Work package description:

Leader of WP1, Verification of the predicted chiral transitions.
The general aim of this WP is the verification of the predicted effect. Experiments with magnetic materials well known to show XMCD (preferably) are performed. Since Energy Loss Magnetic Chiral Dichroism (EMCD) in the TEM is fainter than the XMCD signal and depends on experimental parameters such as thickness, excitation error and momentum resolution in a non-trivial way, a series of experiments on the 3d-metals Fe and Ni has been performed to demonstrate EMCD in the TEM. Work proceeds in parallel along two experimental roads with different approaches to realize the necessary phase shift between plane incident waves:
Task 1 "Intrinsic road": Use of a larger monocrystalline region of the specimen as a beam splitter. This is the default road presenting two main advantages (intrinsic phase lock by the lattice potential, easy implementation in standard TEM-environments).
Task 2 "Biprism road": The use of a biprism as a beamsplitter has the advantage that the phase shift can be tuned independently of the specimen, so it is more flexible. Several experimental difficulties such as instabilities of the beam, the biprism and the specimen are presently being investigated.
Task 3 "Comparison with calculations": Comparison of EMCD spectra with calculational results. XMCD data serve as reference. This task guarantees that the measured EMCDs have been cleaned from spurious effects and artifacts.

Status quo:

Task 1: The "detector shift" and "ESD" technique were tested. Furthermore two new interferometric techniques have been developed. They are variations of the "detector shift" and "ESD" technique. The variation consist in using the Objective Aperture (OA) to select kf and therefore the momentum transfers q and q'. The main advantage is that now the Detector Entrance Aperture (DEA, either of a Energy Filter or of a EEL Spectrometer) is no longer required to be in the diffraction plane and the TEM can be operated in image mode. That way, the spatial resolution can be improved, being it given by the projected DEA in the image plane. The disadvantage is that the smallest OA of the TEM in Vienna is 4.8 mrad, approximately a third of the angular distance between the transmitted beam and the 200 beam in Nickel. It means that the error in q is bigger than when using the DEA to select kf, and the dichroic signal will be a smaller percentage of the detected signal. Moreover this technique will not be suitable for the use of the biprism if the two resulting direct beams form an angle smaller than the OA (as it is likely to happen).
For the "OA shift" technique the results show that the spectra have indeed a better signal-to-noise ratio but a dichroic signal of only 10-15%. It was also possible to obtain so called "chirally filtered" images: four energy-filtered-background-subtracted images are recorded for two position of the OA on the Thales circle in the DP and for two position of the energy slit (on the Ni- L3 or L2 edge). The ratio of two images differing for the position of the energy slit only should show variation of the strength of the dichroic signal according to local variation of thickness and/or orientation (and therefore of the dephasing of the two beams). The difference between two images differing for the position of the OA only should be complementary for the two edges. The preliminary measurements performed with this technique didn't show such behavior and must be therefore more accurately repeated, the problem being most likely related to dark count nonlinearity of the CCD at these low signals.

Task 2: It was found that a "standard" biprism holder used for holography (placed in the SAD plane) is too big to fit in the condenser system. A new construction of the holder was developed, because simple shrinking of dimensions of the standard holder is hardly manufacturable.

A prototype of the new holder was tested and interference fringes with spacing between 6 nm and 18 nm in the specimen plane were achieved. For observation of the chiral phenomenon, fringe spacing has to be either bigger than 10 micrometer or smaller than 0.2 nm. The understanding of a condenser system (accelerator, 3 condenser lenses and one condenser-objective lens) is necessary to reach the required fringe spacing.

As a first step, electron trajectories in a single lens were calculated analytically and a program was written for the visualization of the trajectories. It was found that virtual electron sources created by the biprism have to be imaged into the principal plane of the condenser-objective lens to achieve parallel illumination of the specimen with two overlapping electron waves.
In order to image the virtual sources in a desired position the complete condenser system has to be analyzed. A program calculating the beam path through the condenser system was written. At the moment it operates with arbitrary lens parameters. The actual mocroscope's parameters have to be obtained from the manufacturer of the microscope. Some preliminary data have already been received.
A new problem has occurred. According to the manufacturer, the biprism in the condenser system is placed inside the C2 lens. The action of the electrostatic field of the biprism and a magnetic field of the lens cannot be, in general, treated separately. A new calculation of the electron trajectory is necessary.
The workplan is:
- calculate electron beam trajectory in a lens with a biprism inside
- decipher lens parameters from the data received from the manufacturer of the microscope
- test the stability of the new biprism holder construction
Realization of the "large fringes" appears to be feasible with some additional improvements. The usability of large crystalline regions as a beam splitter has successfully been demonstrated. So far, however, there is no biprism available for first experiments due to charging problems with the current biprism in Dresden. A new biprism will be prepared by the Triebenberg group. For the biprism, the Dresden group reports a successful extraction of the TEM parameters which can be used for a precise input into the simulation program which can be used to optimize the illumination conditions.
For future experiments, a fringe spacing of more than 200 nm should be aimed at, or respectively a fringe spacing of 0.25 mrad in the diffraction plane. An experimental requirement would be to illuminate the specimen within ½ of a period of the biprism fringes in order to control the phase of the illuminating beam. We realized that a reduction of the fringe numbers leads to a higher coherent current available. To use this as an advantage it was proposed that small particles should be used as new specimens, which in themselves will facilitate the required illumination. Such particles should be accessible through other groups (e.g. U. Kaiser, University of Ulm (SmCo particles) or from the Duisburg SFB groups (FePt L10 particles).

Task 3 "Comparison with calculations": It consists of three steps: 1) calculation of partial charges including lms non-diagonal contributions; 2) tetrahedron integration of partial charges over the Brillouin zone; 3) calculation of mixed dynamic formfactors (MDFF). The programs for performing steps 1) and 2) are finished and ready for use and the program for step 3) is near to completion. All codes can take an advantage of the crystal symmetry to save the computing time. There were also made a number of other algorithmical optimizations to further reduce the time required for calculations while maintaining high numeric accuracy. Problems with the sum rule for spin and orbit moments occur in the beginning.

Preliminary results of the simulated dichroic effect on Nickel crystal were achieved. Within the given geometry conditions the switching of imaginary parts of MDFF's was demonstrated for two opposite positions of the detector on the Thales circle, although the calculated effect seems to be of smaller magnitude, than the experimental one. In the first run the sumrule for the dichroic signal was not fulfilled. The program has been further developed by adding user-friendliness and removing approximations to better fit the experimental conditions (i.e. integration over illumination and collection angle would remove the need of assuming parallel illumination or pointlike detector in the simulations). The sum rule is now fulfilled.