Search

The bandedge optical properties of GaAsBi films, as thick as 470 nm, with Bi content varying from 0.7% Bi to 2.8% Bi grown by molecular beam epitaxy on GaAs substrates are measured by photoluminescence (PL) and photothermal deflection spectroscopy (PDS). The PDS spectra were fit with a modified Fernelius model which takes into account multiple reflections within the GaAsBi layer and GaAs substrate. Three undoped samples and two samples that are degenerately doped with silicon are studied. The undoped samples show a clear Urbach absorption edge with a composition dependent bandgap that decreases by 56 meV/% Bi and a composition independent Urbach slope parameter of 25 meV due to absorption by Bi cluster states near the valence band. The doped samples show a long absorption tail possibly due to absorption by gap states and free carriers in addition to a Burstein–Moss bandgap shift. PL of the undoped samples shows a lower energy emission peak due to defects not observed in the usually available thin samples (50 nm or less) grown under similar conditions.

The optical absorption of GaAsN films grown by molecular beam epitaxy on GaAs substrates is measured using the mirage effect photothermal deflection spectroscopy (PDS). The PDS spectra were fitted with a modified Fernelius model, which takes into account multiple reflections within the GaAsN layer and GaAs substrate. This allowed the extraction of bandedge parameters for a series of GaAsN films with N content varying from 0.24% to 1.4% N. All films show a clear Urbach absorption edge with a composition-dependent bandgap consistent with literature and Urbach slope parameters roughly 3 times larger than GaAs values.

Two databases of complex dielectric constants, for InGaAs and InAlAs, near-lattice matched to InP, and around temperatures of 500°C, have been fit using a transfer function model with temperature and composition-dependent coefficients. The parameters of the model are determined with a least-squares algorithm with recursive “whitening” of the error, which shows fast convergence to a near-optimal solution, even when handling a large number of parameters. The model can be inverted by a simple algorithm to retrieve temperature and composition information from optical measurements obtained by spectroscopic ellipsometry. The level of accuracy achieved makes this method adequate for temperature, composition, and thickness determination during MBE growth.

The α-particle energy loss method (AEL) has been implemented in situ to monitor film thickness and composition during growth of GaAs, InP and LaF 3 based materials by molecular beam epitaxy (MBE). In the AEL method, a 228 Th source is used to recoil implant a 5 mm diameter region of the surface of the wafers with the α-emitter daughter isotope 224 Ra prior to growth. The implanted nuclei decay with a half life of 3.7 days through a sequence of daughters which emit alpha particles at different energies. Deposition on the surface causes the emission lines to be shifted to lower energies due to energy loss in the film. For substrates marked with a low activity (∼ 30 kBq; similar to activity of smoke detectors) we are able to measure film thickness with ± 6 nm uncertainty and growth rate with ± 0.01 nm/s uncertainty in real time. By measuring the relative growth rates of the different materials, AEL also allows us to infer the composition of a ternary laver film as well as the sticking coefficients rates directly at different growth temperatures.

Precise control of layer thickness and alloy compositions are very critical for the successful growth of vertical cavity surface emitting lasers (VCSEL) and resonant cavity enhanced photodetectors (RCEPD). In traditional MBE, the growth of these devices requires extensive calibrations. Recently, spectroscopic ellipsometry (SE) was shown to be an accurate method for determining the composition and thicknesses of III–V semiconductor layers in situ during growth. This paper reports the growth of AlAs/GaAs/InGaAs RCEPDs using SE control with no a priori calibration curves. The experimentally measured cavity resonance of the reflectivity spectrum and the photoresponse peak of the fabricated photodiode lie within 3 nm of the target design value of 990 nm. These results show that feedback control during MBE allows quicker and more reproducible growth of RCEPDs with precise control of layer thickness and alloy composition than calibration-based control.

This paper describes the electrical properties of three n-type InP epilayers grown by MOVPE, in the same reactor, with three different bottles of trimethvlindium. Two typical samples show a reduced room temperature Hall mobility along with high temperature excitation to the conduction band. These features are in excellent agreement with a model accounting for a deep center or complex, with a binding energy dependent on the particular source material used for growth. Such a center is a strong scatterer when ionized, thus reducing high temperature mobility and becoming electrically inactive when neutral. The 77 K Hall mobility of the samples is not a measure of their purity and possible depletion effects did not appear to be significant.