The Past, the Present, and the Future of Ferroelectric Memories

Abstract

Ferroelectric materials are characterized by two stable polarization states that can be switched from one to another by applying an electrical field. As one of the most promising effects to realize nonvolatile memories (NVMs), the application of ferroelectrics in NVMs has been studied since the 1950s. In principle, three different ways to read out the ferroelectric polarization are known: measuring the charge-related current that flows during switching of the ferroelectric, measuring the polarization-dependent tunneling current in very thin ferroelectric layers, and measuring the threshold voltage shift of a ferroelectric field effect transistor caused by the polarization change of the ferroelectric integrated into the gate stack. While early attempts used bulk ferroelectric crystals, the first commercial success was reached when the concept was integrated into a MOS process. However, all materials that were known to exhibit ferroelectricity had a very complicated structure, thus making the integration troublesome and leading to a very slow scaling and limiting its application to niche markets. With the discovery of ferroelectricity in hafnium oxide in 2011, the new impetus came into the field for all three variants described above. This article will describe the history of ferroelectric memories and its current status both with respect to the commercialization of ferroelectric memories based on traditional ferroelectric materials and the ongoing research and development activities employing the more recently discovered ferroelectricity in hafnium oxide. This finally leads us to an outlook of the future challenges for ferroelectric memories.