Wang Tian-Yu, Meng Jia-Lin, He Zhen-Yu, Chen Lin, Zhu Hao, Sun Qing-Qing, Ding Shi-Jin, Zhou Peng, Zhang David Wei
State Key Laboratory of ASIC and System, School of Microelectronics, Fudan University, Shanghai 200433, China.
Nanoscale. 2020 Apr 30;12(16):9116-9123. doi: 10.1039/d0nr00919a.
As one of the emerging neuromorphic computing devices, memristors may break through the limitation of traditional computers with a von Neumann architecture. However, the development of flexible memristors is limited by the high-temperature fabrication process, large operating voltage and non-uniform distribution of resistance. The room-temperature process has attracted great attention due to its advantages of low thermal dissipation, low cost and excellent compatibility with flexible electronics. Here, we proposed a fully physical vapour deposition (PVD) process for fabricating a memristor without additional heat treatment. The device showed excellent resistive switching characteristics with ultralow set/reset voltages (0.48 V/-0.39 V), uniform distribution (10%/15%), stable retention characteristic, multilevel storage behavior and reliable flexibility (radius of 10 mm). With continuously modulated conductance, typical synaptic plasticities were simulated by our flexible biomemristor, including excitatory post-synaptic current (EPSC), paired-pulse facilitation (PPF), long-term potentiation/depression (LTP/LTD) and learning-forgetting curve. Furthermore, the array learning behavior like that of the human brain was simulated with these trainable biomemristors. This study paves a new way for developing low-cost, wearable, neuromorphic computing electronics at room temperature and expands the applications of artificial synapse arrays.
作为新兴的神经形态计算设备之一，忆阻器可能突破具有冯·诺依曼架构的传统计算机的限制。然而，柔性忆阻器的发展受到高温制造工艺、大工作电压和电阻分布不均匀的限制。室温工艺因其热耗散低、成本低以及与柔性电子器件的出色兼容性等优点而备受关注。在此，我们提出了一种无需额外热处理的用于制造忆阻器的全物理气相沉积（PVD）工艺。该器件表现出优异的电阻开关特性，具有超低的设置/重置电压（0.48 V/-0.39 V）、均匀分布（10%/15%）、稳定的保持特性、多级存储行为以及可靠的柔韧性（10毫米半径）。通过连续调制电导，我们的柔性生物忆阻器模拟了典型的突触可塑性，包括兴奋性突触后电流（EPSC）、双脉冲易化（PPF）、长时程增强/抑制（LTP/LTD）以及学习-遗忘曲线。此外，利用这些可训练的生物忆阻器模拟了类似人类大脑的阵列学习行为。这项研究为在室温下开发低成本、可穿戴的神经形态计算电子产品开辟了一条新途径，并扩展了人工突触阵列的应用。
