Improvement of InP-GaAs-Si Quality by Thermal-Cycle Growth
- PDF / 2,484,882 Bytes
- 7 Pages / 420.48 x 639 pts Page_size
- 76 Downloads / 234 Views
IMPROVEMENT OF InP-GaAs-Si QUALITY BY THERMAL-CYCLE GROWTH S.M. Vernon, C.J. Keavney, E.D. Gagnon, N.H. Karam, Spire Corporation, Bedford, MA; M.M. Al-Jassim, Solar Energy Research Institute, Golden, CO; N.M. Haegel, V.P. Mazzi, Material Science Dept., UCLA; and C.R. Wie, Electrical and Computer Engineering Dept., SUNY at Buffalo.
ABSTRACT Single-crystal films of InP have been deposited on GaAs, GaAs-coated Si, and InP substrates by metalorganic chemical vapor deposition (MOCVD). Defect-reduction schemes involving various thermal annealing recipes have been developed and characterized. Material quality has been assessed by a variety of methods including transmission electron microscopy, X-ray rocking curve analysis, low-temperature photoluminescence, lifetime measurements, Hall-effect measurements, electrochemical profiling, and Nomarski microscopy. The use of either a thermal-cycle-growth or a thermal-cycle-annealing process leads to heteroepitaxial InP film quality which is significantly improved over that of its as-grown state, with the thermal-cycle growth appearing to be the more effective technique.
INTRODUCTION InP-on-Si and InP-on-GaAs structures have potentially significant applications in the field of high-speed communications because such heteroepitaxial composites may enable the fabrication of monolithic optoelectronic integrated circuits. Wafers of InP-on-Si may be more well suited to opticalfiber communications than are GaAs-on-Si, since InP-based optical components [utilizing GaInAs(P)] can be designed to match the optimum wavelength regions of 1.3 and 1.55 pm. In order for these minority-carrier devices to function effectively, high-quality, low-defect, long-carrier-lifetime InP must be grown in this lattice-mismatched configuration. In the GaAs-on-Si field, one technique that has shown a strong ability to reduce defects is the thermal-cycle growth (TCG) process, in which a repetitive growth, anneal, cooldown procedure is carried out, typically inside an MOCVD reactor [1]. The TCG of InP is more difficult than that of GaAs because the dissociation pressure of P over InP is much higher (by three orders of magnitude) than that of As over GaAs.
EXPERIMENTAL PROCEDURES The structures studied consist of undoped InP films, 3-4 pm in thickness, deposited side-by-side onto GaAs, InP, and GaAs-coated Si substrates; the GaAs layer thickness on the Si substrates is 1-2 pm, an optimum range determined in previous experiments. All layers were grown by atmosphericpressure MOCVD in a SPI-MO CVD 450 system; this reactor has a batch capacity of five two-inch-diameter wafers arranged in a vertical, rotating, barrel-geometry configuration. The details of the GaAs-on-Si growth have been described previously (2] and involve a three-step (bake out, nucleation, deposition) growth process. The InP growth is carried out at 600*C using PH 3 , either trimethylindium or ethyldimethylindium, and a palladiumpurified hydrogen carrier gas. The growth of the InP in some samples utilizes a TCG process which involves several repet
Data Loading...
