Wide Band-Gap Semiconductor Based

Power Electronics for Energy Efficiency

Isik C. Kizilyalli, Eric P. Carlson, Daniel W. Cunningham, Joseph S. Manser, Yanzhi “Ann” Xu, Alan Y. Liu
March 13, 2018
United States Department of Energy
Washington, DC 20585
 TABLE OF CONTENTS

Abstract 2
Introduction 2
Technical Opportunity 2
Application Space 4
Evolution of ARPA-E’s Focused Programs in Power Electronics 6
Broad Exploration of Power Electronics Landscape - ADEPT 7
Solar Photovoltaics Applications – Solar ADEPT 10
Wide-Bandgap Materials and Devices – SWITCHES 11
Addressing Material Challenges - PNDIODES 16
System-Level Advances - CIRCUITS 18
Impacts 21
Conclusions 22
Appendix: ARPA-E Power Electronics Projects 23
ABSTRACT
The U.S. Department of Energy’s Advanced Research Project Agency for Energy (ARPA-E) was established in
2009 to fund creative, out-of-the-box, transformational energy technologies that are too early for private-sector
investment at make-or break points in their technology development cycle. Development of advanced power
electronics with unprecedented functionality, efficiency, reliability, and reduced form factor are required in an
increasingly electrified world economy. Fast switching power semiconductor devices are the key to increasing
the efficiency and reducing the size of power electronic systems. Recent advances in wide band-gap (WBG)
semiconductor materials, such as silicon carbide (SiC) and gallium nitride (GaN) are enabling a new genera-
tion of power semiconductor devices that far exceed the performance of silicon-based devices. Past ARPA-E
programs (ADEPT, Solar ADEPT, and SWITCHES) have enabled innovations throughout the power electronics
value chain, especially in the area of WBG semiconductors. The two recently launched programs by ARPA-E
(CIRCUITS and PNDIODES) continue to investigate the use of WBG semiconductors in power electronics.
From materials and devices to modules and circuits to application-ready systems integration, ARPA-E projects
have demonstrated the potential of WBG semiconductors to lower the cost of high-efficiency power electronics
to enable broad adoption in energy applications.

INTRODUCTION
Electricity generation currently accounts for 40% of primary energy consumption in the U.S.1 and over the next
25 years is projected to increase more than 50% worldwide.2 Electricity continues to be the fastest growing form
of end-use energy. Power electronics are responsible for controlling and converting electrical power to provide
optimal conditions for transmission, distribution, and load-side consumption. Estimates suggest that the fraction
of electricity processed through some form of power electronics could be as high as 80% by 2030 (including
generation and consumption), approximately a twofold increase over the current proportion.3 Development of ad-
vanced power electronic devices with exceptional efficiency, reliability, functionality, and form factor will provide
the U.S. with a competitive advantage in deployment of advanced energy technologies. Additionally, widespread
integration of innovative converters offers substantial energy saving opportunities both directly, by inherently
more efficient designs, and indirectly, by facilitating higher levels of adoption for fundamentally higher performing
applications and technology solutions.

Technical Opportunity
Achieving high power conversion efficiency requires low-loss power semiconductor switches. Today’s incumbent
power silicon (Si) based switch technology includes metal oxide field effect transistors (MOSFET), IGBTs and
thyristors. Silicon power semiconductor devices have several important limitations:
• High Losses: The relatively low silicon bandgap (1.1 eV) and low critical electric field (30 V/μm)
require high voltage devices to have substantial critical thickness. The large thickness translates to
devices with high resistance and associated conduction losses.
• Low Switching Frequency: Silicon high voltage power MOSFETs require large die areas to keep con-
duction losses low. Resulting high gate capacitance and gate charge produce large peak currents
and losses at high switching frequencies. Silicon IGBTs have smaller die than MOSFETs due to
utilization of minority carriers and conductivity modulation, but the relatively long lifetime of minority
carriers reduces the useful switching frequency range of IGBTs.

1   U.S. Energy Information Administration, Monthly Energy Review (March, 2015)
2   U.S. Energy Information Administration, International Energy Outlook 2016 (May, 2016)
3   L.M. Tolbert, et al. Power Electronics for Distributed Energy Systems and Transmission and Distribution Applications: Assessing the
Technical Needs for Utility Applications. Oak Ridge, TN: Oak Ridge National Laboratory (2005)
2
 • Poor High-Temperature Performance: The relatively low silicon bandgap also contributes to high
intrinsic carrier concentrations in silicon-based devices, resulting in high leakage current at elevated
temperatures. Temperature variation of the bipolar gain in IGBTs amplifies the leakage and limits the
maximum junction temperature of many IGBTs to 125°C.

Figure 1: (Left) Relationship of converter power vs switching frequency where the red line indicates the limit of silicon
operation. The dotted line area illustrates the opportunity space where WBG systems can operate and silicon devices
cannot. (Right) Application areas that lie within the dotted line area include motor drives, automotive, power supplies,
aerospace, and distributed energy resources.
Figure 1 illustrates the opportunity space associated with silicon performance limitations. As switching frequency
increases, converter power is reduced. In practice, high power silicon systems operate at low frequencies (<10
kHz). This translates to larger passive components (e.g. inductors, capacitors) which increases volume and
weight. Wide band-gap (WBG) systems can be operated at higher power and higher frequency within the space
silicon devices cannot.

As a result, new opportunities for higher efficiency have emerged with the development of WBG power semi-
conductor devices, driven by the fundamental differences in material properties between Si and semiconductors
such as Silicon Carbide (SiC) and Gallium Nitride (GaN). Higher critical electric fields in these WBG materials
(≥ 2 MV/cm) enable thinner, more highly doped voltage-blocking layers, which can reduce on-resistance by
two orders of magnitude in majority carrier architectures (e.g., Metal Oxide Field Effect Transistors, MOSFETs)
relative to an equivalent Si device.4 Moreover, high breakdown electric field and low conduction losses mean
that WBG materials can achieve the same blocking voltage and on-resistance with a smaller form factor. This
reduced capacitance allows higher frequency operation compared with a Si device. The low intrinsic carrier
concentration of WBG materials enables reduced leakage currents and robust high-temperature performance.
WBG semiconductors therefore provide a pathway to more efficient, lighter, high temperature capable (reduced
cooling requirements), and smaller form factor power converters. However, to unlock the potential of WBG
based devices, intensive and systematic R&D efforts need to take place at every stage of the power electronics
value chain, as depicted in Figure 2.

4   Heffner, A. et al. Recent Advances in High-Voltage, High-Frequency Silicon-Carbide Power Devices. Industry Applications Confer-
ence, 41st IAS Annual Meeting (2006)
3
 Materials Devices / Modules Products Systems
Packaging Integration

Figure 2: Power Electronics Value Chain.

Application Space
High impact opportunities exist across a variety of applications, including:

Motor Drives: Across all sectors, electric motors account for approximately 40% of total U.S. electricity de-
mand.5 It is estimated that 40-60% of currently installed electric motors could benefit from variable frequency
drives (VFDs),6 which enable efficient adaptation to speed and torque demands. Depending on the application,
incorporation of VFDs can reduce energy consumption by 10-30%.7 Conventional VFDs for high power applica-
tions are bulky and occupy significant space. Size, power density and efficiency can be improved, and the overall
system cost reduced, by using WBG-based VFDs.

Automotive: Power electronics such as traction inverters, DC boost converters, and on-board battery chargers
are critical elements in hybrid and electric vehicles (EVs). They affect energy efficiency in two ways: directly
through switching and other losses, and indirectly by adding volume and weight to the vehicle. WBG inverters
can reduce both direct and indirect losses by operating at higher switching frequencies, efficiencies, and tem-
peratures.8 As a result, 15% improvement in energy efficiency has been predicted for representative hybrid EVs
employing SiC traction inverters, with even larger energy savings possible given greater degrees of drivetrain
electrification.9 Assuming aggressive market adoption of EVs in the U.S., use of WBG vehicle power electronics
could save as much as 1 quadrillion Btu of energy per year by 2050 relative to conventional Si-based systems.10
Additionally, efficient, lightweight, and low-cost DC fast charging infrastructure (≥120 kW) enabled by WBG con-
verters will advance the commercial viability and adoption rate of EVs. In conjunction with a cleaner electricity
generation portfolio, this has the potential to significantly reduce the one quarter of total U.S. greenhouse gas
emissions that stem from the transportation sector.11

Data Centers: Energy consumption in data centers accounted for approximately 2% of total electricity use in the
U.S. in 2014.12 The power delivery architecture of most modern data centers consists of a line frequency trans-
former, low voltage power distribution network, centralized backup unit, and inefficient voltage regulators.13 Strat-
egies to improve energy efficiency range from integration of lower loss power converters to complete redesign
of the power delivery network.14 The latter approach often involves converting higher voltages at the rack level,
where space is limited and proper thermal management is imperative. High power density converters based on

5   Waide, P.; Brunner, C. U. Energy-Efficiency Policy Opportunities for Electric Motor-Driven Systems. IEA (2011)
6   Energy Efficiency Roadmap for Electric Motors and Motor Systems. Energy Efficient End-use Equipment, IEA (2015)
7   Energy Efficiency and Power Electronics. Danfoss, ATV Seminar, March 1, 2012
8   Hamada, K. et al. SiC-Emerging Power Device Technology for Next-Generation Electrically Powered Environmentally Friendly Vehi-
cles. IEEE Transactions On Electron Devices (2015)
9   Zhang, H. et al. Impact of SiC Devices on Hybrid Electric and Plug-In Hybrid Electric Vehicles. IEEE Transactions on Industry Appli-
cations (2011)
10   U.S. Department of Energy. Quadrennial Technology Review (2015)
11   Williams, J.H. et al. Pathways to Deep Decarbonization in the United States (2014)
12   Shehabi, A. et al. United States Data Center Energy Usage Report. Berkeley, CA: Lawrence Berkeley National Laboratory (2016)
13   Zhabeloval, G. et al. Data center energy efficiency and power quality: an alternative approach with solid state transformer. 41st
Annual conference of the IEEE Industrial Electronics Society (2015)
14   Candan, E. et al. A Series-Stacked Power Delivery Architecture with Isolated Differential Power Conversion for Data Centers. IEEE
Transactions On Power Electronics (2016)
4
WBG devices can be key enablers for more efficient systems, as higher temperature tolerance can reduce cool-
ing loads and further boost data center grid-to-chip efficiency.

Aerospace: Longer, thinner, and lighter wings can reduce fuel consumption and carbon emissions by 50%
relative to current commercial aircraft.15 Such a reduction would save approximately 1 quadrillion Btu of energy
per year across the U.S. fleet at current demand.16 Achieving this transformative wing design requires elec-
tromechanical actuators that are small and lightweight with robust operation over a wide temperature range.17
Moreover, electrification of environmental controls, fuel pumps, brakes, and de-icing systems can further re-
duce weight and increase efficiency through elimination of bleed air controls and pneumatic/hydraulic systems.18
WBG-based converters, with high gravimetric and volumetric power density plus high temperature operation,
offer a pathway to achieving significant energy savings in air transport. These key attributes will contribute to
weight reduction, enabling new paradigms in body design.

Distributed Energy Resources: In grid applications, such as solar photovoltaic (PV) and wind, as well as the
emerging fields of high voltage direct current (HVDC) and flexible alternating current transmission systems
(FACTS), power conditioners are required to control the flow of electricity. This is achieved by supplying voltages
and currents in a form that is optimally suited to the load. Traditional Si power electronics are responsible for a
loss of approximately 4% of all of the electricity generated in these applications and are the dominant point of fail-
ure for installed systems. For instance, a typical maximum conversion efficiency for a silicon-based PV inverter
is approximately 96% (AC output/DC input)19 including transformer losses, which drops significantly at operating
temperatures above 50 °C. Novel WBG electronic circuits present a route to lower system-level costs by oper-
ating at higher switching frequencies that reduce the size of passive components and lower the overall system
footprint. In addition, WBG circuits will increase system-level efficiency by allowing PV arrays to operate at higher
voltages (e.g. > 1,500VDC). This will enable DC systems with fewer voltage conversions/transformers, replacing
traditional combiner boxes with DC/DC converter. The need for on-site AC transmission lines will ultimately allow
for easier integration of energy storage solutions in the central substation. Together with a higher semiconductor
operating temperature, the advantages of WBG electronics offer a pathway to more robust power converters with
mean time to failure (MTTF) comparable to the PV and wind system lifetime (typically 25 years or longer). This
will lower the equipment replacement cost and total plant Operation & Maintenance and can have a significant
impact on the levelized cost of electricity in distributed resource applications.

15   Slimmed Down Aircraft Wing Expected to Reduce Fuel and Emissions by 50%. NASA, accessed November 29, 2016, https://www.
nasa.gov/image-feature/ames/slimmed-down-aircraft-wing-expected-to-reduce-fuel-and-emissions-by-50
16   Transportation Energy Data Book. Oak Ridge National Laboratory, 35th edition (2016)
17   Thin-Wing Electromechanical Actuation (EMA) Demonstration. Department of Defense Air Force Research Lab.
18   Wheeler, P. The More Electric Aircraft: Why Aerospace Needs Power Electronics. Accessed November 29, 2016, http://www.lboro.
ac.uk/microsites/research/iemrc/Events%20write%20up/Power%20Electronics%2014.05.09/More_Electric_Aircraft_000.pdf
19   SMA Technical Information, Efficiency and Derating, WKG-Derating-US-TI-en-15, Version 1.5, 2016
5
EVOLUTION OF ARPA-E’S FOCUSED PROGRAMS IN POWER
ELECTRONICS
ARPA-E has catalyzed innovations in power electronics ever since its inception in 2009. The exploration began
with OPEN 2009 projects in GaN materials as well as packaging for automotive and LED applications. In 2010,
ARPA-E launched its first focused program in power electronics, Agile Delivery of Electric Power Technology (AD-
EPT), which cast a broad net for innovations in power conversion. ADEPT sought breakthrough technologies in
magnetic materials with high operating flux densities, advanced solid-state switch technologies, advanced circuit
topologies and converter architectures, and advanced charge storage devices. A year later, Solar ADEPT was
launched to bring the general approach of ADEPT to PV power electronics. The successes in ADEPT and Solar
ADEPT, along with four WBG related projects in OPEN 2012, showed that WBG semiconductor devices held ex-
ceptional potential in dramatically improving energy efficiency, broadening the application space of power elec-
tronics. Therefore, ARPA-E initiated the Strategies for Wide-Bandgap, Inexpensive Transistors for Controlling
High-Efficiency Systems (SWITCHES) program with a focus on material- (e.g. GaN substrates) and device-level
technologies (diodes and transistors). The experience and lessons learned from SWITCHES have revealed the
need for further exploration both further up and down the value chain. Down the value chain, the major obsta-
cle experienced by many SWITCHES project teams was selective area p-type doping of GaN. This initiated the
Power Nitride Doping Innovation Offers Devices Enabling SWITCHES (PNDIODES) program, a supplementary
program that would seek the fundamental understanding of material properties and processing technologies for
selective area doping of GaN. Up the value chain, ARPA-E recognized the opportunities that harness recent
advancements at the component level for transformational developments at the circuit and system levels. Hence
the launch of Creating Innovative and Reliable Circuits Using Inventive Topologies and Semiconductors (CIR-
CUITS) Program. The time evolution of ARPA-E power electronics programs is depicted in Figure 3.

Charge Balanced SiC FETS

& plasma based AC/DC
transformer

Moving up & down PE value chain

PNDIODES: Addressing GaN
doping challenges
CIRCUITS: Compact, efficient
designs for WBG switches

Figure 3: ARPA-E’s Power Electronics Timeline.

6
ADEPT
Broad Exploration of Power Electronics Landscape
BROAD EXPLORATION OF POWER ELECTRONICS LANDSCAPE - ADEPT
In 2010, the ADEPT program set out to lower the cost and improve performance of power converters and associ-
ated management systems. The program sought innovations across the entire value chain in power electronics
from advanced component technologies and converter architectures, to packaging and manufacturing process-
es. The program aimed to address three categories of performance and integration levels: 1) fully-integrated,
chip-scale power converters, 2) package-integrated power converters, and 3) lightweight, medium voltage ener-
gy conversion for high power applications. The original program areas of interest included magnetic materials
with high operating flux densities, advanced solid-state switch technologies, advanced circuit topologies and
converter architectures, and advanced charge storage devices. The most successful projects in the ADEPT
program were those that utilized WBG components.

One example of a successful project is the development of a SiC-based EV battery charger, led by Arkansas
Power Electronics International (APEI). The project was titled “10 times smaller EV charger using SiC based
power transistors”, and APEI later won a 2014 R&D100 award.20 On-board battery chargers for EVs and plug-in
hybrid electric vehicles (PHEV) such as the Toyota Prius offer the convenience of universal charging from any
available power outlet, but this functionally adds additional volume and weight to the system. WBG power elec-
tronics based chargers can enable operation at higher switching frequencies as well as elevated temperatures,
thus allowing for a more compact system size/weight via reductions in passive component sizes as well as sim-
plification of the thermal management system.

The APEI team designed a two-stage architecture with topologies that optimized for system density and efficien-
cy. This leveraged the benefits of SiC power devices over traditional Si power diodes into significant system level
improvements. In the first stage AC-DC converter, a bridgeless boost power factor correction (PFC) topology was
chosen. This capitalizes on the negligible reverse recovery current of SiC devices over Si power diodes, result-
ing in performance advantages especially at fre-
quencies in excess of 100 kHz. The second stage
was a DC-DC converter utilizing a phase-shifted
full-bridge (PSFB) topology. This took advantage
of the low output capacitance, high voltage block-
ing capability, and stable on-resistance at elevat-
ed temperatures of SiC devices for greater overall
converter performance.21

The experimental implementation of the charger

utilized commercially available 1200V/20A SiC
MOSFET and Schottky Diodes from Cree pack-
aged in a multi-chip power module designed by
APEI. The full system housed three such models
each with different device configurations. The SiC
based charger, shown in Figure 4, operated at a
switching frequency of 200 kHz with a total weight
of 1.6 kg and volume of 1.2 L. This demonstrated
Figure 4: (bottom) APEI’s prototype charger compared to (top) a 10x increase in volumetric power density and 9x
2010 Toyota Prius plug-in hybrid battery charger (from ref 20). increase in gravimetric power density compared to

20   “2014 Awards Highlight New Trends,” R&D Magazine, Ed. T. Studt, Vol 56, no. 5, pg. 22, October 2014
21   B. Whitaker, A. Barkley, Z. Cole, B. Passmore, D. Martin, T. R. McNutt, A. B. Lostetter, J. S. Lee, and K. Shiozaki, “A High-Density,
High-Efficiency, Isolated On-Board Vehicle Battery Charger Utilizing Silicon Carbide Power Devices,” IEEE Transactions on Power Elec-
tronics, vol. 29, no. 5, May 2014. DOI: 10.1109/TPEL.2013.2279950
8
the 2010, Si based Toyota Prius plug-in battery charger. At the highest tested peak power level of 6.1 kW the
system efficiency was 94%, with an overall peak efficiency of 95% at an output power of 3.1 kW. This is a clear
demonstration of the added value that WBG power electronics provides for automotive technologies.

Two ADEPT projects investigating the potential of GaN devices were also able to make tremendous advance-
ments. Transphorm’s project “High Performance GaN HEMT modules for Agile Power Electronics” aimed to
develop kW class inverters with greater efficiency and power density than incumbent Silicon based Insulated
Gate Bipolar Transistor (IGBT) motor drives. To achieve this, the team had to overcome material challenges as-
sociated with the epitaxial growth of GaN on 6-inch Si with low defect density, fabricate GaN-on-Si High Electron
Mobility Transistors (HEMTs), and demonstrate improved power module performance over the state of the art Si
IGBTs.

Transphorm developed 600V normally-on HEMTs and achieved enhancement mode, or normally-off, operation
by integrating a Si FET in a cascode configuration.22 Under resistive load switching, an inverter using the 600V
GaN e-mode HEMTs hard switched at 100 kHz PWM frequency and demonstrated 98.5% efficiency. When
evaluated in a motor drive test lab side to side with a Si IGBT-based motor-drive inverter operating at 15 kHz
PWM frequency, the GaN inverter showed 8, 4, and 2% improvement in efficiency at low, mid, and full loads
(roughly 500, 1000, and 1500 W) respectively operating at 7x higher PWM frequency23. In 2015, Transphorm,
Inc. announced that its 600V GaN transistor has been fully Joint Electron Device Engineering Council (JEDEC)
qualified and is slated for mass production in an industry standard TO-247 size package.24

The ADEPT project led by the Massachusetts Institute of Technology took a broader approach to the challenge
and addressed three areas needed to improve power conversion: switching devices, inductors, and circuit de-
sign in their work on a novel GaN device, with an application focus on power converters for driving light emitting
diode (LED) loads. In the device arena, MIT pioneered a tri-gate normally off metal-insulator-semiconductor
field-effect transistor and demonstrated it with a breakdown voltage of 565 V at a drain leakage current of 0.6 uA/
mm25. In parallel, MIT also developed novel circuit topologies that leveraged wide-bandgap power electronics
to break the performance constraints of commercially available systems. An LED driver employing a new power
conversion architecture for single-phase AC grid interfaces was developed. The driver operated from 120 VAC,
while supplying a 35 V, 30 W output. Operating at a variable switching frequency of 5-10 MHz with an efficiency
of >93%, the converter had a power factor of 0.89 with an overall box power density of 2.7 W/cm3. For compar-
ison, commercial LED drivers typically operate in the 50-100 kHz frequency range, with maximum efficiencies
around 88% and power densities <1.0 W/cm3.

The technologies and related IP developed at MIT is now being commercialized at several different start-ups.
The circuit design concepts developed during ADEPT and lessons learned informed the design of a new laptop
charger now being commercialized by FINsix, a MIT spin-out. The charger – named DART – is a 65 W charger
that is 3 times smaller and lighter than conventional chargers. FINsix and Lenovo are currently in a partnership
to make the smaller charger available for select ThinkPad models. Another MIT spinout – Cambridge Electronics
– was started by students whose research was funded by ADEPT and are offering state-of-the-art GaN devic-
es for sale. The company has received numerous awards including the Massachusetts’s MassVentures Start
Award as well as the 2016 Compound Semiconductor Industry Award.

22   P. Parikh, Y-F. Wu, L. Shen, “Commercialization of High 600V GaN-on-Silicon Power Devices,” ISCRM, September 2013.
23   Y-F. Wu, D. Kebort, J. Guerrero, S. Yea, J. Honea, K. Shirabe, and J. Kang, “High-Frequency, GaN Diode-Free Motor Drive Inverter
with Pure Sine Wave Output,” Power Transmission Engineering, October 2012.
24   http://www.transphormusa.com/news/transphorm-announces-industrys-first-600v-gan-transistor-247-package/
25   B. Lu, E. Matioli, and T. Palacios, “Tri-Gate Normally-Off GaN Power MISFET,” IEEE Electron Device Letters, vol. 33, no. 3, March
2012. DOI: 10.1109/LED.2011.2179971
9
Solar ADEPT
Solar Photovoltaics Applications
SOLAR PHOTOVOLTAICS APPLICATIONS – SOLAR ADEPT
Solar ADEPT held similar aspirations to ADEPT but specifically focused on packaging and manufacturing pro-
cesses for power converters to reduce costs for photovoltaic systems. Similar to ADEPT, Solar ADEPT aimed
to address technical challenges at different performance and integration levels, from chip-scale all the way to
utility-scale. Tailored for the solar PV application space, Solar ADEPT defined its technical targets into four
categories: 1) fully-integrated, chip-scale power converters for sub-module applications, 2) package-integrated
microinverters with reduced size and improved performance, 3) lightweight inverters for commercial roof-top
and wall-mount applications, and 4) lightweight, solid-state, medium voltage energy conversion for high-power
applications such as utility-scale inverters with direct grid connection.

In their “Bidirectional GaN-on-Si transistors for more compact and reliable power converters” project Transphorm
aimed to develop GaN-on-silicon based four quadrant switches for residential and commercial solar micro-invert-
ers. Most silicon based switches, with the exception of the triode for alternating current (TRIAC), block voltage
only in one direction when gated off - and are thus unidirectional. Bidirectional switches – also called four quad-
rant switches (FQS) - that allow current flow in both directions when on would enable many new circuit topologies
and applications. Building upon their expertise and experience from the ADEPT program, Transphorm demon-
strated high voltage (both 600 V and 1200 V) FQSs with maximum switching frequencies > 25 MHz for a 400 V
bus. The single monolithic GaN FQS, replacing four discrete Si devices, enabled a 4% improvement in converter
efficiency at low power and 0.2% improvement at full power, approximately a 50% lower power device loss.

The ADEPT and Solar ADEPT program successes were significant in advancing commercial applications of SiC
and GaN devices. However, SiC and GaN device technology remained immature relative to Si and carried a
substantial cost premium, limiting their widespread adoption . Many of the largest opportunities for increased
energy efficiency and reduced energy-related emissions exist in extremely cost conscious industries, including
markets for railway traction drives, automotive applications, and industrial motors. Unit cost for equivalent func-
tional system performance remains a major barrier to the widespread adoption of WBG devices, despite oppor-
tunities related to their superior attributes (including reductions in system costs).

10
SWITCHES
Wide-Bandgap Materials and Devices
WIDE-BANDGAP MATERIALS AND DEVICES – SWITCHES
Analysis of market drivers showed that accelerating the commercial uptake of more efficient approaches enabled
by WBG devices requires driving down the costs of the components. This led to the creation of the Strategies
for Wide Bandgap, Inexpensive Transistors for Controlling High-Efficiency Systems (SWITCHES) program. The
SWITCHES program was aimed at transformational advances that address key materials, device fabrication,
and device architecture issues that drive costs for SiC and GaN devices, as well as to evaluate ultra-wide-band-
gap (UWBG) materials such as a diamond. The goal was to enable the development of high voltage (>1,200V),
high current (100A) single die power semiconductor devices that, upon ultimately reaching scale, would have
the potential to reach functional cost parity with Si power transistors. Additionally, they would offer breakthrough
relative circuit performance through low losses, high switching frequencies, and high temperature operation.
These transformational technologies would have promise to reduce the barriers to ubiquitous deployment of low-
loss WBG power semiconductor devices in stationary and transportation energy applications. From a cost basis
perspective, the barrier to market entry needs to be competitive. Low current silicon switching devices (<50A)
can be purchased as low as ȼ5/A 26. However, this is not the case for high current applications. The SWITCHES
cost target for WBG packaged devices was ȼ10/A for a 100A device, which would make them competitive with
the best silicon IGBT devices in the same class.

The SWITCHES program has made tremendous advances in SiC device fabrication. In their project “Lower
cost SiC manufacturing using existing low-cost, high-volume silicon manufacturing” Monolith Semiconductor,
Inc. was able to made great strides in developing large-area, high-voltage, SiC devices. Before the SWITCHES
program, planar and vertical SiC MOSFET power devices were limited in availability and were typically fabricat-
ed in relatively low volume at dedicated facilities that utilized unique process steps. This resulted in high cost
and prohibitively expensive devices. Monolith Semiconductor partnered with XFab Inc., an automotive qualified
Si Complementary Metal-Oxide Semiconductor (CMOS) manufacturing foundry, to develop and implement an
advanced SiC power MOSFET process on their 150mm diameter silicon production line. Working in an active Si
foundry provides significant cost benefits. Using an existing Si manufacturing infrastructure, the overhead costs
to manufacture SiC devices can be reduced. The innovation in Monolith’s approach centered on adopting com-
mercially available 150mm SiC wafers and designing SiC devices and fabrication processes that are compatible
with existing, high-volume Si manufacturing facilities.27

This allows for highly competitive products, rapidly deployed in the market. Monolith successfully demonstrated
fabrication of MOSFETs on XFab’s production line with blocking voltage of 1700V and a specific on-resistance
as low as 3.1 mΩ-cm2 at room temperature, increasing to 6.7 mΩ-cm2 at 175°C. Clamped inductive switching
characterization of the SiC MOSFETs showed turn-off losses as low as 110 µJ. After 750 hours of gate stress at
a gate bias of VGS=+20V and 175°C, only a 250mV shift in the threshold voltage was observed28. Design chang-
es and processes improvements increased the production line yield of the devices to commercially acceptable
levels which illustrated the promise for high-volume production of reliable SiC MOSFETs on 150mm wafers. The
company anticipates to reach ȼ10/A for their devices in the near future.

The SWITCHES program also made significant advances in GaN device design and fabrication. Before the
SWITCHES program, the majority of GaN power device development had been directed toward lateral archi-
tectures, such as high-electron mobility transistors (HEMTs). There were no vertical GaN devices commercially

26   ARPA-E SWITCHES FOA, https://arpa-e-foa.energy.gov/FileContent.aspx?FileID=1f3b152b-a75a-4118-a6a6-30868377dac8

27   S. Banerjee, K. Matocha, K. Chatty, J. Nowak, B. Powell, D. Guttierrez, and C. Hundley, “Manufacturable and rugged 1.2 KV SiC
MOSFETs fabricated in high-volume 150mm CMOS fab.” Paper presented at 2016 28th International Symposium on Power Semicon-
ductor Devices and ICs (ISPSD), Prague, Czech Republic, June 2016, DOI: 10.1109/ISPSD.2016.7520832
28   K. Matocha, S. Banerjee, and K. Chatty, “Advanced SiC Power MOSFETs Manufactured on 150mm SiC Wafers”, Materials Science
Forum, vol. 858, pp. 803-806, 2016, DOI: 10.4028/www.scientific.net/MSF.858.803
12
available. The lateral devices suffered from well-known issues such as current-collapse, dynamic on-resistance,
inability to support avalanche breakdown29, and usable breakdown voltages of no greater than 650V. Vertical
devices on the other hand offer the possibility to realize the potential of GaN including true avalanche-limited
breakdown. Two projects in the SWITCHES program (Avogy, Inc. and Cornell University) were able to demon-
strate near theoretical, high-power vertical GaN diodes exhibiting very high breakdown voltages (1700 - 4000 V)
and figures-of-merit (VB2/RON) greater than 3 GW/cm2.30,31 The SWITCHES sponsored researchers demonstrated
that vertical GaN devices are avalanche capable32 indicating the ruggedness of such devices in breakdown, a
critical requirement for power switching and rectifying applications. The pathway towards achieving ȼ10/A for
GaN devices was shown to be promising. Projects funded by SWITCHES have demonstrated 80% process yield
for 100A, 1200V p-n junctions. With the current cost of 4” GaN wafers and a die size of 12-16 mm2 for a 100A,
1200V device, vertical GaN devices should be capable of reaching the price range of ȼ5/A to ȼ7/A.

Normally-off vertical transistor devices have been more challenging to fabricate compared to diodes, especially
at high current levels, due to the requirement of selective area doping in many vertical device architectures. At-
tempts at selective area doping in vertical GaN devices in the SWITCHES program resulted in devices with large
junction leakage currents, lower than breakdown voltages expected, and avalanche breakdown ruggedness not
demonstrated. This is the major challenge faced by the SWITCHES projects with vertical GaN architectures. To
overcome the selective area doping limitation in GaN two SWITCHES projects have investigated unique vertical
GaN device designs that reduce or eliminate the need for selective area doping. The University of California,
Santa Barbara demonstrated one such novel vertical GaN devices to achieve both low on-resistance (RON) and
enhancement mode operation in a vertical GaN device called an "in-situ oxide, GaN interlayer-based vertical
trench MOSFET" or OG-FET. In the traditional trench MOSFET structure, a dielectric is deposited on an n-p-n
trenched structure and the channel forms via p-GaN inversion at the dielectric/p-GaN interface. However, this
results in a relatively high RON due to poor electron mobility in the channel. By changing the structure to include
a metal-organic chemical vapor deposition regrown un-intentionally doped GaN interlayer followed by an in-situ,
Al2O3 dielectric cap on the n-p-n trenched structure, a pathway (channel) for enhanced electron mobility is cre-
ated, resulting in reduced resistance. The normally-on OG-FETs fabricated by UCSB demonstrated a threshold
voltage of 2V, breakdown voltages of 990 V (EBR ~ 1.6 MV/cm), and an RON of 2.6 mΩ-cm2.33

The SWITCHES project led by Columbia University “High-performance, low-cost vertical GaN devices through
smaller devices and GaN substrate re-use” demonstrated a GaN vertical fin power field-effect-transistor structure
(VFET) on bulk GaN substrates that addresses the selective area doping limitation of conventional power vertical
GaN transistor device structures. The VFET, shown in Figure 5, consists of fin-shaped channels etched into an
8-μm-thick n- doped GaN drift layer using a combined dry/wet etching technology to achieve smooth fin vertical
sidewalls.34 The current flows vertically from the backside drain contact to source contacts deposited on top of

29   E. Zanoni, M. Meneghini, A. Chini, D. Marcon, and G. Meneghesso, “AlGaN/GaN-based HEMTs failure physics and reliability:
Mechanisms affecting gate edge and Schottky junction,” IEEE Trans. Electron Devices, vol. 60, no. 10, pp. 3119–3131, Nov. 2013, DOI:
10.1109/TED.2013.2271954.
30  I. C. Kizilyalli, A. P. Edwards, H. Nie, D. Disney, and D. Bour, “High voltage vertical GaN p-n diodes with avalanche capability”, IEEE
Trans. Electron Devices, vol. 60, no. 10, pp. 3067-3070, 2013, DOI: 10.1109/TED.2013.2266664
31   K. Nomoto, B.Song, Z. Hu, M. Zhu, M. Qi, N. Kaneda, T. Mishima, T. Nakamura, D. Jena, and H.G. Xing, “1.7-kV and 0.55- mΩ⋅cm2
GaN p-n Diodes on Bulk GaN Substrates With Avalanche Capability”, IEEE Electron Device Letters, vol. 37, no. 2, pp 161-164, 2016,
DOI: 10.1109/LED.2015.2506638
32   O.Aktas, and I. Kizilyalli, “Avalanche capability of vertical GaN pn junctions on bulk GaN substrates.” IEEE Electron Device Letters,
vol. 36, no. 9, pp 890-892, 2015, DOI: 10.1109/LED.2015.2456914
33   C. Gupta, C. Lund, S. Chan, A. Agarwal, J. Liu, Y. Enatsu, S. Keller, and U. Mishra, “In Situ Oxide, GaN Interlayer-Based Vertical
Trench MOSFET (OG-FET) on Bulk GaN substrates” IEEE Electron Device Letters, vol: 38, no. 3, pp. 353-355, 2017, DOI: 10.1109/
LED.2017.2649599
34   M. Sun, Y. Zhang, S. Bedell, J. Barth, D. Sadana, K. Shepard, X. Gao and T. Palacios, “Vertical GaN Power Transistors Using Con-
trolled Spalling for Substrate Heterogeneity” Presented at 3rd Annual SWITCHES Review Meeting, Philadelphia, PA, March 2017
14
 the fins. The sub-micron fin channels are surrounded by metal gate
pads which, below the threshold voltage, pinch-off the channel. In this
vertical transistor design only n-GaN layers are needed, no material
regrowth or p-GaN layer is required.35 Fabricated VFETs demonstrat-
ed specific on-resistance of 0.2 mΩ-cm2, threshold voltage of 1 V,
and a breakdown voltage >1200 V with high ON current (>25 kA/cm2)
and low OFF current at 1200 V (<10-4 A/cm2 ). This amounts to a fig-
ure-of-merit (VB2/RON) of ~7.2 GW/cm2. Large devices demonstrated
high current up to 10 A and breakdown voltage >800 V.36, 37

Additionally, the SWITCHES program has made advances in the area

of epitaxial lift-off of GaN materials. Two SWITCHES projects by the
MicroLink Devices and Columbia University teams focused on GaN
substrate re-use
Figure 5: Scanning electron microscope and thinning.
cross sectional view of a fabricated GaN These projects
VFET (from ref 34). demonstrated
large area (>2”)
layers released from bulk GaN wafers by epitaxial liftoff and
spalling without damage to the lifted off layer. The MicroLink
Devices project “High-performance, low-cost vertical GaN
devices through smaller devices and GaN substrate re-use”
demonstrated wafer-scale epitaxial lift-off (ELO) using an In-
GaN release layer and a bandgap-selective photo-enhanced
wet etching to fully released a GaN foil from 2-inch bulk GaN
substrate as shown in Figure 6.38 Perforations spaced at
1mm apart were used to allow for the wet chemical etch ac-
cess to the InGaN release layer. The leakage current in both
reverse and low-forward-bias regimes for thin-film Schottky
diodes was reduced after ELO processing. This is potentially
due to the elimination of leakage paths through the under-
lying n+ buffer layer that is removed in the ELO process.39
Figure 6: A 5-micron-thick GaN epitaxial film released
On the bulk (free standing) GaN substrate development side, from a 4-inch sapphire substrate using perforations
Soraa Inc. developed, in their OPEN 2009 and SWITCHES on a 1-mm pitch. The yellow luminescence of the
projects, a large diameter ammonothermal reactor capable nitrogen face of the released film is visible under
of more than 600ºC operation and a pressure greater than ultraviolet illumination (from ref 39).

35   M. Sun, M. Pan, X. Gao, and T. Palacios, “Vertical GaN power FET on bulk GaN substrate” Paper presented at 2016 74th Device
Research Conference (DRC), Newark, DE, June 2016, DOI: 10.1109/DRC.2016.7548467
36   M. Sun, Y. Zhang, X. Gao, and T. Palacios, “High-Performance GaN Vertical Fin Power Transistors on Bulk GaN Substrates” IEEE
Electron Device Letters, vol: 38, no. 4, pp509-512, 2017, DOI: 10.1109/LED.2017.2670925
37  Y. Zhang, M. Sun, D. Piedra, J. Hu, Z. Liu, Y. Lin, X. Gao, K. Shepard, and T. Palacios, “1200 V GaN Vertical Fin Power Field-Effect
Transistors” Presented at IEEE International Electron Device Meeting, San Francisco, CA, December 2017
38   C. Youtsey, R. McCarthy, R. Reddy, K. Forghani, A. Xie, E. Beam, J. Wang, P. Fay, T. Ciarkowski, E. Carlson, and L. Guido,
“Wafer-scale epitaxial lift-off of GaN using bandgap-selective photoenhanced wet etching” Phys. Status Solidi B, vol: 254, no. 8, pp
1600774, 2017, DOI 10.1002/pssb.201600774
39   J. Wang, C. Youtsey, R. McCarthy, R. Reddy, N. Allen, L. Guido, J. Xie, E. Beam, and P. Fay, “Thin-film GaN Schottky diodes formed
by epitaxial lift-off” Applied Physics Letters, vol: 110, pp 173503, 2017, DOI: 10.1063/1.4982250
15
3,000 atmospheres in order to grow bulk GaN crystals.40 The ammonothermal GaN crystal growth method is
adapted from the hydrothermal method used to grow quartz crystals, which are very inexpensive and repre-
sent the second-largest market for single crystals for electronic applications (after silicon). Soraa successfully
demonstrated growth of GaN crystals that are over two inches in diameter at a rate of at least 10 microns per
hour, and the fabrication of 2 inch GaN wafers from the crystals (Figure 7).41 The wafers met Soraa’s target spec-
ifications for LED crystal quality, dopant levels, dislocation density, miscut, and surface roughness. Soraa has
also shown that with additional processing steps, they have the ability to make wafers with a dislocation density
less than 1x104 cm-2, a breakthrough that will enable higher-performing power electronics devices with a break-
down field greater than 3 MV/cm for GaN.

In the SWITCHES program advances were similarly achieved in the UWBG material diamond. SWITCHES proj-
ects focusing on diamond by Arizona State University and Michigan State University have demonstrated thick
(>1mm) diamond growth by CVD and doping of diamond with >1020 cm-3 boron and phosphorous for p+ and n+
layers, respectively. Using the advances in diamond growth Schottky and p-n didoes with >1000V blocking and
100-500 A/cm2 forward current were demonstrated.42,43 These achievements have yet to reach program targets,
but are nonetheless foundational in the pursuit of ultra-wide-bandgap semiconductor devices.

The projects in the SWITCHES program have made tremen-

dous advances in materials development, vertical device
architecture, and low cost device fabrication. The SWITCH-
ES program set out to achieve three key aggressive targets:
1200V breakdown, 100A single-die current, and cost of pack-
aged discrete device of no more than ȼ10/A. The program is
drawing to a close by the end of 2017 and while no project
has yet to achieve all the targets of the program, the portfolio
of projects are well underway to achieving these targets com-
munally.

Figure 7: 2 inch GaN crystal and wafer fabricated by Soraa’s

process (from ref 40).

40   R. McCarthy, “High-Power Vertical-Junction Field-Effect Transistors Fabricated on Low-Dislocation Dislocation GaN by Epitaxial Lift-
Off” Presented at 3rd Annual SWITCHES Review Meeting, Philadelphia, PA, March 2017
41   “Lower Cost GaN for Lighting and Electronics Efficiency” Advanced Research Projects Agency – Energy Project Impact Sheet,
(March 2016) https://arpa-e.energy.gov/sites/default/files/documents/files/Soraa_Open2009_ExternalImpactSheet_FINAL.pdf
42   M. Dutta, F. Koeck, W. Li, R. Nemanich, and S. Chowdhury, “High Voltage Diodes in Diamond Using (100)- and (111)- Substrates”,
IEEE Electron Device Letters, vol. 38, no. 5, pp. 600-603, 2017, DOI: 10.1109/LED.2017.2681058
43   S. Nicley, S. Zajac, R. Rechenberg, M. Becker, A. Hardy, T. Schuelke, and T. Grotjohn, “Fabrication and characterization of a corner
architecture Schottky barrier diode structure” Phys. Status Solidi A, vol: 212, no. 11, pp 2410-2417, 2057, DOI 10.1002/pssa.201532220
16
PNDIODES
Addressing Material Challenges
ADDRESSING MATERIAL CHALLENGES - PNDIODES
Many SWITCHES project teams experienced a major obstacle in fabricating vertical GaN power electronic de-
vices specifically the lack of viable GaN selective area doping or selective area epitaxial regrowth processes
that yields material of sufficiently high quality to enable defect-free p-n junctions on patterned GaN surfaces. An
example is shown in Figure 8. In GaN selective area p-type doping has proved elusive, because the most obvi-
ous approaches, such as laterally patterned ion implantation with activation or selective area diffusion of p-type
dopants (e.g. Mg, Be, Zn), have not produced p-type regions or satisfactory p-n junctions. Furthermore, selective
area etch and regrowth approaches have resulted in poor electrical performance not sufficient to be useful in
power electronic applications. A breakthrough is needed to enable high performance vertical GaN transistors.
This specific remaining challenge to the SWITCHES program prompted the announcement in late 2016 of the
Power Nitride Doping Innovation Offers Devices Enabling SWITCHES (PNDIODES) program. The PNDIODES
program aims to develop transformational advances and mechanistic understanding in the process of selective
area doping in the III-Nitride wide-bandgap semiconductor material system. The expectation is this will lead to
the demonstration of arbitrarily placed, reliable, contactable, and generally usable p-n junction regions that en-
able high-performance and reliable vertical power semiconductor devices.

Seven projects were selected for funding as part of the PNDIODES program. The project teams will work to de-
velop transformational advances and mechanistic understanding in the process of selective area doping for GaN
using innovative technologies. Projects led by Arizona State University, Sandia National Laboratories, and Yale
University, will focus on selective area doping using patterned etch and regrowth technology. They will attempt
to obtain a deep understanding of the process, including various etching methods, interface impurity control, and
the effect of crystal growth direction. Projects led by Adroit Materials, JR2J, and State University of New York
Polytechnic Institute, will focus on selective area doping using ion implantation and innovative annealing, or heat
treatment. This will include processes such
as laser spike and Gyrotron annealing to re-
move implantation damage and activate the
dopants. The remaining project led by the
University of Missouri will focus on the de-
velopment of neutron transmutation doping,
exposing GaN wafers to neutron radiation to
create a stable network of dopants within,
to fabricate a uniformly doped n-type GaN
wafer to achieve low resistance substrates.

The Department of Energy and Department

of Defense have identified power electron- Figure 8: GaN Selective Area Doping example.
ics based on wide-bandgap semiconductors
as a major area of concern for energy efficiency and the reduction in size and weight, as well as improvement in
the reliability of power conversion systems. Success in the PNDIODES program would offer innovative options
to help drive research, development, and commercialization of vertical GaN power electronic devices.

17
CIRCUITS
System-Level Advances
SYSTEM-LEVEL ADVANCES - CIRCUITS
Previous efforts by ARPA-E have primarily focused on WBG material and device development without focused
consideration and redesign of the circuit topology. Such solutions do not fully exploit the potential performance
improvements enabled by this new class of power semiconductor devices. The circuit design is critical to the
large-scale implementation of more efficient WBG power systems as a result of their ability to operate at high
voltage, high frequency, and high temperature. For WBG power electronics devices, their benefits will not be fully
realized if they are treated as drop-in replacements for Si devices. Instead, new circuit topologies and designs
are needed that take full advantage of the attributes of the WBG semiconductor devices, resulting in minimization
of form factor, cooling systems, and auxiliary circuit components. The Creating Innovative and Reliable Circuits
Using Inventive Topologies and Semiconductors (CIRCUITS) funding opportunity announcement released in
early 2017 seeks to accelerate the development and deployment of a whole new class of efficient, lightweight,
and reliable power converters based on WBG semiconductors. The program will drive transformational sys-
tem-level advances that enable effective operation at high switching frequency, high temperature, and low loss.
With an explicit focus on novel circuit topologies, advanced control and drive electronics, as well as innovative
packaging, CIRCUITS aims to catalyze disruptive improvements for power electronics afforded by cutting edge
materials such as WBG semiconductors. Such technological breakthroughs would catalyze the adoption of
higher performance power converters in various critical applications such as motor drives, automotive, power
supplies, data centers, aerospace, ship propulsion, rail, distributed energy, and the grid. This will enable signifi-
cant direct and indirect energy savings and emissions reductions across electricity generation, transmission and
distribution, and load-side consumption.

Twenty-one innovative projects were selected for funding as part of the CIRCUITS program. The project teams
will accelerate the development and deployment of a new class of circuit topologies optimally designed for WBG
semiconductors to maximize system performance that will save energy and give the United States a critical tech-
nological advantage in an increasingly electrified economy. CIRCUITS projects will establish the building blocks
of this class of power converter by advancing higher efficiency designs that exhibit enhanced reliability and su-
perior total cost of ownership. In addition, a reduced form factor (size and weight) will drive adoption of higher
performance and more efficient power converters relative to today’s state-of-the-art systems. Some examples
of the projects include:

The Eaton Corporation will develop and validate a wireless-power-based computer server supply that enables
distribution of medium voltage (AC or DC) throughout a data center and converts it to the 48 VDC used by com-
puter servers. The Eaton team has targeted the data center sector, as it is quickly becoming a major consumer
of electricity in the United States. If successful, project developments will reduce U.S. data center energy con-
sumption and operating cost while creating a high-volume commercial market for SiC-based power converters.

Marquette University will develop a small, compact, lightweight, and efficient 1 MW battery charger for electric
vehicles with a switching frequency of 1Mhz. The team aims to use state-of-the-art MOSFET switches based on
SiC to ensure the device runs efficiently while handling very large amounts of power in a small package. This
project endeavors to triple the current state-of-the-art in power density and double specific power of chargers
today. If successful, such a device could help to dramatically reduce charging times for big batteries, like those
in electric vehicles, to a matter of minutes.

University of Arkansas will develop a 2 by 250 kW power inverter system for use in the electrification of heavy
equipment and other higher volume transportation applications (e.g., trucks, buses, cars). The team will leverage
SiC power electronics devices to achieve high levels of efficiency while greatly increasing the volumetric and
gravimetric power density of its system over existing ones. If successful, the team will achieve an improvement
of four times the power density and reduce converter cost by 50% compared to today’s technology.

19
IMPACTS
ARPA-E’s sustained exploration in WBG-based power electronics has helped foster a vibrant ecosystem in WBG
R&D. Prior to ARPA-E PE programs, the Department of Defense44 spearheaded R&D efforts in WBG semicon-
ductors for defense applications. For civilian applications, several DOE offices, including the Energy Efficiency
and Renewable Energy (EERE), Advanced Manufacturing Office45 (AMO), the Office of Electricity Delivery and
Energy Reliability46, and the Vehicle Technologies Program47, are now also working to bring WBG devices closer
to widespread adoption. As of February 2017, ARPA-E has awarded 57 projects related to power electronics,
totaling almost $155 million in federal funding. Together, ARPA-E power electronics projects have published 141
peer reviewed technical papers that have been cited 2,204 times, and have been awarded 26 patents. Nine
teams have cumulatively raised almost $386 million in publicly reported funding from the private sector to bring
their technologies into commercial applications.

ARPA-E’s focus is for transformational change within technology to bridge the gap between the laboratory and
follow on funding from private investment companies. Often, follow on funding will come from other DOE Pro-
gram Offices when the next phase of development still has some risk associated with it. For example, two pro-
grams associated with the EERE SunShot Initiative that include PV power electronics development to enhance
energy efficiency are the 2011 program Solar Energy Grid Integration Systems - Advanced Concepts, $25.9M,
and the 2015 program, Sustainable and Holistic Integration of Energy Storage and Solar PV (SHINES), $15M.
Both program FOAs reference ARPA-E and its associated projects in the field of power electronics as influencing
the aim and scope of the activities.

In 2014, the AMO set up Power America as a manufacturing innovation institute to accelerate the development
of next generation of energy-efficient, high-power electronic devices using wide-bandgap semiconductor tech-
nologies. One of the active members Power America was X-FAB Inc. X-FAB has established a 150mm Silicon
Carbide foundry line in Lubbock, Texas with the support from the PowerAmerica Institute. X-FAB’s goal is to
accelerate the commercialization of SiC power devices by leveraging the economies of scale that have been
established in its silicon wafer fabrication line. In parallel, Monolith Semiconductors Inc., of Round Rock, Texas
received funding from ARPA-E through the SWITCHES program to design their next generation SiC diodes and
MOSFETs. By partnering with X-Fab Inc., they were able to establish a pathway to manufacture their devices
adopting a “fab-less” production model. In March 2017, Littelfuse Inc., a leading manufacturer of electrical circuit
protection equipment based in the USA, made an incremental $15M investment in Monolith Semiconductors
Inc., which gave it a majority ownership position in the company.

In 2015, AMO launched the Next Generation Electric Machines: Megawatt Class Motors programs, and award-
ed $22M to five projects aimed at emerging WBG technologies focused on advancements in large-scale motor
control to increase efficiency in high-energy consuming industries. Previous ARPA-E awardee GE is one of the
award recipients in this program. To build capability in the field of WBG power electronics, AMO also launched
a $6M program in 2015 to improve capability in the US workforce, the DOE Traineeship in Power Engineering;
Leveraging Wide-Bandgap Power Electronics48.

44   “DARPA Sets Tough Goals For The Wide-Bandgap Community,” Compound Semiconductor, November 8, 2002
45   Office of Energy Efficiency and Renewable Energy, Advanced Manufacturing Office, “Wide Bandgap Semiconductor for Clean Ener-
gy Workshop: Summary Report,” (Washington, DC: U.S. Department of Energy, 2012)
46   Office of Energy Efficiency and Renewable Energy, Advanced Manufacturing Office, “Wide Bandgap Semiconductor for Clean Ener-
gy Workshop: Summary Report,” (Washington, DC: U.S. Department of Energy, 2012)
47   Office of Energy Efficiency and Renewable Energy, Vehicle Technologies Program, Multi-Year Program Plan 2011-2015, December
2010 (Washington, DC: U.S. Department of Energy)
48   Conversations with P Gradzki of AMO and EERE website
21
CONCLUSIONS
Despite the remarkable advancements in WBG semiconductors, significant work still remains to realize the full
potential of WBG materials in improving energy efficiency. As mentioned above, fundamental research into ma-
terial properties and processing, and continued development up the power electronics value chain into circuits
and systems, are vital steps in ensuring that America can maintain its technological lead in these promising ma-
terials, and reap the energy benefits through wide-ranging applications.

22
APPENDIX: ARPA-E POWER ELECTRONICS PROJECTS
Table 1: ARPA-E Power Electronics Projects and Their Innovation Approaches
Device / Module /
Program Organization Concept Material
Packaging System
Power converters using GaN-on-Si power
Delphi Automotive
Open 2009 transistor and double side cooling mod- x x
Systems, LLC
ules.
2-inch GaN substrates wafers by am-
Open 2009 Soraa, Inc. x
monothermal for high quality, low cost.
Aligned carbon nanotubes ultracapaci-
FastCAP Systems
Open 2009 tors for increased efficiency and energy x
Corp.
density
Virginia Polytech- Small 3D chip integrating GaN-on-Si and
ADEPT nic Institute and high-frequency soft magnetic material for x x
State University power conversion.
Lower cost and size, vapor deposited
ADEPT General Electric x
magnetic films.
Low cost, lighter weight bidirectional EV
HRL Laboratories,
ADEPT charger using GaN based power transis- x x
LLC
tors
High voltage (15kV) SiC power transistors
ADEPT Cree, Inc. x
that are 50% more energy efficient.
Integrated single chip power converter
Teledyne Scientif-
ADEPT for LEDs using iron magnetic alloys and x x x
ic & Imaging, LLC
GaN-on-Si devices.
Efficient power converter integrating
Virginia Polytech-
high-density capacitors, new magnetic
ADEPT nic Institute and x x x
materials, high-frequency integrated cir-
State University
cuits, and a constant-flux transformer.
Georgia Tech Utility-scale power router that uses an
ADEPT Research Corpo- enhanced transformer to more efficiently x
ration direct power on the grid
GeneSiC Semi- Unique SiC device structure for better
ADEPT x
conductor performance
Arkansas Power
10 times smaller EV charger using SiC
ADEPT Electronics Inter- x x
based power transistors.
national, Inc.
Massachusetts More efficient power circuits for LEDs
ADEPT Institute of Tech- using GaN-on-Si, new magnetic materials, x x x
nology and new circuit designs.
Georgia Tech
Compact power converters using low-cost
ADEPT Research Corpo- x
stacked iron alloys as magnetic cores.
ration
GaN-on-Si transistors for power convert-
ADEPT Transphorm, Inc. x
ers

23
 Device / Module /
Program Organization Concept Material
Packaging System
Case Western Re- Smaller and lighter capacitors using titani-
ADEPT x
serve University um that can store 300% more energy.
Lower cost, smaller, and more efficient
City University of
ADEPT power converters for LED lights using x x x
New York (CUNY)
nanoscale material capacitors.
New nanoscale magnetic material for
Carnegie Mellon
Solar ADEPT lower size, weight, and cost power con- x
University
version.
Transformer-less power conversion device
Solar ADEPT Cree, Inc. x
to directly connect solar power to the grid.
University of Colo- Microconverters that can be integrated
Solar ADEPT x
rado, Boulder into individual solar panels.
Bidirectional GaN-on-Si transistors for
Solar ADEPT Transphorm, Inc. more compact and reliable power convert- x
ers.
SiCLAB, Rutgers Unique high voltage (15kV) SiC power
Solar ADEPT x
University, NJ transistors for better performance.
New power conversion technique to
SolarBridge Tech-
Solar ADEPT efficiently and cost-effectively improve the x
nologies, Inc.
energy output of PV power plants.
Bi-directional silicon power switches for
Solar ADEPT Ideal Power, Inc. reduced size, weight, and cost PV invert- x x
er.
Silicon Power High-power and high-voltage optically trig-
Open 2012 x
Corporation gered bi-directional SiC transistor switch.
High-Voltage, Bi-Directional MOS-Grated
Rensselaer Poly-
Open 2012 SiC Power Switches for Smart Grid Utility x
technic Institute
Applications.
Innovative GaN device design for utili-
Open 2012 RamGoss, Inc. x x
ty-scale electronic switches.
High-voltage AlN devices for use in
Open 2012 Hexatech Inc. x x
high-power electronics.
Georgia Tech
Graphene based supercapacitor for in-
Open 2012 Research Corpo- x
creased energy density.
ration
High Voltage, High Power Gas Tube Tech-
Open 2012 General Electric nology and Impact Assessment for HVDC x
Transmission.
High-performance, low-cost vertical GaN
Columbia Univer-
SWITCHES devices thru smaller devices and GaN x x
sity
substrate re-use.
SixPoint Materi- High-quality, low-cost GaN substrates
SWITCHES x
als, Inc. produced via ammonothermal growth.

24
 Device / Module /
Program Organization Concept Material
Packaging System
Lower cost SiC manufacturing using exist-
Monolith Semi-
SWITCHES ing low-cost, high-volume silicon manu- x
conductor, Inc.
facturing.
High-performance, low-cost vertical GaN
SWITCHES MicroLink Devices devices thru smaller devices and GaN x
substrate re-use.
Unique GaN device structure for signifi-
SWITCHES Cornell University cantly smaller size and higher perfor- x
mance.
Kyma Technolo- High-rate large area GaN substrate
SWITCHES x
gies, Inc. growth
Michigan State High-voltage diamond devices for use in
SWITCHES x x
University high-power electronics.
High yielding vertical GaN transistor for
SWITCHES Avogy, Inc. x
lower cost.
Arizona State Low-cost, vertical, diamond bipolar device
SWITCHES x x
University for use in high-power electronics.
iBeam Materials, Low-cost GaN LEDs on flexible metal
SWITCHES x x
Inc. foils.
Follow-on to OPEN2009 project. Large-ar-
SWITCHES Soraa, Inc. x
ea, high-quality, low cost GaN substrates
High-performance, low-cost vertical GaN
HRL Laboratories,
SWITCHES devices at higher power levels than lateral x
LLC
devices.
Fairfield Crystal Unique high-rate GaN boule growth for
SWITCHES x
Technology, LLC lower cost GaN wafers.
University of High-performance, low-cost vertical GaN
SWITCHES California, Santa devices at higher power levels than lateral x
Barbara devices.
High-voltage, solid-state SiC field-effect
Open 2015 General Electric x
transistor charge-balanced device.
Tibbar Technolo- Plasma-based 3-phase AC to DC convert-
Open 2015 x
gies er.
GeneSiC Semi- High-power and voltage vertical GaN
IDEAS x
conductor bipolar junction transistor.
Electromagnetic induction-based static
University of Ne-
IDEAS power converter for efficient low cost AC x
braska, Lincoln
to AC electrical conversions.
Northeastern Innovative universal power converter us-
IDEAS x
University ing SiC based devices for decreased size.
Quora Technolo- Reliable, high-power and voltage, innova-
IDEAS x
gy, Inc. tive lateral GaN transistor.

25
 Device / Module /
Program Organization Concept Material
Packaging System
High-voltage, high-power density, hybrid
Sandia National
IDEAS switched-capacitor DC-DC power convert- x
Laboratory
er using SiC and vertical GaN devices.
University of Colo- Capacitive wireless power transfer archi-
IDEAS x
rado, Boulder tecture to dynamically charge EVs.
Sandia National MVDC/HVDC Power Conversion with
IDEAS x x
Laboratory Optically-Controlled GaN Switches.
Transistor-less Power Supply Technology
Harvard Univer-
IDEAS Based on UWBG Nonlinear Transmission x
sity
Line.
Adriot Materals Selective Area Doping for Nitride Power
PNDIODES x
Inc. Devices.
Effectuve selective area doping for GaN
Arizona State
PNDIODES Vertical Power Transistors Enabled by x
University
Innovative Materials Engineering.
Laser spike anneal technology for the
PNDIODES JR2J activation of implanted dopants in Gallium x
Nitride.
Sandia National High voltage re-grown GaN P-N Diodes
PNDIODES x
Laboratory enabled by defect and doping control.
The Research
Foundation for Demonstration of PN-junctions by implant
PNDIODES x
State University of and growth techniques for GaN.
New York
High quality GaN FETs through trans-
University of Mis-
PNDIODES mutation doping and low temperature x
souri
processing.
Regrwoth and selective area growth of
PNDIODES Yale University x
GaN for vertical power electronics.

26
https://www.arpa-e.energy.gov/