Does the Seebeck coefficient of a single-molecule junction depend on the junction configuration?

A new experimental method for the simultaneous determination of the electric and thermoelectric properties of metal–molecule–metal junctions at the single-molecule level has been developed to test the effects of the junction configuration on the thermopower properties. The method is based on dynamic switching between (thermo)electric current and thermoelectric voltage measurements. Two model systems, 4,4′-bipyridine (1) and 4,4′-diaminostilbene (2), have been scrutinized. Single-molecule conductance (G) and thermopower (S) values were obtained for the two most probable junction configurations of 1 and 2, each having two different conductance values, GH (high) and GL (low), where GH > GL. Thermopower values of S(GH) = −6.4 ± 1.5 μV K−1 and S(GL) = −7.0 ± 1.6 μV K−1 were obtained for the molecular junctions of 1 and values of S(GH) = +14.4 ± 3.5 μV K−1 and S(GL) = +10.4 ± 3.0 μV K−1 were obtained for the molecular junctions of 2. The GH and S(GH) values for 1 and 2 are consistent with previously reported results. Thermopower values obtained simultaneously with conductance measurements for both configurations of 2 during junction evolution are reported for the first time. This work shows that, within experimental error, both S values are the same for each molecule, i.e., S(GH) ≈ S(GL), and they do not depend on the molecular junction configuration. This is an important finding, which supports claims that thermopower is an intensive property of matter. DFT calculations of transmission functions combined with a non-equilibrium Green's function approach complete this study.