Manufacture of synaptic transistor-based neuromorphic systems: from emerging device fabrication to advanced circuit integration

Abstract Neuromorphic computing systems, inspired by biological neural networks in the brain, offer transformative potential for energy-efficient artificial intelligence by breaking through the limitations of conventional von Neumann architectures. Synaptic transistors, serving as core hardware elements that dynamically regulate their conductance states to emulate neurobiological functions, represent a pivotal approach in this direction. This comprehensive review systematically surveys manufacturing pathways for synaptic transistor-based neuromorphic systems, spanning from emerging device fabrication to advanced circuit integration. We first elucidate the motivation for neuromorphic computing and the fundamental role of synaptic transistors as artificial synapses. Core discussions focus on four representative device architectures: electrolyte-gated, ferroelectric, charge-trapping, and optically controlled synaptic transistors, followed by in-depth analysis of advanced manufacturing techniques, including functional layer processing, scalable patterning strategies, and integration schemes. Subsequently, we outline the applications of synaptic transistor-based neuromorphic systems in the field of artificial intelligence and finally discuss the critical challenges, alongside the future prospects in terms of materials, devices, integration, architectures, and codesigned algorithms. This work provides a timely and extensive reference for advancing hardware implementation realization of neuromorphic computing.