Invited Talk

GROWTH OF GaN(0001)A AND B BY MOLECULAR BEAM EPITAXY*

P.I. Cohen, R. Held, G. Nowak, B.E. Ishaug, A.M. Dabiran, I. Grzegory, and S. Porowski, University of Minnesota, Minneapolis, MN 55414

PCSI-26, San Diego, CA, January 17-21, 1999

Abstract

GaN growth on bulk GaN was compared to GaN growth on sapphire in order to remove the complications introduced by the large lattice mismatch of heteroepitaxy. For example on the bulk surface the diffraction pattern was much sharper than for GaN on sapphire. In addition on the bulk surfaces, larger, atomically smooth terraces could be observed, and mixtures of the two c-plane polarities were not present -- all of which simplified the interpretation of the measurements. The bulk surfaces were prepared either by mechanical polishing (GaN-A) or chemo-mechanically polishing (GaN-B). These were identified from the reconstructions. GaN films were grown using Ga from a Knudsen cell and ammonia .

At growth temperatures, reflection high-energy electron diffraction (RHEED) showed that two different surface terminations could be maintained on the B surface, while only one could be maintained on the A surface. We examined their role in the adsorption kinetics of Ga using desorption mass spectroscopy (DMS) and atomic force microscopy (AFM). For the DMS, a differentially pumped, mass spectrometer with a direct line of sight view of the sample surface, was used to measure the desorbed molecules vs time after starting and then stopping the incident Ga flux. The shape of measured signal vs time was different on the A and B surfaces, the shape depended on substrate temperature, and it was observed to change if there was an ammonia flux coincident on the sample. These "uptake'' curves were compared to a rate equation model of the growth. To obtain agreement with the data on the B surface we needed three main terms: (1) a fast desorption term, reducing the effective flux by as much as 50%, (2) a weakly adsorbed Ga state, and (3) a strongly bound Ga state. With these terms, agreement with the data could be obtained over a wide range of substrate temperature, flux, and surface termination. The B surfaces were also examined with AFM at different stages of this adsorption process. The surface termination had a large effect on the observed structure. On the bulk samples, regions corresponding to nitridation could be observed adjacent to step edges. The size of these regions could be varied by quenching the surface at different rates. These nitrided regions could be removed either by adding Ga or heating in vacuum. Locally, these nitrided regions had an island structure; according to the rate equation models, the surface could never be completely nitrided.

Intensity oscillations in the specular diffracted beam were observed on the B surface, but not the A. Growth was limited by decomposition at high temperatures, depending upon the incident Ga and ammonia fluxes, following mass action.

* Supported in part by the Office of Naval Research (N/N00014-97-1-0063) and by the National Science Foundation (DMR-9618656)


Back to my home page