From Silicon to Synapse: Scientists Morph Standard Transistor Into Artificial Neuron

16.04.2025

Researchers at the National University of Singapore (NUS) have demonstrated that a single standard silicon transistor — a core component in modern electronics — can mimic both the behavior of biological neurons and synapses. This discovery represents a major step forward in neuromorphic computing, a field focused on replicating the brain’s efficiency and architecture in electronic systems.

Led by Associate Professor Mario Lanza from NUS’s College of Design and Engineering, the study reveals how traditional transistors, when operated in a nonconventional manner, can emulate neural firing and synaptic plasticity. This was achieved by manipulating the bulk terminal resistance of the transistor to trigger specific physical mechanisms: punch-through impact ionization and charge trapping. These effects enabled the device to behave like both a neuron and a synapse — the two essential components of biological neural networks.

The team also introduced a two-transistor cell design, termed Neuro-Synaptic Random Access Memory (NS-RAM), capable of operating in either neuronal or synaptic mode. Unlike previous approaches that rely on complex multi-transistor circuits or emerging materials not yet ready for mass production, this innovation is fully compatible with existing CMOS (complementary metal-oxide-semiconductor) technology, ensuring scalability and integration into current chip manufacturing processes.

In extensive testing, the NS-RAM cell demonstrated low power consumption, strong operational stability, and consistent behavior across devices — qualities critical for real-world AI applications. This hardware-based approach could enable more compact and energy-efficient artificial neural networks, overcoming the high power demands and inefficiencies of software-based AI systems.

By leveraging commercially available silicon technology, this advancement opens the door to brain-inspired computing systems that are both practical and energy-efficient, potentially revolutionizing fields ranging from edge AI to robotics and beyond.