A systematic roadmap for artificial spiking neurons: From materials to neuromorphic perception and computation

The rapid growth of the Internet of Things and artificial intelligence is driving massive data generation at the edge, from sensors and wearables to distributed terminals. This trend is amplifying a long-standing limitation of von Neumann computing: memory and processing are physically separated, so a large share of energy is spent moving data rather than performing computation. Inspired by the brain’s efficient information processing, neuromorphic computing is widely viewed as a key path beyond Moore’s law. Spiking neural networks are one of its core paradigms, but their practical implementation depends on compact hardware spiking neurons that can reproduce neuron like nonlinear dynamics and enable event driven, low power sensing and computing.

A team led by Professor Chuanyu Han at Xi’an Jiaotong University in China has recently mapped the rapidly evolving landscape of artificial spiking neurons built on emerging electronic devices. The team notes that conventional CMOS neuron circuits can generate spikes, yet they often require complex circuit blocks, large footprint, and suffer from leakage power, which hinders dense integration. By contrast, many emerging devices exhibit intrinsic nonlinear switching and dynamic responses that can embed spiking behavior more deeply into device physics, offering a path to lower circuit overhead, improved energy efficiency, and tighter coupling of sensing and computation.

The team published their review in Nano Research on January 16, 2026.

The team’s open access review has been published in Nano Research. The article is titled “Recent progress on artificial spiking neurons based emerging electronic devices for neuromorphic perception and computation,” with Zheng Fang and Kaiyao Wang as co first authors. Using an integrated materials to devices to circuits to systems perspective, the review aims to provide a clearer and more actionable roadmap for interdisciplinary researchers working across materials science, microelectronics, and neuromorphic engineering. 

The authors emphasize that artificial spiking neurons are not a single category, but can be realized across diverse device platforms. The review offers a comparative discussion of mainstream routes in both two terminal and three terminal implementations. On the two-terminal side, it highlights representative spiking approaches enabled by mechanisms such as Mott transitions, ion diffusion processes, ovonic threshold switching, and phase change dynamics. On the three-terminal side, it examines single transistor neuron concepts as well as more programmable platforms including ferroelectric and antiferroelectric field effect transistors and two-dimensional material transistors. By linking spiking mechanisms to circuit construction templates, the review clarifies how different routes trade off threshold characteristics, energy per spike, operating speed, endurance, and integration feasibility, factors that ultimately determine whether a technology can support scalable neuromorphic hardware systems.

A major challenge in the field is that studies are difficult to compare fairly because metrics and test conditions vary widely. To address this, Han’s team proposes a unified benchmarking framework built around nine key performance indicators designed to reflect both device behavior and system readiness. The framework integrates electrical metrics, reliability and endurance, and integration constraints such as CMOS back-end compatibility, while explicitly accounting for test condition dependence. The review compiles representative benchmark data in Table 1 for ion diffusion, ovonic threshold switching, phase change memory, Mott memristor, and transistor based spiking neurons, providing a practical reference for cross technology evaluation.

At the system level, the review connects device progress to application relevance by surveying representative demonstrations in neuromorphic perception and brain inspired computing. It summarizes prototype explorations spanning tactile and thermal sensing, vision, smell, taste, hearing, and multimodal fusion, and discusses how artificial spiking neurons can support a range of computing architectures, including spiking neural networks and recurrent spiking variants, oscillatory neural networks, pulse coupled neural networks, multi network concepts, and in sensor computing that co locates sensing and processing to reduce data conversion and transfer overhead.

Looking ahead, the team points to new materials and engineering priorities for deployable neuromorphic hardware. Promising directions include organic transistors for soft, low temperature, large area integration and two dimensional materials for heterogeneous stacking and richer tunability. The authors emphasize balancing energy, stability, endurance, and uniformity, while noting that large scale arrays and edge deployment will ultimately be limited by interconnects, variability, peripheral overhead, and long-term reliability under real world conditions.

“We hope to connect materials and device physics, neuron circuit construction, and system level application needs, so researchers can see not only which devices can spike, but also how they can be used in systems, where bottlenecks emerge, and which directions deserve priority in optimization,” said Professor Chuanyu Han, corresponding author of the review. “The unified benchmarking framework is meant to put different technology routes on common ground, helping the community identify the options with real scalability and deployable potential.”

The research team expects the review to accelerate the development of device enabled artificial spiking neurons for configurable, low power neuromorphic perception and computation, helping move the field from isolated demonstrations toward scalable, deployable systems. “By connecting materials and device physics with neuron circuit construction and system level needs, we hope researchers can better identify practical bottlenecks and prioritize the most scalable directions for future optimization,” said Associate Professor Chuanyu Han.

Other contributors include Shujing Zhao, Shiquan Fan, Weihua Liu, Xin Li, and Li Geng from the School of Microelectronics at Xi’an Jiaotong University in Xi’an, China, and the Key Lab of Micro-nano Electronics and System Integration of Xi’an City.

DOI Link:

https://doi.org/10.26599/NR.2026.94908456

About the Author

Chuanyu Han is a tenured associate professor at the School of Microelectronics, Xi’an Jiaotong University. He received his Ph.D. degree from the University of Hong Kong in August 2015, joined Xi’an Jiaotong University in February 2016, and has successively worked as a lecturer, tenure-track associate professor, and currently tenured associate professor; he was a visiting scholar in the Department of Mechanical Engineering, Columbia University, USA from 2017 to 2018. His research mainly focuses on memristors, thin-film transistors, biosensors, and neuromorphic devices and systems. He has published more than 30 SCI papers as first author or corresponding author in journals including IEEE Electron Device Letters, ACS Sensors, and Bioelectronics and Biosensors, and is the principal investigator of one Young Scientist Fund project, two General Program projects of the National Natural Science Foundation of China (NSFC), as well as multiple provincial and ministerial-level research projects. 

About Nano Research

Nano Research is a peer-reviewed, open access, international and interdisciplinary research journal, sponsored by Tsinghua University and the Chinese Chemical Society, published by Tsinghua University Press on the platform SciOpen. It publishes original high-quality research and significant review articles on all aspects of nanoscience and nanotechnology, ranging from basic aspects of the science of nanoscale materials to practical applications of such materials. After 18 years of development, it has become one of the most influential academic journals in the nano field. Nano Research has published more than 1,000 papers every year from 2022, with its cumulative count surpassing 8,000 articles. In 2025 InCites Journal Citation Reports, its 2025 IF is 9.4 (8.3, 5 years), and it continues to be the Q1 area among the four subject classifications. Nano Research Award, established by Nano Research together with TUP and Springer Nature in 2013, and Nano Research Young Innovators (NR45) Awards, established by Nano Research in 2018, have become international academic awards with global influence.