Samedi 22 Juillet 2017
taille du texte
   
Lundi, 09 Janvier 2012 18:51

Aussie Brains Move Chip Design to Quantum Realm

Rate this item
(0 Votes)
Aussie Brains Move Chip Design to Quantum Realm

Are Moore's Law fans happy to see this atomic wire? Image: Bent Weber

The law of the land is Ohm’s Law — even when the land is really, really small.

Contradicting what was previously thought, researchers at the University of New South Wales have announced that the law governing electrical resistivity — how readily electrical current flows through a material — extends into the quantum realm.

This has major significance for chipmakers, who are starting to wrestle with the forces of quantum physics as transistors and interconnect sizes shrink down to a few dozen nanometers. These forces are among the barriers threatening to bring another famous law to a halt. Moore’s Law holds that number of transistors you can fit on a chip doubles about every eighteen months.

Computer chips are made of millions of transistors, tiny semiconductor switches that control the flow of electricity. Groups of transistors are combined to make logic gates, which are the building blocks of all digital devices, and connecting transistors requires wires — very tiny wires.

Previous research showed that resistance in wires narrower than 10 nanometers on silicon chips increases exponentially, but the Australian researchers were undeterred. They built electrical wires that are four atoms wide and one atom tall, and these are as conductive as the copper wiring in your house. One implication is that Moore’s Law will last a few generations longer than expected — at least as far as interconnect wiring is concerned.

“We can make interconnects in silicon all the way to the level of atoms,” says Michelle Simmons, a physics professor at the University of New South Wales.

The researchers made the wires by depositing phosphorus atoms on silicon. By placing the atoms less than a nanometer apart, they overlapped the atoms’ wavefunctions — the quantum fuzziness that governs where and how electrons move around the atoms.

A critical step is encasing the phosphorus atoms in more silicon. Otherwise, electrons moving along the phosphorus atoms would become stuck at the exposed surfaces. “By encapsulating the wires in silicon and having such a high density of phosphorus atoms, we can protect the central conducting core of the wire,” says Simmons.

Of course, the researchers built their wires using a scanning probe microscope, a laboratory tool useful for building one-offs but wholly inadequate to the task of mass-producing computer chips. “To date this technology is not industry compatible,” says Simmons. “That said, there are companies working towards ‘atomically precise manufacturing’, such as Zyvex Labs.”

Innovations are continuously being developed in the lab, but only a small percentage actually find there way into the outside world. “The technology must allow for volume manufacturing at high yields and at a reasonable cost before it will ever be implemented,” says Jim McGregor, chief technology strategist for market research firm In-Stat

There are three pillars to chipmaking innovation: lithography, materials, and transistor design. The Australian advance is in the area of materials. “Right now, lithography appears to be the biggest challenge,” says McGregor.

The industry is looking to packaging technology as the next major advancement. It can change the dynamics of Moore’s Law by stacking layers — going tall as well as small. This should allow for at least a few more generations of innovation, acccording to McGregor.

Quantum chips

Even if the phosphorus atom wires never see the inside of a commercial chipmaking plant, they’re a big step forward for quantum computer chips. Simmons and her colleagues built the wires as part of more-than-decade long project aimed at building chip-based quantum computers. Simmons directs the Australian Centre of Excellence for Quantum Computation and Communication Technology.

The team has built quantum bits, or qubits, from individual phosphorus atoms embedded in silicon. Atoms spin, not unlike a child’s spinning top. By subtly controlling this spin, researchers have been able to perform quantum logic. Quantum computers — still largely confined to be laboratory after several decades of research — promise to be faster than any ordinary computer ever could be at solving certain problems, including cracking secret codes.

“We realized fairly early on that to control these single-atom spins we would need interconnects at the same scale as the atoms themselves,” says Simmons.

Thus were born the atomic-scale conducting wires. Within the world of quantum computing, this is a fairly big deal. Most prototype quantum computers are made of bulky laboratory equipment like ion traps. To be practical, quantum computers will need to be made from chips, just like ordinary computers. And for this to happen, qubits, like ordinary transistors, will need to be connected by very tiny wires.

So while Intel probably views this research as one of thousands of advances to take note of and file away, somewhere, some nascent quantum computer startup is probably very excited.

Authors:

French (Fr)English (United Kingdom)

logo-noemi

Parmi nos clients