We describe the metal-organic chemical vapor deposition (MOCVD) growth of InAsSb/InAs and GaAsSb/GaAs(P) multiple quantum well (MQW) and InAsSb/InAsP and InAsSb/InPSb strained-layer superlattice (SLS) active regions for use in mid-infrared emitters. We also describe the growth and initial characterization of GaAsSbN/GaAs MQW structures. By changing the layer thickness and composition of the InAsSb SLSs and MQWs, we have prepared structures with low temperature (<20 K) photoluminescence wavelengths ranging from 3.2 to 6.0μ m. We have made gain-guided, injection lasers using undoped, p-type AlAs 0.16 Sb 0.84 for optical confinement and both strained InAsSb/InAs MQW and InAsSb/InAsP and InPSb SLS active regions. The lasers and LEDs utilize the semi-metal properties of a p-GaAsSb/n-InAs heterojunction as a source for electrons injected into the active regions. Cascaded, semi-metal, mid-infrared, injection lasers with pseudomorphic InAsSb multiple quantum well active region lasers and LEDs are reported. We also report on GaAsSb/GaAs(P) lasers and LEDs emitting at 1.1 to 1.2 μm grown on GaAs substrates and using AlGaAs layers for confinement.

To gain insight into the nitrogen-related defects of InGaAsN, nitrogen vibrational mode spectra, Hall mobilities, and minority carrier diffusion lengths are examined for InGaAsN (1.1 eV band gap) grown by molecular beam epitaxy (MBE). Annealing promotes the formation of In–N bonding, and lateral carrier transport is limited by large scale (≫mean free path) material inhomogeneities. Comparing solar cell quantum efficiencies with our earlier results for devices grown by metalorganic chemical vapor deposition (MOCVD), we find significant electron diffusion in the MBE material (reversed from the hole diffusion in MOCVD material), and minority carrier diffusion in InGaAsN cannot be explained by a “universal,” nitrogen-related defect.

We have investigated deep levels in 1.05 eV, Sn-doped, n-type In0.075Ga0.925As0.975N0.025 lattice-matched to GaAs. The samples were grown by metal–organic chemical vapor deposition. Capacitance–voltage measurements were used to determine the electron concentration in both as-grown and postgrowth annealed samples, and a decrease in net electron concentration of about 1.5×1017 cm−3 was observed following annealing. Deep level transient spectroscopy measurements are consistent with the presence of four majority-carrier electron traps, E1 (a broad distribution extending from the conduction band edge to approximately EC−0.2 eV), E2(0.36 eV), E3(0.34 eV), and E4(0.82 eV), as well as one minority-carrier hole trap, H1(0.71 eV), in our material. It is shown that E2 and H1, both of which are present in the annealed material only, are likely the same defect observed under conditions of electron and hole emission, respectively. Current–voltage-temperature measurements indicate a thermal activation energy of 0.35 eV for reverse bias current transport, which is in close agreement with the activation energy of E2. It is thus demonstrated that the E2/H1 defect is a recombination–generation center which contributes to current transport in our InGaAsN-based test diode.

Nitrogen vibrational mode spectra, Hall mobilities, and minority carrier diffusion lengths are examined for InGaAsN (≈ 1.1 eV bandgap) grown by molecular beam epitaxy (MBE) and metal-organic chemical vapor deposition (MOCVD). Independent of growth technique, annealing promotes the formation of In-N bonding, and lateral carrier transport is limited by large scale (Ęmean free path ) material inhomogeneities. Comparing solar cell quantum efficiencies for devices grown by MBE and MOCVD, we find significant electron diffusion in the MBE material (reversed from the hole diffusion occurring in MOCVD material), and minority carrier diffusion in InGaAsN cannot be explained by a “universal”, nitrogen-related defect.

InxGa1-xAst-yNy quaternary alloys offer the promise of longer wavelength, greater than or equal to 1.3 pm optical transceivers grown on GaAs substrates. To achieve, acceptable radiative efficiencies at 1.3 pm, highly-strained InGaAsN quantum wells (x approximate to 0.4, y approximate to 0.005) are being developed as laser active regions. By introducing GaAsP layers into the active region for strain-compensation, gain can be increased using multiple InGaAsN quantum wells. In this work, we report the first strain-compensated, 1.3 pm InGaAsN MQW lasers, Our devices were grown by metal-organic chemical vapor deposition. Lasers with InGaAsN quantum well active regions are proving superior to lasers constructed with competing active region materials. Under pulsed operation, our 1.3 pm InGaAsN lasers displayed negligible blue-shift from the low-injection LED emission, and state-of-the-art characteristic temperature (159 K) was obtained for a 1.3 mum laser.

DLTS measurements have been performed on InGaAsN. Four hole traps have been identified in 1.05eV, p-type InGaAsN and the removal of a midgap trap (∼0.5eV) during annealing has been correlated with improved bulk material properties. Improvements in MOCVD growth conditions resulted in a reduction of trap density in 1.05eV, p-type InGaAsN. Increased indium and nitrogen composition has been correlated with higher defect concentrations in p-type InGaAsN. Two electron traps have been identified in 1.15eV, n-type InGaAsN and annealing was found to reduce the density of the shallow electron trap.

The variation of the value of the linewidth of an excitonic transition in InGaAsN alloys (1% and 2% nitrogen) as a function of hydrostatic pressure using photoluminescence spectroscopy is studied at 4K. The excitonic linewidth increases as a function of pressure until about 100 kbar after which it tends to saturate. This pressure dependent excitonic linewidth is used to derive the pressure variation of the exciton reduced mass using a theoretical formalism based on the premise that the broadening of the excitonic transition is caused primarily by compositional fluctuations in a completely disordered alloy. The linewidth derived ambient pressure masses are compared and found to be in agreement with other mass measurements. The variation of this derived mass is compared with the results from a nearly first-principles approach in which calculations based on the local density approximation to the Kohn-Sham density functional theory are corrected using a small amount of experimental input.

By employing a reactive low-temperature wafer bonding technique, we have demonstrated oxide-defined 850 nm vertical-cavity surface-emitting lasers (VCSEL's) on Si substrates. Devices reach a differential quantum efficiency of 53% and a light output power of 7.1 mW under room temperature and continuous-wave operation without a heat sink.

In this paper, we demonstrate propagation losses of less than α=1.0 cm-1 at both λ0=1.55 and 0.63 μm for an oxidized Al0.3Ga0.7As/Al0.85 Ga0.15As waveguide. These loss reductions, and the realization of transparency well above the GaAs band edge, are achieved through both an improved structure (thicker cladding layer, larger Al composition contrast) and a modification to the standard oxidation process (via the controlled addition of O2). The addition of trace amounts of oxygen reduces the oxide surface roughness, increases the index of oxidized AlxGa1-xAs (x<0.5), and increases oxidation rates (0.3<x<0.85). These process enhancements lead to improved mode confinement and reduced absorption and scattering losses

Electron and hole transport in compensated InGaAsN (≈2% N) are examined through Hall mobility, photoconductivity, and solar cell photoresponse measurements. Short minority carrier diffusion lengths, photoconductive-response spectra, and doping dependent, thermally activated Hall mobilities reveal a broad distribution of localized states. At this stage of development, lateral carrier transport appears to be limited by large scale (≫ mean free path) material inhomogeneities, not a random alloy-induced mobility edge.

Selectively oxidised vertical-cavity lasers emitting at 1294nm using InGaAsN/GaAs quantum wells which exhibit continuous wave operation at and above room temperature are reported for the first time. The lasers employ two n-type Al0.94Ga0.06As/GaAs distributed Bragg reflectors each with a selectively oxidised current aperture adjacent to the optical cavity, and the top output mirror contains a tunnel junction to inject holes into the active region. Continuous wave singlemode lasing is observed up to 55 degrees C. These lasers exhibit the longest wavelength reported to date for vertical-cavity surface lasers grown on GaAs substrates.

Vertical-cavity lasers emitting at 1.3 /spl mu/m are extremely attractive for high bandwidth fiber communications where it is advantageous to operate at the dispersion minimum of silica optical fiber. To date, VCSELs based on pseudomorphic GaAs materials have achieved lasing emission only slightly longer than 1.3 /spl mu/m. We report the first 1.3 /spl mu/m selectively oxidized VCSELs using InGaAsN quantum wells which operate at continuous wave at and above room temperature.

Topographic changes and surface contact potential variations near defects on the surface of AlGaAs/GaAs double heterojunction structures grown on Ge substrates are studied using scanning force microscope and electrostatic force microscope. Comparison with transmission electron microscopy results indicates that these surface defects are directly related to stacking faults originated from the GaAs/Ge interface. The surface contact potential inhomogeneities near these defects are consistent with variations in Si dopant concentration.

GaAs/Ge interfaces are receiving renewed attention due to the recent development of III-V/Ge based solar cells for space satellite power, and the development of improved GeSi relaxed buffer layers which may lead to III-V integration onto Si utilizing Ge/GeSi interlayers in the near future. In both cases, the heterovalent nature of the GaAs/Ge interface introduces defects that must be understood and controlled to provide optimal device performance and new device structures. For the latter case however, the presence of residual dislocations emanating from the lattice mismatch between Ge and Si adds an additional factor that must be understood to develop optimum GaAs-based devices on Ge/GeSi/Si substrates. This paper presents the state of progress toward achieving device-quality III-V epitaxy onto both Ge and Ge/GeSi/Si substrates, focusing on the GaAs/Ge interface. For epitaxy on Ge wafers, monolayer-scale control during GaAs/Ge nucleation is shown to completely prevent the formation of antiphase domains (APD's). Moreover, by using migration enhanced epitaxy during initial GaAs growth, atomic interdiffusion and autodoping on either side of the III-V/Ge interface are reduced to negligible levels, with highly abrupt interfaces being achieved. With these growth conditions, high minority carrier lifetimes in GaAs on Ge are obtained without the use of conventional thick GaAs buffer layers. These same growth conditions were successfully transferred to GaAs growth on low dislocation density Ge/GeSi/Si substrates. Microscopy reveals no evidence for APD formation, and the threading dislocation density in thick GaAs overlayers is dictated by the residual density in the underlying relaxed Ge buffer. This indicates that the GaAs/Ge interface on the relaxed Ge/GeSi/Si wafers behaves as an ideal, low mismatched interface, similar to GaAs growth on Ge wafers. The excellent structural quality of these layers directly translates to good electronic quality, with a minority carrier bulk lifetime of 7.7 ns being measured for n-type (1.1x10(17) cm(-3)) GaAs grown on Ge/GeSi/Si substrates. This is the highest lifetime ever reported for GaAs growth on Si substrates, and portends that high performance III-V minority carrier devices grown on Si will soon be feasible.

A high bulk minority-carrier lifetime in GaAs grown on Si-based substrates is demonstrated. This was achieved by utilizing a step-graded Ge/GeSi buffer (threading dislocation density 2×106 cm−2) grown on an offcut (001) Si wafer, coupled with monolayer-scale control of the GaAs nucleation to suppress antiphase domains. Bulk minority-carrier lifetimes (τp) were measured using room-temperature time-resolved photoluminescence applied to a series of Al0.3Ga0.7As/GaAs/Al0.3Ga0.7As double-heterojunction structures doped n=1.1×1017 cm−3 with GaAs thicknesses of 0.5, 1.0, and 1.5 μm. A lifetime τp=7.7 ns was determined for GaAs grown on Si. The extracted interface recombination velocity of 3.9×103 cm/s is comparable to recombination velocities found for Al0.3Ga0.7As/GaAs interfaces grown on both GaAs and Ge wafers, indicating that the crosshatch surface morphology characteristic of strain-relaxed Ge/GeSi surfaces does not impede the formation of high-electronic-quality interfaces. These results hold great promise for future integration of III–V minority-carrier devices with Si wafer technologies.

The nucleation and growth of GaAs films on offcut (001) Ge wafers by solid source molecular beam epitaxy (MBE) is investigated, with the objective of establishing nucleation conditions which reproducibly yield GaAs films which are free of antiphase domains (APDs) and which have suppressed Ge outdiffusion into the GaAs layer. The nucleation process is monitored by in-situ reflection high energy electron diffraction and Auger electron spectroscopy. Several nucleation variables are studied, including the state of the initial Ge surface (single-domain 2×1 or mixed-domain 2×1:1×2), the initial prelayer (As, Ga, or mixed), and the initial GaAs growth temperature (350 or 500°C). Conditions are identified which simultaneously produce APD-free GaAs layers several microns in thickness on Ge wafers with undetectable Ge outdiffusion and with surface roughness equivalent to that of GaAs/GaAs homoepitaxy. APD-free material is obtained using either As or Ga nucleation layers, with the GaAs domain dependent upon the initial exposure chemical species. Key growth steps for APD-free GaAs/Ge growth by solid source MBE include an epitaxial Ge buffer deposited in the MBE chamber to bury carbon contamination from the underlying Ge wafer, an anneal of the Ge buffer at 640°C to generate a predominantly double atomic-height stepped surface, and nucleation of GaAs growth by a ten monolayer migration enhanced epitaxy step initiated with either pure As or Ga. We identify this last step as being responsible for blocking Ge outdiffusion to below 1015 cm−3 within 0.5 microns of the GaAs/Ge interface.

The epitaxial growth of GaAs on Si substrates through the use of a Ge/graded Si1−xGex/Si buffer layer would allow monolithic integration of GaAs-based opto-electronics with Si microelectronics. As an initial step toward this goal, this study examines factors which influence the quality of GaAs growth by molecular beam epitaxy (MBE) on bulk Ge substrates. Key findings include the need for an epitaxial Ge smoothing cap deposited in the MBE chamber, the significant detrimental effect of As overpressure on the resultant GaAs crystalline quality, and the efficiency of a very thin (∼3 nm) migration enhanced epitaxy (MEE) nucleation layer at suppressing both anti-phase domain (APD) formation and interdiffusion across the GaAs/Ge heterointerface. Using this developed optimized growth process, APD-free GaAs on Ge is obtained which has undetectable Ga and Ge cross-diffusion, and As diffusion into the substrate at ⩽1×1018 cm−3. Preliminary results for growths on Ge/Si1−xGex/Si substrates are also presented.

The surface morphology of GaAs films grown on offcut Ge substrates is studied using a scanning force microscope (SFM). We investigated the effects of the Ge buffer layer, growth temperature, film thickness, and prelayer on the GaAs surface morphology. The starting Ge substrates are offcut 6° toward the [110] direction to minimize single steps on the substrates before molecular beam epitaxial film growth. We find that comparing with GaAs samples grown without Ge buffer layers or with unannealed Ge buffer layers, samples with annealed Ge buffer layers are much smoother and contain no antiphase boundaries (APBs) on the surface. For thick (≥1 µm) GaAs films with an annealed Ge buffer layer, the surfaces display crosshatch lines and elongated mounds (along \([\bar 110]\), which are associated with the substrate offcut direction. As the film thickness increases, the crosshatch lines become shorter, denser and rougher, and the mounds grow bigger (an indication of GaAs homoepitaxial growth). We conclude that annealed Ge buffer layers are crucial for growing high quality GaAs films with few APBs generated during the growth. In addition, under optimal conditions, different prelayers make little difference for thick GaAs films with annealed Ge buffer layers.

The quality of GaAs films grown on offcut (001) Ge substrates has been extensively investigated as a function of solid-source molecular beam epitaxy growth parameters using transmission electron microscopy. In the worst case, GaAs/Ge films may feature unexpectedly high threading dislocation densities and numerous anti-phase domains (APDs). The condensation of As point defects incorporated during low temperature (similar to 350 degrees C) GaAs co-evaporation is responsible for the nucleation of dislocation loops with in-plane a/2[110] Burgers vectors. Subsequent loop expansion produces an excess of threading dislocations and an irregular misfit dislocation network. Regarding anti-phase disorder, As-initiated GaAs is sensitive to temperature-dependent Ge surface roughening effects, whereas Ga-initiated growth is highly sensitive to background contamination levels. Nevertheless, APD-free GaAs/Ge films initiated with either As or Ga were achieved by controlling the exposure conditions and relying on an epitaxial Ge buffer annealed at high temperature to provide a clean, single domain surface for growth initiation. A model for APD coalescence is presented to explain observed variations in APD height.