The emergence of wide bandgap semiconductors for power conversion is an important turning point for AC/DC and DC/DC applications in consumer, enterprise, automotive, and clean energy systems. Gallium Nitride (GaN) in particular, enables the world’s fastest power switches, making it make possible for power supplies to deliver a much higher efficiency and superior power density. This new technology also provides a much better use case in the consumer adapter area—faster charging and a smaller form factor. It can also provide much a lower cost of ownership for server applications, thanks to less wasted energy and lower cooling costs.
GaN high-electron mobility transistors (HEMT) are a high-performance alternative to superjunction MOSFETs. They deliver superior switching performance due to their low gate charge, “flat” output capacitance, and zero reverse recovery. This translates into a power device that enables higher energy efficiency, while enabling new topologies. GaN devices are capable of operating several orders of magnitude higher than their silicon counterparts, and have the potential to be manufactured at a lower cost than silicon.
None of this is really new, per se. GaN didn’t just break out onto the power conductor scene overnight. It has been available for well over a decade now, used across compound semiconductor manufacturers, research institutions, and a myriad of small wafer fabs. Today, it has been widely adopted for use in various LED and RF applications. But, despite that, GaN has never really been considered mainstream before—largely because of these preconceived notions of GaN being a very exotic process, held back by a combination of high manufacturing costs and track record of subpar reliability.
Over the last five years, however, we’ve seen GaN take significant strides on both device and process improvements, creating a baseline standard of quality and reliability that is friendlier to mass production and adoption. In fact, many of the biggest foundries across the planet either currently have, or are ramping up, their GaN production programs.
While cost impediments are nothing new for emerging hardware electronic technologies, a path to adoption—at a lower cost than silicon devices—is very viable. Manufacturing improvements have already been made with the goal of achieving cost parity with superjunction MOSFET devices.
The two major cost contributors to GaN are the wafer substrate and subsequent epitaxial deposition processing costs. However, we see three areas that counterbalance and allow for decrease costs:
1) Lower processing costs due to improvements in epitaxial growth (MOCVD) throughput from better manufacturing efficiencies and lower capital expenditures from older and partly depreciated equipment
2) A GaN HEMT device has a much smaller effective area per Rdson (or a higher overall current capability per device area) compared to an equivalent superjunction MOSFET, resulting in higher device density per wafer
3) GaN HEMTs as lateral devices make monolithic integration of multiple devices easier and more cost-effective than superjunction vertical devices—yielding a smaller total solution and a lower cost bill of material (BOM)
One of the challenges for engineering teams to use GaN HEMTs has been the small number of complementary drivers and controllers available in the market. GaN has a tighter requirement to drive the gate, given its unique structure. Therefore, a tuned (or optimized) driver is needed to enable faster switching using GaN.
In addition to drivers, new controllers implementing zero-voltage switching (ZVS) topologies are needed to extract the value of GaN. ZVS controllers enable much smaller magnetic components by increasing switching frequency while saving energy and reducing heat. Superjunction MOSFETs generally fall short of ZVS due to their non-linear output capacitance. GaN enables these new ZVS topologies due to its relatively linear output capacitance. When deploying ZVS topologies, additional savings in BOM costs may be obtained by integrating passive components and reducing the size and cost of magnetic components. Recently, a number of these new GaN-friendly controllers have been introduced to the market.
An important application where GaN delivers superior performance is the 30 W to 65 W consumer adapter. Efficiencies up to 94 percent (a 50 percent power loss reduction) and power density up to 20 W/cubic inch (about half the size) can be achieved using GaN power ICs and ZVS controllers. An integrated GaN power IC, operating in an active clamp topology with ZVS, can deliver a solution with about half the wasted energy and half the overall volume, compared to a traditional silicon-based flyback topology commonly used today. These new solutions allow for a seamless implementation of GaN, avoiding the complex circuitry normally needed to drive discrete GaN power switches.
Beyond consumer adapters, GaN power devices promise to enable more advanced and less energy-hungry power conversion solutions across a broad array of applications. Just to name a few possible use cases: servers, automotive, clean energy, wireless charging, solar inverters, and motor control. The sky’s the limit on GaN, and as more engineers catch on and push GaN further to consumer-level volumes, that will also drive further investments into its development—and in turn, bring manufacturing costs down.
Reaching the necessary scale to decrease costs, provided by selected high-volume applications, has to be the first step on this process. This will create even more momentum for the wider adoption of GaN in power supplies and blueprint for system topologies necessary to take full advantage of GaN technology. Those new topologies, like LLC and active clamp flyback, empower engineers to shrink the power supply, thanks to smaller and more efficient passive and magnetic components. When you reduce the size of these designs, that opens the door to two new avenues: to squeeze the same amount of charging power into more portable, convenient devices, or to wring even more power than before out of same-sized adapters.
At the end of the day, the value proposition GaN has to offer is clear. Engineers are getting a better trade-off when it comes to size and energy efficiency. That trade-off then creates ripple effects, providing systemwide improvements in consumer power adapters—be it for computers, smartphones, or any devices in between—that manage to kill three birds with one stone: reducing size, cost, and power consumption. It’s because of that triple threat, that GaN’s time in the spotlight—and the mainstream—has finally arrived.
