A nearly insulating film became an energy probe
A thin insulating film seems an unpromising way to study electrical conduction. Yet if the film is thin enough, electrons can tunnel through it, and the small current can reveal which electronic states are available on the other side. Giaever used this idea in 1960 to investigate superconductors. The barrier was useful precisely because it coupled the electrodes weakly enough to preserve information about their separate spectra.
The immediate question was whether the energy gap predicted by the recently developed BCS theory could be observed electrically. Resistance measurements could already track the superconducting transition, but they did not reveal the energies of its elementary excitations. Giaever’s experiment converted that invisible energy structure into a voltage-dependent current. To understand the result, three things must be kept together: barrier transmission, occupation of the starting states, and the availability of final states.
Sources: [1][5][2]
The junction had to preserve a barrier without breaking the circuit
The experimental arrangement placed conducting electrodes on opposite sides of a thin insulating region. A voltage source shifted their electrochemical potentials, and the current was measured as that voltage was swept. Cooling one electrode below its superconducting transition changed its excitation spectrum without changing the basic wiring of the experiment. This made the normal-state and superconducting-state curves valuable comparisons.
Fabricating a useful barrier was the central difficulty. If it was too thick, the current became too small to resolve. If metallic bridges or pinholes connected the electrodes, ordinary conduction could bypass tunneling. Merely observing current across a nominal insulator was therefore inconclusive. Giaever later recalled seeking a more decisive experiment after colleagues remained skeptical of tunneling. Combining the barrier with a superconductor provided a characteristic energy scale against which the current could be tested.
Sources: [1][5]
Voltage changes the overlap of occupied and empty states
At equilibrium the two electrodes have aligned electrochemical potentials. Their microscopic states can exchange electrons in both directions, but the opposing rates balance. Applying voltage creates an imbalance. For elastic tunneling, an electron keeps its total energy across the transfer, so bias determines which occupied states on one side can connect to empty states on the other.
A simple normal metal has an approximately smooth density of states over the small energy interval of interest. When both electrodes are normal and the barrier transmission varies slowly, the current is correspondingly smooth. Replacing one electrode by a superconductor changes the available states sharply near its gap. A bend or onset in the current can then carry spectroscopic information. This reasoning also explains why a barrier-transmission probability alone cannot predict the entire current: the reservoirs that supply and accept electrons are essential parts of the measurement.
There are also two different energy scales in the apparatus. The insulating barrier can be much higher than the millielectronvolt structure being probed in an electrode. Over a sufficiently small bias interval its transmission may then change only gradually, allowing the sharper electrode spectrum to dominate the curve. This approximation should be checked, not confused with setting the barrier height equal to the superconducting gap.
Sources: [1][2]
The missing states and the edge pile-up explain the curve
In the ideal isotropic BCS description, a superconductor has no quasiparticle states inside the gap, and a strong density-of-states feature at the gap edge. This relationship gives the density relative to a normal-state reference. Energy is measured from the chemical potential; the gap and energy must use the same units. Taking the real part yields zero inside the ideal gap.
A tunnel spectrum does not mean that electrons have vanished from the material. The gap concerns available quasiparticle excitations of the paired many-electron state. Near the edge, the mapping from ordinary electronic energies to quasiparticle energies concentrates states into a small energy interval. This produces a pronounced conductance feature. Real lifetime effects, inhomogeneity and instrumental resolution smooth the ideal singularity, so an infinitely sharp peak is a feature of the limiting formula rather than a demand on every experimental curve.
Sources: [2][1]
One superconducting electrode and two give different thresholds
For a normal-metal–insulator–superconductor junction at very low temperature, the onset corresponds approximately to one superconducting gap. With two superconducting electrodes, occupied and empty quasiparticle spectra must overlap across both gaps, giving a prominent threshold at their sum. The two experiments should therefore not be assigned the same voltage rule.
Consider an explicitly hypothetical pair of identical electrodes with a gap of two tenths of a millielectronvolt. A normal–superconducting junction would have an ideal onset near two tenths of a millivolt. With both electrodes superconducting, the corresponding sum-gap feature would lie near four tenths of a millivolt. This numerical example follows from electron charge converting voltage into energy; it is not a reconstruction of Giaever’s particular sample. Changing which electrode is superconducting changes the interpretation even when the external voltmeter is unchanged.
Sources: [1][3]
Temperature and defects leave distinguishable fingerprints
Finite temperature populates some quasiparticle states and blurs occupation boundaries. Even an excellent junction can therefore show current below a zero-temperature threshold. The measured current is a convolution involving the density of states and Fermi occupations, not a literal photograph of the density of states. Differentiating the current makes the spectral structure easier to see only under stated assumptions.
Leakage through pinholes, unwanted heating and an incorrect voltage offset can also spoil the apparent gap. Their effects can be investigated by temperature sweeps, normal-state comparisons and changes of measuring current. In a hypothetical sample with the gap used above, a few tenths of a kelvin already correspond to a thermal energy of several hundredths of a millielectronvolt. That scale is small but not negligible compared with the feature being resolved. Cooling and calibration are therefore parts of the spectroscopy.
Sources: [1][2]
The result tested a theory without exhausting superconductivity
The importance of Giaever’s work was the connection between a microscopic energy prediction and a sharply structured electrical measurement. The gap’s evolution with temperature could be compared with the theory and with independent thermal evidence. In his later account, Giaever emphasized that this combination made the tunneling interpretation more convincing than current through an insulating film by itself.
Quasiparticle tunneling turned applied voltage into an energy-selecting tool, making the gap inferred from thermal measurements visible in electrical curves. Changing temperature or which electrodes were superconducting moved the features in predictable ways. Josephson’s subsequent work on coherent pair transfer showed that a thin junction could also probe the phase of superconducting order.
Sources: [5][1][4]