Reliability challenges in 3D NAND Flash memories

摘要

The reliability of 3D NAND Flash memory technology is depending on many factors. Most of them are related to the process-induced variability of the layers. Endurance, data retention capabilities, and cross-temperature immunity are the metrics that become affected by this, turning in peculiar reliability challenges that are difficult to be tackled without the aid of system level solutions. In this work we show how some of the Flash Signal Processing techniques implemented in memory controllers like machine learning and read oversampling algorithms can help in the overall improvement of the 3D NAND Flash reliability.