ALBERT

Notes: Modification of GaAs by Si+-ion implantation is an important process for selective doping of the material. Defects caused by the implantation process often lead to incomplete electrical activation, and annealing procedures are used to recover the crystal quality. Results are presented of variable-energy positron (VEP) and cross-sectional transmission electron microscopy (XTEM) studies of a series of GaAs samples implanted with moderate to high fluences of 3×1013, 3×1014, and 1×1015 Si+ ions cm−2. Samples were irradiated at room temperature, and studied both before and after thermal annealing for one hour at 850 °C. In all cases XTEM results show a high density of small extrinsic dislocations after implantation, and VEP shows high concentrations of point (vacancy type) defects. Annealing leads to a decrease in the point-defect concentration in the lowest-fluence sample, but both XTEM and VEP confirm the formation of macroscopic (i.e., (approximately-greater-than)20 A(ring) diameter) voids following annealing. These data are discussed in the context of microscopic models for defect formation and migration.

Notes: Voids have been found in the near-surface region of GaAs/AlGaAs superlattices in a transmission electron microscopy study. The superlattices were Si- or Al-implanted and subsequently either furnace or rapid thermally annealed. Concurrent with the presence of voids is an inhibition of superlattice layer intermixing enhancement in the near-surface region. This inhibition does not occur in the deeper region of the samples where voids are not found. The voids can form via condensation of the Ga and As vacancies produced by the implantation process. We suggest that voids can depress dopant activation, suppress dopant diffusion, and inhibit the superlattice layer intermixing enhancement.

Notes: The microstructure in epitaxially oriented thin films of YBa2Cu3O7−x grown on (100) MgO by metalorganic decomposition has been studied by transmission electron microscopy and ion channeling. The as-prepared films consisted of single-crystal platelets lying flat on the MgO surface. The majority of the crystallites showed perfect alignment of their c axis with the [100] axis of MgO, while some crystallites were found to have a misorientation of up to 7.5°. Images of the interfacial regions showed good epitaxial growth to within one lattice spacing of the MgO substrate. He++ channeling measurements as a function of energy from 1 to 4.5 MeV indicated a 0.51° spread in crystallite orientation. Extrapolation of the channeling measurements to the limit of zero crystallite spread gave a minimum yield of 0.20 for bulk YBa2Cu3O7−x , which is much larger than the value reported for single crystals. The large backscattering yield is attributed to the grain boundaries in the film. A relatively strain-free interface was indicated by channeling results.

Notes: Proton exchanged samples of LiNbO3 have been profiled by micro-Raman spectroscopy, secondary ion mass spectroscopy, Rutherford backscattering channeling, and by x-ray diffraction (XRD). Following proton exchange (PE) there are two different phases in addition to pure LiNbO3 detected by XRD. After successive annealing steps the outermost phase disappears and an interfacial region forms progressively between PE and LiNbO3. Specific vibrational bands are correlated to electro-optic and nonlinear optical properties of the system, and the recovery of these properties upon annealing is correlated to chemical bonding changes.

Notes: A simple and reproducible process for the open-tube diffusion of zinc from (ZnO)x(SiO2)1−x source films into GaAs, Al0.2Ga0.8As and GaAs0.6P0.4 is reported. (ZnO)x(SiO2)1−x films were deposited onto compound semiconductor substrates by metalorganic chemical vapor deposition. A capping layer of SiO2 was deposited on top of the source films. The diffusions were performed in flowing nitrogen at 650 °C. Diffusion depths from 0.2 μm to several micrometers were readily achieved. The diffusion front in n-type substrates is abrupt and the average hole concentration for diffused layers in GaAs is approximately 8 × 1019/cm3. The dependence of the diffusion depth on the source film composition (x=0.04–x=1.00) is presented. The dependence of the diffusion depth on the source film thickness and the SiO2 cap layer thickness is also reported.

Notes: Voids, formed by the condensation of an excess of implantation-induced vacancies, have been recently identified as the defect directly responsible for dopant diffusion and electrical activation anomalies in Si-implanted and annealed GaAs and GaAs/AlGaAs superlattice materials. Depending on the implanted dose, voids can be distributed either throughout the implanted region or in two bands. We have examined the origin of this void distribution difference. In the as-implanted sample associated with the latter case, a buried continuous band of amorphous GaAs has formed. GaAs formed by the recrystallization of amorphous GaAs does not contain excess vacancies and therefore cannot form voids. However, on either side of the amorphous layer, the excess vacancies can condense to form the observed banded distribution of voids. In the as-implanted sample associated with the former case, a continuous amorphous GaAs layer did not form, and therefore, upon annealing, voids are seen throughout the implanted region.

Notes: Oxygen diffusivity in silicon in the intermediate temperature range 500–650 °C has, for the first time, been directly measured. Oxygen diffusion in this temperature range is enhanced relative to the normal diffusion of interstitial oxygen. The degree of enhancement increases with decreasing temperature, with the enhancement factor reaching 〉102 at ≤550 °C. The diffusivity at 550–750 °C is found to vary among Si wafers by as much as 10×. Oxygen diffusivity generally decreases with increasing annealing time but does not vary proportionally with oxygen concentration as expected from a molecular oxygen model.

Notes: We have used Raman scattering to study the lattice disorder created by the implantation of 1-MeV Si ions into GaAs. Using the change in the longitudinal optical (LO) phonon-line position as the signature for lattice damage, combined with chemical etching for controlled layer removal, we monitored the evolution of the disorder depth profile as a function of implantation dose. The shape of the depth profile of the disorder agrees with the theoretical simulation trim for doses of 1×1014 cm−2 or lower. For higher doses a saturation is observed in the amount of residual disorder. This saturation is a manifestation of dynamic annealing occurring during the high-energy implantations, which we attribute to enhanced defect mobility, induced by the transfer of energy to the lattice, in atomic collision cascade processes. In order to correlate the spectral features in the Raman spectra with structural changes in the ion-implanted samples, we characterized the implantation-induced lattice damage using ion-channeling and transmission electron microscopy (TEM) measurements. The residual defects in the MeV-implanted samples are found to consist of dislocation loops and discrete point defects dispersed in an otherwise perfect (although probably strained) crystalline lattice. An average distance between defects was estimated from the channeling and TEM studies, and compared with the coherence-length parameter L used in the "spatial correlation model,'' which is commonly used to interpret quantitatively the Raman spectra of ion-implanted materials. Although the model gives a good fit to our data in terms of the position and linewidth of the LO phonon peak, no clear correlation could be established between L and the interdefect separations. We also observed the appearance of the broadbands at about 70, 180, and 245 cm−1, in the Raman spectra, which are commonly attributed to amorphous GaAs, although no trace of amorphous material was detected by the TEM analysis. Our results indicate that the quantitative interpretation of Raman spectra to determine crystalline properties of ion-implanted materials, as well as the assignment of Raman spectral features to particular defect structures, is not unambiguously established yet.

Notes: Layer intermixing in MeV Si-implanted GaAs/AlGaAs superlattices (SLs) with doses between 3×1015 and 1×1016 /cm2 has been examined by transmission electron microscopy and secondary-ion mass spectrometry. After either rapid thermal annealing at 1050 °C for 10 s or furnace annealing at 850 °C for 3 h, all the SLs showed a highly crystalline, defect-free zone in the near-surface region followed by a band of secondary defects, with the maximum density located about 1 μm below the surface. A totally mixed region, within the secondary defect band, occurred only in the SL implanted to 1×1016 Si/cm2 and annealed at 850 °C for 3 h. At lower doses or under rapid thermal annealing, only slight Al/Ga interdiffusion was observed, primarily in the layers that contained the high density of dislocation defects. For either annealing condition, the Si concentration profiles showed only slight broadening and they correlated with the distribution of secondary defects as well as with the depth of the intermixed layer. The effects of dynamic annealing and surface on the implantation energy dependence, i.e., MeV vs keV, of layer intermixing are discussed.

Notes: A fast diffusing species of ion-implanted arsenic has been observed in silicon dioxide annealed in a nitrogen annealing ambient. Below 1020 cm−3 concentration this arsenic diffuses 20× faster at 1100 °C than the slow arsenic found in oxidizing ambient. The effective activation energy of diffusion of this fast arsenic is 0.71 eV, in sharp contrast to 4.3 and 4.0 eV for the slow arsenic in O2 and O2 /H2 O, respectively. The fast species is believed to be arsenic occupying either the oxygen site or interstitial site in the SiO2 network.