Gallium nitride is the future of microwave power amps, GaAs has exceeded its half-life, you can quote us on that. More expensive in terms of dollars per die, GaN offers a path to much higher power densities and therefore cheaper dollars per Watt.
Breakdown voltages of 100 Volts are possible on GaN, versus 7-20 volts on comparable GaAs products. Now you can buy parts that are qualified up to 28 volts operation but you can goose them up to 48 volts to witness the full GaN Experience. Ancillary stuff like higher-voltage capacitors and resistors, and backside processes have been developed at some MMIC foundries, in order to participate in this new technology.
GaN devices are typically high-electron mobility transistors, you can think of it as a fancy version of a MESFET. GaN devices can either be discrete or monolithic.
Another niche application of GaN has appeared: robust low noise amplifiers. GaN can provide LNAs with great noise figures, which can withstand much higher power levels than GaAs LNAs (perhaps by a factor of 20 dB!) In future systems you can seriously consider eliminating a limiter in front of an LNA which will save money, reduce module size and further reduce noise figure by the loss of the k so that the US will maintain technological superiority in military programs for the next decade or two. The big DARPA program is called WBGS-II (for wide bandgap semiconductor), and the three teams are PAM-XIAMEN/Lockheed, Raytheon/Cree and Northrop Grumman. No further discussion will appear here, the data is ITAR restricted!
However, as much as the U.S. thinks that GaN technology will be prominent only in one country, it has spread to Europe Asia and even Canada. If you are considering the technology, be sure to ask the suppliers for reliability data and look it over carefully.
Why are native gallium nitride(GaN) wafers impractical? Recall that nitrogen is a gas at room temperature, while gallium is a solid… so how could the two both exist in the liquid state and be forced to solidify into a uniform crystal?
Substrates for GaN are either silicon carbide, sapphire, or silicon. Expensive alchemy is needed to align the GaN crystal onto these mismatched substrates, using molecular-beam epitaxy (MBE) or metal-organic chemical vapor deposition (MOCVD). Four-inch (100mm) SiC substrates are just becoming available for GaN-on-SiC, four inch GaN on silicon wafers are also available with a growth path toward six inch (150mm) and larger. Most MMIC processing lines can handle either 100 mm or 150mm wafers or both, there just isn’t a market that will drive toward 200 mm any time soon. Silicon wafers are dirt cheap ($10 for 200mm diameter) while silicon carbide wafers currently cost 100X more for only 100mm. Sapphire seems to have fallen by the wayside in the past few years.
Silicon carbide is an excellent heat sink, with thermal conductivity similar to the best metals (350 W-m/K around room temperature). Silicon is much lower (40 W/m-K at room temperature), so it doesn’t spread the heat as efficiently and thus for a given power density will result in higher channel temperatures.
If you want to create a MMIC instead of just a discrete device, silicon is at a huge disadvantage because in its most popular form it conducts, just like a semiconductor should! Thus if you were to use ordinary low resistivity silicon (LRS) and print microstrip transmission lines on it, the loss of the interconnects would exceed any gain you’d get out of the transistors, a colossal waste of time! In order to create a silicon MMIC, you can obtain high-resistivity silicon (HRS), which is tricked up to several hundred or even several thousand ohm-cm, which will add measurable loss to the T-lines but just maybe you can design a useful product. HRS is available in diameters up to six inches (150 mm), which potentially gives it a production cost advantage over SiC for the time being.
There’s more bad news for GaN MMICs on HRS: the uniformity of the substrate’s resistivity is imperfect, typically varying by an order of magnitude across the wafer. This will ultimately provide a wider gain variation on MMICs on GaN on silicon than GaN on SiC. Also, if you don’t watch out, the resistivity of the silicon will be reduced during wafer processing. And finally, around 200C, the high-resistivity property of the HRS substrate starts to degrade, so just when we have invented a semiconductor technology that can withstand 200C channel temperature, we have to back off to 175 to stay away from substrate conduction effects. But again, if you are only interested in discrete devices, consider the economy of GaN on silicon.
Maximum channel temperature
GaN can operate up 200C channel temperature (150C is the typically quoted limit of GaAs for 1,000,000 hours operation). Below 2 GHz, expect to see GaN used in base station applications, competing with silicon carbide technology. Higher frequency GaN products will be fielded by the military, with multiple suppliers reporting power amplifiers even up at millimeterwave frequencies.
Cree(GaN on SiC)
Nitronex (GaN on silicon)