A research team led by Ken Shepard, professor of electrical engineering and biomedical engineering, has won a $3 million three-year grant from the U.S. Energy Department's Advanced Research Projects Agency (ARPA-E) for research targeted at developing next-generation power conversion devices that could dramatically transform how power is controlled and converted throughout the grid. His award is one of 14 that are receiving a total of $27 million from the agency's SWITCHES program (Strategies for Wide-Bandgap, Inexpensive Transistors for Controlling High-Efficiency Systems) to find innovative ways to lower the cost and improve the efficiency of power electronics.
In modern energy infrastructure, existing power electronics are based on decades-old technologies and rely on expensive and bulky components. To address these inefficiencies, SWITCHES seeks to lower the cost and improve the energy efficiency of power switching devices used in a variety of power-conversion applications, including data centers, electrical vehicles, and photovoltaics.
"We are really excited to win this funding," says Shepard, who, working with colleagues at MIT, IBM, and Veeco Instruments, is developing a new method to fabricate vertical gallium nitride (GaN) devices in a low-cost matter that would be compatible with traditional silicon semiconductor manufacturing. "We have assembled a world-class industrial-academic team bringing together expertise in circuits, devices, and materials."
He notes that Moore's Law scaling, which states that the number of transistors on integrated circuits doubles approximately every two years, has had tremendous impact on the design of communications and computation devices, allowing smaller, cheaper, and higher performance systems. However, power electronics--electronics that are responsible for the delivery of energy--have largely been unaffected by these trends.
"One of the reasons power electronics have not benefited from Moore's Law is the inability to scale the size and performance of silicon-based power transistors," Shepard explains. "To operate at high voltages, these devices have to be made large and slow. Wide-bandgap semiconductors such as GaN have the potential to make scaled transistors that can operate at high voltages and powers at fast switching speeds."
Shepard's team is using spalling, a method to transfer entire GaN devices to alternate substrates or bases, to create low-cost vertical GaN devices and circuits. The group plans to spall entire fabricated transistors from GaN wafers onto lower-cost silicon substrates and to interconnect the supporting silicon substrates, enabling small-scale integration of its GaN devices.
"Our GaN transfer method will enable the use of low-cost, high-power transistors for kilowatt-scale applications," Shepard adds. "We hope to construct an integrated half-bridge boost converter module at a potential production cost of less than $20. Our success in this program will allow us to bring aspects of Moore's Law scaling to an important market--power electronics--that has not previously seen the benefits in cost, form factor, and performance that scaling brings."