Uneven etching baffles Inspector Jack
Harold Fitch
It all began on a Thursday afternoon as Bob`s boss`s boss returned to his office from the weekly plant manager`s review meeting. Bob was immediately summoned to explain what his department was doing to help the FET (field effect transistor) manufacturing line with its current etch problem.
Bob was not even aware of the problem but replied that he would look into it immediately and report back posthaste. Hurrying back to his office, he exclaimed, “This is a case for Inspector Jack!”
Bob located Inspector Jack in the contamination control laboratory working on filter testing experiments with Doug, the contamination control organization`s expert on filtration. The Inspector hadn`t heard anything about the etching problem in the FET manufacturing line either, but headed right out to the line to investigate.
Inspector Jack entered the line and learned there had been a problem with a thin oxide etch step for about two weeks. Sue, the manufacturing engineer working on the problem, had immediately suspected that, as was often the case in this kind of situation, the problem stemmed from the photoresist used to coat the wafers and define the pattern to be etched. Up to this point, she had tried several different batches of photoresist, varying the photoresist bake time and temperature to solve what she believed was a photoresist adhesion problem. Nothing seemed to work. The etchant was not etching evenly and patterns were being undercut producing uneven lines. Oxide was being left behind in the clear areas that should have been etched away.
Sue suspected the photoresist was flowing during the etch step, causing the uneven etching. After discussing the problem, Inspector Jack suggested checking the room temperature and humidity variations, which often can cause photoresist adhesion problems. Sue agreed.
Inspector Jack checked the temperature and humidity data for the last month in both the room where the photoresist operation occurred and at the actual hoods or workstations where the photoresist was applied, baked, exposed, developed and the oxide etched. This examination showed no significant variations. Everything appeared both normal and in specification — the humidity ran 40 percent ± 2 percent, and the temperature was held within ± 2 degrees Fahrenheit in the room and plus or minus 1 degree in the workstations. Inspector Jack`s experience told him photoresist adhesion problems are not usually caused by humidity problems if humidity is kept under 50 percent.
Next, the Inspector brought in equipment from the contamination control laboratory to check the calibration and accuracy of the on-line temperature and humidity measuring equipment. Again, everything checked out OK.
Time to backtrack …
Inspector Jack was mystified, so he met with the engineer again to discuss his finding and they decided to double-check some of the process elements, such as etch bath temperature, bake oven temperature, the etch process filtration set-up, and the etchant particle count level in both pre- and post-etch situations.
A check on the etch bath temperature control showed the temperature controller was working fine, was in calibration and was controlling the etch bath within the specified limit. A check on the bake oven temperature showed that the controller was in calibration and was controlling the oven to the proper temperature.
Inspector Jack was now truly stumped, so he called on Doug to check out the etchant filtration setup. Doug verified that the correct filter was being used and was working properly. He inspected all of the plumbing and materials of construction in the etch station and confirmed that they were the proper ones and should cause no problems.
The Inspector performed particle counts on the incoming etchant. While they were high, they were within specification and in the normal range. The particle counts on the recirculated, filtered, etchant were considerably lower, and again, within the normal range. Inspector Jack was still stumped.
A couple of weeks after Inspector Jack had first been called and about a month after the first occurrence of the problem, no solution had been found. The contamination control department decided to do a complete audit of the line area under Inspector Jack`s direction.
Everyone put on their bouffant caps; no stone was left unturned. All procedures were checked in great detail. This included items such as personnel dress and movement, handling, cleaning, technical requirement such as filter changes, equipment calibration, storage, process document audits and anything else that came to the minds of the team members.
The Inspector could only conclude that the line was in the best shape ever and that manufacturing was doing a great job monitoring and improving the quality index. This now left the team nearly two months into the problem with no solution in sight.
Solution eludes the Inspector
Inspector Jack continued to work with Sue, the line engineer, who had requested several chemical analyses of both the incoming and the used etchant. No variations from incoming quality specification were found in either. Tempers frayed, fingers were pointed and still no solution was in sight.
A befuddled Inspector Jack happened upon a senior manager from the development pilot line, who recalled that his group had not experienced the problem in the pilot line. This led to an immediate cross-experiment with the pilot line. Sure enough, when product from the manufacturing line was put through the same procedure in the pilot line, the problem went away.
Further cross-experiments revealed that the problem was, in fact, the etchant — it had nothing to do with the photoresist. “Elementary, my dear Inspector Jack! Elementary!” the team members exclaimed. Next, it was discovered that the etchant for the manufacturing line came from a new supplier, but the etchant for the pilot line still came from the old supplier.
While there appeared to be no analytical difference between the two etchants — in fact, the etchant and wetting agent mixture from the new supplier met the material specifications with flying colors — there turned out to be inexplicable differences in functionality.
As a result of this experience, the company now requires extensive functional tests before different materials can be introduced into an existing manufacturing line.
Inspector Jack had been stymied, but his approach to the problem and those of the other people involved could not have been more thorough. As is often the case in contamination control problems, a long, cooperative effort by many people and organizations brings to the eventual solution.
Harold Fitch is president of Future Resource Development, a consulting firm in Burlington, VT, specializing in cleanroom education and problem solving.