If there was one clear message to emerge from the recent Wide Bandgap Electronics Technology Workshop*, it was that over the next five years a major investment is needed to push the US wide bandgap materials effort from the laboratory into systems for military and commercial applications. So say the policy makers and resource sponsors, wide bandgap (WBG) materials researchers, devices and system designers who convened in Columbia last December. But who will pay? Material of the Future Wide bandgap materials such as SiC and GaN exhibit many highly desirable properties: the combination of the large band gap, high breakdown voltage (up to 150 V), good transport properties and ability to form high quality heterostructures in AlGaN/GaN make WBG materials ideal candidates for high power, high temperature applications. It is this robust nature (i.e. temperature resistant/radiation hard) that has the military so fired up, particularly for high power amplifiers, which they believe are vital to enabling systems that maintain a competitive edge in a future theatre of war. The defense industry wants WBG devices for new high power, more compact systems for electronic warfare, missile seeking and higher power radar systems. Radars should have increased range and functionality for no increase in radar footprint, a point which is very important for warships. Similarly, very low noise robust GaN-based LNAs will help combat the effects of stray signals and jamming in communications, and nitride-based sensors could be used by ground forces to detect chemical or biological attack. In its wish list, the airforce would like high bandwidth down-links on its unmanned aeronautical vehicles (UAVs) used for reconnaissance missions. But if WBG materials are not to remain a future technology, a coordinated R&D effort will be needed of similar magnitude to the GaAs MIMIC (Microwave and Millimeter Wave Monolithic IC) program set up in 1987 to develop GaAs technology. Over the 8 years it ran, this $530 million program was responsible for much of today s GaAs IC manufacturing infrastructure. See . According to John Zolper, Program Officer for high power devices in the Electronics Division of the Office of Naval Research (ONR), an additional investment of approximately $50 million per year over 5 years is needed to establish a manufacturing base for WBG microwave materials, devices, and circuits for specific applications. This sum is less than was previously invested by the DoD through DARPA (Defense Advanced Research Projects Agency) under the GaAs MIMIC program. "We think we can live with a lower level of funding [than the MIMIC program] because we can leverage the expertise established in those earlier programs in areas such as materials characterization, electromagnetic design, and processing," says Zolper. "The investment is needed because the technologies that we have now, such as GaAs MMICs, can no longer meet the increasing system requirements. If you look at it from the perspective of a new destroyer, which costs $12 billion, or a new aircraft carrier, which is even more expensive, investing now in certain key technologies is justifiable for these new platforms. Such platforms have a service life of 30 to 50 years, so the technology investment made today will pay dividends for many years. It is the system pull that establishes the technology investment." Still Too Immature? In fact, defense platforms such as the US Navy s DD21 destroyers are now entering the planning stages for service in 2012, and if WBG technology is going to make it into new military system architectures, designers will need to be convinced that this technology is the best solution over the next five years. But is GaN and SiC ready? Most participants agreed that many problems still remain to be solved at both materials and device level before WBG devices find their way into systems. Not least is how these new materials, which have much higher power densities, will be thermally managed and packaged in new system architectures. According to HRL s Dave Grider, at the system level technology needs to be "transparent" to the designer, who should decide on a GaN MMIC device simply by checking specification sheets. "When a system designer looks at a III-V RF circuit, he expects to see a MMIC because that is what he can buy in GaAs. To sell the technology to system users, it is always better to sell the device performance rather than what it s made of." At the device level, reliability and yield still need to be understood. Materials issues such as epitaxial layer uniformity and reproducibility also need considerable investigation. Phenomena that limit RF performance, such as gain compression and gate lagan effect caused by traps found on the surface and in the highly defective GaN material that delay the response between the gate bias and current in the channel of a FETneed to be correlated with material defects and resolved. Substrates Many of these aspects are intimately tied in with substrates, the lack of which for GaN is viewed as one of the most serious obstacles to the commercialization of wide bandgap microelectronic devices. Currently GaN is grown on semi-insulating SiC wafers up to 2-inch in diameter. And while some progress is being made with free-standing GaN substrates grown by HVPE or PVD, a GaN substrate of sufficient diameter appears some years off. By comparison, the GaAs industry now achieves economies of scale with 4- and 6-inch substrates. Litton Airtron s Tom Anderson believes the current drive for larger diameter wide bandgap materials is for compatibility with existing 3- and 4-inch device fabrication lines, rather than reduced costs. "Improvements in material quality and cost are normally associated with an increase in substrate diameter," says Anderson. "The opposite is actually true, and initially, it is always more expensive per unit area to produce a larger diameter crystal and much harder to produce lower densities of defects with the scaled-up diameter. The commercial driver for larger diameters is lower per-die costs and higher per-wafer device yields. Lower device costs expand the market, and economies of scale drive down the substrate cost." SiC substrate research also suffers from the lack of commercial "pull" from the optoelectronics device manufacturers to develop larger wafer diameters. For FETs, SiC is the ideal substrate material because of its excellent thermal conductivity helps to dissipate heat. By contrast, LEDs are small area, commodity devices, and also seem to be miraculously insensitive to large numbers of defects (1010 dislocations per cm2) caused by lattice mismatch. Hence, for optoelectronics inexpensive sapphire substrates suffice in most cases. Some believed that electronic grade SiC is also just far too expensive at present. TRW s Mike Wojtowicz noted that cost, quality and availability of substrates are factors that could seriously impede commercialization. Even a 4-inch InP wafer is only 20% of the cost of a 2-inch SiC substrate, he said. Tom Anderson was able to underline this with a sobering thought: the price per inch2 for 6-inch GaAs is currently $1417, compared to $800 for semi-insulating 2-inch SiC. Commercial Drivers In the commercial sector, the excitement is tempered by the lack of an obvious big market, such as exists for GaAs for wireless handsets or nitride-based high brightness LEDs, which face little competition from other materials. In comparison, wide bandgap materials face stiff competition from existing GaAs and Si devices for RF applications. Some in-roads are being made for SiC devices, which are more mature than GaN, and market opportunities already exist in electric power switching in utility grid systems and control of power in the electrical subsystems of emerging automobile, ship and aircraft technologies. SiC is now also competing with Si for microwave power generation above handset frequencies up to around 7 GHz. Progress is also being made with GaN devices by companies such as Cree and HRL (see ), although most speakers were quick to point out that these results represented demonstrations, and that much work needs to be done on reliability and reproducibility. However, TRW s Mike Wojtowicz said GaN technology was "just too good to pass up". TRW is currently working with a defense contractor on developing a reliable GaN process for LNAs and PAs, he said. ONR s John Zolper added that WBG technology is primarily a DoD unique technology at present, but as with GaAs, the commercial applications will come later. "Although the commercial driver is not there at the moment, I m optimistic that this will come as [the technology] stabilizes," he says. "The size of program required is such that it needs to be a tri-service initiative, and we are actively making our case within the navy and in tri-service committees. In addition, various companies have their own efforts and we would look to leverage and share costs in any potential effort." But as Raytheon s Tom Kazior notes: "If there was a glaring commercial application for this technology we could probably figure out other sources of money to fund a commercial venture." Push Comes to Shove If large investment from the industrial sector is not forthcoming, support for WBG research must fall on the shoulders of the government, which in the near term is the main customer. However, DARPA is currently unable to give substance to plans it may have for WBG research funding past 2001. "DARPA and other agencies in the government are currently in the process of identifying their funding priorities for the fiscal year starting in October 2001," says DARPA s Edgar Martinez. "With the current investment situation in this area, it will be very difficult for this technology to be successfully transferred to industry, where the production of critical components for military systems is likely to take place. If we are successful in demonstrating the potential of this technology minimizing the current technical risks, and establishing an industrial base on which affordable and dependable components can be manufactured on a regular basis, then it will be the job of the DoD services and defense contractors to insert the technology in military systems." With fixed budgets, any increases in defense spending in one sector inevitably leads to cuts in others areas. Nevertheless, the general feeling was that WBG technology is important enough to warrant serious lobbying efforts to raise awareness at Government level. While the situation could ease with a new White House administration eager to spend more on defense, another funding route that was also being openly touted was the congressional "plus up". Plus ups are where Senators lobby for extra money to be set aside for projects at state level, and while this would bring much needed dollars to WBG research, it is criticized by some for diluting other projects and for by-passing the DoDpotentially leading to an uncoordinated allocation of funds and also endangering the usual competition processes. The point on which everyone did agree was that additional dollars will need to be invested soon if wide bandgap microwave materials and devices are to reach maturity in the next 5 to 7 years, and to meet system insertion targets for new military platforms. The wireless industry has adopted GaAs largely because of its enabling performance and because many manufacturing issues had been ironed out in the MIMIC program. While wide bandgap microwave devices certainly promise enabling performance, they also need a critical push to realize the same success in real systems and in the commercial sector. * University of South Carolina, Columbia, SC. December 46, 2000.