A Study Of Scaling Effects On DRAM Reliability
Abstract: 2011 IEEE. Personal use of this material is permitted. However, permission to reprint/republish this material for advertising or promotional purposes or for creating new collective works for resale or redistribution to servers or lists, or to reuse any copyrighted component of this work in other works must first be obtained from the IEEE.
In this study, commercial 512Mb Double Data Rate Synchronous Dynamic Random Access Memory (DDR SDRAM) modules from three progressive technologies – 130nm, 110nm and 90nm - were selected for experimentation to investigate degradation trends as a function of scaling. High temperature, high voltage accelerated stress testing was performed to characterize DRAM reliability and failure rates. Retention time degradation over time as a function of stress was also studied.
For each technology generation, two distinct soft error populations were observed: Tail Distribution, characterized by randomly distributed weak bits with Weibull slope =1, and Main Distribution with Weibull slope greater than 1. Retention time was found to degrade exponentially with time. Analysis reveals multiple failure mechanisms are involved in retention time degradation. Activation energy was found to change with stress temperature for all three technologies.
There are several observations with regard to scaling effects on DRAM reliability. First, the smaller the technology, the larger the operating current increases in percentage after high temperature, high voltage accelerated stress. Second, cell retention time variation decreases as technology scales down. Third, 90nm DRAM has the largest soft-error failure rate among three technologies under equivalent stress, 110nm DRAM has better reliability performance than 130nm at 55°C and 75°C, and 130nm DRAM is the best at 125°C. Studies continue into the scaling effects on reliability of progressive DRAM technologies.
Keywords: Accelerated Life Testing - Failure Rate - Product Reliability - RAMS 2011 Proceedings - Reliability Analysis/Prediction/Estimation