The application of transient junction current and capacitance techniques to the study of imperfection in semiconductor materials is reviewed. An array of perturbation techniques are described which allow direct determination of electronic and atomic structure, as well as electrical and physical properties. The methods are illustrated with silicon materials studies of the divacancy using Polarized Excitation Photocapacitance, the oxygen donor using Stress and Electric Field Modulated DLTS, dislocations using spatially resolved DLTS, and iron impurities employing Charge State Control of Structure.

This talk will review and compare very basic intrinsic and extrinsic paramagnetic defect structures which are characteristic for group IIB-VI, III-V and elemental semiconductors. The case of cation vacancy related defects in ZnS and ZnSe, that of antisite defects in GaP and GaAs and that of deep impurity donors in IIB-VI semiconductors and silicon will be treated as representative examples, thus illustrating the possibilities (and limitations) of the ESR technique.

A review is given of the information that can be obtained on defect centers in semiconductors by optical spectroscopy. Particular emphasis is given to donor and acceptor identification, and symmetry determination of transition metal and axial defect-complex centers. The information that can be obtained from isotope doping effects is discussed.

Knowledge concerning the electronic properties of defects is crucial for the understanding of the behaviour of these defects in semiconductors. Considering the important role defects play in most currently used devices it is not surprising that a great number of different measurement techniques have been developed with the aim of not only quantitatively determining the electronic parameters but also of identifying the defect chemically and evaluating the local environment of such defects. In most cases this goal can not be achieved by applying only a single (defect property) measuring technique.

In this paper an attempt is made, using two different examples, to show what achievements in defect characterization and identification can be obtained by combining two or more different measurement techniques. The first example is a single substitutional impurity, sulfur in silicon, while the second example deals with a more complex defect, the oxygen-vacancy center in silicon. In both cases a number of techniques such as different junction space charge measurements, Hall effect, absorption, ESR, photo-ESR, ENDOR photoconductivity, irradiation and annealing processes, and even theory were used to study the defect. The advantages and limitations of various methodologies will be discussed.

This paper addresses the contributions modern theoretical developments have made to the understanding of the physics of deep defects in semiconductors. The nature of theoretical results will be discussed and underlying approximations and limitations will be examined.

The physics of deep levels is reviewed, with emphasis on the qualitative physics that has been elucidated as a result of the ideas of Lannoo, Lenglart, Hjalmarson, Vogl, Wolford, Hsu, Sankey, Allen, and others.

Thermodynamnies is obviously central to the ultimate goal ol an exquisite control of the electronic properties of scmiconducting devices and to the uinderstanding this will require. It should also be useful for the subsidiary goal of the firm identification of the various electrically active defects. This subsidiary goal will not be achieved unless adequate attention is paid to the reactions among the various atomic and ionization states of the defects, the electrons and the holes.

A theoretical study of native defect complexes in GaAs is reported here. These calculations are based on tight-binding Hamiltonians and employ a large cluster recursion method to obtain the changes of local densities of states. A new description of the tight-binding matrix elements developed by us is used to take into account the lattice relaxation and to provide the interaction matrix elements between the antisite defect and its neigiitors. We have examined the complexes GaAsVGa, (VGa)2, VGaAsGaVGa, VGaAsGaVGa, AsGaVAs and VAsAsGaVAs. The results indicate that the use of realistic interaction between the antisite atom and its neighloors is essential to obtain an accurate result for the midgap state of the isolated antisite and that the GaAsVGa,AsGaVAs, (VGa)2 may be responsible for the A,D hole traps and the electron trap C as proposed recently by Zou et al. using a thermochemical model. Both VGaAsGaVGa and VGaAsGaVAs are found to produce states in the gap. It is not possible to exclude either of them as a candidate for the EL2 state. A detailed analysis of the states induced by the defect complexes is made.

Self-consistent spin-unrestricted all-electron Green's function calculations are reported for the first time for a series of interstitial 3d impurities in silicon. The calculations, performed within the selfinteraction- corrected local-spin-density fromalism show: (i) not all 3d impurities follow Hund's rule: Tio, Ti−, Vo, V+ and Co2+ have a low-spin ground state, (ii) the angular momentum part gL of g-value is quenched due to p-d hybridization effects, (iii) covalency explains also the chemical trends in the central hyperfine coupling constants, (iii) chemical trends in donor andacceptor transitions are reproduced and are consistent with a high-spin to lowspin transition at the low-Z end and high-Z end of the 3d series, (iv) A number of predictions are offered.

We show how self-consistent total-energy calculations can be used to identify the position of defects in semiconductors. Despite intensive experimental research on S, Se and Te point defects in Si, it has remained unclear whether these impurities occupy substitutional or Td-interstitial sites. Our Green-function total-energy calculations show that the substitutional site is favored by several eV and therefore the stable defect position is identified as substitutional. We further consider the formation energies of distant defect pairs consisting of a substitutional chalcogen and a Si self-interstitial and we study the reaction where the two constituents change places.

Theoretical calculations of the finite temperature properties of defects, such as defect concentration and diffusion rate, require a knowledge of the entire 3N-dimensional defect total energy surface. We present a formalism for obtaining such defect energy surfaces from total energy calculations. Preliminary results of the application of this formalism to the silicon selfinterstitial are described. Implications for understanding the remarkable experimental high temperature diffusion entropies are discussed.

The states of lowest energy have been calculated for iron-group (3d) transition-metal impurities in silicon. The donor and acceptor levels reproduce all experimentally observed transitions and trends. The theory predicts that the ground states of the early 3d interstitials and late 3d substitutionals have low spin. This is in conflict with the generally accepted model of Ludwig and Woodbury, if applied to these impurities, but not with existing experiments.

Several recent theoretical studies of the local structure of semiconductor alloys are summarized. First, dilute limit calculations of local bond lengths and mixing enthalpies are discussed. These calculations include effects due to both bond length and bondangle distortions, as well as local chemical rearrangements. Then, a new statistical theory of concentrated alloys is described. Deviations from random alloy distributions (microclusters) are predicted.

The electronic structures of substitutional and tetrahedral-site interstitial Hg+, Auo and Pt− isoelectronic impurities in silicon have been analysed. The centers are theoretically described by the Watson-sphereterminated molecular cluster model within the framework of the multiplescattering Xa formalism. At the substitutional sites the centers are related to the “vacancy” model recently proposed to describe the properties of the elements at the end of the transition-metal series. At the interstitialsites the impurities introduce a hyperdeep s-like level close to the bottom of the valence band and, in agreement with experiments, do not show shallow donor activities. For all the analysed centers the d-states remain fully occupied below, or within, the valence band.

The incorporation process of nonequilibrium vacancies in melt-grown GaAs is strongly complicated by deviations from stoichiometry, and the presence of two sublattices. Many of the microdetects originating in these vacancies and their interactions introduce energy levels (shallow and deep) within the energy gap. The direct identification of the chemical or structural signature of these defects and its direct correlation to their electronic behavior is not generally possible. We must, therefore, rely on indirect methods and phenomenological models and be confronted with the associated pitfalls. EL2, a microdefect introducing a deep donor level, has been in the limelight in recent years because it is believed to be responsible for the semi-insulating behavior of undoped GaAs. Although much progress has been made towards understanding its origin and nature, some relevant questions remain unanswered. We will attempt to assess the present status of understanding of EL2 in the light of the most recent results.

Anion antisite defects are the first convincingly identified intrinsic lattice defect in GaAs. Photo-EPR allowed to determine the energy levels of this double donor, with the single donor level being midgap. This level appears to be very similar to the technologically important “EL2” center in GaAs. However, various evidence indicates that EL2 is not due to ideal, undistorted antisite defects.

Fermi level pinning at Schottky contacts on GaAs has been ascribed to intrinsic lattice defects. Antisite defects are a likely candidate for this process.

Finally, the microscopic identity of the “DX” center in AlGaAs will be discussed. This defect has been generally identified with a complex of a donor atom and an unknown partner. A number of recent experimental observations indicate that this defect might rather be due to isolated donors, which get deep levels because of the change of bandstructure in the alloy system.

Detailed DLTS measurements have revealed the presence of threedifferent midgap levels, labeled EOI, EO2 and EO3, in LEC n-GaAs. The variations of EOl and EO2 concentrations, which are EL2 and ELO in Lagowski's terminology, do not correlate with the distributions of the optical absorption coefficient across wafers, while that of EO3 exhibits good correlation. This suggests that EO3 is the optically detected midgap level. Furthermore, preliminary charged particle activation analysis of oxygen content in GaAs shows that the presence of EO2 (ELO) is not attributed to oxygen. Concentrations of these levels all increase with arsenic mole fraction in the melt.

Anion antisite defects in GaAs have been identified so far by their electron spin resonance (ESR) spectrum as one specific point defect, which was found in as-grown, plastically deformed and electron and neutron irradiated GaAs. By optical detection of electron nuclear double resonance (ODENDOR),it could be shown that several species of anion antisite defects exist having practically the same ESR spectrum, which cannot any longer be taken as “fingerprint” identification for this defect. Recent results of ODESR and ODENDOR investigations in as-grown, plastically deformed and electron irradiated GaAs are reviewed. It is also pointed out that the spin lattice relaxation time is an important quantity for the microscopic identification and the properties of anion antisite defects.

An investigation of the photoquenching behaviour in semi-insulating GaAs at low temperature shows that the ASGa antisite MCD-ODMR is persistently quenched while a broad vacancy-like resonance near g=2, which shows a variation of g-factor with absorption energy, broadens and becomes energy independent. The results are discussed in terms of interacting antisite and vacancy centres in analogy with the distant pair model for donor-acceptor pair recombination and the advantages of the triplet model for the EL2 metastahle state are reviewed.

The photo-response of the ASGa+ antisite electron-paramagnetic-resonance (EPR) has been studied in as-grown GaAs as a function of illumination time and photon energy. The results establish a firm and positive correlation between ASGa and the deep donor level EL2. Spatially resolved EPR measurements show that the ASGa concentration can fluctuate by about a factor of two across a 2-inch semiinsulating GaAs wafer.