Transition voltage spectroscopy: a challenge for vacuum tunneling models at nanoscale
Several studies on the transition voltage ($V_t$) are based on calculations of the tunneling current within the Simmons model. In this paper devoted to vacuum nano-junctions, we compare the Simmons results for $V_t$ with those obtained from the exact Schrodinger equation by exactly including the classical image effects. The comparison reveals that the Simmons estimates for $V_t$ are completely unacceptable for nanogap sizes ($d$) at which image effects are important. The Simmons treatment drastically overestimates these effects, because it misses the famous 1/2 factor related to the fact that the image interaction energy is a self energy. The maximum of the Simmons curve $V_t$ vs. $d$ turns out to be merely an artefact of an inappropriate approximation. Unlike the Simmons approach, the "exact" WKB method yields results, which qualitatively agree with the exact ones; quantitative differences are important, demonstrating that the transmission prefactor has a significant impact on $V_t$. Further, we show that a difference between the work functions of the left and right electrodes, which gives rise to a Volta field, may be important for the ubiquitous asymmetry of the measured $I$-$V$-characteristics in general, and for the different $V_t$-values at positive and negative biases reported in vacuum nanojunctions in particular. The weak dependence $V_t = V_t(d)$ found experimentally in vacuum nanojunctions contrasts to the pronounced dependence obtained by including the exact electrostatic image contribution into the Schrodinger equation. This demonstrates that not only the molecular transport, but also the transport through a vacuum nanogap represents a nontrivial problem, which requires further refinements, e.g., a realistic description of contacts' geometry, retardation effects due to the finite tunneling time, local phonons, surface plasmons or electron image states.
