Northwestern Engineers Print Artificial Neurons That Talk to Living Brain Cells
Researchers at Northwestern University have used aerosol jet printing to create flexible artificial neurons that generate biologically realistic signals and directly activate living neural tissue.
Overview
Engineers at Northwestern University have printed flexible artificial neurons capable of generating electrical signals that directly activate living brain cells — a capability that had eluded researchers for decades. The devices, described in a study published April 15 in Nature Nanotechnology, are made from printable electronic inks and produce spiking patterns that fall within the temporal window required for biological interaction.
The advance marks the first time printed artificial neurons have been shown to operate in a regime compatible with living neural tissue, opening paths toward lower-cost neuroprosthetics and brain-inspired computing hardware.
How It Works
The core of the device is an ink formulated from nanoscale flakes of molybdenum disulfide (MoS₂), a two-dimensional semiconductor, combined with graphene, which serves as an electrical conductor. According to Northwestern Now, the team used aerosol jet printing to deposit these inks onto flexible polymer substrates — a process that keeps fabrication costs low and avoids the rigid silicon wafers required by conventional microelectronics.
The key innovation lies in how the polymer stabilizer in the ink is handled. Rather than removing it entirely — the standard approach — the researchers partially decomposed it. Passing electrical current through the partially decomposed film drives further breakdown, creating spatially localized conductive filaments. These filaments behave as memristors: devices whose resistance depends on their history of applied voltage. The result is a neuron that produces not only simple spikes but also bursting and continuous-firing patterns, matching the multi-order complexity of real neural signaling.
As noted by Interesting Engineering, lead researcher Mark C. Hersam, the Walter P. Murphy Professor of Materials Science and Engineering at Northwestern’s McCormick School, described the challenge his team solved: “We are within a temporal range that was not previously demonstrated for artificial neurons.” Earlier attempts using organic materials spiked too slowly to interact with biology, while metal oxide devices fired too fast; the MoS₂-graphene system hits the right window.
Tested on Living Tissue
To validate biological compatibility, the team applied signals from the printed neurons to slices of mouse cerebellum tissue. The artificial pulses triggered responses in the living cells, demonstrating that the devices can interface directly with biological neural circuits, not just replicate waveform shapes in isolation.
The work was led by Hersam alongside research associate professor Vinod K. Sangwan and Indira M. Raman of Northwestern’s Weinberg College of Arts and Sciences, who contributed expertise in neurophysiology. The paper’s first authors span materials science and electrical engineering, reflecting the interdisciplinary scope of the project.
What It Could Enable
Neuroprosthetics
The most immediate application target is implantable devices that restore lost sensory or motor function. Current cochlear implants and retinal prosthetics rely on rigid metal electrodes that can damage tissue over time. Printed flexible neurons that speak the language of the brain could enable conformable implants for hearing, vision, and limb movement that integrate more naturally with neural anatomy.
Energy-Efficient Computing
Beyond medicine, the work has implications for neuromorphic computing — hardware that processes information the way the brain does rather than executing sequential instructions. Hersam has noted that the human brain is roughly five orders of magnitude more energy efficient than a digital computer. Systems built from spiking artificial neurons that mimic biological dynamics could perform complex inference tasks at a fraction of current data-center power consumption.
The aerosol jet printing process also means these devices can, in principle, be manufactured at low cost and scaled to three-dimensional architectures — something silicon-based neurons cannot easily achieve.
What Remains to Be Done
The current results are from ex vivo mouse tissue. Whether the devices retain their performance inside a living organism, over extended periods, and without causing immune responses remains to be demonstrated. The study also does not address the challenge of precisely routing thousands of printed neurons into coordinated circuits, which would be necessary for practical neuroprosthetic or computing applications.
The paper, titled “Printed MoS₂ memristive nanosheet networks for spiking neurons with multi-order complexity,” is published in Nature Nanotechnology (DOI: 10.1038/s41565-026-02149-6).