Experimental studies on polarization correlations in hard x-ray Rayleigh scattering
This thesis investigates experimentally the elastic scattering of hard x-rays. Combining the novel technologies of a third-generation synchrotron radiation source and a Si(Li) strip detector which acts as a highly efficient x-ray Compton polarimeter allows to measure the linear polarization of the elastically scattered photons for a highly linearly polarized incident beam. Here, such a polarization transfer is considered for the first time in the hard x-ray regime. With a photon energy of 175 keV and gold as scatterer, a highly relativistic regime is chosen where Rayleigh scattering is the only significant elastic scattering contribution. In addition to the polarization of the elastically scattered photons, also the angular distribution is measured. The data are compared to fully relativistic second-order QED calculations. Both observables are well described by these predictions whereas the form factor approximation fails. The simultaneous measurement of angular distribution and polarization allows to identify spurious agreement of the form factor theory in only one observable. At scattering angles around 90°, the assumption that the incident beam is completely linearly polarized is not sufficient to explain the data. The measured linear polarization of the Compton-scattered photons is used to obtain an independent estimate for the incident beam polarization of about 98 % which leads to an agreement between experiment and theory at all measured data points. The significant change introduced by this depolarization of 2 % indicates a strong sensitivity on the polarization of the incident beam. In the present experiment, this sensitivity limits the precision, but on the other hand, it allows a precise reconstruction of the incident beam polarization when the theory is established. Here, such a reconstruction is performed and the result agrees with the 98 % from the Compton polarization, but with a slightly lower uncertainty and with less statistics.