(Invited) Effects of Annealing Capped (0001) Oriented GaN Samples

Changes in the topological, structural, electrical and optical properties produced by annealing (0001) oriented GaN homo- and hetero- epitaxial films that were protected by an AlN cap and annealed at a temperature, TA, ranging from 1150 to 1300 ºC for a time, t A, ranging from 1 to 8 min are discussed. The films were as-grown or Mg implanted unintentionally doped n-type, or they were Mg doped. For as-grown hetero-epitaxial films the step-flow growth patterns on the as-grown surface survive up to TA ≤ 1300 ºC and t A ≤ 2 min. Small angle grain boundaries parallel to the {1100} M-planes become more defined on the surface of the hetero-structure films as TA and/or t A increases so long as it retains its smooth step-flow growth pattern. Small hexagonal etch pits associated with threading edge dislocations decorate these boundaries suggesting that the grain boundaries are created by polygonization. These boundaries are present in the as-grown film, but they do evolve over time at different rates at different temperatures as is illustrated in the (00.2), (10.2), (30.2) x-ray rocking curves, (10.5) ω - 2θ scans, and reciprocal space maps. Similar studies of films grown on ammonothermal bulk single crystals show similar trends, but they are much more muted. All of these studies require an annealing cap to minimize the rate of preferential evaporation of nitrogen. The GaN surface must be clean and smooth, the cap must adhere well, and one must be able to preferentially etch it off after the anneal. A thin AlN film grown epitaxially by MOCVD at low temperatures combined with a thicker sputtered AlN film meets all of these requirements, but it only does so up TA = 1300 ºC for t A < 8 min. It fails at higher TA because the N vapor pressure at 1300 ºC is about equal to the yield stress of AlN. Our caps fail when t A > 8 min at TA 1300 ºC because even some N escapes with a good seal. When the cap does not seal as well, the GaN film decomposes first at the small angle grain boundaries. Even when the surface of the film is smooth, the electrical properties show that the anneal has altered the internal structure. The conductance above the voltage knee of Schotty diodes fabricated on them initially increases with TA and t A, but it begins to drop off at TA = 1200 or 1250 ºC suggesting that at lower TA some of the structural defects are annealed out thereby increasing the mobility, but at higher TA other types of defects, possibly nitrogen vacancy, NV, complexes are created. This explanation is consistent with the observation of the conductance below the voltage knee being larger at the larger TA because the tunneling current through the defects associated with the NV is larger. The reverse leakage current follows the same trend – it being larger at the larger TA. The CV characteristics of the diode suggests that this defect layer is mostly confined to the region near the surface as the net carrier concentration as measured at the end of the depletion layer does not change with TA until TA ≥ 1250 ºC when the defected area is at the end of the depletion layer at zero bias. Defects of a different nature are created by the implant process as determined by the increased breadth of the XRD curves and the creation of another peak. But much of that damage is annealed out as the peak widths return to almost their original breadth and the second peak disappears. However, they do not return to their original breadth. It is not clear whether this is due to not removing all of the implant damage or the damage created by the annealing process contributing to the increased breadth. We are now in the process of deciphering the more complicated electrical measurements made on the Mg implanted or doped samples, and expect to report on them at the meeting.