Results of a seminal study (B. Xu and N. J. Tao, Science, 2003, 301, 1221) on the single-molecule junctions based on bipyridine placed in a solvent have been challenged recently (S. Y. Quek et al., Nat. Nano, 2009, 4, 230) by implicitly assuming a negligible solvent impact on the molecular transport and by merely considering low bias conductance data. In this paper we demonstrate that solvent effects on the molecular transport are important, and to show this we focus our attention on the energy offset epsilon(0) of the dominant molecular orbital (LUMO) relative to the electrode Fermi level. To estimate the energy offset epsilon(sol)(0) from the full I-V curves presented by Xu and Tao for wet junctions, we resort to the recently proposed transition voltage spectroscopy (TVS). TVS, which plays a key role in the present analysis, emphasizes that data beyond the ohmic conductance regime are needed to reveal the solvent impact. We show that epsilon(sol)(0) significantly differs from the energy offset epsilon(0)(0) deduced for dry junctions (J. R. Widawsky et al., Nano Lett., 2012, 12, 354). The present work demonstrates that solvent effects on molecular transport are important and can be understood quantitatively. Results of ab initio calculations with and without solvent are reported that excellently explain the difference delta epsilon(0) = epsilon(sol)(0) - epsilon(0)(0). delta epsilon(0) = Delta Delta G + delta Phi + delta W can be disentangled in contributions with a clear physical content: solvation energies (Delta Delta G), image charges (delta Phi), and work functions (delta W). Accurate analytical formulae for Delta Delta G and delta Phi are reported, which provide experimentalists with a convenient framework to quantify solvent effects obviating demanding numerical efforts.
