Improving the electroforming process in optical recordable media manufacturing via an integrated procedure

In optical recordable media manufacturing, an electroforming process uses glass masters to produce metal shells which then act as stampers to replicate thousands of copies of a disc. The physical characteristics of stampers influence their life time significantly and, moreover, significantly affect the quality performance of the finished optical recordable media. Traditionally, engineers sought the optimal parameter settings in the electroforming process through trial and error, and thus serious losses were experienced owing to the low yield of stampers. This study proposes an integrated procedure for optimizing parameter settings in electroforming to improve stamper quality performance. The proposed procedure combines neural networks, desirability functions and tabu search to solve multi-response parameter design problems. The proposed procedure was applied at a Taiwanese optical recordable media manufacturer, and the implementation results demonstrated its feasibility and effectiveness. Through this work, the average defect rate of stampers can be expected to decline to approximately 4.76%, from over 10% previously. The annual savings through applying the proposed procedure are estimated at US$750,000, easily exceeding the US$80,000 expended on the experiment.