2D materials could be used to simulate brain synapses in computers

Computers could mimic neural networks in the brain – and be much more energy efficient – with a new computing component that mimics how the brain works by acting like a synaptic cell. It’s called electrochemical random access memory (ECRAM), and researchers have developed materials that offer a commercially viable way to build these components.

Researchers from the KTH Royal Institute of Technology and Stanford University have now made a material for computer components that enables the commercial viability of computers that mimic the human brain.

Electrochemical random access memory (ECRAM) components fabricated with 2D titanium carbide have shown exceptional potential to complement conventional transistor technology and contribute to the commercialization of powerful computers modeled after the neural network of the brain. Such neuromorphic computers can be thousands of times more energy efficient than today’s computers.

These computing advances are possible due to some fundamental differences from the classical computer architecture used today and from ECRAM, a component that acts as a kind of synaptic cell in an artificial neural network, explains Max Hamedi, associate professor at KTH.

“Instead of transistors that are on or off, and the need to cycle back and forth between processor and memory, these new computers rely on components that can have multiple states and perform in-memory calculations” , Hamedi said. said.

KTH and Stanford scientists focused on testing better materials to build an ECRAM, a component in which switching occurs by inserting ions into an oxidation channel, in a similar sense to our brains which also works with ions. What was needed to make these chips commercially viable were materials that overcome the slow kinetics of metal oxides and poor temperature stability of plastics.

The key material in the ECRAM units made by the researchers is called MXene – a two-dimensional (2D) compound, just a few atoms thick, made of titanium carbide (Ti3VS2JX). The MXene combines the high speed of organic chemistry with the integration compatibility of inorganic materials in a single device operating at the crossroads of electrochemistry and electronics, explains Hamedi.

Co-author Professor Alberto Salleo of Stanford University says MXene ECRAMs combine speed, linearity, write noise, switching energy, and endurance metrics essential for acceleration parallel artificial neural networks.

“MXenes are an exciting family of materials for this particular application because they combine the temperature stability needed for integration with conventional electronics with the availability of ample composition space to optimize performance,” says Salleo.

Although there are many other hurdles to overcome before consumers can buy their own neuromorphic computers, Hamedi says 2D ECRAMs represent a breakthrough at least in the field of neuromorphic materials, potentially leading to artificial intelligence that can s adapt to confusing inputs and nuances, the way the brain does with thousands of times less energy consumption. It can also enable portable devices capable of performing much heavier computing tasks without having to rely on the cloud.

Source of the story:

Material provided by KTH, Royal Institute of Technology. Original written by David Callahan. Note: Content may be edited for style and length.