Predicting the conductance of strongly correlated molecules: the Kondo effect in perchlorotriphenylmethyl/Au junctions

TL;DR

This paper presents a comprehensive first-principles method combining DFT, NRG, and rSPT to predict conductance and Kondo effects in strongly correlated molecular junctions, validated against recent experiments.

Contribution

It introduces an integrated computational approach to accurately model equilibrium and non-equilibrium transport in correlated molecules, bridging theory and experiment.

Findings

Estimates Kondo temperature for PTM/Au junctions.

Provides qualitative and quantitative transport property predictions.

Assesses validity of equilibrium DFT+NRG calculations for finite bias.

Abstract

Stable organic radicals integrated into molecular junctions represent a practical realization of the single-orbital Anderson impurity model. Motivated by recent experiments for perchlorotriphenylmethyl (PTM) molecules contacted to gold electrodes, we develop a method that combines density functional theory (DFT), quantum transport theory, numerical renormalization group (NRG) calculations and renormalized super-perturbation theory (rSPT) to compute both equilibrium and non-equilibrium properties of strongly correlated nanoscale systems at low temperatures effectively from first principles. We determine the possible atomic structures of the interfaces between the molecule and the electrodes, which allow us to estimate the Kondo temperature and the characteristic transport properties, which compare well with experiments. By using the non-equilibrium rSPT results we assess the range of validity of equilibrium DFT+NRG-based transmission calculations for the evaluation of the finite voltage conductance. The results demonstrate that our method can provide qualitative insights into the properties of molecular junctions when the molecule-metal contacts are amorphous or generally ill-defined, and that it can further give a fully quantitative description when the experimental contact structures are well characterized.