Self-Healing Transistors for Chip-Scale Starships

A pair of stars, Alpha Centauri A and Alpha Centauir B, form part of the closest star system to our own.
Photo: ESA/NASA
Alpha Centauri A and Alpha Centauir B, form part of the closest star system to our own.

Working with the Korea Institute of Science and Technology (KAIST), NASA is pioneering the development of tiny spacecraft made from a single silicon chip that could slash interstellar exploration times.

On Wednesday at the International Electron Devices Meeting in San Francisco, NASA’s Dong-Il Moon will present new technology aimed at ensuring such spacecraft survive the intense radiation they’ll encounter on their journey.

If a silicon chip were used as a spacecraft, calculations suggest that it could travel at one-fifth of the speed of light and reach the nearest stars in just 20 years. That’s one hundred times faster than a conventional spacecraft can offer.

Twenty years in space is still too long for an ordinary silicon chip, because in addition to the frailties it suffers on earth, such as swings in temperature, it is bombarded by radiation of very high energy. This radiation leads to the accumulation of positively charged defects in the chip’s silicon dioxide layer, where they degrade device performance. The most serious of the impairments is an increase in the current that leaks through a transistor when it is supposed to be turned off, according to Yang-Kyu Choi, leader of the team at KAIST, where the work was done. However, there are also other issues, such as a shift in the voltage at which the transistor turns on.

Two options for addressing chip damage are to select a path through space that minimizes radiation exposure and to add shielding. But the former leads to longer missions and constrains exploration, and the latter adds weight and nullifies the advantage of using a miniaturized craft. A far better approach, argues Moon, is to let the devices suffer damage but then to add a an extra contact to the transistors, and use this contact to heal the devices with heating.

“On-chip healing has been around for many, many years,” says Jin-Woo Han, a member of the NASA team. Milestones including the revelation in the 1990s— by a team at the National Microelectronics Research Centre in Cork, Ireland— that heating could drive the recovery of radiation sensors, and far more recently, heat-induced healing of flash memory by Macronix of Taiwan. The critical addition made now, Han says, is the most comprehensive analysis on radiation damage.

This study uses KAIST’s experimental “gate-all-around” nanowire transistor. Gate-all-around nanowire transistors use nanoscale wires as the transistor channel instead of today’s fin-shaped channels. The gate, the electrode that turns on or off the flow of charge through the channel, completely surrounds the nanowire. Adding an extra contact to the gate allows you to pass current through it. That current heats the gate and the channel it surrounds, fixing any radiation-induced defects.

Nanowire transistors are ideal for space, according to KAIST, because they have a relatively high degree of immunity to cosmic rays and because they are very small, with dimensions in the tens of nanometers. “The typical size for [transistor-dimensions on] chips devoted to spacecraft applications is about 500 nanometers,” says Choi. “If you can replace 500 nanometer feature sizes with 20 nanometers feature sizes, the chip size and weight can be reduced.” Costs fall too.

The gate-all-around device may not be that well known today, but production is expected to rocket in the early 2020s, when silicon foundries will use it in place of the today’s FinFET for producing circuits featuring transistors with gate lengths smaller than 5-nm.

KAIST’s has been used to form three key building blocks for a single-chip spacecraft: a microprocessor, a DRAM memory for supporting this, and a flash memory that can serve as a hard disk.

Repairs to radiation-induced damage can be made many times, with experiments showing that flash memory can be recovered up to around 10,000 times and DRAM returned to its pristine state 1012 times. With logic devices, an even higher figure is expected. These results indicate that a lengthy interstellar space mission could take place, with the chip powered down every few years, heated internally to recover its performance, and then brought back to life.

Adding a second gate for heating is not ideal, because it modifies chip design and demands the creation of a new transistor library, which escalates production costs. To address this, those at KAIST are investigating the capability of a junctionless transistor that heats the channel during normal operation when current flows through it. Separately, at NASA researchers are developing on-chip embedded microheaters that are compatible with standard circuits.  

Cutting the costs of self-healing tech is critical to the future of the program. It will help to increase the appeal of the technology, which will require many more years of investment if the launch of the first silicon-chip spacecraft is to get off the ground.

Advertisement

Tech Talk

IEEE Spectrum’s general technology blog, featuring news, analysis, and opinions about engineering, consumer electronics, and technology and society, from the editorial staff and freelance contributors.

Newsletter Sign Up

Sign up for the Tech Alert newsletter and receive ground-breaking technology and science news from IEEE Spectrum every Thursday.

Advertisement