Upcoming Engineer Logo

Have you ever imagined how energy-efficient computing platforms would be?

Global power problems are largely attributable to the enormously energy-intensive rapid expansion of data centres. There is an urgent need to provide alternatives to standard electrical components that will make these systems more energy-efficient given the growing need for faster and more sophisticated computers and gadgets. Researchers from the Centre for Nano Science and Engineering (CeNSE), IISc, announce the creation of a highly energy-efficient computing platform in two recent investigations. This platform holds promise for the construction of next-generation electronic devices.

The researchers employed memristors, components that can both store data and do the computation, in place of complementary metal-oxide semiconductors (CMOS), the modern-day building blocks of most electronic circuits. The team was able to significantly increase speed and efficiency by reducing the number of components required in a circuit by creating novel memristors based on metal-organic complexes.

The two studies that were published in Advanced Materials were led by Sreetosh Goswami, Assistant Professor at CeNSE, who claims, “We have now discovered a molecular circuit element that can capture complex logic functions within itself, facilitating in-memory computations in a smaller number of time steps and using much fewer elements than usual.”

Data processing and storing are currently done at different physical sites in computer designs. The majority of the computational power is used in the back-and-forth communication between two places. “By conducting both processing and storage at the same physical site, we are overcoming this difficulty,” he claims.

According to Goswami, the platform “outperforms” existing cutting-edge technology by orders of magnitude. Even when compared to commercial technology like flash storage, we can now create arrays of devices that are more reliable, consistent, and stable. Because they perform tasks sequentially, earlier designed memristor-based circuits also have speed restrictions and a higher likelihood of mistakes building up. The researchers claim that the new platform’s architecture minimises operating processes, improving speed and lowering error.

Sreebrata Goswami, a Specialist Scientist at CeNSE, created the metal-organic compounds that were utilised to construct their platform. He claims that these “complexes” are similar to electron sponges that can absorb and release electrons without degrading for billions of cycles. Researchers may be able to modify the same circuit for various tasks by making minor chemical changes, such as changing one or two ions in the complexes.

The researchers discovered that the new platform delivered 47 times more energy efficiency and 93 times quicker operating speed while only taking up 9% of the physical footprint when they created circuits that perform mathematical operations and compared them with a normal CMOS circuit.

The platform will be connected to a sensor in the future, such as a touch-sensitive smartphone screen so that the team can examine how well it analyses the data it gathers. A postdoctoral scholar at CeNSE named Santi Prasad Rath adds, “In an Internet of Things (IoT) platform, this computing technique may be highly valuable.” Deepak and Santi developed and built the circuit together.

These initiatives are crucial because, according to experts, the efficiency and performance of CMOS technology will soon be at a point where it cannot be pushed up anymore. In order to maintain Moore’s Law over the next decades, new nanoscale device structures must be developed, according to Navakanta Bhat, a professor at CeNSE and an authority in CMOS technology.

“It is pretty remarkable that a molecular platform that is still in its infancy is surpassing an established technology. This is risky research that might influence the direction of our country’s semiconductor electronics mission in the future.”

You might also be interested in reading A new AI game from DeepMind plays Stratego at a high level.

Leave a Comment