The Inuence of Surface Structure on Scattering

One of the drawbacks of studying Au(111) is that it is known to reconstruct, resulting in a large surface unit cell of √

3×22[32, 33] or√

3×23[34, 35]. So far, for MD calculations on Au(111), the surface reconstruction appears to have been ignored [3, 4] as the reconstruction requires a very large unit cell to be simulated. As will be seen in the next chapter (see Chapter 4), the MD-simulations performed with EMT-JAWK show some discrepancy to the experimental results. To give an estimate of whether surface reconstruction needs to be included into MD simulations and if its inclusion improves the comparison to the experimental results, I studied the eect of surface reconstruction on nonadiabatic scattering. For this, I used a22×6×4slab and modeled the discommensuration lines according to the structures given by Wang et al. [41] and Hanke et al. [43]. Unlike the other calculations presented here, I used a four-layered slab to reduce the calculation time. To test the inuence of small faults on the surface, I introduced one Au atom onto a6×6 surface, leading to a coverage of 1/36.

The scattering probability when surface reconstruction is included (Tab. 3.24) is only a little higher than for the unreconstructed surface. This can be attributed to structure of the

recon-Table 3.24.: Outcomes (in %) resulting from H atom collision with a Au(111) surface for nonadiabatic and adiabatic (in parenthesis) simulations for dierent surface structures. The incidence energy is E_inc = 3.33eV,θ_in = 45^◦ along the [10¯1]surface direction at 300 K, number of simulated trajectories: nonadiabatic: 10⁶, adiabatic: 10⁵.

Scattering Surface Subsurface Transmission

structure Adsorption Absorption

relaxed 55 (81) 23 (4) 21 (7) 1 (8)

1 adatom 56 22 22 1

reconst. 59 (80) 15 (2) 23 (5) 0 (13)

Table 3.25.: Outcomes (%) of scattering for various scattering events resulting from H atom collision with a Au(111) surface for nonadiabatic and adiabatic (in parenthesis) simulations for dierent surface structures. The Surface-column refers to trajectories wherein the H atoms scattered from 1^st layer of the surface. The Roman numerals refer to the lowest subsurface to which penetration occurred. The incidence conditions are Einc = 3.33eV, θ_inc= 45^◦along the[10¯1]surface direction and6×6×6cell at 300 K, number of simulated trajectories: nonadiabatic: 10⁶, adiabatic: 10⁵.

bounce events penetrating bounces

structure single double multi surface I II III >III relaxed 23 (17) 34 (25) 43 (59) 82 (64) 17 (23) 1 (8) 0 (3) 0 (2)

1 adatom 22 33 45 83 15 1 0 0

reconst. 24 (19) 37 (29) 39 (52) 88 (74) 11 (17) 1 (7) 0 (3) 0 (0)

structed surface (see Fig. 2.1(c) and (d)) that is closer than the unreconstructed surface, making it more dicult for an H atom to go subsurface or to resurface again which agrees well with the higher percentage of non-penetrating trajectories (Tab. 3.25). The probability to remain at the surface is lower than for the unreconstructed case, most probably because the way to populate the surface is to a large part via resurfacing. This is more dicult if the surface is packed denser and overlaps subsurface wells in some regions. The absorption and transmission probability is higher than for the relaxed surface. Although it could be expected that the absorption and transmission probability should be lower for a reconstructed surface, the higher ratio can be explained by the previous observation: fewer H atoms might be able to penetrated, but if the H atoms are hindered from resurfacing again due to the reconstruction on the surface, then the number of atoms that remain subsurface or are transmitted can well be higher than in the

Table 3.26.: Energy loss in % of incidence energy for various outcomes resulting from H atom collision with a Au(111) surface for nonadiabatic and adiabatic (in parenthesis) simulations for dierent surface structures. The mean and maximum energy loss are shown for the total and dierential ELD. Number of simulated trajectories: nonadiabatic: 10⁶, adiabatic: 10⁵. The accuracy for specular scattering (θ_out = 45^◦ φ_out = 60^◦ ([10¯1])) has been reduced to account for the lower signal-to-noise ratio in the dierential ELD.

Total θout= 45^◦ φout= 60^◦

structure Mean Peak Mean Peak

relaxed 35.3 (4.65) 14.0 (1.95) 39 (12) 14 (0.75)

1 adatom 35.6 14.3 39 15

reconst. 32.9 (4.05) 15.2 (2.25) 35 (8.7) 15 (1.1)

108

0.0 0.5 1.0 1.5 2.0 2.5 3.0 0.0

0.1 0.2 0.3 0.4 0.5 0.6 0.7

Eloss(eV)

A.U.

(a)

0.0 0.5 1.0 1.5 2.0 2.5 3.0

Eloss(eV) (b)

-1.0 -0.5 0.0 0.5 1.0 r sin(θ)

75°

60°

45°

30°

0° 15°

-15°

-30°

-45°

-60°

-75°

(c)

Figure 3.45.: Energy loss distribution for (a) total and (b) specular scattering (θout= 45^◦ andφout= 60^◦) for the unreconstructed surface (black), the reconstructed surface (dark green) and the unreconstructed surface with an adatom (dark red). (c) Angular distribution for scattering along the [10¯1]-direction. Negative θout correspond to backward scattering, positive ones to forward scattering. A cosine distribution is given in navy.

relaxed case. The ratio of the bounce events stays roughly the same, but the peak of the total energy loss distribution for the reconstructed surface is shifted to higher energy losses and the shoulder stronger (Fig. 3.45(a), green) while the mean energy loss is slightly lower than for the unreconstructed surface (see also Tab. 3.26).

Having an adatom on the surface does not change the dynamics much, neither with respect to reection nor with respect to ab- or adsorption (Tab. 3.24) or ratio of bounce events (Tab. 3.25).

The form of the energy loss distribution and its energy losses remain almost the same.

The angular distribution of scattering along the [10¯1]-direction is for all three considered surfaces much the same (Fig. 3.45(c)). They are all equally broad and peak at she same outgoing angles, showing barely any dierence amongst one another, although the unreconstructed surface has a peak at θout ∼5^◦, the unreconstructed, reconstructed and surface with adatom all show

Table 3.27.: Outcomes (in %) resulting from H and D atom collisions with a Au(111) surface for nonadiabatic and adiabatic (in parenthesis) simulations. The incidence conditions are θinc = 45^◦ along the [10¯1]surface direction, T = 300K with 6×6×6 slab, number of simulated trajectories: 10⁶. For deuterium and hydrogen an incidence energy of 3.27 eV and 3.33 eV, respectively, was used.

Scattering Surface Subsurface Transmission

Isotope Adsorption Absorption

H 55 (82) 23 (4) 21 (6) 1 (8)

D 58 (74) 21 (8) 20 (13) 1 (5)

a peak atθout∼25^◦. The comparison of specular scattering (Fig. 3.45(b)) does not reveal large dierences between the surface types: the rst peak of the dierential energy loss distributions is dominant for all surfaces and although the tail of the dierential energy loss distribution of specular scattering for the reconstructed surface (green) is a bit more lled out than for the unreconstructed surface (black), the increase of a shoulder close to the rst peak is barely visible.

The addition of an adatom (dark red) to the surface does not result in any remarkable changes of the dierential energy loss distribution at specular scattering, either.

It appears therefore that, short of a very severe deviation from the (111) surface structure (perhaps by changing the crystal facet), the surface structure has little inuence on the scattering energy loss distribution.