Recent Developments in Mercury Cadmium Telluride IR Detector Technology

Tuesday, October 13, 2015: 09:20
Curtis A (Hyatt Regency)
J. Antoszewski, N. D. Akhavan, G. Umana-Membreno, R. Gu, W. Lei (The University of Western Australia), and L. Faraone (The University of Western Australia)
In spite of many attempts to find better material for Infrared Detectors, the II-VI compound HgCdTe still dominates the high performance end of the market. Since its invention in 1959, HgCdTe technology evolved from bulk crystal growth and crude single infrared detectors to very sophisticated, epitaxially grown material and large format IR imaging arrays of photovoltaic devices. Today, the applications driven research and development in this field are directed towards high pixel density beyond diffraction limit, high yield, reduced cooling and multiband and hyperspectral operation. In order to meet these goals there is an urgent need for still better quality HgCdTe material and new band engineered devices with reduced dark current and noise levels.

In terms of HgCdTe material, the most challenging is development of large area, high quality epi-ready substrates allowing for growth of HgCdTe with low dislocation defect density – a major source of dark current and noise. Traditionally, lattice matched CdZnTe with 4% Zn content has been used. However, the cost of CdZnTe substrates is prohibitive, their size is limited, and hence not adequate for today’s large size detector arrays. The readily available large size Si and GaAs substrates have been investigated for some time but due to large lattice mismatch (19% and 14.3% respectively) few microns thick buffer layers of CdTe are needed to reduce the high lattice dislocations density in HgCdTe layers. But even with these CdTe buffer layers the best reported dislocation density in HgCdTe is still about two orders of magnitude higher than in HgCdTe epi-layers grown directly on lattice matched CdZnTe. Recently low-doped large epi-ready GaSb substrates become available and have been investigated as an alternative for HgCdTe growth. With relatively low lattice mismatch of only 6.1%, these substrates require much thinner buffer layer, and, as our studies indicate, even un-optimised growth results with CdTe buffer layer quality matching the best reported results for CdTe buffers grown on GaAs, consequently leaving much room for improvement.

In parallel with development of substrates a large effort is directed towards new detector device structures with the main goals focused on reduction of dark current and noise levels, as well as simplified processing leading to higher yield. In recent years so called barrier structures attracted much attention within HgCdTe research community. The concept based on the device consisting of the n-type absorber, undoped barrier, and thin n-type contact (nBn) comes from III-V technology where it was successfully implemented in InAs/GaSb system. Theoretically device like this does not have built in voltage region (reduced dark current and noise) and if externally biased allows for a free flow of minority photo-electrons while blocking majority photo-holes. Implementation of this concept in HgCdTe material system is not straightforward as its band structure is less favourable than that of InAs/GaSb system resulting in a small barrier in the valence band as well. However, this undesirable effect may be compensated by superior HgCdTe electron mobility and photo-generated carrier lifetime. Moreover, few promising concepts for elimination of the valence band barrier in HgCdTe nBn structure have been proposed. They are based on simultaneous modulation of material composition and doping within the barrier, or alternatively, the growth of the HgTe/CdTe superlattice barrier with energy structure optimised by band engineering.