Use of Neural Network to Model the Refractive Property of PECVD Silicon Nitride Films used to Prevent Water Permeability of Piezoresistive Pressure Sensor

A plasma-enhanced chemical vapor deposition (PECVD) silicon nitride is used for the passivation layer of ICs and sensors. For example, in blood pressure sensor, the piezoresistive pressure sensor is immerged into liquid and the silicon nitride film is expected to prevent water permeability into it. In this paper, we first applied a design of experiment (DOE) and neural network to discover the relationship between input parameters and refractive indexes of PECVD silicon nitride films. Later we introduced the metrology developed to quantify the water permeability into the film based on the resistance test structure measurement and the break down measurement. These two performance tests could further use to identify the optimal silicon nitride deposition recipe which could prevent water permeability of piezoresistive pressure sensor.