Neural network approach for ferroelectric hafnium oxide phase identification at the atomistic scale

• DOI: 10.1016/j.mtelec.2023.100027
• Publication Date: 2023-03-02T08:16:05Z
• Authors: Zhiheng Cheng; Xingran Xie; Yimin Yang; Chaolun Wang; Chen Luo; Hengchang Bi; Yan Wang; Junhao Chu; Xing Wu

The hafnia-based ferroelectric oxides with excellent negative-capacitance properties offer a great opportunity to develop high-performance integrated circuits. nanosized multiphase distribution of Hf0.5Zr0.5O2 (HZO) has significant influence on its properties. Transmission electron microscope (TEM) an atomistic resolution could establish the structure-property relationship and guide performance improvement HZO by identifying phase structures. However, high throughput TEM data complexity interpretation make quantitatively extracting physical chemical information from images challenging low-efficiency. Here, we automatic work flow for analysis, which greatly enhances efficiency processing. By interest area training neural network ResNet18, accuracy determination reaches 95.82% low computational cost. Theoretical analysis is conducted unveil advantages ResNet18 network. approach provided in this promote quantitative high-throughput pave way future on-line image stream real-time.