In comparison to silicon and GaAs, GaN compares favourably in terms of power density and power levels, but it is not without its own technical limitations. Noting that GaN power transistors are now capable of power density >10 W/mm² and power levels in excess of 500W, Douglas H. Reep, PhD, senior director of research at TriQuint Infrastructure and Defense Products says, “In a theoretical sense, the technical limitations on GaN are strictly the fundamental material properties and our creativity utilising them.”
Reep suggests that the most important factor to consider with GaN is the relationship between transistor speed and operating voltage, as captured by Johnson's figure of merit. He notes that in this type of comparison, GaN demonstrates an order of magnitude advantage over GaAs, and two orders of magnitude over silicon. He adds that R&D for GaN is frequently focused on thermal management at the semiconductor and package level.
With regard to high voltage parts, GaN Systems recently announced five normally-off 650V GaN transistors optimised for high-speed system design. These 650V devices have reverse current capability, zero reverse recovery charge, and source-sense. (Earlier this year, the company announced 100V GaN power transistors, as well.)
Current technical limits involve pushing operating voltages higher while maintaining reliability, according to Tim Boles, distinguished technology fellow, MACOM (see MACOM commits to GaN at IMS2014). Boles articulates a desire to try to push to higher voltage bias points in order to achieve higher power added efficiency (PAE) and increased power density. “PAE in excess of 70% at 2.5 to 3.5 GHz can reasonably be achieved,” he says. “It is expected that higher bias voltages will improve this parameter.”
Girvan Patterson, president of GaN Systems, agrees that voltage breakdown is key for GaN. He reports that by using GaN on SiC, they have demonstrated more than 2000V in the lab. However, he points out that with today’s GaN on silicon technology, the breakdown voltage is