Effects of Oxygen, Nitrogen, and Hydrogen Annealing on Mg Acceptors in GaN as Monitored by Electron Paramagnetic Resonance Spectroscopy, The

Journal of Electronic Materials, Jan 2005 by Matlock, D M, Zvanut, M E, Wang, Haiyan, DiMaio, Jeffrey R, Et al

Electron paramagnetic resonance (EPR) spectroscopy is used to study the unpassivated Mg-related acceptor in GaN films. As expected, the trends observed before and after O2, N^sub 2^, or forming-gas anneals at temperatures 850°C in O2 or N^sub 2^ permanently removes the signal, contrary to the results of conductivity measurements. Approximately 10^sup 19^ cm^sup -3^ Mg acceptors were detected in some GaN films grown by chemical vapor deposition (CVD) before acceptor activation, suggesting that it is possible to have electrically active Mg in as-grown CVD material.

Key words: Electron paramagnetic resonance (EPR), Mg doped, GaN, annealing, hydrogen, oxygen

INTRODUCTION

As progress toward developing white light-emitting diodes and high-frequency lasers moves forward, understanding the activation of p-type conduction in Mg-doped GaN becomes increasingly more critical. Early work showed that Mg-doped GaN grown by chemical vapor deposition (CVD) must be irradiated with low-energy electrons or annealed at temperatures greater than 700°C in either N^sub 2^ or O2 to reduce the resistivity to 1 ohm-cm.^sup 1-3^ On the other hand, films grown by molecular-beam epitaxy (MBE) exhibit p-type conductivity without any post-growth treatment.4 The results of secondary-ion mass spectroscopy (SIMS), infrared spectroscopy, and electrical measurements indicate that the acceptors in CVD GaN are rendered inactive by hydrogen, and that annealing in N^sub 2^ or O2 releases hydrogen from a Mgacceptor complex.^sup 1-3,5-8^ Annealing is typically performed at a temperature below 850°C because higher temperatures are known to create microscopic defects and trigger nitrogen desorption.9,10 Theoretical calculations based on infrared spectroscopy data suggest that hydrogen bonds to a nitrogen atom forming a Mg-N-H complex rather than a simple hydrogenmagnesium bond.11,12 In addition to the hydrogenrelated work, some studies suggest that other centers limit p-type conductivity by compensating the Mg acceptor.13,14

Electron paramagnetic resonance (EPR) spectroscopy offers a method for studying the acceptor in GaN by focusing on the Mg impurity itself. Typically, EPR measurements provide a detailed microscopic picture of an impurity in a semiconductor, but the necessary spectroscopic features required to identify the defect structure of Mg in GaN are not resolved in the EPR spectrum. Rather, Glaser and co-workers concluded that the EPR signal observed exclusively in Mg-doped GaN is related to the acceptor by correlating the signal intensity in a variety of activated CVD films and as-grown MBE layers with the Mg concentration and hole density.15,16 Furthermore, others have demonstrated that the EPR spectrum changes as the sample is annealed in N^sub 2^ and H^sub 2^ as expected for the Mg acceptor complex.13,17,18 In this work, we offer additional support for the relationship between the EPR signal and the p-type acceptor by examining the temperature dependence of the activation process in an O2 or N^sub 2^ ambient for a variety of CVD samples, as well as MBE films heat-treated in nitrogen. Then, based on the established relationship between the EPR signal and hole density, we draw three conclusions relevant to p-type GaN. (1) The Mg signal may be permanently removed by annealing at temperatures greater than 850°C, independent of the method by which the film was grown. Furthermore, the hightemperature annealed films likely remain conducting despite the absence of the Mg acceptor. (2) The most effective temperatures for re-passivation of the acceptors by hydrogen are similar to the temperature typically used for intentional activation in N^sub 2^. (3) The CVD GaN:Mg may contain electrically active Mg acceptors, but additional defects compensate the holes.

EXPERIMENTAL DETAILS

The EPR experiments were performed on 0.7-2.3μm-thick, Mg-doped GaN epitaxial films grown at four different sources deposited by either orgariometallic chemical vapor deposition (OMCVD) or MBE onto either a-face sapphire substrates or the Si c-face of n-type 6H-SiC. A 0.02-0.125-μm-thick AlN buffer layer was grown on all substrates prior to nitride deposition. Table I provides a description of the samples used for these experiments. Two GaN/SiC wafers were received from North Carolina State University (Raleigh, NC). One half of each was given a rapid thermal anneal (RTA) for 30 sec at 800°C under flowing N^sub 2^; the remaining halves were left unannealed. The Mg concentration is estimated from the molar flow rate of the Cp2Mg precursor delivered by the hydrogen carrier gas. The samples from the Air Force Research Laboratory at Wright Patterson Air Force Base (WPAFB, OH) were grown by a MBE process using ammonia as described elsewhere.19 The carrier concentration was measured using temperature-dependent Hall measurements. Details of the OMCVD growth and additional characterization for the GaN/sapphire samples received from the Naval Research Laboratory (Washington DC) may be found in Refs. 15 and 20.