Engineers create transistors that are 10,000 times faster than brain synapses

Researchers at MIT have developed programmable transistors

working in 10 thousandtimes faster than brain synapses. The technology used for analog machine learning provides not only high data processing speed, but also good energy efficiency.

The working mechanism of the device iselectrochemical incorporation of a tiny ion, a proton, into an insulating oxide to modulate its electronic conductivity, the authors explain. Scientists used a powerful electric field to accelerate protons and put ion transistors into nanosecond mode of operation.

Researchers note that the secret of newdevices using inorganic phosphosilicate glass. It provides ultra-fast movement of protons because it contains many nanometer-sized pores, the surfaces of which provide pathways for the diffusion of elementary particles. It can also withstand very strong pulsed electric fields.

Action potential in biological cellsrises and falls on a time scale of milliseconds as a potential difference of approximately 0.1 volts is limited by the stability of the water. In our work, we apply up to 10 volts through a special, nano-thick, hard glass film that conducts protons without damaging it. And the stronger the field, the faster the ion devices work.

Yu Li, professor at the Massachusetts Institute of Technology and co-author of the work

In the human brain, learning occurs throughstrengthening and weakening of connections between synapses. Deep neural networks use a similar strategy when the weights of the nodes (neurons) are programmed with learning algorithms. When using processors, increasing and decreasing the electrical conductivity of proton resistors provides analog machine learning.

Conductivity is controlled by the movement of protons.To increase conductivity, more protons are pushed into the channel in the resistor, and to decrease conductivity, protons are expelled. This is achieved by using an electrolyte (similar to a battery) that conducts protons but blocks electrons. Increasing the speed of protons significantly speeds up the process of machine learning.

Cover image: Artistic illustration of a proton programmable resistor. Source: Ella Maru Studio, Murat Onen, MIT

Read more:

The supersonic plane will fly at a speed of 2,000 km/h and cross the ocean in 3.5 hours

Scientists filmed a strange creature with tentacles, which they mistook for a flower

Created a quantum computer that "went beyond the binary system"