Ga Interstitials Research Articles

Optical detection of electron nuclear double resonance on a residual donor in wurtzite GaN.

Optically detected electron nuclear double resonance (ODENDOR) was measured in the 2.2-eV yellow luminescence band associated with a residual donor in $n$-type unintentionally doped GaN. The ODENDOR lines are due to gallium and show a quadrupole splitting that can be described with an axial tensor. The quadrupole parameter was estimated to be $\frac{q(^{69}\mathrm{Ga})}{h}=\frac{1}{2}{Q}_{\mathrm{zz}}=0.22$ MHz. A hyperfine interaction for $^{69}\mathrm{Ga}$ of approximately 0.3 MHz for the isotropic part and of approximately 0.15 MHz for the anisotropic part was estimated from the width of the ODENDOR lines. It is tentatively suggested that Ga interstitials are residual donors.

Enhanced Zn diffusion in GaAs pnpn structures: Growth versus annealing

Zn diffusivity, DZn in heavily doped pnpn GaAs structures has been measured after growth and annealing. During growth at 650 °C, DZn∼10−12 cm2/s in the buried p-type layer is found to be ≳104×DZn in the top p-type layer. During annealing at 800 °C, DZn≊5×10−14 cm2/s in the buried layer remains orders of magnitude larger than DZn in the top layer. The measurements provide clear experimental evidence that (1) a large flux of Ga interstitials, IGa, is injected from the surface during the growth of n-type layers, (2) the IGa are trapped in the buried p-type layer by the electric field of the pn junctions, and are thus positively charged, (3) the resulting large concentration of IGa in the buried layer accounts for the enhanced DZn via a kick-out mechanism, and (4) the mobile Zn interstitial is positively charged.

Modeling the diffusion of grown-in Be in molecular beam epitaxy GaAs

The diffusion of Be in GaAs is studied in samples which are molecular beam epitaxy GaAs with grown-in Be. The Be diffusion profiles of samples annealed under various conditions are obtained using secondary ion mass spectrometry. SUPREM-IV.GS, a simulator for GaAs and Si processing technology, is used to compare the experimental results with our models and to extract parameters. The Be diffusion profiles show a kink feature and a time-dependent Be diffusivity which are successfully simulated. The intrinsic Be diffusivity, the Ga interstitial diffusivity, and the equilibrium concentration of Ga interstitials, all as a function of temperature, are obtained from this study: D+1Be=0.17 exp(−3.39 eV/kBT) cm2 s−1, DI=6.4×10−5 exp(−1.28 eV/kBT) cm2 s−1, and CI*int=4.7×1028 exp(−3.25 eV/kBT) cm−3. The role of nonequilibrium Ga point defects in the anomalous Be diffusion behavior is addressed.

Are the Models of the Triply Charged Gallium Vacancy and Doubly Charged Gallium Interstitial Alive or Dead?

AbstractWe have found that the measured diffusivity can be quite different than predicted by recent models of multiply-charged defects. Key problems dealing with the measurement and the interpretation of diffusion mechanisms are discussed. Using a few common variations in GaAs epilayer growth, we have obtained diffusivities which range over several orders of magnitude at the same temperature. Making use of the relatively weak In-As bond, we have used In as a marker to measure group III interdiffusion, DIII. DIII is consistent with the results of others using Al as a marker in n-type GaAs, but orders of magnitude smaller than predicted by the triply charged Ga vacancy, V3−Ga, model of Tan and Gosele. Although diffusion can be attributed to a negatively charged vacancy in n-type GaAs, In is found to often move by a kick-out mechanism in p-type GaAs. It appears likely that many early experiments with n- and p-type GaAs-AlAs interdiffusion were affected by large concentrations of Ga interstitials, IGa, caused by Fermi energy pinning at the growing surface. We present the first direct experimental evidence for the existence of a positively charged Ga interstitial. Our results, combined with those of others, suggest that VGa has a single negative charge associated with it. Because epilayer growth conditions appear to cause the point defect concentrations to deviate substantially from equilibrium, we conclude that the exact charge states of VGa and IGa still remain to be determined.

Effect of Fermi Level Pinning at the Surface During OMVPE Growth

AbstractDuring organometallic vapor phase epitaxy of GaAs, the Fermi energy is found to be pinned ≈200 meV below the intrinsic Fermi energy on a (lOO)-oriented surface. This was determined by making the first comparison of Zn concentration, Nzn. obtained from growth with that obtained at equilibrium, all under the same nominal ambient conditions. Donor solubility has been found to be virtually unchanged during growth of n- or p-type GaAs, and this is strong evidence that the Fermi level remains pinned at essentially the same energy during growth of n- or p-type GaAs. Consistent with Fermi level pinning, we find (1) the NZn is supersaturated in the epilayer for the ambient growth conditions used, (2) a large excess concentration of positively charged Ga interstitials can be grown into n-type GaAs and which diffuses into nearby p-type layers, causing (3) the Zn diffusivity, DZn ∼10−13 cm2/s, out of buried npn layers to be essentially independent of the ambient conditions. Annealing of a heavily Zn-doped layer at the surface can lead to 10−16<DZn<10−13 cm2/s at T=800°C simply by varying the Zn partial pressure over its commonly used range. Use of In-doped spikes in n- and p-type GaAs suggest that interstitial point defects have a strong effect on the group III diffusion.

Eliminating dopant diffusion after ion implantation by surface etching

Implantation damage is believed to play a significant role in dopant diffusion. Described here is an attempt to modify the point defect damage profile of a 40 keV, 29Si implant in GaAs by chemically etching away the top 100 Å of the sample after implantation. In these samples, no Si diffusion was observed after annealing, while significant Si redistribution did occur in a similar sample which received no post-implant etch. trim simulations predict an excess Ga vacancy surface layer and excess Ga interstitials deeper in the sample. It is thought that by removing the vacancy-rich surface layer, the etch alters the implant damage profile and thus the diffusion behavior of the Si. Surface effects from etching as related to the Si diffusion are shown to be consistent with a vacancy-assisted diffusion mechanism. Evidence that this model may be applicable to B implants in Si is also shown.

Time Dependent Diffusion of P-Type Dopants in Gallium Arsenide

ABSTRACTThe diffusion of implanted p-type dopants in GaAs often exhibits transient effects during annealing. This diffusion can greatly affect device performance and must be understood in order to create accurate, physically based process modeling simulation tools for compound semiconductors. The transient diffusion is believed to be related to defects created during implantation, but a direct correlation between implant damage and dopant redistribution has been difficult to establish. Results are presented in this paper of experiments designed to gain insights into the fundamental mechanisms underlying the relationship between defects and diffusion. Many anomalies in previous data are the result of implanting dopants under poorly controlled conditions, particularly implant dose rate and temperature. Special care was taken in this work to ensure that the implants were performed under well controlled conditions. Al was co-implanted with Mg and Be to study the effect of implant damage without affecting the position of the Fermi level. A correlation was found between the initial transient diffusion observed in low dose Be implants and the presence of dislocation loops. This may present a viable method for suppressing diffusion of implanted Be. SIMS profiles of Mg and Al co-implants were anomalous and difficult to interpret. Cap failure, dopant segregation to the surface and/or defects may be responsible for the discrepancies in the data. P-type doping appears to stabilize dislocation loops, probably by increasing the concentration of positively charged Ga interstitials. RBS measurements of implanted Mg indicates a link between damage production, implant dose rate, and diffusion. This result agrees with previous work on damage production in Si implanted GaAs.

Theory of Zn-enhanced disordering in GaAs/AlAs superlattices.

The microscopic mechanisms for Zn diffusion in GaAs and Zn-induced interdiffusion in GaAs/AlAs superlattices are investigated by ab initio molecular dynamics. Among the various proposed mechanisms for Zn diffusion, kick-out by Ga interstitials has the lowest activation energy. Zn in-diffusion generates nonequilibrium group-III interstitials, which are bound to Zn by Coulomb forces. The interstitials follow the Zn diffusion front and disorder the superlattice. The calculated activation energies for these processes are in good agreement with the experimental data.

Comparison of Current-Induced Migration of Be and C in GaAs/AlGaAs HBTs

ABSTRACTWe contrast the stability under bias-aging conditions of GaAs/AlGaAs HBTs utilizing highly Be- or C-doped base layers. Devices with Be doping display a rapid degradation of dc current gain and junction ideality factor. At 200°C, a 2 × 10 μm2 Be-doped device (4 × 1019cm−3 base doping) operated at a current density of 2.5 × 104 A. cm−2 shows a decrease in gain from 16 to 1.5 within 2h. Under the same conditions a C-doped device with even higher base-doping (7 × 1019 cm−3) is stable over periods of 36h, the longest time we tested our structures. The degradation of Be-doped devices is consistent with the mechanism of recombination-enhanced diffusion of interstitials into the adjoining layers. Similar results are obtained with Zn-doped devices. Since C occupies the As sub-lattice rather than the Ga sublattice as with Be and Zn, it is not susceptible to reaction with Ga interstitials injected during growth or bias-aging.

Ga self-diffusion in GaAs

Evidence is summarized which indicates that the self-diffusion of Ga in GaAs may proceed by either the vacancy or interstitialcy mechanisms. It is possible to switch between these mechanisms with a variation of either the Fermi level or As pressure. Self-diffusion appears to be associated with Ga vacancies with charge of -3 in n-type GaAs, and Ga interstitials with charge of +2 in p-type GaAs. For intrinsic GaAs, the commonly observed V-shape in the Ga diffusivity vs As pressure plot is found to be consistent with these two diffusion mechanisms. Simple calculations can be used to predict the near-equilibrium Ga diffusivity from a single intrinsic Ga diffusion coefficient, when defined at appropriate standard conditions. A new model is presented to link the rollover in Ga diffusivity and electron concentration at high donor concentrations.

Be diffusion mechanism in GaAs investigated by slow positron beam

The slow positron beam technique was applied on undoped and Be-doped GaAs to study the effects of Be impurities on both creation and migration of Ga vacancies, VGa, during annealing. It is observed that a monovacancy of VGa is created in Be-doped GaAs to result in enhanced Coulombic interaction between As vacancy, VAs, and Be acceptor, BeGa. In undoped GaAs, the formation of divacancies, VGa-VAs, is dominant. The migration depth of vacancies is shorter in Be-doped GaAs than in undoped GaAs. This suggests the existence of Ga interstitials, IGa, in the Be diffused layer which interact with VGa introduced from the surface. Based on these observations, we suggest the kick-out mechanism for Be diffusion in GaAs.

Effects of impurities Si and Be on the creation of Ga vacancies and Ga interstitials in GaAs grown by molecular beam epitaxy

Two type of top layers, namely, undoped and Si-doped GaAs layers, were grown, respectively, on a Be-doped epilayer by molecular beam epitaxy, and the effects of impurities of Be and Si on the creation of Ga interstitials IGa and Ga vacancies VGa during the annealing of the specimens were investigated by a slow positron beam. The concentration of VGa created from the surface of the undoped GaAs during the annealing decreases drastically when the annealed GaAs was kept at room temperature for one month. This implies that the Be atoms diffusing to the undoped GaAs during the growth and/or the annealing cause the creation of IGa in the undoped layer and it makes the recombination with VGa. On the other hand, no decrease in the concentration of VGa was observed in the Si-doped GaAs. These support the creation of IGa and of VGa, respectively, in the Be-diffused GaAs and in Si-diffused GaAs.

Diffusion-induced defects in GaAs by zinc and the effects of post-diffusion anneal

The crystallographic defects generated during the diffusion of zinc into GaAs have been studied using transmission electron microscopy and chemical etching techniques. By observing the effects of arsenic pressure on the defect density and the zinc penetration, the annihilation of the diffusion-induced defects in relation to the surface stoichiometry was investigated. It was found that the dissolution of the defects within the crystal is mainly caused by the out-diffusion of the diffusion-induced Ga interstitials to the surface. This is also supported by results obtained from post-diffusion annealing of the samples under different arsenic pressures. In addition, the possible inclusion of arsenic atoms in the interstitial dislocation loops found in GaAs is also discussed.

Application of the charged point-defect model to diffusion and interdiffusion in GaAs

Diffusion and interdiffusion in GaAs and GaAs/AlGaAs superlattices are shown to be consistent with the charged point-defect model. The charged Ga vacancies V3−Ga and interstitials I2+Ga appear to control group II, III, and probably V element diffusion. After adjustment for carrier concentration and As pressure, these elements are found to have a nearly identical intrinsic diffusivity and activation energy over a wide range of temperature. A natural consequence of Ga diffusion via negative or positive point defects is that enhanced group-III interdiffusion is expected with either n- or p-type doping. Anomalous enhancements in group-II dopant diffusivity have been related to the supersaturation of Ga interstitials.

Modelling of the generation of Ga interstitials during the diffusion of Zn into GaAs

AbstractThe diffusion of Zn into a GaAs crystal is modelled using numerical techniques similar to those of Zahari and Tuck.1 This technique does not directly solve any differential equations, the physical processes are directly modelled. The diffusion of Zn is assumed to be by the kick‐out mechanism. In this mechanism the Zn atoms diffuse into the crystal interstitially with a constant diffusion coefficient. The interstitial Zn atoms transfer to the substitutional lattice site by kicking out a Ga atom. The incorporation of the Zn atoms on to the lattice sites is assumed to take place at a rate much greater than the diffusion of the interstitial Zn. The diffusion of the generated Ga interstitial is also contained in the model. The effect on the Zn concentration profile and Ga interstitial concentration of varying both the Ga interstitial diffusivity and equilibrium concentration has been examined. Finally, the implications of these results on Zn induced disordering of GaAs/AlAs superlattices is discussed.

Diffusion of Ion Implanted Mg and Be in GaAs

AbstractAnnealed Mg implants into GaAs show three diffusion regions: 1) rapid uphill diffusion in the peak of the implant; 2) rapid concentration-independent diffusion in the tail; and 3) slow concentration-dependent diffusion in between. Implanted Be, in contrast, exhibits only concentration-dependent diffusion. Constant Fermi level experiments show that this diffusion is actually hole-dependent. Uphill diffusion can be induced in Be implants by co-implanting with a heavier element such as Ar. Paradoxically, this retards the concentration-dependent diffusion. This behavior can be explained with the Substitutional-Interstitial-Diffusion (SID) mechanism and an understanding of the defect chemistry after implantation. In the region of uphill diffusion, the dopants are seen to getter from areas of excess Ga interstitials toward areas of excess Ga vacancies. The magnitude of the Ga interstitial gradient with respect to the dopant concentration is shown to be critical for the uphill diffusion. The reduction in concentration-dependent diffusion with co-implants is thought to be caused by implant damage allowing dopant atoms to shift from interstitial to substitutional sites.

Native defects in gallium arsenide

We describe information which has been obtained on point defects detected in various types of GaAs materials using electron paramagnetic resonance as well as electrical and optical techniques. From a comparison of their characteristics and those of simple intrinsic defects (As and Ga interstitials, vacancies and antisites) it is concluded that native defects are not simple intrinsic defects, with the exception of the antisites, but complexes formed by the interaction of such defects between themselves or with impurities. Particular emphasis is given to the As antisite complexed with an As interstitial, the so-called EL2 defect which plays a major role in the electrical properties of bulk materials. Differential thermal analysis, positron annihilation, and x-ray diffraction demonstrate that bulk materials contain a large concentration of vacancy-related defects and As precipitates located along dislocations which play the role of gettering centers. Presumably, bulk materials also contain other As clusters of various sizes although only the smallest ones (EL2) have been detected. All these As clusters are sources of As interstitials which play an important role in thermal treatments. As to semi-insulating materials, their electrical properties result mainly from the compensation between the double donor, called EL2, associated with the As antisite and the double acceptor ascribed to the Ga antisite.

Optically detected magnetic resonance of native defects in AlxGa

The structure and recombination properties of native defects in ${\mathrm{Al}}_{\mathrm{x}}$${\mathrm{Ga}}_{1\mathrm{\ensuremath{-}}\mathrm{x}}$As have been studied with use of optically detected magnetic resonance (ODMR). Three types of spectra were observed. Extended results are presented for the first type, which has been attributed to Ga interstitials. Spectra taken at different microwave frequencies suggest that alloy disorder affects the line shape. The dependencies on a variety of parameters in molecular-beam and organometallic-vapor-phase epitaxy are explored. Spectral dependence of the ODMR reveals that the +/2+ level of the interstitial is 0.56\ifmmode\pm\else\textpm\fi{}0.1 eV above the acceptor level. A brief comparison with other interstitials is presented. The second type of spectrum is sharp with a donorlike g value (1.947\ifmmode\pm\else\textpm\fi{}0.003). This ODMR line is attributed to Si donors by comparison to the electron-paramagnetic-resonance work of Bottcher et al. [Phys. Status Solidi B 58, K23 (1973)]. The third type of spectrum has an acceptorlike g value (2.183\ifmmode\pm\else\textpm\fi{}0.009 for B parallel to [110]). It may be due to a transition-metal impurity or an Al or Ga antisite.

Identification of the Ga interstitial in AlxGa1-xAs by optically detected magnetic resonance.

A new optically detected magnetic resonance spectrum in ${\mathrm{Al}}_{x}{\mathrm{Ga}}_{1\ensuremath{-}x}\mathrm{As}$ is reported and assigned to native Ga interstitials. Luminescence-quenching signals were observed over the energy region from 0.75 to 1.1 eV. The optically detected magnetic resonance is nearly isotropic with spin-Hamiltonian parameters $g=2.025\ifmmode\pm\else\textpm\fi{}0.006$, central hyperfine splitting $A(^{69}\mathrm{Ga})=0.050\ifmmode\pm\else\textpm\fi{}0.001$ ${\mathrm{cm}}^{\ensuremath{-}1}$, and $A(^{71}\mathrm{Ga})=0.064\ifmmode\pm\else\textpm\fi{}0.001$ ${\mathrm{cm}}^{\ensuremath{-}1}$ for H near [001]. The strong hyperfine coupling denotes an electronic state of ${A}_{1}$ symmetry, which current theories predict for the Ga interstitial but not the Ga antisite. The slight anisotropy probably indicates that the ${\mathrm{Ga}}_{i}$ is paired with a second, unknown defect.

A numerical study of manganese redistribution in GaAs employing an interstitial-substitutional model

A finite element simulation of the anomalous out-diffusion of Mn in GaAs employing an interstitial-substitutional model combined with defect-chemical kinetics has been performed. The major species considered were Ga vacancies (VGa ), Mn substitutionals (MnGa), and interstitials (Mni) augmented by shallow donors (DAs ) and Ga vacancy-donor pairs (VGa-DAs). Mathematically, the model can be represented by two ordinary differential equations which are the kinetic relations describing the formation of MnGa and VGa-DAs coupled with three partial differential equations, including sink and source terms, for the diffusion of VGa, Mni, and DAs . The analysis required advanced numerical techniques using dynamic spatial and time meshes. We have achieved a very good theoretical description of Mn out-diffusion data for both 90 min and 24 h annealing cycles up to a depth of 0.5 μm from the surface [P.B. Klein, P. R. Nordquist, and P. G. Siebenmann, J. Appl. Phys. 51, 4861 (1980)]. Moreover, eventual depletion of the Mn contamination has been demonstrated. However, an inflection beyond 0.5 μm observed in the experimental results cannot be explained by the present treatment.