Graphene Flakes Make Laser Neuron Superfast

Chip's transistors are really lasers that spike with bursts of photons

2 min read
Graphene Flakes Make Laser Neuron Superfast
Illustration: Alexey Kotelnikov/Alamy

Tiny flakes of graphene may hold the key to building computer chips that can process information similar to the way the human brain does—only far faster—potentially leading to everything from better image recognition to control systems for hypersonic aircraft.

Researchers are developing so-called neuromorphic chips consisting of networks of transistors that interact the way neurons do, allowing them to process analog input, such as visual information, more quickly and accurately than traditional chips can.

One way of building such transistors is to construct them of lasers that rely on an encoding approach called “spiking.” Depending on the input, the laser will either provide a brief spike in its output of photons or not respond at all. Instead of using the on or off state of the transistor to represent the 1s and 0s of digital data, these neural transistors rely on the time intervals between spikes.

“We’re essentially using time as a way of encoding information,” says Bhavin Shastri, a Banting postdoctoral fellow in electrical engineering at Princeton University. Computation is based on the spatial and temporal positions of the pulses. “This is sort of the fundamental way neurons communicate with other neurons,” he says.

To see the advantages of this kind of processing, just try to swat a fly

Shastri describes his work, done along with Paul Prucnal, a professor of electrical engineering at Princeton University, in the current issue of Nature Scientific Reports. One of the challenges to building such chips, he says, is to get the laser to spike at picosecond time scales, one trillionth of a second. The team achieved this by placing a tiny piece of graphene inside a semiconductor laser. The graphene acts as a “saturable absorber,” soaking up photons and then emitting them in a quick burst.

Graphene, it turns out, makes a good saturable absorber because it can take up and release a lot of photons extremely fast, and it works at any wavelength; so lasers emitting different colors could be used simultaneously, without interfering with each other—speeding processing. It also stands up very well to all the energy produced inside a laser. The team demonstrated the effect using a relatively large fiber laser sitting on a benchtop, but it shouldn’t be difficult to scale it down to place many such lasers on a single chip, Shastri says.

To see the advantages of this kind of processing, Shastri says, just try to swat a fly. The fly’s neural network processes the shifting pattern of light it sees as your hand moves to strike it, allowing it to quickly take evasive action. A similar system could be useful to control supersonic jets, or it could activate an ejector seat when a fighter pilot is only just registering that a missile is about to hit. Such potential applications led the Defense Advanced Projects Research Agency to fund similar research into this kind of processor.

This post was corrected on 13 January 2016.

The Conversation (0)

3D-Stacked CMOS Takes Moore’s Law to New Heights

When transistors can’t get any smaller, the only direction is up

10 min read
An image of stacked squares with yellow flat bars through them.
Emily Cooper

Perhaps the most far-reaching technological achievement over the last 50 years has been the steady march toward ever smaller transistors, fitting them more tightly together, and reducing their power consumption. And yet, ever since the two of us started our careers at Intel more than 20 years ago, we’ve been hearing the alarms that the descent into the infinitesimal was about to end. Yet year after year, brilliant new innovations continue to propel the semiconductor industry further.

Along this journey, we engineers had to change the transistor’s architecture as we continued to scale down area and power consumption while boosting performance. The “planar” transistor designs that took us through the last half of the 20th century gave way to 3D fin-shaped devices by the first half of the 2010s. Now, these too have an end date in sight, with a new gate-all-around (GAA) structure rolling into production soon. But we have to look even further ahead because our ability to scale down even this new transistor architecture, which we call RibbonFET, has its limits.

Keep Reading ↓Show less