Transistor mimics the activities of a brain neuron
doi:10.1038/nindia.2020.92 Published online 2 June 2020
An international research team has designed a neurotransistor that mimics the activities of a brain neuron1. The team merged the learning and memory functions of a neuron in this device, making it act as an artificial neuron.
The device could potentially be used for artificial intelligence systems that can process information using a network.
Transistors, according to recent studies, could potentially be used to emulate the functions of neurons.
The scientists, including an Indian researcher from the University of Calcutta in Kolkata, India,, invented the neurotransistor using silicon nanowires on an eight-inch insulated silicon wafer. They then covered the transistor with a metal-ion-doped sol-gel-derived silicate film that has multiple input gates and output channels.
Experiments and simulation studies show that the transistor functions like a random-access-memory device. It doesn’t need an extra device that stores electrical charge, such as a capacitor.
The film that covers the transistor mimics the local or global ion movement through the ion channels of neuronal membrane. This allows mobility of metal and non-metal ions in the film.
Unlike existing devices, which have a fixed memory, the transistor’s memory can be tuned by varying its charging capacity in response to various voltages. The mobile ions in the film aid in this process.
The researchers say that it is possible to scale up the production of the transistor using a conventional transistor-making process. Such a possibility means that this device could be used to make complex computing devices, they add.
1. Baek, E. et al. Intrinsic plasticity of silicon nanowire neurotransistors for dynamic memory and learning functions. Nat. Electron.(2020) doi: 10.1038/s41928-020-0412-1