Multivariate regression analysis using the gradient descent method in the analysis of aluminum nitride films formedby magnetron discharge

Aluminum nitride films are used in the design of devices for selection and generation of signals in the microwave range on longitudinal acoustic waves. The quality of aluminum nitride thin films during magnetron formation is influenced by the following technological parameters: pressure in the vacuum chamber, substrate temperature, power supplied to the target, the ratio of argon and nitrogen gas flows. A large number of technological parameters have an ambiguous effect on the surface quality and properties of films. For A multivariate regression analysis was carried out to determine the optimal technological modes.