Atomic-Level Computing

Andreas Heinrich of IBM Research — New research involving antiferromagnets by Andreas Heinrich and his IBM colleagues is exploring the transition from quantum mechanics to classical physics.

The defining characteristics of quantum computing and “classical” digital computing have long been boiled down fairly concisely: A digital computer must represent any given value as either a 1 or a 0, and a quantum computer does not. It may express that value as either 1 or 0 or a simultaneous superposition of the two.

However, recent atomic research from the University of New South Wales (UNSW), in which researchers built a single-atom transistor, and IBM Research, which scaled an antiferromagnetic storage array 100 times denser than current state-of-the-art technology, demonstrates the boundary between the two is narrowing to the extent that even describing how the principles of one may affect the other, and how quickly scientific discovery may become more viable applied research, are taking on a decidedly amorphous or even “quantum” nature.

For instance, Gabriel Lansbergen, a principal engineer at Taiwan Semiconductor Manufacturing Co., in commenting about the UNSW project in Nature Nanotechnology, noted “the electrical characteristics of the resulting single-atom transistor are, in a way, similar to those of a conventional transistor in that the current between the source and drain can be controlled by applying a voltage to the gate electrodes. However, when the gate voltage is just below the threshold for transistor operation, the single-atom transistor behaves as a small quantum dot.”

The IBM research also straddles the quantum-classical line. Andreas Heinrich, lead investigator of atomic storage at IBM Research Almaden, says their work addresses the transition from quantum mechanics to classical physics. “An antiferromagnet the way we present it here—which I call the classical physics—and the quantum mechanical antiferromagnet behave quite differently,” Heinrich says. “Quantum antiferromagnets align their spins such that in total there is no magnetic moment left, the two electrons together are nonmagnetic, they lose their spin.”

No spin, of course, means no readable data. So, Heinrich says, the research explores that boundary. “‘How many atoms does it take that you can ignore quantum mechanics?’ in a nutshell is what we’re describing.”

But IBM Fellow Don Eigler, who pioneered atom manipulation in 1989 and was also a member of the IBM research team on the latest project, says it may be difficult to translate their results into a wider search for scalable atomic componentry.

“Germans have the perfect way of expressing that,” says Eigler, who recently retired from IBM. “They say ‘jein,’ a fluid combination of ‘ja’ and ‘nein.’ You can’t say that in English because you can’t say ‘yesandno.’ Jein is perfect. We’re slowly kind of getting there. The great problem is, despite our body of knowledge in the laboratory, we don’t know of, at least to my knowledge, a way to economically manufacture anything where we have the kind of control we need over where atoms go in relation to one another.”

Cold, Hard Facts

Essentially, the projects—and a related atomic-level wire project also recently unveiled at UNSW—demonstrate the first times researchers have been able to manipulate and stabilize atomic-size structures in such a way that they can persistently maintain the qualities needed to be classified as viable, though still strictly experimental, computational devices. The UNSW researchers placed a single phosphorus atom channeled in a silicon substrate within one atomic lattice site, ± 3.8 Angstroms, within the parameters necessary to create a stable platform for building a working circuit at that level. The IBM researchers discovered that an iron-on copper nitride-on copper structure resulted in a naturally antiferromagnetic structure that shrinks the space needed for storing a bit of information on 12 iron atoms to about nine square nanometers, which is about 100 times denser than the current state-of-the-art storage technology.

The IBM researchers calculate it will take about 150 atoms to create a stable storage mechanism at room temperature.

In practical terms, the experiments provide both a bridge to the conventional thinking around CMOS technology and a simultaneous caution for speculative thinking about the likelihood of achieving viable quantum computing technologies. Perhaps the most obvious example of the duality of the projects lays in the extremely cold temperatures necessary to stabilize the atoms involved. The UNSW transistor operated at the temperature of liquid helium (lower than −270 Celsius or −452 Fahrenheit), while the IBM storage array was stable (with less than one spontaneous switching event per 17 hours) at temperatures of about .5 kelvins (−272.3 Celsius). Another example of the difficulty of extrapolating laboratory advances into a larger context was the tool both groups used to construct their examples, the scanning tunneling microscope, is one for which no industrial scaling has yet been perfected.

However, the IBM researchers made two observations in the course of their work that may hold portent for future research. The first was that, at a specific number of atoms aligned antiferromagnetically, the atoms’ tendency to exhibit quantum superpositions diminished greatly. The second was that they could calculate the number of atoms it would likely take to create a stable storage mechanism at room temperature—about 150.

The fundamental property the antiferromagnets bring to scaling storage down so small is that their defining property—in which the atomic spin of each atom is opposite to instead of aligned with its next-door neighbor—can eliminate the problem inherent with trying to pack traditional ferromagnetic bits, whose magnetic spins are all aligned, closer together. The problem is that, as the magnetically aligned atoms get closer together, the magnetic field of each is more likely to affect or be affected by its neighbor. Interestingly enough, Heinrich says the antiferromagnetic nature of the structure was serendipitous.

“We didn’t set out saying ‘Let’s use antiferromagnets because that would be so cool,'” he says. “It was more like, as quite often happens in this kind of exploratory science, you try things out and go with whatever nature wants you to do with it. At least, in principle, if you have this kind of control that we have here, you can pack magnetic bits much more densely together without them interfering with each other. The antiferromagnetism allows you to cancel out the interaction between the bits. It wasn’t designed from the beginning that way, but that’s in the end what is probably the most interesting technological application of this.”

Silicon Plays Nice

While the IBM project demonstrated a realistic possibility of much denser storage capabilities at room temperature, the UNSW project offered proof of concept of not only the ability to control the placement of a single-atom transistor with scientifically significant accuracy, but also the ability to use industry-standard silicon, instead of a less universally applicable or economically viable material, as the device’s substrate. However, the UNSW researchers also discovered they could not employ atom relocation through STM manipulation alone. They needed to find a way to place atoms upon the silicon and then etch away all but the single atom they needed to stabilize.

The Wales and IBM experiments provide a bridge to the conventional thinking around CMOS technology and a simultaneous caution for speculative thinking about the likelihood of achieving viable quantum computing technologies.

Michelle Simmons, the project’s leader and director of UNSW’s ARC Center for Quantum Computation and Communication, says the idea of just moving atoms as Eigler did in his pioneering STM experiment, which employed xenon on nickel, would be impossible on silicon.

“You cannot just pick up atoms, like IBM did with its original demonstration, because silicon is too strongly bound,” Simmons says. “So you basically destroy the substrate if you try to do the same kind of experiment.”

The UNSW team used a combination of hydrogen-resist lithography and STM microscopy to create the stable phosphorus atom that served as the transistor, employing phosphine as the dopant precursor. The phosphine molecules, placed within three adjacent pairs of surface silicon atoms, were successively dosed at room temperature and at 350 Celsius. This resulted in progressive breaking up of the component phosphorus and hydrogen atoms, ultimately resulting in the single phosphorus atom being incorporated into the silicon, substituting for one of the original six silicon atoms. Nanowires were then attached via silicon guides, and the entire structure was covered with a 180-nanometer layer of silicon.

Eigler says observers should bear in mind not only that silicon was the substrate for the UNSW work, but also the structure of the device itself.

“The other point is that it’s in silicon as opposed to on,” says Eigler. “That’s one of the wonderful things about what Michelle has learned to do, is how to build things with atomic-scale precision and then bury them.”

That capability might well lead to the development of three-dimensional, layered atomic-scale structures, Eigler says. It also allows developers to completely control the environment those atomic elements inhabit by blocking contaminants and also restricting the atoms’ movement on the surface.

And, as cautious as Eigler is about the scalability of the discoveries, he also thinks he and his colleagues crossed a critical barrier.

“I saw the beginnings of the transition from science to engineering,” he says. “If you asked anybody else in the lab, I don’t think they paid much attention to it. They didn’t want to emphasize it in the paper we wrote. But I saw that and thought, ‘OK, there’s a little bit of a demarcation there, a little bit of a landmark. You’re beginning to cross that line, right here, right now.'”

Further Reading

Fuechsle, M., et al.
A single-atom transistor, Nature Nanotechnology 7, 4, April 2012.

IBM Research Almaden
IBM researchers store one bit of magnetic information in just 12 atoms, http://www.youtube.com/watch?v=hpKMShooDBo&feature=player_embedded.

Loth, S., Baumann, S., Lutz, C.P., Eigler, D.M., and Heinrich, A.J.
Bistability in atomic-scale antiferromagnets, Science 335, 6065, Jan. 13, 2012.

Morton, J. J. L., McCamey, D. R., Eriksson, M. A., and Lyon, S. A.
Embracing the quantum limit in silicon computing, Nature 479, 7373, Nov. 17, 2011.

Weber, B., et al.
Ohm’s law survives to the atomic scale, Science 335, 6064, Jan. 6, 2012.

Figures

Figure. New research involving antiferromagnets by Andreas Heinrich and his IBM colleagues is exploring the transition from quantum mechanics to classical physics.