Not as catchy as The Magic Flute, but a lot more practical: antiferromagnetism is a phenomenon vital to the layered structures of today's advanced computer hard-disk read heads and to the memory devices of the future.

THE FRONT LAYER IN THIS READ HEAD (GREEN) IS A FERROMAGNET WHOSE MAGNETIC DOMAINS ARE FREE TO RESPOND TO THE EXTERNAL FIELD OF A BIT PASSING BENEATH ON THE HARD DISK.  IN THE SECOND FERROMAGNETIC LAYER, DOMAINS ARE PINNED BY THE ANTIFERROMAGNET IN BACK. WHEN THE DOMAINS IN THE FERROMAGNETIC LAYERS ARE ORIENTED DIFFERENTLY, ELECTRICAL RESISTANCE INCREASES GREATLY. Image courtesy of IBM Corp.

"A modern read head uses layers of very thin films with different magnetic properties," explains Andreas Scholl of the ALS. "As the head passes over the hard disk, these layers sense the orientation of the domains on the disk and cause the head's electrical resistance to change in response."

At normal temperatures, the electronic spins in the magnetic domains of ferromagnetic materials such as iron or cobalt are parallel and point in the same direction; the domains change their orientation in the presence of an applied magnetic field. A read head takes advantage of the fact that if two ferromagnetic layers share the same orientation, they exhibit less electrical resistance than when they are opposed.

While spins in the magnetic domains of antiferromagnetic materials are also parallel, they alternately point in opposite directions, so on average the material is insensitive to applied magnetic fields. This comes in handy for read heads: in order for one of the ferromagnetic layers in a read head to switch independently of the other, one of them must be "pinned" by an underlying antiferromagnetic layer.

The more formal term for pinning is "exchange bias" -- a phenomenon known for more than 45 years. "Even though we have used the effect and even built devices by trial and error, we haven't understood how it works," says Frithjof Nolting, an ALS researcher visiting from Stanford University.

Two things were needed for a better understanding, Nolting says: "first, a method of imaging the configuration of domains in antiferromagnetic thin films, which requires a resolution better than 100 nanometers" (100 billionths of a meter) "and second, a way to image the interface between ferromagnetic and antiferromagnetic domains in adjacent layers," which requires distinguishing between layers containing different chemical elements.

"The only method that can do all this is photo-electron emission microscopy, or PEEM," says Simone Anders, leader of the team that built the PEEM2 microscope on ALS beamline 7.3.1.1. When an x-ray beam is incident upon a sample, PEEM2 uses electrons ejected from the sample to form an image with ten-thousand-fold magnification and a resolution of 20 nanometers.

X-rays of different energies stimulate photoelectrons characteristic of different elements; thus by tuning the energy of the beam, layers containing different elements can be distinguished. And if the beam is polarized, it can reveal magnetic domains: linear polarization yields images of antiferromagnetic domains, while circular polarization reveals ferromagnetic domains.

IMAGES MADE BY THE PEEM2 MICROSCOPE SHOW THAT AT THE INTERFACE OF A THIN LAYER OF COBALT, A FERROMAGNET, AND LANTHANUM IRON OXIDE, AN ANTIFERROMAGNET, MAGNETIC DOMAINS ARE PRECISELY ALIGNED.

With PEEM2 in place at the ALS, the curtain was set to rise. Act 1 appeared in Science magazine this February, when the researchers reported the first images that clearly revealed the alignment of domains in an antiferromagnetic thin film. When these PEEM2 images, each only a few hundred nanometers in area, were compared to transmission electron microscope images of the same sample, the magnetic domains corresponded exactly to the orientation of its crystals.

Act 2 was a letter to Nature in June, when the researchers announced another first: direct images of the alignment of magnetic domains on both sides of an interface between ferromagnetic and antiferromagnetic layers.

In this work the sample was a ferromagnetic cobalt film, less than three nanometers thick, deposited on a film of antiferromagnetic lanthanum iron oxide. "By tuning the photon energy of the beam, we were able to record separate images of the antiferromagnetic and ferromagnetic layers in exactly the same place," Nolting says.

The perfectly registered images show precise correspondence between the spin orientation of microscopic domains in the lanthanum iron oxide layer and the domains of the cobalt layer immediately adjacent to them, demonstrating that exchange coupling aligns the magnetic structure of both layers, domain by domain.

But the end of Act 2 held a surprise: when the researchers measured the strength of the coupling by applying a magnetic field, they discovered an unexpected phenomenon.

Because exchange-bias devices such as read heads depend upon an overall preferred magnetic orientation in a ferromagnetic layer coupled to an antiferromagnet layer, a "bias" is set during the manufacturing process.

"The usual method is to set a bias by annealing the multilayer in a magnetic field," Nolting explains, which takes advantage of the fact that magnetic materials lose their magnetism above a critical temperature, then regain it as they cool. However, Nolting says, "we imaged samples just as they were grown, without any additional processing."

The surpise: "We found that prior to any setting procedure, there is already a bias locally, within each individual domain. Apparently exchange bias is an intrinsic property of the interface, caused by the common alignment of the magnetic structure of both materials, even though initially there may be no total bias, averaged over a large area."

Says Nolting, "This opens the door to new investigations, which may affect the way devices based on the exchange bias effect are manufactured" -- as well as the materials chosen to make them.

Thus the latest plot twist suggests that many more surprises are in store, in the ongoing drama of antiferromagnetism's secrets.

In addition to Andreas Scholl, Frithjof Nolting and Simone Anders, other members of the research team are Joachim Stöhr of Stanford University, formerly of the IBM Almaden Research Center in San Jose, who led the project; Jin Won Seo of the University of Neuchâtel and IBM's Zürich Research Laboratory; Jean Fompeyrine, Heinz Siegwart, and Jean-Pierre Locquet of IBM's Zürich Research Laboratory; Jan Lüning, now with the Stanford Synchrotron Radiation Laboratory, Eric. E. Fullerton, and Michael F. Toney of IBM's Almaden Research Center; Michael R. Scheinfeld of Arizona State University; and Howard A. Padmore of the ALS. PEEM2 was built under a corporate research and development agreement (CRADA) between IBM and Berkeley Lab, in collaboration with Arizona State University.

Additional information:

• IBM-produced animation explaining the function of multilayered magnetic devices in read heads
• Microscope produces first images of the domain structure of an antiferromagnetic thin film