(Nanowerk Spotlight) The human brain processes complex information while consuming merely the power of a dim light bulb. This remarkable efficiency stems from synapses, the connections between brain cells that can strengthen or weaken based on patterns of electrical signals. These synapses allow the brain to form both fleeting impressions and lasting memories through precise chemical interactions that modify their structure.
Electronic devices have struggled to achieve this adaptability. Conventional computer memory operates in fixed states – either volatile memory like DRAM that erases when powered off but switches quickly, or non-volatile memory like flash storage that retains information but requires high voltage to change states. Manufacturing these devices requires specialized facilities operating at temperatures above 800 degrees Celsius, consuming substantial energy and limiting options for creating more brain-like computing systems. The field particularly lacks materials that can smoothly transition between temporary and permanent memory states through controlled chemical processes similar to biological synapses.
Scientists from the Shenzhen Institute of Advanced Technology have developed artificial synapses that capture this biological adaptability through careful chemical modification of graphene. Their approach relies on the interaction between silver ions and phosphate groups attached to graphene surfaces. These phosphate groups reduce the energy required to form conductive pathways of silver atoms, enabling devices that can switch between short-term and long-term memory storage using much lower voltages than conventional electronics.
The team developed a two-step process to create their modified graphene. First, they mixed graphite powder with red phosphorus and used high-energy ball milling to break it into uniform flakes 50 to 100 nanometers wide. They then treated these flakes with nitric acid, which converted the phosphorus into phosphate groups chemically bonded to the graphene surfaces. Unlike traditional semiconductor manufacturing that requires high temperatures and vacuum conditions, this entire process occurs at room temperature using common laboratory equipment.
The researchers used these phosphorylated graphene films to build two types of memory devices called memristors. The first type, using just the modified graphene and silver electrodes, demonstrated short-term memory characteristics. When electrical signals passed through this device, silver ions from the electrode formed temporary conductive filaments within the graphene film. The phosphate groups lowered the energy barrier for this process, allowing filaments to form at voltages below 1 volt – far lower than the 10-15 volts required in flash memory. These filaments naturally dissolved when power was removed, resetting the device’s electrical resistance similar to how short-term memories fade without reinforcement.
The second type incorporated additional silver ions that could coordinate strongly with the phosphate groups on the graphene surfaces. This chemical interaction stabilized the conductive filaments through stronger bonds, allowing them to persist even without power – mimicking how long-term memories are stored through lasting structural changes in biological synapses. Microscopic analysis revealed that these stabilized filaments were both larger (30-120 nanometers versus less than 30 nanometers) and carried 20% more current than in the temporary-memory devices, explaining their permanence.
a) Schematic illustration of the synaptic plasticity in the human brain, forgetting/re-recognition function of volatile artificial synapses, and construction of an artificial neural network based on non-volatile artificial synapses; b) schematic illustration of the scale armor structure which inspires the design of phos-GPs based films and memristors. (Image: Reprinted with permission by Wiley-VCH Verlag)
The devices demonstrated sophisticated synaptic behaviors beyond simple memory storage. They showed paired-pulse facilitation and depression – temporary strengthening or weakening of signals based on the timing between electrical pulses – matching how biological synapses process information. When arranged into artificial neural networks, they achieved 94.7% accuracy in recognizing handwritten numbers and 81.7% accuracy on complex image classification tasks. These results match the performance of conventional silicon-based neural networks while potentially consuming far less power due to their lower operating voltages and simpler manufacturing process.
These artificial synapses maintained consistent performance through 500 operating cycles and continued functioning after 1,000 bending cycles, demonstrating both durability and flexibility not found in conventional electronics. The combination of room-temperature processing, mechanical flexibility, and sophisticated memory behavior could enable new types of adaptive computing systems that better match the efficiency of biological brains.
Moving this technology from laboratory demonstrations to practical applications requires addressing several key challenges. Manufacturing processes must be scaled up while maintaining the uniform size and chemical modification of the graphene flakes. Methods for precisely controlling the silver ion concentration and distribution need refinement to ensure consistent device performance. Additionally, while the lower operating voltages suggest improved energy efficiency, exact power consumption measurements in complex neural networks remain to be determined.
Nevertheless, this research demonstrates that carefully engineered chemical interactions can create memory devices matching the sophistication of biological synapses while maintaining the reliability needed for practical applications. Rather than requiring new fundamental breakthroughs, advancing these artificial synapses toward use in neuromorphic computing systems now depends primarily on engineering optimization.
Get our Nanotechnology Spotlight updates to your inbox!
Thank you!
You have successfully joined our subscriber list.
Become a Spotlight guest author! Join our large and growing group of guest contributors. Have you just published a scientific paper or have other exciting developments to share with the nanotechnology community? Here is how to publish on nanowerk.com.
