Authors: L.R.C. Fonseca and A.A. Demkov
Affilation: Motorola Inc, United States
Pages: 86 - 89
Keywords: transport, scattering theory, tunneling, device physics, molecular devices, nanoelectronics
Ab-initio calculations of transport through nanometer scale structures, such as individual atoms or molecules and ultra-thin oxide barriers directly link atomic structure and chemistry, which are difficult to probe, to transport properties, which are easier to access experimentally. However, in order to deliver results that are qualitatively and quantitatively meaningful, such calculations face three major challenges: (1) the band gap problem, inherited from density functional theory (DFT) which is the mostly used approach when a self-consistent band structure is required, (2) a proper description of the wavefunction tails of the tunneling states, (3) the interaction between the device and the electrodes. In this work we address issues (2), which may have a qualitative effect on the tunneling current, and (3), which in general has a more quantitative character. We selected two systems for investigation: Si/SiO2/Si MOS device and Au/benzene-1,4-dithiol/Au. Transport was calculated using non-perturbative scattering theory operating on the tight-binding-like Hamiltonian generated by the local orbital SIESTA code. On issue (2) we show that the exponential leakage current decay with barrier thickness of ~1 decade/2 Å measured in Si/SiO2/Si  can only be reproduced theoretically using at least a single-zeta plus polarization (SZP) basis set. The same basis set comparison is repeated for Au/benzene-1,4-dithiol/Au where the calculated conductance is compared to the experimental values. On issue (3) we show that, independently of surface orientation or detailed surface atomic arrangement, the number of electrode atomic layers affected by the presence of the device is large for both systems. We show that underestimating that number may result in an overestimation of the tunneling current by several orders of magnitude.