Source-induced RDF overwhelms RTN in nanowire transistor: Statistical analysis with full device EMC/MD simulation accelerated by GPU computing

We numerically demonstrate that a random dopant fluctuation (RDF) in a source region causes a noticeable variability in the on-current of Si nanowire (NW) transistors, and its effect is much larger than that of a random telegraph noise (RTN). This work assesses the static and dynamic variability of NW device characteristics using the ensemble Monte Carlo/molecular dynamics (EMC/MD) simulation, which employs parallel computing technique using a graphic processing unit (GPU). The current flow in a one-dimensional NW device is determined by the number of dopants at the source edge, indicating the importance of forming an abrupt source-channel boundary to suppress the variability.