We have found that co-implantation of phosphorus (P) with Si into SI-GaAs caq improve the quality of ion implanted layer. Hall mobilities of 4600–4700 cm2/Vs and activation efficiencies of 75–85% are obtained. These results are better than those obtained from the samples without P co-implantation. Schottky-barrier technique measuerments show a better mobility profile for the samples with P co-implantation. DLTS measurements show that the number of deep levels and their concentrations decrease. We think that P atoms occupy As vacancies and enhance the activation efficiency and their uniformity. The damage produced under the Si-implanted layer by P co-implantation increases the average Hall mobility.

Ion-beam-assisted etching of GaAs/AlGaAs and InP/GaInAsP, ion-beam disordering of GaAs/AlGaAs multiple-quantum-well structures, and ion implantation in InP are discussed.

We have investigated the thermal stability of high resistivity regions introduced by ion bombardment of GaAs, AlGaAs, InP and InGaAs. For low doses in which the ion species density is below that of the doping density in the target material, we obtain the usual damage-related compensation in which deep levels created by the bombardment trap the charge carriers. By this method one creates material with resistivities around 108 Ω/□ (n- or p-type GaAs and AlGaAs, p-type InP), around 106 Ω/□ (n-type InP) or around 105 Ω/□ (n-type InGaAs or p-type InGaAs), with a return of the initial resistivity after elevated temperature annealing (∼600°C for GaAs and AIGaAs, ∼500°C for InP and InGaAs). The more interesting case is the use of higher dose implants of species which create chemical deep levels. This occurs for O in n-type AlGaAs where O creates a deep acceptor (Ec-0.49 eV), and Fe in n-type InP and InGaAs, where it is also a deep acceptor. When the concentration of these species exceeds the doping density in the material, the bombarded regions retain their high resistivity even after high temperature annealing (> 1000°C for GaAs and AIGaAs, >850°C for InP and InGaAs). The case of O in GaAs appears to represent a third mechanism; it creates thermally stable material only in the case of Be-doped GaAs, suggesting an ion-pairing reaction.

The effects of tilt and rotation angle on implant channeling are explored for Si implants into (100)-oriented GaAs substrates using Monte Carlo simulations, SIMS profiling, capacitance profiling, and Hall measurements. It is shown that the 7° tilt angle often used to prevent axial channeling is not large enough, and 11–13° angles are required to obtain the sharpest and most reproducible implant profiles. Rotation angles have a much smaller effect on the profiles as long as the beam incidence is kept away from the < 110 > direction.

Low energy ion implantation followed by rapid thermal annealing (RTA) was utilized to modify exciton transition energies of MBE- rown GaAs/AlGaAs quantum wells (QW). The samples were irradiated with an 75As ion beam with an energy low enough that the depth of the disordered region was spatially separated from the QWs. After RTA, exciton energies (determined using optical spectroscopy) showed large increases which were dependent on QW widths and the implantation fluence with no significant increases in peak linewidths. These energy shifts were interpreted as resulting from the modification of the shapes of the as-grown QWs from square (abrupt interfaces) to rounded due to enhanced Ga and Al interdiffusion in irradiated areas. These results are similar to our data on the RTA of the same structures capped with SiO2 and are consistent with the model of enhanced intermixing of Al and Ga atoms due to diffusion of vacancies generated near the surface.

Magnesium implanted at high doses (1×1014cm−2 for a 120keV implant energy) into GaAs diffuses out of the crystal upon annealing. The outdiffusion is reduced, but not eliminated, when rapid thermal annealing is employed rather than furnace annealing. We have investigated various mechanisms that may explain this large outdiffusion. The high vapor pressure of the magnesium is not responsible for the outdiffusion since beryllium, which has a low vapor pressure, also outdiffuses. The presence of oxygen in the anneal ambient mostly leads to significant GaAs surface etching. The etching is more pronounced in the case of shallow energy and high dose implants, which suggests some oxygen- magnesium interaction mechanism. A high level of damage at the GaAs surface in the case of shallow magnesium implantation at high dose may also favor rapid outdiffusion of interstitial magnesium.

In this paper we report on the depth distributions of implanted hydrogen in GaP crystals and the subsequent changes produced by post- implantation furnace annealing. A sulfur doped n+ GaP wafer has been implanted with 333 keV protons to a fluence of 5E15/cm+2. A similar wafer was implanted with 350 keV deuterons to the same fluence. Portions of each wafer have been furnace annealed at temperatures up to 500°C. The implanted hydrogen and the dopant S atoms were then depth profiled using secondary ion mass spectrometry (SIMS). The measurements show that the redistribution of hydrogen begins with annealing at about 300°C and proceeds both towards the surface and deeper into the substrate. The overall behavior is similar to that found previously for hydrogen in GaAs. However, in GaP crystals this redistribution begins at a higher temperature and proceeds more slowly in the implanted region. Based on the SIMS profiles, diffusion coefficients for hydrogen migrating into substrate are obtained.

Layer intermixing in MeV Si-implanted quantum well superlattices (SLs) has been studied by transmission electron microscopy, secondary ion mass spectrometry and Rutherford backscattering spectroscopy. Molecular beam epitaxially grown GaAs(200Å) - Al0. 5Ga0.5As(200Å) SLs were implanted with 1 MeV Si+ at doses between 3 × 1014 and 1 × 1016/cm2. The implanted SLs were either furnace annealed at 850°C for 3 hr or rapid thermally annealed at 1050°C for 10 sec, both under GaAs proximity capping conditions. Totally mixed regions were observed only for the SLs implanted with 1 × 1016 Si/cm2 and then furnace annealed at 850°C for 3 hr. For the same dose, the RTA annealed SLs only showed slight layer intermixing. At lower doses, no appreciable intermixing was detected in either furnace or RTA annealed samples. By contrast, under either annealing condition extensive intermixing has been demonstrated for lower energy (220 keV) implantation, but at doses almost two orders of magnitude lowerl XTEM showed that in all the annealed samples, a defect-free zone existed in the near-surface region, followed by a band of secondary defects, with the maximum density located at about 1 μm below the surface. In the disordered samples, the position of the intermixed layers correlated with the defect band maximum. Under both annealing conditions, Si concentration profiles only showed slight broadening, and they correlated with the distribution of secondary defects as well as with the depth of the intermixed layers. The effects of dynamic annealing and surface on the implantation energy dependence, i.e., MeV vs. keV, of layer intermixing are discussed.

Raman scattering, double-crystal x-ray diffraction, and electron beam electroreflectance have been used to assess the damage produced in undoped (100)-GaAs by boron ion implants and the influence of post-implant anneals. Both conventional furnace and rapid thermal annealing treatments were found to remove much of the lattice strain created by the implants. However, considerable disorder also remains after these anneals.

Recently, GaAs-AlGaAs interfaces have been interdiffused following furnace and rapid thermal annealing (RTA) of samples containing impurities. Two models are proposed to explain the events occurring during the annealing processes. Te was chosen as the impurity for furnace annealing, whereas for the case of RTA, the impurities were introduced via ion implantation. The Te-induced interdiffusion process is modeled using an ensemble Monte Carlo simulation. The results of the simulation are in good agreement with the Te intermixing data. In the case of ion implantation followed by RTA, the intermixing model is based on the solution of coupled diffusion equations involving the excess vacancies and the Al distribution following ion implantation. Both initial distributions are obtained from the solution of a three-dimensional Monte Carlo simulation of ion implantation into a heterostructure sample. The excess interstitials calculated in the simulation are coupled to the vacancy diffusion via a phenomenological decay constant. Having established the value of the decay constant from one set of experimental results, the model is able to predict accurately other independent experimental results.

Mg ion-implantation was carried out into very pure GaAs made by molecular beam epitaxy, in which the dose of Mg atoms, [Mg] was varied from l×1015 to 1×1020 cm−3 . As a method of characterization, low temperature photoluminescence (PL) studies were carried out as a function of [Mg]. A new emission, denoted by [g-g] was observed between bound exciton emissions and the emission due to 1 and-to-acceptor pairs. It was found that [g-g] appears around [Mg]= 2×1016 cm−3 and becomes a dominant emission at [Mg] over l×1017 cm−3. A significant shift of [g-g] towards the lower energy side with increasing [Mg] was reproducibly obtained and its PL properties were carefully investigated in terms of annealing temperature, TA and excitation density. Results revealed that the implanted Mg atoms become optically active at TA over 600 [g-g] was suggested to be acceptor-acceptor pair, which is consistent with the theoretical explanation that its radiative mechanisms is ascribed to the pair between excited-state acceptors. The systematic energy shift inherent to [g-g] was attributed to the growing ionization energy of the pair with increasing pair distance.

We have studied the disordering phenomenon in GaAs-AlGaAs superlattices induced by Si implantation followed by rapid thermal annealing. Layer intermixing has been detected in superlattices implanted with 220 keV Si+ at doses ≥ 1×l015 cm−2 and annealed at 1050°C for 10 s. The amount of intermixing saturates with time after 10 s annealing, whence the lattice damage caused by the implantation is predominantly annealed out. The transient disordering is attributed to defect-induced layer intermixing occurring during the annealing of the implantation damage. Concurrent with the disordering, Si diffusion is observed to be minimal, which indicates that layer intermixing due to Si diffusion and other impurity effects lags behind that due to defects in the time scales of the present experiment.

Rapid thermal annealing (RTA) (800–1000°C, 1–60 s) was performed on capless and silicon nitride (SixNy) capped GaAs samples implanted with Si (30 keV, 4.5×1013cm−2 ) and SiF+ (50 keV, 4.5×1013cm−2 ). The maximum activation for the Si+ implants saturated at 25% (capless) and 42% (capped) and that for the SiF+ implants saturated at 20% (capless) and 35% (capped). The activation degraded at temperatures > 825°C due to surface decomposition in the capless annealing case and at > 925°C due to the cap failure in the capped annealing case. For all the RTA conditions studied here, higher activation always occurred with the Si+ rather SiF+ implants. An activation energy of 0.48 eV for the annealing process was determined from the electrical data. It was also observed that the heating and/or cooling rates can significantly influence the electrical activation.

Dry etching of indium-based III–V materials in chlorine-based gases is difficult because of the low vapor pressure of the indium chloride by-product. Recently, reactive ion etching of InP(1,2), and GaAs(3,4) using methane, hydrogen and argon mixture in which volatile organometallic group III compounds are formed has been employed. In this paper, we report the reactive ion etching of indium-based materials using CH4:H2:Ar and SiCl4:Ar mixtures.

Electrical characterizations (I–V,C–V) of TiPtAu/n-type GaAs Schottky diodes were performed in order to investigate the electrical damage introduced by CH4JH2 RIE. The effective barrier height measured according to the thermionic conduction model decreases from 0.79 eV to 0.53 eV after RIE carried out at 1.3 W/cm2.

Recovery of the control I–V characteristics requires a wet chemical removal of 50 nm of the dry etched surface.

Doping profiles, derived from conventional C–V analysis at 80 K and 300 K, show a decrease in free carrier concentration within 500 nm from the surface. Hall measurementsbetween 4 K and 300 K show an increase in carrier mobility after RIE. These effects are essentially attributable to chemical neutralization of the silicon dopants by hydrogen. Thermal annealing at 360°C leads to an almost complete recovery of the electrical activity and of the control Schottky barrier height.

Cleaning treatments of GaAs(001) by UV decomposition of NH3 has been studied by means of X-ray photoemission spectroscopy (XPS) and X-ray photoelectron diffraction (XPD). This photolysis treatment is shown to decompose the surperficial oxides and to remove the carbon of contamination. The surface cleaning is followed by the formation of gallium nitride overlayers. When used prior to UVCVD silicon nitride deposition, this treatment is shown to improve the electrical characteristics of the Si3N4/GaAs MIS structure.

We have demonstrated a novel downstream plasma activated vapor phase etching of III-V compound semiconductors using ethylene di bromide as the etchant. Highly reproducible surfaces and etch rates were obtained by monitoring the reaction process with a mass spectrometer. Etch rates on InP of up to 4500Å/min, at temperatures as low as 160°C, at an rf power input of 25 watts were achieved with no evident damage to the surface as indicated by PL measurements. Apparent activation energy measurements seem to suggest that the etch products might be organometallic in nature, in contrast to inorganic etch products seen in most etching studies. Etching is partially anisotropic and resulted in smoother surfaces when H2 was used as the carrier gas for the etchant instead of N2.

During heat treatment of GaAs there is a tendency for arsenic to evaporate. This process can have deleterious consequences and must be controlled during the annealing of GaAs by using encapsulants. In this work we use a Cr collector to catch the evaporated atoms. This collector is placed on top of GaAs or GaAs/cap during annealing. We survey the effectiveness of W, Hf and HfN as capping materials for thermal annealings of 10 min. for 450–700 C and compare them to Si3N4. The backscattering spectrometry results show that Si3N4 and W were excellent caps up to 700 C for 10 min. No evaporation was defected through HfN either, but hillock formation was detected on the surface. Hf failed because of its chemical reaction with GaAs at 650 C, 10 min.

The effect of Ar sputtering, in combination with different dielectric cappings, on the disordering of AlGaAs/GaAs multiple quantum wells was studied. It was found that the combination of Ar sputter etch and SiO2 capping provided the strongest enhancement in disordering. This effect generates enough change in refractive index to produce an index guided laser.

A novel GaAs metal insulator semiconductor (MIS) structure in which the surface of the GaAs is unpinned and inverted is reported. The structure utilizes a thinsingle crystalline Si interlayer between the GaAs surface and a SiO2 insulator. The Si and SiO2 are deposited using low temperature remote plasma chemical vapor deposition techniques. Characteristics of the Si layer have been examined using reflecting high energy electron diffraction, Xray photoelectron spectroscopy, and He ion scattering spectroscopy. The complete MIS structure has been characterized using capacitance-voltage data taken at frequencies ranging from near D.C. (quasistatic) to 100 MHz. The structure exhibits deep depletion and inversion behavior. Calculation of the surface potential variation using quasistatic data indicates that the surface potential is varied by 0.83 V with an applied electric field from −1.6 to 3.2 MV- cm−1.