Search

High-resolution core-level photoemission data from the 1s level of Li(110) have been obtained between 77 and 280 K. Analysis of the data reveals a significant difference in the zero-temperature phonon broadening between the bulk and surface atoms but only a small difference in the effective surface and bulk Debye temperatures. This latter result is in good agreement with an embedded-atom-method calculation of the bulk and surface Debye temperatures of Li. Implications of these results to surface core-level phonon broadening and surface lattice dynamics of the heavier alkali metals are discussed. (c) 2000 The American Physical Society.

High-resolution photoelectron spectroscopy using synchrotron radiation is used to measure the binding energy of the W 4f{sub 7/2} core level associated with bulk atoms of a W(111) crystal in the temperature range from 80 to 780 K. As temperature increases the lattice expands and the binding energy of bulk W 4f{sub 7/2} decreases. A total shift of {minus}45 meV is measured at 780 K with respect to the binding energy of bulk W 4f{sub 7/2} at 80 K. The mechanism responsible for this shift can be understood quantitatively in terms of a Friedel model, in which the narrowing of the d band as well as the shift of the center of the d band relative to the bottom of the conduction band due to the lattice expansion play dominant roles in shifting the binding energy toward lower energies. {copyright} {ital 1997} {ital The American Physical Society}

Above the threshold for K 3p excitation, the conduction band photoemission spectrum of metallic potassium is dominated by core excitation coupled to autoionization. This accounts for the observation of a strong response in regions where k-conserving transitions are forbidden, as well as for the angle-integrated character of data taken in an angle-resolved mode. The anomalous triangular peak at the Fermi level observed with photon energies near 25 eV is due to the selective enhancement of the 4p character of the conduction band by autoionization involving electrons excited into the normally empty 3d bands.

The difference between threshold theory and X-ray absorption-edge data in the near-edge region of simple metals is indicative of a reduction in screening response beyond threshold. An empirical model in which the edge exponent decreases linearly above threshold successfully reproduces the experimental Al K-edge absorption spectrum over an energy range of 5 eV and confirms the previously determined threshold exponent. The K-edge exponent of Mg, determined using this model, is in agreement with a prediction based on the singularity index and L-edge exponent.

Photoemission from the CsCl structure, n-type semiconductor CsAu confirms the presence of an adsorbed Cs layer on films produced by vapor phase reaction between bulk Au and Cs vapor. Cs 4d and 5p spectra yield a well-defined Cs core-electron binding-energy shift of −0.58 ± 0.06 eV for this layer relative to the Cs in bulk CsAu. The Au 4f and 5d spectra do not exhibit a measurable surface-atom binding-energy shift. Annealing of reacted CsAu films produced by the reaction of Au(100) with Cs vapor show ordered diffraction patterns that are coverage dependent.

Photoemission spectra of CaxC60, SrxC60, and BaxC60 show successive filling of the tlu-derived LUMO and the tlg-derived LUMO + 1 bands, for 0 < x < 6. For x < 2 compounds of Ca and Ba exhibit a well defined Fermi cut-off; compounds of Sr do not. In this range of composition, the resistivity remains activated, suggesting that the metallic character in photoemission is due to a surface phase. Stoichiometric compounds with formula A3C60 differ in both structure and electronic properties. The Ca compound retains the fcc stacking of C60, and is an insulator. The Sr compound transforms partially to the A15 structure but remains an insulator. The Ba compound has the A15 structure and is weakly metallic. For 3 < x < 6 metallic behavior is weak in the Ca and Sr compounds, but well defined in the Ba compound, indicating that the LUMO + 1 band is filled progressively by Ba. This is in accord with the superconductivity of these systems, which is found in compounds with x near 5, when the Fermi level is in the tlg band.

Angle-integrated valence-band photoemission spectra from ${\mathrm{C}}_{60}$ yield a total bandwidth of 0.52\ifmmode\pm\else\textpm\fi{}0.05 eV for the highest occupied molecular orbital in the surface layer of ${\mathrm{C}}_{60}$. The bulk bandwidth is estimated to be \ensuremath{\sim}0.8 eV, in reasonable agreement with a quasiparticle band-structure calculation. The phonon broadening in the surface layer at room temperature is 0.32\ifmmode\pm\else\textpm\fi{}0.04 eV.

Surface-ion core-level shifts from evaporated films of K, Rb, and Cs chlorides, bromides and iodides have been identified in high resolution photoemission spectra. The measured shifts of the alkali-metal surface ions in the six NaCl-structure compounds average 0.45 ± 0.03 eV to higher binding energy. Subtracting from these shifts the change in Madelung potential at the surface gives the corresponding change in the surface polarization energies, averaging 0.17 ± 0.02 eV. Shifts of the halide surface ions are found to be significantly smaller. All of these results are in qualitative agreement with theoretical expectations of surface Madelung potential and polarization energy changes, both of which are positive for the cation but of opposite sign for the anion. The data from the Cs compounds have a much larger scatter due to the polymorphism upon thin film growth.

CsxC60 compounds were prepared by the reaction of C60 films with Cs vapor and investigated by means of photoemission spectroscopy. Samples of the whole range of stoichiometry 0⩽x⩽6 were characterized. Upon doping the lowest unoccupied molecular orbital of C60 which is derived from the t1u orbital becomes filled, but the material does not show metallic properties at any composition indicating that correlation effects remain significant in the doped compound. The diffusion of Cs into the closely packed fcc lattice of weakly doped C60 is very slow at room temperature. It becomes much faster after the phase transition to the more open bct and bcc structures.

The phonon widths of outermost core-electron photoemission spectra from (110)-oriented overlayers of Na, K, and Rb metals show the expected Debye behavior for the bulk atoms, but significant deviations for the surface atoms. The data indicate a softening of the surface vibrational mode above 200 K. This effect, which is weak in Na but strong in K and Rb, demonstrates that the vibrational mode normal to the surface is anharmonic.

Layers of C 60 deposited on W(100) show charge donation into the LUMO of the molecules in contact with the metal surface, but there is no evidence for formation of a bulk compound. We present a simple formulation which accounts for the known successes and failures in metal−C 60 compound formation. Hypothetical compounds with the transition metals are unstable against precipitation of the metal clusters.

Photoemission spectra from thin films of ${\mathrm{K}}_{\mathit{x}}$${\mathrm{C}}_{60}$ prepared by vapor-phase reaction of ${\mathrm{C}}_{60}$ with K vapor consist of superpositions of distinct bulk and surface contributions. The spectra of ${\mathrm{K}}_{3}$${\mathrm{C}}_{60}$ and ${\mathrm{K}}_{6}$${\mathrm{C}}_{60}$ are identified and shown to be in very good agreement with band-structure calculations. The surface layer of molecules does not contribute to the metallic conductivity for any bulk composition.

Compounds of the form BaxC60 with x=3 and 6 resulting from the reaction of C60 with Ba have been prepared and characterized by photoemission. For x=3 the t1u band derived from the LUMO fo C60 is filled, resulting in an electronic structure similar to that of K6C60. Upon further exposure to Ba, the t1g band becomes occupied, but the system remains metallic at the largest Ba concentration that could be achieved. The superconductivity is associated with the t1g band.

High-resolution 4f core-level spectra of the Ta(110) surface region have been obtained at 80 and 300 K with 70- and 100-eV synchrotron radiation. The data show that the subsurface core-level binding-energy shift (compared to deeper-lying atoms) for a close-packed bcc(110) surface can be substantial: 65\ifmmode\pm\else\textpm\fi{}15 meV for the first underlayer atoms of Ta(110). The surface core-level shift is 360\ifmmode\pm\else\textpm\fi{}12 meV at 80 K and decreases by 13\ifmmode\pm\else\textpm\fi{}2 meV at 300 K. Final-state screening in both the bulk and surface layers is well described by a constant singularity index of 0.133\ifmmode\pm\else\textpm\fi{}0.012. An enhanced phonon broadening at the surface corresponds to a reduced perpendicular Debye temperature for the surface atoms of 128\ifmmode\pm\else\textpm\fi{}18 K compared to the bulk Debye temperature of 225 K.

Photoemission spectra of vacuum deposited layers of C 60 , before and after exposure to K vapor, show that the K donates its conduction electron into the band derived from the lowest unoccupied molecular orbital. A compound with composition of K 3 C 60 , corresponding to the maximum conductivity, has been prepared. In it the potassium atoms presumably occupy both the octahedral and the two tetrahedral interstitial sites of the face-centered-cubic (fcc) C 60 structure.

Methods used to obtain information from data with unresolved lines are described and applied to core-level photoemission spectra. It is shown that the parameter correlation matrix provides an invaluable guide to least-squares optimization. Difficulties, signaled by large elements in this matrix, can be overcome by introducing physically motivated constraints. This approach yields a surface-atom core-level shift of 195 meV from the 3p spectrum of the K(110) surface, as well as surface and subsurface shifts of 375 and 60 meV from the 4f spectrum of Ta(110).

Photoemission data on alkali-doped C 60 are reinterpreted in the light of the finding that the electron mean free path (escape depth) in typical experiments is comparable to or less than the diameter of the molecular unit. Data are consequently dominated by the signal from the surface layer of molecules which have distinct electronic properties. This accounts for the weakness of the Fermi cut-off found in all photoemission studies.

Several recent structural studies of the Si(111)–B ((3)1/2×(3)1/2) surface have established that the boron atoms occupy substitutional sites in the second full Si layer and have Si adatoms directly above them. High‐resolution (∼80–100 meV) Si(2p) core‐level photoemission was used to determine the B‐induced perturbation of the surface Si atoms. The samples were prepared by surface segregation from Si(111) wafers (∼1.5 at. % B) after thermally removing the thin protective oxide layer on the surface. Photoemission spectra for photon energies from 110 to 140 eV show three peaks, indicative of at least two well‐separated spin‐orbit doublet components. Attempts to fit the data with only two components were unsatisfactory unless a Doniach–Sunjic line shape was used for the B‐induced component. This suggests that the surface region influenced by the B impurities is metallic, i.e., the Fermi level lies within the B‐modified valence band. The intensity of the bulk component goes through a minimum at a photon energy...

The size-dependent insulator-metal transition in supported metal clusters manifests itself as a deviation from inverse radius dependence of the core-electron binding energy. Data for mono- and polydisperse supported gold clusters give evidence that the transition to the metallic state occurs in clusters containing circa 100 atoms. Simple theoretical considerations account for this observation.

The H-induced shift of the surface-atom core-level binding energy in W(110) is shown to arise from two distinct effects, one chemical in nature and the other structural. The structural shift supports a recently proposed (1{ital p}{times}1) reconstruction that turns on at {similar to}0.5 monolayer coverage. These new findings are used to provide a self-consistent interpretation of previously reported shifts from H-covered W(111) and W(100) surfaces.

High resolution 4f photoemission spectra from clean W(110) show that the natural lifetime width and the singularity index, which characterizes the conduction electron screening, are both larger in the first atomic layer than in the bulk. The phonon broadening of the surface and bulk components are smaller than earlier theoretical estimates. The excess broadening at the surface is compatible with a simple Debye model. These findings are very different from the interpretation previously given surface-atom core-level line shapes and have implications extending to other systems.

The photoemission signal from the first atomic layer of W(110) is used to assess the nature of the interaction between the surface atoms of the metal substrate and the adsorbates Na, K, and Cs for coverages up to 1 atomic layer. Our results indicate that there is little or no charge transfer from the alkali metal to the W surface, even in the limit of low coverage. The satellite structure of the photoemission lines of the outermost p shell of the alkali metals confirms this conclusion. While contrary to the conventional picture of alkali-metal-charge donation, these findings fully support recent theoretical calculations.

It is shown that empirical sensitivity factors do not provide a reliable basis for quantitative chemical analysis by electron Spectroscopy for chemical analysis (ESCA), because the fraction of photoelectric events accompanied by intrinsic excitations depends on the electronic structure of the solid. This provides a serious source of error in surface analysis. It is pointed out that the area of the entire photoemission line, including the loss tail, is directly proportional to the photoelectric cross-section, and provides a way of using ESCA for bulk quantitative analysis. A method of chemical analysis based on this observation is outlined. It has the advantage of being largely insensitive to surface segregation and surface contamination.

The growth of a passivating layer on a In0.53Ga0.47As(001)-4 × 2 surface by atomic-layer deposition of tetrakis[ethylmethylamino]Hafnium (TEMAHf)) followed by the water pulse was investigated by synchrotron radiation photoemission. The Hf atoms maintain four-fold coordination, both after the initial TEMAHf deposition and the subsequent water pulse. The Hf atoms initially bond to the As dangling bonds of the surface As atom located on the edges of the raised ridges. One EMA ligand is removed in this process. Subsequent water exposure substitutes OH ligand for one or more remaining EMA ligands. These in turn react with TEMAHf to form Hf-O-Hf bonds allowing the hafnium oxides to grow. The surface In atoms on the terrace of the raised ridges were partially removed, but none bonded of the precursor atoms. Correlations between the interfacial electronic structure and the electric performance are discussed.

Photoemission spectra of vacuum deposited layers of C(60), before and after exposure to K vapor, show that the K donates its conduction electron into the band derived from the lowest unoccupied molecular orbital. A compound with composition of K(3)C(60), corresponding to the maximum conductivity, has been prepared. In it the potassium atoms presumably occupy both the octahedral and the two tetrahedral interstitial sites of the face-centered-cubic (fcc) C(60) structure.

Rb 4p core-electron photoemission spectra obtained with He II resonance radiation exhibit a well-resolved response from the first atomic layer of atoms. The surface-atom core-level shift is 190 ± 3 meV, and the singularity index of the conduction electron screening response is 0.205 ± 0.005. Within the quoted experimental uncertainties both are identical to those of Na. The phonon broadenings at 100K are 65 meV for the bulk and 85 meV for the surface, showing the enhanced vibrational broadening at the surface previously found in Na.

Photoemission spectra from the shallow 3{ital p}{sub 3/2} core levels of the surface atoms of metallic potassium exhibit the effects of a small but measurable crystal-field splitting of {similar_to}38 meV. It manifests itself mainly as an apparent angle-dependent modulation of the spin-orbit splitting. This phenomenon may, in general, interfere with accurate determinations of surface-atom core-level shifts.

Valence-band photoemission spectra from a variety of alkali halides are shown to be well represented by a phonon-broadened density-of-states histogram with well-determined bandwidths and three or four main peaks. The bandwidths are generally in good agreement with band-structure calculations. The phonon broadening is typically 85% of that calculated for anion core levels, indicating that the valence-band holes are localized.

The change in the average surface-atom core-level shift (SCS) produced by submonolayer coverages of alkali adsorbates is quite small for both Ta(110) and W(110), less than 45 meV in magnitude. The small change of the measured SCS's as a function of coverage decisively supports the covalent-bonding picture of alkali adsorption proposed by Ishida and Terakura.

Two previously unappreciated features in photoemission spectra from alkali atoms adsorbed on W(110), namely, the sign of the alkali-induced surface-atom core-level shift of the substrate at low coverage and the very large alkali shallow core-hole lifetime width at all coverages, show that the alkali-substrate interaction is not well described by a transfer of alkali charge. Instead, both features point to the formation of a charge cloud between the alkali adatom and substrate that is derived largely from alkali valence states.

Photoemission spectroscopy is used to demonstrate that Ge segregates to the first atomic layer of Ge0.5Si0.5(100)2×1 and that the second layer is predominantly Si. Comparison of the resolved signals from the dimer atoms of the reconstructed (100)2×1 surfaces of Ge, Si, and equiatomic Ge‐Si alloy shows that the surface layer of the alloy is extremely Ge rich and the second layer is occupied mainly by Si atoms. This result is in good agreement with theoretical predictions.

Photoemission spectra of compounds prepared by the reaction of C(60) films with calcium show two distinct metallic phases, whereas alkali-doped C(60) films have only one. In the first phase the bulk t(1u) band, derived from the lowest unoccupied molecular orbital of C(60), is partially occupied. This is followed by an insulating phase that has the composition Ca(3)C(60) in which the t(1u) band is filled and has properties analogous to those of K(6)C(60). Continued exposure to calcium produces a second metallic phase in which electrons are donated into the t(1g) band. The superconductivity of Ca(5)C(60) is associated with the t(1g) band.

Photoemission data in which the signal from the first atomic layer is well resolved from that of the bulk are used to determine accurately the kinetic-energy dependence of the inelastic-electron mean free path in the alkali metals. At the higher kinetic energies, the data are in very good agreement with the theory of Penn. Below about 10 eV, the mean free path in the heavier alkali metals drops markedly below the theoretical values. This is attributed to electron decay processes involving the unoccupied d bands

Photoemission data from (110) films of Li, Na, and Rb, in which the signal from the first atomic layer is well resolved, show that the core-hole-screening singularity index is \ensuremath{\sim}40% larger at the surface than in the bulk for all three metals. This result, which is indicative of the more atomiclike character of metal surface atoms, in general, is particularly large for the alkali metals because their conduction-electron screening is mainly s-like. In addition to quantifying the difference in screening at the surface, the data provide bulk singularity indices of 0.22, 0.16, and 0.14 for Li, Na, and Rb, respectively. These new values are in better agreement with theory and with the threshold exponents than earlier values derived from incompletely bulklike x-ray photoemission data.

High-resolution photoemission spectra from the shallow core levels of alkali metals and of In have been obtained between 78 K and room temperature. The data yield values for the alkali-metal surface-atom core-level shift and show thermal shifts of comparable size for bulk and surface. The positive surface shifts are due to the spill-out of conduction-electron charge, which is responsible for the surface dipole layer. The surface shifts are in good agreement with values obtained from a Born-Haber cycle expressed in terms of surface energies. The thermal shifts are proportional to the lattice expansion, and arise from both initial-state and final-state effects. As the lattice expands, the Fermi level decreases, decreasing the core-electron binding energy. At the same time, the expansion of the conduction-electron charge increases {ital r}{sub {ital s}}, thereby decreasing the potential at the core level and increasing the binding energy. The expansion also decreases the relaxation energy, further increasing the core-electron binding energy. In the alkali metals, the combined potential- and relaxation-energy terms dominate the Fermi-level term, making the shifts positive. In divalent metals the three terms tend to cancel, while in trivalent metals it is the Fermi-level term that dominates, making the shifts negative.

Using core-level photoemission, a coverage-dependent transition in the adsorption of K on Si(100)2\ifmmode\times\else\texttimes\fi{}1 is observed. Below \ensuremath{\sim}0.25 monolayers, a single adsorption is occupied, the asymmetry of the Si-dimer reconstruction is enhanced, and no more than \ensuremath{\sim}0.05e is transferred from K to Si. Above this coverage, multiple sites are occupied, the dimer configuration becomes more symmetric, and the K overlayer becomes increasingly metallic. These findings resolve a number of conflicting studies of this system.

High-resolution temperature-dependent photoemission data from Na 2p core levels reveal substantially larger phonon broadening in the first atomic layer of Na(110) than in the bulk. We show that the enhanced width is due primarily to the excitation of relatively soft phonon modes perpendicular to the surface. Soft surface-phonon modes also account for previously reported but uninterpreted broadening of transition-metal surface-atom core levels.

In Cs 5p core-electron photoemission spectra obtained with He resonance radiation, the response from the first atomic layer is well resolved and shifted 228 meV from the bulk line at 11.79 eV. The singularity index of the conduction-electron screening response is 0.18, i.e., comparable to the values close to 0.20 obtained for the other alkali metals. In contrast, the edge exponent of the Cs 5p absorption edge is only 1/6 that of the Na 2p edge and is incompatible with the phase shifts deduced from the Cs singularity index. This result sharpens a long-standing problem regarding the interpretation of the emission and absorption edges.

We report a modification of the AT&T Bell Laboratories 6m toroidal grating monochromator at the National Synchrotron Light Source at Brookhaven National Laboratory which provides greatly improved resolution at midrange without the use of a movable exit slit. This modification is based on detailed ray tracing calculation, using the precedures developed by Haber, Howells, and McKinney and Howells, modified to include the holographic correction terms to the focusing equations. To maintain the optimum resolution requires an exit slit assembly movable over a distance of ∼20 cm. However, we show that an additional fixed slit 16.8 cm closer to the grating than the standard exit slit position gives near optimum resolution in the range 70–140 eV for the 2400 line/mm grating. Resolution measurements, using the Fermi cut‐off of various metals, confirm that the full width at half maximum (FWHM) can be kept below 100 meV for photon energies from 70 to 140 eV. Further improvement (perhaps to as low as 60 meV, FWHM) can be obtained by masking the width of the grating. The modified instrument has been successfully used in a number of studies in the far‐ultraviolet and soft x‐ray regions.

High-resolution soft-x-ray photoemission spectra of the C 1s line of graphite have been analyzed to determine the various contributions to the linewidth and line shape. The natural lifetime width \ensuremath{\Gamma} is 210 meV, much larger than those of small molecules (for which \ensuremath{\Gamma}\ensuremath{\simeq}110 meV), or that calculated for the isolated carbon atom (for which \ensuremath{\Gamma}=56 meV). The phonon width is no larger than 50 meV. The general trend toward an enhanced Auger decay rate from the isolated atom to the small molecule and solid is indicative of significant contributions from interatomic Auger processes. However, the large increase in width from molecules to solids must be due to another phenomenon, perhaps to the formation of the excitonic final state in the solids. Our line-shape analysis confirms that the line has an asymmetry like that associated with metallic screening, which had been earlier related to the excitonic final state in graphite, but the singularity index is significantly smaller than that obtained in the previous analysis.