Neuromorphic computing requires electronic systems that can perform massively parallel computational tasks with low energy consumption. Such systems have traditionally been based on complementary metal–oxide–semiconductor circuits, but further advances in computational performance will probably require devices that can offer high-order complexity combined with area and energy efficiency. Halide perovskites can handle both ions and electronic charges, and could be used to create adaptive computing systems based on intrinsic device dynamics. The materials also offer exotic switching phenomena, providing opportunities for multimodal systems. Here we explore the development of neuromorphic hardware systems based on halide perovskites. We examine how devices based on these materials can serve as synapses and neurons, and can be used in neuromorphic computing networks. We also consider the challenges involved in developing practical perovskite neuromorphic systems, and highlight how these systems could augment and complement digital circuits.

神经形态计算需要能够以低能耗执行大规模并行计算任务的电子系统。此类系统传统上基于互补金属氧化物半导体电路，但计算性能的进一步进步可能需要能够提供高阶复杂性以及面积和能源效率的设备。卤化物钙钛矿可以处理离子和电子电荷，并可用于创建基于固有器件动力学的自适应计算系统。这些材料还提供奇异的开关现象，为多模式系统提供了机会。在这里，我们探索基于卤化物钙钛矿的神经形态硬件系统的开发。我们研究基于这些材料的设备如何充当突触和神经元，并可用于神经形态计算网络。我们还考虑了开发实用的钙钛矿神经形态系统所面临的挑战，并强调了这些系统如何增强和补充数字电路。