CIRCULARLY POLARIZED X RAYS IMAGE THE MAGNETIC DOMAINS IN THE FERROMAGNETIC LAYER, WHICH IS IDENTIFIED BY THE EMISSION OF ELECTRONS FROM COBALT.  LINEARLY POLARIZED X RAYS IMAGE MAGNETIC DOMAINS IN THE ANTIFERROMAGNETIC LAYER, IDENTIFIED BY THE EMISSION OF ELECTRONS FROM IRON.  AT THE INTERFACE OF THE TWO KINDS OF MATERIALS, THE DOMAINS ARE PRECISELY ALIGNED.

At the Advanced Light Source (ALS), located at the Department of Energy's Lawrence Berkeley National Laboratory, a consortium of researchers from Berkeley Lab, Stanford University, IBM Corporation, the University of Neuchâtel in Switzerland, and Arizona State University obtained images of tiny magnetic domains in cobalt, a ferromagnet, coupled to domains with the same orientation in an adjacent layer of lanthanum iron oxide, an antiferromagnet. The phenomenon is known as exchange bias, or pinning.

"The exchange bias phenomenon has been known for more than 45 years, but even though we have used the effect and even built sophisticated devices by trial and error, we haven't understood how it works," says Frithjof Nolting, an ALS researcher visiting from Stanford University.

In ferromagnets, the electronic spins in magnetic domains are aligned parallel to one another. In antiferromagnets they alternately point in opposite directions, which renders antiferromagnets insensitive to applied magnetic fields.

When these materials are layered to create computer read-heads or other magnetization sensors, one ferromagnetic layer is pinned by an adjacent antiferromagnetic layer -- and thus acts as a magnetic reference -- while another is free to change its orientation in response to an applied field.

When the spins of the ferromagnetic layers are opposed, resistance to an electric current is greater than if they are parallel; this is the so-called "giant magnetoresistance" effect. By varying the orientation of domains in one of the ferromagnetic layers and thus varying the resistance, information can be written, stored, and recovered.

For a better understanding of pinning, Nolting says, two things were needed: "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.

PEEM2, the photoemission electron microscope at ALS beamline 7.3.1.1, meets these requirements. When an x-ray beam is incident upon a sample, electrons ejected from the sample can be used 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; for example, near 710 eV (710 electron volts) iron emits copious photoelectrons, while cobalt does the same near 780 eV. By tuning the energy of the beam, layers containing different elements can be distinguished.

If the x-ray beam is polarized, it can reveal magnetic domains: linear polarization produces high-contrast images of antiferromagnetic domains; circular polarization yields high-contrast images of ferromagnetic domains.

In the sample used by the researchers, a cobalt film less than three nanometers thick was deposited on a film of 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 resulting perfectly registered images show precise correspondence between the spin orientation of microscopic domains in the lanthanum iron oxide layer and the domains of the ferromagnetic layer immediately adjacent to them, demonstrating that magnetic exchange coupling aligns the magnetic structure of both layers, domain by domain.

When they measured the strength of the coupling by applying a magnetic field, the researchers discovered an unexpected phenomenon.

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

REFERENCE FERROMAGNETIC LAYER (BLUE) IS PINNED BY EXCHANGE BIASING TO ANTIFERROMAGNETIC LAYER (GREEN)

"The usual method is to set a bias by annealing the multilayer in a magnetic field," Nolting explains. "But we imaged samples just as they were grown, without any additional processing. And we found that prior to any setting procedure, there is already a bias locally, within each individual domain" -- although because of their random orientation, the effect vanishes over large areas.

"Apparently exchange bias is an intrinsic property of the interface, caused by the common alignment of the magnetic structure of both materials," Nolting adds, "even though initially there may be no total bias, averaged over a large area."

By using photoemission electron microscopy at the ALS, the researchers were able to make the first direct observation of the link between antiferromagnetic and ferromagnetic spin at the interface between the two kinds of materials. Their measurements imply that the setting procedure reorients domains, shifting the balance from a random bias in opposing directions to a preferred bias direction. But it does not create that bias.

"This opens the door to new investigations, which may affect the way devices based on the exchange bias effect are manufactured and the materials that can be used in them," says Nolting.

Frithjof Nolting and his colleagues report their findings in a letter to Nature, "Direct observation of the alignment of ferromagnetic spins by antiferromagnetic spins," June 15, 2000.

Other authors of the letter are Andreas Scholl of the ALS; 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; Simone Anders of the ALS; 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.

The Berkeley Lab is a U.S. Department of Energy national laboratory located in Berkeley, California. It conducts unclassified scientific research and is managed by the University of California.

Additional Information:

• Microscope produces first images of the domain structure of an
  antiferromagnetic thin film -- a material vital to the read-heads of computer hard-disk drives