Monte Carlo Prediction of PPM Failure Rate Using a Parametric Reduced Order Model
- Resource Type
- Conference
- Authors
- Bornoff, Robin; Luiten, Wendy
- Source
- 2020 26th International Workshop on Thermal Investigations of ICs and Systems (THERMINIC) Thermal Investigations of ICs and Systems (THERMINIC), 2020 26th International Workshop on. :1-5 Sep, 2020
- Subject
- Components, Circuits, Devices and Systems
Integrated circuits
Temperature distribution
MOSFET
Histograms
Monte Carlo methods
Thermal conductivity
Reduced order systems
- Language
- ISSN
- 2474-1523
Parametric thermal reduced order models of a power semiconductor device are exercised in a Monte Carlo framework. Normal distributions of some critical thermal conductivities are considered, and the resulting histogram distribution of junction temperature predicted. Considering a maximum junction temperature constraint, a parts per million (PPM) failure rate may be determined by relating the total number of Monte Carlo junction temperature predictions that exceed this constraint to the total number of Monte Carlo simulations performed. The relationship between the number of Monte Carlo samples and the predicted failure rate is investigated.