Compact model of a metal oxide molecule sensor for self-heating control

Abstract

Metal oxide molecular sensors have notable advantages in their low cost and small size, and they are useful to establish a low-power sensory system for massive data accumulation. However, most of them require high temperatures to cause chemical reactions, and an external heater is needed. Thus, their sensory system consumes around 1 W or higher power for operation. A candidate solution to decrease power consumption is using self-heated sensors. Since the heating area is limited in the sensor itself, mW-order and quick operation are possible. Since the sensors need careful temperature management, a dedicated analog circuit is demanded. Therefore, it is important to create their compact models in MATLAB or Verilog-A and predict the performance of the system as a whole with simulations. In this paper, an experimental self-heated sensor and its compact model are developed. To check the model’s validity, some critical model parameters are determined firstly by experiments without self-heating. Then, the simulation outputs are compared with experimental results with self-heating. The comparison shows that the model predicts the saturation value and transient time constant of the gas reaction well. In addition, the error caused by the sensor’s drift increases particularly if the sensor is operated in an inert gas.