Recent results from our laboratory on the characterization of optical waveguides are reviewed. In particular, two means of experimental characterization that provide spatially local information are presented. Raman microprobe spectroscopy is used to explore the role of stress in a GaAlAs channel waveguide and to characterize the nature of impurity induced compositional mixing in a multiple quantum well structure suitable for optical waveguiding. Finally, photon tunneling microscopy is shown to probe the optical waveguide evanescent field and thus the local surface and index variations in an optical waveguide.

The neutral impurities boron and fluorine have been studied as species for impurity induced disordering. In the GaAs/AlGaAs system fluorine disordered multiple quantum well waveguide structures exhibited blue shifts of up to 100 meV in the absorption edge (representing complete disordering) accompanied by substantial changes, > 1%, in the refractive index. The absorption coefficient in partially disordered structures at near band-edge wavelengths was as low as 4.7 dB cm−1. Integrated extended cavity lasers have been fabricated with low losses (19 ± 8.4 dB cm−1) in the passive waveguide. Disordering of GalnAs/AlGalnAs and GalnAs/GalnAsP quantum well structures lattice matched to InP has also been investigated. The temperature stability of as-grown phosphorus-quaternary material is poor, with blue shifts of the exciton peak occuring at temperatures greater than 500°C, but the aluminium-quaternary is stable to at least 650°C. Large blue shifts (up to 90 meV for phosphorus quaternary and 45 meV for aluminium quaternary samples) were observed in the fluorine-implanted samples. The estimated loss in fluorine-disordered phosphorus quaternary samples is typically around 8 dB cm“−1.

Raman scattering has been used to characterize lattice damage and impurity-induced compositional disordering in AlGaAs superlattice suitable for optical waveguiding. The degree of damage induced by both conventional ion beam (CIB) implantation and focused ion beam (FIB) implantation is studied using a spatial correlation model to interpret the Raman spectra. FIB implantation is found to induce slightly more damage than CIB implantation for doses of 8×1013 cm−2 and 4×1014 cm−2. and significantly more damage with 2×1015 cm2 compared to CIB implantations of the same dose. Suitable FIB implantation and rapid thermal annealing (RTA) conditions which provide compositional mixing were determined using Raman and photoluminescence spectroscopy. Using these conditions, an optical channel waveguide in AlGaAs superlattice formed by FIB-induced compositional intermixing is demonstrated.

Low temperature photoluminescence spectra have been used to characterize conventional ion beam (CIB) and focused ion beam (FIB) implanted superlattices. The excitation dependence of the single scan FIB is found to be significantly different from CIB and multiple scan FIB implantations which are similar. The peak position of the donor-acceptor transition is observed to change to higher energies significantly slower with excitation intensity for the single scan FIB case when compared to the multiple scan FIB and CIB cases. Simple models to describe these effects are briefly discussed.

Results are presented on FIB mixing of an Al0.3Ga0.7As/GaAs superlattice with equal 3.5 nm barrier and well widths. Si++ was accelerated to 100 kV and implanted parallel to sample normal at doses ranging from 1013 to 1015/cm2. The level of inter-layer disordering was measured primarily by Auger depth profiling. The mixing effect of RTA-only increased roughly linearly with annealing time at both 950 and 1000°C, with the latter exhibiting a sharper slope. At either temperature, 90 s was sufficient to produce complete mixing within the accuracy of the measurement. An anneal of 10 s at 950°C, which was utilized in subsequent post-implantation annealing, resulted in ∼ 25% thermally-induced mixing, with a corresponding PL “blue shift” of 6 meV. The level of mixing by an ion dose of 1 × 1014/cm2 yielded a mixing parameter of 0.87, where 1.0 represents complete mixing. This is the lowest ion beam dose necessary for nearly complete mixing reported to date for either FIB or BB implantation. Doubling the dose to 2 × 1014/cm2 results in an increase in mixing by only 0.05 to 0.92. Larger doses produced a diminishing increase in mixing parameter.

Undoped AlAs/AlxGa1−xAs superlattice structures were grown by molecular beam epitaxy and annealed under Si3N4, SiO2 or WNX encapsulant films, both with and without the presence of implanted Sn. Enhancement of the Al-Ga interdiffusion coefficient occurred under the Si3N4 film due to in-diffusion of Si. Enhancement was even greater during diffusion of the Sn implant under both Si3N4 and SiO2. Underneath the WNX film, however, interdiffusion was suppressed even in the presence of Sn. We simulated these results with SUPREM IV and show that both the Fermi level effect and vacancy injection from the cap are necessary to cause significant enhancement of Al-Ga superlattice disordering.

Recent experimental results indicate that diffusion of implanted p-type dopants in GaAs is time dependent under certain conditions. For Mg implanted at a dose of 1 × 1014 cm−2, the diffusion is constant for approximately an hour, then decreases by an order of magnitude or more. Be implanted at 1 × 1013 and 1 ×1014 cm−2 exhibits similar behavior, but with a shorter time before the diffusivity decreases. The diffusivity in 1 × 1013 Mg cm−2 implants, in contrast, remains constant for up to 16 hours. TEM micrographs of Be and Mg implants reveal dislocation loops in the higher dose samples, but not in the lower dose ones. During annealing, the loops grow and decrease in density, eventually disappearing completely from the crystal. This annealing of the loops appears to correlate to the time dependence of the diffusion. This behavior can be explained in terms of the substitutional-interstitial diffusion (SID) mechanism and point defect equilibria.

Al-Ga interdiffusion of undoped AlGaAs/GaAs multiple quantum wells was studied by varying the crystal defect concentrations and is shown to be mediated predominantly by column III vacancies from the Ga-rich to the As-rich side of the solidus. An increase of two orders of magnitude has been observed in the interdiffusion coefficient when going from the inclusion of excess Ga to excess As in the evacuated ampoule. The in-diffusion of column HI vacancies from the surface in an As-rich ambient, as well as the out-diffusion of column III vacancies from the n-type substrate in an As-poor ambient, were observed using secondary ion mass spectroscopy. The role of SiO2 and Si3N4 encapsulations in controlling the interdiffusion was also determined. We have also observed that the photoluminescence intensity in the Ga-rich crystal is several orders of magnitude stronger than that of the As-rich crystal. Finally, cathodoluminescence spectroscopy reveals a deep level which has been associated with a Ga vacancy complex in the samples annealed with excess As that correlates with the extent of Al-Ga intermixing.

The diffusion rate of Zn in AlxGa1−xAs alloys has been determined, for values of x up to 1.0, as a function of temperature. At 625°C the diffusion coefficient shows a strong dependence on x, reaching a maximum at about 65% Al. A similar rapid increase in diffusion coefficient with x is seen at 700°C. The zinc concentration profiles are similar for all values of x, showing a very abrupt diffusion front. The activation energy for diffusion decreases with increasing x.

Most of AIGaAs laser diodes (LDs) contain the doublehetero (DH) structure. The DH structure consists of AIGaAs layers with high Al composition as cladding layers and undoped GaAs or AIGaAs with low Al composition. Therefore, it is important for improvement of device characteristics to understand and control the diffusion of dopants. However, most work on the diffusion of dopants have been carried out on the diffusion in GaAs. In this paper, we compared electrical and optical properties of Si-doped AIGaAs with those of Se-doped AIGaAs and investigated the diffusion of Si, Se and Zn in the GaAs/Al0.48Ga0.52As DH structure by secondary ion mass spectroscopy (SIMS). Doping profile of Si is controllable rather than that of Se. However, from the viewpoint of device characteristics, Se is more suitable than Si.

A quantitative determination of the contributions of the triply-negatively charged Ga vacancies and of the doubly-positively charged Ga self-interstitials to Ga self-diffusion coefficient in GaAs has been carried out. Unde thermal equilibrium and intrinsic conditions, the contribution is characterized by an activation enthalpy of 6 eV for As-rich crystals and of 7.52 eV for Ga-rich crystals, while the contribution is characterized by an activation enthalpy of 4.89 eV for As-rich crystals and of 3.37 eV for Ga-rich crystals.

Diffusion of substitutional Cr atoms (Crs) in GaAs results from the rapid migration of interstitial Cr atoms (Cri) and their subsequent changeover to occupy Ga sites (or vise versa), a typical substitutional-interstitial diffusion (SID) process. There are two possible ways for the Cri-Crs changeover to occur: the kick-out mechanism in which Ga self-interstitials are involved, and the dissociative mechanism in which Ga vacancies are involved. The Crs indiffusion profiles are of characteristic shapes indicating the dominance of the kick-out mechanism, while the Crs outdiffusion profiles are error-function shaped, indicating the dominance of the dissociative mechanism. In this study, an integrated SID mechanism, which takes into account the effects of both the kick-out and dissociative mechanisms, is used to analyze Cr diffusion results. Going beyond just qualitative consistency, the Cr in- and outdiffusion features in GaAs are explained on a quantitative basis. In this model the kick-out mechanism dominates Cr indiffusion while the dissociative mechanism dominates Cr outdiffusion. Parameters used to fit existing experimental results provided quantitative information on the Ga self-interstitial contribution to the Ga self-diffusion coefficient.

We have measured the Shubnikov-de Haas (SdH) effect in the MBE grown Al0.6Ga0.4Sb/InAs/Al0.6Ga0.4Sh single quantum well (QW) for the magnetic fields up to 4.5T and the temperatures from 1.1 K to 4.2K. The carrier concentration of the two dimensional electron gas (2DEG) was varied via the negative persistent photoconductivity (NPPC) effect. By illuminating the sample with a red light-emitting diode (LED) at the low temperature, the carrier concentration of the 2DEG in the InAs well was reduced from 5.8×1011 cm−2 to 3.6×1011 cm−2 and the corresponding quantum lifetime increased from 0.16ps to 0.21ps. The effective mass was determined by the temperature dependence of SdH oscillations and equal to (0.0317 ± 0.0005)m0.

The 2-dimensional electron gas (2DEG) in an Al0.6Ga0.4Sb/InAs single quantum well (SQW) is studied using cyclotron resonance (CR) and Shubnikov - de Haas (SdH) techniques. SdH results show spin-splitting in Landau levels at magnetic field strength (B) as low as 1.5T. The effective mass (m*) of the 2DEG was obtained from the peak positions of the CR transmission spectra. The results exhibit oscillatory behavior as a function of B. The m* value extracted from die temperature dependence of the SdH oscillations is in good agreement with the average value of m* obtained from CR measurements. The effective mass is calculated as a function of B using an electron self-energy model based on the Hartree-Fock approximation. The calculated m* values also show oscillatory behavior similar to that of the measured CR m*. Both experiment and theory show that m* maxima are shifted from the integral values (both odd and even) of the filling factors.

Data are presented showing the effects of diffusion sources, surface encapsulation, and As overpressure on Mn diffusion in GaAs. Four different Mn-containing sources are used, including Mn, Mn,MN3As, and MnAs granules as well as Mn thin film deposited directly onto GaAs substrate. Smooth surface morphology with high surface Mn concentration can be obtained using MnAs (and in certain conditions, Mn3As) as diffusion source; different degrees of surface degradation are observed if otherwise sources are used as diffusion source. Data also show surface encapsulation and As overpressure have significant effects on the reaction between the source and GaAs and the Mn diffusion in GaAs.

InGaP thin films are evaluated as wide-bandgap materials for GaAs surface passivation. A 200-Å InGaP thin film increases GaAs photoluminescence intensity 25-fold and enables Schottky barrier heights of more than 0.6 eV on n-type GaAs layers with a carrier concentration of 3×1018 /cm3. These effects persist after annealing at 800 °C for 10 min. InGaP thin films are thus suitable as surface passivation films for high-performance GaAs-MESFETs.

Extended defect formation is studied in ion implanted GaAs. A number of different species including Si+, Al+, Mg+, Ge+, As+, and Sn* have been investigated. Cross-sectional TEM studies have been done comparing the as-implanted structure (amorphous or crystalline) with the final defect location and morphology. The defects are identified by the same classification scheme used for implanted and annealed silicon. It is found that the threshold dose for type I defect formation is very sensitive to the implant energy for heavier ion masses. Type II, III and IV defects are unstable at annealing temperatures below 900°C. Type V defects are of a loop morphology for Si* and Ge* implants. The source of the interstitials may be a kickout process as the implanted species moves onto substitutional sites. Type V defects for Sn implants appear as precipitates which at the annealing temperature appear to be migrating in the liquid phase. Upon cooling the Sn precipitates, in many cases, solidify as grey (α) Sn.

The activation of Si+ and Be+ ions implanted into InGaP, InGaAs or InAlAs grown by GSMBE and OMVPE was investigated as a function of ion dose and annealing temperature. Activation efficiencies close to 100% were obtained in InGaP and InGaAs for Be doses up to ∼1014 cm−2 and annealing temperatures of 700–850°C. Activation of Be was less efficient in InAlAs. By contrast, implanted Si displayed a saturation in active sheet electron densities at 1–3 × 1013 cm−2 and required higher annealing temperatures for optimum activation efficiency. High sheet resistance (≤108 μ/□) regions were created by O+ implantation into n+ InGaP or InAlAs, with hopping conduction dominating carrier transport in the bombarded material. For post-implant annealing temperatures above 750°C, the conductivity was restored to its initial value. No evidence was found for the creation of electrically active oxygen-related deep levels in either material.

Microscopic voids, formed from the condensation of supersaturated vacancy point defects, were recently discovered in implanted and annealed GaAs. These defects have been shown to suppress carrier concentrations. Since voids are formed only at relatively high temperatures (> 650 °C), the possibility exists that voids can be used for thermally stable implant isolation. In this paper, we report on the formation of highly resistive layers in GaAs, created by Al+ implantation and annealing in the 700–900 °C range. In samples containing voids, their sheet resistivities increased by about six orders of magnitude from the as-grown value. Formation of these thermally stable, high resistivity regions is different from the conventional H or O implant isolation techniques, which use lattice damage to create the isolation characteristics. However, since lattice damage is annealed out between 400–700 °C, this type of isolation becomes ineffective at high processing temperatures. By contrast, voids are stable at high processing temperatures, and potential advantages of using such defects for device isolation in GaAs are pointed out.

We have undertaken a systematic study of the effect of co-implantation on the electrical properties of C implanted in GaAs. Two effects have been studied, the additional damage caused by co-implantation and the stoichiometry in the implanted layer. A series of co-implant ions were used: group III (B, Al, Ga), group V (N, P, As) and noble gases (Ar, Kr). Co-implantation of ions which create an amorphous layer was found to increase the electrical activity of C Once damage was created, maintaining stoichiometric balance by co-implantation of a group III further increased the fraction of electrically active carbon impurities. Co-implantation of Ga and rapid thermal annealing at 950°C for 10 s resulted in carbon activation as high as 68%, the highest value ever reported.