Corrective Maintenance

Corrective maintenance (CM) is unscheduled repair work performed after a tool has already failed or drifted out of spec — as opposed to preventive maintenance (PM), which happens on a fixed schedule before anything has gone wrong. CM starts with a fault (an alarm, a failed qual wafer, a chamber that won't strike) and ends with the tool restored to its qualified state; PM starts with a calendar or usage interval and aims to prevent that fault from happening at all. A fab tracks both, but for opposite reasons: PM frequency is a cost-and-risk tradeoff a fab controls, while CM frequency is a symptom of how well that tradeoff — and the equipment's underlying reliability — is actually working.

Mean time to repair (MTTR) is the metric CM is measured against, and every minute of it is lost chamber capacity with no process running. Unlike PM, which a fab schedules during planned downtime windows that minimize production impact, CM happens whenever the fault occurs — potentially mid-shift, mid-lot, with wafers already committed to a process that can't finish. Fabs invest heavily in fast fault diagnosis (clear alarm codes, accessible diagnostic logs, spare-parts inventory for common failures) specifically to compress MTTR, because unlike a PM's planned duration, CM duration is pure unplanned downtime stacking directly against the tool's overall availability number.

A high CM rate is diagnostic information about the PM schedule, not just a record of bad luck. If a specific failure mode keeps generating corrective-maintenance events between scheduled PM intervals, that's a signal the PM interval for the component involved is set too loosely — reliability engineering teams specifically track CM events by failure mode and feed that data back into PM interval decisions, tightening the schedule for components that fail early and potentially loosening it for ones that reliably outlast their current interval with margin to spare. This feedback loop is what preventive maintenance scheduling actually optimizes against over time, rather than being set once and left fixed.

Requalification after CM is non-negotiable in the same way it is after chamber-mount-docking or a process-kit swap, because a repaired tool isn't automatically a qualified one. Replacing a failed RF cable, retuning a drifted match network, or swapping a worn process kit as a corrective action all change the chamber's physical state in ways that need confirmation before product wafers resume — a short monitor-wafer run checked against the golden reference chamber's historical performance, the same qualification gate this site's chamber calibration entry describes for new or recalibrated chambers, applies equally to a chamber coming out of corrective maintenance.

Corrective maintenance will always exist alongside preventive maintenance — no PM schedule eliminates all unscheduled failures — but a fab's CM rate, tracked over time and broken down by failure mode, is one of the clearest signals of whether its overall maintenance strategy is actually working or just reacting.