Issue link: https://iconnect007.uberflip.com/i/1527952
76 PCB007 MAGAZINE I OCTOBER 2024 When we resumed running production the next day the room was suddenly filled with ammonia fumes that made it impossible to get anywhere near the etcher. e system had a warm air dryer at the end and a vent between the preceding rinse and dryer to prevent air from the dryer blowing through the system. A quick check on the flow through the vent showed the airflow was less than half of what it had been when installed. e building had a central ventilation sys- tem and investigation showed that a new plat- ing line had just been installed in the next room with the ventilation lines installed ahead of the feed to the alkaline etch room. When those vents were on, it reduced the airflow to the alkaline etch room, so there wasn't enough flow in the vent to prevent the air from the dryer from flowing through the system, and it was blowing ammonia fumes into the room. We solved the problem by reducing the airflow from the dryer to the minimum needed so we could dry the panels and not overwhelm the vent. is was a somewhat obvious cause-and- effect situation, but other things we could have considered might have been changes in chem- istry or suppliers, materials or suppliers, or processing in prior steps. A Problem on the DES Line An example of isolating the problem hap- pened more than 20 years ago when a customer had just installed a new DES line and reported that line widths on panels were frequently out of spec. ey ran developed panels from their other DES line through the new etcher with- out any problems with line width specifica- tions, but developed panels from the new line etched in the other etcher showed the same out-of-spec line widths; therefore, not unrea- sonably, the problem was with the developer in the new line. A trip to the customer site could find noth- ing obviously wrong with the new developer. Fortunately, the customer had a standard test pattern they ran through each system daily and an AOI to get quick results. ey ran the test patterns in groups of five to check repeatabil- ity and the data was stored. A review of the data from the new line revealed that from each group of five, some panels were well within specifications, some were barely in specifica- tions, some were wildly out of specifications, and none of the groups of five were entirely in or out of specifications. A problem with the developer would show consistent failure modes, not on and off prob- lems, so I concluded the problem was with the exposure unit. But could it be? e exposure unit was brand new with the DES line and was the latest and greatest unit available; it cost more than the DES line. Some exposed but undeveloped panels were collected along with some unexposed panels and brought back to our lab. e exposed panels were developed in the lab etcher and the unexposed panels were exposed in our exposure unit (not the latest and greatest by any means). e panels were then brought back to the customer site. e panels exposed there and developed in our lab showed the same problem of inconsistent line " Don't assume you know the problem, so look at alternative scenarios that might be the cause. "