JOURNAL OF MATERIALS SCIENCE: MATERIALS IN ELECTRONICS ! (1990) 75-78 Deep levels in GaAs/AIGaAs multi-quantum well structures A. ARBAOUI, B. TUCK, C. J. PAULL, M. HENINI* Department of Electrical and Electronic Engineering, and *Department of Physics, University of Nottingham, Nottingham, UK The study of the behaviour of deep traps in AIxGal_xAs/n-GaAs multi-quantum well struc- tures with three different well widths (1.7, 2.5 and 3.4 nm) has been performed using deep level transient spectroscopy. The measurements showed the presence of two closely spaced levels in each structure in the temperature range 160 to 270 K. It was found that as the GaAs well width increased, the apparent activation energies and the concentrations of both levels decreased. It is seen that in the 1.7 nm well samples, the lower temperature level is dominant, but in the 2.4 and 3.4 nm welt samples, both levels were comparable in concentration. 1. Introduction Since the pioneering work of Esaki and Tsu [1], there has been considerable interest in superlattices either as active layers in the device structures or as substrates for optoelectronics devices. For successful operation of both minority and majority carrier devices with superlattices, it is important to detect and control the presence of electrically active defects in the super- lattice structures. Previous studies have shown that by replacing the doped A1GaAs layer in the MODFET structure with a superlattice structure [2, 3], it is possible to reduce the troublesome persistent photo- conductivity effect associated with the DX centre in the A1GaAs alloy. Moreover, it has been reported that with regard to deep traps within the superlattices both the period magnitude and the number of periods involved may play an important role affecting the concentration of deep levels and the interpretation of their characteristics [4-6]. Thi s paper reports the results of an investigation into the behaviour of deep traps in A10.4Ga06As/ n-GaAs multi-quantum well structures using the deep level transient spectroscopy (DLTS) technique to obtain results as a function of GaAs well width. 2. Experimental procedure The samples were prepared by molecular beam epitaxy (MBE) on (1 00) oriented GaAs substrates grown by the liquid encapsulated Czochralski process. The GaAs substrates were heavily doped with silicon (2 x t0 ~ cm-3). Each sample consisted of a GaAs buffer layer grown on top of a conducting GaAs substrate, followed by a multi-quantum well structure (MQW) consisting of alternating layers of silicon- doped GaAs and undoped A1,Ga~ ~As (x = 0.4). A total of 60 periods of GaAs/A1,.Ga~_,As were grown. Fig. t shows the layer structure of the samples. Three samples were grown, with GaAs layer thicknesses of 3.4, 2.5 and 1.7 nm, respectively; the AI~Ga~_,As layer thickness of 29.7 nm was identical for all samples. The 0957-4522/90 $03.00 + .12 © 1990 Chapman and Hall Ltd. GaAs layers in all samples were doped to 10 ~scm ~. A total MQW thickness of 1.904 #m or greater was used in each sample to avoid depleting the substrate in the measurement process. Growth parameters were iden- tical for the three samples with the sole exception of the GaAs well layer thicknesses. For diode fabrication, ohmic contacts were first formed on the backside of n + silicon-doped GaAs substrate by evaporation of Ni-Ge-Au contacts and alloying at 400 ° C in an Ar/H2 atmosphere for 15 sec. Prior to the evaporation of Schottky barriers, surface cleaning was carried out on the semiconductor slices. The cleaning procedure was employed to remove as much surface contamination as possible. Schottky barrier diodes were formed by evaporating aluminium on the top of a sample in a vacuum of 10 -6 torr (1 torr = 133.322Pa). The experimental techniques used for electrical characterization of the Schottky barrier diodes were current-voltage (I-V) measurement, capacitance- voltage (C-V) measurement, and DLTS. The I-V characteristic gives a general description of the diode quality; C-V characterization is used to obtain the shallow dopant concentrations and the doping profile. In this paper the MQW supertattice structure will be viewed as a perturbed bulk crystal rather than as a series of junctions [1]. All diodes show a forward turn-on voltage of about 0.5 V and a reverse break- down voltage between 8 and 15V. C-V profiling con- firmed the doping level, but did not show the periodic variation of carrier concentration in any of the samples. A microcomputer-based DLTS system was used to measure deep-level parameters. The principle of the measurement and the experimental procedure have been reported elsewhere [7]. Typically the capacitance transients are recorded under isothermal conditions rather than during temperature scans using a fixed apparatus time constant. DLTS measurements were performed only on 75