The DX-center model is widely used to explain data for the persistent photoconductivity (PPC) effect. An analysis of the DX-center model suggests a new experiment to test its correctness. In this experiment, photons near the threshold energy of the photoionization cross-section for the DX-center induce transitions from the partially occupied conduction band to empty DX-centers. This mechanism, which we call photocapture, competes with the usual photoionization which empties the DX-centers. The photocapture cross-section is estimated and an experimental attempt is made to detect photocapture. The significance of the null result is discussed.

We report the measurement of the thermal activation energy for the DX- center in Se-doped AlxGa1-xAs grown by metal-organic chemical vapor deposition (MOCVD) for different alloy compositions (x=0.19, 0.23, 0.27, 0.31). The peaks obtained from conventional DLTS are often broad or asymmetric with shoulders on one or both sides. These phenomena often arise from two or more traps which are active in the same temperature range.

The capacitive transients are recorded digitally and analyzed directly by applying a nonlinear double exponential fitting routine to the data. This fitting produces two Arrhenius plots and yields the densities of the defect states. From the Arrhenius plots, the capture cross sections at infinite temperature and the thermal activation energies are calculated. These results are then used to simulate the DLTS spectra. Excellent agreement between real and simulated spectra is shown.

The dependence of the DX centers thermal emission rates on the electric field has been measured by Isothermal Transient Spectroscopy (ITS) and differential ITS (DITS) techniques. These techniques allow us to achieve a more accurate analysis of these defects than the DLTS, which ability for the study of DX centers is discussed. This dependence has been found to be mainly due to a Poole-Frenkel effect and is the cause of the non-exponentiality of the thermal emission transients.

The defects present in AlGaAs high electron mobility transistor (HEMT) devices subjected to accelerated lifetime tests have been studied by transient current, and when possible, by transient capacitance techniques. By measuring transient source-drain currents following the application of a voltage pulse to the gate it is possible to perform deep level transient spectros-copy (DLTS) experiments on HEMT devices which are too small for the conventional capacitance DLTS. The capacitance and current spectra for both stressed and unstressed HEMTs contain the AlGaAs DX defect. The current DLTS spectra for stressed devices contain an additional feature which is not found in capacitance DLTS measurements on the stressed HEMTs. This additional current DLTS feature is anomalous in that the transient has a sign which is opposite to that expected for a majority carrier trap. The absence of the new defect in the capacitance DLTS suggests that the defect is located in the channel between the gate and either the source or the drain. The current DLTS line shape of the stressed induced defect depends upon the polarity and size of the source-drain voltage.

Transient conductivity and photoconductivity experiments have been performed on hydrogenated Al0.37Ga0.63As:Si epilayers. The recapture kinetics of free electrons on the remaining deep donors is significantly affected under hydrogena-tion. The energy barrier height for capture on the deep donors drops from 210mev in the reference sample to 40mev in the hydrogenated sample. We also note a shift toward higher energies of the photoionization cross section versus photon energy curve. Different models are proposed for explaining the data.

The metastability of the EL2 and DX defects in GaAs and Ga1-x AlxAs alloys is discussed in the context of recent hydrostatic pressure and photo-EPR results. A unified model is presented based on the metastable trapping of carriers in excited effective-mass (EM) states derived from secondary conduction band (CB) minima. The metastability is attributed to small lattice relaxation effects.

The material covered in the talk given under this title has already been published [1] or has been accepted for publication [2]. For this reason, the material to be presented below will consist only of a summary with references.

Photo-induced changes in the infrared (IR) absorption spectrum of the EL2 defect in undoped and lightly alloyed unannealed GaAs materials grown by the liquid-encapsulated Czochralski technique were observed under a monochromatic light irradiation. These changes were attributed to the change of the charge states of EL2. The spectrum which is believed to be due to the charged EL2 exhibits a complex structure with two peaks at 1.07 and 1.32 eV and a shoulder at 0.94 eV. The EL2+/2+ → EL20/+ and EL20/+ → EL2+/2+ transitions were obtained by illuminating the sample with 0.7 ≤ hv ≤ 0.95 eV and 1.3 ≤ hv ≤1.5 eV, respectively. The transformation EL2+/2+ ↔ EL20/+ can be achieved in less than 10 sec and can be repeatedly switched back and forth between the two states. Fourier-transform photoconductivity (FTPC) measurements were performed on a large number of samples both semi-insulating and n-type. Two broad peaks were observed in the FTPC spectra with thresholds at 0.80 and 0.95 eV. The photoquenching and thermal recovery properties of these two peaks are identical to those of EL2. Both IR and FTPC results suggest strongly that EL2 is a double donor.

We report the results of DLTS experiments under uniaxial stress on the second ionization level of EL2(++/+) in p-type GaAs. We measured the shift in the hole emission rate as a function of stress applied in the [100] and [110] directions. By modeling the valence band with two independently displacing bands and appropriately derived effective masses, we obtain a small absolute hydrostatic pressure derivative for the defect, 39 ±15 meV GPa-1. The shear contribution is negligible. This result is very different than for the first ionization level, EL2(+/o) with a emission energy pressure derivative of 90 ±15 meV GPa-1. The difference can be accounted for by the pressure dependence of the electron capture barrier of EL2(+/o), 49 ±15meV GPa-1. The absolute pressure derivatives of the two levels are then comparable and in good agreement with simple theory for Ga site point defects.

Linear dichroism has been measured in the broad absorption band of the EL2 defect in GaAs under uniaxial stress. In addition, the splittings of the EL2 zero phonon line (ZPL) at 8378 cm-1 under uniaxial stress applied along [100], [111], and [110] directions have been measured. Splitting of the ZPL under [100] stress is over one order of magnitude smaller than under [111] stress, on the other hand, the linear dichroism in the broad absorption band is roughly equal for these two directions of stress. This is an evidence for the quenching of the coupling to tetragonal strains due to interaction with trigonal modes of the lattice (the Ham effect). Therefore, it is established, that the excited T2 state of EL2 is a localized state subject to dynamical Jahn-Teller coupling to trigonal modes of the lattice. The possibility that the excited T2 state has hydrogenic nature associated with the L minima is ruled out by the present results. The observed splittings of the ZPL together with polarization selection rules clearly indicate the tetrahedral Td symmetry of the EL2 defect ruling out any other point group in particular trigonal C3v In view of the presented experimental results, their interpretation, and recent theoretical investigations, the isolated arsenic antisite AsGa most successfully accounts for the properties of the neutral charge state of the EL2 defect.

The infrared absorption technique is used to study the recovery of the EL2 metastable state in semi-insulating GaAs under monochromatic light illumination. The induced optical recovery is monitored after low intensity (≤ 2mW/cm2) irradiation in the energy range 0.7 ≤ hv ≤ 1.5 eV. The data exhibit a complex structure consisting of a broad band around 0.9 eV and a set of multiple sharp peaks between 1.44 and 1.5 eV. This recovery is strongly dependent on the sample, temperature and illumination time. The present results suggest that a) the existing data and theoretical predictions for the isolated AsGa antisite structure are not compatable with the optical recovery data, b) EL2 is affected dramatically by other defects (traps) present in the sample and c) the peaks observed in the optical recovery data are coincident with the arsenic vacancy energy levels and therefore the present results support the proposed complex models involving an arsenic vacancy.

The anti-site defect AsGa, EL-2, is used to understand the nature of arsenic surface species during the Organometallic Vapor Phase Epitaxy(OMVPE) of GaAs. The concentration of EL-2 in unintentionally doped n-GaAs, measured by Deep Level Transient Spectroscopy, is presented as a function of AsH3 partial pressure, TMGa partial pressure and the growth temperature. Based on this data, a model for EL-2 incorporation in OMVPE GaAs is developed in which all surface species As-H, are converted to As2 at around 765 ° C. Under the same set of growth conditions, relative carbon levels measured by 4K Photoluminescence, suggest that the increase in carbon levels with growth temperature is due to the gas-phase loss of H radical from the As-H species.

The mobility lifetime products for photo-electrons in semi-insulating GaAs, which fit successfully the results of photorefractive studies undertaken in the presence of electric fields are three orders of magnitude smaller than those inferred from transport measurements or from the photorefractive effect with no applied electrical field. Consideration of enhanced recombination via EL2 effective-mass states linked to the L-conduction band minimum allows us to fit the dependence of the photorefractive beam coupling gain coefficient on the grating period for both AC fields and moving gratings. A cascade-capture process, which is three orders of magnitude faster than recombination by multiphonon emission from the Γ band to EL2, leads to greatly reduced mobility-lifetime products for field strengths greater than 1 kV/cm. Our results establish the dominant influence of the EL2 defect properties on the recombination processes essential for modelling and optimizing the photorefractive effect in semi-insulating GaAs.

Formation of native defects in GaAs is described in terms of the amphoteric native defect model. It is shown that Fermi energy induced formation of gallium vacancies is responsible for the limitations of maximum free electron concentration in GaAs. The effect of the defects on electron mobility in heavily doped n-GaAs is quantitatively evaluated. Defect scattering explains the abrupt reduction of electron mobility at high doping levels. Also, it is demonstrated that native defects are responsible for the mobility reduction in inverted modulation doped GaAs/AlGaAs heterostructures. The amphoteric defect model also explains a distinct asymmetry in defect formation in n- and p-GaAs. In p-GaAs the Fermi level induced reduction of the defect formation energy is much smaller, and therefore the concentration of the native defects is negligible compared with the hole concentration.

We have investigated the formation and diffusion of delta-doped layers in GaAs employing the acceptors Zn, Cd, and C. Organometallic Vapor Phase Epitaxy was used for the growth of the Zn and Cd δ-doped layers while Metalorganic Molecular Beam Epitaxy was utilized to achieve C δ-doping. The narrowest atomic profiles for Zn had full width at half maxima of 80 Å for peak Zn concentrations of ≤ 3 × 1018 cm-3, as measured by SIMS. An effective diffusion coefficient of ≤ 7 × 1017 cm2/s is estimated for a growth temperature of 625°C, based on multiple Zn δ-doped layers. For carbon, a doping spike of 7 × 1019cm-3 with a full width at half maximum of 50Å, as measured by electrochemical capacitance voltage profiling, was achieved and represents the highest doping level yet reported for planar doping. By contrast, it was difficult to achieve doping levels ≥ 4 × 1017 cm-3 for Cd due to its high vapor pressure.

Ultra-high Be doping of Ga0.47In0.53 As layers grown by gas source molecular beam epitaxy has shown that for each growth temperature, there exists a maximum hole concentration (≥1×1020cm-3). Increasing the Be flux above that which produces the maximum hole concentration results in a degradation of the crystalline quality of the films. The degradation of film quality results from precipitation of a Be-rich phase on the surface during growth and nucleation of dislocations at each precipitate. Below that concentration, some of the Be segregates and floats on the growing surface.

Electronic and photonic device applications of the InGaAs/InP materials system often require the growth of epitaxial material doped to or near the solubility limit of the impurity in the host material. These requirements present an extreme challenge for the crystal grower. To produce devices with abrupt dopant profiles, preserve the junction during subsequent growth, and retain a high degree of crystalline perfection, it is necessary to understand the limits of dopant incorporation and the behavior of the impurity in the material.

In this study, N-type doping above 1019 cm-3 has been achieved in InP and InGaAs using Sn as a dopant P-type Zn doping at these levels has also been achieved in these materials but p type activation above ~3 × 1018 cm-3 in InP has not been seen. All materials were grown by the metalorganic chemical vapor deposition (MOCVD) crystal growth technique. Effective diffusion coefficients have been measured for Zn and Sn in both materials from analysis of secondary ion mass spectra (SIMS) of specially grown and annealed samples.

Using SIMS profiling, temperature-dependent Hall measurements, electronic Raman scattering, and infra-red absorption, a detailed study is presented examining the effects of Si dopant behaviour on intersubband absorption in GaAs/Al .32Ga .68As MQW’s. The samples were grown by molecular beam epitaxy at substrate temperatures ranging from 520 to 680°C with Si doping in the GaAs quantum wells only. SIMS profiling reveals that, with increasing substrate temperature, substantial Si migration into the Al .32Ga.68 As barrier layers occurs during epitaxial growth. Hall measurements indicate that both at elevated growth temperatures and under reduced incident As4 beam fluxes, the electron sheet concentration is reduced in the QW’s. In both cases, loss of free carriers is attributed to enhanced Si compensation. Shifts in the absolute frequency of the infra-red absorption resonance, determined by electronic Raman scattering, as well as variations in the integrated absorption intensity, are both shown to directly reflect the growth-induced variations in the free-carrier concentration in the quantum wells.

Beryllium is an effective p-dopant in GaAs and AlGaAs and plays an important role in device characterizations of hetero bipolar transistors. This work addresses the doping and mobility properties for two series of beryllium-doped samples: GaAs and AlGaAs. Within each series the doping ranged between 3×1015cm-3 to levels of 5×1019cm-3. Mobility and carrier concentrations were obtained through Hall and Polaron measurements. The doping concentration results suggest the onset of carrier compensation at higher doping levels. One possible explanation is that for high doping levels, Be is incorporated as interstitial donors. A thermodynamic model is used to explain the observations.

Carbon doping in InxGa1-xAs is investigated using solid source MBE. InAs mole fractions, x, from 0 to 1 are tested. Under a constant doping level, hole concentration decreases with increasing x. When x is higher than 0.6, the conduction type is n and electron concentration increases with x. The conduction-type inversion and the behavior of carrier mobilities are explained by a self-compensation mechanism.