The Cu(110) (5 × 2)pg-methoxy structure

So far, the possibility of coadsorption of two different sites, as suggested in several previous experimental studies, has not been addressed, nor the related issue of the structure of the ordered (5×2) phase that has been reported in LEED and STM experiments [132]. Notice that although the long-range order of the surface investigated experimentally by PhD [11] was not known, it is known that methoxy can form this ordered (5×2) phase on Cu(110), and even if the surface was not well-ordered, the main local structural ingredients were likely to be closely similar to those of the ordered phase.

5.1.5.1 Long range structural information

The reported LEED pattern of the (5×2) phase showed certain diffracted beams to be missing at normal incidence [131], characteristic of the presence of glide

Figure 5.3: Plan view of the structure of the Cu(110) (5×2)-methoxy adsorption structure proposed on the basis of earlier DFT calculations [142]. The white rectangle shows the (5×2) unit mesh. The Cu adatoms and substrate in this structure, but not all atoms in the methoxy adsorbates, show the glide symmetry of the pg space group; the white dashed lines show the locations of these glide symmetry lines.

symmetry lines in the [001] direction [148]; this implies that the space group of this ordered phase is either pg or 2 mg, and a later STM study narrowed this down to pg [137]. One further important result obtained from STM [131] is that islands of (2×1)-O, surrounded by the ordered (5×2)-methoxy phase, could be eroded and converted into the (5×2)-methoxy phase by further exposure to methanol, without the creation of large defect regions or of adatom islands. This strongly suggests that most or all of the 0.5 ML of Cu adatoms in the (2×1)-O phase are retained in the (5×2)-methoxy structure. Indeed, through the use of experiments to monitor by STM step movements resulting from desorption or dissociation of methoxy from the surface, Leibsle et al. [132] estimated the Cu adatom coverage

in the (5×2)-methoxy phase to be not less than 0.4 ML. A coverage of 0.5 ML of adatoms would imply 5 such Cu atoms per (5×2) unit mesh, but an odd number is incompatible with the glide symmetry. As a result, Leibsle et al. [132] proposed several possible structures that incorporated either 6 or 4 Cu adatoms in each unit mesh. All these models assume 4 methoxy species per unit mesh (symmetry similarly dictates that there must be an even number). More recently, Sakong and Gross [142] have investigated these proposed structures, involving several different local methoxy adsorption sites, using DFT calculations; they concluded that the lowest-energy structure is that shown in Figure 5.3. This is a slight modification of one of the structures originally proposed by Leibsle et al. in which the methoxy species occupy long-bridge sites on pairs of adatoms, and three-fold hollow sites in which bonding is to both one Cu adatom and two Cu atoms in the underlying surface. The DFT-optimised structure shown in Figure 5.3 differs from this initial proposal in that half of the methoxy species that originally occupied 3-fold hollow sites have moved to two-fold coordinated edge-bridge sites. Notice that the shift of some molecules from the 3-fold hollow to edge bridge sites formally breaks the glide-line symmetry, as does the tilting of some of the molecules. However, it is certainly possible that the associated LEED pattern would still show very weak intensities in the otherwise symmetry-forbidden diffracted beams, because the arrangement of the more strongly-scattering Cu adatoms does retain the required symmetry. Sakong and Gross reported that the average adsorption energy per methoxy species in this structure was only 40 meV lower than in the short-bridge site on a non-reconstructed Cu(110) surface.

5.1.5.2 DFT results

A PhD simulation based on the (5×2) structure proposed by Sakong and Gross [142] failed to achieve a R-factor value lower than 0.97 [11], even after searching for minor changes to minimise this value. Nevertheless, as a starting point to the DFT investigation of the (5×2) phase, a geometric optimisation of this struc- ture was conducted. Figure 5.3 shows the exact geometry obtained from these calculations. The average adsorption energy per methoxy species calculated for this structure, including the energy cost of the Cu adatom structure, is 1.69 eV (see also Table 5.2), a value that is 0.39 eV smaller than for methoxy adsorbed in short-bridge sites on the non-reconstructed Cu(110) surface. This contrasts with an energy difference quoted by Sakong and Gross for the same two structures of only 40 meV. It seems likely, however, that these authors did not include the

Table 5.2: Comparison of the results for the optimised structures found in DFT calculations for different adsorption sites and reconstruction models of methoxy adsorbed on Cu(110) in the (2×2) and (5×2) unit meshes using both five-layer and seven-layer slabs to represent the substrate. Ea and Ea0 are as defined in

section 5.1.3 and in the caption to Table 5.1.

5-layer substrate slab 7-layer substrate slab

Model Ea (eV) Ea0 (eV) Ea (eV) Ea0 (eV)

Short bridge (0.25 ML - (2×2)) 2.08 2.08 2.11 2.11 Short bridge (0.50 ML - c(2×2)) 2.13 2.13 2.15 2.15 [1¯10] added-row short bridge 2.26 2.16 2.25 2.16

(5×2) - mixed sites 2.02 1.69 2.02 1.69

(5×2) - new model (Figure 5.4a) 2.15 1.98 2.16 1.99

energy cost of the Cu adatom structure; without this correction, the average ad- sorption energy in the present calculations is 2.02 eV, a value that is only 60 meV per methoxy species less than that of the unreconstructed short-bridge geometry. Nevertheless, the key conclusion is that the energy cost of the Cu adatom distri- bution in this structure is such that this (5×2) structure is clearly energetically unfavourable.

In view of both the incompatibility of this model of the (5×2) phase with the experimental PhD results [11], and the energetic considerations, alternative models have been explored that are based on occupation of short-bridge sites only, but involve half of the methoxy species occupying sites above sections of [1¯10] adatoms rows. As in the originally-proposed models of Leibsle et al., a coverage of 4 methoxy species per unit mesh has been assumed, and either 4 or 6 Cu adatoms. The geometrically-optimised versions of the two proposed structures of this type that have been considered are shown in Figure 5.4. Both structures are fully consistent with the pg space group, and are energetically very significantly more favourable than the structure of Figure 5.3. For the model incorporating 4 Cu adatoms per unit mesh (see Figure 5.4(a)), the average adsorption energy of the methoxy species on the reconstructed surface is 2.15 eV, a value that falls to 1.98 eV after taking account of the energy cost of the Cu adatom distribution. For the model incorporating 6 Cu adatoms per unit mesh (see Figure 5.4(b)), the reconstruction-adjusted average adsorption energy, Ea0, is 110 meV lower; the

Figure 5.4: Plan view of the two mixed-short-bridge models of the Cu(110) (5×2)- methoxy adsorption structure proposed here on the basis of the new DFT calcu- lations. The white rectangle shows the (5×2) unit mesh. The dashed white lines show the glide symmetry lines of the pg space group.