Transphorm Releases Quality and Reliability data
Early Life Failure and Field Failure rates provides industry’s first complete validation set for 650V GaN
Transphorm, a US developer of high reliability 650V GaN semiconductors, has announced the industry’s first complete validation data set for GaN power FETs in the 600 V and higher range.
This information expands on Transphorm’s December 2018 announcement indicating that the company has shipped over 250,000 GaN FETs to date. What follows provides details behind public Transphorm customer statements citing reliability as a major influencer behind their GaN device selection.
“In the power conversion market—quality, reliability, and performance are the three main factors used when vetting transistors. We’ve always operated with the mindset that prioritising quality and reliability will result in high voltage GaN’s success,” explained Philip Zuk, VP of technical marketing worldwide, Transphorm. “We’re proud to see that strategy deliver positive results and felt it important to release the validation data so that potential and existing customers can understand GaN’s true capabilities.”
Availability of the two new data types, Early Life Failure (ELF) and Field Failure, that round out the validation data set marks another major milestone for high voltage GaN technology. They further position Transphorm’s GaN reliability as competitive to and poised to likely surpass that of alternative solutions - Silicon and SiC - given that Transphorm’s power conversion technology is in its first maturation stage whereas Silicon transistors have long-since matured and SiC is a decade into its development.
High Voltage GaN Reliability: A Complete View
Transphorm’s complete data set includes five components of product reliability:
- Product Qualification: defined by JEDEC and AEC-Q101 Standards.
- Testing beyond standard requirements:includes high voltage switching, single event burnout (SEB), HTOL, and HTGB testing at elevated temperatures and voltages.
- Intrinsic Lifetime: measures device’s “wear-out” lifetime; also defines failure modes and acceleration factors.
- Extrinsic Lifetime or ELF:forecasts Field Failure rates in Failure in Time (FIT) or Parts per Million (ppm) per year rates; used for warranty calculations.
- Field Failure:measures device’s actual field performance in customer applications.
“Intrinsic data alone is not enough,” said Ron Barr, VP of quality and Reliability, Transphorm. “Intrinsic testing gives us the acceleration factors that we use in conjunction with Early Life Failure testing to determine the product’s Infant Mortality rate. This makes it easy for customers to accurately vet GaN devices. Pairing intrinsic data with extrinsic data and Field Failure rates provides a completebaseline for how GaN FETs will perform.”
Extrinsic Early Lifetime Failure
Intrinsic failure rates give unrealistically optimistic views of product reliability. Also called Infant Mortality, ELF provides the most realistic view via FIT and ppm rates. Early Lifetime Failure assesses potential defects in materials, design, and process control that may cause parts to fail. Notably, ELF causes most customer warranty claims and typically occurs sooner and at higher rates than wear-out failures. Given this, customers use ELF data to determine warranty risks and costs.
Transphorm’s ELF:
FIT: 0.45
ppm: 4
Field Failure
Field Failure measures the number of devices that fail in customer systems in production in relation to the total number of parts sold. Transphorm has shipped over 250,000 FETs and accumulated over 1.3 billion field hours of operation resulting in the following Field Failure rates:
FIT: 3.1
ppm: 27.4 [conservative estimate]
Transphorm’s Field Failures align with that of SiC, which is reported to be less than 5 FIT. Further, Transphorm’s ppm rate continues to decrease over time regardless of application, suggesting that reliability is better than currently reported.
Intrinsic Lifetime
Intrinsic Lifetime is essentially a device’s theoretical lifetime assuming that material wear-out is the only contributor to the part’s longevity. The data is created using the “Physics of Failure” methodology, which involves measuring time to failure when stressing parts with voltage and temperature, and building related models used to predict ultimate lifetime.
Transphorm’s Intrinsic GaN wear-out:
Mean Time Before Failure [MTBF - is the time when 63.2 percent of the population has failed]: 1e11 hours [11M+ years]
Lifetime [wear-out phase of the devices where 100 ppm/year is used]: 100M hours [11,415 years]
