We first rule out the possibility that the observed effects arise from depinned magnetic vortices. First, all experiments were conducted under zero-field cooling conditions, with the ambient magnetic field reduced to below 1 μT using dual magnetic shielding. Such conditions ensure that no pre-existing magnetic vortices are present within the sample area (width ≈ 10 μm × length ≈ 30 μm). Second, even when neglecting the finite time required for vortex entry and depinning under picosecond current drive, the maximum displacement of a vortex would be less than a few tens of nanometres, given maximum vortex velocities on the order of a few tens of kilometres per second^{7,19,20,21,22,23}. Full vortex penetration is therefore implausible. This explains why vortex-induced resistive responses in superconductors typically manifest only on nanosecond timescales^17,18,19. Third, the distinct responses observed in the s-wave superconductor NbN and the d-wave superconductor YBCO suggest that the observed picosecond suppression of superconductivity is closely related to the microscopic gap symmetry.

Taken together, these three considerations support the conclusion that the observed depairing current density J_c* in the s-wave type-II superconductor NbN originates from the competition between quasiparticle energy shift and the superconducting gap. In superconductors with an anisotropic gap in momentum space, such as d-wave superconductors, J_c* is not a well-defined quantity and is therefore not observable in the ultrafast transport measurements on YBCO. We note that, the monocrystallinity of both NbN and YBCO thin films (Supplementary Information, section 1) is important here. In polycrystalline superconductor thin films, the phase slip dynamics at domain boundary dominate the sample’s nonlinear response²⁷ and impede the observation of intradomain depairing processes.

Figure 5a compares the conventional critical current density J_c and the intrinsic depairing current density J_c* as functions of temperature in NbN (see raw data in Supplementary Information, section 7). As the temperature increases, J_c decreases much faster than J_c*; at 0.8 × T_c, J_c is nearly an order of magnitude smaller than J_c*.

Because J_c* is fundamentally determined by the microscopic properties of the s-wave superconductor, it can be calculated directly from microscopic parameters. Assuming at J_c*, the quasiparticle energy shift \(\hslash {{\bf{k}}_{\rm{F}}} \cdot {{\bf{v}}_{\rm{s}}}\) equals the superconducting energy gap Δ, yielding

where m is the effective mass of the electron and ξ is the superconducting coherence length.

The depairing current density is then given by

where n_s is the superfluid density. By substituting the temperature-dependent functions n_s(T) and ξ (T) (Supplementary Information, section 9), the calculated temperature dependence of J_c*(T) shows excellent agreement with the experiments, with the experimentally measured values slightly smaller than the calculated ones. This is possibly because the depairing process partially starts before the quasiparticle energy shift \(\hslash {{\bf{k}}_{\rm{F}}} \cdot {{\bf{v}}_{\rm{s}}}\) reaches the superconducting energy gap Δ due to thermal fluctuations. Details of the calculation are provided in Supplementary Information, section 9.

We simulated the depairing process in the s-wave superconductor NbN, using both time-dependent Ginzburg–Landau (tDGL) theory and microscopic Bardeen–Cooper–Schrieffer (BCS) theory in the dirty limit relying on the Usadel equation³⁰. The microscopic theory directly takes parameters from experimental characterizations of the sample (see details in Supplementary Information, section 8), apart from Dynes broadening, which is set to guarantee numerical convergence and to reproduce the experimental results. In general, the sharp transmission drop is attributed to the strong-nonlinearity of the superfluid density, which depends on the s-wave nature and strong disorder of the system. tDGL cannot capture the strong nonlinearity necessary to model this behaviour, as it relies on expanding the free energy in powers of the order parameter and is only valid near T_c (refs. ^31,32). This consistency further confirms that the observed sudden change in the sample’s response originates from the rapid depairing of Cooper pairs in s-wave superconductors. We note that the BCS-based simulations do not fully reproduce the tail behaviour as the response approaches the normal state. This discrepancy suggests the involvement of additional relaxation channels, such as phonon interactions, which is not included in the current model and require further theoretical investigations (see Supplementary Information, section 11 for discussions). For YBCO, constructing a fully microscopic description of the dynamics under strong current pulses remains challenging, as a quantitative microscopic theory for cuprate superconductors is not yet established even in equilibrium. We therefore use a tDGL-based approach to simulate the qualitative features of the dynamics, which is discussed in Supplementary Information, section 11.4.

In summary, we have reported ultrafast electrical transport measurements in which vortex motion and subsequent self-heating are bypassed—thereby enabling the extraction of intrinsic microscopic properties of type-II superconductors. The findings reported here also offer an approach to probing the gap symmetry in systems that cannot be directly accessed using spectroscopic techniques. From an applications perspective, our results further reveal how maximum supercurrents in a type-II superconductor can be reached on a picosecond timescale, providing a potential platform for generating ultrashort, strong magnetic field pulses.