Attaining process capability using multiple linear regression and other techniques

This paper describes the development and refinement of a series of models, based around simultaneous multiple linear regression models, to predict the layer thickness resulting from a manufacturing process. The process is the makin of optical films by the depsition of very thin layers of titanium dioxide and silicone dioxide. As the model enable the company (Omitec Thin Films Ltd. of Plympton) to improve its prediction of layer thickness, so it is able to improve process capability, reduce scrap, and improve its record of short lead time, on-time delivery of conforming products.