The use of electron paramagnetic resonance (EPR) in the study of defects in semiconductors is briefly reviewed, including group IV (C-diamond, Si, Ge, SiC), III-V (AlSb, GaAs, GaSb, GaP, InAs, InP, InSb), II-VI (BaO, BaS, BeO, CaO, CaS, CaSe, CdO, CdS, CdSe, CdTe, MgO, SrO, SrS, ZnO, ZnS, ZnSe, ZnTe) and miscellaneous svstems. The identification of defects via EPR is described as is the exploitation of that identification as a tool in future studies. Particular attention is paid to Si, where is emerging an integrated panorama of identified defects ranging from point defects to aggregates through intermediate defect configurations (as discussed by Tan) to dislocations and stacking faults; EPR results in Si as a testing ground for the theory of shallow donors, in the understanding of diffusion at high temperatures and in the study of heat-treatment defects are discussed as examples of the use of EPR as a tool in defect studies.

A defect has negative-U properties if it can trap two electrons (or holes) with the second bound more strongly than the first. It is as if there were a net attraction between the two carriers (negative Hubbard correlation energy U) at the defect, and the defect energy levels in the gap are therefore inverted from their normal order. Experimental evidence is presented that interstitial boron and the lattice vacancy, both common simple point defects produced by electron irradiation of silicon, have this unusual property. These defects represent the first and only concrete examples of negative-U centers in any material and serve as models for an understanding of the phenomenon.

Our present knowledge on self-interstitials in silicon and the rôle these defects play under widely different experimental conditions are surveyed. In particular, the following phenomena involving self-interstitials either in supersaturations or under high-temperature thermal-equilibrium conditions are considered: mobility-enhanced diffusion of self-interstitials below liquid-helium temperature, thermally activated diffusion of self-interstitials at inter-mediate temperatures (14O K to 600 K), concentration-enhanced diffusion of Group-III or Group-V elements in silicon at higher temperatures, and— as examples for high-temperature equilibrium phenomena — self-diffusion and diffusion of gold in silicon. This leads to the picture that the self-interstitials in silicon may occur in different electrical charge states and possess dumbbell configurations or are extended over several atomic volumes at intermediate or high temperatures, respectively.

The paper introduces the basic point-defect models proposed for silicon, which involve either vacancies or self-interstitials only, or both types of point defects simultaneously under thermal-equilibrium conditions. The growth and shrinkage kinetics of oxidation-induced stacking faults as well as oxidation-enhanced or -retarded diffusion phenomena are discussed within the frame work of these models. Whereas no unambiguous conclusions on the dominant diffusion mechanism can be drawn from the available oxidation-related experiments, recent investigations on so-called anomalous diffusion phenomena (e.g., the ‘emitter-push effect’) and on the diffusion of gold in silicon demonstrate Si self-interstitials to be the point defects governing self- and impurity diffusion. The possibility of a coexistence of vacancies and self-interstitials in thermal equilibrium is discussed in this context. The paper concludes with speculations on how carbon in conjunction with self-interstitials may influence the nucleation process of oxygen precipitates in silicon.

Si crystals were doped with 0.1–0.2 at% 11B in the near surface region by ion implantation followed by thermal diffusion at 1373 K or by ruby laser annealing. The position of the B atoms in the Si lattice was determined by channeling measurements, utilizing both the yield of H+ ions (of incident energy 0.7 MeV) backscattered from Si atoms and the yield of alpha particles from the 11B(p,α)8Be nuclear reaction. Initially, 95–99% of the B atoms were substitutional. Irradiation at 35 K or 293 K with 0.7 MeV H+ displaced B atoms from lattice sites. The displacement rate was greater at 293 K than at 35 K, and was greater for diffused samples than for laser annealed samples. Following 35 K irradiations, a large increase in the fraction fdB of displaced B atoms occurred during annealing near 240 K. At higher annealing temperatures, fdB decreased over a broad temperature range from 425–825 K. Angular scans through <110> channels for the laser annealed samples after 293 K irradiation or after 35 K irradiation plus 293 K annealing showed a pronounced narrowing of the dip in 11B(p,α)8Be yields compared with the dip in yields from Si, whereas no narrowing was observed for <100> channels. These results indicate that B atoms were displaced into specific lattice sites by the migration of an interstitial B defect (the EPR G28 defect) near 240 K.

The properties of sulfur-related defects in silicon are shown to differ dramatically from those that would have been expected on the basis of effective mass theory for a simple substitutional double donor. The ratio of the densities of the sulfur states as measured by capacitance-voltage techniques has been observed to vary in specimens fabricated from the same starting resistivity. Optical absorption studies have shown that the deepest sulfur level has a manifold of ground states which anneal at unequal rates at 550°C. Deep-level measurements show that the thermal emission rate at a given temperature and the variety of effects produced depends on annealing history and total sulfur density. The variability of properties of samples of sulfur-doped silicon is similar to those found for the oxygen donors in silicon, thus suggesting a chemical trend for the column VI impurities in silicon.

Junction spectroscopic techniques provide a unique tool for the study of imperfections in semiconductor materials. New, fundamental defect processes have been discovered and direct observations of the role of defects in device manufacture and operation have been made. The principles of application of capacitance transient spectroscopy, electroluminescence, and SEM-charge collection microscopy are briefly reviewed. Examples of singular accomplishments of these techniques are given, and directions of future research are outlined.

The channeling and blocking phenomena are well enough understood that they can be combined with ion scattering or ion induced reactions to investigate defects and surface structures. A brief review is given of the origins of these phenomena, of associated concepts which are useful for defect and surface analysis, and of experimental constraints and requirements. Applications to selected problems are used to illustrate the advantages and limitations of these techniques.

Channeling of fast, light ions in crystals has been widely used as a tool for studying crystal defects. Backscattering yield measurement on ions incident along major axial or planar crystalline directions provides information on the depth distribution of the structural defects in the first few microns. The channeling technique in defect detection is not as sensitive as Transmission Electron Spectroscopy, nor is it accurate in measuring the absolute numbers of defect density. Channeling measurements can give only an indication of the degree of lattice disorder. It is possible to distinguish one type of defect from another by carefully studying the energy dependence of the dechanneling. The dechanneling interpretation is not always unique, and in practice it is difficult to obtain structure information through that method. Despite these negative qualities, channeling is an attractive and unique method in certain defect studies. For example, it is sensitive for studying the lattice location of impurity atoms at substitutional or interstitial sites. Clustering of substitutional impurity atoms will show a displacement of the impurity atoms from lattice sites due to the change of bond distance. Channeling is sensitive for measuring impurity displacement as small as 0.1A°. This has been demonstrated in the study of arsenic clustering formation in Si. Interfacial relaxation and contraction in a multi-layered structure made by molecular beam epitaxy has been detected by dechanneling along various axial directions. Channeling study on surface and interface structures has developed over the past few years. In this paper, I will use examples to illustrate the unique features of the channeling technique and its application to defect studies in single crystals.

The use of ion channeling to characterize disorder in semiconductors is briefly reviewed. At high defect densities the ion channeling/backscattering technique can give the depth distribution of defects with a resolution ∼ 10 nm. Quantitative analysis of the defect depth profile requires that a single type of defect dominates the scattering of partiicles out of channeling trajectories. This scattering process is characterized by two defect-specific quantities: the direct scattering factor and the dechanneling cross-section. For impurity-associated defects the lattice location of the impurity atoms can be determined to within ∼ 0.1 Å by simultaneously measuring the impurity and host atom signals as a function of tilt angle about channeling directions. The channeling technique can detect a wide range of intrinsic defects including interstitials, dislocations, stacking faults, microtwins, and amorphous clusters. Examples of the application of channeling to study defects in silicon will be given for the cases: 1) hydrogen trapping at defects; 2) ion implantation doping; and 3) epitaxial growth of layers.

X-ray diffuse scattering can be used to study the size, concentration, and nature of lattice defects and defect clusters in crystalline materials. The availability of analytical models and detailed numerical calculations for the “Huang” diffuse scattering close to Bragg reflections and the “asymptotic” diffuse scattering at somewhat larger distances from Bragg reflections has made it possible to carry out detailed analyses of the intensity and angular distribution of the diffuse scattering in terms of specific defect parameters. Accordingly, diffuse scattering has been used to study point defects, dislocation loops, and precipitates in metals, semiconductors, and insulators and has provided information on defect geometries, size distributions, and concentrations. In this paper, the theoretical framework necessary for the interpretation of the “Huang” and “asymptotic” diffuse scattering from defect clusters is presented and examples of diffuse scattering calculations for defects in silicon are given. Diffuse scattering from clustered defects in silicon is discussed in terms of the type of information available from this scattering and the relative merits of diffuse scattering as an investigative tool for defect clusters. Diffuse scattering analysis techniques for the determination of separate size distributions for vacancy and interstitial dislocation loops in silicon are presented and applied to diffuse scattering measurements on neutron irradiated silicon. Dislocation loop densities and sizes determined in the as-irradiated state and after thermal anneals are compared with electron microscopy results.

The process of dislocation nucleation from point defect condensations in Si(Ge) is discussed. Based on the assumption that during the dislocation nucleation stage, the dominant factor in the configurational energy is the number of dangling bonds per point defect incorporated, rather than the more commonly recognized factor of strain energy, it is possible to model the dislocation nucleation process. In order to minimize the number of dangling bonds, point defects would condense into row configurations elongated in <110>, called intermediate defects (IDC), and then the IDCs would evolute into undissociated 90° edge –, 60°, and Frank partial dislocations.

Undissociated dislocation dipoles and intermediate defects are detected in As+ ion damaged Si using the lattice resolution technique of transmission electron microscopy. The present results are consistent with our proposed dislocation nucleation models.

An atomic model with fully bonded interstitial atoms is proposed for the {113} stacking faults (SF). It is shown that the development of the {113} habit plane is a result of minimizing bond length changes. The fault may be represented by a partial dislocation and a partial will facilitate an unfaulting reaction to result in a edge dislocation. Lattice images were obtained for a {113} SF which showed that the model is essentially correct.

In a lattice imaging study of As+ ion damaged Si, we have detected =110> chain type defects which are not associated with any significant strain or configurational changes. By image matching of the experimental and calculated micrographs of vacancies and interstitials, it is established that about 100% more interstitial atoms may incorporate into a defective chain. A structure model of this defect is proposed wherein a di-interstitial, occupying the =100> split position, is incorporated into every available site along a =110> chain.

Residual damage in the form of point defect clusters and amorphous regions has been investigated in ion and neutron irradiated silicon specimens. Annealing of this damage during conventional heating, flame annealing, and pulsed laser irradiation has been studied by plan-view and cross-section electron microscopy techniques. These results provide detailed information on annealing mechanisms, and emphasize the characterization of damage in the as-irradiated state.

The clustering of isolated interstitial silicon, implanted atoms, and vacant lattice sites produced by low temperature and room temperature ion implantation during subsequent annealing is reviewed. An electron microscope method for studying the kinetics of the amorphous to crystalline transformation in silicon is described. The technique is applied to measurement of the activation energy for interface migration and the formation of microtwins for different growth directions. A very brief review of the effects of laser annealing after ion implantation is included.

Neutron transmutation doping (NTD) has become the preferred method of phosphorus doping of float zone silicon for the Si power device industry. This doping process is accomplished by thermal neutron irradiation in a nuclear reactor. One of the stable isotopes, 30Si, is transmuted to the desired dopant, 31p, by thermal neutron capture and beta decay. This transmuted product becomes electrically active only after suitable annealing of the accompanying radiation damage which is the result of a number of different displacement mechanisms. This paper will present a review of our present understanding of the production and annealing of these displacement defects.

A new electron trap has been observed in electronirradiated n-type silicon at Ec − ET = 0.105 eV using transient capacitance techniques. It is found that the maximum transient capacitance response is observed only when the majority carrier pulse width is much smaller than some characteristic time constant, and when the time between pulses is much larger than another characteristic time constant. It is shown that this defect is related to an electron trap at Ec−ET = 0.172 eV (probably the oxygenvacancy or A-center); it is believed that this trap is induced by the electric field found in the depletion region of a p-n junction.

Trace radioactive impurities found in all semiconductor devices induce soft errors in semiconductor memories by α-particle emission. Data taken on 16K d-RAMs which have been fast neutron irradiated to fluences from 1013 to 1016 n/cm2 show that soft errors in these devices can be significantly reduced while maintaining acceptable device operation. Reduction in soft error rate by factors as high as 80 are observed following irradiation and thermal annealing. The effect on device parameters is discussed as well as the defects responsible for this beneficial radiation processing. Estimates of the soft error rate improvement to be expected on higher density memory devices (64K and 256K d-RAMs) will also be presented.