Real-space method for highly parallelizable electronic transport calculations

Abstract

We present a real-space method for first-principles nano-scale electronic transport calculations. We use the non-equilibrium Green's function method with density functional theory and implement absorbing boundary conditions (ABCs, also known as complex absorbing potentials, or CAPs) to represent the effects of the semi-infinite leads. In real space, the Kohn-Sham Hamiltonian matrix is highly sparse. As a result, the transport problem parallelizes naturally and can scale favorably with system size, enabling the computation of conductance in relatively large molecular junction models. Our use of ABCs circumvents the demanding task of explicitly calculating the leads' self-energies from surface Green's functions, and is expected to be more accurate than the use of the jellium approximation. In addition, we take advantage of the sparsity in real space to solve efficiently for the Green's function over the entire energy range relevant to low-bias transport. We illustrate the advantages of our method with calculations on several challenging test systems and find good agreement with reference calculation results.