Drawn to the sound: Supercomputers reveal phonon magnetism

Joseph Heremans, Ohio eminent scholar in nanotechnology, holds an artist’s rendering of a phonon heating solid material. Image courtesy Renee Ripley, Kevin Fitzsimons, and Ohio State University.

Did you know that sound waves have magnetic properties?

Ohio State University researchers recently demonstrated this fact and performed calculations at the Ohio Supercomputing Center (OSC) identifying the physical origins of the effect. 

This discovery is the product of a collaboration between Joseph Heremans, Ohio eminent scholar in nanotechnology and professor of mechanical engineering, Wolfgang Windl, professor of materials science and engineering, and Roberto Myers, associate professor of materials science and engineering and electrical and computer engineering.

What they’ve found has the potential to rewrite our science textbooks.

Physicists have long known the molecular principles behind sound: Essentially, it is the vibration of atoms carried through a medium. Heat is the storage of energy in those vibrations. Scientists conceive of sound and heat much like light — as both a particle and a wave. Light particles are called photons; sound and heat particles are known as phonons.

Building on these fundamental scientific facts, Heremans' team is the first to see sound as something subject to magnetic forces. 

To demonstrate this, Heremans created a lopsided tuning fork made from a crystal of the semiconductor indium antimonide, one branch 4mm wide and the other 1mm wide. He chilled the fork to -450 F (-268 C), and then heated the two parts, sending phonons through each branch of the fork independently.

The phonons in the larger portion traveled faster and bumped into the walls of the crystal less frequently than the phonons in the smaller portion.

Heremans then turned on a 7 Tesla magnet, and discovered that the phonon speed in the larger portion of the fork slowed when a magnetic field was applied, due to phonon-phonon collisions, while their collisions with the wall remained the same. The phonon-phonon collisions are thus sensitive to the magnetic field.

The tiny tuning fork used in the experiment. Researchers were able to slow the movement of heat through the wider arm of the fork using a magnetic field. Image courtesy Kevin Fitzsimons and Ohio State University.

To quantify and model this effect, Heremans approached Windl, Nikolas Antolin, and Oscar Restrepo from the Department of Materials Science and Engineering at Ohio State University.

They determined the phonons slowed due to an interaction between the externally applied magnetic field and the intrinsic magnetism in the electrons within the crystal tuning fork. This is diamagnetism, an effect occurring even in traditionally non-magnetic substances.

Although diamagnetism is usually the smallest magnetic interaction and should be much too weak to have a large effect on phonon interactions, Heremans' team found the magnet slowed the phonons by 12% — it took intensive theoretical analysis and computation to explain the measurements.

To study the effect of magnetic fields on phonons on the OSC Oakley cluster, Windl’s group used a new method based on density functional theory, a quantum mechanical modeling strategy that identifies electron distribution to determine the vibrational properties and forces between atoms.

Determining the electronic structure tends to be one of the most computationally intensive maneuvers scientists undertake and requires supercomputing facilities. In fact, it consumed a large portion of the 1.5 million computer hours the project required on the Oakley cluster. “Completing this research would not be possible at all without significant parallel computing resources,” Windl says.

With the help of the engineers at the OSC, Windl’s group switched to the high-throughput Lustre file system for the read/write speeds needed to handle the size of the datasets generated.

Windl says the diamagnetic interaction was surprising to him. “If you think about typical strengths, magnetic forces are about 1,000 times weaker than phonon interactions, so how can a David-like magnetic force cause these huge changes in the Goliath-size thermal conductivity?”

What Windl’s computations illuminate is that Heremans chose the temperature in his experiments perfectly, allowing the diamagnetic forces to have maximum impact.

Though he cautioned against expecting rapid application in the near future, Heremans suspects the ability to control heat and sound waves magnetically could have an impact on energy production in the future.

“Heat flow is crucial to many industries, power generation, engine cooling, cooling of electronics.  Developing a solid-state heat switch could make solid-state heat engines more efficient, particularly those based on magnetocaloric effects.”

But more relevant than far-off applications, Heremans reasons, is the doorway opened by the Oakely cluster.

“The most important part, for a supercomputer center, is that the effect can be modeled,” says Heremans. “There is no need to use the Edisonian approach to find materials where this can be an important and potentially useful effect. Now that we have the concept, you can calculate the materials.”