Identifying Quantum Interference Effects from Joint Conductance-Thermopower Statistics

Although quantum effects are thought to dominate the heat and charge transport through molecular junctions, large uncertainties in chemical structure, lead-molecule coupling strengths, and energy levels make it difficult to definitively identify these effects from the measured thermopower S and conductance G distributions alone. Here, we develop a simple statistical method to identify destructive quantum interference features (nodes) through the anticorrelation between simultaneously measured G and S values. We find these correlations can be used to unambiguously identify far-detuned nodes, even when G and S distributions alone cannot. As an example, we consider several para- and meta-configured systems, including benzenediamine and diiodo-terphenyl-based junctions, finding that nodes can be identified in ensembles with broad level-alignment and lead-molecule coupling distributions, and with significant anodal transport contributions, including from vacuum tunneling. The efficacy and limitations of this method are analyzed.