A new computer component developed by researchers at KTH Royal Institute of Technology and Stanford University mimics the human brain by acting like a synaptic cell. The new component is called an electrochemical random access memory (ECRAM).
Rise of Neuromorphic Computers
The ECRAM memory components were made with 2D titanium carbide, and they demonstrated an impressive ability to complement classical transistor technology. They are enabling the commercialization of powerful computers modeled after the brain’s neural network. These neuromorphic computers have the potential to be far more energy efficient than today’s computers.
The ECRAM has an architecture that is dramatically different from classic computing, and it acts as a synaptic cell in an artificial network.
Max Hamedi is a KTH Associate Professor.
“Instead of transistors that are either on or off, and the need for information to be carried back and forth between the processor and memory — these new computers rely on components that can have multiple states, and perform in-memory computation,” Hamedi says.
The team of scientists at KTH and Stanford have been working towards testing more efficient materials for building an ECRAM. In order to make these chips commercially viable, they require materials that can overcome the slow kinetics of metal oxides, as well as the unstable temperature of plastics.
The researchers fabricated a material known as MXene, which is a 2D compound that is just a few atoms thick and consists of titanium carbide. MXene combines the high speed of organic chemistry and the integration compatibility of inorganic materials.
MXene ECRAMs combine the speed, write noise, linearity, switching energy, and endurance metrics that are needed for parallel acceleration of artificial neural networks (ANNs).
Professor Alberto Salleo at Stanford University is co-author of the research.
“MXenes are an exciting materials family for this particular application as they combine the temperature stability needed for integration with conventional electronics with the availability of a vast composition space to optimize performance,” Salleo says.
According to Hamedi, there are still many barriers that must be overcome if consumers are going to be able to buy their own neuromorphic computers. However, the 2D ECRAMs are a major breakthrough in the area of neuromorphic materials. They could enable AI that is capable of adapting to confusing input and nuance, similar to the human brain. At the same time, it would require far less energy consumption.