Domain‐Engineered Flexible Ferrite Membrane for Novel Machine Learning Based Multimodal Flexible Sensing

Abstract

Abstract Flexible materials and devices that can simultaneously reflect multimodal information are highly desired for novel flexible electronics and intelligent flexible sensing systems. In this regard, flexible magnetic films have great potential for wireless multimodal flexible sensor due to the curvature and azimuth angle‐dependent ferromagnetic resonance. However, a key challenge now is to build the precise relationship among the mechanical bending, azimuth angle, and the ferromagnetic resonance of the film, which involves multi‐physics and coupled process. In this work, the physical problem is solved by combining material engineering and machine learning. Material domain engineering is applied to form localized multi‐peak ferromagnetic resonance features for increasing sensitivity. Besides, convolutional neural network algorithm is utilized to help recognize the bending and azimuth angle modulated ferromagnetic resonance in flexible film systems. It is found that the bending information for the flexible film with engineered domain structure can be mapped to the ferromagnetic profile with accuracy over 99%, while the accuracy sharply decreases to less than 50% in the control group of high‐quality film. This study provides a versatile platform for developing machine learning‐based novel sensing materials.

Keywords

• epitaxial oxide thin film
• flexible devices
• flexible spintronics
• machine learning
• magnetic materials