An approximate method to correct band structures for the effects of the nonorthogonality of a basis set consisting of degenerate orbitals is derived and applied to the bond-orbital model for tetrahedrally coordinated semiconductors. It is shown that the band structure corrected for the nonorthogonality of the basis orbitals is approximately related through a simple transformation to the band structure calculated by disregarding the nonorthogonality. Expansion of this transformation shows that nearest-neighbor overlap can produce terms in the band structure which are similar to those produced by more distant-neighbor interactions. For the bond-orbital model, the most important effect of nonorthogonality correction is to increase the curvature of the $p$-like bands at $\ensuremath{\Gamma}$, bringing the overall agreement of the model band structure into still better agreement with experiment.

Energy distribution spectra of photoelectrons excited by monochromatized $\mathrm{Al} K\ensuremath{\alpha}$ radiation (XPS) from the valence bands of the ${\mathrm{Mg}}_{2}X$ semiconductors are presented. The results are compared with calculated densities of valence states obtained from parametrized tight-binding bands. The nearest-neighbor interaction parameters are determined so as to obtain theoretical total densities of valence states which fit the experimental XPS spectra and partial densities of states which agree with soft-x-ray emission data. The upper, $p$-like valence bands so determined agree well with the empirical pseudopotential calculations of Au-Yang and Cohen while discrepancies appear for the lower $s$ bands. The parameters of the tight-binding model are compared with those reported recently by several authors for the elemental semiconductors.

AgCl and AgBr have been investigated by photoemission for photon energies $h\ensuremath{\nu}=16.8, 21.2, ,26.9, 40.8, \mathrm{and} 1486.6$ eV. By exploiting the strong dependence on $h\ensuremath{\nu}$ of the photoionization cross sections for the atomic orbitals composing the valence bands, we have been able to deduce approximate partial $p$ and $d$ densities of valence states for these compounds. The most conspicuous feature of the photoelectron distribution curves is a sharp peak near the center of the valence band, which our partial density of states shows to be mostly $d$-like. The $p$ density of states exibits two main peaks centered at 2- and 5-eV binding energy and a gap at the energy of the sharp $d$ peak. From the $p$ and $d$ densities of states we conclude that, except for the sharp $d$-like peak and the associated $p$ gap, the $p$ and $d$ functions are nearly equally and uniformly mixed throughout the valence band. The results are compared with partial densities of states calculated with a simplified tight-binding scheme and the band structure is discussed in detail.

Photoemission from AgCl, AgBr, and AgI has been studied at photon energies hω = 16.8, 21.2, 26.9, 40.8, 48.4 and 1486.6 eV. By exploiting the strong

By exploiting the strong dependence of the photoionization cross section for the Cu 3d and I 5p levels on photon energy, p and d partial densities of valence states in CuI have been determined. The results compare well with other experimental information on CuI.

The densities of valence states (DOVS) of the amorphous and crystalline forms of GaP, GaAs, GaSb, InP, InAs, InSb, AlSb, ZnTe, and CdTe have been determined from the energy-distribution spectra of photoelectrons emitted by high-energy photons (16.9, 21.2, 40.8, and 1486.6 eV). In general the DOVS of the amorphous forms can be represented by a broadened version of those of the corresponding crystalline forms. Fine structure which appears in the upper valence bands of the crystalline materials, due to critical points at $L$, $X$, and $W$, is completely washed out in the amorphous phase. The core-level spectra have nearly the same positions and widths in the amorphous as in the crystalline modifications. This fact indicates that the fluctuations in the Coulombic environment about each type of atom are small, suggesting that the structure is homogeneous and contains an insignificant number of odd-membered rings. The plasma frequencies, determined from the plasma-loss spectra associated with core levels, are the same in the amorphous as in the crystalline phases to within 3%. This fact enables us to conclude that the densities of both modification differ by less than 6%. We present a simple bond-charge model which can simulate realistically the density of valence states of germanium and zinc-blende-type semiconductors. The valence bands at any point of the Brillouin zone are obtained in this model as the solution of a 4\ifmmode\times\else\texttimes\fi{}4 secular equation. Within this model, the structure of the top $p$-like valence bands depends primarily on overlap between second-neighbor bonds. Thus fluctuations in the position of second neighbors can be invoked to explain the smearing of the fine structure of these bands in the amorphous modifications. A simple model which relates the chemical shifts of the compounds to their ionicity is also discussed.

The photoelectron energy distribution curves for CuCl, CuBr, CuI, and AgI have been measured for $\ensuremath{\hbar}\ensuremath{\omega}=16.8, 21.2, 26.9, 40.8, 48.4, \mathrm{and} 1486.6$ eV. By exploiting the strong dependence on $\ensuremath{\hbar}\ensuremath{\omega}$ of the photoionization cross sections of the atomic levels comprising the valence band, we have been able to determine the $s$, $p$, and $d$ partial densities of valence states. These results compare well with those obtained by x-ray fluorescence and x-ray photoemission, but have better resolution. It is found that the $d$ levels of ${\ensuremath{\Gamma}}_{12}$ symmetry remain corelike, while the ${\ensuremath{\Gamma}}_{15}$ $d$ levels band significantly. In the Cu compounds the ${\ensuremath{\Gamma}}_{15}$ levels of predominantly $d$ character lie above the ${\ensuremath{\Gamma}}_{12}$ bands, which in turn lie above the mainly $p$-like ${\ensuremath{\Gamma}}_{15}$ levels. This level ordering is reversed in AgI. The band structure and densities of valence states computed from a seven-function basis set, consisting of four bonding $s\ensuremath{-}p$ orbitals and three ${d}_{\mathrm{xy}}$ orbitals, are shown to reproduce the trends observed in the density of valence states of these compounds.

AbstractThe density of valence states, the energies of the core levels, and the characteristic energy loss spectra of crystalline and amorphous GeTe have been measured by X‐ray (1486.6 eV) and UV (21.2 eV) photoemission spectroscopy. The valence band densities of states of the two GeTe phases are in general similar, but exhibit distinct differences in their detailed structure. That of the amorphous form of GeTe may be approximated by a sum of the densities of states of the elemental forms of amorphous Ge and Te. A model for the valence band density of states is proposed which may be generally applicable to random covalent networks. No difference in the separation of the core level of the Ge and Te atoms is found between the amorphous and crystalline phases, thus indicating that there is no significant ionicity difference in the bonding character of these two phases. In light of the current structural models for the amorphous form, the similarities between the photoemission from the two phases are surprising. The plasma frequencies of the amorphous and crystalline forms are found to be 16.4 and 16.9 eV, respectively.

Photoemission experiments have been performed on amorphous and crystalline films of Se and Te using photon excitation energies of 1486.6 ($\mathrm{Al}K\ensuremath{\alpha}$), 40.8, and 21.2 eV (He II and He I). From these experiments we have determined the densities of valence states, the binding energies of the core levels, the characteristic loss functions, and the plasma frequencies. Comparisons of the experimental to the theoretical densities of states are made for both the amorphous and crystalline forms. The optical properties of Te in the region of interband transitions are shown to be described well by a simple model based on the density of valence states. Evidence that the optical absorption peak from the Se $3d$ level may be of excitonic origin is presented.

AbstractThe energy distribution spectra of photoelectrons (XPS) and Auger electrons (AES) excited with AlKα radiation in Mg2Si, Mg2Ge, and Mg2Sn are reported. For the purpose of determining core shifts, measurements were also performed on the elemental constituents. These measurements yield information about the density of valence states, the position and the chemical shifts of the core levels. The energy of the valence plasmons is also obtained. This paper (I) is followed by a second one (II) in which similar results obtained with far UV excitation are reported and discussed in the context of the present work.

AbstractThe density of valence states of the II‐VI compounds ZnSe, CdS, and HgS has been determined with far uv (h v = 21.1 and 40.8 eV) photoemission. Discrepancies in the widths of the top portions of the valence bands between theory and experiment, similar to those reported for other II‐VI compounds in part I, are also found here. The results for ZnSe are compared to those obtained by others using synchrotron radiation and monochromatized X‐rays.

In the previous paper (I) the energy distribution spectra of photoelectrons (XPS) and Auger electrons (AES) excited with AlKα radiation in Mg 2 Si, Mg 2 Ge, and Mg 2 Sn were presented. Here similar measurements are reported performed with He resonant radiation (48.4, 40.8, and 21.2 eV). The effect of surface contamination by O 2 or H 2 O is shown to be very important in these UPS spectra and results meaningful of the material under study can only be obtained if such contamination is kept to less than 0.2 monolayers. The strong structure which appears in less clean samples has been identified as related to the compounds MgO and Mg(OH) 2 . The UPS spectra of the clean materials yield information about the densities of valence states and about the outermost d electrons of the Ge and Sn core. Very accurate values for the corresponding spin‐orbit splittings and chemical shifts are obtained.

It is shown that Auger spectroscopy can be used to determine the density of valence states in compound semiconductors. We determined the density of valence states by this method in ${\mathrm{Mg}}_{2}$Sn and find it in basic agreement with those which we obtained by x-ray-and uv-induced photoelectron spectroscopy. A comparison is made with the calculated ${\mathrm{Mg}}_{2}$Sn valence band. The ${\mathrm{Mg}}_{2}$Sn valence band obtained by Auger spectroscopy (involving the $\mathrm{KL}$ transition in Mg) is contrasted with the analogous spectrum for Mg metal.

The densities of states of the entire valence bands of amoprhous and trigonal selenium have been measured by X-ray (hv = 1486.6 eV) and u.v. (hv = 21.2 eV) photoemission. It is found that the primary differences in the density of states of the two forms occurs in the p-like valence bands, in particular the lower bonding set. The density of states calculated by Kramer et al. for the amorphous form is shown to be in better agreement with the experimental density of states of the trigonal form than with that of the amorphous form.

The use of both far UV (h u = 21.2 and 40.8 eV) and X-ray (h o = 1 486.6 eV) excitation sources is shown to expand the power of photoemission experiments in determining the density of valence states in solids. When data from both excitation sources are used, the effects of photoionization cross-sections, final densities of states, secondary electrons, and resolution can be identified and eliminated. Thus, a better picture of the density of states can be obtained than with each source used independently. It is also shown that the 21.2 eV line can be used to determine large portions of the density of valence states in 11-VI compounds with better resolution than can be obtained with monochromatized X-rays. High energy photoelectron spectroscopy is now a well established, valuable technique for determining the density of valence states in solids [ I ] . The most troublesome difficulties remaining in the extraction of the density of valence states from the photoernission data arise from instrumental resolution, energy dependent photoionization cross sections, s t r ~ ~ c t u r e in the final density of states, and contributions froni inelastically scattered electrons. Most of tlie recent works on senliconductors have been performed on complex experimental syste~iis using either synclirotron radiation [2] o r monoc111-omatized X-rays [3] as excitation sources. Less ideal, but simpler experimental systems, equipped with a helium discharge lamp o r a no~i-moiiocliro~~i~~tized X-ray tube, have no t been ~ ~ s e d with as much success in the past. Here, we wish to sliow that by retaining the best features of data obtained from a system equipped with a differentially pumped h e l i ~ ~ m discharge lamp and a standard Al X-ray tube, we can overcome many of the difficulties hindering the interpretation of the data. Especially, we wish to sliow that with these sources used together, we can greatly expand the power of photoemission experiments in determining the electronic structure of solids. We have listed in table I some typical data of the photoelectron spectra provided by the heliu~n discharge lamp (He I and He 11) and the A1 X-ray tube (AIKu). Other gases, such as neon (Ne I, 17 eV ; Ne 11, 27 eV) can be used in the discharge lamp. Here we consider only helium since it provides the most commonly used and the most useful type of excitation. The dominant line in each spectrum occurs at energies of 21.2, 40.8 and 1 486.6 eV, thus providing a wider range than has been used in experimental systems equipped with a synchrotron source [2]. With a wide range of energies available one can infer variations in the photo-ionization cross sections across the bands. Also, since lighter contaminating elements rapidly lose their photoelectric cross sections in the X-ray region, the X-ray spectra are less subjected to their efrects. Agreement found with all three spectra (He I , He 1 1 , and AlKu) assures 11s that tlie density of valence states characteristic of the material is obtained. Each spectrum has a weaker component occurring a t higher energies than the stronger component ; for tlie He 11 and AIKu spectra, tlie weaker component has 10 "/, of the strength of the stronger conlponent and for the He I spectra, it has 2 '::; of the strength. Unfortunately, the electron energy distribution curves consist of the sum of electrons c~nit ted by each component, thus some complications may arise in interpreting data obtained from them. As we shall see shortly, tliese weaker

The diffracted x-ray intensity and Raman scattering of amorphous ${\mathrm{Ge}}_{0.5}$${\mathrm{Si}}_{0.5}$ prepared by sputtering has been measured. The x-ray data indicate that the structure is most likely a random mixing of Ge and Si atoms in a continuous random network structure similar to that proposed for amorphous Ge and Si. The reduced Raman intensity indicates vibrational features that can be associated with Ge-Ge, Ge-Si, and Si-Si bonds. The high-frequency Si-Si-like vibrations are more severely influenced by the compositional disorder than the low-frequency Ge-Ge-like modes.

AbstractThe energy distribution spectra of electrons photoemitted from the lead chalcogenides by 48.4, 40.8, and 21.2 eV photons have been measured. Both the valence band and the 5d electrons of Pb are observed in these spectra. The valence band portions exhibit considerable structure which is compared with theoretical calculations and with recent X‐rays photoemission work. The d‐electrons spectra are extremely sharp (line width 0.5 eV). They are also compared with theoretical predictions (OPW, KKR, and pseudopotential calculations), with the result of X‐ray photoemission, and with the corresponding spectra of pure Pb. A discussion of the relative intensities of the valence and d‐electron energy distribution spectra is presented.

AbstractThe density of valence states of the II–VI compounds ZnTe, CdSe, CdTe, HgSe, and HgTe have been determined with far uv (h v = 21.2 and 40.8 eV) photoemission (UPS). Although the upper portions of the valence bands are found to have shapes in qualitative agreement with theory, they are consistently 1 to 2 eV wider than predicted. With the high resolution obtainable in uv photoemission, the shapes of the d‐bands of the cation can be resolved to permit an accurate determination of their positions relative to the top of the valence band and their spin‐orbit splittings. The “apparent” spin‐orbit splittings of the uppermost d‐orbitals of Zn and Cd are found to be 50% larger in the metallic form then in the compounds.

The photoemission from A1Sb films sputtered on low and high temperature substrates has been measured for both X-ray (1486.6 eV) and far u.v. (40.8 and 21.2 eV) incident photon energies. The X-ray photoemission from the valence band is found to resemble the valence band density of states typically calculated for III–V semiconductors. The valence band photoemission for the u.v. incident photons, however, looks considerably different; it is proposed that this difference is produced by the small escape depth of the photoelectrons. The u.v. photoemission is thus more representative of the density of electronic states near the surface. Also, the core levels of A1Sb, and the plasma frequency of the valence band electrons, determined from the X-ray photoemission, are presented.