Chen Lijun, Saleh Saad, Tavormina Filippo, Di Mario Lorenzo, Li Jiaxiong, Xie Zhiqiang, Masciocchi Norberto, Brabec Christoph J, Koldehofe Boris, Loi Maria Antonietta
Zernike Institute for Advanced Materials, University of Groningen, Nijenborgh 4, Groningen, 9747 AG, The Netherlands.
CogniGron (Groningen Cognitive Systems and Materials Center), University of Groningen, Nijenborgh 4, Groningen, 9747 AG, The Netherlands.
Adv Mater. 2025 May;37(20):e2414430. doi: 10.1002/adma.202414430. Epub 2025 Mar 31.
Metal halide perovskites have drawn great attention for neuromorphic electronic devices in recent years, however, the toxicity of lead as well as the variability and energy consumption of operational devices still pose great challenges for further consideration of this material in neuromorphic computing applications. Here, a 2D Ruddlesden-Popper (RP) metal halides system of formulation BAPbSnI (BA = n-butylammonium) is prepared that exhibits outstanding resistive switching memory performance after cesium carbonate (CsCO) deposition. In particular, the device exhibits excellent switching characteristics (endurance of 5 × 10 cycles, ON/OFF ratio ≈10) and achieves 90.1% accuracy on the MNIST dataset. More importantly, a novel energy-efficient content addressable memory (CAM) architecture building on perovskite memristive devices for neuromorphic applications, called nCAM, is proposed, which has a minimum energy consumption of ≈0.025 fJ bit/cell. A mechanism involving the manipulation of trapping states through CsCO deposition is proposed to explain the resistive switching behavior of the memristive device.
近年来,金属卤化物钙钛矿在神经形态电子器件方面引起了极大关注,然而,铅的毒性以及运行器件的可变性和能耗,仍然对该材料在神经形态计算应用中的进一步考量构成巨大挑战。在此,制备了一种化学式为BAPbSnI(BA = 正丁基铵)的二维Ruddlesden-Popper(RP)金属卤化物体系,该体系在沉积碳酸铯(CsCO)后展现出出色的电阻开关记忆性能。特别地,该器件展现出优异的开关特性(5×10次循环的耐久性,开/关比≈10),并在MNIST数据集上实现了90.1%的准确率。更重要的是,提出了一种基于钙钛矿忆阻器件用于神经形态应用的新型节能内容可寻址存储器(CAM)架构,称为nCAM,其最小能耗约为0.025 fJ/比特/单元。提出了一种通过CsCO沉积操纵俘获态的机制来解释忆阻器件的电阻开关行为。