Fig. 1. ZnTe crystal orientation for arbitrary angle (a), standard/balanced configuration (b), and DEOS configuration (c).

Fig. 2. Transfer function values in absolute units for different chirp parameters and different frequencies.

Fig. 3. Simulated spectral datasets for different polarization configurations. The top left shows the expected signals under the standard configuration with α=π4 and the bottom left shows the classic DEOS configuration. The zeros of the transfer functions (equal in these two configurations) are indicated with dotted lines. On the right the simulated DEOS polarization readout with THz polarized at α=π4 is used to showcase the difference between a correct reconstruction resulting from the knowledge of the incoming THz polarization and a traditional DEOS reconstruction.

Fig. 4. Simulated datasets with the actual temporal waveforms displayed on the top and noted with A and measured waveforms that are modulated by the transfer functions on the bottom noted with M.

Fig. 5. Simulated refractive index retrieval with real part on the left and imaginary part on the right. Ext. represents the analytical extraction and Fit is the numerically fitted values. The zeros of the transfer functions are presented in vertical dashed and dotted lines to highlight the frequencies where information cannot be accurately retrieved.

Fig. 6. THz source experimental setup. HWP and QWP refer to a half- or quarter-waveplate and Pel-BS is shortened for pellicle beamsplitter. The red path represents the IR pump path that is pulse front tilted by the grating and two lenses noted as “f” that can be adjusted in position when varying the pulse compression. The Mg:LiNbO3 crystal converts IR to THz that is then collimated by a Teflon lens and focused down to the ZnTe crystal by a TPX lens. The space in between the two lenses is used for a wiregrid polarizer and/or a sample. The orange path represents the probe path for EOS that is split off and recombined with the THz radiation by pellicle beamsplitters. After the THz-induced modulation, the probe is propagated through polarization optics before being split into horizontal and vertical polarization and then spectrally analyzed. A portion of the probe is split off for a reference line that is not modulated by THz.

Fig. 7. Measured and reconstructed signals for different wiregrid polarizer orientations. In the top left are the temporal profiles under the standard configuration polarization readout. Γx and Γy refer to the measured polarization components that are modulated by the transfer functions. Γ refers to the reconstructed waveforms with a DEOS-based measurement and reconstruction as comparison. We note that for the 0° polarization angle in the standard configuration, no accurate reconstruction can be performed as the Γx component is not captured. The top right shows the spectral amplitudes of the individual polarizations for different wiregrid orientations with the transfer functions highlighting the loss of information. The bottom left shows the peak amplitude of Γx and Γy with theoretical expectations for different wiregrid orientations. The bottom right shows the reconstructed spectra with a DEOS dataset as a comparison. The change in amplitude here is due to the filtering effect of the wiregrid polarizer when it is oriented at non-zero angles in respect to the y-axis.

Fig. 8. Real part of the ordinary and extraordinary refractive indices no and ne from y-cut LiNbO3. Three different types of measurements are portrayed: Ref. denotes the refractive indices retrieved through two separate measurements by orienting the main axes of the sample to the THz polarization and then fit the index ellipsoid; Ext. refers to the single-measurement technique in which the sample is oriented 45° to the THz polarization, allowing both refractive indices to be sampled simultaneously with the cost of losing information at the zeros of the transfer function; Fit refers to the same setup as Ext. but uses a fitting algorithm to reconstruct the index ellipsoid. The vertical dotted lines are the zeros of the transfer function to visualize where information cannot be analytically retrieved under the single-measurement configuration. The spectral amplitude of the THz signal without the sample is shown in the bottom in green to show how refractive index extraction errors are correlated with low spectral amplitude.