Material Usage Could Hold the Key to an Improved IoT Infrastructure

Researchers at Purdue University are working with a new material that could pack more data processing punch with lower energy requirements.

New research on the atomic properties of a two-dimensional compound could change what’s possible in the expanding world of the IoT (Image courtesy of Purdue University.).

New research on the atomic properties of a two-dimensional compound could change what’s possible in the expanding world of the IoT (Image courtesy of Purdue University).

From the home to the factory floor, the rapid proliferation of the Internet of Things is stoking demand for devices that can stay up-to-speed on huge data volumes. Inherent in devices with such capacity, however, is a tendency to suck up correspondingly huge amounts of power. That trade-off has been one that product designers and end users are increasingly unwilling to accept. Enter Purdue University’s electrical engineering department, which has recently published research presenting a possible solution to this limitation.

In conjunction with the National Institute of Standards and Technology and Theiss Research Inc., the Purdue team’s work centers on a compound called molybdenum ditelluride. By layering this two-dimensional material into a computer chip, it could be possible to provide more number-crunching speed and save power simultaneously. The key could lie in leveraging a technology chipmakers have worked on for years known as resistive random access memory (RRAM).

How RRAM Works

In this memory technology, electricity flows through memory cells comprised of materials arranged in layers. Each of these layers represents a difference in resistance that archives data as a series of 0s and 1s. The layout of the 0s and 1s acts as an identifier for various bits of data that can then be processed by a computer and re-stored as memory. The sheer repetitive volume of this process necessitates that any materials used in such layering be hardy enough to handle at least several trillion of these operations.

That’s the handicap that explains why RRAM has yet to gain much traction in spite of its obvious potential in IoT and AI applications. The compounds used in stacked memory cells, to this point, have failed to hold up. The Purdue research suggests that molybdenum ditelluride could be different. It allows for speedier switching between 0s and 1s than previously tried materials, which in turn accelerates the sequence of storage and retrieval that characterizes RRAM. The lower but still present resistance between layers of molybdenum ditelluride strikes the important balance of allowing for the basic functionality of a RRAM system without creating a needlessly high barrier—leading to longer battery life.

Looking Ahead

The next step for the researchers could be coming up with a fully integrated chip system that uses molybdenum ditelluride-enabled RRAM. They envision a stacked memory cell that incorporates logic and interconnects, which would require some creative material usage beyond just molybdenum ditelluride. Their thinking is that if logic and interconnects could come together in a two-dimensional space, lower friction would improve both processing power and battery usage. Considering the Purdue Office of Technology Commercialization has already filed two related patents, a mainstream rollout of high-speed, long-life IoT devices could be closer than ever before.

The notion that increased battery efficiency will be essential in unlocking IoT’s full potential is widely held. For more on this topic, check out this article on how one chipmaker is looking to develop a “forever battery.”