Optimal Machine Learning Model for the Relationship Between Grain Size, Channel Thickness, and Grain Boundary Trap Density in 3D NAND Strings

Abstract

The rapid growth of the semiconductor industry has resulted in the development of three-dimensional NAND (3D NAND) strings, which offer increased memory density and greater performance over classical planar NAND designs. In this work, we look at the critical aspects that determine the performance and reliability of 3D NAND strings. Using simulation results and machine learning algorithms we focus on the effects of grain size, channel thickness, and grain boundary trap density. We address their individual and cumulative effects on memory cell behavior, with the current state-of-the-art strategies used to optimize these characteristics. This work will be helpful in the ongoing development of 3D NAND flash technology.