Published tunneling results of Binnig et al interpreted as related to surface superconductivity in SrTiO3

TL;DR

This paper reinterprets tunneling results on SrTiO3, suggesting surface superconductivity rather than bulk two-band superconductivity, supported by fitting models to experimental data with adjustable parameters.

Contribution

It offers an alternative explanation for tunneling results in SrTiO3, emphasizing surface layer effects over bulk two-band superconductivity, with detailed modeling and data fitting.

Findings

Surface layer superconductivity explains tunneling results.

Best fit phonon energy is 21 meV.

Large ratio of mean-field to actual Tc suggests additional band effects.

Abstract

In 1980 Binnig et al. reported tunneling measurements on Nb-doped SrTiO3, and interpreted their results as indicating two-band superconductivity in the bulk of SrTiO3. However: (1) Effective masses determined from tunneling results in the normal state by Sroubek in 1969 and 1970 are much smaller than those determined by most other methods. The much smaller masses were attributed to properties of a surface layer by the present author in 1971; (2) The only other reports of two-band superconductivity in bulk SrTiO3 can be used to infer much smaller values of band separations than found by Binnig et al.. In this paper we give an alternative explanation of the results of Binnig et al. in terms of superconductivity in a surface layer. We obtain fair fits to the band gaps versus Fermi energy for the two bands in three samples where two surface subbands are occupied and to the temperature dependence of the gaps in one crystal, using a model with three adjustable interaction parameters, an adjustable energy for the phonons which dominate the pairing, and an adjustable ratio for the mean-field Tc to the actual Tc. We show results for a combined fit to the low-temoerature band gaps and to the T-dependence in one crystal. The phonon energy which gives the best fit is 21 meV. This is probably an appropriate average over the three longitudinal polar modes and acoustic modes in the material. A large value of about two is found for the ratio T_(cmf}/Tc, and we conjecture that this arises because a band with a small Fermi energy not seen in the tunneling results plays a part in increasing T_{cmf}/Tc.