AVS2004 Session SC-WeA: Narrow Gap Semiconductors

Wednesday, November 17, 2004 2:00 PM in Room 304C

Wednesday Afternoon

Start	Invited?	Item
2:00 PM	Invited	SC-WeA-1 Growth and Properties of Dilute Nitride Semiconductors with a Bismuth Surfactant T. Tiedje, D,A, Beaton, S. Tixier, M.B. Whitwick, E.C. Young, N.R. Zangenberg (University of British Columbia, Canada); S. Francoeur (National Renewable Energy Laboratory) The dilute nitride semiconductor In_yGa_1-yAs_1-xN_x is a promising new III-V alloy for solar cells and lasers in which low concentrations of N cause an anomalous reduction in the bandgap. N alloying degrades the electronic properties, possibly through formation of localized N cluster states. We show that the electronic properties of GaAs_1-xN_x can be improved with the use of a Bi surfactant during plasma assisted MBE growth. The effect of modifying the active nitrogen species (ratio of atoms to excited state molecules) in plasma-assisted MBE growth has also been explored. The Bi surfactant reduces the surface roughness by a factor of ten and produces step flow growth at temperatures as low as 460°C. The surfactant also improves the room temperature photoluminescence (PL) efficiency in as-grown and annealed samples, and reduces the density of shallow gap states, as determined from the shape of the PL spectra. The N incorporation increases with the Bi coverage by up to 50% when the surface is saturated with Bi. In-situ mass spectroscopy experiments show that NAs (but not NBi) is present in the gas phase, during growth and that the partial pressure of NAs increases slightly with Bi flux, suggesting that the Bi surface layer enhances the reaction of active nitrogen with GaAs. The Bi covers the surface but does not incorporate (<2x10¹⁷ cm^-3) under normal growth conditions, in which there is an As overpressure and the substrate temperature is >450°C. If the As:Ga flux ratio is reduced to approximately 1:1, Bi will incorporate at low substrate temperatures (380°C). The dilute Bi alloys show room temperature photoluminescence footnote1 and similar to the dilute nitrides the bandgap has an anomalously large concentration dependence. footnote1S. Tixier et al, Appl. Phys. Lett. 82, 2245 (2003).
2:40 PM		SC-WeA-3 Nitrogen Incorporation in GaAsN Films and GaAsN/GaAs Superlattices H.A. McKay, M. Reason, X. Weng, N. Rudawski, W. Ye, V. Rotberg, R.S. Goldman (University of Michigan) GaAsN and InGaAsN alloys with a few percent N have potential applications in infrared laser diodes, high efficiency solar cells, and other electronic devices. However, as-grown materials often exhibit poor photoluminescence efficiencies and lower than expected carrier concentrations and mobilities. A few studies have suggested that control of N incorporation via ex-situ annealing or superlattice (SL) growth may lead to improved optical and electronic properties. In this work, we are exploring in-situ approaches to controlling N incorporation during the growth of GaAsN films and GaAsN/GaAs SLs. In the case of GaAsN films, we have investigated N incorporation in GaAsN films grown by solid-source molecular beam epitaxy (MBE) using a 10% N₂/90%Ar or pure N₂ RF plasma source, with As₂ or As₄, and a variety of growth temperatures, Si-doping, and V/III ratios. Nuclear Reaction Analysis and Rutherford Backscattering Spectrometry in channeling and non-channeling conditions reveal significant composition-dependent non-substitutional incorporation of N, presumably as N-N or N-As split interstitials. We find that non-substitutional N incorporation is minimized for films grown at 400°C with pure N₂, apparently independent of As species, Si-doping, and V/III ratio. In the case of GaAsN/GaAs SLs, we have developed an in-situ approach to prevent incorporation of N into the GaAs barriers of the superlattice, using an independently-pumped plasma source chamber, separated from the MBE via a gate valve. High-resolution x-ray diffraction studies reveal significant improvements in the interface quality for superlattices prepared with the active N flux controlled via the gate valve in comparison with a conventional shuttering approach. The effects of N incorporation on the electrical and optical properties of GaAsN films and GaAsN/GaAs SLs will also be presented. This work is supported by DOE, AFOSR-MURI, NSF-NER, NASA-Lewis, and TRW.
3:00 PM		SC-WeA-4 Optical Properties of GaAs_1-xN_x: A Tight Binding and Variable Angle Spectroscopic Ellipsometry Study S. Turcotte, S. Larouche, J.-N. Beaudry, N. Shtinkov, L. Martinu, R.A. Masut (École Polytechnique de Montréal, Canada); R. Leonelli (Université de Montréal, Canada); P. Desjardins (École Polytechnique de Montréal, Canada) We have carried out a series of tight-binding (TB) calculations and variable-angle spectroscopic ellipsometry (VASE) measurements in order to investigate the excited states above the band gap of GaAs_1-xN_x layers on GaAs(001). The calculations are carried out using an empirical TB model in the sp³d⁵s*s_N parameterization, which provides a band-anticrossing description of the GaAs_1-xN_x band structure over the entire Brillouin zone for energies up to 5 eV above the valence band maximum. A series of fully-coherent GaAs_1-xN_x layers with x up to 0.012 were grown on GaAs(001) by organometallic vapor phase epitaxy using trimethylgallium, tertiarybutylarsine, and dimethylhydrazine precursors. The GaAsN dielectric function was reliably determined from 0.76 to 4.4 eV through point by point fitting of the experimental data with a model which also takes into account the GaAs substrate and the oxide overlayer. In addition to the band edge transition E_-, the L point related optical transition E₁ and their split-off replica, we also observe, for the first time in room temperature VASE measurements, the E₊ level signature which appears as a well defined critical point contribution to the dielectric function. The comparison between experimental data and TB calculations provides information on the specific contributions of the different Brillouin zone points to the GaAsN dispersion curves.
3:20 PM		SC-WeA-5 Growth Modes of InN on Sapphire (0001) with GaN Buffer Layers S.R. Leone (University of California); B. Liu, D.X. Chen (Lawrence Berkeley National Laboratory); T. Kitajima (National Defense Academy of Japan) InN is an important group-III nitride semiconductor. Use of InN can extend the working wavelength of the nitride-based optoelectronic devices from ultraviolet to infrared. However, heteroepitaxy of InN has encountered difficulties due to the thermal instability of InN and the large lattice mismatches between InN and the commonly used substrates (e.g., sapphire). During growth, the high equilibrium pressure of nitrogen requires a high V/III flux ratio to suppress InN decomposition, which often results in undesired three-dimensional (3D) rough surfaces. In this work, using atomic force microscopy (AFM) and scanning tunneling microscopy (STM), we study the growth modes and the surface morphologies of InN grown by plasma-assisted molecular beam epitaxy on sapphire (0001) substrates with intermediate GaN buffer layers. With smooth GaN buffer layers, 3D InN islands are observed to have mesa-like shapes with atomically flat tops. However, prolonged growth in this mode does not produce continuous two-dimensional (2D) InN films by coalescence of InN islands. With 3D rough GaN buffer layers, continuous 2D InN films are obtained showing the characteristics of step-flow growth. STM imaging reveals the defect-mediated surface morphology of the 2D InN films, including surface termination of screw (or mixed) dislocations and a high density of shallow surface pits with depths of about 0.1 nm. The mechanisms of the different growth modes and surface defect formation are also discussed.
3:40 PM		SC-WeA-6 Energy Gap and Stokes-like Shift in Cubic In_x Ga_1-xN Epitaxial Layers DG Pacheco-Salazar, J.R.L. Fernandez (University of Sao Paulo, Brazil); J. Soares (University of Illinois at Urbana Champaign); J.R. Leite (University of Sao Paulo, Brazil); F. Cerdeira, E.A. Meneses (University of Campinas, Brazil); S.F. Li, O. Husberg, D.J. As, K. Lischka (University of Paderborn, Germany) Group-III nitrides have been intensively investigated due to their recent applications in optoelectronic and electronic devices. The light emission mechanism of efficient blue-green-ultraviolet LED and laser diodes based on these materials is still object of investigation, and several works have been carried out recently on the optical properties of the In_xGa_1-xN alloy, the active medium in these optoelectronic devices. In the present work photoluminescence (PL), photoluminescence excitation (PLE) and Cathodoluminescence (CL) are used to investigate emission and absorption mechanisms in cubic 100 nm thick In_xGa_1-xN epitaxial layers grown on thick GaN/GaAs(001) buffer layers by MBE. High resolution x-ray diffraction (HRXRD) revealed that the InGaN layers were pseudomorphic with the GaN buffer. The In fraction x was calculated from the strained lattice constants obtained from the HRXRD reciprocal space maps. Reflectivity was used to measure the thickness of our films. PL and PLE spectra were recorded at 7 K and 300 K. CL spectra were performed at room temperature. The main features observed in the PL spectra are the characteristic emission from the band edge region of c-GaN and a lower energy peak which we ascribe to the InGaN layer. From the PLE spectra, we have determinated the alloy energy gap as a function of the In content. Comparing results from PL and PLE we observe a large Stokes-like shift for all samples. Another absortion band at lower energy than that of the alloy energy gap is observed on some of the PLE spectra. We tentatively ascribe this band to an absorption mechanism taking place in In-rich regions in the InGaN alloy. These findings are supported by a multi-peak structure obtained by PL and depth-resolved CL. In depht-resolved CL, the relative intensity of the peaks changes with increasing excitation depth, indicating a second InGaN fase close to the GaN/InGaN interface.
4:00 PM	Invited	SC-WeA-7 InAs-based Heterojunction Bipolar Transistors P.W. Deelman, P.D. Brewer, D.H. Chow, K.R. Elliott, T. Hussain, R.D. Rajavel, S.S. Thomas, III (HRL Laboratories) We have demonstrated the first InAs-based heterojunction bipolar transistors (HBTs) with f_T and f_max simultaneously exceeding 100GHz. Our best small-area (0.4 × 3.0 μm²) HBTs feature an f_T of 215GHz, a V_be of 0.35V, and a β of 60. Using similar devices, we have successfully fabricated the world’s first InAs-based integrated circuit - - - a divide-by-sixteen circuit operating at 27GHz. The divider consists of 62 transistors fabricated on a 3" wafer. The promise that HBTs nearly-lattice-matched to InAs potentially operate at higher frequencies and with lower power-delay products than current technologies motivated this work. Nevertheless, we have had to overcome several materials-related obstacles in order to achieve these results. For example, insulating substrates for the 6.1Å family of materials do not exist, and the nature of the band alignments in these materials dictate that at least one, if not all, the layers in the structure be strained. We employed two approaches to deal with the substrate issues. InAs transistors were grown by molecular beam epitaxy either on InP substrates using semi-insulating, strain-relaxed AlGaAsSb buffers or coherently on 3" InAs substrates. The latter approach required subsequent wafer bonding to sapphire carriers. Our baseline structures utilized either an InAsP or AlInAs emitter and an InAs base. For some devices, the emitter was graded to reduce strain energy. Current gain decreases for base doping levels above 1×10¹⁹ cm^-3; however, even with heavily Be-doped (~2 × 19 cm^-3) InAs bases, we obtain high β (>60). Numerical simulations provide an insight into the deleterious implications of Be diffusion, the degree of which can be manipulated through the MBE growth process. The work described in this talk was sponsored by DARPA/SPAWAR under contract number N66000-01-C-8033.
4:40 PM		SC-WeA-9 Lattice Constant Differences and Their Affect on the Surface Bonding of In₂O and Ga₂O on GaAs and InAs M.J. Hale, D.L. Winn, J.Z. Sexton, M. Passlack, A.C. Kummel (University of California, San Diego) The atomic bonding structures formed upon deposition of In₂O on GaAs(001)-c(2x8)/(2x4) and InAs(001)-c(2x8)/(2x4) were identified through scanning tunneling microscopy and density functional theory calculations. These results were compared to the surface bonding structure of Ga₂O deposited on GaAs(001)-c(2x8)/(2x4). It was found that In₂O forms three distinct bonding configurations on the GaAs(001)-c(2x8)/(2x4) surface: two different configurations which span the trough, and one which inserts into row arsenic dimer pairs. When In₂O was deposited on the InAs(001)-c(2x8)/(2x4) surface only the row As dimer pair site was observed. This effect is attributed to the In₂O molecule being large enough to span the 8.0Å trough of the GaAs(001)-c(2x8)/(2x4) surface, but too small to span the 8.6Å trough of InAs(001)-c(2x8)/(2x4). The In₂O bonding site on InAs(001)-c(2x8)/(2x4) is analogous to the binding site for Ga₂O on GaAs(001)-c(2x8)/(2x4), where Ga₂O is found to bond only in row arsenic dimer pairs. The change in lattice constant between InAs and GaAs is also found to have an affect on Fermi level pinning. Through experiments and DFT calculations, we have shown that In₂O pins GaAs whereas Ga₂O does not. DFT calculations also show that we should not expect In₂O to pin InAs. Our results show that a modest change in lattice constant can dramatically alter the adsorbate bonding and electronic structure via a change in site selectivity.
5:00 PM		SC-WeA-10 An Atomic Understanding of the Sub-Monolayer Interface formed Upon the Deposition of SiO on GaAs(001)-c(2x8)/(2x4)} D.L. Winn, M.J. Hale, J.Z. Sexton, M. Passlack, A.C. Kummel (University of California, San Diego) Scientists have been trying to develop a GaAs-based MOSFET device in an effort to reduce standby power and gate leakage. To achieve this it is important to understand the chemistry at the oxide/semiconductor interface. The interface formed upon deposition of SiO on GaAs(001)-c(2x8)/(2x4) was studied using STM, scanning tunneling spectroscopy (STS) and DFT. STM images show that SiO molecules bond Si end down in three distinct locations: into row and trough As dimers and between row As dimers. It was also observed that when SiO bonds between row As dimers, the two adjacent dimers along the [1bar10] direction each contain an inserted SiO molecule forming a chain of three molecules (triple site). STS measurements show that ~5% of a monolayer of SiO pins the Fermi level at mid gap. This is consistent with SiO absorbates withdrawing charge from surface As atoms, causing charge on the As atoms to grossly deviate from the charge on bulk As atoms. Multiple SiO sites were simulated using DFT, the lowest energy structure however was not observed in STM. An energy versus chemical potential plot was used to explain and identify the most stable bonding structures in the coverage studied with STM. This plot showed that at low SiO coverages the most stable sites were single sites but at higher coverages the most stable sites shifted to sites involving three SiO molecules including the triple site. This work demonstrates the key role of coverage in determining the most stable bonding structure. For oxides on semiconductors there are usually multiple nearly degenerate bonding configurations and the coverage is the key variable in determining the relative chemical potentials