MIT and Ericsson Set Goals for Zero-power Devices and a New Field—“Lithionics”

What if the future of neuromorphic computing depended on lithium, not silicon? Researchers from MIT and Ericsson are now exploring this question in more detail—specifically in researching lithium-oxide-based memristors as alternate computing technology.

Furthermore, the combined engineering teams from both MIT Materials Research Laboratory and Ericsson will also investigate an energy harvesting system from ambient RF signals.

 

"Frontier Research" in a New Field: Lithionics

In electronics, lithium is one of the most popular elements from the periodic table, right after copper.

Most of us are familiar with this material being used in lithium-based batteries. That's because lithium is a low-density metal with a very high energy-to-weight ratio and high electrochemical potential. These properties allow lithium-polymer and lithium-ion batteries to hold a charge in a relatively small footprint while being able to recharge using simple, straightforward circuits. 

Now, however, MIT and Ericsson researchers are rethinking how lithium can be used in ways beyond energy storage.

 

The merits of lithium oxide in brain-inspired computing. Image used courtesy of MIT MRSEC

 

In the past few years, two MIT professors Dr. Jennifer Rupp and Dr. Martin Bazant have discovered that in certain instances, battery electrodes made with lithium are also highly effective for other applications, including computing. This new field of research into alternative use cases is dubbed "lithionics." 

 

Does Moore's Law Hinge on Material Science Advances? 

Semiconductor computing is nearing its physical limitations because transistors cannot be smaller than the materials they’re made of. As such, physicists and engineers are ever searching for new materials to develop smaller and more powerful circuits.

Cue lithium oxides.

The joint research teams have recently found that these materials can be used as the basis of neuromorphic computer systems inspired by the human brain. Namely, lithium oxides can be manufactured into memristors for the circuits used in these systems. 

 

MIT researchers use a non-volatile memory effect called "resistive switching" in lithium-titanate compounds to adapt device performance by controlling the lithiation degree at the nanoscale level. Image used courtesy of MIT MRSEC

 

Memristors, also known as memory resistors, are two-terminal electrical components that can combine both the functions of data storage and data processing in one unit. These components can operate at multiple resistance states in order to encode information beyond just binary data. They also use much less energy than transistors. 

According to one of the leading researchers from Ericsson, Saeed Bastani, this project is "frontier research." It has required participants to find the best lithium compositions to develop memristors while also creating a manufacturing process for future memristor processing chips.

The goal of this new field—lithionics—is to overcome the obstacles of semiconductor computing while also developing small, fast, and low-power circuits that can match the energy efficiency of the human brain.

 

The Goal of Zero-power Devices

The second project that MIT and Ericsson are jointly working on is partially inspired by the RF waves typically used for TV and communication signals. The goal of this project is to develop self-powered systems that don’t require any batteries or external charging—which, of course, is of significant value to IoT developers. 

MIT and Ericsson teams are specifically investigating "zero-power devices" that perform completely autonomously with no outside wires attached to them. The researchers plan to achieve this feat via RF energy harvesting, collecting energy from radio waves to be used for wireless communication.

 

Working principle behind RF power harvesting. Image used courtesy of Micro and Nano Systems Letters

 

Because many devices transfer data wirelessly—phones, computers, TVs, and even cars—there is no shortage of devices capable of providing a wireless signal. This project aims to harvest energy from these types of signals to power smaller autonomous electronics via a larger network.

It's important to note that these zero-power devices will not be able to harvest power from ambient signals. Rather, they will harvest energy from specifically emitted, directed signals to the autonomous devices themselves (though these emitters could be embedded in appliances like TVs, cars, computers, etc.). 

 

MIT and Ericsson Divide and Conquer the Task

This wireless, over-the-air charging technology will require three different areas of research by multiple teams from MIT and Ericsson.

One team will focus on collecting the energy itself and developing circuitry to efficiently transform the RF energy into voltage. A second team will design the actual electronics, which must use extremely low voltages and operate with the collected energy from RF signals. And finally, a third team of researchers will design a networked system that will direct the power signal from the source device toward the smaller autonomous devices.

 

Last year, MIT conducted similar research to harness terahertz radiation into usable power. Image used courtesy of MIT

 

According to Jonas Hansryd from Ericsson, not every device needs to be powered via RF energy harvesting. Even having some devices powered through a network such as this would be a revolutionary and disruptive technology.  

Ericsson is also working with other companies to standardize this technology for potential use and more widespread consumer adoption.

 

Implications of Lithionics and Wireless Charging

Both of these research projects have the potential to drastically change the ways in which we use electronic devices. 

In certain use cases (like neuromorphic computing), lithium oxides may one day be a successor to conventional semiconductor chips. Future generations of memristor processors could vastly outperform current semiconductor ones using a fraction of the energy in an even smaller footprint. 

While it's difficult to power an entire electronic device through RF energy alone, the prospect of so-called zero-power devices may drive engineers to develop functional circuitry (that use extremely low voltages and appropriate emitters) to direct a power signal toward a nearby autonomous device.

MIT professor Tomas Palacios asserts that another important part of both of these projects is the interdisciplinary teamwork it will require. “This is an amazing example of what a true collaboration between industry and academia should be,” he notes.