Normal-superconducting phase transition mimicked by current noise

As a superconductor goes from the normal state into the superconducting state, the voltage versus current characteristics at low currents change from linear to nonlinear. We show theoretically and experimentally that the addition of current noise to nonlinear voltage versus current curves will create ohmic behavior. Ohmic response at low currents for temperatures below the critical temperature $T_c$ mimics the phase transition and leads to incorrect values for $T_c$ and the critical exponents $\nu$ and $z$. The ohmic response occurs at low currents, and will occur in both the zero-field transition and the vortex-glass transition. Our results indicate that the transition temperature and critical exponents extracted from the conventional scaling analysis are inaccurate if current noise is not filtered out. This is a possible explanation for the wide range of critical exponents found in the literature.

Figure 1

Two sets of $I\text{−}V$ curves for a $2100\text{−}\mathrm{\AA }$-thick film with bridge dimensions $8\times 40\mu {\mathrm{m}}^2$. The solid lines are isotherms taken with low-pass $\pi$ filters ($3\mathrm{dB}$ at $4\mathrm{kHz}$) at the screened room wall and low-pass double-$\mathrm{T}$ filters ($3\mathrm{dB}$ at $2\mathrm{kHz}$) at the top of the probe. The dashed lines are isotherms taken without filtering. The isotherms are separated by $200\mathrm{mK}$, and the error bars are shown (when larger than the lines). The highest-temperature isotherms $\left(91.7\mathrm{K}\right)$ are ohmic, and fully in the normal state. In the transition region, we see (especially in the isotherms at $91.1$ and $90.9\mathrm{K}$) that noise creates ohmic tails in nonlinear signals.

Figure 2

Three sets of $I\text{−}V$ curves for one sample. The solid and dashed lines are the same isotherms from Fig. 1. The dotted-dashed lines are isotherms taken with $\mathrm{T}$ filters and copper-powder filters at the cold end of the probe ($3\mathrm{dB}$ at $70\mathrm{kHz}$). The isotherms are separated by $200\mathrm{mK}$ and the error bars are suppressed for clarity. We can see that a $3\text{−}\mathrm{dB}$ point of $70\mathrm{kHz}$ is not low enough to filter the isotherm completely.