Evaluation of the predictive capability of Berkeley's hot-carrier reliability model

A model for hot-carrier reliability prediction developed at the University of California is widely used in industry. Confusion exists, however, as to how the model is to be applied. The model's devisers suggest that hot-carrier degradation is most rapid when substrate current is maximum, yet the model itself is in strong disagreement with this. Since the minimum lifetime for a given drain voltage is of most interest, the minimum prediction of the model needs to be evaluated. The necessity of using a novel hot-carrier stress method in the evaluation is demonstrated.