能信耦合的类神经形态器件技术展望

随着智能化时代的到来，云计算、人工智能、物联网等领域的兴起，全球电子终端设备数量剧增，电子终端设备的性能和功耗矛盾日益凸显。类神经形态器件相较于传统CMOS器件在算力、能效方面优势显著。为了更好地了解类神经形态器件的发展现状，明确其未来发展方向，首先综合评述了典型的类神经形态器件研究进展，而后分析了现有类神经形态器件技术的瓶颈问题，进而提出了通过能量与信息耦合，推动类神经形态器件进一步提高性能、降低功耗的策略，并讨论了能信耦合类神经形态器件的物理可行性和应用潜力，最后展望了智能化时代对于电子信息器件的发展需求，以及未来能信耦合器件的发展思路和研究重点。

With the advent of intelligent age and the rise of cloud computing, artificial intelligence, Internet-of-Things, and other information technologies, the number of electronic terminal equipment around the world has increased dramatically, and it has also caused a sharp increase in power consumption. Compared with traditional CMOS devices, neuromorphic devices have significant advantages in computing power and energy efficiency. To better understand the development status of neuromorphic devices and clarify their future development direction, this article firstly summarizes the research progress of various types of neuromorphic devices, analyzes the existing technology bottleneck of these devices, and then put forward a strategy to further improve the performance and energy efficiency of neuromorphic devices via an energy-information coupling route. The fundamentals and feasibility of the presented energy-information coupling devices are also discussed. Finally, this article will envision the development of information devices in the intelligent age, and the development ideas and research focus are discussed on the energy-information coupling devices.