Detection of Magnetic Circular Dichroism Using TEM and EELS

Abstract

Magnetic Circular Dichroism (MCD) is a phenomenon that occurs in magnetic materials whereby the intensity of transitions from core states to available states above the Fermi energy depends on the circular polarization of the exciting radiation. This is due to the fact that spin-orbit coupling breaks the degeneracy of core states with different total angular momentum (J) and the magnetic field gives rise to difference in spin-up/spin-down density of available states. Traditionally, those transitions are excited with X-rays (XMCD). We present here a new technique by which a virtual circularly polarized photon is absorbed in Electron Energy-Loss Spectroscopy (EELS), giving rise to EELS-MCD (EMCD). The basis of this work is the equivalence between photon polarization (ε) and electron momentum transfer (h q) in the determination of the scattering cross section. The equivalent of circular polarization in EELS is achieved through special scattering conditions as interference between Bloch waves in a crystal. Angular resolved EELS is then used to measure spectra under different polarization conditions, from which information about the magnetic properties can be extracted, in particular the ratio of spin to orbital contribution to the magnetization. Measurements can be performed by simply recording spectra at two particular symmetric points in reciprocal space or by acquiring the whole diffraction pattern trough a series of energy filtered images. Spatial dichroic maps can be obtained too either in EELS STEM or by EFTEM. The advantage with respect to XMCD lies in the higher spatial resolution (2 nm). However, since the magnetic information is intrinsically entangled with dynamical diffraction, DFT simulations of the momentum space are required to extract quantitative information from the measurement. We present an example in this work. This also means that the measured difference depends on parameters such as crystal thickness and orientation. It has been recently proposed that a way to overcome this limitation would be to use so-called electron vortex beams. EMCD would then be applied to a larger class of samples and become a true complementary alternative to XMCD.