Expanding the dimensions of neuromorphic computing in the age of artificial intelligence

Originally based on a 1945 document, von Neumann architecture has long been a conventional computer design with memory space that shares both instructions and data. But now, in the age of artificial intelligence, it is increasingly constrained by limitations that cannot adequately meet current computing demands. In its place, neuromorphic computer designs, which are inspired by the human brain, are emerging and feature artificial synaptic devices that can significantly expand learning and memory functions.

Wu et al. designed a mixed-dimensional, heterojunction synaptic transistor, utilizing a core structure of quantum dots comprised of zero-dimensional molybdenum disulfide (MoS₂) and nanowires made from one-dimensional indium zinc oxide (InZnO) nanowires.

“A key advantage of this architecture is its ability to significantly enhance the separation and trapping efficiency of interface carriers,” said author Fengyun Wang. “In essence, the mixed-dimensional interface allows for more effective control over electron behavior, enabling more linear and controllable synaptic weight updates.”

The researchers constructed a mixed-dimensional heterostructure by integrating hydrothermal synthesis and electrospinning techniques, which can enable making certain materials in artificial synaptic devices, and further expanded the single-device configuration into a 3×3 array. This array successfully emulated the key processes of image memory, forgetting, and relearning observed in biological synapses. By leveraging the structure’s mixed-dimensional interface, the team successfully demonstrated the device’s superior linearity and learning performance.

“We believe the structure shows considerable promise for guiding the design and construction of efficient, low-power photonic neuromorphic vision systems,” said Wang. “In short, we have demonstrated a device prototype capable of ‘sensing, memorizing, and learning,’ offering a new direction for brain-inspired visual computing.”

Source: “0D/1D mixed-dimensional heterojunction synaptic transistor for visual information processing,” by Liren Wu, Shuwen Xin, Renjie Li, Xudong Zhang, Mengyao Wei, Feiyang Xu, Jiaqi Xu, Peilong Xu, Lei Liu, Yuanbin Qin, and Fengyun Wang, Applied Physics Letters (2025). The article can be accessed at: https://doi.org/10.1063/5.0299679 .