Development of a TCAD-ready, analytical electron mobility model for strained bulk Si and low electric field has been completed. The model includes doping dependence, temperature dependence, substrate orientation dependence and dependence on the field direction. An analytical model for electron velocity saturation at high fields has been developed and fitted to full-band Monte Carlo results for strained Si. Two approaches have been pursued, namely a three-valley model taking into account valley population as internal variables, and a more empirical, direct fit of the velocity-versus-field curve. A widely used high-field mobility model has been augmented, and the dependences of all parameters on the valley splitting are described by simple, analytical expressions. The physical effects of electron transport in strained Si channels have been studied in detail. Energy-resolved velocity profiles are a suitable means to demonstrate effects of degenerate statistics, surface roughness scattering, and screening on the surface mobility. As already observed by other authors, the theoretically predicted mobility enhancement at high normal field is still too small in comparison with measurement data.
A project on the modeling of silicon multi-gate FETs near the scaling limit based on the Wigner equation has been continued. Inclusion of size quantization of the carrier motion into the Wigner function approach has been addressed in detail. It has been shown that an approach treating size quantization directly within the Wigner function formalism leads to a complicated set of integro-differential equations for complex valued quantities, which could hardly be handled without introduction of additional, restrictive approximations. A practical approach which was found to be applicable to MOSFET simulation is based on the separation of the quantized motion in transverse direction from the motion in the direction of propagation. In this case the eigen-energy depends on the position along the propagating direction and plays the role of the coordinate-dependent subband minimum. The electron motion within each subband is then described by the corresponding Wigner equation, with the potential energy determined by the corresponding quantization energy of the transverse motion. The subband decomposition has been shown to provide a good approximation for double-gate and triple-gate silicon-on-insulator FETs. To prove this, we expanded the density matrix in the subband-related basis set and explicitly found the inter-subband coupling Hamiltonian. The inter-subband coupling elements were computed for different FET geometries and found to be much smaller than inter-subband energy, which allows them to be safely neglected in practically relevant cases. An existing numerical Monte Carlo simulator for solving the Wigner equation has been extended and tested. Special attention was paid to ultra-scaled devices in which the potential in the transport direction along the channel changes quite rapidly, and tunneling through the barrier is becoming important. Numerical stability of the simulation method is improved by a spectral separation of the potential along the channel into a classical and quantum mechanical part. The classical potential accommodates the voltage applied to the structure. The proposed potential separation allows Wigner function-based simulations of practically relevant double-gate SOI FETs.
Various architectures of carbon nanotube (CNT) FETs have been studied using Minimos-NT. Assuming ballistic transport, a Schroedinger solver coupled with the three-dimensional Poisson solver of Minimos-NT is used to analyze both Schottky-type and ohmic-type CNT-FETs. The current is calculated using the Landauer-Buettiker formula or by solving the Schroedinger equation with open boundary conditions. The charge on the tube is taken into account self-consistently. To optimize the off-state characteristics of the CNT-FET, a dual-gate structure has been proposed. The second gate effectively suppresses hole tunneling at the drain contact. The dynamic response of CNT-FETs has been analyzed employing the quasi-static approximation. It has been shown that through appropriate selection of the gate-drain spacer both the DC and AC response of ohmic contact CNT-FETs are improved. Through an increase in the gate-drain spacer the ambipolar behavior is suppressed and the parasitic capacitance between the gate and drain contacts is reduced. Suppressing the ambipolar behavior increases the on-to-off current ratio by three-orders of magnitude. Reducing the parasitic capacitances increases the cutoff frequency about 30%.
The theory and practice of conjugated pi-electron systems and their simulation by three-dimensional networks of energetically disordered localized states has been studied. Amorphous and polycrystalline zinc-phtalocyanine (ZnPc) samples at different doping levels have been adopted as appropriate compounds for the development of a dynamic Monte Carlo simulator covering diffusion and recombination phenomena in carbon-based semiconductors. The extremely low intrinsic conductivity can effectively be enhanced by p-doping, achieved by co-evaporation with tetrafluoro-tetracyano-quinidimethane (F4-TCNQ). A Gaussian disorder model has been implemented with an Abrahams-Miller-like jumping rate, ignoring polaronic effects and tacitly assuming the polymer's coupling to a heat bath. The ability to study the interplay between the spatial and the energetic disorder for various molecules and crystallographic phases has been considered to be an important feature of the simulator. As a basic discretization principle, two-level jumping sites, capable of simultaneously holding 2 LUMO and 2 HOMO-electrons (exclusion principle), repulsive on-level Coulomb interaction (Hubbard-model-like) and an excitonic binding energy are at this time implemented as the basic building blocks for the molecular orbitals. A second activity is the extension of a drift-diffusion-based device simulator with respect to organic devices such as OFETs and OLEDs. For several parameters specific models are required, such as mobility, density of states, trap distribution, and band alignment. Various models for the mobility in organic semiconductors exist in literature. Currently, a model based on the variable range hopping theory of Vissenberg is being investigated in more detail. The model should cover a wide range of temperatures and field strengths. Effects at the contacts play an important role. Contact models including thermionic emission, tunneling and interface recombination are implemented.