9604
J. Phys. Chem. B 2005, 109, 9604-9612
Coverage-Dependent Adsorption of Atomic Sulfur on Fe(110): A DFT Study Michelle J. S. Spencer, Ian K. Snook, and Irene Yarovsky* Applied Physics, RMIT UniVersity, GPO Box 2476V, Victoria, 3001 Australia ReceiVed: December 9, 2004; In Final Form: March 17, 2005
Adsorption of atomic sulfur at different coverages on the Fe(110) surface is examined using density functional theory (DFT) in order to investigate the effect that adsorbate-adsorbate interactions may have on the surface properties. S is adsorbed in the high-symmetry adsorption sites: 4-fold hollow, bridge, and atop sites in the following surface arrangements: c(2 × 2) and p(1 × 1) which correspond to coverages of 1/2 and 1 monolayer, respectively. The binding energy, work function change, adsorption geometry, charge density distribution, magnetic properties, and density of states are examined and compared to our previous study of S adsorbed at 1/4 monolayer coverage and p(2 × 2) arrangement [Spencer et al. Surf. Sci. 2003, 540, 420]. It was found that S forms polar covalent bonds to the surface. The bonding goes from being S-Fe dominated at the low coverages to being S-S dominated at the higher coverages where the S atoms are located closer together on the surface and interact with each other.
1. Introduction
2. Method
Sulfur (S) contamination of iron (Fe) has important implications in many industrial processes and the effect it has on the metal’s surface properties is a major factor in the current and future applications of the metal. Previous experimental studies have shown S to adsorb in various arrangements and coverages on the Fe(110) surface. At low coverages, a p(2 × 2) arrangement is observed corresponding to 1/4 monolayer (ML),1 with S adsorbing in 4-fold hollow sites. At higher coverages, different overlayer arrangements have been found which correspond to 1/3 ML,2-4 whereas at higher temperatures and pressures more complicated arrangements and coverages can form.2-4 Due to the different overlayer arrangements found for the 1/3 ML coverage, this coverage will be considered elsewhere.5 Previous computational studies appear to be limited to our own studies of S adsorbed at 1/4 ML coverage in the p(2 × 2) arrangement.6,7 Despite these studies, it is not known how different coverages of S contamination affect the structural, electronic, and magnetic properties of the surface and the importance of adsorbate-adsorbate interactions to these properties. In the present study, we systematically examine the adsorption of atomic S on Fe(110) at different coverages and arrangements using density functional theory (DFT). S is adsorbed in three high-symmetry adsorption sites (atop, bridge, and 4-fold hollow) in two arrangements: c(2 × 2) and p(1 × 1) corresponding to coverages of 1/2 and 1 ML, respectively. We note that the 4-fold hollow site is in fact more like a long-bridge site, however, we refer to it as 4-fold hollow after the original paper by Shih et al.,1 who determined the geometry and structure of the Fe(110)p(2 × 2)-S system. We investigate the changes in energetics, work function values, electron density distribution, magnetic properties, and density of states of the Fe(110) surface at the different S coverages and compare the findings with our previous study that examined S adsorbed in the p(2 × 2) arrangement and 1/4 ML coverage.7
2.1. Computational Details. All calculations were performed using the plane wave pseudopotential Vienna ab initio Simulation Package (VASP),8-10 which performs fully self-consistent DFT calculations to solve the Kohn-Sham equations.11 The generalized gradient spin approximation (GGSA), using the functional of Perdew and Wang (PW91),12 was employed. The plane wave cutoff energy was 308.76 eV. Core electrons were replaced by ultrasoft pseudopotentials.13 K-space sampling was performed using the scheme of Monkhorst and Pack,14 with a mesh of 6 × 6 × 1 for both surface coverages as the cells were the same size as used previously for the 1/4 ML coverage.7 Our previous work showed this approach to give a good description of bulk, surface, and interfacial properties of Fe,15-18 FeS2,19-21 as well as S/Fe(110) systems.7,22 For the calculation of fractional occupancies, a broadening approach by Methfessel-Paxton23 was used with order N ) 1 and smearing width 0.1 eV, for the Fe(110) and S/Fe(110) slabs. The isolated S atom was modeled in a 15 × 15 × 15 Å cell, using a Gaussian method, a smearing width of 0.1, and a cutoff energy equal to that used for Fe. For accurate calculation of total energies and magnetic moment values, the tetrahedron scheme24 was employed. Weigner-Seitz radii of 1.3 and 1.164 Å were employed for Fe and S, respectively. As these radii are different to those used previously for our study of the p(2 × 2) S/Fe(110) system,7 the relevant parameters were recalculated and are presented here. For the calculated work function values, a dipole correction was added in the direction perpendicular to the surface (as we have an asymmetric slab with the adsorbate placed on only one side of the slab). Calculations of the electrostatic potential within the vacuum region of each supercell then showed a clear distinction between the vacuum level of each side of the slab. The work function values, Φ, were then calculated as
* Corresponding author: Email:
[email protected]. Telephone: +61 3 9925 2571. Fax: +61 3 9925 5290.
Φ ) Evac - EF where Evac is the electrostatic potential in the vacuum region of the supercell on the adsorbate side and EF is the energy of the Fermi level.
10.1021/jp044378l CCC: $30.25 © 2005 American Chemical Society Published on Web 04/22/2005
Adsorption of Atomic Sulfur on Fe(110)
J. Phys. Chem. B, Vol. 109, No. 19, 2005 9605
Figure 2. Calculated binding energies for sulfur adsorbed on Fe(110) in atop, bridge, and 4-fold hollow sites as a function of S coverage.
3. Results and Discussion
Figure 1. (a) Top view of the Fe(110) surface showing unit cells used to model the different S coverages and arrangements indicated. S was adsorbed in atop, bridge and 4-fold hollow sites (only the supercells for the atop site are shown); (b) side view of the relaxed 1/2 and 1 ML coverage surface models. The atoms in gray represent the atom pairs that the shortest S-Fe distances were measured between. When two Fe atoms are equivalent distance to the adsorbed S atom (for the bridge and 4-fold hollow sites), then both atoms are shaded.
2.2. Surface Models. The Fe surfaces were modeled using the supercell approach, where periodic boundary conditions are applied to the central supercell so that it is reproduced periodically throughout space. Surfaces were cleaved from a crystal structure of (bcc) Fe, corresponding to the (110) Miller plane. For the p(2 × 2), c(2 × 2), and p(1 × 1) S-adsorption arrangements, a [2 × 2] cell was used with 1, 2, and 4 adsorbed S atoms, respectively (Figure 1). The cells comprised 5 Fe layers, and the S atoms were adsorbed on one side of the surface slab only. The surface was produced by replication of the central supercell in the x,y directions. A vacuum spacer of ∼10 Å was inserted in the z direction. A lattice constant value of 2.855Å was used, as this is the lattice constant obtained by bulk cell optimization using the same computational parameters.7 Calculations were performed allowing the S and top 3 Fe layers to relax while keeping the bottom two Fe layers fixed at the bulk geometry. For each coverage and adsorption site, the S binding energy (BE) was calculated as
BE ) [Eslab+nS - nES - Eslab]/n where Eslab+nS is the total energy of the relaxed S/Fe(110) system with n ) 1, 2, or 4 adsorbed S atoms depending on the coverage, Eslab is the total energy of the relaxed clean Fe(110) slab and ES is the total energy of an isolated S atom. Hence, a more negative binding energy indicates a more favorable structure. Even though a [1 × 1] cell is sufficient to model a p(1 × 1) coverage, a [2 × 2] cell was employed for consistency with the supercell size of the p(2 × 2) and c(2 × 2) coverages.
3.1. Binding Energy and Work Function Measurements. The calculated binding energies (BE) for S adsorbed in atop, bridge, and 4-fold hollow sites at the different coverages are shown in Figure 2. All are stationary points on the potential energy surface and the classification of these points is included in Table 1. For the 1/4 ML coverage, the 4-fold hollow site is a minimum, whereas for the 1/2 ML coverage, none of the highsymmetry sites are minima. Also, for the 1 ML coverage, no sites considered here are minima. We did consider adsorption in the slightly lower symmetry 3-fold hollow site but found that it was not a stationary point for the coverages and arrangements considered here. However, we have been able to identify a minimum, where the S adsorbs in 3-fold hollow sites at 1/2 ML coverage, but the arrangement of atoms does not correspond to the c(2 × 2) considered here. It is possible, though, that other lower symmetry sites or different surface arrangements could be considered, but this it outside the scope of this work, which focuses on the high-symmetry sites. The binding energies indicate that S is most stable in the 4-fold hollow site, followed by the bridge and then atop sites at 1/4 and 1/2 monolayer coverage. At monolayer coverage, however, the bridge site is actually more stable than the 4-fold hollow site, but it should be noted that there is only a small difference of less than 2.7% between the BE for all sites at this coverage. For the other coverages, the binding energy in the 4-fold hollow site is approximately 17 and 29 % larger than the atop site, for the 1/2 and 1/4 ML coverages, respectively. For all adsorption sites, the S is more stable on the surface at the lowest coverage, being less favored at 1/2 ML. At 1 ML, the S is least stable of all coverages. Calculations of the change in work function, ∆Φ, of the Fe(110) surface after adsorption of S can give an indication of the charge reorganization which affects the surface dipole moment. A decrease in the work function after adsorption indicates a transfer of charge from the adsorbate to the substrate, while an increase in the work function indicates a transfer of charge from the substrate to the adsorbate. The calculated work function changes after S adsorption in different sites and coverages are shown in Figure 3 and were determined by subtracting the calculated work function value of the clean surface (4.86 eV) from those of the S adsorbed surface values. At coverages of 1/4 and 1/2 ML, the work function is clearly greater than that of the clean surface, leading to a positive work function change and suggesting the presence of a negatively charged surface species. Interestingly, at monolayer coverage, in contrast to the lower coverages, the work function becomes
9606 J. Phys. Chem. B, Vol. 109, No. 19, 2005
Spencer et al.
TABLE 1: Calculated Properties of the Different S/Fe(110) Systemsa S coverage arrangement 1/
1/
7 4 ML p(2 × 2)
2
ML c(2 × 2)
1 ML p(1 × 1)
buckling of layer
magnitude of relaxation (Å)
type of SP
d(S-Fe) (Å)
d⊥(S-Fe1) (Å)
1
2
3
δz1
atop
26
2.06
1.80
+
+
+
bridge
16
2.15
1.60
+
+
+
4-fold hollow
06
2.19
1.49
+
+
+
0.21 0.02 -0.05 0.00 0.10 0.09 0.01
atop
3
2.09
1.80
+
+
+
bridge 4-fold hollow atop bridge 4-fold hollow
1 1
2.19 2.18
1.78 1.47
+
+
+
3 2 3
2.23 2.35 2.38
2.23 1.99 1.90
site
0.26 -0.04 0.08 0.18 0.01 0.03 0.05 -0.09
δz2
δz3
0.08 0.02
0.01 0.03
0.04 0.05 0.06 0.04 0.03 0.08 0.02 0.05 0.07 0.03 0.04 0.06 -0.01
0.02 0.03 0.03 0.02 0.03 0.02 0.03 0.04 0.02 0.03 0.03 0.01
a Type of stationary point (SP): 0)minimum, 1) transition state, 2 ) 2nd order saddle point, 3 ) 3rd order saddle point; shortest S-Fe distance, d(S-Fe), see Figure 1; perpendicular height of S above the top surface Fe layer, d⊥(S-Fe1); magnitude of Fe surface relaxation for top three layers. A negative value indicates a contraction of the layer toward the bulk, while a positive value indicates an expansion. The presence (+) or absence (blank) of surface buckling is also indicated.
Figure 3. Calculated work function changes, ∆Φ, for S adsorbed on Fe(110) in atop, bridge, and 4-fold hollow sites as a function of coverage.
smaller than that of the clean surface, leading to a negative work function change, indicating that the S is behaving as an electropositive species. We relate this different behavior of the work function change at the different coverages to the charge density differences in section C. It is interesting to note, though, that the trend in the change of work function values with increasing S coverage appears to be correlated to the trend in binding energy values. Specifically, as the binding energy decreases at a particular coverage, the work function change increases, and vice versa. This observation may be useful experimentally to estimate general trends in binding energy values from measured work function changes. 3.2. Surface Geometry. 3.2.1. Adsorbate-Induced Relaxation and Reconstruction. Adsorption of S on the Fe(110) surface at different coverages and adsorption sites causes the Fe surface layers to relax from their positions on the clean surface. For some sites and coverages, the surface Fe layers were also found to display small adsorbate-induced surface reconstructions, showing movements in the x-y plane of the surface. Previously, we showed that small reconstructions were also seen for S adsorbed at 1/4 ML coverage and p(2 × 2) arrangement in 4-fold hollow and bridge sites but not the atop site.7 At 1/2 ML, S does not cause the Fe surface to reconstruct for the atop and 4-fold hollow sites, however, the topmost Fe layer
atoms do reconstruct slightly, by
