Synaptic memristors based on BaTiO3 thin films irradiated by swift heavy ions for neuromorphic computing

Minghui Xu; Tao Liu; Hailian Li; Yong Liu; Pengshun Shan; Ruowei Wang; Weijin Kong; Minghao Zhang; Shuangqing Fan; Jie Su

doi:10.1039/d4mh00716f

Synaptic memristors based on BaTiO₃ thin films irradiated by swift heavy ions for neuromorphic computing

Mater Horiz. 2024 Oct 28;11(21):5429-5437. doi: 10.1039/d4mh00716f.

Authors

Minghui Xu¹, Tao Liu¹, Hailian Li¹, Yong Liu², Pengshun Shan¹, Ruowei Wang¹, Weijin Kong¹, Minghao Zhang¹, Shuangqing Fan¹, Jie Su^{1

3}

Affiliations

¹ College of Electronic and Information Engineering, College of Physics Science, Qingdao University, Qingdao 266071, China. [email protected].
² School of Energy and Power Engineering, Shandong University, Jinan 250061, China. [email protected].
³ National Laboratory of Solid State Microstructures, Physics Department, Nanjing University, Nanjing 210093, China.

PMID: 39188220
DOI: 10.1039/d4mh00716f

Abstract

Swift heavy ion (SHI) irradiation is an effective method for modulating the properties of thin oxide films by introducing defects, strains, and structural transformations. Here, we applied 516 MeV Xe³¹⁺ irradiation to BaTiO₃ (BTO) thin films grown on Nb:SrTiO₃ substrates to induce the generation of tracks and nanohillocks. Memristors with BTO films irradiated at a fluence of 5 × 10¹⁰ ions cm^-2 displayed excellent retention and endurance characteristics. Moreover, the memristors exhibited highly stable synaptic plasticity functions such as excitatory/inhibitory post-synaptic currents (E/IPSC) and paired-pulse facilitation/depression (PPF/D). The memristors achieved a discrimination accuracy of 92.5% on given handwritten digit data by an artificial neural network with supervised learning. These results verify that the judicious application of SHI irradiation on thin oxide films is a viable strategy for exploring neuromorphic computation.