When the photon energy is tuned near an absorption edge in either scattering or absorption experiments, a considerable change occurs in the cross section. Most importantly an intermediate excited state is formed, which corresponds to the promotion of a core electron into electronic states above the Fermi level. This process is then followed by the decay of the intermediate into the original state with photon re-emission. As pointed out some ten years ago [1,2], there is a special interest when the electron shells involved in the excited state are spin polarised. The resulting theories and experiments have given rise to two new fields of synchrotron magnetic studies, X-ray magnetic circular dichroism (XMCD) and resonant X-ray magnetic scattering (RXMS). XMCD is sensitive to only the imaginary part of the anomalous scattering amplitude, while RXMS measures the modulus squared of the total scattering amplitude including both the real and imaginary parts of the anomalous scattering. Both techniques have the special advantage that they are element and electron shell specific. In the case of XMCD, sum rules [3], which are based on atomic models, are available for extracting the spin and orbital contributions to the magnetic moments. Of course, practitioners of these two spectroscopic methods have focused almost without exception on elements that have appreciable magnetic moments. However, a series of recent experiments at the ESRF using both dichroism and scattering techniques have been performed at the anion K edge of various uranium compounds. The size of the resonant signal has been a considerable surprise.

Dichroism experiments were performed at beamline ID12A on ferromagnetic uranium sulphide. This material orders below Tc = 174 K with a saturation magnetic moment of 1.7 µB. The measurements were performed at 160 K in an applied external magnetic field of 2 Tesla, using total fluorescence detection mode. The differential XMCD spectrum is obtained by flipping the helicity of the X-rays at each point of the energy scan using the recently installed ElectroMagnet/Permanent magnet Hybrid Undulator [4]. The dichroic signal obtained at the sulphur K-edge is much larger than expected and can be easily seen by simply comparing the XANES spectra recorded for both circular polarisations shown in Figure 66. To check that the experimental results are free from any eventual artifacts, the same spectra were recorded for the opposite direction of the external magnetic field. If one corrects the experimental XMCD spectrum for the circular polarisation rate of the monochromatic beam and for the sample magnetisation (which is only 15% in this energy range), the amplitude of the signal at the sulphur K-edge is about 25%. This effect is 250 times bigger than the signal at the iron K-edge in pure ferromagnetic iron and also much greater than found at the K edge of sulphur in EuS [5].

Scattering experiments were first performed on single crystals of antiferromagnetic UGa3 at the ID20 magnetic scattering beamline. This material orders below TN = 67 K with a moment of 0.6 µB on the uranium site and with a magnetic propagation vector of (1/21/21/2). The photon energy was tuned to the K edge of gallium (10.37 keV) at a temperature of 10 K. A very large (by a factor of ~ 1000) enhancement of the small non-resonant magnetic scattering intensity was observed. This is demonstrated by plotting the integrated intensity as a function of incident photon energy in Figure 67a (solid points). The energy width is ~ 4 eV, essentially the expected core-hole lifetime of the intermediate state. Also shown in Figure 67a is the fluorescence spectrum measured in the same experiment (dashed line). The position of the peak in energy strongly suggests that the resonance is dipolar in nature, i.e. involves the 4p electron states of Ga. The signal is present only in the channel with s incident polarisation from the synchrotron, again consistent with a dipole transition. Figure 67b shows the intensity as a function of position in reciprocal space. The signal has a periodicity identical to that of the uranium magnetic structure, is long range in real space, and disappears by TN. Similar effects were subsequently observed in UAs when photons were tuned to the K edge of arsenic. In these experiments intensities of up to 80,000 cts/sec in full polarisation mode were observed on ID20. Such enhancements of the non-resonant scattering signal are much greater than found on other K edges, even those of the magnetic 3d elements.

In all experiments reported here, the signal does not arise from a simple magnetic dipole situated at the non-magnetic element ­ its size would have to be so large that it would have been seen easily by neutron scattering. The effect is undoubtedly more subtle than that, perhaps involving the hybridisation of the anion p orbitals with the uranium 5f magnetic states. Theory continues to address the underlying physics, but the experiments open new perspectives using resonant effects, which will certainly be the subject of future efforts.

References
[1] D. Gibbs et al., Phys. Rev. Lett., 61, 1241 (1988); J.P. Hannon et al., Phys. Rev. Lett., 61, 1245 (1988).
[2] G. Schütz et al., Phys. Rev. Lett., 58, 737 (1987).
[3] P. Carra, M. Altarelli, Phys. Rev. Lett., 64, 1286 (1990).
[4] A. Rogalev et al., Proc. SPIE, 4774, 275 (1999).
[5] A. Rogalev et al., J. Phys. Cond. Matter, 11, 1115 (1999).

Authors
RXMS: D. Mannix (a,b), L. Paolasini (b), N. Bernhoeft (c), A. Stunault (b), C. Vettier (b,d), D. Kaczorowski (e), G.H. Lander (a); XMCD: A. Rogalev (b), J. Goulon (b), J.P. Sanchez (c), N. Kernavanois (c).

(a) ITU, Karlsruhe (Germany)
(b) ESRF
(c) CEA-Grenoble (France)
(d) ILL, Grenoble (France)
(e) Polish Academy of Sciences, Wroclaw (Poland)