COMMUNICATION 1800195 (1 of 11) © 2018 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.advmat.de Ion Gated Synaptic Transistors Based on 2D van der Waals Crystals with Tunable Diffusive Dynamics Jiadi Zhu, Yuchao Yang,* Rundong Jia, Zhongxin Liang, Wen Zhu, Zia Ur Rehman, Lin Bao, Xiaoxian Zhang, Yimao Cai, Li Song, and Ru Huang J. Zhu, Prof. Y. Yang, R. Jia, Z. Liang, L. Bao, Prof. Y. Cai, Prof. R. Huang Key Laboratory of Microelectronic Devices and Circuits (MOE) Institute of Microelectronics Peking University Beijing 100871, China E-mail: yuchaoyang@pku.edu.cn W. Zhu, Z. U. Rehman, Prof. L. Song National Synchrotron Radiation Laboratory CAS Center for Excellence in Nanoscience University of Science and Technology of China Hefei, Anhui 230029, China Prof. X. Zhang CAS Key Laboratory of Standardization and Measurement for Nanotechnology CAS Center for Excellence in Nanoscience National Center for Nanoscience and Technology Beijing 100190, China The ORCID identification number(s) for the author(s) of this article can be found under https://doi.org/10.1002/adma.201800195. DOI: 10.1002/adma.201800195 that fundamentally overcomes the von Neumann bottleneck in conventional dig- ital computers. [2–4] As a result, there exists a tremendous upsurge of research interests on building neuromorphic systems, espe- cially by exploiting the scalability and func- tionality of emerging devices such as phase change memories, [5–7] spintronic devices, [8] metal–insulator transition devices, [9,10] atomic switches, [11] and memristors. [12,13] In particular, conceived neuromorphic archi- tectures using these devices as synaptic elements, by implementing similar tunable weights, analog behavior or ionic dynamics with biological synapses, [14–19] have been widely pursued and investigated. [20–22] Although encouraging processes have been made so far, previous studies usually utilize these emerging devices as nonvola- tile elements that can store analog weights and accelerate vector-matrix multiplica- tions, without exploiting the rich function- ality and dynamics of biological synapses. This is a large simplification on synaptic functions, given the fact that biological syn- apses possess much more extensive and complicated functionali- ties that are described as “synaptic plasticity” and are deemed cru- cial for neural signal transmission, filtering as well as memory, learning, and other cognitive functions in the brain. [23,24] It is therefore essential to develop and fabricate biorealistic synaptic elements, whose strong functionalities and rich temporal/spatial dynamics [13,25,26] shall be indispensable for the construction of truly intelligent yet low power neuromorphic hardware. It has also been noticed in existing studies that previous syn- aptic devices usually show much higher energy consumption compared with their biological counterparts with extremely low energy consumption of ≈10 fJ per spike. [3] In order to scale up the network toward the capacity of human brain while main- taining the constraint of acceptable energy consumptions on the chip, it is necessary to aggressively reduce the energy consump- tion of synaptic devices at least down to pJ per spike level. [3] Recently, there is a growing interest on 3-terminal synaptic architectures, whose additional input terminal and modi- fied device configuration have proven favorable for achieving complicated synaptic functions (such as heterosynaptic plas- ticity [26,27] ) and low power consumption. [27–32] However, a biore- alistic artificial synapse with excellent linearity, symmetry, and sufficiently low energy requirement is still missing. Neuromorphic computing represents an innovative technology that can perform intelligent and energy-efficient computation, whereas construction of neuromorphic systems requires biorealistic synaptic elements with rich dynamics that can be tuned based on a robust mechanism. Here, an ionic- gating-modulated synaptic transistor based on layered crystals of transitional metal dichalcogenides and phosphorus trichalcogenides is demonstrated, which produce a diversity of short-term and long-term plasticity including excitatory postsynaptic current, paired pulse facilitation, spiking-rate- dependent plasticity, dynamic filtering, etc., with remarkable linearity and ultralow energy consumption of ≈30 fJ per spike. Detailed transmission elec- tron microscopy characterization and ab initio calculation reveal two-stage ionic gating effects, namely, surface adsorption and internal intercalation in the channel medium, causing different poststimulation diffusive dynamics and thus accounting for the observed short-term and long-term plasticity, respectively. The synaptic activity can therefore be effectively manipulated by tailoring the ionic gating and consequent diffusion dynamics with varied thickness and structure of the van der Waals material as well as the number, duration, rate, and polarity of gate stimulations, making the present synaptic transistors intriguing candidates for low-power neuromorphic systems. Neuromorphic Computing Envisioned by Carver Mead in 1990, [1] neuromorphic computing seeks inspirations from the massive parallelism, robust compu- tation, and high energy efficiency of the human brain and can potentially give rise to a revolutionary computing technology Adv. Mater. 2018, 1800195