Coherent Light with Tunable Polarization from Single-Pass Free-Electron Lasers

Tunable polarization over a wide spectral range is a required feature of light sources employed to investigate the properties of local symmetry in matter. In this Letter, we provide the first experimental characterization of the polarization of the harmonic light produced by a free-electron laser and demonstrate a method to obtain free-electron laser harmonics with tunable polarization. Experimental results are successfully compared with theory. Our findings can be expected to have a deep impact on the design and realization of experiments requiring full control of light polarization.

Figure 1

Schematic layout of a helical (APPLE-type) undulator. When the magnetic arrays are aligned, the magnetic field describes a sinusoid confined in the $(y,z)$ plane (dotted line). The polarization of the emitted light is in this case linear horizontal. Circular polarization is obtained by shifting the magnetic arrays so as to make $B_x=B_y$ and the phase between the field components equal to ${\lambda }_u/4$. In this case, the total magnetic field describes a helix (continuous line).

Figure 2

Scheme of the experimental setup used to measure the polarization of the light emitted by the radiator.

Figure 3

Fundamental synchrotron emission at 195 nm. Upper panel: Normalized Stokes’ parameters $M/I$ (circles) and $S/I$ (squares) as a function of the ratio $B_x/B_y$. Lower panel: On-axis intensity (normalized to the intensity recorded when the polarization is linear horizontal), as a function of the ratio $B_x/B_y$. Relative errors are estimated to be about 10%.

Figure 4

Nonlinear harmonic emission at 195 nm. Upper panel: Normalized Stokes’ parameters $M/I$ (triangles, experiments; dashed line, theory) and $S/I$ (circles, experiments; continuous line, theory), as a function of the ratio $B_x/B_y$. Lower panel: On-axis intensity (circles, experiments; continuous line, theory), normalized to the intensity recorded when the polarization is linear horizontal, as a function of the ratio $B_x/B_y$. Relative errors are estimated to be about 15%.