Electrical properties of granular semiconductors: Modelling and experiments on metal-oxide gas sensors

DC, AC, and the transient characteristics of granular n-type semiconductors are modelled using the drift-diffusion theory. The transient model describes the electrical large-signal response to both voltage and temperature changes. The analysis is based on the dynamic electrical model of the grain-boundary region and electronic trapping in the acceptor-type electronic interface states at the grain boundaries. The use of different approximations in the model derivation results in a simple fully analytical model and a semianalytical model, which requires numerical methods in the solution. The models are verified by performing numerical device simulations with SILVACO ATLAS. The models are also fitted to experimental data. They are in excellent agreement with ATLAS and the experimental data. Compared to ATLAS the transient calculations employing the semianalytical model are four orders of magnitude faster on a standard PC computer, yet having the same accuracy. The existence of electronic traps at grain boundaries results in nonlinear DC, extraordinary AC, and highly complex and nonlinear transient electrical characteristics. The current-voltage curves can be divided into four characteristic regions: linear, sublinear, superlinear, and series resistance limited regions. The electrical-equivalent-circuit presentations of the AC characteristics have, in addition to the common resistors and capacitors, special RL and RC circuit branches associated with the electronic trapping. These circuit branches have negative admittance. In the experimental part an atomic-layer-deposited SnO2 microhotplate gas sensor was designed and fabricated for the first time. The sensors exhibit good response and recovery to ethanol, acetone, and acrylonitrile vapours, and good stability. In addition, the developed model is extended to the case of n-type gas-sensitive surface-type metal oxides. The adsorption of gases is described by a surface-state model. The model is employed in the quantitative explanation of the new effects in metal-oxide gas sensors: the bias-dependent sensitivity and negative admittance effects, which were observed experimentally in commercial WO3 gas sensors. These effects can be used for increasing the selectivity of the gas sensors.