The July 2022 issue of IEEE Spectrum is here!

Close bar

Atom-Thin Memristors Discovered

Experts thought memory in 2D materials was impossible. Then engineers in Texas discovered “Atomristors”

2 min read
Illustration of a voltage-induced memory effect in monolayer nanomaterials, which layer to create 'atomristors,' the thinnest memory storage device that could lead to faster, smaller and smarter computer chips.
Illustration: Cockrell School of Engineering

Two-dimensional atom-thin materials are good for a lot of things, but they don’t make good memory devices. At least that’s what everyone thought until  Ruijing Ge, a first-year graduate student at the University of Texas, Austin, persuaded her mentor—flexible electronics guru Deji Akinwande—to let her try. They sandwiched an atom-thick layer of molybdenum disulfide between two electrodes and found that, contrary to expectation, the structure displayed memristance; it can be set to a high resistance or low resistance state by particular voltages and remain stable long after the voltage is removed.

It’s not completely clear why it works, but these “atomristors,” as Akinwande has christened them, could have a big impact—and not just as memory devices. They could serve as switches in radios of 5G smartphones and Internet-of-Things gadgets, and as computational elements in brain-inspired artificial intelligence circuits.

Ordinary memristors are made of oxide materials sandwiched between two conductors. The resistance across the oxide changes when a high current in one direction moves oxygen atoms vertically through the oxide. The original resistance is restored by switching the direction of the current, putting the oxygen back in its place.

But that can’t be what’s happening in an atomristor. There is neither oxygen nor a vertical direction for it to move. Instead, Akinwande hypothesizes that defects in the 2D crystal lattice—the holes left by occasional missing sulfur atoms, for instance—are what are moving around. Voltage of one polarity attracts the defects, bunching them together in a way that decreases the resistance across the material. Switching the polarity scatters the defects, ramping the resistance back up.

That’s the theory, at least. Akinwande says his group is collaborating with one of the U.S. National Labs, which have the kind of microscope that should allow them to see defects moving during operation.

Importantly, this memristance property seems to be common to the whole class of 2D materials to which molybdenum disulfide belongs. That class is called transition metal dichalcogenides, and they are “the premier 2D semiconductor,” says Akinwande. Collaborating with materials scientists in Yanfeng Zhang’s group at Peking University, which supplied several of the materials, they tested molybdenum diselenide, tungsten disulfide, and tungsten diselenide as well (all less than 1 nanometer thick). In every case, it worked. “We tried four or five, and there are possibly hundreds of them,” says Akinwande.

In addition to clarifying how atomristors work and determining how many types there are, his lab plans to work on some interesting applications for them. They’ve already demonstrated an ultra-thin 2D version of the memristor cell by building the conductive parts out of graphene. And Akinwande believes it should be possible to integrate 3D arrays of memristors atop finished silicon logic chips, because the materials can be transferred at room temperature. The end result could be neuromorphic systems with a brain-like density of connections. “The sheer density of memory storage that can be made possible by layering these synthetic atomic sheets onto each other, coupled with integrated transistor design, means we can potentially make computers that learn and remember the same way our brains do,” Akinwande said in a press release.

But atomristors might have a much more down-to-earth use. The RF switches used to channel radio signals into and out of wireless devices are best when they consume little power when not switching but can pass a lot of current through them. Atomristors seem to fit that job quite well, as they don’t need any power to hold their states. Akinwande’s lab has shown that the atomristors can channel frequencies as high as 50 GHz—which includes the key 5G bands—through them.  

The Conversation (0)

3 Ways 3D Chip Tech Is Upending Computing

AMD, Graphcore, and Intel show why the industry’s leading edge is going vertical

8 min read
Vertical
A stack of 3 images.  One of a chip, another is a group of chips and a single grey chip.
Intel; Graphcore; AMD
DarkBlue1

A crop of high-performance processors is showing that the new direction for continuing Moore’s Law is all about up. Each generation of processor needs to perform better than the last, and, at its most basic, that means integrating more logic onto the silicon. But there are two problems: One is that our ability to shrink transistors and the logic and memory blocks they make up is slowing down. The other is that chips have reached their size limits. Photolithography tools can pattern only an area of about 850 square millimeters, which is about the size of a top-of-the-line Nvidia GPU.

For a few years now, developers of systems-on-chips have begun to break up their ever-larger designs into smaller chiplets and link them together inside the same package to effectively increase the silicon area, among other advantages. In CPUs, these links have mostly been so-called 2.5D, where the chiplets are set beside each other and connected using short, dense interconnects. Momentum for this type of integration will likely only grow now that most of the major manufacturers have agreed on a 2.5D chiplet-to-chiplet communications standard.

Keep Reading ↓Show less
{"imageShortcodeIds":[]}