Anablog

IR just made an announcement about its development of gallium nitride (GaN) transistors. Not included in the press release was whether they are HBT bipolar transistors (doubtful) or HEMT FET transistors (more likely considering this recent IR patent). Also not mentioned is whether the FETs are depletion mode, as is common, or enhancement mode, as the patent suggests. One thing is clear is that IR won’t be growing the transistors on sapphire, this is GaN on silicon or silicon carbide as the patent suggests, if they want to play in the low-cost power field as opposed to the high-cost RF transistor market. The patent calls out a bidirectional switch with multiple gates and a 100 times reduction in specific on-resistance for a 300-volt part. So far the best article about GaN FETS that I can find comes from the May 2002 post of Science Blog. A salient quote:

The bonding between nitrogen and gallium atoms in a unit cell of gallium nitride is very tight. The strength of the bond between the individual atoms determines the width of the gap between the valence and conduction bands of a semiconductor. The stronger the bonds, the larger the separation between the valence and the conduction bands, and the greater the band gap.

The larger the band gap, the more voltage that can be applied, and that is what is key for electronic devices. Power is the product of voltage and current. Gallium nitride enables a transistor that can handle not only high voltage but also high current. So the product--power--goes up between 10 and 100 times over silicon or gallium arsenide.

Here is a neat diagram from Gain Microwave that shows the advantages of GaN over silicon and GaAs:

Here is a Google patent search with IR as the assignee. And here is the salient diagram from the International Rectifier GaN patent mentioned above.

Other articles about GaN:

• Gallium nitride alternative to silicon for more powerful electronics
• Gallium Nitride roadmap
• Current Issues in GaN
• Let there be gallium nitride lite (from Australia, with better-than-US tech reporting evident)
• Gallium Nitride Power Transistors in Narrow and Wide Bandwidth Applications (webcast that asks for too much DirectX for me to watch)
• GaN essentials from Nitronex
• Gallium Nitride 2005 - Technology Status, Applications, and Market Forecasts report price: $ 3,950.

EDN has published an article about a student that developed a GaN MOSFET which is based of this: Alternative To Silicon Chip Invented By Student, but despite how much I would like to think a rural boy could revolutionize the transistor, it is obvious he worked under the direction of a professor and they have done an improvement but not necessarily invented a new transistor.

If you need a university, besides UCSB to do some GaN research:

• The Cambridge Centre for Gallium Nitride

And lest you think IR is the only company pursuing GaN transistors (not to mention all the companies doing GaN optoelectronics):

• TriQuint Semiconductor Reveals Gallium Nitride Products
•  NEC Develops High-Power Gallium Nitride Transistor Amplifier
• Toshiba Gallium Nitride Power FET
• Freescale Gallium Nitride
• Aethercomm Gallium Nitride Amplifiers
• Panasonic Develops a Gallium Nitride (GaN) Power Transistor with Ultra High Breakdown Voltage over 10000V
• http://www.veloxsemi.com/
• Velox press release 1200V 100A GaN on Si Transistors (pdf)
• Nitronex
• Fujitsu’s Technology for Gallium Nitride HEMT  
• Oki Starts Sample Provisioning of Gallium Nitride Power Transistor
• Raytheon Added to Research List for Gallium Nitride Semiconductors
• RF Micro Devices Gallium Nitride (GaN) High-Power Transistors

I will keep you posted on what these “devices” that IR alludes to in the press release, including silly facts like breakdown voltage, on resistance, gate charge, and temperature rating, all of which should be better than silicon. Don’t you miss the days when the press release was the price and datasheets of real parts you could use? Now we get a release about the plan for what the thing is supposed to do, followed by ten more regarding business partnerships and VC funding, but no parts. Knowing IR they will be following up with real parts pretty soon. It is obvious this was why they valued the company more than Vishay did. When I saw that the new CEO was an expert in acquisitions and mergers it was pretty plain to me that IR was in play, maybe now they will get a sweetheart deal from some giant multinational. This also explains why they sold off a lot of the silicon discrete business. IR makes a lot of money off other peoples FETs since they still hold patents on MOSFET technology and many companies pay IR to license the IP. International Rectifier bet the company once on HexFets and it looks like this is another bold move. IR is a great outfit and I wish them luck in this new endeavor.