The intensity of contrast and width of the image of a single dislocation in silicon were measured versus bias of Schottky barrier at room temperature. A regime of measurements was established, which reduced dramatically the number of variable parameters involved in contrast formation. In addition, using the right regime of SEM-CCM one can determine the position of a defect relative to the surface of the specimen without doing TEM measurements. It was shown for the first time that from contrast readout one can determine the dislocation electrical potential. A model of image formation is discussed.

The elemental semiconductors silicon and germanium can be purified to electrically active impurity concentrations as low as ∼1010cm−3. Highly sensitive, energy dispersive analytical techniques have been developed to identify and measure the concentration of the residual elemental impurities. The application of these techniques to very pure materials has also resulted in the discovery of a large number of new levels which are due to impurity/defect complexes. Photothermal ionization spectroscopy using uniaxial stress or a magnetic field, electron paramagnetic resonance, and doping experiments using stable and radioactive elements have been used in combination to identify the composition and the structure of some of the new centers.

The paper reviews isotope effects observed in the photoluminescence (or absorption) of deep defects in silicon introduced by irradiation damage, heating, quenching or during crystal growth. Three examples are discussed in more detail where photoluminescence methods, partially in combination with other analytical methods, have led to microscopic defect models.

A theoretical analysis has been made of infrared absorption spectra measured earlier for four deep-lying sulfur centers in Si. The data consist of absorption frequencies and polarization selection rules determined for uniaxial stress applied along [001], [111], and [110] axes. Both ls + np transitions and ls → ls transitions are observed. The A and B centers (binding energies 0.1090 eV and 0.1872 eV, respectively) are found to be He-like, while the C and D centers (binding energies 0.3683 eV and 0.6116 eV, respectively) are found to be He+-like. The D center spectra are consistent with Td symmetry, while the A, B and C spectra are consistent in most respects with C3v or D3d symmetry. One noteworthy result is the finding that the ls + np spectral frequencies observed under stress for the A center are consistent with a ls(E) ground state.

Iron-acceptor impurity pairs, consisting of a positively charged iron ion trapped on an interstitial site in the vicinity of an ionized acceptor, in silicon were observed by electron paramagnetic resonance for all common acceptor dopants (B, Al, Ga, In). The Zeeman splittings of these pairs, to which both spin and orbital momenta contribute, cover the range between 1.1 and 6.4. An interpretation of these spectroscopic splitting factors is presented, which considers the effects of the crystal field - of cubic, axial, or lower symmetry - and of spin-orbit interaction on the 4F ground state of the iron ion in a (3d)7 configuration. It is concluded that the apparent quenching of the orbital angular momentum is not due to a dynamical Jahn-Teller effect, nor due to hybridization. Rather, it is proposed that a significant reduction, by about 80%, of the orbital magnetism arises from covalency.

We present theoretical results for the optical prooerties of the Pb center at the Si/SiO2 interface. Using wave functions obtainec from semiempirical (MINDO/3) cluster calculations, we have calculated electric dipole matrix elements connecting the singly occupied (neutral) defect state to the unoccupied conduction-band-like states, as well as those connecting the occupied valence-band-like states to the singly occupied defect state and to the unoccupied defect state. We predict the absorption cross section for excitation from the valence band to the unoccupied state to be of order 10−19 cm2 and that for excitation from the valence band to the occupied state and from the occupied state to the conduction band to be an order of magnitude larger. We also predict that the absorption will in some cases be strongly dependent on the direction of the polarization. Effects of symmetry lowering in the oxide and of distortions in the silicon are discussed.

The silicon “dangling bond” defect plays a large part in controlling the electronic properties of a-Si, polycrystalline silicon, and the Si/Si0 2 interface. Jackson et al. have suggested that transitions of electrons occupying this defect produce the Urbach-like sub-gap absorption tail seen in two of these materials. We have performed optical and electron spin resonance measurements on polycrystalline silicon, plastically deformed silicon, and Si/Si02 interfaces to further examine this contention. In addition to seeing no measurable absorptance due to dangling bond interface states in the latter system, we conclude from the poor correlation of ESR signals with optical data that the Urbach tail in polycrystalline and deformed silicon is not due to transitions of dangling bond electrons.

The optical properties of the trivalent silicon dangling bond defect in hydrogenated amorphous silicon and at the Si/SiO2 interface are compared. While both defects give rise to ambipolar deep levels within the gap, significant differences in the optical properties between the two systems are found. The absorption in a-Si:H is significantly stronger and is dominated by a transition from the defect to the conduction band while the absorption at the interface is dominated by hole emission. The average dipole matrix element squared is roughly independent of energy in both systems with a magnitude of ∼30Å2. Implications of these results for optical measurements in other silicon systems are discussed.

EBIC evidence of the key role of dislocations in hydrogen passivation of EFG ribbon is presented. Dislocations, and not bulk point defects, are shown to be both the principal defect being passivated and the means whereby rapid thermal diffusion of hydrogen takes place. Passivation of dislocation arrays to depths of the order of hundreds of microns has been observed. Low stemperature EBIC (96°K < T < 300°K) has shown that shallow electron traps associated with dislocations and within ∼0.1 eV of the conduction band are not passivated while (proposed) deeper mid-gap levels at the same spatial location are readily passivated. A general dislocation model for hydrogen passivation applicable to all polycrystalline silicon is presented.

The near-surface hydrogen profile was measured using the 15N hydrogen profiling technique in silicon ribbon grown by the edge-defined film-fed growth (EFG) process. This direct method has a high depth resolution of 10 - 30 Å, can be used to a depth of several microns, and can measure hydrogen in concentration of 100 at. ppm. By appropriate surface treatment we were able to observe for the first time the separation of the near-surface hydrogen profile in silicon from the surface hydrogen contamination layer. Using our technique, hydrogen profiling of the near-surface region revealed the existence of a subsurface hydrogen layer which acts as a barrier to the transfer of hydrogen into the bulk of both passivated and untreated silicon. The structure of this hydrogen barrier was measured for different plasma treatments.

Secondary electron imaging, electron channeling and electron-beaminduced current (EBIC) are used alternately in a scanning electron microscope to characterize and correlate the morphology, crystallographic orientation, and electronic quality (types and spatial distribution of defects) of individual grains in polycrystalline semiconductor samples. The technique is discussed in some detail, and a number of applications and results in the study of edge-supported-pulling (ESP) silicon sheet and low-angle silicon sheet (LASS) materials are reported.

The energy-distribution of interface states at grain boundaries in fine-grained silicon films is measured by two independent methods. The analysis reveals deep exponential band tails in the two-dimensional density of states within the band gap. It is proposed that these tails originate from Anderson localization of free carriers within potential fluctuations along the grain boundary plane. The measurements are in good agreement with this model.

Thin films of cubic SiC were grown on Si substrates by a new CVD heteroepitaxial technique. The cubic polytype and crystal quality were verified by room temperature Raman spectroscopy. Low temperature photoluminescence (PL) studies detected nitrogen donor bound excitons, N-Al donor-acceptor pairs in Al-doped samples, and some defect complexes usually observed only in irradiated samples. Evidence of strain due to the Si-SiC lattice mismatch was observed in both the Raman and PL spectra and the effects of substrate removal and high temperature annealing on these optical spectra were studied.

The N1 centre in Type Ia diamonds is a di-nitrogen centre with S = ½. One nitrogen's interaction with the unpaired electron at room temperature gives a large splitting, All = 46.4 G, A ⊥ = 32.2 G with <111> symmetry while the hyperfine splitting due to the other one has <110> symmetry with parameters All = 2.95 ±.03 G, A⊥ = 2.81 ±.03 G. As the temperature is increased averaging effects start; first some hyperfine lines broaden at about 150°C and disappear at 200°C. At still higher temperatures new lines appear and at about 850°C the averaging process is completed. The interpretation of the dynamics leads to a model of a non-planar N-C-C-N configuration with the electron jumping between the two N-C bonds. The measured activation energy is 0.4 eV which is about midway between the measured values for a similar effect in the isolated nitrogen centre, P1, (E = 0.7eV) and the N-C-N, W7, centre (E = 0.24 eV).

Techniques have been developed at NCSU for fabricating cross-sectional transmission electron microscopy (XTEM) foils from monocrystalline beta silicon carbide thin films grown by chemical vapor deposition. The results of the TEM observations are utilized to discern the efficacy of the various processing parameters in terms of film quality and defect structure as well as oxidation, ion implantation and annealing procedures.