Characterisation of gallium arsenide using deep level transient spectroscopy.

The origins and characteristic parameters of levels occurring deep in the band gap of semiconductors have been summarised. The effect of these levels on capacitance and carrier concentration measurements of metal-semiconductor diodes has been outlined. In particular the technique of Deep Level Transient Spectroscopy, which is based on these effects, has been explained and several developments which extend the scope of the basic technique have been described. A comparison with other techniques for the investigation of deep levels has been made and the experimental implementation of DLTS related. Its application to several classes of deep levels in GaAs materials and devices and the experimental results obtained have been reported and discussed. The effects of light ion damage as typified by proton bombardment have been investigated in detail and the similarities to both electron and heavier dopant ion damage highlighted. The isolation of IMPATT diodes on a single chip by means of bombardment with high doses of protons has been investigated and found not to cause degradation of these devices. Deep levels characteristically found in the very thin n type layers of FET structures grown on semi-insulating substrates have also been examined. The problems caused by outdiffusion of chromium from the substrates and its insensitivity to detection by DLTS have been explained. Problems related to ion implantation of GaAs for FET fabrication and subsequent thermal annealing of the damage after encapsulation in silicon nitride to prevent surface decomposition have been investigated. The newer technique of electron beam annealing which eliminates the need for silicon nitride deposition has also been investigated Finally DLTS has been used to search for possible recombination centres associated with the degradation of light output in GaAlAs double heterostructure stripe geometry lasers. Trapping levels, including the donor related DX centre were observed and identified but were not seen to increase in concentration during accelerated lifetesting of the lasers. Possible reasons for the lack of detection of degradation related recombination centres have been discussed. In conclusion the usefulness of DLTS as an assessment technique has been analysed