Extended Data Fig. 2: The robustness of superconductivity in trained and untrained configurations.

a, dV/dI as a function of I_dc and ν_tTLG measured at B = 0 and D = − 100 mV/nm for the electron doped superconducting phase. The measurement is performed with the superconducting diode effect after field training (left panel), and without the superconducting diode effect after ‘un-training’ with a large DC current (right panel). b, The reciprocal (top panel) and non-reciprocal (bottom panel) component of the critical current, \(({I}_{c}^{+}+{I}_{c}^{-})/2\) and ΔI_c, as a function of ν_tTLG extracted from a. It is worth noting that several experimental works have reported interplay between DC current flow and the sign of magnetic order: in orbital ferromagnetic states, a large DC current is shown to induce sign-reversal in the magnetic order ^33,34. It is hypothesized that the mechanism underlying current-induced switching stems from the interaction between different magnetic domains and current flow around the edge of the domain. Our observation that a large DC current couples to the underlying time-reversal symmetry is consistent with previous experimental results. However, the sample interior of an orbital ferromagnet is insulating and current flows along the edge of the magnetic domain. Whereas the sample interior in the nonreciprocal superconducting phase is highly conductive. As such, we anticipate the interplay between DC current and the underlying time-reversal symmetry breaking to be different. Notably, how DC current interacts with the magnetic order remains an open question for graphene moiré systems in general ^33,34.

Source data