Onboarding at a semiconductor company is thick. Cleanroom gowning, chemical and gas safety, process fundamentals, tool architecture and SOPs take weeks. Yet the first moment a new engineer is truly stuck usually comes from outside the textbook. On night shift, a post-etch CD measurement falls outside the SPC control limits, and the operator asks: "If I re-measure and it comes back in, can I just move the lot on?"
The difference between control limits and spec limits is in the textbook. What to do first when that question comes at 3 a.m. is something nobody has practiced. The Solmaru Semiconductor Fab map in First Run collects moments like this as quests. Solmaru Semiconductor and its customer, Orvane Automotive, are both fictional, and the quests teach the order of judgment rather than any company's recipes or numbers.
How the fab map is set up
Pick the semiconductor fab in a new game and you start as an intern in a bunny suit. There are five roles, and the goal is Line Manager.
- Process Engineer: protects wafer quality through recipes and process conditions. SPC alarms and OCAPs, recipe change control (MOC), scoping an excursion.
- Equipment Engineer: PM due dates under production pressure, interlocks, LOTO before a vendor starts.
- Production Planning & Control: hot lots and Q-time, shift handover, keeping MES records.
- Quality & Yield Engineer: yield drops, the root cause in an 8D, whether to trust a new gauge.
- New Line Setup PM: tool move-in clashing with hook-up, hot-work permits, reporting a slip early.
Add two shared tutorials, three customer quests on audits, delivery commitments and a product change notice (PCN), and two quests from the Line Manager's seat, and the map has 22 quests.
Why the tutorials start with safety, not process
The first two quests in the fab are about safety, not process.
Your first gowning
On your first cleanroom tour, you find a small tear in your glove in the gowning room, and the tour group has already gone in. Tucking the glove cuff under your sleeve to hide it, or skipping the air shower just this once, is Poor. Patching it with whatever tape you can find and hurrying in is Risky. Excellent is changing the glove in the proper sequence, checking the mirror, telling the group you'll be late, then going in.
What this quest teaches is less the gowning sequence than the idea that running late is never a reason to skip a step. That idea carries into every quest that follows.
When the gas alarm sounds
A toxic gas alarm goes off in the bay, and a nearby worker says it misfires sometimes and not to worry. Excellent is treating it as real: stop, evacuate by the marked route to the assembly point, and stay out until the all-clear. Finishing the five-minute step you're on first is Risky. Walking over to the gas cabinet to check for a leak is Poor. New hires need to meet, once, the case where wanting to help turns into a dangerous act.
Five calls that stop new fab engineers
1. Control limits vs. spec limits: the first response to an OOC point
This is the night-shift OOC from the opening. The options split like this.
- Keep re-measuring until a reading comes back in control, then move the lot on: Poor
- Adjust the etch-time offset before the next lot arrives to pull CD back to target: Poor
- It's still inside spec, so the product is good; release it and leave a note for day shift: Risky
- Hold the lot, work through the OCAP, record the disposition and tell the owner engineer: Excellent
Measuring until it passes is picking the data you like. Touching the recipe before you know the cause can throw a healthy process off if the gauge was the problem, and an unapproved adjustment also breaks change control (MOC). "It's in spec" may be true today, but it means nobody notices until the process drifts out of spec. Control limits and spec limits answer different questions. As a sequence: hold, OCAP, record, share.
2. Interlocks and LOTO: no bypassing
The exhaust interlock on a wet bench keeps tripping, and a technician suggests one jumper on the sensor would get the bench through this shift. Allowing the jumper for this shift only is Poor, and so is lowering the trip setpoint. Running the bench with an operator stationed at the exhaust gauge is Risky. Excellent is refusing the jumper, keeping the tool down, and checking the real exhaust flow and the sensor with facilities.
The LOTO quest has the same shape. The line says the tool is idle and the panel is dark, so skip the 20-minute LOTO and start. Excellent is isolating every energy source, having each worker apply their own lock and tag, verifying zero energy, then starting. Switching off the main breaker yourself and having a colleague guard it is Risky, because protection vanishes the moment the guard steps away, and other energy sources such as gas and pneumatics are still live.
3. Between a hot lot and Q-time
Sales wants a customer sample lot flagged as hot and put first at every step. But at the clean-to-furnace segment, lots close to their Q-time limit are already waiting.
- Put the hot lot first everywhere; any lot that blows its Q-time can be re-cleaned: Poor
- Decline the hot lot, since first-in, first-out keeps the line stable: Risky
- Flag it as hot in MES and let the operator at each tool work out the order: Risky
- Approve the hot lot, but let the expiring Q-time lots go first at the furnace and give sales a realistic date: Excellent
What a new production planner needs to learn is not how to pick one priority but how to split priorities by segment. Declining the hot lot looks principled, but it erases a real constraint, the business weight of a customer sample. A conditional yes that protects the constraint beats a no.
4. Shift handover
Thirty minutes before night shift ends, a tool went down at 4 a.m., two lots are on hold waiting for engineering disposition, and one lot is running short on Q-time. Telling the day lead "all fine except one tool, details are in MES" and heading home is Poor.
The two Risky options are the interesting ones. Pasting the whole night's MES event log into the handover note, and staying past shift end to close every open item yourself. Both look conscientious. The first buries what's urgent in a log the next shift has to interpret while Q-time keeps running, and the second leaves the next shift starting without knowing what happened. Shift work assumes handing over. Excellent is writing up tool status, held lots and owners, Q-time left and open actions, then walking the day lead through it in person.
5. Yield drop: evidence before blame
Weekly probe yield on one product drops a few points, and someone in the meeting says it must be the new etcher and wants it shut down now.
- Watch another week of data first: Poor
- Tighten final test screening to catch bad dies and keep shipping: Poor
- Inhibit the new etcher now; if yield recovers, you've found the cause: Risky
- Start with wafer map patterns, fail bins and a tool and chamber commonality check on the affected lots: Excellent
Stopping a tool without evidence costs capacity, and if the real cause is elsewhere, yield stays down. The first habit of a quality and yield engineer is to slice the data before lining up the most likely suspect.
What changes when you become Line Manager
The executive quests later in the map ask about the same principles under heavier pressure.
The quarter closes in three days, and hitting the target means shipping ten held lots that are under investigation for a borderline defect signal. Releasing all ten, or shipping them labelled as engineering samples, is Poor. Releasing five and holding five is Risky: it looks like a compromise, but there is no evidence for which half is safe. Excellent is keeping the hold, fast-tracking lot-by-lot disposition with extra testing, shipping what passes and explaining the gap early.
The other is reporting culture. An operator nearly opened a chamber that was still under pressure, and nobody reported it for two weeks for fear of getting in trouble. Disciplining the operator and shift lead is Poor. Rewarding the shift with the fewest near-misses each month is also Poor, because the easiest way to lower that number is to stop reporting, so the KPI rewards exactly the wrong behaviour. A poster campaign is Risky: the message is right, but what changes culture is what happens after someone reports. Excellent is thanking the reporter, holding a no-blame review of system causes and fixing things visibly on every shift.
Cautions when building fab onboarding scenarios
- Quests teach the order of decisions, not numbers and recipes. Real control limits, re-measurement rules and LOTO steps must come from the company's own OCAPs and SOPs.
- A fictional company is the safer setting. Real tool or customer names lengthen security review and make sharing harder.
- Safety quests do not replace mandatory safety training or on-site training. Use them to practice recalling the rules under pressure.
- Keep the links between roles. The same alarm is an OCAP question for a process engineer, a Q-time question for production control and a customer notification question for quality.
Checklist for semiconductor onboarding scenarios
- Is there a quest that separates control limits from spec limits?
- Is bypassing an interlock or skipping LOTO wrong in every situation?
- Are there quests that resolve priority conflicts by segment and scope?
- Does the handover quest practice what to write and who to walk through it with?
- In yield problems, does data analysis come before naming a culprit?
- Do the manager quests show a culture where near-misses are not punished?
- Are real recipes, figures and customer names kept out?
For the design rules shared across industries, read Industry-specific onboarding scenarios. The construction map is covered in Construction site onboarding scenarios. Root cause analysis and management of change, both central in a fab, are covered in more depth in Root cause analysis beyond 'operator error' and Management of change.
Try workplace scenarios as onboarding quests
Pick the semiconductor fab in a new game and start from your first day in a bunny suit. Play the Onboarding