Communications of the ACM
Quantum Computer Chips Now One Step Closer to Reality
"We have a technique that manufacturers could potentially use to fabricate quantum devices directly on a silicon chip, side-by-side with their regular circuits and switches," says Ohio State University Professor Paul Berger.
Credit: Ohio State University
In the quest for smaller, faster computer chips, researchers are increasingly turning to quantum mechanics — the exotic physics of the small. The problem: the manufacturing techniques required to make quantum devices have been equally exotic.
That is, until now.
Researchers at Ohio State University have discovered a way to make quantum devices using technology common to the chip-making industry today. This work might one day enable faster, low-power computer chips. It could also lead to high-resolution cameras for security and public safety, and cameras that provide clear vision through bad weather.
Paul Berger, professor of electrical and computer engineering and professor of physics at Ohio State University, and his colleagues report their findings in an upcoming issue of IEEE Electron Device Letters.
The team fabricated a device called a tunneling diode using the most common chip-making technique, called chemical vapor deposition.
"We wanted to do this using only the tools found in the typical chip-maker's toolbox," Berger says. "Here we have a technique that manufacturers could potentially use to fabricate quantum devices directly on a silicon chip, side-by-side with their regular circuits and switches."
The quantum device in question is a resonant interband tunneling diode (RITD) — a device that enables large amounts of current to be regulated through a circuit, but at very low voltages. That means that such devices run on very little power.
RITDs have been difficult to manufacture because they contain dopants — chemical elements — that don't easily fit within a silicon crystal.
Atoms of the RITD dopants antimony or phosphorus, for example, are large compared to atoms of silicon. Because they don't fit into the natural openings inside a silicon crystal, the dopants tend to collect on the surface of a chip.
"It's like when you'e playing Tetris and you have a big block raining down, and only a small square to fit it in. The block has to sit on top," Berger says. "When you're building up layers of silicon, these dopants don't readily fit in. Eventually, they clump together on top of the chip."
In the past, researchers have tried adding the dopants while growing the silicon wafer one crystal layer at a time — using a slow and expensive process called molecular beam epitaxy, a method which is challenging for high-volume manufacturing. That process also creates too many defects within the silicon.
Berger discovered that RITD dopants could be added during chemical vapor deposition, in which a gas carries the chemical elements to the surface of a wafer many layers at a time. The key was determining the right reactor conditions to deliver the dopants to the silicon, he found.
"One key is hydrogen," he says. "It binds to the silicon surface and keeps the dopants from clumping. So you don't have to grow chips at 320° Celsius [approximately 600° Fahrenheit] like you do when using molecular beam epitaxy. You can actually grow them at a higher temperature like 600° Celsius [more than 1,100° Fahrenheit] at a lower cost, and with fewer crystal defects."
Tunneling diodes are so named because they exploit a quantum mechanical effect known as tunneling, which lets electrons pass through thin barriers unhindered.
In theory, interband tunneling diodes could form very dense, very efficient micro-circuits in computer chips. A large amount of data could be stored in a small area on a chip with very little energy required.
Researchers judge the usefulness of tunneling diodes by the abrupt change in the current densities they carry, a characteristic known as "peak-to-valley ratio." Different ratios are appropriate for different kinds of devices. Logic circuits such as those on a computer chip are best suited by a ratio of about 2.
The RITDs that Berger's team fabricated had a ratio of 1.85. "We're close, and I'm sure we can do better," he says.
He envisions his RITDs being used for ultra-low-power computer chips operating with small voltages and producing less wasted heat.
"Chip makers today are having a great difficulty boosting performance in each generation, so they pack chips with more and more circuitry, and end up generating a lot of heat," Berger says. "That's why a laptop computer is often too hot to actually sit atop your lap. Soon, their heat output will rival that of a nuclear reactor per unit volume. That's why moving to quantum devices will be a game-changer."
RITDs could form high-resolution detectors for imaging devices called focal plane arrays. These arrays operate at wavelengths beyond the human eye and can permit detection of concealed weapons and improvised explosive devices. They can also provide vision through rain, snow, fog, and even mild dust storms, for improved airplane and automobile safety, Berger said. Medical imaging of cancerous tumors is another potential application.
His coauthors on the paper included Si-Young Park, and R. Anisha, both doctoral students in electrical engineering at Ohio State; and Roger Loo, Ngoc Duy Nguyen, Shotaro Takeuchi, and Matty Caymax, all of IMEC, an industrial research center in Belgium.
This work was partially supported by the U.S. National Science Foundation.
No entries found