Modeling of electron transport in hybrid silicon-molecule devices

Geng-Chiau Liang, Purdue University


Electron transport through metal-molecule-metal junctions has been widely studied both experimentally and theoretically. Recently, new experiments involving molecules assembled on semiconductor surfaces such as Silicon and GaAs have been reported. Compared to metal contacts, silicon substrates exhibit stronger covalent bonding with the molecules, and their properties can be tuned by doping. Moreover; the sharp band-edge characteristic of semiconductors leads to a bias-dependent negative differential resistance (NDR) for degenerately N-type and P-type doped substrates which was predicted by theory and has been confirmed by recent experiments. Modeling molecule-semiconductor heterostructure is a challenge due to the semiconductor's unique properties such as its band-edge, surface reconstruction, and surface states. These features play important roles in electron transport through molecules and contacts under non-equilibrium conditions. Localized basis-sets used in first-principle calculations describe molecules very well, but are not efficient in characterizing bulk properties of the contacts which are very important in determining the current-voltage characteristics. Semi-empirical calculations like Extended Huckel Theory (EHT) or effective mass method are typically better equipped to model both the bulk and the surface physics of contacts. In this thesis, we develop a formalism to couple two different basis functions in order to accurately model both molecules and contacts under non-equilibrium conditions. As an example, we use EHT parameters optimized to describe bulk silicon, and couple it with an ab-initio basis set, 6-31g(d), to simulate the contact surface atoms and the molecules grown on silicon. Such a coupling is achieved by matching the surface green's function in real space in both basis sets. Moreover, we use this hybrid-basis formalism to couple the contacts with a density-functional treatment of the molecule to simulate scanning tunneling spectroscopy (STS) measurements of C60 on a silicon substrate. Our results demonstrate that several experimentally observed variations in conductance-voltage (G-V) characteristics can be quantitatively explained simply by invoking variations in the detailed bonding geometries and the varying separations of the scanning probe from the molecules influence their peak amplitudes. In addition, our simulations exhibit a prominent negative-differential resistance (NDR) in such molecular I-Vs due to the interaction between the molecular levels and the silicon band-edge. We are able to use these results to qualitatively interpret experimental observations of room-temperature negative differential resistance.




Datta, Purdue University.

Subject Area

Electrical engineering

Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server