|Login | Register | Logout|
Nanodsim is a NEGF-DFT transport package for modeling solid state devices that involve large number of atoms and atomistic disorder. It implements a semi-local exchange that accurately predicts band gaps and dispersions for many important semiconductors and insulators.
Realistic device structures always involve some random disorder such as defects, vacancies, dopants, impurities, roughness and irregularities. Physical quantities calculated by modeling should be averaged over disorder configurations. In many applications, one not only needs the mean value of the physical quantity, but also the variance surrounding the mean which gives the device-to-device variability. In brute force super-cell calculations, one generates and calculates many atomic configurations to obtain the mean results and variance. This quickly leads to prohibitively lengthy calculations. For semiconcutors and insulators, DFT with local functionals does not correctly predict band gaps and dispersions. These problems severely limit the applicability of NEGF-DFT methods to semiconductor systems.
Nanodsim resolved these problems. It does disorder averaging by the non-equilibrium vertex correction (NVC) theory. NVC analytically derives a self-energy due to multiple impurity scattering at the non-equilibrium density matrix level, thus disorder averaging and device-to-device variebility are calculated in one shot. In particular, the random impurity averaging is done within coherent potential approximation (CPA) for single particle quantities (e.g. Hamiltonian) and propagators, and within NVC for NEGF and transport properties. These advanced techniques allow the mean as well as the variance of the physical quantities (e.g. conductance) to be predicted accurately and efficiently. Another very important feature of nanodsim is its implementation of a semi-local exchange functional that accurately predicts band gaps and dispersions for many important semiconductors.
Different from the general purpose tool nanodcal, nanodsim is specially designed for calculating solid-state devices that tend to involve large number of atoms. The DFT of nanodsim is based on the TB-LMTO method. We have applied nanodsim routinely on systems with more than a thousand atoms in the device scattering region. Running on a parallel cluster with 600 cores, we have analyzed Si nanoFET channel structures having 9,960 Si atoms. Nanodsim is fully parallelized in computation and distributed in memory.
Nanodsim have been applied for quantum transport modeling of the following systems: