DX centres are deep levels associated with donors in III-V semiconductors. As these deep levels disturb the band-gap structure of the semiconductor, they are unwanted. This Action aimed to understand the physics and control of these deep levels in high band-gap III-V semiconductors. The structural, optical and electrical characteristics of DX centres, as well as their effects on various device structures, were studied in GaAs and AlGaAs alloys. The ultimate goal was to provide guidelines for materials anddopants selection and device structure design that avoids or minimises DX centre effects. It is expected that the knowledge obtained about AlGaAs in this respect will be applicable to other III-V and compound semiconductors. .
High Bandgap III-V (gallium arsenic like) materials, especially aluminium gallium arsenic alloys, were the focus of a study into n-type dopant behaviour in semiconductor materials and devices. The electronic structure and the electrical and optical properties structure and the electrical and optical properties of these materials doped with usual and more exotic donors were systematically examined. Processing conditions and device structures that avoid or minimize 'DX centre' effects were defined.
Tellurium, tin and selenium donors have been characterized and their advantages over silicon for device behaviour at low temperatures confirmed.
The electron capture kinetics have been studied by both capacitance and magnetotransport techniques, clarifying the origin of the strong nonexponential processes. There is evidence that the donor local environment affects the energy level structure of the DX centres. The radiative properties of the DX centres are being investigated. The presence of intrinsic point defects in aluminium gallium arsenic has been analysed to understand the origin of infrared emission in n-type aluminium gallium arsenic alloys. The near bandgap emissions have been probed by piezophotoluminescence techniques.
By magnetotransport experiments in aluminium gallium arsenic and gallium arsenic phosphorus under hydrostatic pressure, it has been shown the tellurium introduces a DX centre in gallium arsenic and gallium arsenic phosphorus that is more than 100 meV above the corresponding related and tin related DX centres.
Magnetic resonance experiments provided breakthroughs. Results on the tin, tellurium and silicon donors in aluminium gallium arsenic layers gave evidence for a diamagnetic DX ground state. Low temperature photoexcitation process are rise to the formation of paramagnetic donor states.
In order to minimize the effects of the DX centres in devices, guidelines for the selection of donor species have been presented favouring the use of group VI donors. The growth of aluminium gallium arsenic on silicon substrates indicated that the biaxial strain achieved did not induce significant changes in the DX behaviour. The benefits of using gamma doping in high electron mobility transistor (HEMT) structures received special attention. Indium aluminium arsenic and indium gallium phosphates doped with silicon were studied in detail due to their technological importance. No DX centres were detected for compositions below tellurium 60% or phosphorus 60%, repectively.
Because the DX state is intrinsic to n-type doping, no real way to suppress DX centres in aluminium gallium arsenic has been found.
APPROACH AND METHODS
A selected set of samples that include variations of the Al fraction in AlGaAs, doping level and donor chemical species have been defined and fabricated by metal organic vapour phase epitaxy (MOVPE) and molecular beam epitaxy (MBE) techniques. Si, Sn, Se,Te, and Pb donors have been selected as dopants.
A variety of structural, optical, magnetotransport and space charge spectroscopy techniques have been systematically applied to the set of samples. A theoretical modelling effort is also included.
In a second stage, neutron transmutation experiments, superlattice structures, double-doping, delta-doping, atomic layer epitaxy, strain effects, and high flux donor incorporation was planned. Basic DX microstructure knowledge was obtained, while donors, materials, doping procedures and device structures was also compared for DX minimisation.
PROGRESS AND RESULTS
-Te, Sn and Se donors have been characterised to a large extent, and their advantages over Si for device behaviour at low temperatures confirmed. HEMT structures using Se have shown a much reduced III-V collapse and persistent photoconductivity effect at 77 K as compared to Si-doped devices. In GaAs doped with Te, resonant DX states lie at significantly higher energy levels than those due to Si and Sn.
-The electron capture kinetics have been studied by both capacitance and magnetotransport techniques, clarifying the origin of the strong non-exponential processes. There is evidence that the donor local environment may affect the energy level structureof the DX centres. The subject of the radiative properties of the DX centres is being fully investigated. The presence of intrinsic point defects in AlGaAs, claimed to be As antisites, has been analysed to understand the origin of infrared emission in n-type AlGaAs alloys. The near-bandgap emissions have been probed by piezo-photoluminescence techniques.
-By magnetotransport experiments in AlGaAs and GaAsP under hydrostatic pressure, it has been shown the Te introduces a DX center in GaAs and GaAsP that is more than 100 meV above the corresponding Si and Sn-related DX centers. This result has practical i mplications in achieving high doping levels in GaAs. By transport analysis in Si- and Sn-codoped AlGaAs, strong support for the negative-U, DX- model has also been obtained.
-Magnetic resonance experiments provided breakthroughs. By independent techniques, results on the Sn, Te and Si donors in AlGaAs layers gave strong evidence for a diamagnetic DX ground state, in agreement with negative-U models. Low-temperature photoexci tation process gives rise to the formation of paramagnetic donor states.
-In order to minimise the effects of the DX centers in devices, guidelines for the selection of donor species have been presented favouring the use of group VI donors. In alloying has not been shown to be really helpful. It was shown that Pb is not an al ternative. The growth of AlGaAs on Si substrates indicated that the biaxial strain achieved did not induce significant changes in the DX behaviour. The benefits of using gamma-doping in HEMT structures received special attention. InAlAs and InGaP doped with Si were studied in detail due to their technological importance. No DX centers were detected for compositions below Al 60% or P 60%, respectively.
-The results achieved in this Action have contributed to significantly advancing the understanding of the behaviour of donors in III-V compounds. However, this behaviour has shown to be even more complex than it was initially thought. Donor atoms introdu ce shallow EM states, thought to correspond to pure substitutional behaviour, and the DX states, that are understood to be due to a distorted position when filled. Finally, because the DX state is intrinsic to n-type doping, no real way to suppress DX centers in AlGaAs has been found.
Because of the wide use of n-type regions in III-V device technology, the understanding of n-type dopant behaviour, and the corresponding understanding and control of DX centres, will have a profound impact on material doping procedures and device design.Knowledge that makes possible better alloy composition and device design will help minimise the deleterious effects of deep donors in III-V optoelectronics, high-speed devices and ICs.
221 00 Lund