Adsorption of gold on low index planes of rhenium

On atomically rough areas of a thermally cleaned rhenium field emitter, adsorbed gold behaves like it does on tungsten. The average work function $\text{}\text{?}$gf increases at low average gold coverage $\text{}\text{?}$gq due to formation of gold-rhenium dipoles, and at high coverage a structural transformation in the gold layer leads to a $\text{}\text{?}$gq-independent work function. Broadly similar behaviour is found for gold on the low-index planes of tungsten, but on low-index rhenium planes gold behaves rather differently. When thermally cleaned at > 2200 K and annealed below 800 K, the work function, φ(clean), of (101&#x0304;1&#x0304;) takes one of two values 5.25 ± 0.04 eV, and 5.36 ± 0.04 eV, which are tentatively attributed to the two possible structures of this plane. Similar behaviour is expected and observed for (101&#x0304;0),but the values taken by φ(clean) are not well defined. Both forms of (101&#x0304;1&#x0304;) are thought to undergo reconstruction above 800 K forming a single structure with φ(clean) = 5.55 ± 0.03 eV. (112&#x0304;0) and (112&#x0304;2&#x0304;) each have only one possible structure, and in keeping with this, φ(clean) has a single well-defined value for each plane. The flatness of (101&#x0304;1&#x0304;) and (101&#x0304;0) leads to field reduction at their centres which produces an increase in their measured work functions by up to 10%. The initial increase in φ produced by gold condensed at 78 K and spread at low equilibration temperatures T_s on (112&#x0304;2&#x0304;), (101&#x0304;1&#x0304;) and (112&#x0304;0) is attributed to gold-rhenium dipoles, which, on the latter two planes approximate to the Topping model, giving dipoles characterised by μ₀(1011) = 0.1 × 10^?30 C-m with α = 10 ų and μ₀(112&#x0304;0) = 0.32 × 10^?30 C-m with α = 22 ų, where μ₀ is the zero-coverage dipole moment and α its polarizability. Failure of the Topping model on (112&#x0304;2&#x0304;) is attributed to its atomically rough structure. No dipole effect is seen on (101&#x0304;0). Energy spectroscopy of electrons field emitted at (202&#x0304;1&#x0304;) and (101&#x0304;1&#x0304;) demonstrates the non-free character of electrons in rhenium, while the small effect of adsorbed gold strengthens the belief that gold is bound through a greatly broadened 6s level centred 5.6 eV below the Fermi level and the dipolar nature of the bond supports this model. At higher values of T_s and $\text{}\text{?}$gq gold appears to form states which are well-characterised by a coverage-independent work function. (101&#x0304;0), (101&#x0304;1&#x0304;) and (112&#x0304;0) each form two such states, one in the range 2 < $\text{}\text{?}$gq < 4 (state 1), and the second at $\text{}\text{?}$gq > 4 (state 2). The atomic radii of gold and rhenium are thought to be sufficiently similar to allow the possibility that state 1 is a replication of the Re plane structure by gold. The high work function and thermal stability of state 2, taken together with the observed temperature dependence of the transformation of state 1 to state 2, encourages the belief that state 2 results from atomic rearrangement of state 1 into a close-packed Au(111) structure. State 2 also forms on (112&#x0304;2&#x0304;) and the absence of state 1 on this plane suggests some surface alloying at coverages below 4 $\text{}\text{?}$gq.     

Adsorption of lead on low-index planes of tungsten

Using probe-hole field emission microscopy the effect of adsorbed lead on the work function of the 100 and 110 planes of tungsten hasbeen studied and compared with the findings of Bauer et al. who studied the same system using LEED/Auger techniques. The effect of lead on the average work function $\text{}\text{?}$gf and that of (211) is also reported. Sub-monolayer lead increases φ(211) and this is ascribed to formation of a lead-tungsten dipole, the lead being negatively charged, with dipole moment 0.035 × 10^?30 C-m and polarizability 2.0 ų. On (110) lead reduces φ and behaves as a dipole with positively charged lead of moment 0.15 × 10^?30 C-m and polarizability 2.5 ų. φ(100) is also observed to decrease at low coverages equilibrated at low temperatures. This contrasts with Bauer's findings but is considered to result from failure of the Fowler—Nordheim model. With increasing lead coverage on all planes φ(hkl) tends to a constant value φ_sat. By comparison with Bauer et al. we can identify φ_sat on (110) as a compressed monolayer of lead. Likewise φ_sat produced by low temperature (～450 K) spreading on (100) is also associated with a compressed (1 × 1) structure. The lower value of φ(100) produced at higher temperatures (～850 K) is identified with the microfacetted surface observed by Bauer et al. Lead is observed to be absent from (110) when mean adatom densities as high as 8 × 10¹⁴ atoms cm^?2 are thermally equilibrated, and this is shown to result from the relatively low binding energy of lead on (110). The general agreement between the present findings and those of Bauer lends confidence to the belief that both techniques can detect the same behaviour despite the very large (10¹⁰) difference in the size of the area examined.     

Adsorption of oxygen on a Cu{110} surface with and without the influence of A keV Ne+ beam: Stage I: Coverages up to half a monolayer

Low Energy Ion Scattering has been used to study the interaction of molecular oxygen with a Cu{110} surface. The amount of adsorbed atomic oxygen was monitored by the 4 keV Ne⁺¦O reflection signal. In the first adsorption stage (coverage less than half a monolayer) the sticking probability varied proportional to the number of empty adsorption sites: S = S₀ (1 ? $\text{}\text{?}$gq). It turned out not to be influenced by the Ne⁺ bombardment. The initial sticking probability S₀ was found to be ≈ 0.24. In this first adsorption stage the oxygen-covered surface is reconstructed according to the “missing row” model, leading to a (2 × 1) LEED pattern.     

Site-selective adsorption of xenon on a stepped Ru(0001) surface

The adsorption of xenon has been studied with UV photoemission (UPS), flash desorption (TDS) and work function measurements on differently conditioned Ru(0001) surfaces at 100 K and at pressures up to 3 × 10^?5 Torr. Low energy electron diffraction (LEED) and Auger electron spectroscopy (AES) served to ascertain the surface perfectness. On a perfect Ru(0001) surface only one Xe adsorption state is observed, which is characterized by$\text{Xe}\text{5}\text{p}{}_{\mathrm{<mtext>3</mtext><mtext>2</mtext>}}$,$\text{1}\text{2}$ electron binding energies of 5.40 and 6.65 eV, an adsorption energy of E_ad≈ 5 kcal/mole and dipole moment of μ'_T ≈ 0.25 D. On a stepped-kinked Ru(0001) surface, the terrace-width, the step-height and step-orientation of which are well characterized with LEED, however, two coexisting xenon adsorption states are distinguishable by an unprecedented separation in$\text{Xe}\text{5}\text{p}{}_{\mathrm{<mtext>3</mtext><mtext>2</mtext>}}$,$\text{1}\text{2}$ electron binding energies of 800 meV, by their different UPS intensities and line shapes, by their difference in adsorption energy ofΔE_ad ≈ 3 kcal/mole and finally by their strongly deviating dipole moments of μ_S = 1.0 D and μ_T = 0.34 D. The two xenon states (which are also observed on a slightly sputtered surface) are identified as corresponding to xenon atoms being adsorbed at step and terrace sites, respectively. Their relative concentrations as deduced from the UPS intensities quantitatively correlate with the abundance of step and terrace sites of the ideal TLK surface structure model as derived from LEED. Furthermore, ledge-sites and kink-sites are distinguishable via E_ad. The E_ad heterogeneity on the stepped-kinked Ru(0001) surface is interpreted in terms of different coordination and/or different charge-transfer-bonding at the various surface sites. The enormous increase in Xe 5p electron binding energy of 0.8 eV for Xe atoms at step sites is interpreted as a pure surface dipole potential shift. —The observed effects suggest selective xenon adsorption as a tool for local surface structure determination.     

Hydrogen chemisorption on W {110}: Angle-resolved photoemission

Normal photoemission spectra (k_∥ = 0) from a W {110} surface reveal 2 major resonances during hydrogen chemisorption, at 2.0 and 4 eV below the Fermi level, E_F. The former appears at 300 K during filling of the low coverage β₂ state, and the latter grows as the β₁ state is filled. Detailed spectra obtained along the $\text{}\text{?}$gS and $\text{}\text{?}$gD azimuths are reported, showing additional resonances at ~0.5, 1,3,6 and ~7 eV below E_F. A structural model is proposed, based on a consideration of the present results in relation to RHEED and HRELS data, for the H-saturated surface in which a p(1 × 2) structure is formed, at a H: surface W ratio of unity, with β₁ adatoms occupying a close bridge position and β₂ adatoms in a 3-fold hollow site.     

The preparation and surface structure of clean V(110)

The first quantitative determination of the surface structure of a group VB metal is reported. A procedure is described for the preparation of a clean, well-ordered surface of V(110), free from the major bulk impurity, oxygen. The clean surface exhibits the two-dimensional periodicity of the corresponding bulk plane. Experimental LEED intensity measurements are compared to the results of multiple-scattering model calculations, with very good agreement being obtained for a value of the first interlayer spacing d, close to the bulk value. Minimization of the r factor for the comparison of experimental and calculated intensity spectra leads to a value of $\text{d = 2.12 ± 0.02}\text{A}\text{?}$, compared to the bulk value of 2.14 Å.     

Carbon monoxide and carbon dioxide interaction with tantalum

The adsorption of carbon monoxide and carbon dioxide on tantalum and the dissolution of these gases in the adsorbent at T ? 300 K have been studied. The flash-filament method (FFM) in a monopole mass-spectrometer and a field emission microscopy was used in the same apparatus. Carbon monoxide and carbon dioxide dissociate on the tantalum surface, carbon monoxide being desorbed in both cases during the flash. The desorption curves of CO reveal three different binding states: two of them (α and $\text{}\text{?}$gb₁) for the adsorbed particles whereas the high temperature desorption state relates to the adsorbate dissolved in the metal, For the $\text{}\text{?}$gb₁ state of CO the activation energy, the pre-exponential factor and the kinetic order in the kinetic equation of desorption have been estimated. They turned out to be E = 110 kcal/mol, C = 3 × 10¹²sec^?1, and ν = 1. The activation energy of diffusion for CO in tantalum and the energy of outgassing for the metal were found to be 9.4 and 49 kcal/mole, respectively.     

Adsorption of CO on Pd single crystal surfaces

Studies of CO adsorption on Pd(110), (210) and (311) surfaces as well as with a (111) plane with periodic step arrays were performed by means of LEED, contact potential and flash desorption measurements. Isosteric heats of adsorption were evaluated from adsorption isotherms. Earlier work with Pd(111) and Pd (100) surfaces is briefly reviewed, yielding the following general picture: The initial adsorption energies vary between 34 and $\text{40}\text{kcal}\text{mole}$ and close similarities exist for the dipole moments, the maximum densities of adsorbed particles and for the adsorption kinetics. At low and medium coverage the adsorbed particles are located at highly symmetrical adsorption sites, whereas saturation is characterized by the tendency for formation of close-packed layers.     

Surface chemistry of the metal-halogen interface: Bromine chemisorption and dibromide formation on palladium (111)

At 300 K and in the coverage regime ($\text{0<θ<}\text{1}\text{3}$) bromine chemisorbs rapidly on Pd(111); the sticking probability and dipole moment per adatom remain constant at 0.8 ± 0.2 and 1.2 D, respectively. This stage is marked by the appearance of a √3 structure: desorption occurs exclusively as atomic Br. At higher coverages, desorption of molecular Br₂ begins (desorption energy ～130 kJ mol^?1) as does the nucleation and growth of PdBr₂ on the surface. This latter stage is signalled by the appearance of a √2 LEED pattern and the observation of PdBr₂ as a desorption product (desorption energy ～37 kJ mol^?1). Some PdBr₂ is also lost by surface decomposition and subsequent evaporation of atomic Br. The data indicate that the transition state to Br adatom desorption is localised and that PdBr₂(a) ? Br(a) interconversion occurs; these surface species do not appear to be in thermodynamic equilibrium during the desorption process.     

Thermodynamics of xenon adsorption on Pd(s)[8(100) × (110)]: From steps to multilayers

The thermodynamic properties of the adsorption of xenon on the stepped Pd(s)[8(100)×(110)] surface have been studied over a wide range of pressure (5×10^?11 to 1×10^?4 Torr) and temperature (40–140 K). We have measured adsorption isobars using AES in order to evaluate the surface coverage. By choosing pressure and temperature we have studied under equilibrium conditions, the successive adsorption of xenon on the steps and on the terraces until the first layer is formed, the condensation of the second layer as well as the formation of xenon multilayers. For a small range of pressure and temperature, adsorption takes place only on the atomic steps. The LEED pattern shows that only every other site along the steps is occupied. The extrapolated initial heat of adsorption for steps is E_ad^S = 10.2 kcal/mol, decreasing monotonically by about 2 kcal/mol as the relative coverage of the step sites increases. The dipole moment of the Xe atoms adsorbed on steps is 1.12 D. During adsorption on the terraces the LEED observations suggest that the xenon adlayer is non-localized up to completion of the hexagonally close packed monolayer. The initial heat of adsorption on the terraces, E_ad^T is 8.2 kcal/mol and decreases continuously to a value of 6.9 kcal/mol for a complete monolayer due to lateral repulsive interactions between the adsorbed xenon atoms. The induced dipole moment of Xe on terraces is reduced to 0.49 D. The 5p_{$\text{1}\text{2}$} binding energy of Xe adsorbed on terrace sites is 0.3 eV smaller than that of Xe occuping step sites. The differential molar entropy of the adsorbed layer on the terraces as a function of coverage compares fairly well with the calculated value for an ideally mobile two-dimensional gas. No indication of the growth of two-dimensional xenon islands has been found under these conditions. The isosteric heat of adsorption for the second layer is E_ad^sec = 5.8 kcal/mol independently of the coverage. The condensation of the second layer is a first order two-dimensional gas ? two-dimensional solid phase transition in opposition to the continuous nature of the adsorption of the first layer (extending over a wide range of temperature for a given pressure). The induced dipole moment is further reduced for the Xe second layer to a value of 0.11 D. Finally, the condensation of multilayers proceeds with a latent heat of transformation of E_cond = 3.8 kcal/mol in excellent agreement with the known bulk value for the heat of sublimation of xenon. The line shape of the NVV low energy Auger transitions of xenon or the UPS binding energies of the Xe 5p_{$\text{3}\text{2}$,$\text{1}\text{2}$} spectra allow a clear distinction between first, second and higher layer Xe atoms. We have also established the temperature/pressure conditions for equilibrium between first, second and bulk xenon layers, i.e. a so-called “roughening point”.     

The epitaxy of gold on (110) tungsten studied by leed

Growth of gold condensed on the (110) plane of tungsten has been studied using LEED and AES. Three ordered surface structures were observed when condensation takes place at or above 700 K, and no detectable order is seen below this temperature. Structure 1 is developed as the coverage approaches one monolayer and has gold atoms held in the W(110) array with a resultant 2% reduction in gold atom diameter. The second gold layer adopts the Au(111) structure with $\text{Au}\text{[1}\text{2}\text{1]}$ rotated by 2.5° from $\text{W}\text{[1}\text{1}\text{0]}$ and the first gold layer may also be constrained to adopt this structure. Deposition of more gold produces three dimensional crystallites with Au(111)∥W(110) which are double-positioned with their 121 directions parallel to the 121 directions of tungsten. Addition of half a monolayer of oxygen before condensation, completely prevents formation of structures 1 and 2. Instead, at coverages of 3 or more monolayers, three dimensional crystallites develop with Au(111) ∥ W(110) and $\text{Au}\text{[1}\text{2}\text{1] ∥}\text{W}\text{[1}\text{1}\text{0]}$. This behaviour is compared with the reported behaviour of copper and silver on W(110).     

Chemical shifts in auger electron spectra from silicon in silicon nitride

The chemisorption of nitric oxide on (110) nickel has been investigated by Auger electron spectroscopy, LEED and thermal desorption. The NO adsorbed irreversibly at 300 K and a faint (2 × 3) structure was observed. At 500 K this pattern intensified, the nitrogen Auger signal increased and the oxygen signal decreased. This is interpreted as the dissociation of NO which had been bound via nitrogen to the surface. By measuring the rate of the decomposition as a function of temperature the dissociation energy is calculated at 125 kJ mol^?1. At ～860 K nitrogen desorbs. The rate of this desorption has been measured by AES and by quantitative thermal desorption. It is shown that the desorption of N₂ is first order and that the binding energy is 213 kJ mol^?1. The small increase in desorption temperature with increasing coverage is interpreted as due to an attractive interaction between adsorbed molecules of ～14 kJ mol^?1 for a monolayer. The (2 × 3) LEED pattern which persists from 500–800 K is shown to be associated with nitrogen only. The same pattern is obtained on a carbon contaminated crystal from which oxygen has desorbed as CO and CO₂. The (2 × 3) pattern has spots split along the (0.1) direction as $\text{(m,}\text{n}\text{3}\text{)}$ and $\text{(}\text{m}\text{2, n}\text{)}$. This is interpreted as domains of (2 × 3) structures separated by boundaries which give phase differences of $\text{2π}\text{3}$ and π. The split spots coalesce as the nitrogen starts to desorb. A (2 × 1) pattern due to adsorbed oxygen was then observed to 1100 K when the oxygen dissolved in the crystal leaving the nickel (110) pattern.     

Surface self-diffusion on Ni(110): Temperature dependence and directional anisotropy

The surface self-diffusion coefficients, D_s, on a Ni(110) crystal are measured by a mass transfer technique in [1$\text{1}$0] and [001] directions in the temperature range 773–1573 K. The surface cleanliness was checked by Auger electron spectroscopy. LEED investigations showed that the sinusoidal surface profile consisted of (110) terraces and monatomic steps. The temperature dependence of D_s can be expressed by $\text{D}{}_{\mathrm{s}}\text{[1}\text{1}\text{0] = 0.009}\text{exp}\text{(}\text{?17.5}\text{kcal}\text{mole}\text{· RT)}$ and $\text{D}{}_{\mathrm{<mtext>s</mtext>}}\text{[001] = 470}\text{exp}\text{(}\text{?45}\text{kcal}\text{mole}\text{· RT)}$ at temperatures below 1150 K. Theoretical values for the activation energies of surface migration were calculated in the framework of the pairwise interaction model. Together with an estimate for the formation energy of adatoms of $\text{16.3}\text{kcal}\text{mole}$, one obtains for the activation energy of surface self-diffusion 17 and $\text{51 kcal}\text{mole}$ for [1$\text{1}$0] and [001] direction, respectively. At T > 1150 K the anisotropy in D_s begins to vanish. Surface diffusion in [1$\text{1}$0] direction at T < 1150 K is most likely taking place by a simple adatom hopping process. Circumstantial evidence indicates that diffusion in [001] direction does not occur by a simple hopping process but by a more complex mechanism involving higher energy surface diffusion states. This isotropic process is suggested to take place for both directions at T < 1150 K.     

Chlorine adsorption on copper: II. Photoemission from Cu(001)c(2 × 2)-C1 and Cu(111)(√3 × √3)R30°-Cl

We have studied high-resolution angle-resolved and photon-polarization dependent photoemission from chlorine adsorbed on Cu(OOl) and Cu(111). Chlorine forms a c(2 × 2) saturation overlayer on Cu(OO1) and adsorbs dissociatively as revealed by LEED and XPS. Several two-dimensional energy bands on Cu(001)c(2 × 2)-Cl could be iden along the $\text{}\text{?}$gG M? and $\text{}\text{?}$gG M? lines of the surface Brillouin zone, their respective mirror symmetry and their orbital character could be determined. An interpretation of these bands is given in terms of the interaction of the ordered overlayers with particular substrate bulk bands. Besides the appearance of adsorbate-induced two-dimensional bands drastic changes are resolved in the substrate d-band emission region. These can be explained almost exclusively by surface umklapp processes involving reciprocal lattice vectors of the ordered adsorbate mesh. Supplementary studies of the Cu(111) (√3 × √3)R30°-Cl system support our ideas. We discuss some important implications of our results for the interpretation of angle-resolved photoenussion spectra from ordered adsorbate layers.     

Hydrogen-induced reconstruction on a high step density W(001) surface

The hydrogen-induced reconstruction on a high step density W(001) crystal, ($\text{2}\text{×}\text{2}$)R45°-H, with steps oriented parallel to the [110] and ～ 28 Å average terrace width has been investigated using LEED symmetry, beam shape analyses, and EELS. The symmetry of the LEED pattern is observed to change from p2mg for the ($\text{2}\text{×}\text{2}$)R45° clean surface reconstruction to c2mm for the commensurate phase ($\text{2}\text{×}\text{2}$)R45°-H reconstruction. Correspondingly, the shapes of the half-order beams indicate that the hydrogen-induced reconstruction domains are much less elongated than the clean surface domains. A splitting of each half-order beam into four beams at higher exposures indicates the existence of two domains of the incommensurate phase. A commensurate phase v₁ vibrational loss peak centered at 160 meV in the EELS spectrum broadens on the low-energy side during the incommensurate phase and then shifts toward 130 meV and narrows as the (1×1)-H saturation structure develops. These observations imply that there is no long-range inhibition ( ～ 20 Å) to the formation of either commensurate or incommensurate phase; hydrogen induces a switching of the atomic displacements from 〈110〉 directions on a clean surface to 〈100〉 directions, even with steps oriented parallel to the [110]; and in the incommensurate phase there is a distribution of hydrogen site geometries with the most probable geometry more like the commensurate phase geometry than the saturation phase geometry.     

Dielectric constants of adsorbed CO on Cu and Ag

Adsorption of CO on evaporated Ag and Cu films was studied by ellipsometry and resistivity measurements. Changes in ellipsometric angles δΔ and δψ due to adsorption of CO were analyzed by an improved linear approximation of the stratified layer model of adsorption. Adsorption of CO on Cu induced a 23% increase in the relative resistivity change (δR/R) which was proportional to δψ, while adsorption of CO on Ag induced a 1% increase without the proportionality. The dielectric constant of CO adsorbed on Ag is $\text{?}{}_1\text{= 2.2?0.7i}$ at λ = 1152 nm in comparison to ?₁ = 2.1 for gas phase CO; that of CO on Cu is $\text{?}{}_1\text{= ?4.8 ? 8.5i}$, which is consistent with the prediction by the Bennett and Penn theory. The large difference in $\text{}\text{?}$ge₁, of adsorbed CO on Ag and Cu is understood in terms of the energy level of the 2 π molecular orbital of adsorbed CO relative to the Fermi level. A possibility of adlayer plasmon excitation is discussed.     

Chemisorption and surface reactivity of nitric oxide on clean and sodium-dosed Ag(110)

The chemisorption of NO on clean and Na-dosed Ag(110) has been studied by LEED, Auger spectroscopy, and thermal desorption. On the clean surface, non-dissociative adsorption into the α-state occurs at 300 K with an initial sticking probability of ～0.1, and the surface is saturated at a coverage of about $\text{1}\text{25}$. Desorption occurs without decomposition, and is characterised by an enthalpy of E_d ～104 kJ mol^?1 — comparable with that for NO desorption from transition metals. Surface defects do not seem to play a significant role in the chemistry of NO on clean Ag, and the presence of surface Na inhibits the adsorption of αNO. However, in the presence of both surface and subsurface Na, both the strength and the extent of NO adsorption are markedly increased and a new state (β₁NO) with E_d ～121 kJ mol^?1 appears. Adsorption into this state occurs with destruction of the Ag(110)-(1 × 2)Na ordered phase. Desorption of β₁NO occurs with significant decomposition, N₂ and N₂O are observed as geseous products, and the system's behaviour towards NO resembles that of a transition metal. Incorporation of subsurface oxygen in addition to subsurface Na increases the desorption enthalpy (β₂NO), maximum coverage, and surface reactivity of NO still further: only about half the adsorbed layer desorbs without decomposition. The bonding of NO to Ag is discussed, and comparisons are made with the properties of α and βNO on Pt(110).     

A leed-aes study of the structure of sulfur monolayers on the Mo(100) crystal face

The formation of ordered phases of sulfur on the molybdenum (100) crystal face has been studied by Low Energy Electron Diffraction (LEED), Auger Electron Spectroscopy (AES) and Thermal Desorption Spectroscopies (TDS). Sulfur was deposited from a S₂ molecular flux streaming out of an Ag₂S containing electrochemical cell inside the UHV chamber. The use of a controlled flux of S₂ allowed the careful determination of saturation values for the monolayer, as well as the formation of multilayers of sulfur. This allowed the calibration of Auger intensities in terms of sulfur coverage. Various ordered structures, c(2 × 2), (1 × 2), $\text{2}\text{1}\text{?}\text{1}\text{1}$ and c(2 × 4), were observed by LEED for different values of the S coverage. Real space models for these structures are proposed that satisfy the coverage values observed and place sulfur atoms only on high symmetry four-fold sites on the (100) molybdenum surface.