Monte-Carlo Simulation of Electromigration Failure Distributions of Submicron Contacts and Vias: A New Extrapolation Methodology for Reliability Estimate

We have studied the electromigration failure distributions as a function of current density, and we shoe experimentally that the forms deviate from lognormal distributions as j approaches critical current density. The form change in failure distributions can be well described by Monte-Carlo simulation based on our physical electromigration failure model. The model predicts that median time to fail (MTF) and failure time dispersion (sigma) approach infinity as j -> j sub c, and we show that this behavior results from a change in the functional form of failure with current density as j -> j sub c. We propose a new methodology for the extrapolation of contact and via electromigration data to account for the change in the form of the failure distribution.