Abstract
We present a SPICE-based percolative model for the simulation of randomly aligned networks of carbon nanotubes (CNT). The network generation is based on a stochastic algorithm that randomly generates nanotubes over a substrate according to some statistical distribution inferred from the measurements of real devices. The transport mechanisms are modeled in a multiscale framework. The current of each nanotube is first pre-computed following the theory of one-dimensional channels. The electrical behavior of the entire network is afterwards simulated by coupling a SPICE program with an iterative algorithm that calculates self-consistently the electrostatic potential in each node of the network. Comparisons with AFM images allowed us to validate the model for the network generation, while the results from the simulations have been compared with measurements of devices designed as gas sensors. Particular attention has been focused on the response of the network in the AC regime.