Litcius/Paper detail

One-Step Annealing-Configured Hf <sub>0.2</sub> Zr <sub>0.8</sub> O <sub>2</sub> Memristive–Antiferroelectric Devices for Bioinspired CSNN Neuromorphic Computing

Jinhao Zhang, Kangli Xu, Lu Chen, Lin Lü, Qingxin Chen, Zhigang Li, Yongkai Liu, Jiajie Yu, Jialin Meng, Qingqing Sun, David Wei Zhang, Tianyu Wang, Lin Chen

2025Nano Letters5 citationsDOI

Abstract

Traditional computing systems are limited by separated architectures, inspiring the development of compatible artificial neuron and synaptic devices for hybrid neuromorphic computing. Here, we present a CMOS-compatible, single-stack Hf 0.2 Zr 0.8 O 2 (HZO) platform in which the as-deposited film serves as a memristive synapse, while a one-step postdeposition anneal yields an antiferroelectric (AFE) neuron device. Both device roles share identical CMOS-compatible premanufacturing steps; a single, nonreversible postanneal diverges the same TiN/HZO/TiN stack into the AFE-neuron path. A memristive device enables analogue conductance modulation for convolutional feature extraction. After annealing, antiferroelectric devices achieve spontaneous depolarization behavior, paving the way for spike-based encoding and biologically plausible neuronal dynamics. By integrating this process-compatible dual-mode device set within a unified material platform, a convolutional spiking neural network was constructed with 97.9% accuracy in dynamic gestures. This work highlights CMOS compatible neuromorphic electronics for hybrid neuromorphic computing within compact neuromorphic hardware.

Topics & Concepts

Neuromorphic engineeringMemristorCMOSComputer scienceMaterials scienceConvolutional neural networkElectronic engineeringSpiking neural networkStack (abstract data type)Modulation (music)Set (abstract data type)Synaptic weightNanoelectronicsArtificial neural networkEncoding (memory)SiliconUnconventional computingOptoelectronicsNanotechnologyConductanceElectronicsComputer architectureKey (lock)Electronic circuitTopology (electrical circuits)Feature (linguistics)Flexible electronicsAdvanced Memory and Neural ComputingFerroelectric and Negative Capacitance DevicesNeural Networks and Reservoir Computing
One-Step Annealing-Configured Hf <sub>0.2</sub> Zr <sub>0.8</sub> O <sub>2</sub> Memristive–Antiferroelectric Devices for Bioinspired CSNN Neuromorphic Computing | Litcius