A high resolution optical study was carried out on GaAs crystals containing only the EL2 midgap level. The no-phonon line at 1.039 eV was identified as an intracenter transition within AsGa. An excellent correlation was found between the intensity of the no-phot1n AsGa line and the characteristic EL2 absorption. A correlation was also found between the characteristic optical absorption of EL2 and its concentration as determined by DLTS measurements. These results prove the direct association of EL2 with a neutral state of AsGa The presence of another midgap level ELO, contained in heavily doped oxygen crystals had no effect on the optical spectra of EL2, but it did alter the DLTS measurements. Thus, an accurate calibration was obtained for the determination of EL2 by optical absorption based on DLTS measurements on crystals containing only EL2 and thus involving no uncertainties introduced by other midgap levels.

2" diameter wafers cut from indium doped, liquid encapsulated Czochralski (LEC) grown GaAs ingots have been examined by two different methods. Dislocations were revealed by etching, and isolated regions of enhanced [EL2], the concentration of the deep donor level EL2, weredetected by infrared transmission microscopy. In regions of low dislocation density (∼5 × 102 cm−2) single isolated dislocations were identified and at the identical sites infra-red micrographs showed increased absorption. The results demonstrate that enhancement of [EL2]occurs at single dislocations, and that the enhancement can be detected by suitable etching techniques.

Measurements are reported, for 0.5 mm thick semi-insulating (S.I.) GaAs wafers, of the neutral EL2 distribution, the pattern of dislocations, and of MESFET electrical parameters - these devices built on the same wafers. These results indicate that the device properties are more affected by EL2 nonuniformity than by dislocation density pattern.

The application of electron nuclear double resonance (ENDOR) for identification and characterization of point defects in silicon is reviewed. Taking the vacancy and the boron-vacancy complex as examples it is discussed how ENDOR can provide information on the atomic and electronic structure of paramagnetic centers.

With Electron Nuclear Double Resonance (ENDOR) ligand superhyperfine (shf) interactions were investigated for 7 neighbor shells of the interstitial Al ++ center (Si-G18) and for 10 neighbor shells of the Al ++-Al-pair center (Si-G19). All shells of the interstitial center exhibit undistorted Td symmetry confirming the defect model derived from earlier experiments. In the pair center the shf interaction of the Al++ with the neighboring Al nucleus shows a pure [111]-symmetry, also the shf interactions with the outer Si-neighbor shells exactly reflect this axial [111]-symmetry of the defect. The magnitude of the quadrupole interaction of the neighboring Al nucleus clearly shows, that this nucleus is a nearest neighbor of the Al++. From the angular dependence and the intensity ratios of ENDOR lines of the pair center it can be excluded that any additional impurity is associated with the Al++-Al- pair at least within the first 3 neighbor shells of the AI++. The influence of the Al- neighbor on the hyperfine axes of the first [111] Si-neighbors is negligible and is also remarkably small for the other Si-neighbors. The magnitudes of the ligand shf interactions for the pair center are closely related to those of the Al++ center. A comparison of these values for both centers allows a definite assignment of the measured shf interactions to their corresponding neighbor shells for the Al++ center.

In this brief review we consider the vacancy-related, the vacancy oxygen-related, and the vacancy-hydrogen-related defects. We note the common opportunity for chemically-driven partial dissociation of defects. With this background we briefly survey what is known of the oxygen agglomerates, noting that we favor the ylid ( the saddle-point for oxygen diffusion) as the thermal donor core, but that the (vacancy + di-oxygen) complex can also be a core, and that the latter defect can occur when the strain-energy reaches the.point that a vacancy-interstitial pair can be created, causing interstitial emission from the oxygen agglomerate. We note as well that the emission of Si≡0 is energetically favored versus emission of an unbonded interstitial.

The effect of uniaxial stress on the NL8 EPR spectrum in 450°C heattreated silicon containing thermal donors (TD's) is described. Changes in the relative amplitudes of the NL8 spectral components are interpreted as arising from electronic redistribution between the differently oriented defects. These changes are consistent in sign and magnitude with those predicted for TD+ from observed stress effects on the TD IR and DLTS spectra.

Rapid thermal annealing of thermal donors in Si with 10 sec anneal times at temperatures between 600 and 1000 °C has been investigated by infrared absorption at 80 K. Thermal donors A through D, which are identified by excited state absorption, are present in as-grown Czochralski Si; whereas excited states for donors A through F as well as photoionization of thermal donors are observed after extended heating at 450 °C. The temperature required for rapid thermal annealing is lower when only donors A through D are present. Removal of thermal donors A through F by rapid thermal annealing at temperatures > 800°C restores 7 to 8 oxygen atoms to interstitial sites per electricallӯ measured donor removed. This ratio supports oxygen cluster models for thermal donors but does not support previous suggestions that such clusters are embryonic forms of high temperature oxygen precipitates.

Spatial variation of interstitial oxygen concentrations ([Oi]) have been observed in wafers sliced from CZ grown ingots. Oxygen precipitation can occur with subsequent heat treatments causing non-uniform electrical device characteristics across a processed wafer. A new method is presented using infrared absorption measurements to map [Oi]in silicon wafers for correlation with wafer electrical measurements. The new infrared mapping method uses a spectrometer fitted with a software controllable wafer positioner. 100 m m diameter wafers were mapped to show the concentration variations and oxygen precipitation effects resulting from thermal heat treatments. Electrical measurements were made on nearly adjacent wafers from the same ingot and subjected to the same thermal cycling plus a diode-capacitor fabrication process. Two types of electrical measurements were used to map diode-capacitor arrays: reverse leakage current (Ir) and transient reverse recovery lifetime (τr. The measured values for each device tested were positionally correlated to the fourier transform infrared spectrometer (FT-IR) [Oi] data. Scatter plots were generated by plotting the electrical measurements as a function of the interpolated concentration data or differential concentration data parameterized at equivalent positions on the wafers. The scatter plots show that the electrical characteristics are strongly dependent on the precipitated oxygen (∆ [Oi]) for larger values of ∆[Oi], but nearly independent for lower ∆[Oi] values.

A dominant electron trap in boron-doped CZ silicon at Ec −0.41 ± 0.02 eV with a electron capture cross section >10−14 cm2 has been observed with DLTS in 2-step annealed (16 hrs at 800°C + 16 hrs at 1050°C) seed end wafers where oxygen precipitation is pronounced. A strong correlation between the generation lifetime τg (as calculated from junction reverse generation currents) and the density of this trap is also observed. Electron microscopy showed the dominant precipitates to be {100} plate type containing only Si and 0, with dislocations and punched-out loops in close proximity. Wafers from the center- or tang-sections of the ingot or those given a 1-step anneal (conditions which result in much less oxygen precipitation) contain a low density (<1.2 × 1012 cm−3) of a hole trap at Ev +0.25 eV not correlated with Tg. These wafers contain precipitates with a different morphology with much rower or minimal dislocation densities. Possible origins of the lifetime controlling electron trap is discussed.

Interaction of oxygen and nitrogen in ion-implanted crystalline silicon layers has been investigated by infrared absorption. Infrared absorption bands which are not produced by either 14N or 160 alone are observed after sequential implantation of 14N and 160 into overlapping profiles followed by annealing at 500°C. The new bands indicate oxygen - nitrogen interactions and are ascribed to Si-N vibrational modes with oxygen in nearby interstitial sites.

Annealing of oxygen-rich Czochralski silicon at temperatures between 700°C and 1000°C introduces a new scattering mechanism which correlates very well with the presence of oxide precipitates. In lightly doped crystals (n,p ∼1015cm−3) this mechanism becomes dominant at T<170°C, and it determines to a large extent the value of the electron (hole) mobilities. The effect vanishes upon annealing at 1300°C; such heat treatment causes the dissolution of oxide precipitates. The new component mobility related to precipitates was found to exhibit a characteristic temperature dependence μ ∝ T0.5. This dependence was explained in terms of scattering of free carriers by spacially slowly varying potential associated with oxide precipitates of varying size.

The photoionization cross section of the platinum-acceptor level in silicon was measured (in relative units) as a function of photon energy. Capacitance transients due to electron emission from this level were studied in a p+n gated photodiode at temperatures of 40, 60, and 80 K. Measurements were made over the wavelength range of 2 to 5 μm with light from a prism monochromator with a constant bandpass of 10 meV. The platinum density in the diode was about 1014 cm−3, providing a ratio of deep to shallow (phosphorus) levels of about 0.1. The data are in good agreement with the Ridley-Amato lattice-coupling model when a Huang-Rhys parameter of S = 0.3 is used, corresponding to a Frank-Condon shift of 15 meV if an average phonon energy of 50 meV is assumed. The electronic energy of the acceptor level was 226 ± 5 meV below the conduction band, independent of temperature and in agreement with previous studies of thermal ionization. The present results provide the first clear experimental evidence of lattice relaxation associated with a deep level in silicon. However, the observed Huang-Rhys parameter is smaller than the theoretical estimates of Lowther (S ≅ 1), suggesting that multiphonon emission may not be the only mechanism for carrier recombination involving this level.

The use of electron paramagnetic resonance to investigate intrinsic defects in the III-V semiconductors is reviewed. Particular attention is given to lattice vacancies, antisites and their complexes in GaP, GaAs, and InP. The role of EPR in arriving at an understanding of these defects is emphasized and the interplay between experiment and theory is discussed.

The role of the luminescence killer center iron in GaP has been studied by the optically detected magnetic resonance technique. Observations of strong Fe°(3d5) and O0 resonance in the (Zn,O) donor-acceptor pair (DAP) and the 0.841 eV electron capture (EC) luminescence bands show that the luminescence killing action is via (Fe,O) DAP recombination.

Two distinct but similar (Fe,O) DAP processes were detected: The (Fe0,O0) process with a slower recombination rate and (Fe0,O0*) process, with a faster effective relaxation rate, involving the oxygen excited state.

Simple atomic models are proposed for three metastable centers in InP, which have been called “M centers”. All undergo “large lattice relaxation” and emit or absorb multiple electrons on changing atomic configuration. The one which has been most thoroughly studied is called M and is introduced by MeV electron irradiation. The second, called M:P, is present in as grown device quality InP and is a problem for MISFET devices. The third, called M:Fe, is present in Fe doped InP that has been processed at high T to produce a p+ region. Our model for M consists of a V−InP+In defect complex in one configuration and 2P+In + V+p in the other configuration. Transformation between these two configurations occurs by nearest neighbor hopping of a P atom between vacant sites. The “negative U” character of the M center is a simple consequence of the P atom hop. Our model for the M:P center is a V+p one configuration and a V−InIn−p in the other, the transformation occurs by In atom hopping. For the M:Fe center we tentatively propose Fe−In + V−In in one configuration and Fe−In V+PP+In for the other, the transformation again occurring by P atom hopping. It is also possible that the Fe occupies a bond centered interstitial position, Fe−2bc, in the first configuration and Fe−InV+P in the second, the transformation occurring by Fe atom motion. A distinction between these two possibilities can be made with a pressure dependence measurement. It is noted that the activation energies for these transformations support the Ballistic Model for vacancy migration.

Electronic deep levels in GaAs have been investigated by correlated luminescence and DLTS measurements on material in which the defects were systematically perturbed by rapid thermal annealing. The samples were grown by MOCVD and encapsulated with silicon nitride. Annealing was performed at temperatures from 800°C to 950°C for 5 seconds. The luminescence spectra were measured from 0.7 to 1.6 eV at temperatures from 4-80 K and have features at 0.97, 1.17, 1.35, 1.40 and 1.5 eV. The band-to-acceptor luminescence shows the formation of Si acceptors, starting at an anneal temperature of about 850°C. The luminescence peak at 1,35 eV, attributed to As vacancy - acceptor complexes, is observed to increase in intensity with higher annealing temperatures. In contrast, the Ga vacancy-donor complex peak at 1,17 eV decreases in intensity. DLTS data show an increase in the density of deep levels and new levels not present in the unannealed material. Capacitance-voltage data find a reduction in carrier concentration, although type conversion is not observed even after annealing to 950°C.

High resolution photoluminescence for the identification of impurities and defects in semiconductors has emerged as a powerful technique. We review the technique with emphasis on GaAs realizing that it is applicable to many different semiconductor materials. The binding energies of thle ground state and of several low-lying excited states of the impurity centers are determined by studying the radiative transitions associated with excitons bound to neutral donors or acceptors. Recent developments in the studies of GaAs-AlGaAs multi-quantum-well (MQW) structures using high resolution photoluminescence are also reported. Variations of the energies of the various transitions in MQW structures as a function of well size are presented. Estimates of the interfacial quality in these heterostructures are made from line shape analysis.

The double acceptor states of chromium, cobalt and nickel were identified by means of the optical and transport measurements under external fields such as hydrostatic pressure, uniaxial stress and magnetic field. The axial fields enable us to observe optical line splittings, which give information about defect symmetries. This technique has been used to identify a double acceptor state of nickel in the GaAs energy gap. High pressure eliminates a degeneracy between defect levels and crystal bands. This leads to activation of new charge states of some impurities. In the case of chromium and cobalt impurities, their double acceptor levels were found to be degenerate with the conduction band of GaAs. A trend existing for the energies of the double acceptor levels permits us to conclude that there will be no other acceptors among the transition metal impurities in GaAs.