Electrically controllable superconducting memory effect in UTe₂

Now we turn to considering the likely microscopic origin of the memory effect in UTe 2 . First, given that we are applying substantial electrical current pulses at dilution fridge temperatures, we should consider any extrinsic heating effects that might be present. In the Supplementary Information, we plot the measured temperature as a function of time during the current pulse measurement protocols and present further discussion as to why we do not believe that Joule heating effects greatly affect our measurements. One strong argument against Joule heating artefacts comes from the sensitivity of the memory effect to the relaxation rate of the dc stimulus. In Fig. 2, we showed that stronger memory effect correlates with larger amplitude and duration tuning. For both of these, the larger stimulus correlates with more energy deposited into the system (through lead/contact resistances). However, the memory effect is also strongly correlated with faster relaxation rates α, which anticorrelates with the total energy transferred into the system. If the memory effect were because of some relatively straightforward Bean–Livingston barrier or Bean model pinning effect process46, this should be insensitive to how the current is turned off, depending only on the total energy imparted. We note, however, that it is therefore hard to rule out the possibility of thermal quenching playing a role. We posit that the acute sensitivity of the memory effect to the turn-off speed implies some dynamic reorganization or annealing of vortex domains, probably because of some interesting intrinsic physical property (or properties) of UTe 2 .

Further arguments against extrinsic heating effects originate from the fact that this phenomenon is only observed in the narrow regime of B–θ space directly underneath SC2, with no hysteretic V(J) phenomena observed in purely SC1 or SC2. The memory effect is most pronounced at the lowest temperatures and quickly diminishes on warming (Fig. 3), providing perhaps the strongest argument against any heating-induced artefacts. This is in sharp contrast to the case of NbSe 2 , in which hysteretic V(J) profiles are enabled close to T c (ref. 23). Instead, our observations strongly indicate that the hysteretic V(J) behaviour of UTe 2 is because of some intrinsic low-energy phenomenon, with an energy scale much lower than T c .