Ferroelectricity is the very well-studied phenomenon for the last 100 years, even though it is less known than its famous brother, ferromagnetism, which is known to people in every civilizations from the ancient years as a form of permanent magnet. Despite the different physics involved, the characteristics of two phenomena, ferroelectricity and ferromagnetism are astonishingly similar, in that the external electrical or magnetic field causes the flipping of the internal material structure to generate spontaneous electrical or magnetic polarization. As the states of the spontaneous polarization can be maintained even after the external field disappears, information can be stored, which is the characteristics of “memory”.

The most common ferroelectric materials are metal oxides with a perovskite structure, such as BaTiO3, PZT, SrTiO3, CaTiO3…. etc. Ti cation moves between two stable positions, which define up and down polarization states. Utilizing perovskite ferroelectric for fast and nonvolatile random access memory gained much attention in early 2000. The effort was partially successful for FRAM array with tens of Mb were implemented in SOCs and a few embedded systems. Despite the ambitious original goal to replace DRAM, the early FERAM could not be successful as high density mainstream memory due to lack of scalability of Perovskite materials. They lose ferroelectric property when the thickness becomes thinner than ~100nm. The reason is ironically a huge spontaneous polarization value, which causes de-polarization field to become bigger in a thin structure and eventually nullify the ferroelectricity.

In 2010s, it has been realized that Hf based ferroelectric materials can be scaled down to ~5nm to be integrated in the modern nano-scale devices. Since then, there have been enormous interests in ferroelectricity for high density memory and logic devices. Hf based ferroelectric materials were also blessed by the development of atomic layer deposition (ALD) technique, which allows to create thin multi component layers with precise atomic level layer-by-layer deposition control. The composition and thermal process can be tuned to have hysteretic ferroelectric, to low-field hysteresis-less anti-ferroelectric and negative capacitance. Such a versatile electrical property is beneficial for various applications from memory to logic type applications. 

There are many applications for Hf based ferroelectric layer for DRAM-like fast memory, NAND-like high capacity memory, and the logic types using the charge boosting effect. Recently, the possibility to use FE-FET for machine learning, has been reported, as it uses FE-FET multi-states to store and retrieve the weights of the states. 

In the RTC, we are interested in research of ferroelectric materials to extend DRAM and NAND technologies beyond the scaling limits. We published a paper on 1T-1C FE-RAM utilizing HfZrOx ferroelectric material integrated in highly scaled DRAM node in IEDM 2021. We also demonstrated HfSiOx based low voltage 3D Ferro-NAND in IMW 2022. Even though our main interest is to use properties of ferroelectric for memory, our research area is not limited to traditional memory devices. As ferroelectric devices have been of great interest in recent years, we are open to explore various areas for emerging applications as well as traditional memory applications.