Process variability modeling in devices; variation-aware modeling of devices


The shrinking volume of aggressively scaled semiconductor devices induces the emergence of novel stochastic phenomena collectively denoted as device variability. The random nature associated to technological parameters (e.g., atomistic doping, device and electrode size uncertainty, etc.) determine variations in the electrical features (e.g. threshold voltage, transconductance, etc.) since they are no longer mitigated by the device volume. These phenomena can be modelled exploiting the same techniques developed for physics-based noise analysis, traced back to the model Green's functions. Using such techniques, statistical analysis of the device subject to random variations can be addressed with high computational efficiency.  Similarly, the (deterministic) sensitivity to technological and physical parameters can be exploited for the optimization of the device performance and the sensitivity analysis of the device embedded in realistic circuits. This activity will leverage the concept of the Conversion Green's Functions for variability and sensitivity analysis of AC (small-signal) and large-signal device parameters. In recent years, simulation capabilities have been demonstrated for the statistical analysis of nanometer scale devices (in particular FinFETs) and of microwave devices, also embedded in microwave circuits (power amplifiers) with concurrent variations of device and circuit parameters (external loads).

ERC sectors 

  • PE7_5 (Micro- and nano-) electronic, optoelectronic and photonic components


  • Semiconductor device modelling
  • Variability
  • Device optimization