Semiconductor manufacturing process visualization

This paper describes a new graphical display of semiconductor manufacturing processes. The display shows the tool usage at each step and the frequency of transitions between adjacent steps using a normalized nonlinear color map. We use normalized entropy and conditional entropy to quantify how close the tool usage and transition frequencies are to an ideal uniform mixing process. The display also shows the steps for which a fraction of the data was lost. Throughout the display we used color to emphasize different process conditions: red represents alarm conditions and is designed to draw attention; white represents normal, expected conditions; and blue represents lower than normal usage. We designed the display to convey useful information and help process engineers visualize large data sets. This enables fast identification of violations in the assumptions of typical statistical process control techniques and data integrity problems.