We use Monte-Carlo Simulations to study the conductance switching generated by gas-induced electron trapping/-releasing in films of sintered metal oxide nanoparticles by using a site-bond percolation model. We explore the possibilities of gas sensors based on these mechanisms. In our study, we model films of different thicknesses where the conductance values of the grains (sites) and of the contacts (bonds) between these grains depend on the surface density Nr of adsorbed gas molecules from the ambient atmosphere. Below a critical density Nr=Nr,c , the system is insulating due to the interruption of current flow, either through the connecting bonds or through the grain interior. This leads to two competing critical gas covering thresholds N(bond)r,c and N(site)r,c , respectively, that separate the insulating from the conducting phase. For N(site)r,c>N(bond)r,c , the characteristic curve of monodisperse sensors shows a noticeable jump from zero to a finite conductance at Nr=N(site)r,c , while for polydisperse sensors site percolation effects modify the jump into a steep increase of the characteristic curve and thus lead to an enhanced sensitivity. For N(site)r,c<N(bond)r,c , both mono- and polydisperse systems follow the same curves that show a smoother characteristic increase ∝(Nr−N(bond)r,c)2 which reveals that, despite the occurrence of an inherent bond percolation effect close to Nr,c , the increase of the bonds is the dominating effect.
