The use of X-rays for the investigation of the magnetic properties of matter is a relatively recent field. The first magnetic diffraction experiment was published in 1972, while the first magnetic circular dichroism experiment was reported in 1987.

At the ESRF, diffraction- or absorption-based techniques to investigate magnetic materials have been not only extensively used, but often improved and pushed to ever more ambitious goals. This is not only due to the high brilliance typical of third generation synchrotron sources, but also to the development of specially designed insertion devices and of optical elements which are able to generate photons with a controllable state of polarisation, and which make it possible for the experimentalist to switch between different states rapidly. 

The beamlines ID12A, ID12B and ID20 are mainly devoted to magnetic studies, but other beamlines such as ID3, ID10, ID15 and ID24 have also given important contributions.

The techniques used range from non-resonant magnetic diffraction (which allows the orbital and the spin part of the magnetic form factor to be determined separately) to resonant scattering (which allows element-specific information to be obtained and surface and few-layer systems to be investigated) to circular dichroism, which has imaging as well as time-resolved capabilities.

The experiments performed at the ESRF include real breakthroughs such as the detection of the magnetisation of a single atomic layer by resonant surface magnetic scattering, the high-resolution imaging of magnetic domain walls, and the study of the time-dependent magnetic response on a nanosecond time-scale by circular dichroism. In a field closely related to magnetic studies, the first unambiguous evidence for X-ray natural circular dichroism in a gyrotropic system has also been provided.


Magnetism in 3d transition metal compounds

Two areas in magnetism made accessible by synchrotron X-ray sources have been exploited at ID20: the possibility to separate the spin (S) and the orbital (L) contributions to the magnetisation density in solids and the determination of the magnetic components from various electronic shells. The separation of L and S is achieved by analysing the non-resonant magnetic scattering of X-rays, while contributions from different electronic shells are probed by resonant scattering. This is illustrated by the determination of the L/S ratio in NiO and the energy dependence of the resonant scattering at the K-edge of manganese in RbMnF3.

NiO was the first material to be studied by magnetic X-ray diffraction in 1972 by de Bergevin. Since then, seven orders of magnitude in scattered intensity have been gained. The analysis of the linear polarisation of the scattered X-ray beams at several magnetic reflections has allowed the ratio of the orbital to spin momenta to be determined as a function of the scattering vector Q (Figure 22). The observed Q-dependence arises from the individual form factors which correspond to different spatial extent of the spin and orbital densities. The extrapolation to Q = 0 of the L/S ratio indicates that the orbital contribution to the magnetisation (17%) is larger than expected from bulk measurements and neutron data. At large Q values, the dipolar approximation used to predict the Q dependence may no longer be valid. These results will bolster new calculations of magnetisation densities in solids and demonstrate that L/S determination can be achieved with magnetic X-ray scattering experiments.

Resonant magnetic scattering is usually interpreted in terms of electric multipole transitions between a core level and excited "magnetic" electronic states available at the Fermi level. At the K edge of 3d transition elements, the core level is the 1s level, and the excited levels are the strongly spin polarised 3d level (quadrupolar transition) and the 4p level (dipolar transition). Quadrupolar resonances are weak and the 4p level is not expected to be strongly spin polarised. The results from the antiferromagnet RbMnF3 demonstrate that the resonant process is far from being understood. The scattered intensity (Figure 23a) shows enhancement of a factor of 10, close to what is observed at the L edges of the light rare-earths. Furthermore, polarisation analysis of the diffracted beam shows that the resonance can only be observed in the rotated channel, implying a dipolar origin and a sizeable spin polarisation of p electron bands. The resonance has a striking line shape which follows closely the white line (Figure 23b). The white line of 3d transition elements is known as being of dipolar origin, and related to the formation of excitons at the 4p level.

The fact that the resonance at the K edge of manganese in RbMnF3 is as strong as the resonance in light rare-earths is completely unexpected. Moreover, the line shape of this resonance cannot be explained in an atomic-like picture and suggests that solid state effects also have to be taken into account.



F. de Bergevin (a), N. Bernhoeft (b), T. Brückel (c), V. Fernandez (d), C. Giles (e), W. Neubeck (d), A. Stunault (d), C. Vettier (d), D. Wermeille (d), to be published.

(a) CNRS, Lab. de Cristallographie, Grenoble (France)
(b) ILL, Grenoble (France)
(c) HASYLAB, Hamburg (Germany)
(d) ESRF
(e) LNLS, Campinas (Brazil)




Observation of changes in 5d band character

Tuning the photon energy to the L-edges of the rare-earths (7 - 9 keV) has been applied to study the magnetism of elements and compounds of this series for the last 10 years. Experiments at ID20 show how the influence of electronic band structure can be unravelled by X-ray magnetic scattering methods. We have studied the ordered compound DyFe4Al8. It is known from Mössbauer and neutron experiments that the Fe sublattice orders at TFe = 165 K, whereas the 4f moments of the Dy sublattice do not achieve long-range order until a much lower temperature of TDy ~ 20 K. Early X-ray experiments have shown that the magnetic structure is a cycloid. The temperature dependence of the L3 signal is shown in Figure 24 (yellow points). Despite the fact that the L-edge signal is element-specific to the Dy site, effects are seen at both TFe and TDy. The L3 signal appears small at TFe and we associate this with a small polarisation of the 5d states by the molecular field from the Fe 3d moments. It rises sharply when the Dy 4f moments order at TDy. This increase in intensity is associated with a small (~ 2 eV) increase (on cooling) of the energy of the centroid of the L3 resonance. Such an intriguing energy shift would not be expected on the basis of a simple rigid band model, and implies that the character of the 5d bands changes when the Dy 4f moments polarise. Hence the magnetic scattering energy dependence may be used as a spectroscopic tool to test calculations.

Even more exciting results have now come from the temperature dependence at the Dy L2 edge. The intensities are shown (filled points) on the same scale as those of the L3 in Figure 24. At low temperature the L3 signal is the stronger, in accordance with experiment and theory. We can best illustrate this by plotting the branching ratio, which is the ratio of the intensities at the two edges as a function of temperature as in Figure 25. The low-temperature value of ~ 5 is in good agreement with dichroism and other experiments on ordered Dy compounds. The energy dependence of the L2 resonance peak shows the same increase by ~ 2 eV as the compound is cooled through TDy. Certainly the most puzzling feature of the data is the dominance of the L2 signal in the temperature range when only the Fe 3d moments are ordered, leading to a branching ratio less than one for T > TDy. It is now clear that the disordering of the 4f moments at TDy actually gives rise to a much larger change in the character of the 5d bands than it had been previously imagined. There is currently no explanation of this, but calculations are in progress to examine carefully what happens to both the character and energy dependence of the 5d band of this material. Solid state physics effects should be taken into account to explain the resonant process.


S. Langridge (a), J.A. Paixão (b), G.H. Lander (c), N. Bernhoeft (d), C. Vettier (e), to be published.

(a) ISIS Facility, Rutherford Appleton Laboratory (UK)
(b) University of Coimbra (Portugal)
(c) EITU, Karlsruhe (Germany)
(d) ILL, Grenoble (France)
(e) ESRF




Element specific magnetism in actinide compounds

Tuning the photon energy to the M edges of actinide elements is known to greatly increase the sensitivity to magnetism. We have used this technique, with its inherent element specificity, together with the high brilliance of the undulator beams at the ESRF, on ID20, to examine the magnetic properties of small single crystals containing uranium and neptunium. The successful outcome of this project required the development of a technique to position the beam on the sample surface (with an area of ~ 0.2 mm2 and masses less than 1 mg) with the crystal encapsulated for safety reasons and therefore not visible.

The crystals studied were made by slow cooling alloys of (U1-xNpx)Ru2Si2, which has a tetragonal crystal structure, and is of interest because the parent compound with x = 0 is a well-known heavy fermion. Compounds throughout the solid solution are antiferromagnets and this gives rise to additional peaks in the diffraction pattern from the magnetic periodicity, which is different to that of the charge. The crystals were prepared and mounted on Ge(111) wafers at the Institute for Transurarium Elements, Karlsruhe, Germany.

Figure 26 shows the intensity of the (003) magnetic peak as a function of photon energy for the x = 0.5 compound. Marked on the figure are the M4 and M5 resonant peaks that arise because both the U and Np atoms carry magnetic moments. Although more data analysis is required (the experiments ran in late June!) this figure already shows that both atoms carry a moment and that the larger moment is on the Np atoms. Since this quantity can be determined also by Mössbauer spectroscopy, the synchrotron radiation experiments provide a means for determining the U moments. Measurements as a function of temperature have shown that the temperature-dependence of the U and Np moments are different, suggesting that the magnetism of the U atoms is driven by the molecular field produced by the larger Np moments.

In the pure Np compounds (x = 1) the magnetism ordering is incommensurate and we have determined the wavevector as a function of temperature. The T-dependence of the first and third harmonics of the modulation are shown in Figure 27. It is interesting to note that the weak third order peak was also measured as 0.13 cts/sec in a "tour de force" neutron experiment from a 2 mg sample.

A rough calculation shows that the (intensity/volume) ratio for this synchrotron radiation experiment is many orders of magnitude greater than in the neutron experiment. The point of these comparisons is that completely new science can be carried out with the synchrotron beams, e.g. microgram as well as surface and thin film magnetism on actinide samples.

The use of small crystals for these studies is motivated by three different reasons. First, smaller crystals tend to be easier to produce and have a smaller mosaic spread than larger samples. Second, the radioactive inventory is reduced to a level that is acceptable at the ESRF with the necessary precautions and procedures. Third, the results allow us to imagine a new frontier of magnetism by being able to measure the microscopic magnetic properties on samples of curium (Cm), berkelium (Bk), and californium (Cf), which are in an uncharted (for solid-state physics) part of the periodic table, and where only very small samples are available. With almost 106 cts/sec. in the magnetic peaks from the samples examined in our present experiments, we may extrapolate to experiments on microgram samples, as well as the magnetism of nanometer-thick actinide films!


E. Lidström (a,b), A. Hiess (c), P. Dervenagas (d), M. Longfield (e), G.H. Lander (b), J. Rebizant (b), C. Vettier (a), to be published.

(a) ESRF
(b) EITU, Karlsruhe (Germany)
(c) ILL, Grenoble (France)
(d) CEA-Grenoble (France)
(e) University of Liverpool (UK)




Magnetic properties of ultra-thin holmium films

The magnetism of epitaxially-grown thin films and multilayers is a field of growing interest adressing both fundamental as well as application-related questions. While magnetic multilayers are particularly suited to study the coupling mechanism between two or more magnetic layers separated by paramagnetic or non-magnetic spacer layers, the dependence of the magnetic properties on the film thickness can best be studied in single-crystalline films grown epitaxially on a suitable substrate.

We have started a programme at beamline ID10A with the goal to study the magnetic properties of ultra-thin rare-earth films by resonant magnetic X-ray scattering. The first experiments were performed on thin holmium films (370 Å < d < 780 Å) taking advantage of the resonant enhancement of the magnetic scattering intensity at the Ho LIII absorption edge. These films were grown epitaxially on a W(110) single crystal with the surface normal parallel to the Ho c-axis. The growth process as well as the film characterisation were performed in situ inside a small UHV chamber mounted on the ID10A diffractometer.

Our main interest was to investigate the influence of the film thickness on the magnetic structure of holmium. Bulk holmium orders magnetically below TN = 138 K and forms a helical magnetic structure. The magnetic spiral is oriented along the c-axis with the moments lying in the ab-planes. The repeat distance of the spiral depends upon the temperature and is about 10 atomic layers at T = 40 K. For film thicknesses below this value, no complete spiral can be formed and the magnetic structure in such a film is unknown. A spiral antiferromagnetic structure manifests itself in magnetic satellite reflections of the type (0, 0, L) +/- , where is the modulation wavevector. Figure 28 shows the results obtained with a 370 Å film at 40 K in a scan along the reciprocal c* direction. Charge scattering from the (0, 0, 2) Bragg reflection was strongly reduced by employing a polarisation analyser operated in the charge forbidden - channel. The (0, 0, 2) - magnetic satellite is clearly visible in this film. The temperature dependence of the modulation wavevector as well as the transition temperature were studied in the 370 Å thick film and in a 780 Å thick film, and showed agreement to within 1 K with the bulk behaviour. This indicates that the magnetic behaviour of holmium films with thicknesses down to 370 Å follows essentially the one found in bulk single crystals. Investigations of thinner films are presently in progress.


E. Weschke (a), C. Schüssler-Langeheine (a), R. Meier (a), G. Kaindl (a), C. Sutter (b), D. Abernathy (b) and G. Grübel (b), to be published.

(a) Institut für Experimentalphysik, FU Berlin (Germany)
(b) ESRF




Resonant magnetic scattering as a probe of induced spin polarisation in magnetic multilayers

In highly-correlated Ce/Fe multilayers, Cerium adopts an -phase like electronic structure within ca. 20 Å from the Fe interface. X-ray Magnetic Circular Dichroism (XMCD) experiments at the L and M edges, have confirmed that the hybridisation of the Ce 5d and 4f states with the 3d states at the Fe interface induces at room temperature a magnetic order with a weak magnetic moment on both the 5d and 4f states of Ce. XMCD results reveal a fundamental difference between the 4f polarisation which is restricted at the Fe interface and the 5d one which extends over 20 Å, though it decreases with Ce thickness.

In order to keep more insight into the complex behaviour of the induced 5d states polarisation, we have performed a resonant magnetic scattering experiment at the Ce L2 edge on a Ce10Å / Fe30Å multilayer. Ce being amorphous, it does not provide diffraction peaks at large angles. We thus had to investigate the low-angle Bragg peaks of the multilayer. For such an experiment, a circular polarisation of the X-ray beam is required to enhance the magnetic contribution to the scattered intensities. Measurements have been performed on the ID12A beamline with the benefit of a high circular polarisation rate (c = 0.84 after monochromator), with magnetic field applied parallel (I+) or antiparallel (I-) to the diffraction plane. Figure 29 displays the asymmetry ratio (I+-I-)/(I++I-) measured at the Ce L2 edge on top of the first 4 Bragg peaks. Their amplitudes vary from 1 x 10-3 for the first order, up to 1.8 x 10-2 for the fourth order. Each spectrum is the sum of 3 to 4 scans of one hour. The overall shape of the energy dependence of the asymmetry ratio compares well with the L2 XMCD spectra in which there are two lobes associated with the 4fo and 4f1 configurations in the final state. The increase of the amplitudes with the Bragg order directly evidences a non-constant magnetic profile across the Ce layer.

To analyse the data, we used a kinematical approach, the Ce layer being described by a discrete stacking of atomic plane, each carrying a polarisation amplitude. The shape of the magnetic resonance is taken from the XMCD measurements. Both interface planes are refined for concentration in Ce. The distribution tentatively derived from the dependence of the XMCD spectra on the Ce thickness, with the polarisation decreasing with the distance from the Fe interface, does not allow the fit of the data, the sign of the 3rd order asymmetry ratio even being reversed with respect to the experimental data. A different refinement procedure exploiting our XRMS data yields an unexpected oscillating polarisation distribution shown in Figure 30, together with the Ce concentration at the interfaces. We point out that the average moment deduced from the fitting is in agreement with its XMCD evaluation within 20% depending on the actual values of the structural parameters used in the calculation.

This first attempt to determine the induced spin polarisation profile in a magnetic multilayer demonstrates the usefulness of the XRMS method in complement with the XMCD one.


L. Sève (a), J.M. Tonnerre (a), F. Bartolomé (a), D. Raoux (a), M. Arend (b), W. Felsch (b), A. Rogalev (c), J. Goulon (c), to be published

(a) Laboratoire de Cristallographie, CNRS, Grenoble (France)
(b) I. Physikalisches Institut, Göttingen (Germany)
(c) ESRF




Anomalous thermal expansion due to magnetism in EuAs3 and MnS2

The good penetration power of high-energy X-rays, together with the high-resolution triple axis diffractometer at beamline ID15A, gives access to the bulk of materials with a resolution for relative lattice parameter deviations d/d down to 10­5. This has been employed to measure the evolution of the lattice parameters as a function of temperature in simple antiferromagnetic systems where the orbit momentum L = 0 and where no crystal field effects are involved in the magnetism. Systems with only a spin contribution as EuAs3 with S = 7/2 and MnS2 with S = 5/2 studied in the present work are supposed to have a small response of the lattice parameter to the magnetic ordering, which has not been accessible by diffraction methods before.

The magnetic semiconductor MnS2 crystallises in the cubic pyrite structure with a = 6.104 Å at room temperature. The compound undergoes a paramagnetic to antiferromagnetic phase transition of first order at TN = 48 K. Scans probing for the lattice parameter show a sudden jump from a single peak to a well separated double peak when descending through the phase transition. The peak positions are translated into relative lattice parameter deviations with respect to their value just above TN, and they are displayed in Figure 31. The appearence of the second peak can be understood by a tetragonal deformation together with intensity contributions of three populated domain orientations. The less intense peaks in both 800 and 080 orientations with respect to the cubic lattice corresponding to a positive deviation of the lattice parameter would suggest an enlargement of the tetragonal axis and a compression of the dimensions in the tetragonal plane.

EuAs3 is an example of a semi-metallic system in which p-f hybridisation effects give rise to its unusual magnetic properties. This compound orders below TN 11 K to an incommensurate sine wave which undergoes an incommensurate to commensurate lock-in first order phase transition at TL 10.3 K. The inset in Figure 32 shows the intensity of the 1, 0, 5/2 non-resonant magnetic reflection with increasing temperature. It drops abruptly to zero at TL. The temperature variation of the interatomic spacings measured at the ­6, 0, 6 charge reflection of the monoclinic lattice shows a thermal expansion anomaly due to magneto-elastic effects below the magnetic phase transition. For higher temperatures, let's say above 15 K, the normal phonon contribution describes the thermal expansion.


T. Chattopadhyay (a), K.-D. Liss (b), Th. Tschentscher (b), to be published.

(a) Institut Laue-Langevin, Grenoble (France)
(b) ESRF



Measurement of the magnetism of a single atomic plane

The interaction of X-rays with the magnetic moment of atoms is usually too weak to be observed in very diluted samples such as a single atomic plane in a macroscopic crystal. Recently, however, it was discovered that under some resonant conditions (the photon energy has to coincide with the energy of some atomic absorption edges), the magnetic coupling of X-rays with the magnetic moment of atoms in resonant condition is largely enhanced. The effect is so intense that it has allowed the magnetism of a single atomic plane of atoms in a buried interface of Co/Pt to be observed. The enhancement effect has the additional advantage of being selective to only the resonant atoms. In our case, as the photon energy was tuned to an absorption edge of Pt, only Pt atoms were probed by the measurements.

We have grown a few atomic layers (from 2 to 12) of cobalt on the (111) surface of a Pt crystal. This was done in the ultra-high vacuum system at the ID3 beamline. An external magnetic field parallel to the surface was applied to magnetise the cobalt films at saturation which, in turn, induced a magnetisation of the Pt atoms at the Co/Pt interface. This magnetisation was shown by tuning the photon energy to the LIII absorption edge of Pt and by isolating the magnetic part of the diffracted intensity from the surface. To achieve surface sensitivity, one has to measure in regions of reciprocal space away from bulk Bragg peaks. Usually the measurements are taken along the so-called diffraction rods which are directions in reciprocal space normal to the surface of the crystal. In order to isolate the magnetic part of the surface-sensitive diffracted intensity in a diffraction rod, one has to measure the difference between the intensities corresponding to the application of the field in two opposite directions. That asymmetry ratio allows one to extract direct information on the localisation within the crystal of the magnetised Pt atoms and on their magnetic moment.

The results depicted in Figure 33 allow one to conclude that the spatial distribution of the magnetised Pt atoms in the direction normal to the film surface is very abrupt; basically only the atomic layer of Pt in contact with the cobalt is magnetised (µ 0.2 µB), the magnetisation of the second atomic layer of Pt being about ten times smaller.


S. Ferrer (a), J. Alvarez (a), X. Torrelles (a), E. Lundgren (a), P. Fajardo (a) and F. Boscherini (b), to appear in Phys. Rev. B. (15 Oct. 97)

(a) ESRF
(b) INFN, Frascati (Italy)




Magnetic interactions in europium chalcogenides revealed by spin-polarised XAFS

Europium chalcogenides form an important class of magnetic semiconductors and Heisenberg magnets with a highly symmetric rock salt structure and strongly localised spin-only 4f moments at the europium cation. These moments are considered to be coupled (i) by a ferromagnetic exchange between the nearest Eu neighbour sites and (ii) by an antiferromagnetic superexchange between the next nearest neighbours. Even though the corresponding exchange parameters have been determined experimentally, the underlying mechanism of individual interactions is not yet clear. It was envisaged that the anions in the europium chalcogenides are involved in the magnetic interactions and do not serve simply to expand or contract the lattice. X-ray Magnetic Circular Dichroism (XMCD), being an element and orbital selective tool to study local magnetic structure, is expected to help the understanding of the exchange interactions in ferro(i)magnetic compounds.

XMCD study of EuS and EuSe in the low temperature ferromagnetic phase has been performed at the ID12A beamline using a helical undulator source. Analysis of the near edge XMCD spectra recorded at the Eu L-edges has shown that the signal is mainly controlled by intra-atomic 4f-5d exchange and is fairly little influenced by the inter-atomic interactions. In contrast, the dichroic signals observed in the EXAFS range strongly depend on the nature of the anion. The presence of the Eu-S (Eu-Se) signatures in the corresponding spin-polarised FT-EXAFS spectra (see Figure 34) indicates that the chalcogen anions are magnetically polarised and therefore are involved in magnetic interactions. This observation is further supported by the detection of an intense XMCD signal at the Se K-edge (Figure 35) which has evidenced spin polarisation of the Se p valence band via its strong hybridisation with the Eu 4f and 5d states. Such hybridisation results in the existence of small moments carried by the chalcogen anions and promoting the indirect exchange coupling between Eu moments.


A. Rogalev (a), C. Neumann (a), V. Gotte (a) and J. Goulon (a), submitted to Phy. Rev. Lett.

(a) ESRF




Magnetic microstructures observed with a photoemission microscope

Understanding the physical properties of magnetic microstructures is of great technological importance, particularly in the field of magnetic storage devices. The improvement in storage density has reduced the average bit size down to some 100 nm. In order to characterise these magnetic systems, imaging techniques with a lateral resolution in the range of the domain boundaries, i.e. several 10 nm, are needed. Photoelectron emission microscopy represents a very promising approach to meet this challenge. Exploiting soft X-ray magnetic circular dichroism (XMCD) as a magnetic contrast mechanism, magnetic sensitivity and elemental selectivity are combined in a unique manner.

Our experiments made use of the circularly polarised undulator radiation at ID12B and employed a newly-developed photoelectron emission microscope (Focus IS-PEEM) with an immersion lens objective and an integrated sample stage. The parallel imaging capability, combined with the high brightness at ID12B, significantly cut down the acquisition times and in some cases permitted even real-time studies. The microscope itself achieves a lateral resolution of about 30 nm.

Figure 36 shows the magnetic domain image of a micropatterned CoPt multilayer recorded at the Co L3 edge. Each square has a side length of ~ 20 µm and exhibits a very complex domain structure which is due to the strong intrinsic magnetic anisotropy of the CoPt system. The filigran feather-like domain pattern suggests that the local magnetisation vector has a large number of possible spatial orientations. This is compatible with the film being polycrystalline. In this case the easy axes of magnetisation change with the spatial orientation of the crystallite more or less randomly. The result is a rather complex variation of M along the sample surface which is sometimes called a magnetisation "ripple".

The transition region between two neighbouring domains, the so-called domain wall, represents a magnetic microstructure on a yet smaller lateral scale. Using the orientation dependence of XMCD, we were able to selectively image the magnetisation component within the wall along the direction of light incidence. An example is given in Figure 37 which shows a small section from an Fe(001) single crystal surface. We have chosen a region of the sample where the magnetisation in the domains points perpendicular to the incoming light beam. Consequently the domains themselves do not show a magnetic contrast. Their boundaries, however, show up as narrow bright and dark lines.

In order to understand this finding we must consider the difference in the behaviour of a domain wall in the bulk and at the surface. The domains in Figure 37 are oppositely magnetised and are thus separated by a so-called "180° wall" in which the magnetisation vector M rotates continuously by 180° from one orientation to the other. In the bulk the axis of rotation is normal to the wall. This case is called a Bloch wall. Note that in the centre of the Bloch wall M stands perpendicular to the surface plane. This energetically very unfavorable situation is avoided, if M rotates within the surface, i.e., the rotational axis lies within the wall plane and normal to the surface. This behaviour is said to be Néel-like. As a result, a bulk Bloch wall takes a Néel-like surface termination. It should be pointed out that the rotations of M in the Bloch- and Néel-like part of the wall are not uniquely connected. Experimentally, we may therefore observe a change of contrast along the wall. This change of rotational sense of M is limited to the near-surface region and forms a surface magnetic vortex (indicated by the circles). The width of the walls which are known as "V-lines" is found to be of the order of 500 nm. These V-lines reflect a complex closure domain pattern caused by the bulk magnetisation.



C. M. Schneider (a, b), R. Frömter (b), C. Ziethen (a), W. Swiech (c), N.B. Brookes (d), G. Schönhense (a), J. Kirschner (b), to be published.

(a) Inst. f. Physik, Joh. Gutenberg-Universität, Mainz (Germany)
(b) Max-Planck-Institut für Mikrostrukturphysik, Halle (Germany)
(c) F. Seitz Materials Research Lab., University of Illinois (USA)
(d) ESRF




Nanosecond-resolved XMCD

The dynamics of magnetisation switching is an essential issue in recording technology. XMCD is a unique element-selective, orbital symmetry-selective spectroscopy. It has been enriched by the addition of the time resolution thanks to microcoils which generate pulses of magnetic field as large as 1 Tesla at the 357 kHz ESRF frequency. The pump (magnetic field of a few ns) ­ probe (X-ray pulse ~ 100 ps-long) scheme has been implemented taking advantage of the single-bunch filling at the ESRF which opens up tuneable delay between pump and probe pulses from zero up to 2.8 µs. The use of the ultra-fast electro-acoustic chopper based on the diffraction of the surface acoustic waves generated on a multilayer-coated LiNb03 crystal developed by Tucoulou et al., enables us to enlarge the accessible window for phasing the pump with respect to the probe.

The hair-pin shaped, 50 µm-thick copper microcoil with a 50 µm-wide gap takes full advantage of the focusing optics of the energy dispersive XAS spectrometer of ID24 which combines a Kirpatrick-Baez double mirror configuration and a Bragg dispersing and focusing geometry to focus the polychromatic beam to a spot size as small as 30 x 100 µm2.

The full spectrum of the transmitted beam is collected by a CCD-based position sensitive detector in a few tens of milliseconds due to the typical flux of 5 x 1010 photons per second focused on the sample in the single bunch mode.

The last (but not least) optical element is the quarter wave plate (QWP) which has been implemented on ID24 last year. Thanks to the high quality of diamond crystals, the birefringent properties in the vicinity of a Bragg reflection are able to transform X-rays with the natural horizontal polarisation produced by the flat undulator into circular polarised photons. In addition, the flipping from one side of the diamond (111) Bragg peak to the other allows the helicity to be tuned from left to right with a circular polarisation rate close to 1 (0.99 at 7243 eV, the Gd L3 edge energy) (Figure 38).

The first study deals with a Gd-Co3 amorphous thin film which saturates at about 0.5 T. The film is slightly anisotropic with an easy axis of magnetisation in the plane of the film. In the present experiment the field produced by the microcoil is applied perpendicular to the film, along the X-ray propagation direction. The XMCD signal comes from the difference of absorption by flipping the helicity which is unusual in most experimental stations in the world except at the ESRF on ID12 where the helicity flipping comes from the phase flipping of the helical undulator.

The XMCD-based Gd magnetisation has an unexpected oscillatory character (Figure 39). Since each time-dependent XMCD data point is averaged over 80 millions of minor hysteresis cycles generated by the repeated magnetic field pulses, the oscillations of the Gd magnetisation within the duration of the pulse are not expected to be stochastic. In addition when the pulse is back to zero, the magnetisation of the probed area (30 x 100 µm2) oscillates, going first to a negative value, and has been observed many times.


M. Bonfim (a), K. Mackay (a), S. Pizzini (a), A. San Miguel (b), H. Tolentino (a), C. Giles (b), T. Neisius (b), M. Hagelstein (b), F. Baudelet (c), L. Varga (d), C. Malgrange (b, d), A. Fontaine (a, b), to be published in SRI'97 proceedings.

(a) LLN, CNRS, Grenoble (France)
(b) ESRF
(c) LURE- CNRS-CEA-MEN, Orsay (France)
(d) LMCP Université de Paris 6 & 7 (France)




X-ray natural circular dichroism (XNCD) in gyrotropic single crystals

Natural optical activity of crystals has fascinated generations of physicists since the discovery by Arago in 1811 that crystalline quartz was optically active. The roots of modern stereochemistry are indeed in the realisation by Pasteur and Fresnel that molecules or crystals which exhibit optical rotation must have specific symmetry properties: natural optical activity can only be observed in gyrotropic crystals, i.e. a subset of the 21 non-centrosymmetric crystal classes. From a purely mathematical point of view, optical activity is represented by a cartesian tensor which can be decomposed into 3 irreducible representations with respect to the Group of 3D rotations: (i) a pseudo-scalar; (ii) a vector; (iii) a rank-2 tensor called "pseudo-deviator". All crystal classes compatible with chiral enantiomorphism, i.e. with the separation into right-handed (R) or left-handed (L) isomers or helices, have a non-zero pseudoscalar part. It was discovered much later that the pseudodeviator part, if non-zero, was associated with remarkable properties in non-linear optics: such crystals are most efficient for harmonic generation and frequency doubling in lasers. Thus optical activity is an important concept, not only in stereochemistry, but also in materials science.

Natural Circular Dichroism (NCD) refers to the differential spectral absorption between left- and right-handed circularly polarised light. This spectroscopy has been known for a very long time at optical wavelengths. However, all attempts made so far to detect Natural Circular Dichroism in the X-ray range had failed. This is due to a number of experimental problems which could not be solved for a long time. Also reflecting the existence of such difficulties is the fact that X-ray Magnetic Circular Dichroism (XMCD) was measured for the first time only in 1987 by G. Schütz et al. This technique has now been recognised as a very useful technique giving access to the orbital or spin components of the local magnetic moments in ferromagnetic materials. Ten years later, a series of experiments carried out at the ESRF beamline ID12A produced the very first unambiguous evidence of X-ray Natural Circular Dichroism (XNCD) in a gyrotropic inorganic crystal.

Our team contributed to initiate two complementary projects concentrating on the experimental detection of XNCD in oriented crystals:

(i) the first project which was developed in collaboration with external groups of ESRF users from U.K. and Italy focused on the detection of XNCD in a stereogenic organometallic complex in which the metal centre (i.e. a rare-earth cation) is itself in a chiral ligand field. The strategy was to grow single crystals in which the (+) and (-) enantiomers were fully resolved.

(ii) the second project concerned the investigation of inorganic crystals which may - or may not - be chiral but are known to exhibit a large non-linear susceptibility at optical wavelengths. The results were discussed in close collaboration with another group at the University of Paris-VI.

We report here briefly only on the second project for which we obtained more complete results. For the sake of simplicity, we chose a uniaxial single crystal of a-LiIO3 which is often used to double the medium power frequency of [Ti: Sapphire] or [Cr: LiSrAlF6] lasers. The high quality crystal was grown by the Chinese company CASIX Inc.

Whereas in XMCD experiments it is more convenient to flip the direction of the magnetic field used to magnetically order the sample, it is mandatory in XNCD experiments to flip the helicity of the incoming photons without changing the direction of emission of the X-ray beam by more than 1 µrad. This turned out to be feasible with the ESRF undulator Helios-II. Nicely structured XNCD spectra were measured in the Fluorescence Yield (FY) technique at the LI , LII and LIII absorption edges of iodine. The quality of the data is not limited by the performances of the beamline but by the long-term angular stability of the source.

From the spectroscopic point of view, the measured XNCD signal was assigned to electric dipole (E1)-electric quadrupole (E2) interference terms. Our interpretation is fully supported by band structure calculations. Keeping in mind that sum rules have also to be satisfied for XNCD, it is noteworthy that the signatures recorded at the LII-III edges have the same sign (Figure 40). This is in contrast with XMCD: while spin-orbit interaction is well known to be the driving term in XMCD, this is clearly not the case for XNCD. Our result also leaves very little space for a contribution of the electric dipole-magnetic dipole (E1.M1) interference terms which are usually dominant at optical wavelengths but should be hardly detectable in the X-ray range.


J. Goulon (a), C. Goulon-Ginet (a, b), A. Rogalev (a) and V. Gotte (a), to be published.

(a) ESRF
(b) Université Joseph Fourier - Faculté de Pharmacie, Grenoble (France)