Reliability and physics of failure in eGaN devices: Part 1

March 27, 2016 // By Alex Lidow
Alex Lidow
Alex Lidow, Chris Jakubiec and Robert Strittmatter discuss their company's enhancement-mode gallium nitride (eGaN) FETs.

Efficient Power Conversion (EPC) Corp.’s enhancement-mode gallium nitride (eGaN) FETs continue to expand into new market applications due to the competitive performance advantages over traditional power MOSFETs. Wireless power, DC-DC conversion, RF base station transmission, satellite systems, audio amplifiers, and LiDAR are just a few example applications that can take advantage of the superior performance of eGaN FETs.

Thirty years of silicon power-MOSFET development taught us that one of the key variables controlling the adoption rate of a disruptive technology is whether or not the product is reliable enough to use in the application. This principle has guided the design of EPC’s enhancement mode gallium nitride devices.

In this series we will look at the various ways the reliability of eGaN technology has been validated, and how we are developing models from our understanding of the physics of failures that can help predict failure rates under almost any operating condition. Detailed reports of the reliability studies and testing results discussed in this series can be found here.

In this first installment and the next, we will look at the field experience from the past six years of GaN transistors use in a variety of applications from vehicle headlamps to medical systems to 4G/LTE telecom systems. Diving into the failure of each and every part leads to some valuable lessons learned.

In the third installment, we will review the basic qualification testing results from over seven million actual hours of maximum stress on over 10,000 parts. Results from traditional electrical stress, environmental stress, and thermo-mechanical stress testing demonstrates that GaN technology can certainly pass any of the tests required of MOSFETs. We also will review early life failure rates based on several thousands of parts tested. However, unless we understand the underlying physics of failure we do not know how to extrapolate these maximum-stress results to real-world stress conditions.

The fourth and fifth installments will discuss the basic physics behind