No full-text available
To read the full-text of this research,
you can request a copy directly from the author.
Summary Abstract: Localized Auger final states in covalent systems
Abstract
Scitation is the online home of leading journals and conference proceedings from AIP Publishing and AIP Member Societies
Photon- or electron stimulated desorption (PSD or ESD) from solid surfaces results from the excitation of localized surface or adsorbate electronic states. The fundamental issues include the identification of electronic excitations with lead to desorption, the mechanisms for localization of the excitation, and the competition between nuclear motion and quenching of the excited states. We discuss these basic issues and their role in understanding stimulated desorption of molecular adsorbates from metal surfaces, and the desorption processes important in electron or photon bombardment of ionic solids. We focus on simple systems which serve as useful prototypes: CO chemisorbed on metals for adsorbate systems, and alkali-halides tor ionic solids. We review the current understanding of PSD and ESD for these systems, including a variety of very recent results. Finally, we discuss an example of the application of ESD to provide precise control of surface chemistry in the synthesis of electronic materials.
We have studied the electron-induced desorption (EID) of O+ ions from oxygen-covered polycrystalline W, Mo and Cr surfaces. As the incident electron energy is increased, desorption begins at about 25 eV, and increases dramatically at the binding energy major low-lying substrate core levels. In the range of electron energy studied, below about 200 eV, there are also other variations in ion yield not associated with known core levels. The peak ion kinetic energy is 7.8, 5.3 and 3.4 eV for W, Mo and Cr respectively, and is independent of incident electron energy. These measurements imply that the desorption is initiated by a core level ionization event, although the subsequent electronic transitions leading to desorption apparently differ from those occurring on metallic oxide surfaces.
Excited states with two holes and one electron (2h-le states) are known to be crucial in electron stimulated desorption. Starting with a positively ionized adsorbate (1h state), we show in this note that two-hole one-electron states (2h-le) can be generated by means of an Auger-like mechanism. Explicitly, when the Coulomb interaction between the adsorbate hole and the d-band of a transition metal surface is stronger than a critical value, a bound state is formed between the adsorbed hole and electron-hole pairs. Charge transfer to the adsorbate affinity level is strongly suppressed.
Electron Stimulated Desorption (ESD) of O+ ions from oxygen-covered Ni(100) has been investigated at 390 K and 500 eV primary energy. The ion energy distribution is found to peak at 7.5 eV and to extend to 11 eV, over our whole exposure range (0–1000 L). The 7.5 eV peak height as a function of exposure shows that desorption takes place both in the chemisorption and the oxidation region. Emission of O+ occurs preferentially along the surface normal, with a base width of ≈ 60°. No azimuthal structure is observed. Additional electron energy dependent measurements clearly show a threshold near the oxygen 2s level.
The density of states of a hydrogen atom on a linear chain is calculated using the Anderson-Newns Hamiltonian with a many-electron basis set. The total cross-section of EELS is considered as a superposition of five different excitations represented by operators, taking into account both adsorbate and substrate levels; the dynamic form factors of the five excitations are calculated and their main features discussed, for different occupations and widths of the band. The Hamiltonian is supplemented with a Coulomb interaction Umm between the substrate electrons and another, Uam, between a substrate electron and an adsorbate electron/hole, their effects in the results being discussed.
We have measured the alkali resonance photon intensity due to electron stimulated desorption of excited particles from NaCl and KCl targets as a function of incident electron energy (20 to 1000 eV) and sample temperature (room temperature to 300 °C). The energy dependence data suggest a correlation between excited Na and K desorption thresholds and substrate core levels. These are the first experimental correlations of the desorption of neutral atoms, specifically excited neutral atoms, with the creation of core holes. The intensity of emitted radiation is observed to rise sharply and then saturate with increasing temperature.
The nature of chemical information in Auger electron spectroscopy (AES) data is reviewed with special emphasis on data from solid surface systems. Two strategies are most frequently used to extract this information: (i) measuring and analyzing energy (chemical) shifts in Auger peaks; and (ii) making use of the shapes of Auger signals to determine the chemical environment at the site of the initial core hole. Chemical shift data are primarily illustrated by highlighting the interaction of oxygen with solids; and analyses of these data based on core‐level binding‐energy shifts, relaxation, and hole–hole interactions are outlined and discussed. Auger transitions that involve valence electrons are usually those for which lineshapes are taken as indications of the local chemistry at the initial core‐hole site. Attempts at extracting valence band density‐of‐states information from lineshapes are proving successful and this approach to the surface chemical information in AES is illustrated with the aid of examples dealing with the interaction of silicon with hydrogen and with oxygen. The use of the AES lineshapes simply as ’’fingerprints’’ of the core‐hole‐site chemistry is examined and illustrated by examples which include studies of silicon nitride properties, of solid surface properties related to catalytic reactions, and of passive films on iron. Auger decay activated desorption processes are briefly examined and found to promise new and unique chemical information when combined with conventional AES. Some gas phase AES studies are also briefly reviewed.
We have measured the incident electron energy dependence of excited alkali neutral desorption arising from electron bombardment of alkali halide surfaces. Substrate core hole formation is shown to play an important role in the initial energy transfer process. In each case only the lowest resonance lines are observed. The yields for excited alkali (10−5/electron) are measured to be at least two orders of magnitude greater than those obtained for ion ejection due to electron stimulated desorption. Our measurements provide added insight into the mechanisms of electronic excitation, ejection, and subsequent particle-surface interaction responsible for the desorption of excited alkali neutrals.
A review of the electron-stimulated-desorption (ESD) literature shows
that many of the features of ESD that are difficult to rationalize
within the model of Menzel, Gomer, and Redhead can easily be interpreted
using the Auger decay model, which has recently been developed to
explain ESD from transition-metal oxide surfaces. Specifically, the
Auger model helps to explain the charge state of the desorbing species,
the high-energy (~ 30-40 eV) onset behavior that is seen, the
differences in thresholds for positive and neutral desorbates, ESD
cross-section and isotope-effect data, and the high kinetic energies of
desorbing particles. The success of the Auger picture for ionically
bonded surfaces suggests a number of new applications of ESD, including
the deduction of reaction paths in surface chemistry and the study of
the evolution of surface oxides.
We present evidence for a fundamentally new mechanism for impact-induced desorption, viz., core-hole Auger decay. We thereby explain why observed thresholds for electron-stimulated desorption (ESD) of positive ions (O+, OH+, and F+) from certain d-band metal oxides (TiO2, V2O5, and WO3) correlate in energy with the ionization potential of the highest-lying atomic core levels. We conclude that electron-stimulated desorption is in many interesting cases an atom-specific, valence-sensitive probe of surfaces.
Measurements are reported of the M23VV and M1VV Auger spectra of copper and nickel and the results compared with other Auger transitions, like LVV and ion neutralization spectroscopy as well as with X-ray M23 band emission data in terms of peak structure, peak width and energy shifts. The conditions under which quasiatomic or band like spectra can be obtained are investigated. The data reveal a strong dependence of Auger lineshape on the involved initial state. A careful analysis of the likely sources of MVV structure is carried out and some effects possibly related to the band structure become apparent. In particular, the appearance of certain peaks in the high energy side of the main line is tentatively interpreted in terms of interatomic de-excitation processes.
The work of Asaad and Burhop and of Shirley on atoms is Shown to form the basis of a useful theory for the analysis of molecular core-valence-valence Auger spectra. If the static relaxation term of Shirley is neglected, the theory requires only a single SCF calculation on the neutral molecule. Since this approximation seems justified for the class of itinerant holes, the theory not only pemits the interpretation of the spectra of molecules considerably larger than previously analyzed, but potentially the spectra of chemisorbed molecules from cluster calculations. Theory and experiment are compared for the carbon KW spectra ofCH4, C2H2, C2H4 and CH3OH.
We report photon-stimulated desorption (PSD) of ions by the same core-hole Auger-decay mechanism observed in electron-stimulated desorption (ESD). Ions desorbed by synchrotron radiation were identified by time-of-flight mass spectroscopy. The thresholds for the PSD and ESD of H/sup +/, OH/sup +/, and F/sup +/ from TiOâ are identical. The PSD excitation spectra consists of sharp peaks, compared to broad steps in ESD. PSD is shown to be a powerful surface probe with the potential of providing unique information on bonding sites.
Noted are pronounced configuration-interaction effects upon Auger decay probabilities in CO and CO2. In CO, transition probabilities for carbon KVV decay to the 1π−2 final states are reduced by a factor of 0.09 due to initial-state Cl mixing with the unoccupied 2π orbital. In CO2, these amplitudes for decay to the 1π−2g final states are increased by ≈102 due to mixing with the virtual 2πu orbital.
Auger-electron spectroscopy is shown to measured something quite different from photoemission: the distribution of atomic (as opposed to overlap) charge populations across the valence bands. While matrix-elements effects must be considered in s-p band materials, their inclusion in calculations still lead to poor agreement with experiment. Good agreement may obtained, however, if one divides the electronic charge into atomic and overlap (bonding) LCAO components and notes that the latter does not contribute to the Auger current.
In this Letter it is shown that quasiatomiclike Auger spectra in narrow-band metals are a direct result of electron correlation effects. It is shown that if the "on-site" Coulomb interaction is much larger than the one-electron-band width, the Auger spectrum consists of an intense narrow atomiclike peak together with a weak broad bandlike peak at higher kinetic energy. On the other hand, if the Coulomb interaction is small the Auger spectrum will resemble the convolution of the band density of states convoluted with itself.
A calculation of the Auger spectrum for an initially filled simple-cubic tight-binding s band is presented. The spectra are obtained for various values of the Coulomb correlation energy as compared to the one-electron bandwidth. The results are compared to various approximate Auger line shapes. The Auger line shape is found to be strongly dependent on the Coulomb interaction between the two holes in the final state and deviate strongly from the self-convolution of the density of states even for small Coulomb interactions.
The Auger spectrum of beryllium metal is presented and analyzed. This analysis suggests that the valenceband screening of the initial-state core hole alters the observed line shape from that predicted on the basis of band structure alone. Decomposition of the spectrum into components describing the angular momenta of the final-state holes indicates that the screening charge is largely s-like. A rigorous description of the screening-charge effect is presented together with an approximate theory which preserves the analysis in terms of band structure. The approximate theory is in qualitative agreement with the results of ab initio calculations on a 13 atom cluster.
The copper M1VV Auger line shape, experimentally shown to contain both bandlike and atomiclike features, is analyzed. It is found that the atomiclike part arises from decays resulting in the creation of two d-like final-state holes, while the bandlike portion arises from the creation of one s-like and one d-like hole. The relative intensity of the two processes agrees well with theoretical predictions which also explain why the s-d process is not visible in the L2,3VV or M2,3VV spectra. In addition, this analysis shows that it is likely that the polarization charge, screening the initial-state M1 hole, appreciably affects the relative Auger amplitudes.
The LMM Auger spectra of Cu, Zn, Ga, and Ge are presented and discussed. Transition probability calculations are described and with these a clear assignment of the peaks can be made. It is further shown that from the L3M45M45 Auger lines the term splittings and the effective on-site electron-electron interaction can be determined. The latter is shown to be strongly reduced from the free-atom value. This has important consequences for the description of the band structure using one-electron theories. The satellite structure for Cu and Zn in the L3M45M45 region of the spectrum is shown to be a result of strong Coster-Kronig processes involving the L2 and L3 core levels.
The L2M45M45 and L3M45M45 Auger spectra of clean metallic copper and zinc were measured. A theory was developed to predict the Auger energies. The theory employs experimental electron binding energies, which were also measured, two-electron integrals, and Hartree-Fock energies. It accounts for multiplet splitting in the d8 final state, predicting structure in excellent agreement with experiment in zinc and in very good agreement in copper. It also accounts for "static" atomic relaxation and for static extra-atomic relaxation (screening), which is related to the Friedel theory of alloys. The theory developed here predicts the Auger energies to within 1 eV in zinc and 2-3 eV in copper. Since atomic integrals were used, the success of the theory implies that an atomistic approach to Auger energies is valid for these metals. The magnitude of the extra-atomic relaxation energy (∼ 10 eV) suggests that it may be a crucial factor in Auger energy shifts arising from chemical environment or surface condition.
Interatomic Auger transitions associated with valence electrons from nearest-neighbor atoms to the initial hole-state site have been measured in the ionic compounds NaF, MgF2, and Al2O3. Both low-(< 100 eV) and high-(> 1000 eV) energy interatomic transitions were observed, corresponding to decay of shallow and deep core hole states, respectively. Of the former group, transitions were identified in which the final states are characterized either by a single vacancy on a site adjacent to the initial hole-state site (interatomic Coster-Kroning decay), by double vacancies localized on an adjacent site, or by double vacancies delocalized on different adjacent sites. A simple model for calculating the energies of these transitions is presented in which corrections assuming complete ionicity and dielectric response are added to empirically determined one-electron binding energies. The corrections take into account the additional electronic polarization and hole-hole interaction energies absent in single-vacancy final states. Electron binding energies measured by x-ray photoemission were self-consistently referenced to the measured kinetic Auger energies from the ionic compounds. These latter energies were then compared with those calculated using the simple model. For both high- and low-energy interatomic transitions the overall agreement, typically in the range of 1-2 eV, was within the uncertainties of the Auger measurements and comparable to that found in the studies of intraatomic core-level transitions.
Final-state correlation effects in Auger line shapes are considered within the cluster linear combination of atomic orbitals-molecular orbitals-configuration interaction theory with a parametrized Hamiltonian. A model problem is solved analytically to elucidate the role of final-state hole-hole correlation and to understand the localization of the holes on rather small subclusters of the system. The relationship of the correlation effects to the relative magnitudes of the one-center hole-hole repulsion u and the bandwidth Γ has been previously reported; however, this previous work has been limited to metallic single element conductors. This work extends the theory to covalently bonded insulators (and possibly semiconductors) consisting of more than one element. Application of the theory is made to the O KVV and Si L23VV Auger line shapes from SiO2. A high-energy shoulder at 511 eV in the O KVV line shape is interpreted as arising directly from correlation effects. A peak at 50 eV in the Si L23VV line shape, its intensity significantly underestimated by the previous theory, is now accounted for; a peak at 70 eV previously suggested to be a shake satellite is now indicated also to arise from correlation effects. Both line shapes reveal a density of states primarily localized on an Si2O subcluster. The magnitude of the hole-hole repulsion on the subcluster and between neighboring Si2O subclusters is empirically determined from the Auger line shapes to be ∼ 11 and 4 eV, respectively. The oxygen 2p nonbonding bandwidth is estimated to be ∼ 6 eV, but in light of other theoretical and experimental results, our result is believed to be 1-2 eV too large. Reasons for our overestimate are discussed.
It is shown that one-electron band theory predicts the experimentally observed L2,3VV Auger line shape of silicon and the KVV line shape of lithium, provided that the partial densities of states are properly normalized for the atomic orbital (AO) basis used to calculate the matrix elements. This normalization, when combined with matrix-element effects, is responsible for the dominance of p-p hole final states in the experimental spectra. The effect is equivalent to noting that with the atomic-orbital basis, the electronic charge is divided into atomic and overlap populations. Due to matrix-element effects, the latter does not contribute to the Auger process. Thus, Auger-electron spectroscopy is sensitive to the variation of the local atomic charge density across the valence band. Since the s AO contributes more to the overlap (bonding) charge than the p AO does, the s-like contribution is suppressed in the Auger line shape. The quality of the agreement with experiment suggests that the combined effects of the surface, many-body phenomena, and the distortion of the valence band in response to the core hole are small for the above spectra.
Considerable recent theoretical work has shown that core–valence–valence Auger lineshape analysis may often be successfully made in terms of the molecular orbital structure of the molecule or the band structure of the solid. This approach assumes a priori that the final state hole motion is uncorrelated. Limitations to this approach include the highly correlated final states observed in ionic and narrow band solids (as Cu). Complications to this approach include the initial state valence screening of the core hole which affects the Auger decay probabilities. However, much useful information about the local density of states (local chemical environment) may be extracted in covalent and metallic materials. Examples from molecular (CH 4 , C 2 H 4 , C 2 H 6 , and C 3 H 8 ) and solid state (Cu and Be) spectra are presented and the state of our understanding of lineshape analysis is summarized.
The availability of empty electron states above the Fermi level and the presence of strong intra-atomic correlations may be expected to lead to new features in the Auger XVV spectra of conduction bands. Here, the qualitative aspects of the problem are studied within a simplified model. The case of a low equilibrium hole density nh in the band is considered in detail and the various propagators are calculated in a low density approximation. It is found that correlation and shake up effects can be approximately factored out and the singularity exponent relevant to the problem is linear in nh. In the single particle density of states we find a peak in the high binding energy side, which appears to be consistent with the experimental XPS spectra of Fe, Co and Ni. The Auger XVV spectra are obtained in the same approximation. As in the case of closed bands (nh = 0) two-hole resonances appear in the spectra provided that intra-atomic correlations are strong enough (quasiatomic case), but the condition for their occurrence is found to involve the total band width, rather than the width of the occupied portion of the band only. Moreover, even in the quasiatomic case, the resonances have a width that is related to the energy difference between the top of the band and the Fermi level.
We present criteria for the stability of ionic materials in ionizing environments, confining ourselves to cases where the core hole Auger decay mechanism of Knotek and Feibelman is applicable. The main result is that Auger induced decomposition will not occur unless the cation species in the solid is ionized down to a relatively deep filled shell. This shell must be sufficiently deep that an Auger decay starting from it will release the energy necessary for decomposition. The degree to which covalency in bonding affects stability is discussed. We show how these concepts can be applied by examination of the periodic table and a table of electron binding energies.
A theory of XVV Auger spectra of atoms in solids is proposed, that takes hole-hole repulsion into account. The interaction between the holes is taken to be localized at the atomic site. Within this model, the theory is exact. The relationship between atomic-like and band-like spectra is discussed and several previously unexplained features of the experimental spectra are shown to find a natural justification within the context of the theory.
If two holes are suddenly created in the same band and at the same atomic site e.g. by an Auger process in a solid, their density of states N(ω) will depend on their Coulomb interaction. In a tight binding model, we present the exact N(ω), in the limit of zero bandwidth. In the case of a general band, we give an exact integral equation that allows calculating N(ω) once the single electron density of states is known. The interaction is shown to produce a characteristic distortion of N(ω) and hence of the Auger spectrum.