Engineers from Northwestern University have created printed artificial neurons that generate electrical signals close enough to those in the brain to activate living nerve cells. The achievement can change the future of neuroprosthetics and pave the way for a new generation of energy-efficient computing systems inspired by the brain.
The study, published on Wednesday in the journal "Nature Nanotechnology", shows that flexible devices made of nanoscale molybdenum disulfide and graphene-based inks can reproduce complex "spike" patterns - electrical impulses that match the signals of biological neurons in shape and duration. In tests on mouse cerebellar slices, artificial neurons consistently triggered a response in Purkinje cells, activating neural circuits in a way very close to natural stimulation.
A new material approach to artificial neurons
The devices are made using aerosol jet printing, which applies conductive and semiconductive inks to flexible polymer substrates. Instead of removing the stabilizing polymer from the ink - a step long considered a manufacturing obstacle - the team partially decomposes it, forming conductive filaments capable of generating sharp electrical impulses similar to neuronal ones.
The result is an artificial neuron that is not limited to simple single impulses, but creates a rich set of signal patterns, including solitary spikes, continuous tonic activity, and burst discharges. This diversity means that each device can encode more information, which makes it possible to reduce the number of components needed to build computing systems that resemble the structure and dynamics of the brain.
"The cream achieves complexity through billions of identical devices," explains Prof. Mark C. Hersam, the Walter P. Murphy Professor of Materials Science and Engineering at Northwestern's McCormick Engineering School, who led the project with Adjunct Professor Vinod K. Sangwan. "The brain is the complete opposite. It is heterogeneous, dynamic and three-dimensional. To get close to it, we need new materials and new ways of building electronics."
On the border between biology and computation
Biological validation was conducted in collaboration with Indira M. Raman - a neurobiologist from the Weinberg College of Arts and Sciences at Northwestern University. Hersam notes that previous attempts at artificial neurons with organic materials gave too slow impulses, and devices based on metal oxides - too fast. "We have achieved a time range that has not been demonstrated in artificial neurons before," he says.
The significance of the work goes far beyond medical applications. Hersam draws attention to the rapidly growing energy needs of artificial intelligence, emphasizing that the human brain is approximately 100,000 times more efficient than a digital computer in terms of energy consumption. "To meet the energy needs of AI, technology companies are building data centers with a power of gigawatts, powered by separate nuclear power plants," he notes. "Anyway, we need to develop much more energy-efficient hardware for AI."
Printed artificial neurons that can be integrated on flexible surfaces and communicate directly with living cells are emerging as a potential key to future neural implants and brain-inspired hardware accelerators - a step towards technologies that combine biology and computation on a whole new level.