Zero Bias Anomalies in Tunneling: A Review

Abstract

The measured conductance σ (V) of a tunnel junction between two metals may depart from the simple idealization of a constant σ₀, independent of applied bias voltage V for several more or less basic reasons, which will be the topic of this review. First, a nearly temperature independent quadratic increase in conductance σ = σ₀ + σ₁ (V−V₀)² is often observed, and may reasonably be ascribed to the distortion of a rectangular or trapezoidal insulating tunnel barrier with variation of V. An offset V₀ of the structure may arise from a built-in electric field across the barrier. Such an effect is almost an intrinsic feature, therefore, of a tunnel junction. Two other frequently observed temperature-dependent structures which occur precisely at V = 0, on the other hand, are evidence for tunneling mechanisms additional to direct elastic tunneling and are associated with additional localized electron states in or near the barrier. A “giant resistance peak,” or conductance minimum has convincingly been ascribed to two-step tunneling across the barrier via real intermediate states occurring on small metal particles imbedded in the barrier. Variations of this effect have been seen in other contexts, including that of metal-semiconductor tunneling. The narrower and weaker conductance peak can be attributed to magnetic scattering of tunneling electrons from paramagnetic impurities at the edge of the barrier, effects which may be described as Kondo and spin-flip scattering across the tunneling barrier. Finally, the effect of a narrow conduction band in lead­ing to an asymmetric tunnel conductance will be briefly described.