Inspector Jack`s Case of Deja Vu
By Harold D. Fitch
When Inspector Jack is called to an older production line to help solve a high foreign material count at a post-resist strip inspection point, he immediately sensed a feeling of deja vu.
This type of production process is fairly straight forward. The product is being coated with photoresist. The photoresist is then exposed through a photomask. The exposed image is developed to remove the resist in selected areas. The product is etched, rinsed and dried, before going to the resist strip operation. The problem occurs when the product is inspected after the strip operation.
Inspection takes place by shining a bright light, very similar to a Kodak carousel, on the product`s surface at a shallow oblique angle. An inspector then looks down at the product and counts any contamination highlighted by the light beam. This is very similar to sunlight highlighting dust in room air when a bright ray shines in through a window. Reflected light makes particles down to about 5 microns visible, whereas the unaided eye would cut off at about 50 microns.
The inspection limit for this product is a pass-fail cut-off of 16. Any product with more than 16 reflected spots fails and has to be reworked. The product goes through 100 percent inspection. Currently, 30 to 40 percent are failing each pass through the resist strip inspection. In addition, products that pass inspection average a count of 14. This inspection had recently been implemented as part of an enhanced effort to improve process yields throughout the line.
Finding the correlation
Inspector Jack immediately recognizes that this problem is very similar to problems on the advanced semiconductor line and that similar solutions should apply. Even though the dimensional requirements in this case are not as rigid as those on the advanced semiconductor line, the principles of contamination control should be the same.
Inspector Jack decides to do a preliminary check before making any recommendations. His first action is to check the particle count of the solvent used in the resist strip operation. The particle count on the incoming strip solvent shows a count of over 100,000 particles per liter at a size of 2 microns and above, when checked with a liquid particle counter.
He knows from experience that adding continuous circulation filtration to the strip bath will both reduce the initial particle count in the strip bath and keep it low during the photoresist strip operation. At this point, Inspector Jack reviews the situation with the contamination control department`s filtration expert, Doug.
Doug checks the solvent being used in the stripping process for chemical compatibility and matches this information with the process requirements and available filters. He then provides Inspector Jack with his filter selection and details on setting up the recirculation filtration tank. This includes material recommendations for the filter housing, plumbing, pump selection, and the tank.
The test set-up
Inspector Jack constructs the test set-up that Doug recommended. The continuous recirculation filtration system demonstrates that it can rapidly reduce the count in the initial strip solution to less than 5,000 particles per liter at 2 microns and above.
Next, Inspector Jack sends products exhibiting excess contamination to the failure analysis laboratory to be sure that the contamination is indeed residual photoresist. When this is confirmed, he is ready to test his solution.
First, he runs extensive tests–stripping several hundred pieces of freshly resisted test product with the filtered strip bath. The stripper particle counts stay well below the 5,000 level. Jack knows each change must be carefully tested before implementing it in the production line. On semiconductors, the initial introduction of recirculation filtration to the etching step removed the wetting agent in the etchant and impaired the etch control. After much hard work and the introduction of a new wetting agent, there finally was a process that stood up in actual manufacturing conditions.
Inspector Jack continues by setting up a small strip process in the contamination control laboratory. He uses two subsequent strip tanks with recirculation filtration, followed by a DI water rinse with nitrogen agitation, and a final DI water spray rinse. He tests the proposed process again using rejected product, but this time the product goes through the actual manufacturing line inspection and the product passes inspection with flying colors.
Inspector Jack is finally ready to present his plan to manufacturing. He outlines the plan and includes Doug`s information on the optimal tighter control. The manufacturing group decides that it is best to get the process in control, under current requirements, rather than tighten them.
The next step is to run several lots of specially identified actual product through the laboratory strip set-up, followed by actual manufacturing line inspection, and then continue tracking the product through final test in manufacturing to make sure the process changes have not caused any other product problems.
Upon completion and testing of 10 actual product lots at final product test, the results show that the product processed through the new procedure has four to five percent higher overall yield at final test. The bath particle count runs consistently under 300 particles per liter at 2 microns and above. The percent of the product passing inspection on first pass increases to over 95 percent and the average count on passing product drops to about five particles.
The increase in final test yield probably results from less contamination on the good product now coming out of the strip process than there used to be (five particles vs. 14). This reduces the contamination proceeding on through subsequent steps. n
Harold Fitch is president of Future Resource Development, a consulting firm in Burlington, VT, specializing in cleanroom education and problem-solving.