Atomically Abrupt and Smooth Heterointerfaces: An Optical Investigation
- PDF / 337,255 Bytes
- 6 Pages / 420.48 x 639 pts Page_size
- 45 Downloads / 160 Views
ATOMICALLY ABRUPT AND SMOOTH HETEROINTERFACES: AN OPTICAL INVESTIGATION COLIN A. WARWICK*, WILLIAM Y. JAN*, ABBAS OURMAZD*, TIMOTHY D. HARRIS** AND JUDRGEN CHRISTEN*** *AT&T Bell Labs, Crawfords Corner Road, Holmdel, NJ 07733. **AT&T Bell Laboratories, 600 Mountain Avenue, Murray Hill, NJ, 07974 ***Technische UniversitAt Berlin, FRG
ABSTRACT Luminescence spectra from quantum wells are routinely interpreted in terms of atomically smooth and atomically abrupt interfaces. Here we show that this interpretation is inconsistent with photoluminescence, photoluminescence excitation, and quantitative microscopic (chemical lattice imaging) results. We argue that the discussion of interfacial roughness in terms of "an island size" is too naive. A full characterization of an interface requires the description of a "roughness spectrum", specifying the amplitude of the interfacial corrugation vs corrugation wavelength over the relevant length scale. INTRODUCTION Photoluminescence (PL), scanning cathodoluminescence, and PL excitation (PLE) spectra from high quality single quantum wells can consist of two or three sharp peaks, separated in energy by a few meV. It is routinely claimed that these peaks occur at energies corresponding to excitonic recombination (or absorption for PLE) in quantum wells exactly an integral number of monolayers (MLs) thick, and that the spacings between the peaks indicate abrupt ML changes in the well thickness [1-6]. This interpretation has become fundamental to current understanding of the effect of structure on luminescence properties. So much so, that it is now the guiding principle in the growth of interfaces of supposedly ultimate atomic perfection. The correct interpretation of the luminescence data is thus of crucial scientific and technological importance. The prevailing interpretation of luminescence data advocates the existence of atomically perfect (i.e. atomically smooth and abrupt) interfaces, which change position abruptly by exactly 1 ML. On this basis, a quantum well of nominal thickness n in fact consists of regions (islands), within each of which the thickness is exactly (n -1), n, or (n + 1) MLs, between which the interfacial position changes abruptly by 1 ML. These islands have been claimed to be as large as 10 pm in diameter [6], but are generally thought to lie in the micron range [4], and in any case to be much larger than the exciton diameter (-15 nm). On this model, each PL (PLE) peak arises from free exciton recombination (absorption) within an island, over which the quantum well is an integral number of layers thick. Thus, the peak separations must correspond to the difference in the energies of excitons that recombine in regions of the well differing in thickness by exactly 1 ML. In practice, the splittings rarely correspond exactly to ML changes in well width. Departures from "ML values" are generally ascribed to experimental uncertainties in determining the peak positions, to fluctuations in the composition of the material, to impurities [4], or to exotic configurations o
Data Loading...
