As the result of recent rapid development in advanced device designs, the power density of GaN HEMTs has rocketed to >30 W/mm at C and X bands, a level exceeding what was predicted for diamond devices. While this is almost beyond what even the high thermal conductivity of the SiC substrates can handle, it is a good problem to have. However, the GaN HEMT has been performing poorly at millimeter-wave bands. A major hurdle is that the cut-off frequencies do not scale up well with the shrinkage of gate length. Traditionally, the electron velocity of III-V FETs overshoots as the gate length reduces to sub-micron scale, which overcomes the higher percentage of parasitic charging times as the intrinsic input capacitance reduces. This overshoot is not yet seen in GaN HEMTs. What is worse, the ft of these devices peaks at only 25-30% of the full-channel current and deteriorates severely at higher current levels. These problems are responsible for the measly 2-5 W/mm and 15-40% efficiency produced by GaN HEMTs at >30 GHz. So, what is limiting the speed of the GaN HEMT? Will its electron velocity ever overshoot? What will be the best power device at millimeter-wave frequencies. Please join the discussion on the future of ultra-high speed power electronics.
|Original language||English (US)|
|Number of pages||1|
|Journal||Device Research Conference - Conference Digest, DRC|
|State||Published - Dec 1 2004|
|Event||Device Research Conference - Conference Digest, 62nd DRC - Notre Dame, IN, United States|
Duration: Jun 21 2004 → Jun 23 2004
ASJC Scopus subject areas