Preventing Lithography-Induced Maverick Yield Events With A Dispense System Advanced Equipment Control Method

As semiconductor manufacturers march to the drum beat of Moore's law there is very little room for yield mavericks, especially those that can be prevented. Critical process errors are costly and photolithography is one of the few processes in semiconductor manufacturing where there is an opportunity to correct errors. Small changes in photo resist dispensed volume may have severe impact on film thickness uniformity and can ultimately affect patterning. It is important to monitor photo-dispense conditions to detect real-time events that may have a direct negative impact on process yield and be able to react to these events as quickly as possible. This paper presents an evaluation of the IntelliGenreg Mini, a photo resist dispense system manufactured by Entegris, Inc. This system utilizes advanced equipment control software, known as dispense confirmation, to detect variations in photo dispense. These variations, caused by bubbles in the dispense line, valve errors, and accidentally changed chemistries can all create maverick yield events that can go undetected until metrology, defect inspection, or wafer final test. The ability of an advanced dispense system to detect events and create alerts is a very powerful tool, but it can be most effective when that information is collected and analyzed by an automated system. In a modern fabricator this is most likely a statistical process control chart that is monitoring a track's progress and is ready to stop the track when a maverick event occurs or alert personnel to trends they may not otherwise catch with other inline metrology data. Dispense confirmation, when combined with networking capabilities, can meet this need. After a brief description of the pump, data from simulated yield-affecting events will be examined to evaluate the IntelliGenreg Mini's ability to detect them. This discussion will conclude with a brief analysis of the ultimate time and cost savings of utilizing dispense confirmation with networking capabilities to detect and eliminate poorly coated wafers.