Diffusive memristor for artificial neurons

Researchers from the University of Southern California, University of Massachusetts, University of California Los Angeles, Syracuse University, and the Air Force Research Laboratory developed artificial neurons that replicate the complex electrochemical behavior of biological brain cells.

“Our existing computing systems were never intended to process massive amounts of data or to learn from just a few examples on their own. One way to boost both energy and learning efficiency is to build artificial systems that operate according to principles observed in the brain,” said Joshua Yang, a computer and electrical engineering professor at USC, in a press release. “Ions are a better medium than electrons for embodying principles of the brain. Because electrons are lightweight and volatile, computing with them enables software-based learning rather than hardware-based learning, which is fundamentally different from how the brain operates. The brain learns by moving ions across membranes, achieving energy-efficient and adaptive learning directly in hardware, or more precisely, in what people may call ‘wetware’.”

Based on a ‘diffusive memristor,’ the artificial neuron uses silver ions in oxide to generate an electrical pulse and emulate the physical processes of the brain, in which electrical signals are converted into chemical signals (like potassium, sodium, or calcium ions) and back to electrical as they pass from one neuron to the next through a synapse.

“Even though it’s not exactly the same ions in our artificial synapses and neurons, the physics governing the ion motion and the dynamics are very similar,” said Yang. “Silver is easy to diffuse and gives us the dynamics we need to emulate the biosystem so that we can achieve the function of the neurons, with a very simple structure.”

While the silver is not compatible with conventional semiconductor manufacturing, the researchers plan to investigate alternative ionic species that could provide similar functionalities. [1]

Magnetic tunnel junction synapses

Researchers from the University of Texas at Dallas, Everspin Technologies, and Texas Instruments built a small-scale prototype of a neuromorphic computer that learns patterns and makes predictions using fewer training computations than conventional AI systems.

“Our work shows a potential new path for building brain-inspired computers that can learn on their own. Since neuromorphic computers do not need massive amounts of training computations, they could power smart devices without huge energy costs,” said Joseph S. Friedman, associate professor of electrical and computer engineering at The University of Texas at Dallas, in a statement. “The principle that we use for a computer to learn on its own is that if one artificial neuron causes another artificial neuron to fire, the synapse connecting them becomes more conductive.”

The device relies on spin-transfer torque (STT) magnetic tunnel junctions (MTJs) to mimic the way the brain processes and learns patterns. As signals pass through MTJs in a coordinated manner, their connections adjust to strengthen certain pathways, similar to how synaptic connections in the brain are reinforced during learning. Additionally, binary switching makes the MTJs reliable for storing information. [2]

Metal-organic framework nanofluidics

Researchers at Monash University, University of Science and Technology of China, and National University of Singapore developed a nanofluidic chip that mimics the neural pathways of the brain. The coin-size chip uses a metal-organic framework (MOF) that channels ions through nanoscale pathways, enabling it to behave like electronic transistors while also remembering previous signals.

“For the first time, we’ve observed saturation nonlinear conduction of protons in a nanofluidic device,” said Huanting Wang, a professor and deputy director of the Monash Centre for Membrane Innovation, in a statement. “This opens up new opportunities for designing iontronic systems with memory and even learning capabilities.”

“Our chip can selectively control the flow of protons and metal ions, and it remembers previous voltage changes, giving it a form of short-term memory,” added Jun Lu of the Monash Department of Chemical and Biological Engineering, in a statement. “What makes our device truly special is its hierarchical structure, which allows it to control protons and metal ions in entirely different ways. This kind of selective, nonlinear ion transport hasn’t been seen before in nanofluidics.” [3]

References

[1] R. Zhao, T. Wang, T. Moon, et al. A spiking artificial neuron based on one diffusive memristor, one transistor and one resistor. Nat Electron (2025). https://doi.org/10.1038/s41928-025-01488-x

[2] P. Zhou, A.J. Edwards, F.B. Mancoff, et al. Neuromorphic Hebbian learning with magnetic tunnel junction synapses. Commun Eng 4, 142 (2025). https://doi.org/10.1038/s44172-025-00479-2

[3] X. Hu, H. Xu, J. Lu, et al. Selective ion transport of nonlinear resistive switching by hierarchical nanometer-to-angstrom channels for nanofluidic transistors. Sci. Adv. 11, eadw7882 (2025). https://doi.org/10.1126/sciadv.adw7882