Missing row model

Although it was shown in section 7.3.3.1 that a buckled c(2×2) model cannot account for the observed c(6×2) phase of methanethiolate adsorption on Cu(100), there is another plausible model for this phase, proposed by Kondoh et al. [192] and introduced in section 7.1, in which every third thiolate row of a c(2×2) structure is vacant. Shown in Figure 7.5 is the energetically preferred solution of this type, with a reaction energy, as defined in section 7.2, of 860 meV per thiolate. Indeed, this relatively large reaction energy corresponds to a lower surface energy for the system than both the p(2×2) and c(2×2) lowest energy models by 8 meVÅ−2_{. Although all models were initialised without the imposition}

of surface symmetry, as shown in Figure 7.5, this particular optimised modeldoes

contain a mirror plane in the [011] direction. The vacant 4-fold hollows in this missing row structure clearly allow good relief of the surface stress, the two Cu- Cu diagonals under the adsorbed methanethiolate increasing by 1.7% and 1.8%. In contrast to the large compressive stress of the c(2×2) model, the resulting structure exhibits only a very small compressive surface stress of about 0.15 Nm−1 in the principle directions. Another interesting feature of this lowest energy model is that consecutive rows of thiolates are tilted in opposite directions, an effect which maximises the physical separation of the methyl head groups. Furthermore, the azimuthal orientation of the methyl head groups, which are rotated by 60◦ relative to those of the tilted species in the c(2×2) model (see Figure 7.3), also maximises this separation. Indeed, the smallest distance between atomic locations of atoms associated with different thiolates is 2.84 Å for the c(6×2) structure, significantly larger than that found for the c(2×2) structure, 2.17 Å.

A test of the validity of this model can be made by comparing its total energy with that of the lowest energy (3×2) structure with the same adsorbate coverage. As shown in Table 7.1, the surface energy is indeed lower for the c(6×2) structure, by 1.2 meVÅ−2_{. A likely reason for this is that the surface layer Cu atoms, which}

in the c(2×2) reconstruction would be bonded to the ‘missing’ thiolates, can accommodate strain without azimuthally skewing the relative locations of Cu atoms in the two rows under these ‘missing’ adsorbates.

[011] [010] [011] [010] [011] [010] [011] [010]

Figure 7.5: The preferred optimised missing row model for the c(6×2) phase of thiolate adsorption on Cu(100), structural parameters for which are presented in Table 7.1. A c(6×2) unit cell is enclosed by the dashed line while the mirror plane of the unit cell is indicated by the dot-dash line.

7.3.3.3 STM images

Since STM images of the thiolate covered surface are available in the open litera- ture, a comparison can be made between these and theoretical STM images of the proposed c(6×2) missing row structure. Figure 7.6 shows three simulated STM images of the missing row structure (upper row of panels), generated using the Tersoff-Hamann approximation (see section 3.3.5), along with two published STM images of the c(6×2) phase (lower row of panels). The simulated images, which are recorded at distances from the outer Cu layer of the surface in the range 5-9 Å, vary substantially in appearance. In the right panel, which shows the image taken closest to the surface, features that have the distinctive triangular shape of the methyl headgroups are clearly displayed. As the distance from the surface increases, however, the bright STM features become circular (centre panel). At even greater distance from the surface, pairs of features begin to merge, creating the appearance of just two larger elliptical features per (6×2) unit cell.

Attempting to describe the real STM images in Figure 7.6 is a much more formidable task, in part due to limited spacial resolution, in part due to the inhomogeneity of the surface. The c(6×2) regions in both images do clearly contain a regular array of two bright features per (6×2) unit cell, which, in itself,

Figure 7.6: Displayed in the upper row of panels are three simulated images of the c(6×2) missing row structure, recorded at decreasing distance from the surface from left to right, at 9 Å, 7 Å and 5 Å. The bias is -0.33 V. The lower row of panels contains two experimental room temperature STM images, published by Driver et al. in reference [198]. Both images show regions of c(6×2) periodicity, one subsequent to adsorption of DMDS (left panel) using a bias voltage of -0.33 V and tunnel current of 1.1 nA, the other following adsorption of methanethiol (right panel) using a bias of -0.25 V and tunnel current of 1.4 nA. In each image a c(6×2) unit cell is enclosed by a solid white line. The lower left image also contains a region of c(2×2) periodicity running down the image, slightly right of centre. A unit cell of this phase is also highlighted in this image.

is consistent with the large-distance simulated image in the upper left panel of the figure. In the lower left panel of Figure 7.6 however, which contains an STM image of the surface recorded during exposure of a clean Cu(100) surface to DMDS, these strong features are approximately circular, in contrast to the elliptical features of the simulated image. In addition, these features are accompanied by the same number of less prominent, elongated features aligned approximately in the [010] direction. Conversely, in the image in the lower right panel of Figure 7.6, recorded following exposure of the clean surface to methanethiolate, the bright features themselves appear to be elongated in the direction of the diagonal of the (6×2) unit cell, and no weaker accompanying features are visible.

What is clear is that the same reconstruction can produce rather varied STM images depending on the conditions, both in the experimental and simulated cases. A further complication in making a direct comparison between the simu- lated and experimental images comes from the fact that the simulated images are for the optimised ground state structure, whereas the experimental ones consist of a time-averaged representation of the surface at approximately room temper- ature. In this regard, by observing the number of similar structures found in relatively close proximity to the lowest energy structure and by monitoring the slow convergence of the models, the DFT simulations performed in this work have shown that there is a rather flat energy minima in which the orientation of the thiolates lie. Hence, there will certainly have been significant movement of the methanethiolate species at the temperature at which the experimental images were recorded. Almost freely rotating methyl head groups are likely to lead to rounding and smearing of features in the experimental images, while a changing angle of the S-C bond relative to the Cu surface is likely to result in elongation and further smearing of these features.

An additional feature of sequential STM images of this reaction is the emer- gence of metastable regions of c(2×2) periodicity, an example of which can be seen in the centre of the lower left image of Figure 7.6, which are subsequently replaced by the c(6×2) phase. In fact, while the adsorption energy per adsorbate is great- est in the p(2×2) reconstruction, and is thus observed at low coverage, as shown in Table 7.1, the c(2×2) reconstruction incorporating the tilted methanethiolate species has a lower surface energy, albeit by just 0.03 eVÅ−2_{. In order to transform}

the surface from the p(2×2) to the c(6×2) reconstruction, it is necessary to relo- cate 1/3 of the adsorbed methanethiolate species and adsorb an additional 1/3. In contrast, creating the c(2×2) reconstruction from the p(2×2) reconstruction only requires the adsorption of additional methanethiolate species in the four-fold

hollows not already occupied. It appears, therefore, that the energy barrier to restructure the surface is sufficient to prevent the immediate emergence of the c(6×2) phase. Interestingly, since the 0.33 ML c(6×2) phase contains a lower adsorbate coverage than the metastable 0.5 ML c(2×2) phase, the local coverage islowered by the restructuring to the c(6×2) phase. This is rather unusual since one normally associates the saturated surface with the phase of highest coverage. In order for this restructuring to occur, c(2×2) regions must either engulf regions of the surface that still exhibit the lower coverage p(2×2) phase or desorption of thiolates must occur, probably as DMDS.