Search

We have performed DFT calculations using both plane wave-pseudopotential and Gaussian basis set approaches on the (1 1 1) and (2 1 0) surfaces of pyrite (FeS2). Our calculations indicate that the (1 1 1) surface is more stable than the (2 1 0) surface, which is predicted to have a higher surface energy. The (2 1 0) surface is predicted to be essentially bulk terminated, with a relatively small amount of surface relaxation. Bridging S atoms on the (1 1 1) surface undergo significant lateral displacement to coordinate to neighbouring undercoordinated surface Fe atoms. Electrostatic effects induced by loss of coordination at surface Fe atoms are likely to be responsible for surface ionic displacements. Our calculations indicate that surface Fe states dominate at the Fermi level on the (2 1 0) surface, but that surface S states make significant contributions to the valence band on the (1 1 1) surface. On the (2 1 0) surface, Fe atoms of 4-fold coordination are spin polarized and therefore paramagnetic, while 5-fold coordinated Fe are diamagnetic. For the (1 1 1) surface, both 5- and 6-fold coordinated Fe atoms are predicted to be spin polarized, although nominally the unpaired electrons are expected to be localised on the surface S− species. This is likely to be due to the removal of spin polarization from the surface S by charge transfer from the surface Fe.

The ability of a recently derived transferable variable-charge Ti–O interatomic potential to simulate low-coordination environments has been tested by application to the (1 1 0) and (0 0 1) rutile surfaces. Hartree–Fock and density functional theory calculations of the (1 1 0) rutile surface were also performed. Comparison of the variable-charge results with data from first principles and two other standard forcefield calculations, along with experimental data, reveals that the variable-charge model appears to be inadequate for surface simulations. The atomic relaxations predicted by all the forcefields are not only mutually inconsistent but are also in disagreement with ab initio and experimental data. The quantum mechanical treatments, on the other hand, yield relaxation data in agreement with each other, but different to experimental data on the (1 1 0) surface. The surface formation energies obtained for the relaxed (1 1 0), (1 0 0), and (0 0 1) surfaces are also not in the right order when computed with the variable-charge model. The low ionic charges predicted by the model appear to be a major reason for this failure. Inclusion of surface data in the parameterisation of the model is likely to improve the transferability to low-coordination environments.

Adhesion between clean, bulk-terminated bcc Fe(1 0 0) and Fe(1 1 0) matched and mismatched surfaces was simulated within the theoretical framework of the density functional theory. The generalized-gradient spin approximation exchange-correlation functional was used in conjunction with a plane wave-ultrasoft pseudopotential representation. The structure and properties of bulk bcc Fe were calculated in order to establish the reliability of the methodology employed, as well as to determine suitably converged values of computational parameters to be used in subsequent surface calculations. Interfaces were modelled using a single supercell approach, with the interfacial separation distance manipulated by the size of vacuum separation between vertically adjacent surface cells. The adhesive energies at discrete interfacial separations were calculated for each interface and the resulting data fitted to the universal binding energy relation (UBER) of Rose et al. [Phys. Rev. Lett. 47 (1981) 675]. An interpretation of the values of the fitted UBER parameters for the four Fe interfaces studied is given. In addition, a discussion on the validity of the employed computational methodology is presented.

The results of a study of the physical, elastic and electronic properties of PbS using both first principles theoretical and experimental X-ray and ultraviolet photoelectron techniques are presented. We have investigated the influence of the computational approximations such as the basis set, pseudopotential and the treatment of exchange and correlation on the accuracy of the theoretical results. The current study work confirms that the binding energy of the S 3s state lies at 13 eV, in agreement with previous experimental work and contrary to earlier ab initio Hartree–Fock (HF) calculations. We establish that the reason for the discrepancy between previous experimental and theoretical determinations of the electronic structure of PbS is due to the treatment of the exchange interaction. We demonstrate that calculations using the HF approximation tend to overestimate the binding energies of the valence and core states yielding densities of states in poor agreement with experiment whereas calculations performed with density functional theory within either the local density or the generalised gradient approximation give a significantly better description of the electronic structure. Finally, we confirm that hybrid functional techniques such as B3LYP provide an excellent description of the physical and electronic structure of PbS.

We present the relaxed structure of the (001) surface of TiO2 obtained from fully converged ab initio calculations using both Hartee–Fock and density functional theories. The calculated surface structure is in reasonable agreement with previous plane wave density functional and empirical tight binding calculations but is significantly different to either of the proposed structures based on LEED–IV experiments. We discuss possible reasons for the discrepancy with the LEED–IV results. The current calculations also reveal that the (001) termination has a significantly higher surface energy than the (100) or (110) surfaces confirming that it is the least stable low index termination of rutile TiO2. Finally, we examine the electronic structure of the relaxed surface and find no evidence for the presence of band gap states as reported in early photoemission experiments and tight binding calculations. The current results suggest that such states will only occur on the defective surface and are likely to be localised Ti3+ states induced by oxygen reduction.

Static and time resolved Fourier transform infrared spectroscopic studies were combined with threshold temperature programmed desorption studies to examine the effects of adsorbed sulfur on the kinetics of CO desorption from Ni(100). The appearance of new desorption states upon the addition of sulfur to Ni(100) can be attributed to the occupation of four-fold hollow and modified bridge-bonded species produced by the presence of sulfur. The observed kinetics of desorption from these surfaces is very sensitive to the initial binding site distribution of the adsorbed CO. Adsorbing at 100 K without anneal leads to a reduction in the measured desorption energy, except at the highest sulfur coverages studied. Apparently, relaxation of the adsorbed phase from a nonequilibrium distribution among sites on the surface during temperature programmed desorption is incomplete, leading to stronger repulsive interactions on the average and reduced desorption energies following adsorption at low temperature.

The effect of binding site distribution of CO on its desorption kinetics from Ni(100) was investigated using temperature program desorption (TPD) and reflection absorption infrared spectroscopy (RAIRS). Desorption of CO was studied from either surfaces presaturated at 100 K and annealed to 250 K (θco = 0.59 ML), 342 K (θco = 0.49 ML), and 414 K (θco = 0.21 ML) or by dosing to similar coverages without anneal at 100 K. The TPD spectra indicate that the initial desorption kinetics are significantly different for the two methods of preparation. Desorption kinetic parameters, Ed and vd, obtained using the threshold temperature program desorption (TTPD) analysis without anneal were 114 ± 2 kJ mol and 8.1 × 10 15 s −1 at 0.18 ML and 32 ± 2 kJ mol and 107 s−1 at 0.50 ML. At both of these coverages CO is initially distributed nearly equally between atop and bridged sites. Annealing the adlayer led to the dominance of atop sites in the adsorbate population, yielding less coveragesensitive desorption kinetics. At θ co = 0.21 ML, E d and v d were 128 ± 4 kJ mol and 3.9 × 1016s−1 and at 0.50 ML they were 92 ± 3 kJ mol and 1.3 × 1014 s−1.

The oxidation of Si(111)7 × 7 covered with potassium atoms has been investigated with high resolution electron energy loss (HREEL) spectroscopy and temperature programmed desorption (TPD). The amount of SiO 2 produced (monitored using the 140 meV HREEL excitation) increased in proportion to the K coverage up to 25 ML. This result indicates that the promotion effect of adding potassium atoms is strongly local, yet the HREEL and TPD spectra clearly demonstrate that the details of the promotion mechanism depend on the K coverage. The HREEL spectrum recorded after a 0.8 ML potassium layer was exposed to oxygen shows evidence only of O-Si bonding, even though the K TPD spectrum exhibits a shift toward higher temperature and a change in peak shape when oxygen was present. Oxygen interacts with the K atoms in a submonolayer, but indirectly via the Si atoms to which they are bound. In contrast, oxygen interacts with the K atoms in a multilayer directly since the HREEL spectrum shows that stable potassium oxides K x O y form. Heating these compounds above 300 K causes them to decompose yielding K atoms which desorb and reactive O species which diffuse to make SiO 2 .

The longitudinal and transverse components of the image force acting on a moving charge near a metal surface are calculated. In particular, it is found that the transverse force is finite at the surface in contrast to the classical image force. The transverse force leads to deviations in trajectory in grazing incidence reflection inelastic electron scattering experiments, which are estimated. A consequence is the existence of a limiting maximum probability for excitation of surface plasmons.

Our method is based on the assumption that the H to metal d-orbital interaction is predominantly a local one between the H and the nearest neighbour substrate atoms, whereas the H-sp orbital interaction is not localised and is best treated within a jellium-like framework. Our early simple but successful model for H adsorption on jellium is generalised by including a single muffin tin sphere representing the transition metal atom. This model should work best for the narrow band case of nickel and at present applies only to atop adsorption but is being generalised to multi-atom embedded clusters. However, it is felt that general features of the adsorbate-substrate interaction should also be evident for atop adsorption. The hydrogen-metal and barrier-transition metal distances are the main uncertainties in our calculation. For reasonable values of these we obtain a virtual bound state below the d-band, but lying in the sp-band. This virtual bound state has strong d-orbital character especially for Pd and Pt. The width of the state is markedly less for Pd and Pt than for Ni, in apparent agreement with UPS observations.

We have developed an Embedded Cluster Model for treating H chemisorption on Ni, Pd and Pt, which is also applicable to other systems. Our model tries to embody in the simplest way the essential physics, believed to be as follows: (i) the d-component of the bond will be local so we take into account the H atom and only its three (or four) nearest neighbours via muffin tins surrounding the atoms; (ii) the sp component is best described in a free-electron picture requiring that the cluster be embedded in the electron gas; (iii) the rise in potential from inner to vacuum potential. The model can be transformed into a novel Anderson-like model. Results show a chemisorption-induced resonance below the d-band in approximately the position observed in UPS measurements. Consequences for the chemisorption energy and (for Ni) magnetization are pointed out. The results imply hybridization between the H 1s and transition metal d and s orbitals.

The interaction between H adatoms on a surface is calculated within the embedded cluster model of chemisorption. The model is first applied to the case of two H atoms on a free electron surface. The interaction energy is found to be an oscillatory function of the H-H separation R ab . Application of the free electron model to the problem of chemisorption on transition metal surfaces leads to unphysical results with the prediction of formation of ordered H overlayers which are not observed in LEED experiments. We next include the l = 2 TM muffin tins. Results for H adsorption on the low index faces of Ni and Pd substrates are presented. Graphitic structures are predicted for the (111) faces of both Ni and Pd with the H atoms occupying both types of three-fold hollow sites on the surface. This agrees with the results of LEED experiments for H/Ni(111). Comparison with experiment is not possible in the case of H/Pd(111) owing to the lack of low temperature studies for that system. Zig-zag chains with the H atoms adsorbed in sites of three-fold coordination on alternate sides of the TM(110) rows are predicted for both Ni and Pd. This is in agreement with the results of He diffraction experiments for H/Ni(110). No structure determination has been done for H/Pd(110). Adsorption in the four-fold centre sites for H on the (100) faces of Ni and Pd is found to be unfavourable. The H atoms are expected to adsorb in sites of three-fold symmetry below the (100) surface for H on Pd with formation of a c(2 × 2) structure in agreement with the LEED observations. For H/Ni(100) the H atoms are believed to adsorb above the surface, away from the centre site and to bond to two surface Ni atoms. No short-range ordered structures are predicted in this case.

The electronic properties of H adsorbed on a Pd(111) surface are described within an embedded cluster model of chemisorption and compared to recent ARUPS data. Best agreement with the low temperature spectra is obtained for H adsorbed in the three-fold hollow sites a distance of d = 1.02 A above the surface. Binding energy calculations indicate a minimum in the Δ E versus d curve at about the same position of the H atom (d = 1.05 A). Calculation of the interaction energies between H adatoms also indicate the presence of strong attractive interactions which are responsible for the formation of an ordered overlayer of (2 × 2) structure. Bridge bonded H was found to be slightly more attractive than centre bonded H, but with considerably more repulsive inter-adatom interactions indicating not only the lack of order on the surface but also a reduced H saturation coverage for this bonding configuration. The binding energy of H atoms at intermediate sites (between the centre and bridge site) was found to be in the range between the 2 values for centre and bridge adsorption indicating the possibility of high mobility at higher temperatures, thereby resulting in a more diffuse H-metal bonding state whose position may be coincident with the bottom of the Pd d-band. Subsurface H impurities result in markedly more repulsive HPd energies than obtained for H above the Pd surface.

We have generalised our earlier work calculating the induced density of states in alkali chemisorption to include the valence np level in addition to the s level. This allows the theory to take account of the possibility of sp hybridization. Hybridization is said to be weak is the sp level separation ΔEsP exceeds twice the width of the s or pz resonance. In the converse case hybridization is strong and the induced density of states no longer reflects distinguishable s and pz resonances. We argue that ΔEsP for the chemisorbed alkali exceeds considerably that for the free alkali atom. In the case of Na on rs = 2 jellium (considered typical of the smaller alkali on rs = 2−3 substrates) hybridization is strong for the atomic value of ΔEsp but weak when a somewhat larger value is taken, making the conclusions a little indefinite. For Cs on Cu(111), we find a clear situation of weak hybridization with the atomic ΔEsP and a fortiori with a more realistic value. There should thus be a well defined 6s resonance in this system. Our estimate of the width of the 6s resonance is in good agreement with photoemission data.

Application of a simple wave function-matching technique is made to the problem of alkali adsorption on free-electron like substrates, giving in particular an estimate of the width of the resonance at a given adatom-substrate separation, which is then compared to existing first-principles calculations. The model is then applied to the alkali/transition-metal problem, thus assuming the non-importance of the transition-metal d-bands relative to their sp-bands. The results indicate a fair amount of hybridization between the ns and np valence levels of the alkali adatom.

A theoretical model of chemisorption on narrow band transition metals is presented. The model is based on the KKR formalism, and concentrates on calculations of quantities which may be associated with UPS measurements. These include for instance changes in density of states Δ N (ϵ) induced by adsorption of an atom on a clean surface, or localised density of states ρL(ϵ) of a given angular momentum L . Results are presented in a form reminiscent of the old Anderson model of chemisorption, thus clarifying considerably the nature of the bond in these systems.

The response function for a jellium model with a finite discontinuity in potential, is expressed analytically in reciprocal space for the system designed to represent metal-vacuum and metal-metal interfaces. Sum rules which the response function must satisfy are derived and verified in both cases. Mention is made of the many interesting applications in which a knowledge of the response function is fundamental.

Hydrogen adsorption on Fe(110) is modelled within the embedded cluster formalism of chemisorption. The most favourable sites for the H adatoms are the hollow sites of three-fold coordination. The magnitude of the interaction energies was found to be strongly dependent on the H-Fe distances. For short H-Fe bond lengths a graphitic (2 × 2) structure is expected from the present calculation in disagreement with the LEED observations. For larger H-Fe distances the predictions from the present model are consistent with the LEED results in that a sequence of ordered structures starting with a (2 × 1) structure and culminating in a (1 × 1) structure is expected on the Fe(110) surface.

A systematic study of H adsorption on the close-packed surfaces of the transition metals in the 3d and 4d series is presented. The effect of the TM d band on the chemisorption bond is investigated, by embedding a cluster of TM muffin tins at the surface of an effective jellium-like medium. It is found that the broad and incomplete H/jellium resonance is narrowed, shifted down and made to contain more electrons as a result of hybridization with the TM d states. These effects are larger in the case of the 4d metals, thus indicating a greater participation in the chemisorption bond of the d electrons for these metals than for the 3d metals. Calculation of one-electron energy differences on going from the H on jellium system to that of H on the TM cluster are presented. Trends for the one-electron energy differences are compared to trends in experimental chemisorption energy. H adsorption in the three-fold hollow site with no secondlayer TM atom below the H site is favoured for the hcp metals, while no discernible preference between the two hollow sites is recorded for the fcc metals, with the exception of Rh where the site with no second-layer TM atom below is preferred.

The interaction energy between adatoms on a metal surface is investigated within multiple-scattering theory. Interaction energies for groups of up to six adatoms are calculated and compared to the sum of the corresponding pairwise interactions. The contribution arising from non-pairwise interactions is found to be small but of the same order of magnitude as the long-range pair interactions which are invoked in the formation of ordered overlayers on the surface. Application is made to the H/Fe(110) chemisorption system. Calculation of the one-electron energy change on moving the H atom from a free-electron environment to an Fe(110) environment suggests that short-bridge sites are occupied by the H adatoms. Nearest-neighbour pair interactions are found to be considerably repulsive indicating that H saturation coverages should not be greater than 1. On the basis of calculation of pair interactions only, it is found that at lower H coverages, two ordered structures may be formed, however when including multiple-adatom effects a marginal preference for the experimentally observed c (2 × 2) structure is recorded. Beyond saturation of this ordered overlayer, other ordered overlayers of (3 × 3) structure may be formed, however since more substantial repulsive interactions have to be involved in the formation of such structures one would expect these to be accompanied by a considerable reduction in the hydrogen heat of adsorption.