nanoparticles.Second, proximity effects, indicating interactions between CO ad molecules adsorbed on non-neighbored Pt surface atoms, i.e., Pt surface atoms that are separated by at least one inert surface atom, seem to be negligible or very small for this system.

Conclusions
Using periodic density functional theory calculations and extending our previous combined experimental and theoretical work, 25 we could provide detailed insights on the stability, electronic properties and CO adsorption behavior of pseudomorphic bimetallic PtAg surfaces supported on a Pt(111) substrate.These include surfaces covered by pseudomorphic Ag film of 1-3 layers in thickness, Pt(111) layers with one or two Ag layers underneath, and monolayer PtAg surface alloys, either on Pt(111) or with 1 or 2 Ag layers underneath and then Pt(111).In addition to finding that Ag segregation is always favorable compared with Ag underlayer formation, and intermixing into surface alloys is more favorable than separation into Ag and Pt surface phases, the data provide quantitative information on the stability of different surface configurations.Calculations of the LDOS on the Pt surface atoms in the monolayer surface alloy reveal an increasing up-shift of the d-band center with increasing number of neighboring Ag surface atoms.Furthermore, they show a decoupling of the Pt surface atoms from the bulk in a Pt 1 Ag 8 / Pt(111) monolayer surface alloy in the case of Ag underlayers beneath, with a dominant state about 1.5 eV below the Fermi level.
Systematic evaluation of the CO adsorption properties on various different Pt n ensembles (n = 1-3) in a matrix of Ag surface atoms reveal very small changes in the adsorption energy with increasing number of CO ad on the ensemble, up to 1 CO ad per Pt surface atom.Furthermore, the differences between different adsorption sites for CO on Pt 2 dimers and compact Pt 3 trimers are very small.On the other hand, the data show a weak, but clear increase of the adsorption energy on Pt with increasing number of neighboring Ag surface atoms, which is explained by dominant electronic ligand effects.The results are consistent with experimental findings of rather narrow desorption peaks and the (coverage dependent) occupation of different adsorption sites.
The formation of multicarbonyl species, with two CO binding to a single, individual Pt surface atom, can be excluded at finite temperatures.With a differential adsorption energy of 0.02 eV, the second CO ad is too weakly bound to be populated under these conditions, in good agreement with experimental findings for supported Pt nanoparticles.Despite the stronger adsorption on the PtAg 6 site and despite the possibility of lowering the CO ad -CO ad repulsions by tilting the molecules away from each other, the effective repulsions, both by direct repulsion and by indirect electronic effects, are too strong to facilitate multicarbonyl formation.On the other hand, proximity effects, reflecting interactions between CO ad on two Pt surface atoms separated by at least one Ag surface atom, were found to be negligible, consistent with the result that even for 2CO ad on a Pt 2 dimer the effective interaction is negligible.Overall, the interactions between CO ad species on small Pt n ensembles (n = 1-3) and also on one-dimensional Pt strings in PtAg/Pt(111) surface alloys seem to be very weak.
Finally, a more general understanding of bimetallic surface properties was obtained by systematic comparison with similar data obtained previously for bimetallic PdAg/Pd(111) and Ag/ Pd(111) surfaces, which despite their close structural similarity exhibit characteristic differences in their stability, electronic properties and CO adsorption behavior.The results derived in this model study provide detailed insights into the chemistry of bimetallic PtAg surfaces, which is also useful for a better understanding and possible improvement in the performance of bimetallic PtAg catalysts.

Fig. 1
Fig. 1 Formation enthalpies DH in meV Å À2 for the Pt 1Àx Ag x /Ag nL /Pt(111) monolayer surface alloys as a function of the substitution ratio x.The index n L denotes the number of pseudomorphic Ag layers between Pt(111) and the monolayer surface alloy, the substitution ratio x defines the number of Pt atoms replaced by Ag atoms in the topmost layer relative to the total number of surface atoms (here: 9).The 1 À x scale at the upper x-axis describes the fraction of Pt atoms in the topmost layer, which also corresponds to the Pt coverage Y Pt in the topmost layer.The results were obtained using PBE/PAW, for RPBE/PAW values see Fig. S1, ESI.†

Fig. 2
Fig. 2 Local density of d-states calculated by projection of the d-wave functions onto the atomic orbitals at a Pt top site (for Ag(111) on a Ag top site).In the upper box (a) we compare the pure Pt(111) surface and the PtAg/Pt 4L surface alloys containing Pt 1 , Pt 2 and Pt 3 ensembles.All energies refer to the Fermi energy.Up-arrows denote the positions of the d-band centers for the Pt(111) surface and down-arrows denote the positions of the d-band centers for the surface alloys.In the lower box (b) we compare the LDOS at the Pt atom of PtAg surface alloys containing only Pt 1 monomers for different numbers of Ag underlayers (0, 1, 2, see figure).For comparison we also show projected LDOS for the pure Pt(111) and Ag(111) surfaces.Up-arrows denote the center of the band of the pure surfaces, while down-arrows denote the center of the band of the surface alloys.The results presented here were obtained using PBE/PAW.

Fig. 3
Fig. 3 Schematic representation of CO adsorption on different sites of Pt 1 , Pt 2 and Pt 3 ensembles in Pt 1 Ag 8 /Pt(111), Pt 2 Ag 7 /Pt(111) and Pt 3 Ag 6 /Pt(111) surface alloys, respectively (energies in eV per molecule, angles refer to the tilt of the CO molecules with respect to the surface normal (see the text), for further details see the figure).Energies are calculated via PBE/PAW, and values obtained via RPBE/PAW are listed in the ESI † in TableS4.

Fig. 4
Fig. 4 Schematic representation of increasing CO adsorption on a string of Pt surface atoms in a PtAg/Pt(111) monolayer surface alloy along a close-packed h110i direction, both for on-top adsorption and for adsorption on bridge sites (energies in eV per molecule, angles refer to the tilt of the CO molecules with respect to the surface normal (see the text), for further details see the figure).Energies are calculated via PBE/PAW, and values obtained via RPBE/PAW are listed in the ESI † in TableS4.

Fig. 5
Fig. 5 Schematic representation of increasing CO adsorption on a double string of Pt surface atoms on a PtAg/Pt(111) monolayer surface alloy along a close-packed h110i direction, both for on-top adsorption and for adsorption on bridge sites (energies in eV per molecule, angles refer to the tilt of the CO molecules with respect to the surface normal (see the text), for further details see the figure).Energies are calculated via PBE/PAW, and values obtained via RPBE/PAW are listed in the ESI † in TableS4.

Fig. 6
Fig. 6 CO adsorption on Pt 1 in Pt 1 Ag 8 /Pt(111) (energies in eV per molecule, angles refer to the tilt of the CO molecules with respect to the surface normal (see the text), for further details see the figure).Energies are calculated via PBE/PAW, and values obtained via RPBE/PAW are listed in the ESI † in TableS4.
As an extreme example, multicarbonyl species with more than one CO bound to a Pt atom were proposed by Tsang et al. and Oduro et al. for Co-doped Pt catalysts.
for the layered systems, where E Ag nL /Pt(111) and E Pt(111) are the total energies of the Ag nL /Pt(111) and Pt(111) slabs per surface atom, respectively.x denotes the fraction of Pt atoms replaced by Ag atoms in the slab relative to the number of surface atoms, E Ag and E Pt are the bulk energies per atom of these metals, and DH represents the mean formation enthalpy per surface atom.

Table 1
Formation enthalpies (DH), surface energies (E S ) and estimated surface energies of the top surface (E S,top , see text) for pure Pt(111) and Ag(111) surfaces and for Ag nL /Pt(111) pseudomorphic overlayers, varying the number of Ag layers (PBE/PAW).Surface energies and formation enthalpies are given in meV Å À2 .Ag 5L /Pt 0L denotes the pure Ag(111) surface, keeping the lateral lattice parameter of Pt(111), whereas Ag 5L denotes the pure Ag(111) surface with the Ag lattice parameter.Note the conversion: 1 meV Å À2 E 16.02 Â 10 À3 J m À2 .For the systems with a Pt(111) lattice, 1 meV per surface atom corresponds to 0.1462 meV Å À2 for a slab of identical thickness.Results obtained via RPBE/PAW are shown in TableS1in the ESI

Table 2
Mean enthalpies of formation (upper lines), differential enthalpies of formation (middle lines), and surface energies (bottom lines), for the PtAg/Ag nL /Pt(111) surface alloys indicated.Surface energies and formation energies are given in meV Å À2 (1 meV Å À2 E 16.02 Â 10 À3 J m À2 ).For the systems with a Pt(111) lattice, 1 meV per surface atom corresponds to 0.1462 meV Å À2 for a slab of identical thickness.Results were obtained using PBE/PAW, and data using RPBE/PAW are collected in TableS2in the ESI

Table 3
Center of the d-band on the Pt surface atoms and relative shifts compared to Pt(111) in eV for different PtAg surface alloys.The results were obtained using PBE/PAW, and data using RPBE/PAW are collected in TableS3in the ESI Pd(111), where the center of the d-band on the Pd surface atoms is rather similar for Pd 1 Ag 8 , Pd 2 Ag 7 and Pd 3 Ag 6 systems, at about À1.34 eV, upshifted from the value for Pd(111) of À1.53 eV.