| After entering the Internet era,driven by cloud computing,artificial intelligence and the Internet of Things,global data generation has exploded and the demand for computing power has increased significantly.The human brain’s neuromorphic system can process information in parallel and consume less energy,thus promoting research on neuromorphic applications is a powerful means to solve the current arithmetic challenges.Memristors,which are similar in structure to neurons in the human brain,have a strong potential for building artificial neuromorphic hardware systems.In this paper,we investigate the enhancement of the resistive properties of zirconia-based memristors and simulate synaptic properties to achieve artificial nociceptors and tactile perception simulations,thus providing a scientific basis for the application of memristors in neuromorphic research.The Pt/ZrO2/TiN single layer and Pt/ZrO2/IGZO/TiN bilayer memristors were firstly prepared.Compared with the Pt/ZrO2/TiN single-layer memristor,the Forming voltage of the Pt/ZrO2/IGZO/TiN bilayer was reduced by 2.1 V,the window value was increased by 11.3 times and the VReset dispersion coefficient was increased from 3.6%to 2.7%after the insertion of a3 nm IGZO functional layer.These results show that the bilayer memristor has better cycling stability and consistent transition voltage distribution.Furthermore,the bilayer structure memristor has good multi-level storage and data retention characteristics.The thesis further employs pulse signals to modulate the conductance properties of bilayer structured memristors and to investigate their simulation in terms of biological synaptic plasticity.Two consecutive pulse pairs are applied to the device at the same electrode to achieve the paired-pulse facilitation（PPF）property,pulse pairs with time differences are applied to the device at different electrodes to achieve the spike-timing-dependent plasticity（STDP）property,pulse sequences with different frequencies are applied to the device to achieve the spike-rate-dependent plasticity（SRDP）property,pulse sequences with different polarities are applied to the device to long-term depression（LTD）and long-term potentiation（LTP）characteristics are achieved.Based on the excellent synaptic plasticity of Pt/ZrO2/IGZO/TiN bilayer structured memristors,the thesis investigates its application to biological nociceptors by modulating the amplitude,width and frequency of the pulses to achieve brain-like behaviors of biological nociceptors such as threshold properties,no adaptation properties,hyperalgesia properties and allodynia properties.Finally,an integrated circuit for sensory storage and computation was constructed.The behavior of the amnestic current response in response to the force applied to the pressure sensor was realized in the circuit,which is consistent with the behavioral changes of the human tactile system in response to the different pressures perceived by the skin,thus simulating human tactile perception.In summary,this paper compares the resistive properties of single-layer and bilayer memristors and finds that all resistive properties of the bilayer memristors are superior.The synaptic bionic function and the artificial nociceptor function were further investigated and implemented to build a sensory storage and calculation integrated electrical system and successfully simulate the human tactile perception behavior.This has significant implications for the application of memristors in the field of neuromorphic,and provides practical ideas for the subsequent successful development of multi-modal perceptual simulation and integration of sensory storage and computation. |
