Molecular electronics on silicon

It is conceivable that molecules with their myriad and tunable properties including size, shape, absorption spectrum and flexibility will create new functional possibilities on silicon. In a recent theoretical work we predicted that a hybrid silicon-molecule structure can exhibit negative differential resistance (NDR) in the current-voltage (I-V) characteristics. The structure is a two-terminal one consisting of molecules grown on heavily doped silicon substrates and a scanning tunneling microscope (STM) tip. It is conceivable that the presence of the silicon bandgap can give rise to NDR when the potential of the STM tip drives a molecular level into the bandgap cutting off transmission. NDR is expected to occur in the positive bias direction for molecules on degenerately doped p-type silicon substrates when a molecular level crosses the valence band-edge (Ev) and in the negative bias direction for degenerately doped n-type substrates when a molecular level crosses the conduction band-edge (Ec). Experimental observations of bias dependent NDR (Hersam group, Northwestern) for differently doped substrates are in good qualitative agreement with the theoretical predictions. In my thesis, we first explain the physics of the polarity-dependent NDR from an energy band diagram. Then we present a simulator that is based on Density Functional Theory (DFT) and Non-equilibrium Green\u27s Function (NEGF) formalism that self-consistently solves for the potential and charge density inside the molecule in the presence of electrodes at non-equilibrium. Next we show I-Vs of molecules docked onto silicon obtained from the simulator and point out what features of the experimental I-Vs are well reproduced and what features are still unexplained. We discuss possible new physics that may help understand the experiments better and suggest experiments to verify the ideas. We also address the question of whether a molecule used in a three-terminal device configuration can offer any performance advantages relative to standard silicon devices. We derive a general result that can be used to evaluate and compare transconductance of different field-effect mechanisms in molecular transistors, both electrostatic and conformational. We show that the conformational component helps the electrostatics of gate control but can lead to significant advantages only if the dipole moment of the molecule is comparable to the charge times the thickness of the gate oxide