• Tompkins, R. P.
• Smith, J. R.
• Leach, J. H.
• Jones, K. A.
• Kirchner, K. W.
• Shahedipour-Sandvik, F.
• Leathersich, J. M.
• Udwary, K.
• Suvarna, P.

Abstract

<p>We have examined the performance of gallium nitride (GaN) high-power Schottky diodes fabricated on unintentionally doped (UID) metalorganic chemical vapor deposition (MOCVD) films grown simultaneously on four substrates ranging in threading dislocation density from 5 x 10(3) cm (- 2) to 10(10) cm (- 2). The substrates were an intentionally doped and a UID freestanding hydride vapor phase epitaxy substrate, an MOCVD GaN template grown on a sapphire wafer, and a bulk GaN substrate grown via an ammonothermal method. Capacitance-voltage (C-V) results showed the carrier concentration was similar to 2 x 10(16) cm(-3) for films grown on each of the four substrates. With that doping level, the theoretical breakdown voltage (V (b)) is similar to 1600 V. However, measured V (b) for the devices tested on each of the four substrates fell short of this value. Also, the breakdown voltages across each of the four substrates were not substantially different. This result was especially surprising for films grown on bulk GaN substrates, because of their superior crystal quality, as determined from their x-ray rocking curve widths. Simple probability calculations showed that most of the diodes tested on the bulk substrate did not cover a single threading dislocation. Although optimization of edge-termination schemes is likely to improve V (b), we believe that point defects, not threading dislocations, are the main reason for the reduced performance of these devices.</p>

Topics

• density
• impedance spectroscopy
• phase
• nitride
• dislocation
• chemical vapor deposition
• Gallium
• point defect