You Should Also Think about Complicated Things (to Improve Computation): Exorcising Numerical Ghosts from Electron Transport Calculations
Oak Ridge National Laboratory
Understanding how molecules conduct electric current is imperative for the development and/or improvement of solar cells, catalysts, scanning probe microscopies and molecular electronics. Theoretical frameworks for describing this electron transport are well established, and their recent marriage with quantum chemistry/physics packages should enable ab initio computational simulations of this physics. Unfortunately, the use of bigger or better or “more complete” basis sets can artificially increase (by several orders of magnitude) the calculated electron transport properties, and the cause of this “ghost transmission” has proven elusive [C. Herrmann et al., J. Chem. Phys. 132, 024103 (2010)]. In this talk I will discuss the computational challenges in simulating electron transport and also diagnose the cause of ghost transmission. In effect, basis set non-orthogonality causes a redefinition of several key quantities that can have serious implications when applying the inherent open-system boundary conditions. Finally, I will present strategies for exorcising ghost transmission, paving the way for accurate simulations of electric current through molecular wires.