Institute of Microelectronics and Tsinghua National Laboratory for Information Science and Technology (TNList), Tsinghua University , Beijing 100084, China.
Department of Electrical Engineering, Princeton University , Princeton, New Jersey 08544, United States.
ACS Nano. 2017 Dec 26;11(12):12247-12256. doi: 10.1021/acsnano.7b05726. Epub 2017 Dec 11.
Extremely low energy consumption neuromorphic computing is required to achieve massively parallel information processing on par with the human brain. To achieve this goal, resistive memories based on materials with ionic transport and extremely low operating current are required. Extremely low operating current allows for low power operation by minimizing the program, erase, and read currents. However, materials currently used in resistive memories, such as defective HfO, AlO, TaO, etc., cannot suppress electronic transport (i.e., leakage current) while allowing good ionic transport. Here, we show that 2D Ruddlesden-Popper phase hybrid lead bromide perovskite single crystals are promising materials for low operating current nanodevice applications because of their mixed electronic and ionic transport and ease of fabrication. Ionic transport in the exfoliated 2D perovskite layer is evident via the migration of bromide ions. Filaments with a diameter of approximately 20 nm are visualized, and resistive memories with extremely low program current down to 10 pA are achieved, a value at least 1 order of magnitude lower than conventional materials. The ionic migration and diffusion as an artificial synapse is realized in the 2D layered perovskites at the pA level, which can enable extremely low energy neuromorphic computing.
超低能耗的神经形态计算对于实现与大脑相媲美的大规模并行信息处理至关重要。为了实现这一目标,需要基于具有离子输运和极低工作电流的材料的电阻式存储器。极低的工作电流可通过将编程、擦除和读取电流最小化来实现低功耗操作。然而,目前在电阻式存储器中使用的材料,如缺陷 HfO、AlO、TaO 等,在允许良好离子输运的同时,无法抑制电子输运(即漏电流)。在这里,我们表明 2D Ruddlesden-Popper 相混合卤化铅钙钛矿单晶是低工作电流纳米器件应用的有前途的材料,因为它们具有混合电子和离子输运以及易于制造的特点。通过溴离子的迁移,在剥落的二维钙钛矿层中可以明显看出离子输运。可视化了直径约为 20nm 的细丝,并实现了工作电流低至 10pA 的超低编程电流的电阻式存储器,这一数值比传统材料至少低 1 个数量级。在二维层状钙钛矿中,离子迁移和扩散可以实现人工突触,这可以实现极低能耗的神经形态计算。