On the Memory Window Variability in a 3-D Multi-Granular Ferroelectric FET Including Grain Boundaries

This paper analyzes the variability in the memory window of a ferroelectric FET induced by distinct grain patterns. The Poisson Voronoi algorithm nucleates the random grain texture in MATLAB, which is coupled with TCAD-Sentaurus to obtain the Id - Vg characteristics. The grain texture with boundaries exhibits larger variability than the grain texture without boundaries. The memory window decreases with the emergence of grain boundaries. Furthermore, a multi-grain pattern along the ferroelectric thickness direction is also incorporated in the study. The multigranular ferroelectric layer possesses a higher variability than the mono-grain layer. Additionally, linearly increasing and decreasing polarization profiles are assumed in the ferroelectric grains, and such polarization distribution's impact on the memory window variability is analyzed. The random nucleation of grains in the ferroelectric layer establishes a replica of the experimental growth mechanism.