GaN is Finally Here for Commercial RF Applications! Eric Higham Director of GaAs & Compound Semiconductor Technologies Strategy Analytics Gallium Nitride (GaN) has been a technology with so much promise for commercial RF applications, but it always seemed a year or so away! The unique material properties of GaN have been of interest to scientists and researchers for some time. Literature references preliminary research into GaN for LED applications in the 1970s. GaN HEMT devices for RF applications began appearing in the early part of the 2000s and Efficient Power Conversion Corporation (EPC) introduced the first enhancement mode transistor aimed at power electronics applications in 2009. Forecast the GaN opportunity in the LED, RF and power management markets and a case can be made for revenues easily into the $10 - $15 billion range. The overwhelming majority of this revenue comes from the LED market where GaN-on-sapphire or GaN-on SiC-based LEDs are currently being produced in high volume, at very low prices. The volume in this market is also pushing development of GaN-on-silicon and GaN-on-GaN processes to further reduce costs and ease some of the manufacturing challenges. While GaN has been very successful in the LED market, the same has not generally been true for RF or power electronics applications, but there are some very strong signs that this situation is changing rapidly. The allure of GaN-based devices stems largely from the attractive intrinsic physical properties of the material. The material exhibits wide bandgap, high breakdown voltage, extremely high power density and high gain at microwave frequencies. These properties, coupled with excellent thermal conductivity make GaN devices well suited for high power, high frequency and wide bandwidth applications in extreme environments. The table below shows a comparison of some physical properties of GaN with other semiconductor device materials. Si GaAs InP SiGe SiC GaN Lattice constant (Å) 5.4 5.7 5.9 5.5 3.1 3.2 Saturation velocity (cm/s) 1 x 10 7 0.8 x 10 7 2.2 x 10 7 --- 2 x 10 7 2.5 x 10 7 e - mobility (cm 2 /Vs) 1350 8000 10000 3000 900 1500 E g bandgap (ev) 1.1 1.4 1.3 0.7-1.1 3.3 (4H) 3.4 F t (GHz) FET 20 150 300 50 20 150 Power density (W/mm) 0.2 0.5 --- 0.3 10 >30 Thermal conductivity (W/cmK) 1.5 0.5 --- --- 4.9 ~2.0 Emission wavelength (µm) N/A 0.6-0.9 1.0-1.5 N/A N/A 0.4 Source: Strategy Analytics The bandgap, approaching 3.4eV, enables GaN devices to support peak internal electric fields approximately five times higher than either silicon or GaAs. This allows for higher breakdown
voltages, which is a critical attribute for high-power requirements and for achieving higher electrical efficiencies with higher supply voltages. GaN material also has low intrinsic carrier concentrations at device operating temperatures, which allows high-temperature operation and high radiation stability. For power electronics applications, these material parameters result in commonly used figures of merit that are orders of magnitude higher for GaN than the incumbent silicon-based technologies. If we focus on RF applications, the material characteristics of GaN offer clear advantages. GaN RF transistors offer many times the theoretical maximum output power density of GaAs or silicon transistors. Additional key characteristics of GaN transistors include high cut-off frequency and good thermal conductivity. GaN devices offer the best solution for simultaneous high power, high frequency and high temperature operation. The chart below captures frequency versus output power capabilities for several compound semiconductor and traveling wave tube technologies used in power amplifier applications. The combination of high frequency, wide bandwidth and high power capabilities make GaN a natural fit for military applications. Military agencies have fostered, as well benefited from GaN device and process development. Opportunities for GaN devices in military applications are numerous but the three largest segments are Electronic Warfare (EW), communications and radars, specifically Active Electronically Scanned Arrays (AESA). The US military has already deployed GaN-based RF devices for EW applications. The largest application has been in anti-ied (improvised explosive devices) systems. In these applications, GaN-based PAs amplify broadband microwave noise to disrupt and jam RF signals used to detonate the IEDs. The combination of broadband and high-power microwave emission is ideal for radio signal jamming applications. In the early part of the decade, this represented the largest military application for GaN devices. With US troop withdrawals from Iraq and Afghanistan, it is likely that this opportunity will diminish along with the US troop presence.
With asymmetrical warfare expected to play a large role in future conflicts, the US has identified this anti-ied capability as central to its future battle plans. Mirroring the tremendous growth in data consumption in the commercial market, military battle strategies have adopted a network-centric approach. This battle philosophy aims to connect all the military assets, whether they are land, sea, air or space into a high bandwidth network that promotes sharing of resources and information among these resources. With the range of legacy radios and different forces, this idea relies on very broad bandwidth, smart and agile communications devices to stitch together all the information sources into a coherent network. Many of the latest tactical radios are incorporating GaN to fulfill this requirement. The largest area for the growth of GaN in military applications is phased array radars. Because of their flexibility, performance and reliability, phased array radar use in military applications is growing quickly. These radars can produce very high-pulsed powers for surveillance applications or multiple simultaneous beams for shorter distance targeting and acquisition applications. These radars typically contain a number of array elements, operating in bands from UHF to X-band. The military has installed these phased arrays on ground, air, ship and mobile platforms. Active electronically scanned array (AESA) radar systems contain a (relatively) low power solid-state transmit/receive (T/R) module connected to each antenna element. Because the T/R modules are combined at the antenna output to achieve the proper transmit power and beam configuration, the transmit power of each module is likely below 10W. Because of the output power, existing AESA systems have historically contained GaAs semiconductor components. Despite the GaAs heritage, this is an area squarely in the sights of GaN component manufacturers. Since many of the platforms using AESA radars are aircraft, size and weight become critically important. In addition to the performance characteristics we have discussed, the ability to achieve equivalent output power at a smaller size and handle higher temperatures means heat sinks and cooling plates can be smaller, lighter and less costly. Despite undeniable performance advantageous for power applications and widespread usage in military applications, commercial adoption of the technology for RF applications has been much slower than expected. Initial concerns about reliability and repeatability have become moot as deployed systems build a compelling set of actual metrics. The cost issue is much thornier. To take full advantage of the material advantages of GaN, just about every RF manufacturer uses a GaN-on-SiC wafer scheme. Low volumes, the cost of the SiC wafers, coupled with wafer diameters in the 2 4 range all contribute to GaN devices being much more expensive than competitive technologies. Research that we conducted three years ago had GaN running about three times the price of GaAs and LDMOS in power applications. Despite these challenges, the interest in GaN for commercial applications is on the rise and the technology finally appears to be getting substantial traction. So, what has changed? While there will always be price pressure in commercial applications, the price differential to other technologies is shrinking quickly. At various industry events, we have heard some GaN manufacturers say the price of GaN is now comparable to LDMOS and GaAs for high power applications. Proponents of the technology point to the economies of scale reducing the price of SiC wafers as volume increases. They also point out that, the assembled cost of a GaN
device entails much more than the cost of the SiC wafer. In 2013, RFMD introduced the first 6- inch GaN-on-SiC wafers for RF power transistors and M/A-COM technology introduced a line of GaN devices in plastic packaging. Companies like TriQuint, Cree and UMS continue to expand their GaN product and process portfolios. Developments like these and ongoing process improvements will continue to reduce the cost of GaN devices. In addition to manufacturing, process and cost improvements, there has been a subtle change in strategy and acceptance in the commercial segments. Initially, the value proposition for GaN acceptance was linked to better output power characteristics. A system could get better performance out of the same form factor or smaller devices could produce equivalent performance. This was a big advantage in military systems where improved performance meant increased functionality and reductions in size, weight and power. The same reasoning did not hold the same appeal to commercial applications. In these applications, price is paramount, so better performance was not necessarily a must have, especially if the price was higher as a result. Initial attempts to penetrate market applications based solely on higher output power performance were not very successful, but manufacturers and service providers quickly discovered a much more compelling benefit from GaN. Being able to produce equivalent RF output power with lower DC power dissipation has proven very attractive. This feature is driving much more commercial market acceptance than the superior power capabilities of GaN. The same performance at lower DC dissipation feature translates to lower operating costs and service providers are touting green networks that consume less energy. This has been the biggest boost to GaN use by service providers and operators. CATV became the earliest market segment to experience widespread GaN adoption, but other commercial market segments are following quickly. As increasing data consumption forces architectural changes and more challenging performance requirements in wireless infrastructure networks, GaN is quickly capturing market share in the power amplifier function. Military satellite communications applications have been early targets for GaN, but as more companies develop and space-qualify higher frequency products, the commercial satcom and VSAT market segments are using more GaN products. Finally, we are starting to see much more evidence of the high frequency, broad bandwidth and higher linearity performance of GaN devices being essential to the requirements for point-to-point radios used in mobile backhaul. The chart below shows out latest estimate of the GaN market for RF and high power electronics applications.
Source: Strategy Analytics Strategy Analytics forecasts that the market for GaN microelectronic devices will grow to reach nearly $335 million in 2017. Aerospace and Defense applications will continue to represent the largest portion of the demand and this sector will still grow, but commercial applications will grow even faster. At the close of the forecast period, aerospace and defense applications will have dropped to slightly more than 50 percent of the total market. We believe that the high power electronics segment will see the fastest growth and be the largest commercial segment at the end of the forecast period. It does appear that the RF market segments that have been discussed have passed the tipping point. We believe GaN is entering a period of strong adoption in commercial markets and we estimate GaN devices will account for nearly $85 million in revenue in 2017. We will be watching this very dynamic market closely. With most of the major device manufacturers and foundries having access to GaN capabilities, we anticipate a rapid increase in product and process developments. It will be very interesting to see the rise of commercial opportunities and the fundamental research that result from GaN finally reaching critical mass.