Shining Soft X-rays on Magnetic Structures

Ulrich Hillebrecht*

The study of magnetic materials with x-rays is developing at a breathtaking pace. The methods used in these studies extend magneto-optics--the response of a material to optical excitation as a function of the relative orientation of magnetization and light polarization--to the x-ray regime. One of the key advantages of these techniques is that they are element specific, because the photon wavelength (or energy) in the x-ray regime can be tuned to match the excitation energy of a discrete core electronic level. Also, layer thicknesses of magnetic multilayers and lateral structures in novel magnetic devices that are presently being developed are comparable to soft x-ray wavelengths. Therefore, detailed insight into the magnetic properties of these technologically important magnetic structures can be obtained by use of soft x-rays. The soft x-ray regime is of importance because all technically relevant magnetic materials contain 3d transition metals, for which the 2p excitation that shows the highest sensitivity to the magnetic state lies between 500 and 1000 eV.

 Synchrotron radiation is a particularly powerful tool for soft x-ray magneto-optics, offering tunability to virtually any core level resonance, high intensity and brilliance, good collimation, and complete control over polarization. Recently developed polarizers and wave retarders allow us to analyze and modify the state of polarization for soft x-rays (1), further increasing the scope of these magneto-optic studies. On page 2166 of this issue, Dürr et al. (2) demonstrate the power of x-ray magneto-optics in a synchroton study of single crystalline FePd layers, which provides detailed insights into their complex magnetic domain structure.

 In a classical picture, magneto-optic effects occur when the magnetic field associated with a magnetic material acts on the electron currents induced by an incident electromagnetic wave. A charge moving in a magnetic field experiences the so-called Lorentz force, the direction of which is normal to the velocity of the charge and the magnetic field (see top figure on page 2100). The Lorentz force leads to light with polarization perpendicular to the incident light in the reflected beam. Microscopic theories (3) show that this magneto-optic effect is connected to excitations from the core shell to spin-polarized unoccupied electronic states and the interaction of magnetic moments due to the spin of the electrons and their orbital motion.


Phenomenological model of the magneto-optic response. An incident linearly polarized wave generates an oscillating current I in the plane of incidence with components iz and ix along the x and z directions. For magnetization M along the x direction, the current along z experiences a Lorentz force along y. This gives rise to polarization perpendicular to the incident beam; the normal reflected beam was omitted for clarity. In the soft x-ray regime, this effect can be detected as a change of the polarization on reflection or a change in the magnitude of the reflected intensity.

The change of polarization when light is passed through a magnetic material, known as the Faraday effect, is in principle the most direct way to measure the optical constants responsible for magneto-optics (1). The late Theo Thole, Paolo Carra, and co-workers at the European Synchrotron Radiation Facility in Grenoble, France, showed how one can determine local magnetic moments, distinguishing between different constituents and spin and orbital parts, from magneto-optic spectroscopy in the core level regime (4). Also, the study of magnetic properties by scattering experiments, both with hard (5) and soft (2, 6, 7) x-rays is increasing rapidly. In a scattering experiment, one measures the number of outgoing photons as a function of direction, or transferred wave vector, thus probing the spatial arrangement of scattering centers. The wavelength of the photons to be scattered then has to be comparable to the distances between atoms. Another property of central importance for scattering experiments is the coherence of the incident light: For interference to take place between waves emerging from different atoms, the light incident on these atoms has to have a fixed phase relationship.

 Research on magnetic materials is currently focused on layered samples of increasing complexity, containing magnetic and nonmagnetic layers configured so as to provide the best performance for a particular application. Surface and interface roughness play an important role in the magnetic properties. Freeland et al. (7) showed in diffusely scattered light measurements that the magnetic moment distribution appears to be smoother than the chemical interface. A multilayer reflects and transmits light at every interface, and the interference between these waves gives rise to a periodic modulation of the overall reflected and transmitted intensity. This allows one to determine geometry and interface properties of the multilayer. For regular assemblies of magnetic units--multilayer films (6) or regular domain patterns (2)--the magneto-optic signal shows similar modulations, which reveal information about the magnetic structure.

 Magneto-optical scattering was used recently in a beautiful experiment on a magnetic sandwich consisting of two Co layers separated by a Cr layer a few nanometers in thickness (6). The Cr layer mediates an exchange coupling between the two ferromagnetic Co layers, such that their magnetizations are not completely independent. Monitoring the reflected intensity at the Co 2p excitation energy as a function of applied field yields the ferromagnetic hysteresis loop. Using visible light, one would just see the overall magnetization, because the wavelength and probing depth would be large compared with the film thickness. In contrast, measurements with x-rays with wavelengths comparable to the film thickness reveal modulations of the reflected intensity as a function of incidence angle, typical for layered structures. In addition, the hysteresis loops change qualitatively with incidence angle. In principle, the loops should show inversion symmetry, but as can be seen in the figure to the right, for certain incidence angles, this is not the case. The asymmetric shape results from the interference of the magneto-optic response from the two layers, which are not equivalent because of the differing sequence of interfaces: The bottom film is sandwiched between a semiconductor substrate and Cr, whereas the top film is between Cr and an Al cap layer. Taking this into account, one obtains symmetric loops for each layer. However, these loops are not identical (see figure at right), contrary to naïve expectation for films of the same material, thickness, and crystal structure. This shows that anisotropies and magnetic switching are influenced by the layer sequence and provides information about the coupling strength and magnetic correlation function between the two layers, all of which are important aspects for the development of novel magnetic devices based on interlayer exchange coupling.

 The report by Dürr et al. (2) in this issue uses a similar approach to characterize the lateral magnetic structure in a chemically homogenous thin film. In a thin film with perpendicular magnetic anisotropy, the shape of the film may cause a competing shape anisotropy, leading to a domain pattern with some regularity. The typical length scale for this domain structure is on the order of a few tens of nanometers, and the resulting best suited wavelength for scattering is close to the 2p excitation threshold of Fe, where a large magneto-optic response can be exploited. Magnetic scattering occurs because scatterers are magnetically inequivalent, as in an antiferromagnet. Dürr et al. transfer this idea to oppositely magnetized domains, which by way of self-organization form a regular array. The experiment shows the presence and size of closure domains that are not visible by magnetic force microscopy or other methods. These closure domains stabilize the magnetic structure and control the speed of magnetization reversal, which is a key issue in device applications.

An asymmetric response. Magnetic hysteresis loops measured for a Co/Cr/Co trilayer at different incidence angles q [adapted from (6)], with circularly polarized radiation tuned to the Co 2p excitation. A trilayer consisted of 5 nm of Co, 3.5 nm of Cr, and 5 nm of Co, grown on a semiconductor substrate. The reflected intensity was measured as a function of magnetic field applied along intersection of sample surface and plane of incidence, which coincides with the axis of easy magnetization. For some incidence angles, symmetric hysteresis loops are observed (top left), whereas extremely asymmetric loops appear at other angles (top right). From a collection of such curves, the hysteresis loops for the top and bottom Co films can be deduced separately.

The importance and the scientific attraction of the type of investigation described here are reflected in the broad activities at numerous synchrotron radiation sources. Novel magnetic devices such as magnetoresistive and spin-polarized tunneling read heads or nonvolatile memories depend on controlling the properties of ever smaller magnetic structures, to which the new analytical methods that use soft x-rays will provide a substantial input.

References and Notes

  1. J. B. Kortright et al., J. Magn. Magn. Mater. 191, 79 (1999).
  2. H. A. Dürr et al., Science 284, 2166 (1999).
  3. H. Ebert, Rep. Progr. Phys. 59, 1665 (1996) [ADS].
  4. B. T. Thole et al., Phys. Rev. Lett. 68, 1943 (1992) [APS]; P. Carra et al., ibid. 70, 694 (1993) [APS]; C. T. Chen et al., ibid. 75, 152 (1995) [APS]. Because it is easier to generate circularly polarized light at a synchrotron than to measure the change of light polarization, the equivalent way to obtain this information by measuring the absorption of circularly polarized light is used more widely.
  5. D. Gibbs et al., Phys. Rev. Lett. 61, 1241 (1988) [APS].
  6. Y. U. Idzerda, V. Chakarian, J. W. Freeland, ibid. 82, 1562 (1999) [APS].
  7. J. W. Freeland et al., J. Appl. Phys. 83, 6290 (1999).
  8. The author is grateful to N. Weber for useful remarks and discussions and to Bundesministerium für Bildung und Forschung for support under grant 05 621 PFA 7.

The author is at the Institut für Angewandte Physik, Heinrich-Heine-Universität Düsseldorf, 40225 Düsseldorf, Germany. E-mail:

Related articles in Science:

Chiral Magnetic Domain Structures in Ultrathin FePd Films.
H. A. Dürr, E. Dudzik, S. S. Dhesi, J. B. Goedkoop, G. van der Laan, M. Belakhovsky, C. Mocuta, A. Marty, and Y. Samson

Science 1999 284: 2166-2168. (in Reports)
Volume 284, Number 5423 Issue of 25 Jun 1999, pp. 2099 - 2101 
©1999 by The American Association for the Advancement of Science.