device engineering section manager, Motorola, Mesa, Arizona
Pages: 11 Published: Jan 1987
Tightening the distributions of key device parameters is increasingly important in the enhancement and prediction of yields, and in improving the quality of the products of a semiconductor fabrication area. Extensive experimentation is not always feasible in a production area. Statistical tools and knowledge of the effects of processing factors on the device characteristics are used to help identify the most significant causes of variability with the fewest experiments. The Generation of System Moments method, combined with reliable mathematical models or simulation programs which relate device parameters to processing factors partitions the device parameter variance to the causes according to partial derivatives obtained from the model. In identifying the most significant factor, this method is an effective tool in evaluating methods and tradeoffs in efforts to improve process control. The most significant factors involved in threshold and breakdown voltage variability were identified using this approach.
Parametric variability, statistical modelling, device sensitivity, generation of system moments, error propagation, Pareto analysis
Paper ID: STP25778S