Broadband Non-Reciprocal THz Isolator
A Faraday isolator is an electromagnetic non-reciprocal device. It is required to shield electromagnetic sources against the detrimental effects of back-reflected light. Although it was proposed by Rayleigh  and demonstrated in various spectral regimes, it has been so far absent from the THz regime due to the limitations on handing the associated broad bandwidth. The THz isolator requires a broadband non-reciprocal phase retardation of 45º. This has always been extremely challenging to achieve over a wide bandwidth. We have recently demonstrated an isolator based on a magnetic non-reciprocal retarder operating with broadband THz pulses. Our device depends on permanently magnetized Strontium Iron Oxide (SrFe12O19), where Faraday rotations up to (194º/T) were obtained .
In comparison with previous studies on THz Faraday rotations , our sample has three main advantages that allowed for the realization of an isolator. First, it has a ferromagnetic resonance in the sub-THz band ~50-60 THz , i.e. the rotation is flat in the THz region (above the resonance). Second, it belongs to the class of insulating ferromagnets. This means it shows similar magnetic properties to its conducting counterparts, the conductivity being reduced by many orders of magnitude. This leads to very low absorption, which is a fundamental requirement for isolators. Third, it is a permanent magnet, which allows for its use without the need for an external magnetic field bias.
To build the isolator, we magnetized the sample to give a rotation of 45º upon single propagation, i.e. the reflected pulse experienced a 90º rotation and got blocked by a 0º-aligned polarizer. Figure 1(a) shows a schematic diagram of the back-reflection measurement setup that was used to characterize the isolator. The sample was placed between two polarizers. As for a typical isolator configuration, WGP4 ensured that the horizontal polarization of both the generated and the detected THz and WGP5 were aligned to 45º. The THz reflected off the mirror is detected in a ZnTe crystal by electro-optical sampling. Figure 1(b) shows the phase retardation map where the back-reflected THz is plotted for different levels of magnetic field. Two cases are presented in Fig. 1(c), corresponding to the sample providing 0º and 45º rotations. As shown, in the case of an unmagnetized sample (0º), the THz is reflected back from a mirror towards the source. On the other hand, the latter case demonstrates a fully functioning isolator where the back-reflection is forbidden. The demonstrated isolator is just a specific case of a general THz non-reciprocal phase retarder. Thanks to the large achievable retardations, by tuning the applied magnetic field, we managed to even reach phase reversal of the input field as shown in the phase retardation graph (Fig. 1(b)).
In conclusion, we demonstrated a fully functional THz isolator. Our device is capable of shielding back-reflected THz in the range (0.08-0.8) THz. We believe that our result will pave the way to a new era of non-reciprocal THz devices, whether operating alone or coupled to other reciprocal systems.