Research from Osaka University developed nanomaterial based device that partly replicating brain functions.
The human brain requires less energy to adapt and learn according to the surroundings. Furthermore, it can make ambiguous recognitions and exhibit high recognition ability and intelligence along with performing complex information processing. The neural circuits of brain possess learning ability of synapses and nerve impulses or spikes. Although various scientific progresses have gradually clarified brain structure, it is complicated to entirely emulate nerve impulse generation. Now, a research by Kyushu Institute of Technology and Osaka University studied control of rectifications in various molecular junctions and particles absorbed on single-walled carbon nanotube (SWNT). The team used current sensing atomic force microscopy (CS-AFM), and found that a negative differential resistance was produced in polyoxometalate (POM) molecules absorbed on SWNT. This resistance reveals the unstable dynamic non-equilibrium state in molecular junctions. Furthermore, the researchers created SWNT/POM network molecular neuromorphic devices are remarkable devices composed of a graphite sheet rolled into a seamless cylinder that are extremely dense and random and capable of generating spontaneous spikes similar to nerve impulses of neurons.
POM is composed of metal atoms and oxygen atoms blended together to form a 3-dimensional framework and are capable of storing charges in a single molecule. The researches theorized that negative differential resistance that increased voltage across the device’s terminals to decrease electric current through it and spike generation from the network were a result of non-equilibrium charge dynamics in molecular junctions. Hence, simulation calculations of the random molecular network models were conducted with POM molecules. These molecules store electric charges by replicating spikes generated from the random molecular network. Such molecular model can replace as a component of reservoir computing devices, which is a next-generation artificial intelligence (AI). The research was published Nature Communications on July 12, 2018 is expected to contribute to the development of neuromorphic devices of the future.