神经形态工程学
铁电性
钛酸钡
材料科学
薄膜
纳米技术
光电子学
计算机科学
人工智能
电介质
人工神经网络
作者
Chenxi Wang,Lin Guo,Junjie Hu,Titao Li,Fangping Zhuo,Hong‐Hui Wu,Xiaoqiang Lu,Min Zhu
摘要
The growing interest in ferroelectric materials has witnessed the thriving prospect of bio-inspired artificial neuromorphic system, where multi-level polarization states play a crucial role. In this work, with typical BaTiO3 ferroelectric thin film as the model system, we explore the physical effects of inhomogeneity on polarization switching dynamics and neuromorphic performance. Inhomogeneous films exhibited pinched polarization–electric field hysteresis loops, leading to a high recognition accuracy of 96.03% for hand-written digits, compared to about 10.31% for homogeneous films. The inhomogeneity in switching dynamics was analyzed by inhomogeneous field mechanism. Diffusive distributions of switching time and local electric fields were observed, aligning with experimental results and the expected inhomogeneity. The prolonged domain wall depinning time and lowered energy consumption suggest the potential for multi-level polarization states, a possibility further confirmed by phase-field simulations that demonstrated their presence during long-term potentiation/depression. Our work highlights the positive influence of inhomogeneity in enhancing the performance of ferroelectric-based neuromorphic systems.
科研通智能强力驱动
Strongly Powered by AbleSci AI