We report the fabrication of a 10 nm thick, self-supporting, single-crystal silicon membrane. The fabrication process can be broken up into four major stages. First, a buried SiO2 layer was formed by implantation of oxygen at a depth of 200 nm into a (100) silicon wafer. The size of the membrane was then established by removing the bulk of the silicon over a 1 mm area using a fast acid etch. After this the sample was etched in a hot EDP solution which stops at the buried SiO2 layer. The sample was then cleaned and the SiO2 layers removed, after which it was introduced into a plasma-etching chamber. The membrane was thinned down to a final thickness of 10 nm by RF plasma etching in a gas mixture of carbon tetrafluoride and oxygen. The thickness was monitored during plasma etching by measuring the intensity of He–Ne laser light transmitted through the membrane. The electron energy loss spectrum of the membrane has been measured and shows two features due to single and double plasmon excitation. The plasmon energy was 17.05 eV, in good agreement with previous measurements. Membrane thickness has also been estimated from the area of the plasmon energy loss peak. The final sample had good crystalline quality, was of even thickness over the membrane diameter and showed only a small amount of surface contamination due to the plasma etching stage.

Electron-momentum spectroscopy based on the (e,2e) reaction has been used to observe the energy-momentum density of valence electrons in the [110] direction for an ultrathin, free-standing film of crystalline silicon. An asymmetric scattering geometry is used in which the incident, scattered and ejected electron energies are 20.8, 19.6, and 1.2 keV, respectively. The measurement is complicated by the possibility of diffraction of the free electrons. The theory of the reaction including diffraction is summarized and applied to experiments with different target orientations. The orientation is determined from an independent electron diffraction experiment. Very good agreement between theory and experiment is observed.

We have measured the energy-momentum density of amorphous silicon dioxide using an (e,2e) spectrometer with 20.8 keV incident, 19.6 keV scattered, and 1.2-keV ejected electron energies. The amorphous SiO2 sample was prepared by oxidizing a thin silicon membrane. The experimental data show a valence electronic structure characteristic of upper p-like and lower s-like bands. The width of the upper valence band is 10 eV. This is separated by 9 eV from the lower valence band, which exhibits 2-eV dispersion. We have calculated the energy-momentum density of alpha-quartz using the ab initio linear muffin-tin orbital method and the result is spherically averaged over all crystal directions to enable comparison with the experiment. The calculated electron momentum densities show very good agreement with experiment for both the upper and lower valence bands. The theoretical prediction of the energy separation between the upper and lower valence bands is about 2 eV smaller than that measured and this discrepancy is discussed. The agreement between theory and experiment suggests that the short-range order in silicon dioxide plays an important role in determining the electronic structure of this material. [S0163-1829(98)04908-X].

A brief overview of the (e,2e) technique as applied to solids is reported, including the spectrometer used in these studies. In particular, we describe how the energy resolution of our spectrometer has been improved by the addition of an electron monochromator for production of the incident electron beam. This monochromator is also discussed in some detail. Results obtained using the monochromated beam are compared with previous data collected with a standard electron gun source.

We describe in detail the procedures used for the preparation of ultrathin (∼10 nm) free-standing membranes for (e,2e) spectroscopy. Such a thin target is needed to minimize electron multiple scattering before and after an (e,2e) event. The development of a rf plasma source which allows in situ thinning and thickness monitoring is of key importance to the success of the target preparation. Materials (C, Si, Ni, Cu, Al2O3, SiO2, CuO) with different properties and structures are usually prepared in different ways. For insulating targets it is important to have a conducting sublayer to avoid the charging problem. A well prepared target usually has a thin area larger than the (e,2e) beam size (∼0.2 mm in diameter) and yields high quality (e,2e) data from which the electron energy-momentum density in a chosen direction is determined. Efforts demonstrated in this article indicate that the preparation of ultrathin free-standing films is a challenging area where significant technical development is needed.

Recent developments in (e,2e) momentum spectroscopy of thin films have resulted in the study of a diverse range of solid targets. These studies have revealed the electronic structure of solids in much more detail than has been previously available using this technique. A summary of the developments which have led up to this is presented here. Some details of a spectrometer that represents the state of the art are given. Recent results from this spectrometer are discussed.

We have measured the spectral momentum densities of thin foils of diamond-like carbon using (e,2e) spectroscopy. Transmission electron energy loss spectra and (e,2e) spectra were measured before and after annealing a thin foil at around 900 degrees C and before and after thinning the foil using reactive ion etching in an argon-oxygen plasma. The valence band spectral momentum densities are compared with spherically averaged graphite and diamond band theory calculations. After annealing the surface sensitive (e,2e) data are closer to the graphite theory for the foil. Before annealing and also after plasma etching the (e,2e) data compare more favourably with the diamond theory. Bulk-sensitive transmission energy loss spectra for the annealed sample show a weak graphitic plasmon at around 6 eV energy loss which disappears after subsequent plasma etching. These measurements show that the diamond-like carbon films become graphitic only at the surface after annealing and that the graphitic surface layer can be easily removed by reactive ion etching.

The spectral momentum density of the valence band of are evaporated amorphous carbon has been measured by (e, 2e) spectroscopy with significantly improved energy resolution relative to earlier studies. The valence band has been studied over a range of momenta from 0 to 1.6 a.u. with a resolution of 0.15 a.u. and over a range of binding energies from 9 eV above to 35 eV below the Fermi energy with a resolution of 1.5 eV. As seen in earlier studies, two major peaks are observed in the spectral momentum density which previously have been associated with sigma and pi bands in graphite. A third feature in the spectra, a weak shoulder approximately 4 eV below the Fermi energy, is observed. A heuristic model is introduced based on the assumption that the spectral momentum density of evaporated amorphous carbon is an angular average of the spectral momentum density of graphite. The behavior of the strongest feature in the experimental spectra is described well by this model, but the other two features, which are in the energy range of the graphitic pi, sigma(2), and sigma(3) bands, are poorly represented by the model. It is suggested that the poor agreement is due to rehybridization of these graphitic bands.