Introduction by A. Rogalev (ESRF)

X-ray Magnetic Circular Dichroism (XMCD) is considered to be one of the most important discoveries in the field of magnetism in the last two decades. In an XMCD experiment, the quantity studied is the difference in the X-ray absorption spectra recorded with left and right circularly-polarised photons while the sample magnetisation is kept parallel or antiparallel to the direction of the propagation of the incident X-ray beam. The major strength of the XMCD is the capability to deduce from the experimental spectra the orbital-projected magnetic moments both in magnitude and direction with the full benefit of the element selectivity inherent to X-ray absorption spectroscopy.

The physical origin of XMCD can be most easily explained with the so-called two-step model. The first step describes the excitation of a core electron by a circularly- polarised X-ray photon that carries an angular momentum (+ for a right-handed photon and - for a left-handed photon), the corresponding helicity vector being parallel (right) or antiparallel (left) to the propagation direction. As a consequence of the conservation of angular momentum in the absorption process, the photon's angular momentum is entirely transferred to the photoelectron. Let us consider the photoelectron being excited from a spin-orbit-split core level (e.g., LII,III absorption edges): then part of the angular momentum carried by the photon will be converted into spin via spin-orbit coupling. The acquired spin moment is always parallel to the photon propagation direction but its sign depends on the helicity of the incident X-ray photon and on the spin-orbit coupling (l+s at the LIII and l-s at the LII). The magnetic properties of the sample are driving the second step. A polarised photoelectron occupies the states above the Fermi level and, if there is any imbalance in either spin or orbital momentum in the final states, the XMCD spectrum reflects the difference in the density of states with different spin or orbital moments. The sum of the XMCD spectra recorded at the LII and LIII edges reflects only a difference in the orbital moments of the final states, while the difference is proportional to a spin polarisation of the valence states. This is precisely the content of the magneto-optical sum rules. It is worth mentioning that the summation over two spin-orbit split edges is equivalent to what can be measured for a core level with no spin-orbit interaction. This implies that a dichroic effect at the K-edges is only due to the orbital moments in the valence shell.

The XMCD technique is now widely used to unravel the microscopic origin of magnetism in various ferro- and ferrimagnetic systems [1]. Furthermore, it has recently been shown that high-quality XMCD spectra could be recorded on paramagnetic systems, including Pauli and van Vleck paramagnets, subjected to high magnetic field [2].

XMCD has also become a remarkable element-specific magnetometry tool for heteromagnetic systems. A particularly outstanding example is presented below for DyFe2/YFe2 magnetic superlattices. The high sensitivity of XMCD also makes it unique for the study of the magnetic properties of reduced-dimensionality structures: thin magnetic films [3] and multilayers [4], magnetic quantum wires and dots [5]. This is nicely illustrated below in the study of magnetic moments of 3d-impurities on alkali films.

References
[1] M. Besse et al., EPL 60, 608 (2002).
[2] F. Wilhelm et al. To be published.
[3] M. Marangolo et al., PRL 88, 217202 (2002).
[4] F. Wilhelm, PRL 85, 413 (2000).
[5] P. Gambardella et al., Nature 416, 301 (2002).