Scaling DC lifetests on GaN HEMT to RF conditions

Abstract

The assumptions behind a new lifetesting approach are documented, evaluated, and tested where possible. This approach utilizes “signature parameters” to track individual degradation mechanisms in both DC and RF lifetests, and determines the mean time to failure (MTTF) Arrhenius curves for each mechanism individually, during RF operation. This is important for GaN HEMT because most studies indicate that several mechanisms contribute to its wearout, simultaneously, and this makes it impossible to extrapolate conventional RF wearout curves to other temperatures or longer times. A key assumption is that degradation mechanisms can be identified, and associated with unique signature parameters. A second key assumption is the integrity of degradation mechanisms, whether they occur under steady DC biases, or under oscillating RF biases. This allows us to deduce MTTF's under RF operation from the ratio of degradation rates of the individual mechanisms in DC lifetests, and the rates in RF operation, integrated over the RF waveform. This can be found by means of DC and RF lifetests, monitoring the signature parameters. Then the DC Arrhenius curves can be scaled to the RF conditions. After evaluation of these, and several other assumptions, we find the net uncertainties for one of the GaN HEMT technologies that we used for development of our approach. They amount to — 30%, +100% in MTTF; this is entirely adequate for high-reliability parts evaluation, where a margin of at least 10× (900%) is required.