Electron transport in GaN expitaxial layers

The electron transport mechanisms in MOCVD-grown GaN films have been investigated. Mobilities, carrier concentrations and resistivities were assessed by means of the temperature dependent Hall Effect, Differential Hall Effect, and resistivity measurements from 10 - 600 K. Unintentionally-doped and Si-doped films all showed n-type conduction with 300 K carrier densities and mobilities ranging from 5x1016 - 6x1018 cm-3 and 50 - 600 cm2/Vs respectively. The experimental findings cannot be analysed by assuming single carrier transport. A parallel conducting channel affects the bulk electron concentration and mobility from low temperature to high temperature. The electron concentration shows a minimum as the temperature is decreased and then starts increasing until it reaches a saturation point below 40 K, with mobilities decreasing faster than predicted by scattering theory. The behaviour of the experimental data over the entire temperature range could be satisfactorily accounted for on the basis of simultaneous conduction by two kinds of carriers (of the same type) in different energy bands and with different mobilities. These two bands have been taken as the conduction band and a donor impurity band. Using a two-band analysis in conjunction with Fermi-Dirac statistics our Hall data were successfully analysed and correct values of bulk electron concentration and mobility have been obtained with the corrected 300 K mobility being always higher than the uncorrected one. In an attempt to identify the effect of the interface region, we undertook plasma etching experiments, combined with the Hall effect, to isolate the interfacial region from the rest of the epilayer. Our results showed that multiple parallel conduction paths were present in the set of samples studied, and were associated with shallow impurity bands in the Si-doped region of the epilayer and/or in the oxygen-doped defective interfacial region. By applying a Differential Hall analysis we have managed to study the spatial distribution of electrons and deconvolve the properties of each of these channels. This allowed values of donor and acceptor concentrations as well as activation energies to be derived from the theoretical fitting of mobility and carrier concentration data.