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

摘要

The application of plasma-enhanced chemical vapor deposition (PECVD) silicon nitride, commonly used as the passivation layer for ICs, has been applied to advanced sensors as well. For example, the silicon nitride film coating the piezoresistive pressure sensor of a blood pressure sensor is used to prevent water permeation when immerged into liquid. In this paper, we applied a combination of design-of-experiment (DOE) and neural network to reveal the relationship between input parameters and refractive indexes of PECVD silicon nitride films. We also introduced a metrology to quantify the water permeability into the film based on the corresponding resistance test structure measurement and the break down measurement. These two performance tests are further used to optimize the silicon nitride deposition.