Surface State Transitions Research Articles

UV-photoelectron spectroscopy at highest resolution —direct access to lifetime effects in solids?

Lineshapes and linewidth in angle-resolved photoemission spectra from solid surfaces contain a wealth of contributions from e.g. the photohole lifetime, the lifetime of the final state electron, and from their respective interactions with phonons and lattice imperfections. In addition, finite energy and angular resolution contribute to the experimentally observed linewidths. Using photoelectron spectra from bulk and surface state transitions on copper as an example, we discuss to which extent the various contributions may be distinguished experimentally. The results indicate that relevant spectroscopic information can be directly derived from such studies at very high resolution. This will lead beyond the kinematical analysis of photoelectron data in terms of band structures and may enable us to extract quantities which refer to the dynamical properties of the many-electron system.

Unoccupied bulk and surface bands in the Gamma KX- Gamma KX plane of GaAs as observed by inverse photoemission spectroscopy.

An extensive set of isochromat inverse photoemission data has been collected from the (110) face of a cleaved GaAs crystal for angles of electron incidence up to 63\ifmmode^\circ\else\textdegree\fi{} from the normal. In these data, transitions are observed across the full width of the \ensuremath{\Gamma}KX-\ensuremath{\Gamma}KX plane of the GaAs Brillouin zone, and the bands involved in these transitions are assigned with the guidance of an ab initio linear muffin-tin orbital (LMTO) calculation. A symmetry method for absolute K-space determination is outlined, and this is used to locate several of these transitions. Along the \ensuremath{\Gamma}KX symmetry line, an unoccupied state on the LMTO band eight saddle is found at 6.3 eV above the valence-band maximum (VBM) and another transition is observed at 8.6 eV, near the LMTO band 9 minimum. The band 5 maximum along the \ensuremath{\Gamma}\ensuremath{\Delta}X symmetry line is estimated to be 3.2 eV above VBM. An intense resonance is observed for angles of electron incidence near 10\ifmmode^\circ\else\textdegree\fi{}, where the energy of surface-state transitions approaches that of bulk transitions. The surface state is found to disperse to a minimum energy of 2.0 eV, and to have a bandwidth of at least 0.7 eV. \textcopyright{} 1996 The American Physical Society.

Reflectance anisotropy from non-III–V systems: Si and SiGe growth on (001) Si and adsorbate-induced reconstruction of Cu(110)

Reflectance anisotropy (RA) data are presented for the study of two discrete systems — the dynamic growth of Si and SiGe on Si(001) and the formation of ordered overlayers on Cu(110). In the first of these studies first examples are presented of high quality RA oscillations and simultaneously recorded RHEED oscillations obtained during the growth of Si and SiGe on Si (001) by gas source MBE using disilane and germane at temperatures of 600 °C. RA measurements clearly demonstrate an enhancement in growth rate upon introduction of germane to the disilane flow. For both systems, the RA was found to oscillate at a period corresponding to bilayer growth which is explained in terms of a model based upon the changes in relative domain size and hence Si dimer “concentration” on the two orthogonal [110] azimuths. The RA data are also compared with variations in the total reflectance of the growing surfaces which is recorded simultaneously. In the second study it is demonstrated here that the RA technique may also be applied to the study of surface processes occurring at a metal single crystal surface. A marked RA response was observed from a Cu(110) single crystal under conditions where adsorption of O2, O2 + formic acid, and benzoic acid results in a re-structuring of the surface as determined independently using LEED. Dynamic RA responses corresponding to (R110 — R001)/Rtot are correlated with exposure of the various adsorbates and the appearance of particular overlayer structures. It is suggested that the RA response observed comes from a quenching of an allowed surface transition of the clean Cu(110) surface by the electronic perturbation induced by the adsorbates. If correct, this explanation would add much weight to the currently held interpretation of RA data from III-V systems which invokes similar geometrically oriented surface state transitions. However, at present, the possibility of involvement of new adsorbate-induced surface states cannot be ruled out.

Experimental results from spectroscopic ellipsometry on the (7 × 7)Si(111) surface reconstruction: dielectric function determination

The dielectric function of the (7 × 7) Si(111) surface has been directly determined in UHV conditions by ellipsometric measurements on a Si(111) clean surface. The dielectric function obtained from ellipsometric spectra near the pseudo-Brewster angle (PB) has been calculated by means of a three-phase model (bulk, (7 × 7) transition layer and vacuum) in which optical properties of bulk Si are deduced from literature data and the thickness of the (7 × 7) transition layer is taken from the DAS model. Our calculation allows us to confirm the metallic character of the (7 × 7) transition layer and to point out the previously found surface-state transitions at 2.6 and 3.4 eV and perhaps at 1.6 eV. Furthermore, a surface-state transition has also been confirmed at 3 eV. Ellipsometric results depend on the crystallographic direction. This optical anisotropy is ascribed to some roughness anisotropy induced by steps.

Resonant second-harmonic generation on Cu(111) by a surface-state to image-potential-state transition.

Dispersion measurements of second-harmonic generation from a clean Cu(111) surface show a resonance feature at the second-harmonic (SH) photon energy of 4.1 eV, consistent with an image-potential-state to surface-state transition in agreement with polarization selection rules. The sensitivity of the SH efficiency to surface defects on Cu(111), observed by Bloch et al. [Phys. Rev. B 45, 12 011 (1992)], is explained in terms of an inherent energy broadening of the surface-state wave function.

Modelling the optical response of surfaces measured by spectroscopic ellipsometry: application to Si and Ge

We present a convenient formula for interpreting the surface contribution to ellipsometric measurements, which should facilitate the further development of such optical studies. The formalism has been generalized to allow for optical anisotropy at the surface. This generalized form allows direct comparison to theoretical models of surface optical response. The method is used to analyze measurements of the Ge(111)(2 × 1), Si(111)(2 × 1) and Si(100)(2 × 1) reconstructed surfaces. The results are compared to previously published results. We conclude that surface state transitions do not dominate the surface contributions above the bulk energy gap.

Temperature dependence of the inverse photoemission from copper surfaces.

The temperature dependence of the intensity of inverse photoemission transitions into bulk and surface states has been studied on Cu(100), (110), and (111) surfaces. The temperature dependence is strongly influenced by multiple-scattering effects. For Cu(100) and (111) it can be fitted very well with a Debye-Waller law where the mean-square displacements of the surface atoms increase linearly with temperature. For Cu(110) we find an enhancement of the mean-square displacements above 400 K in agreement with surface-sensitive scattering experiments. Through the comparison of a given bulk transition on different surfaces, we find a strong influence of surface vibrations on the intensities of bulk transitions. A classification of the temperature dependence of the investigated bulk transitions is attempted in a two-band model, in a nearly-free-electron model, and in the one-step model of inverse photoemission incorporating bulk and surface phonons. The results are not quite satisfying, although they seem to show the right trend: the temperature dependence is stronger the higher the energy is. The temperature dependence of the surface-state transitions depends on their character. Transitions into image-potential surface states are temperature independent. Transitions into crystal-induced surface states respond to temperature changes. Transitions into states that are located in the same band gap and have nearly the same relative position in it show the same temperature dependence. The temperature dependence gets smaller the more bulklike the surface state gets.

Surface phonons and electronic surface states at si (111) 2×1: energy shifts and polarizaticn effects

Using high-resolution electron energy loss spectrosoopy and low energy electron diffraction we have studied the peak position and intensity of the surface phonon and the electronic surface state transition on the cleaved Si(111) 2×1 surface. The peak position of the phonon was found to vary between 48 and 56 meV, depending on the cleave, and this variation correlated with the peak position of the surface state transition. These empirical results find a possible interpretation in terms of the theory of Alerhand ., [Phys. Rev. Lett. 55 , 2700 (1985)], in which the energy positions of these two peaks depend on the tilt angle of the p-bonded chains with respect to the surface. A variety of factors - cleavage direction, electron scattering plane, and distribution of domains - influence the observed intensity behavior. Hydrogen or water uptake completely quench these transitions and the 2×1 reconstruction is converted to a 1×1.

Ge-derived Si(1 1 1)(7 × 7) structure studied by low-energy electron diffraction and angle-resolved electron-energy-loss spectroscopy

Using low-energy electron diffraction (LEED) and angle-resolved electron-energy-loss spectroscopy (AR-EELS), we find that a new Ge-derived Si(7 × 7) superstructure is formed as the stable surface phase at a Ge coverage of 1.5 monolayers. A new surface one-electron transition appears at 1.15 eV ( at δk| ( the momentum transfer) = 0 A ̊ −1 ) and the Si dangling bond surface state transition at 1.5 eV ( at δk| = 0 A ̊ −1 ) disappears, while loss peaks due to SiSi back-bond surface states still remain. We ascribe the new peak to the Ge dangling-bond surface state transition and propose that the Ge atoms substitute the Si surface atoms which have surface dangling bonds.

The effect of surface states and band bending change on reflectivity of cleaved GaAs(110) and GaP(110)

New accurate surface differential reflectivity data of the GaAs and GaP cleavage faces are presented. Surface state transitions have been detected near 2.9 eV in GaAs(110), while in GaP(110) they appear at 2.8 and 3.5 eV. In both cases a small bulk-like Franz-Keldish effect is superimposed on the surface contribution to differential reflectivity. A method for subtracting this term out is described. The relevance of the present results in the context of the debate on surface excitons is also discussed.

Optical transitions on GaAs [110] surface

Modulation techniques have been employed in optical reflection measurements on GaAs [110] cleaved surface. Spectral bands at 2.62 and 2.83 eV observed for the light polarization E‖[ 110] have been assigned to the surface state transitions. The E 1 and E 1 + Δ 1 bulk transitions were found to be sensitive to the surface conditions.

Surface State and Direct Transitions on Al (111) Investigated with Angle-Resolved Photoelectron Spectroscopy

Al(111) has been studied with angle-resolved photoelectron spectroscopy. A previously not observed surface state band located around the K point in the surface Brillouin zone is reported. The measured dispersion of the surface state band falls in an absolute electron energy band gap when the three-dimensional band structure from a new LAPW (Linearized-Augmented-Plane-Wave) calculation is projected onto the surface Brillouin zone of the (111) crystal face. The existence of this surface state has been theoretically predicted by several authors.Direct transitions between bulk bands are found to give significant contributions to the photocurrent. Experimentally deduced initial-state energy dispersions of the bulk transitions are compared to the corresponding dispersions obtained from the LAPW band structure calculation.

Wavevector-resolved electron-energy-loss spectroscopy on the dangling-bond states of Ge(111)-(2 × 1)

Electronic surface state transitions at loss energies up to about 1.5 eV are studied by high resolution electron energy loss spectroscopy on Ge(111)-(2 × 1) surfaces prepared by cleavage in UHV. By variation of the primary energy and the detection angle the dependence of the transition energy hω on the wavevector transfer q ∥ is measured. The obtained “dispersion” hω( q ∥) is compared with results from angular resolved photoemission spectroscopy and with band structure calculations for the dangling bond bands.

Nature of Angle-Resolved Electron Energy Loss Spectrum

Angle-resolved electron energy loss spectroscopy (AREELS) has been studied on the Si(111)7×7 surface by using a retarding field analyzer in a medium electron energy range (150≤ E p ≤1000 eV) at the glancing incidence. Some fine structures are distinctly observed of surface and bulk state transitions as well as plasmon losses in the energy distribution curves N ( E ) at the specular Bragg reflection, but no transitions except for plasmon losses are observed at the back-ground position. AREELS spectra at the specular Bragg reflection are compared with those at the background position and with conventional angle-integrated electron energy loss spectra. AREELS spectra at the specular reflection are measured as a function of the primary energy around Bragg conditions.

Energy-dependent electron-energy-loss spectroscopy: Application to the surface and bulk electronic structure of MgO

The various effects that can occur when the primary electron energy, ${E}_{p}$, is varied in a reflection-energy-loss experiment are considered. For ${E}_{p}\ensuremath{\ge}100$ eV, the dominant effect is an increase of the electron mean free path with increasing ${E}_{p}$. By performing energy-loss measurements with $100 \mathrm{eV}\ensuremath{\le}{E}_{p}\ensuremath{\le}2000 \mathrm{eV}$, it is possible to unambiguously separate bulk and surface features in loss spectra. That technique has been applied to the MgO (100) surface, and both Mg intraionic and O-to-Mg interionic transitions have been studied. The former transitions are found to agree well with the excited-state spectra of free ${\mathrm{Mg}}^{2+}$ ions, while the latter agree more closely with itinerant-electron calculations of MgO. Intrinsic surface-state transitions are seen in both ${\mathrm{Mg}}^{2+}$ and O-to-Mg spectra. The Mg core-level surface-state spectra can be explained by Stark splitting of the surface ${\mathrm{Mg}}^{2+}$ levels in the intense electric fields at the crystal surface. The surface-state structure seen in O-to-Mg loss spectra agrees with discrete variational $X\ensuremath{\alpha}$ calculations of the MgO (100) surface and disagrees with linear combination of atomic orbitals calculations of the same surface. A low-energy-loss peak, possibly associated with surface defects, is seen on some surfaces; the exact nature of the surface defects involved is not yet clear.

Protoluminescence study of ultra-high vacuum cleaved germanium

We have studied the photoluminescence of Ge samples cleaved in ultra-high vacuum to better understand the role of intrinsic surface states in radiative and non-radiative recombination. We find that the surface states associated with the as-cleaved surface (2 × 1 reconstructed 〈111〉) are very efficient in quenching the band-to-band recombination at T ∼ 90 K. The transitions through surface states are predominantly non-radiative. The adsorption of oxygen, up to a half monolayer, markedly reduces this quenching effect by removing the surface state bands. Some preliminary experimental results on the question of radiative transitions involving the surface states is presented. No luminescence unambiguously attributable to surface state transitions is observed for wavelengths shorter than 5 μm.

Electron energy loss spectroscopy of the Si(111)—simple-metal interface

New data are presented on the formation of Schottky barriers on Si(111) 7 \ifmmode\times\else\texttimes\fi{} 7 with evaporated Al, Ga, or In metal. Electron-energy-loss spectra (ELS) have been taken as a function of metal overlayer coverage. The removal of clean surface state transitions has been observed at submonolayer coverage. The behavior of interface collective excitations of bulk-silicon-like transitions and of transitions from the metal core levels confirm the covalent character of the interface chemical bonds.

Electronic structure of cleaved clean and oxygen-covered GaAs (110) surfaces

GaAs (110) surfaces of $p$- and $n$-type crystals cleaved in ultrahigh vacuum are investigated by ellipsometry, surface photovoltage (SPV) spectroscopy and low-energy-electron loss spectroscopy (ELS). Changes of the ellipsometric angle $\ensuremath{\delta}\ensuremath{\Delta}$ which are induced by a Franz-Keldysh effect in the space-charge layer due to adsorbed oxygen indicate that oxygen adsorption changes the band bending of the clean surface at dosages (\ensuremath{\sim} 1 langmuir, 1 L = ${10}^{\ensuremath{-}6}$ Torr sec) much lower than those which produce measurable Auger-electron signals. In SPV spectroscopy on a clean perfectly cleaved (110) surface empty or occupied surface states can not be detected in the forbidden band whereas oxygen adsorption and/or crystallographic defects produce such states. On crystallographic irregularities oxygen also induces a new set of extrinsic surface states close to the valence or the conduction-band edge. The ELS data in combination with SPV results support the interpretation of the $\mathrm{Ga}(3d)\ensuremath{-}\mathrm{G}\mathrm{a}$ (surface state) transition in terms of a surface exciton. ELS furthermore suggests a contribution of Ga surface atoms to the chemisorption bond of oxygen. The results from ellipsometry and SPV spectroscopy can be understood by means of a model in which two discrete sets of surface states near midgap are induced by adsorbed oxygen.

Comment on Selection-Rule Effects in Electron-Energy-Loss Spectroscopy of Ge and GaAs Surfaces

Recent energy-loss and photoemission-partial-yield studies of (spin-orbit-split) $d$-core to surface-state transitions imply relative transition probabilities for the ${d}_{\frac{5}{2}}$ and ${d}_{\frac{3}{2}}$ components differing from their orbital degeneracies. This has previously been interpreted as implying $s$ character for the unoccupied surface states of GaAs(110). I present an alternative explanation for this effect assuming only $p$ character for these states and a small exchange interaction.

Observation of Excitonic Surface States on MgO—Stark-Model Interpretation

Using the energy and angle-of-incidence dependence of the electron-energy-loss spectra of MgO, we have separated the excitonic transitions from the Mg core levels to the excited states into those of bulk and surface origin. The bulk transitions are very nearly those of the free ${\mathrm{Mg}}^{2+}$ ion. The surface-state transitions can be described by Stark splitting of the energy levels of the surface ${\mathrm{Mg}}^{2+}$ ions in the intense Madelung electric fields at the crystal surface.