100% defect inspection with neural network integrated system of in-situ particle monitor and surfscan

The ever-growing demand to increase speed by shrinking device geometries has driven manufacturers to take more drastic measures in defect reduction. The installation of the state-of-the-art air filtration systems has shifted the focus from human-caused contamination to process-induced defects. For this reason, in-situ particle monitor (ISPM) has been adopted in the industry for optimizing cleaning cycles and detecting equipment abnormalities. However, if the equipment is well-maintained and yield-dominating excursions rarely happen, ISPM signal is usually too weak for yield monitoring. In this work, the ISPM dependence with various process settings for an etching process under nonexcursion condition was carefully studied. An alternative approach of integrating ISPM in a yield monitor system was developed, experiments were carried out and encouraging results were obtained. We were able to integrate ISPM into our daily manufacturing processes so that effective 100% sampling can be exercised and all wafers with potential particle problems are identified for careful study and remedy action