Exploring Materials for Faster Computers

Note de la rédaction : L’article suivant s’appuie sur une traduction accélérée conforme à la norme ISO 18587, qui rend le sens, mais peut comporter des imperfections. L’article original est reproduit dans la version anglaise de cette page Web.

Australian scientists accessed the CNBC in search of the right materials for the next generation of computer processors.

Image: Intel

The 1980s and 1990s saw amazing transformations in personal computers. However, this evolution has slowed considerably in recent years. And while devices are smaller and more useful than ever, the processing power of the average personal computer hasn’t changed all that much.

“Silicon-based computer processors are now a mature technology. In order to make further breakthroughs in processing power, new materials must be identified.”

A number of economic and scientific factors are limiting advancements in silicon-based computer processors. For instance, a lack of improvement in regulating the heat generated in processor chips has limited chip speed since 2004. Furthermore, the continual drive towards smaller transistors is becoming increasingly challenging and cost-prohibitive.

While the cost of creating ever-smaller components might one day be overcome with technological developments, a fundamental barrier to further shrinking will soon be reached. Indeed, today’s leading computer processors already include unbelievably tiny components, some of which measure only about 50 atoms (i.e., 14 nanometres) across. When components reach about 10 atoms across, however, their electrons begin to act differently in accordance with the laws of quantum mechanics. Thus, even if such ultra-tiny components were to be developed, the processors would no longer function.

While some improvements in silicon-based processors can still be expected, major breakthroughs will require totally new technologies based on different materials. That is why scientists from the Australian Nuclear Science and Technology Organisation (ANSTO) and Macquarie University in Australia are participating in the search for new materials that have the potential to solve the heat problem while packing even more computing power. For these Australian researchers, that means finding materials that are conducive to spin electronics.

“Spin electronics—or ‘spintronics’—is a way to do more in the same tiny space because spintronic devices encode information using the electron spin as well as its charge,” says ANSTO scientist Dr. Frank Klose. As he further explains, “Electron spins exist as one of two states, either ‘up’ or ‘down,’ similar to transistors, which have two states, either ‘on’ or ‘off’.”

An electron’s spin creates a tiny magnetic field with a preferred direction, either up or down—just like a kitchen magnet, which has to be facing the right way before it will stick to the fridge. (Illustration: IBM)

“Early spintronic devices for specialized data storage applications are already on the market,” adds Dr. David Cortie, a post-doctoral researcher at ANSTO. “And scientists have been searching through thousands of materials for the right properties for spintronics in a processor chip.”

For processor chips, the ‘right properties’ include both silicon-like semiconducting properties as well as strong magnetism for storing the electron spin state. Identifying the ideal material for this combination, however, has proven challenging.

The ANSTO group has focused on a family of materials called ‘rare-earth nitrides.’ Although rare-earth compounds make some of the best magnets, they are difficult to study because they are not stable in air or in bulk form. Thus, the researchers had to not only create an extremely thin film of rare-earth nitrides (similar to the form that could be used in tiny circuits), but also find a way to protect this film from the air.

For this, they turned to their collaborators at Macquarie University, who specialize in synthesizing these exotic nitrides. The Macquarie group, led by James Downes, developed a method for creating thin films of these materials on an aluminum substrate while protecting them from air with a ‘capping’ layer of zirconia.

The ANSTO team then accessed the Canadian Neutron Beam Centre (CNBC) to study the film using a method called ‘polarized neutron reflectometry.’ This incredibly precise technique allowed the researchers to obtain depth-sensitive chemical and magnetic data with nanometre precision, all while holding the rare-earth nitride samples at requisite low temperatures and at very high magnetic fields of up to three tesla—a capability not available on reflectometers at other neutron beam facilities.

Such high magnetic fields were required to drive the system to saturation (i.e., to make all the tiny magnets inside the film point in the same direction). Without this perfect alignment, the true strength of the tiny magnets could not have been measured precisely. “Being able to take the measurements under these extreme conditions was critical to the scientific results,” says Dr. Helmut Fritzsche, the CNBC scientist who collaborated with the ANSTO team to ensure the experiment’s success.

These experiments led the team to three findings. First, they verified that the thin films were homogenous, indicating that the capping layer was working effectively to preserve the integrity of the rare-earth nitride films.

Second, they verified that the thin films were intrinsic magnets, implying that they would retain their magnetic properties on the smallest of scales—even down to a few atoms. This is an advantage over materials that have to be doped (i.e., adding elements to alter a substance’s properties), because doped materials rely on effects produced by averaging over many atoms, thereby limiting the ability to shrink them.

Tomorrow’s computer processors could require ultra-cold storage facilities. (Image: Digital Storm)

Third, and perhaps most interestingly, they found that the magnetic strength of the rare-earth nitride films was two–thirds lower than predicted by theoretical calculations. This finding, published in 2014, has sparked other researchers to search for explanations, with some ideas already being published. “This discrepancy shows that these materials hold more secrets than we expected,” says Cortie, who at the time of the experiments was working with the ANSTO team as a graduate student at Australia’s University of Wollongong. “And even with only about 30 percent of the expected magnetization, they could still be very useful.”

These materials do have one notable limitation: they require extremely cold temperatures. The ‘warmest’ of the three rare-earth nitrides in this study was dysprosium nitride (DyN), which had to be cooled to −230 °C to achieve the key magnetic property necessary for spintronics.

The frigid temperatures have not deterred these intrepid explorers, however. “At this stage, we’re doing fundamental science,” explains Klose. “These materials have to be studied in their most fundamental form before you can make sense of any more complicated forms.”

Although rare-earth compounds have only just begun to be explored for computer applications, there have already been examples of raising the critical temperatures by over 100 °C by alloying.

“Someday, we might achieve the desired properties at room temperature so that we can make a more powerful laptop computer using these materials,” says Cortie. “But even if we don’t, you could imagine a future where your laptop connects seamlessly to supercomputing power centrally located in ultra-cold facilities, like how we use cloud-computing today.”

DOI: 10.1103/PhysRevB.89.064424

This research story was republished with the permission of the Canadian Institute for Neutron Scattering.

Tagged in:
Featured Institutions:

Subscribe to Our Newsletter

If you are interested in keeping up with news in the field of Neutron Beams, fill out the form below and join our mailing list.

"*" indicates required fields

Name*
Interests
This field is for validation purposes and should be left unchanged.