It’s no secret that the brain does not work like the electronic circuits of a computer. They have different architectures that more than one generation of scientists have dreamed of bringing together. While the brain stores and processes data in one place, the computer constantly transfers data between the processor and memory banks. The main problem is the lack of a suitable memory cell, which also plays the role of a transistor, which US scientists promised to help.

A team of scientists from Northwestern University, Boston College, and Massachusetts Institute of Technology (MIT) reported that they created and tested a transistor called a synaptic transistor. Neural networks with association. Scientists think that the main advantage of the development is that the transistor can operate at room temperature with an extremely low consumption of 20 pW (picowatts).

In living nervous tissue, the synapse is the gap between the end of one neuron and the beginning of another (if we are talking about the brain or spinal cord). In this interval, biochemical reactions occur that are responsible for further transmission or blocking of the nerve impulse. The transistor presented by scientists performs a similar function, but uses physical phenomena and processes in its work.

The technology generally belongs to the field of moire quantum materials. In most cases such materials operate under cryogenic cooling conditions. Therefore, it was important for the research team to demonstrate the effect they successfully achieved at room temperature.

A transistor, if you can call it that, is two atomic-thick overlapping layers of material slightly offset relative to each other in the horizontal plane. One layer is graphene and the second layer is boron nitride with a hexagonal cage. Turning one of them to a certain angle creates a moire pattern of two interconnected structures, and that’s the magic. Right angles from which expressive interactions occur are even called magical.

At certain angles of rotation, Coulomb interactions between two materials fall into the category of exotic electrical interactions that do not occur in ordinary materials, opening the potential for using such structures with precisely known functionality in future electronics.

To the researchers’ credit, they went further and created a series of neural circuits that demonstrated the ability for associative learning based on the conditional transistor transitions presented. Experimental circuits were trained to recognize groups of digits in binary coding, and they coped with this successfully. For example, neural circuits distinguished combinations 000 and 111 from 101, indicating the associative connectivity of the first and their difference from the third combination. Thus, the article was included in summary in the magazine. Nature Scientists, “The Moiré synaptic transistor enables efficient in-memory computing circuits and envisions advanced hardware accelerators for artificial intelligence and machine learning.”