The day the CPK report came out, the whole room fell silent for three seconds, and then the boss's face turned green.
"Damn it! Why is it down again? Didn't we say we replaced the parts last time?!" At 2 AM, staring at the alarm from the production line, my head was about to explode. It was a rare weekend without being on duty, but an urgent call dragged me back from my dreams to reality. It turned out that a new piece of equipment, recently put online, had its critical process parameters drifted yet again, resulting in a batch of scrapped products. The process engineer was so angry they wanted to smash the machine, the equipment engineer's face was ashen, and I, as the "senior" representative for the initial equipment evaluation, could only silently endure pressure from all sides. In the meeting, the boss looked at the dreadful Cpk 1.08 report, his face looking worse than a machine breakdown. I knew it was time to review "how the equipment was evaluated in the first place."
What Exactly Went Wrong? FMEA, in essence, is "Proactive Deployment."
To be frank, you and I both know that such annoying incidents are almost always due to "insufficient planning." We often only look at specifications and price when purchasing equipment, neglecting "potential risks." "FMEA," short for "Failure Mode and Effect Analysis," sounds academic, but frankly, it's a "proactive deployment checklist." Imagine, when you buy a new car, would you consider where it might break down in the future? Would you worry about brake failure? Engine trouble? FMEA does exactly this: it requires you to anticipate all potential failure points before the equipment even enters the factory, or even during the design phase, and then consider how to prevent or reduce these risks.
The key point is that FMEA is not a document for post-incident review, but a tool to make you "forewarned and forearmed, preventing future troubles."
How to Do It in Practice? Give It a Number to Reveal the Risk.
The FMEA commonly used in our semiconductor fabs is usually divided into DFMEA (Design FMEA) and PFMEA (Process FMEA). During equipment planning, we focus more on DFMEA.
- Identify Failure Modes: Where might this equipment fail? How might it fail? What negative outcomes might result? For example: vacuum pump seal aging leading to leakage, pump motor overheating and burning out, chamber scratches causing particle contamination.
- Assess Severity (S): If it fails, how significant is the impact? Will it lead to scrap? Downtime? Personnel injury? We typically assign a score of 1-10, with 10 being catastrophic, such as "million-dollar wafer scrap."
- Assess Occurrence (O): How high is the probability of this event occurring? Does it happen frequently, or is it a rare occurrence every five years? Again, a score of 1-10 is given, with 10 being "high frequency of occurrence."
- Assess Detection (D): If it truly happens, how quickly can we detect it? Are there real-time alarms? Or will we only find out when product issues arise? A score of 1-10, with 10 being "very difficult to detect."
Then, multiply these three scores to obtain a "Risk Priority Number (RPN) = S x O x D." Our goal is to identify the points with the highest RPNs and then find ways to reduce them. For example, if a chamber scratch leads to particle contamination, and S=9 (high scrap risk), O=7 (high probability of occurrence), and D=8 (difficult to detect), then the RPN is 9 x 7 x 8 = 504. Seeing a number like this, you know this is definitely a critical issue that needs priority attention.
In other words, RPN transforms abstract "risks" into concrete numbers, allowing you to instantly identify the most dangerous areas.
The Most Common Pitfall: Treating FMEA as a Mere Paperwork Exercise.
The most common pitfall I've encountered is treating FMEA as a "compliance document." Suppliers hastily write it up just to deliver the equipment, and we just glance at it and sign off. What's the result? After the equipment arrives, the "preventive measures" so well-documented in the FMEA are simply not implemented. For instance, the FMEA might recommend "replacing critical consumables every six months," but after the equipment is installed, this turns into once a year to save costs or avoid downtime. Then, we just wait for breakdowns, scrap, and the boss's face turning green.
Another pitfall is completing the FMEA and then shelving it without periodic review. Production line environments and process conditions are constantly changing; new failure modes might emerge, and old risks might decrease due to improvements. FMEA is a living document, not a one-time assignment.
One Thing You Can Do Today
Open the equipment FMEA you have on hand, look at the five points with the highest RPNs, and check if their preventive measures are truly being implemented.