Charge-based nonvolatile memory: Near the end of the roadmap?

The floating gate cell has been the workhorse for nonvolatile memories for over 40 years now. Indeed, the original concept as proposed by Kahng and Sze in 1967 [1] has survived the subsequent generations of nonvolatile memory types going from EPROMs and One-Time-Programmable (OTP) memories to EEPROMs and finally ending with the major success technology based on the same concept: Flash memory. The success of Flash has recently made it into the technology driver for the entire silicon industry, not in the least in the area of lithography.However, with this major success, final scaling limitations come into sight at an increasingly more rapid pace and several red brick walls have been identified on the not-so-far horizon. This paper will elaborate on these limitations and challenges and will also propose solutions or onsets of solutions to allow for further conceptual improvement in the area of these charge-based memories with long-term data retention.

► This review paper summarizes the main challenges for further Flash scaling in a consistent way.
► Solutions are given for the different challenges that have been identified including imec-specific results such as hybrid and ultimate floating gate cells.
► Finally, the paper provides a personal roadmap for extending the lifetime of charge-based memory as well as the potential entry points for emerging memory technologies such as phase change memory and Ressistance RAMs.