Comparison of InGaSb/InAs superlattice structures grown by MBE on GaSb, GaAs, and compliant GaAs substrates

Journal of Electronic Materials, Jul 2000 by Tomich, D H, Eyink, K G, Grazulis, L, Brown, G L, Et al

This paper contains the characterization results for indium arsenide/indium gallium antimonide (InAs/InGaSb) superlattices (SL) that were grown by molecular beam epitaxy (MBE) on standard gallium arsenide (GaAs), standard GaSb, and compliant GaAs substrates. The atomic force microscopy (AFM) images, peak to valley (P-V) measurement, and surface roughness (RMS) measurements are reported for each sample. For the 5 jim x 5 pm images, the PV heights and RMS measurements were 37 A and 17 A, 12 A and 2 A, and 10 A and 1.8 A for the standard GaAs, standard GaSb, and compliant GaAs respectively. The high resolution x-ray diffraction (HRY-RD) analysis found different 011, order SL peak to GaSb peak spacings for the structures grown on the different substrates. These peak separations are consistent with different residual strain states within the SL structures. Depending on the constants used to determine the relative shift of the valance and conduction bands as a function of strain for the individual layers, the change in the InAs conduction band to InGaSb valance band spacing could range from 7 meV to -47 meV for a lattice constant of 6.1532 A. The cutoff wavelength for the SL structure on the compliant GaAs, control GaSb, and control GaAs was 13.9 gm, 11 gm, and no significant response, respectively. This difference in cutoff wavelength corresponds to approximately a -23 meV change in the optical gap of the SL on the compliant GaAs substrate compared to the same SL on the control GaSb substrate.

Key words: InGaSb/InAs, MBE, compliant GaAs substrates

INTRODUCTION

The InGaSb and InAs material system is a promising technology that may enable the creation of long and very long wavelength detectors. Theoretical calculations predict that these detectors could be created using strained layer superlattices of InGaSb and InAs.1,2 These structures have been previously designed and experimentally tested to operate with photoresponse cut-off wavelength of 10 gm.3

Currently, these structures are grown on GaSb or GaAs substrates. The GaSb substrates have a lattice constant very close to that of the SL structure. This allows the SL structure to have a low density of threading dislocations. Unfortunately, the GaSb substrates are not transparent at the desired operational wavelength. This prevents the structures grown on GaSb structures from being backside illuminated without thinning. In addition, the GaSb substrates are expensive, vary between manufacturers, and highly p-type conductive. This conduction reduces the detectivity of the SL structure by increasing the background noise. The more mature GaAs substrates can be used which are cheaper, more reproducible, and semi-insulating. However, the large lattice mismatch between the GaAs substrate and the SL structure results in a high density of threading dislocations. Numerous buffer growth techniques have been attempted to eliminate the threading dislocations." While some of these attempts have reduced the number the dislocations, the density of the threading dislocations has remained substantial. Due to these dislocations, SL structures grown on GaAs substrates typically exhibit very poor photoresponse.

Recently, the ability to grown lattice mismatched antimonides on compliant substrates without threading dislocations has been demonstrated.7,8 A SL structure grown on a compliant substrate should attain the free standing lattice constant of the SL. This free standing lattice constant can be calculated from the thickness of each layer within the SL period. This change in strain states is important in controlling the cutoff wavelength of the SL photoconductor. The new lattice constant when larger than that of GaSb results in less strain for the InGaSb layer and more strain for the InAs layer. Conversely, when the new lattice constant is smaller than that of GaSb there is more strain in the InGaSb layer and less strain for the InAs layer. This change in strain is critical since the strain influences the band offsets.9

EXPERIMENTAL

The InAs/InGaSb SL structures were designed to be photoconductors with a 10 tm cut-off. The SL structure with 100 repetitions consisted of 1.6 A of InSb, 42 A of InAs, 1.6 A of InSb, and 15.7 A of In.,,Gao.7?Sb for a total period of 60.9 A. The calculated free standing lattice constant of this SL is 6.1128 A. These structures were grown in a Riber 32 MBE system, equipped with EPI valued arsenic, valued phosphorus, and antimony cracker effusion cells. Three different samples were used in this experiment. A semi-insulating GaAs (001) substrate was used for comparison (#2483). A p-type GaSb (001) substrate was used for comparison (#2228). A GaSb buffer was grown prior to the SL on both the control GaAs substrate and the compliant GaAs substrate. The compliant substrate used in run #2491 was prepared as previously discussed.10 Briefly, the compliant substrates were bonded at temperature and pressure with a primary flat misorientation of approximately 450. The compliant substrate used a In.Gal_-As layer with a target thickness of 20 A. The HRXRD characterization on all three samples was performed using a five crystal Philips MRD system. The HRXRD models for the SL structure was calculated using a full dynamical simulation package. The AFM images were generated using a Park CP AFM system operating in tapping mode. For comparison, 5 x 5, 10 x 10, 20 x 20, and 50 x 50 micron images were taken for each sample. The peak to valley height and surface roughness (rms) was found from average line scan measurements. Cross-sectional transmission electron microscopy (TEM) was used to examine the structure of the individual SL layers and interfaces. The samples were prepared by ion milling. The images were taken using a Philips CM 200 FEG TEM operating at 200 kV. The samples were imaged under conventional brightfield conditions with the sample oriented in the twobeam condition so as to excite the (000) and (200) reflections. The samples were mounted to a copper heatsink and attached to the cold finger of a liquid nitrogen cryostat equipped with a temperature controller. Spectral response was measured with a Fourier transform infrared (FTIR) spectrometer. The samples were illuminated through the front side at normal incidence.