Thermally stable, oxidation resistant capping technology for Ti/Al ohmic contacts to n-GaN

The intermetallic TiAl(3) has been used as a thermally stable cap for Ti/Al ohmic contacts to n-GaN. The electrical performance of the TiAl(3)-capped contact is nearly the same as that of a standard Ti/Al/Ni/Au contact processed on the same substrate, but the Ti/Al/TiAl(3) contact's performance is optimized at a much lower temperature. The Ti/Al/TiAl(3) contact achieved a lowest specific contact resistance (rho(c)) of 2.1x10(-5) Omega cm(2) following 1 min at 700 degreesC in flowing, oxygen-gettered ultrahigh purity (UHP) Ar. The Ti/Al/Ni/Au contact standard achieved a rho(c) of 1.8x10(-5) Omega cm(2) following a 15 s anneal at 900 degreesC in flowing, oxygen-gettered UHP Ar. The TiAl(3)-capped contact structure shows little sensitivity to the amount of oxygen in the annealing ambient for optimization, and we found that it could achieve a rho(c) of 1.1x10(-5) Omega cm(2) following 5 min at 600 degreesC in air. This performance is almost identical to that attained when the contact was annealed in oxygen-gettered UHP Ar and ordinary Ar. Anneals were extended to a total time of 20 min in the three ambient atmospheres, and the Ti/Al/TiAl(3) contact showed no significant difference in its performance. The fact that this contact structure can withstand optimization anneals in air suggests that it could be annealed alongside a Ni/Au contact to p-GaN in air and still achieve a low contact resistance. The performance of the TiAl(3)-capped bilayer was found to be stable following thermal aging for more than 100 h at 350 degreesC in air, which was also comparable to an optimally annealed Ti/Al/Ni/Au contact aged at the same time. The TiAl(3) material should be an oxidation cap solution for many other Ti/Al contact structures, almost regardless of the Ti:Al layer thickness ratio, since the TiAl(3) will be stable on the upper Al layer. Use of this cap eliminates the need to alter a previously optimized bilayer, thus it is a means of enhancing any existing Ti/Al bilayer contact's performance without necessitating the reoptimization of the layers to accommodate the cap. (C) 2002 American Institute of Physics.