Transphorming the power transistor industry using GaN

Since 2010, new products have challenged the lead superjunction (SJ) MOSFETs retain today in the high-voltage power device market. SJ MOSFETs may be cheaper and offer lower on-resistance, gate and output charges than conventional MOSFETs, but they face competition from GaN-on-silicon (Si) HEMTs.

“GaN-on-Si HEMTs offer new capabilities, such as the possibility to work at higher frequencies,” writes Elena Barbarini, Senior Cost Analyst at System Plus Consulting in her latest reverse costing report, ‘GaN-on-Si HEMT vs SJ MOSFET: Technology and Cost Comparison’. “They are also competitive on manufacturing cost.” The report analyzes the technology innovations and manufacturing processes of the main SJ MOSFET and GaN players to identify cost drivers and understand GaN devices’ success.
To gain further insights into GaN-on-Si technology, Yole Développement and its sister company, System Plus Consulting, interviewed Philip Zuk, Senior Director Technical Marketing at Silicon Valley’s Transphorm’s office located in San Jose, California’s. Philip discussed his company’s strategy, position and challenges in supplying GaN power devices, and shared his vision of the industry.

GaN HEMT vs SJ MOSFET cost evolution
(Source: GaN-on-Si HEMT vs SJ MOSFET: Technology and Cost Comparison report,
System Plus Consulting, March 2016)

Yole Développement: How is Transphorm positioned? What is your star product?
Philip Zuk: We offer a number of star products in our 650V, second generation cascode switch portfolio, from 35-290m?. Different power levels require different devices, so we are well positioned, offering solutions from 250W-4.5kW.

YD: What products is Transphorm planning to introduce?
PZ: Now that our 650V second generation cascode switch portfolio has been released, we are working on our third generation 650V family that will go into production during the 2017 calendar year. Our third generation family will include ~30m? in a TO-247 package and 50m? in a PQFN 8×8 package. Along with our 650V innovation we are moving up the high voltage ladder to 900V products that will be available within the next 12-15 months. This will offer our customers additional options in the industrial, renewable energy, and automotive markets. We have additional R&D activities going on, but I’m unable to speak about them at this time.

YD: Will you release enhancement-mode (E-mode), normally off, GaN transistors? How do you compare E-mode to your normally-off cascode solution? Which markets do E-mode and cascade target, respectively?
PZ: E-Mode is a technology that will complement our cascode product portfolio. Looking into the past and comparing traditional single epitaxial layer silicon planar MOSFETs to superjunction silicon MOSFETs there was a definite technological advantage moving from the former to the latter. When comparing E-mode versus cascode it is more about offering flexibility to our customers. It ultimately comes down more to the customer, their application, and design preferences. For example, do they want to use TO-xxx or SMT packaging? What is the frequency of operation? Is the design continuous current mode (CCM) or critical continuous mode (CrCM)? And so on.

GaN HEMT Cross-Section – SEM view
(Source: GaN-on-Si HEMT vs SJ MOSFET: Technology and Cost Comparison report,
System Plus Consulting, March 2016)

YD: What will be the leading applications for GaN?
PZ: At Transphorm we are focused on four markets:
o Renewable Energy (solar inverters)
o Industrial (motor drives/servo)
o Automotive (electric vehicles and charging)
o Telecom/Server/Industrial (power supplies)

Any of these have the potential to be leaders in the GaN space. We see any application that converts AC to DC and DC to AC as a possible solution for GaN. It ultimately comes down to the value proposition of GaN. Is the customer looking for a reduction in weight/size (increased power density), a reduced cost with respect to the overall system bill-of-material (BOM) cost, and/or a performance increase? Depending on how these three “value propositions” are defined in one’s design will dictate how, when, and what will be used.

YD: What are the challenges facing GaN?
PZ: Adoption and incumbent technology. We still see adoption as a challenge for GaN, not from the standpoint of the GaN technology, more on being risk adverse and being an early adopter. GaN has been talked about for many years; many power supply companies have limited or no experience with it and require both hardware and firmware support.. It is like any new technology: time and energy are required to see the fruits of our labour. Our biggest incumbent is silicon superjunction technology and it is well known and can offer great performance at a low price. Silicon is not going away, but as customers demand higher power levels, increased efficiencies, and power densities GaN will surface as the only suitable solution.

YD: How do you see the market evolving?
PZ: There will be steady growth over the next couple of years as GaN adoption increases in production applications, with exponential growth being projected towards the end of the decade.

YD: Have you considered low voltage GaN products, below 650V?
PZ: Low voltage GaN is a technology that can be realized out of our existing technology, but today we are focusing on 650V and higher. As the technology becomes more mature this is an area that we may look at.

YD: Do you think that there will be a high voltage GaN application, for example, in automotive applications?
PZ: The two biggest semiconductor consumer markets – PCs and smartphones – are slowing down while the automotive market is growing, as semiconductor content in cars is growing. This trend is likely to intensify as the market progresses to self-driving cars. GaN fits nicely within this market as it can reduce size, weight, and overall BOM costs and increase performance over what silicon can offer today.

YD: What about GaN-on-GaN? Do you see potential applications in power or anything other than LEDs?
PZ: At this point I do not see GaN-on-GaN as a competitive threat due to manufacturing costs and the slow evolution of the technology.

YD: What about the cost comparison with SJ MOSFET?
PZ: GaN and SJ technology will continue to decrease in cost, but will it become cost parity, for products with the same on-resistance? In my opinion, unlikely in the high voltage markets. GaN is not an effective drop-in replacement technology due to the flexibility and advantages that it offers. It offers a system solution versus a device solution. In my experience when comparing GaN to SJ, especially in a bridgeless totem pole power factor correction (PFC) circuit, a GaN device with double to triple the on-resistance can still offer better performance than SJ in the traditional, standard PFC topology. In many cases we are not comparing like-to-like devices, which is why the overall BOM needs to be looked at.

Wide Band Gap value proposition (Source: Transphorm)

YD: What drives cost reduction in GaN manufacturing process?
PZ: As with any new technology, design, process, epitaxy, and packaging. Transphorm offers a unique value chain were we have internal control of the design, epi, and manufacturing processes. This gives us the ability to be nimble, offer new generations quicker, and to maximize our operational efficiencies. As we move from our second to our third generation cascode, cost and performance advantages will continue to be realized.

YD: What is the average manufacturing yield for GaN structures? Do you think there will be rapid improvement in the next five years?
PZ: When looking at our fully-qualified cascode technology, manufacturing yield is not a problem, as we do not have leakage or high temperature dynamic on-resistance issues. Though I cannot speak to yield information, as it is proprietary, what I can say in some cases I see it higher than SJ.

YD: Do you think you will face new competitors in the future?
PZ: With GaN’s adoption and the increased demands for power, more competitors will enter this market in order to grab market share to grow their business. GaN is the next evolution in power solutions and Transphorm will continue to lead this revolution.

Philip Zuk, Senior Director Technical Marketing, Transphorm
Philip leads the market strategy and market adoption of high voltage GaN technology in high power applications for Transphorm. He worked previously for Vishay (Siliconix) heading up their high voltage superjunction technology, Microsemi PPG running marketing efforts on their high voltage MOSFET, FRED diodes, IGBTs, and SiC efforts, Medallion Instrumentation Systems and Fairchild Semiconductor. He has expertise in microcontroller and power supply designed systems and applications, high power semiconductor devices and project management.
He holds a MBA (Hons) from I.H. Asper School of Business, University of Manitoba and has a Bachelor of Science in Electrical Engineering, University of Manitoba, and an Electronic Engineering Technology Associate Degree, Red River College.

Sources: www.yole.fr  /  www.systemplus.fr  /  www.transphormusa.com 

RELATED REPORT

GaN on Si HEMT vs SJ MOSFET: Technology and Cost Comparison
Will SJ MOSFETs still be attractive compared to GaN devices? More HERE.