Application of the ferroelectric materials to ULSI memories

Abstract

Memory is essential to electronic data processing and continuous efforts are being made to develop improved memory devices. In the era of VLSI, difficulties have arisen with respect to storage capacitance, which must be kept to a certain value while the device dimensions are reduced. This has prompted the adoption of complicated structures like the trench or stack, causing the number of process steps to be increased. The use of high dielectric constant materials has been researched for the extension of DRAM development. Recently, the development of the memories which use the polarization reversal current of the ferroelectric material is rapidly progressing because it enables high speed nonvolatile memory action which generally needed in recent electronic systems. These memories will replace a large portion of the existing memory systems in the near future. However, this is not a perfect solution to the problem, because they are not in accordance with the scaling rule. In this paper, it is shown that ferroelectric memories using the field effect current of a semiconductor by the remanent polarization of the ferroelectric material are in accordance with the scaling rule. The first experimental verification of the non-volatile memory action was reported by Moll and Tarui in 1963 [1]. This basic memory action has been successively used in MFS (metal-ferroelectric-semiconductor) transistors. The ferroelectric memories are nonvolatile and are expected to be high-speed devices, making them suitable for universal applications. However, it is necessary to optimize the interface between the semiconductor and ferroelectric material. Experiments for the prospective devices using CeO 2 or Ce x Zr (1− x) O 2 as the buffer insulator layers of the MFIS (metal-ferroelectric-insulator-semiconductor) transistors are described.