An extended linear hazard regression model with application to time-dependent dielectric breakdown of thermal oxides

We develop a generalized Extended Linear Hazard Regression (ELHR) model with linear time-varying coefficients to estimate reliability under normal operating conditions using failure time data obtained from accelerated conditions. The model considers the effect of proportional hazards, time-varying coefficients and also time-scale changes. Extensive simulation experiments demonstrate that the ELHR model provides more accurate reliability estimates than those obtained using existing accelerated life testing models. We utilize the ELHR model to study the time-dependent dielectric breakdown of thermal oxides on n-type 6H-SiC using laboratory data. The results show that oxide reliability considerations effectively preclude SiC MOS devices from many high-temperature applications.

Acknowledgements

The authors would like to thank Professor J. A. Cooper of Purdue University for providing us with a valuable reference and the real laboratory data. We would like to thank the referees for their helpful comments and suggestions. We would like to acknowledge the financial support of the NSF/Industry-University Cooperative Research Center on Quality and Reliability Engineering, Rutgers University-Arizona State University.

Notes

*Extrapolated from Fig. 5.