Since its discovery, optical activity in the visible light range has become one of the most powerful spectroscopic tools in physics, chemistry and biology and has fascinated many successive generations of scientists. Optical activity is a manifestation of spatial dispersion, i.e. the dependence of optical properties of matter on the wavevector of light. Being a first order spatial dispersion effect (linear in the wavevector), optical activity is associated with transition probabilities which mix multipoles of opposite parity, e.g., electric dipole - magnetic dipole (E1.M1) or electric dipole - electric quadrupole (E1.E2). The Curie symmetry principle states that this is only possible in systems with broken inversion symmetry. However, there was rather little motivation to look for optical activity in the X-ray range until X-ray natural circular dichroism (XNCD) was unambiguously detected at the ESRF in quite a few non-centrosymmetric crystals [1]. It stems from the E1.E2 interference terms which can be large in the hard X-ray range, as opposed to the E1.M1 term which is known to be responsible for optical activity in the visible but is regarded as negligible in core level spectroscopies.

Given that the E1.E2 interference term is a traceless tensor, not all crystals with broken space inversion symmetry, only 13 crystal classes, can exhibit optical activity in the X-ray range associated with the E1.E2 term. There is virtually no hope of detecting XNCD spectra in isotropic samples (powders or liquids) except if the E1.M1 contribution becomes non-negligible: in this case the system should exhibit enantiomorphism, i.e. it can exist in both right- and left-handed forms. We tried very hard to measure XNCD spectra in solutions of chiral complexes but strong radiation damage of the sample prevented us from obtaining any reproducible result.

In order to establish what could be the ultimate order of magnitude of the weak E1.M1 interference terms, we decided to carry out careful XNCD measurements on a single crystal of -NiSO4x6H2O. This uniaxial crystal (with four formula units per unit cell) belongs to the enantiomorphous tetragonal space groups, P41212 or P43212 (see Figure 84). This compound has long been known to exhibit natural optical activity only in the crystalline state, due to a chiral arrangement of non-chiral units: four (nearly) perfect Ni(H2O)62+ octahedrons are located along a screw axis which is parallel to the tetragonal axis of the crystal, i.e. the crystallographic c axis which is the optical axis.

Fig. 84: The pre-edge part of the Ni K-edge XNCD spectra for two enantiomeric uniaxial single crystals of -NiSO4x6H2O. a) XNCD spectra assigned to the E1.M1 interference term, measured at the magic angle. b) XNCD spectra recorded with the X-ray wavevector parallel and perpendicular to the optic axis. The isotropic XANES spectrum is also added for the sake of comparison. c) Crystal structure of the two enantiomers.

For such a uniaxial crystal, the angular dependence of the XNCD signal should vary as (3cos2-1), where denotes the angle between the X-ray wavevector and the crystal optical axis. As illustrated with Figure 84b, this is exactly what we observed: the XNCD spectrum recorded with the X-ray wavevector perpendicular to the optical axis is about twice as weak and it has the opposite sign with respect to the XNCD spectrum recorded in the parallel configuration. This result provided us with an excellent illustration of the fact that the E1.E2 mechanism is quite efficient at the K-edge of transition metals. Unfortunately, it is also strong enough to mask any significant contribution of the weaker E1.M1 interference term, at least under such experimental conditions.

Actually, the only chance to detect a weak E1.M1 XNCD signal is when the optical axis of the crystal is set at the magic angle, ca. 54.73° with respect to the X-ray wavevector. This is because the XNCD signal due to E1.E2 interference term should vanish in this geometry so that only the E1.M1 term could ultimately contribute to the XNCD spectrum. We have performed a whole series of angle dependent XNCD measurements for both enantiomers. The results are shown in Figure 84a, where we focused on the XNCD spectra recorded in the pre-edge region of the Ni K-edge X-ray absorption spectrum. The dichroism spectra reproduced in Figure 84a may reasonably be assigned to the E1.M1 interference term. As expected, the XNCD spectra have the opposite sign for the two enantiomers and their amplitudes are as small as 3x10-5 with respect to the edge-jump. To the best of our knowledge, this appears to be the smallest static circular dichroism signal ever measured in the X-ray range. It is noteworthy that the E1.M1 XNCD signal is ca. 60 times smaller than the signal assigned to the dominant E1.E2 mechanism and has systematically the opposite sign with respect to the E1.E2 contribution measured with a crystal of the same enantiomer.



[1] J. Goulon, A. Rogalev, F. Wilhelm, N. Jaouen, C. Goulon-Ginet and C. Brouder, J. Phys.: Condens. Matter 15, S633 (2003).


Principal publication and authors

A. Rogalev, J. Goulon, F. Wilhelm and A. Bosak, in Magnetism and Synchrotron Radiation, Springer Proceedings in Physics 133, 169-190 (2010).