New Memory Device Can Take the Heat

A new memory device based on gallium nitride can operate at 300 degrees Celsius

2 min read
Researchers (from left) Houqiang Fu, Yuji Zhao, Kai Fu work on the memory device at a probe station.
Researchers [from left] Houqiang Fu, Yuji Zhao, and Kai Fu work on the memory device at a probe station.
Photo: Yuji Zhao

On the bleak, cratered surface of Mercury, temperatures can reach some 430 degrees Celsius in the daytime. Almost twice the distance away from the sun, Venus has a similar surface temperature of around 462 Celsius, thanks to an atmosphere rich in carbon dioxide.

“Everything that you send there has to work under these temperatures,” says Yuji Zhao, a materials scientist at Arizona State University. That includes electronic systems needed in instruments, sensors, and probes. While no mission has landed on Mercury’s surface to date, the longest-surviving probe that landed on Venus in 1982—the Soviet Union’s Venera 13—lasted only 127 minutes before failing. (In comparison, the Curiosity rover landed on Mars in 2012 and is still going strong.)

Toward the next generation of electronics that can survive high temperatures, Zhao and his team recently reported gallium nitride memory devices that operate from 25 to 300 degrees Celsius in IEEE Electron Device Letters. The research is funded by NASA’s Hot Operating Temperature Technology (HOTTech) program to support future missions to Mercury and Venus.

“There’s empty space here,” says Zhao. “There’s no technology that can sustain this high temperature.”

Gallium nitride is a good candidate for high-temperature electronics because of its large bandgap. Traditional silicon has a bandgap of only 1.12 eV. That means, with a small temperature rise, electrons will easily get excited and transition from the valence band into the conduction band, explains Zhao. “As a result, you lose control of the carriers and your device will malfunction.” In contrast, gallium nitride has a bandgap of 3.4 eV, allowing devices to tolerate higher temperatures before their electrons go rogue. Gallium nitride is not the only semiconducting material with a wide bandgap being studied for high-temperature electronics; NASA has also invested in a close rival, silicon carbide, for the HOTTech program.

The memory device was fabricated by chemical vapor deposition on a gallium nitride substrate. Key to the device’s performance were the etching and regrowth processes during fabrication, says Zhao. After several layers of gallium nitride were deposited, some areas were etched away with plasma, then regrown. That created an interface layer with vacancy sites that are missing nitrogen atoms, says Zhao. “The interface layer is critical for the memory effect,” he says. The researchers believe that the nitrogen vacancies are responsible for capturing and releasing electrons, giving rise to high- and low-resistance states—or 0 and 1 states—in the device.

At room temperature, the device showed stable switching with almost no deterioration between 0 and 1 states to a thousand cycles. The researchers then tested its performance at high temperatures going up to 300 degrees Celsius, at which the device also maintained stable switching between 0 and 1 states for another thousand cycles. Above 350 degrees Celsius, the device lost its memory effect. But its performance returned after the device was brought back to room temperature, says Zhao. “This device is actually very robust,” he says.

Further evaluation is required: Zhao and his team are now testing another version of the device for stability up to 500 degrees Celsius, and for long-term stability. The team is also investigating the role of the nitrogen vacancies for the device’s performance. Once NASA deems a prototype good enough, it will have to undergo testing in controlled chambers that mimic the harsh environments on Mercury and Venus at NASA facilities, says Zhao. “I will say there is several years of work to do, but the initial result is definitely very, very encouraging and exciting,” he says.

The Conversation (0)

The Future of Deep Learning Is Photonic

Computing with light could slash the energy needs of neural networks

10 min read

This computer rendering depicts the pattern on a photonic chip that the author and his colleagues have devised for performing neural-network calculations using light.

Alexander Sludds
DarkBlue1

Think of the many tasks to which computers are being applied that in the not-so-distant past required human intuition. Computers routinely identify objects in images, transcribe speech, translate between languages, diagnose medical conditions, play complex games, and drive cars.

The technique that has empowered these stunning developments is called deep learning, a term that refers to mathematical models known as artificial neural networks. Deep learning is a subfield of machine learning, a branch of computer science based on fitting complex models to data.

Keep Reading ↓ Show less