Qualifying reliable systems from unreliable components

Summary form only given. In an era of increasing performance targets and ever more stringent power requirements, design margins are shrinking more than ever. It is therefore critical to be able to understand the factors which impact product reliability and to accurately predict their effects. Overestimate the impact of defects, variability, NBTI, HCI, TDDB and EM on the product, and critical performance or power is left on the table. Underestimate these effects and the product suffers an unacceptable failure rate. Further complicating this picture is the fact that the relative importance of these different degradation mechanisms can change drastically from one process generation to the next. It is shown how a combination of careful modeling combined with product measurement data is used to set the reliability screens and margins. The role of design in building-in reliability is then discussed to help qualify reliable systems from unreliable components.