神经形态工程学
材料科学
记忆电阻器
电导
电阻式触摸屏
异质结
蚀刻(微加工)
调制(音乐)
信号(编程语言)
光电子学
纳米技术
可扩展性
接口(物质)
纳米尺度
复合数
人工神经网络
等离子体子
制作
分子开关
路径(计算)
实现(概率)
电阻随机存取存储器
微型加热器
化学气相沉积
突触可塑性
电致变色
光子学
计算机科学
突触
作者
Liangchao Guo,Peng Wang,Chunyu Du,Sheng Hu,Feng Zhao,Junfeng Cui,Qilong Yuan,Chao Zhang
摘要
Memristive devices that integrate gas sensing with neuromorphic functionality have been identified as a promising route to compact, low-power, and multifunctional artificial olfaction. Herein, we presented a chemically gated memristive device based on a V2CTx MXene/PANI:PSS nanocomposite. A dense, strongly bonded heterostructure was constructed via HF etching followed by in situ oxidative polymerisation, enabling molecular interactions to directly modulate device conductance and thus realise gas-electrical co-operative synaptic operation. As a gas sensor, the device shows a 31.8% response to 5 ppm NH3, with response/recovery times of 26 s/52 s and >95% signal retention over 15 days at room temperature. As a memristor, it exhibits stable bipolar resistive switching (ON/OFF > 102) with a switching time of ∼10 ms; under preset gaseous atmospheres and pre-synaptic electrical pulses, EPSC/IPSC and PPD-type temporal plasticity are elicited. Mechanistically, protonation/deprotonation of PANI sets the chemical working point, while interfacial trap charging and barrier modulation at the V2CTx/PANI interface enable reversible conductance updates, mapping molecular events into time-amplitude fingerprints. At the array level, a 2 × 2 crossbar with a multi-task neural network enables parallel identification of gas type and concentration, achieving 96.2% and 94.1% accuracy, respectively. This strategy enables an end-to-end chain-from molecular transduction through synaptic modulation to algorithmic inference-within a single device, providing a scalable path to compact, low-power artificial olfaction hardware.
科研通智能强力驱动
Strongly Powered by AbleSci AI