Terahertz range polarization rotation in the candidate time-reversal symmetry breaking superconductor BiNi Ralph Romero III, Zhenisbek Tagay, Jiahao Liang, Jason Y. Yan, Di Yue, N. P. Armitage Abstract Here we report the observation of time-reversal symmetry (TRS) breaking superconductivity in a BiNi bilayer using terahertz (THz) polarimetry. Leveraging a novel high-precision THz polarimetry technique, we detect, in the superconducting state and at zero magnetic field, the smallest polarization rotation of THz light measured to date. By using the MgO substrate itself as an optical resonator, we can reference the Faraday and Kerr rotations to each other. We observe a low-frequency Kerr rotation on the order of several hundred microradians in the superconducting phase, a clear signature consistent with TRS-breaking superconductivity. Our measurements enable direct access to the THz-range Hall conductivity. Through a Kramers-Kronig analysis, we link these low-energy measurements to prior high-frequency magneto-optic Kerr effect (MOKE) data. This connection provides critical insight into the nature of the TRS-breaking state, supporting a multiband superconducting scenario over a disordered single-band interpretation for the origin of the Kerr effect. Chiral superconductors Kallin and Berlinsky 2016 that break time-reversal symmetry (TRS), have been the subject of intense current study as, aside from fundamental interest, they have been proposed to host Majorana zero modes, a promising platform for topological quantum computation Kitaev 2003 ; Sarma et al. 2015 . Sr 2 RuO 4 was a leading candidate to host such chiral superconductivity Sato and Ando 2017 , however recent results are incompatible with a p_{x}\pm ip_{y} type order parameter leaving the form of its superconducting order parameter an open question Mackenzie et al. 2017 ; Yaguchi et al. 2002 ; Mao et al. 2000 ; Taniguchi et al. 2015 ; Pustogow et al. 2019 . Moreover, its T c is fairly low, \sim 1.5 K, making studies and application challenging. In 2015, epitaxially grown BiNi bilayers were shown to superconduct at 4 K Gong et al. 2015 . This was surprising as neither of the constituent materials exhibit superconductivity by themselves at these temperatures or pressures. It was shown shortly thereafter using a Sagnac interferometer to measure the magneto-optical Kerr effect (MOKE) that the superconducting state of the BiNi bilayer broke TRS. It was claimed to be an even parity, d_{xy}\pm d_{x^{2}-y^{2}} , order parameter, due to the fact that superconductivity was believed to be occurring at the Bi surface Gong et al. 2017 . This assignment was later disputed with the observation that SC appeared to occur throughout the bulk of the film Chauhan et al. 2019 ; Wang et al. 2023 . Much like Sr 2 RuO 4 , consensus on the symmetry of the order parameter in BiNi remains lacking to this day. Two leading methods of directly detecting TRS breaking superconductivity are \mu SR and MOKE Wysokiński 2019 ; Ghosh et al. 2020 . These are powerful probes, with MOKE via Sagnac interferometry possessing a remarkable sensitivity on the order of tens of nanoradians. However, both are limited in the sense they are single frequency and spatially local, making them difficult to cross check with theory and possibly misinterpretable as arriving from magnetic impurities. Furthermore, the energy of the optical or infrared photons ( \sim 1 eV) used in MOKE are far from typical superconducting gap energies ( \sim 1 meV or less in many discussed systems). The mismatch of the photon energy in conventional MOKE to the natural energy scale of superconductivity inherently requires high sensitivity as the expectation is that when superconductivity onsets the majority of the spectral weight rearrangement will occur at much lower optical frequencies. With this in mind, it would be beneficial to have a probe of TRS breaking with lower photon energies in order to get spectroscopic information on the most relevant energy scales. In this Letter, we report the detection of TRS breaking SC in a BiNi bilayer via THz range polarimetry. By using a novel scheme for high precision THz polarimetry we observe in the superconducting state of BiNi at zero magnetic field the smallest rotation of the polarization of THz light to date. By using the MgO substrate as an optical resonator, we find low frequency Kerr rotations on order of a few hundred \mu radians in the superconducting state. This is consistent with TRS breaking superconductivity. Our experiment enables direct measurement of the THz-range Hall conductivity. Through a Kramers-Kronig analysis we can connect the Hall effect (measured via previous MOKE) measured at eV frequencies to this data, which is measured much closer to the gap energy scale. With this perspective we show signatures of the TRS breaking state in BiNi is more likely to originate from a multiband scenario than a dirty single band case. Figure 1: a),b) Temperature dependence of the real and imaginary parts of the conductivity measured using TDTS. A clear superconducting transition can be seen near 3.8 K. c), d) Magnetic field dependence of the complex conductivity. Complete suppression of superconductivity can be seen around 20 kG. This upper critical field is consistent with superconductivity coming from the Bi/Ni bilayer as opposed to NiBi 3 . A 10 nm thick rhombohedral Bi(110) layer was epitaxially grown on a 1 nm Ni(100) layer at 110 K, which is seeded at 300 K on a 0.5 mm thick MgO(100) substrate. Conventional time-domain THz spectroscopy (TDTS) was performed on this sample to measure the longitudinal complex conductivity, \tilde{\sigma}(\nu) . By measuring \tilde{\sigma}(\nu) in the THz range while varying temperature or out-of-plane magnetic field, one can clearly identify the superconducting transition temperature, T_{c} , and the upper critical field at which SC is destroyed, H_{c2} . Figs. 1 a) and b) show the real, \sigma_{1}(\nu) , and imaginary, \sigma_{2}(\nu) , parts of the conductivity from 0.2 - 2.0 THz at temperatures from 7 to 1.6 K. At 7 K, \sigma_{1}(\nu) exhibits a slightly decreasing frequency dependence while \sigma_{2}(\nu) is slightly increasing with frequency. This is indicative of a dirty metal Drude-like response with a scattering rate somewhat larger than our frequency window. As discussed in previous work, the normal state is a disordered metal whose electrodynamic response looks like neither Bi or Ni alone Chauhan et al. 2019 . As temperature is lowered, the onset of superconductivity is signaled by spectral weight in \sigma_{1}(\nu) being pushed to a delta function at zero frequency and a corresponding 1/ \nu behavior developing in \sigma_{2}(\nu) Tinkham 2004 . T_{c} can thus be seen to be around 3.8 - 3.9 K. This is consistent with Ref. Chauhan et al. 2019 who measures T_{c} = 4.15 K and \Delta(1.6K) = 0.67 meV. Figs. 1 c) and d) show the real and imaginary conductivities at 1.6 K as we vary an external out-of-plane magnetic field. At zero field, the optical signatures of superconductivity are clear. As the field is ramped up, superconductivity is slowly suppressed and the frequency dependence begins to resemble that of higher temperatures, with superconductivity being completely killed at 20 kG. An H_{c2} of this scale is consistent with the optical response being dominated by the BiNi bilayer and not NiBi 3 which could have been formed inadvertently at the interface Shang et al. 2023 ; Wang et al. 2023 . Figure 2: The operational principle of our polarization resolved time-domain THz spectrometer. A WGP is placed at [formula] relative to the propagation direction of the THz pulse, its wires oriented at [formula] relative the initial pulse polarization. This makes the WGP a polarizing beam splitter, passing the vertical component and reflecting the horizontal component of the elliptically polarized, transmitted THz pulse. The scheme and its calibration is discussed extensively in Ref. Tagay et al. 2024 . Next we investigate the possibility of TRS breaking in the superconducting state using a novel scheme for high-precision THz polarimetry Tagay et al. 2024 . Upon being transmitted through (reflected from) a sample which breaks TRS, the initially linearly polarized THz pulse is expected to emerge elliptically polarized and rotated due to the magneto optical Faraday (Kerr) effect. We use a wire grid polarizer (WGP) as a polarizing beam splitter after the sample in order to resolve the orthogonal components of the transmitted pulse. One polarization is transmitted through the WGP and one is reflected, each towards a separate detector. In this fashion \tilde{E}_{x}(\nu) and \tilde{E}_{y}(\nu) are measured simultaneously and the complex rotation can be extracted via the relation, \theta_{K,1}+i\theta_{K,2}=\frac{\tilde{E}_{y}(\nu)}{\tilde{E}_{x}(\nu)} , with a precision of better than 20 \mu rad ( \approx 1 millidegree) Tagay et al. 2024 . Figure 3: a) Our referencing scheme b) Raw (before antisymmetrization) real part of the Kerr rotation, \theta_{K,1} , from 0.3 to 1.0 THz for the B_{train,1}=+75G (solid) and B_{train,2}=-75G (dashed) at 1.6 and 7 K. b) Antisymmetrized \theta_{K,1} for the same experiment. A clear inverse dependence on frequency can be seen in \theta_{K,1} at 1.6 K, when the sample is in the superconducting state. Figure 4: a) Real part of the Kerr rotation, \theta_{K,1} , from 0.3 to 1 THz at 1.6 and 7 K for the five separate experiments in Table 1. The bold curves are the averages of the five experiments with the error bars being one standard deviation. b) Imaginary part of the Kerr rotation, \theta_{K,2} , for the separate experiments. The insets show the same quantities for B_{train,1/2}=\pm 75G on bare Bi (10 nm) and bare Ni (1nm). No onset of \theta_{K,1} can be seen at 1.6 K in either film. c) Average of the real part of the Hall conductivity of the five experiments computed from Eq ( 1 ) using \tilde{\sigma}_{xx} in superconducting and normal state from Fig 1 and \tilde{\theta}_{K} from above. d) Imaginary part of the Hall conductivity computed from Eq ( 1 ) In order to achieve such precision, the two detectors must be calibrated such that any difference in path length and spectral response are accounted for, a reference electric field must be defined, and the measured electric fields must be antisymmetrized with respect to magnetic field. Typically the reference is taken to be the pulse transmitted through a bare substrate, however moving the sample stage vertically introduces considerable irreproducibility when the signal is on the order of the precision of the spectrometer. In order to meet this challenge, we have used the MgO substrate as an optical resonator to measure the Kerr rotation of the first reflected THz pulse using the initial transmitted pulse’s Faraday rotation as the reference (see Fig 3a). As the pulse is first incident on the metallic thin film, the transmitted part incurs a Faraday rotation as it passes through the TRS breaking film. We use this as our reference. A part of this initial pulse which is reflected by the sample feels a Kerr rotation, is reflected off the back of the substrate and transmitted through the sample picking up a Faraday rotation similar to the first pulse. Thus by subtracting the rotation of the first pulse from the rotation of the second pulse we are left behind with a measure of the complex Kerr rotation, \tilde{\theta}_{K}^{film}(\nu) . We use this high precision THz polarimetry to characterize TRS breaking in this material. As is customary in the diagnosis of TRS breaking states of matter, a small magnetic training field is employed to ensure a monodomain sample. A small external magnetic field ( B_{train}0 and \sigma_{xy,2}<0 . In the normal state no appreciable Hall conductivity is evident. The above analysis further corroborates the existence of TRS breaking in the superconducting state but is this scale of the Hall conductivity consistent with previous MOKE experiments Gong et al. 2017 ? As discussed in the SM, from general Kramers-Kronig considerations one expects that the high frequency \sigma_{xy,1} goes like (\delta/\nu)^{2} down to a cutoff energy scale \delta that reflects either the superconducting gap (2 \Delta ) or the scale of interband transitions ( E_{g} ). The expectation is then that the Hall conductivity at high frequencies goes as \sigma_{xy,1}(h\nu\gg E_{g})\sim\sigma_{xy,1}(0)(\frac{E_{g}}{h\nu})^{2} . To extract the high frequency Hall conductivity from existing Kerr data. we performed a transmission experiment on our sample over a large spectral range. \theta_{K}^{surface} from light reflected off a semi-infinite sample is related to \tilde{\sigma}_{xy} via the equation Okamura et al. 2020 \tilde{\theta}_{K}^{surface}(\nu)=-\frac{\tilde{\sigma}_{xy}}{\tilde{\sigma}_{xx}\sqrt{\tilde{\epsilon}_{xx}}}, (2) where all quantities are complex and its been assumed again that the Hall conductivity is much smaller than the longitudinal conductivity. From our transmission experiment (SI Figs. 3 and 4), \sigma_{xx}(h\nu=0.8eV)\approx 11.3+i0\ (m\Omega^{-1}cm^{-1}) , \epsilon_{xx}(h\nu=0.8eV)\approx 2+i25 and \tilde{\theta}_{K}(h\nu=0.8eV)=80+i0 nrad from Gong et al. 2017 . We can plug these values into Eq. ( 2 ) and solve for \sigma_{xy,1} , giving us \sigma_{xy,1}(h\nu=0.8eV)\approx-3.33\times 10^{-6}\ (m\Omega^{-1}cm^{-1}) . Finally we can use our derived expression \frac{\sigma_{xy,1}(\nu\gg 1)}{\sigma_{xy,1}(\nu=0)}\sim-(\frac{E_{g}}{h\nu})^{2} to extract an energy scale, E_{g} . Reading off the value of \sigma_{xy,1}(h\nu\ =1.2\ meV\sim 0)=0.05\ m\Omega^{-1}cm^{-1} from Fig. 4 c) gives us a value of E_{g}\approx 7 meV. This is obviously too large to correspond to the superconducting gap of BiNi ( \approx 0.7 meV), but may correspond to the inverted bandgap at the L point of Bi of \approx 15 meV Aguilera et al. 2015 . Thereby we can conclude that there is possible rough correspondence between the Kerr rotation data at 0.8 eV and our THz data at meV scales. Moreover, since our extracted E_{g} is an order of magnitude larger than the superconducting gap, it is most likely that this TRS breaking superconducting state originates from a multiband scenario as alluded to in the second scenario proposed in the SI. This is corroborated by the more complicated form for the real and imaginary Hall conductivities found in Fig. 4. More explicit calculations of the Hall response are needed for possible TRS breaking superconducting order parameters BiNi bilayers. We would like to thank V. Yakovenko for helpful conversations, and P.-C. Xu and X. Jin for preparation of the BiNi film. Measurements at JHU were supported by the ARO MURI “Implementation of axion electrodynamics in topological films and device” W911NF2020166. The instrumentation development at JHU that made these measurements possible was supported by the Gordon and Betty Moore Foundation EPiQS Initiative Grant GBMF- 9454. The data that support the findings of this article are openly available Ralph Romero III, Zhenisbek Tagay, Jiahao Liang, Jason Y. Yan, Di Yue, N. P. Armitage 2025