APC patents in this corpus cover the feedback systems that adjust fab equipment settings based on sensor and metrology data, including run-to-run control, virtual metrology, and chamber-drift compensation. They typically pair a control-systems claim (IPC G05B) with a semiconductor-device or process claim (IPC H01L), which is why both subclasses appear in the majority of the 82 families reviewed here. The distinguishing feature versus general statistical process control is the closed-loop adjustment of equipment recipes rather than simple monitoring.

The filing trend in this dataset peaked at 4 in 2018 and dropped to around 1 per year by the 2022 midpoint, suggesting the core control-loop architectures were claimed early and later filings have been incremental refinements. Part of the apparent decline toward 2026 is a publication-lag artifact, since patents filed in the last 18 months typically have not published yet. Even accounting for that lag, the absence of a filing surge suggests this is a well-established rather than fast-moving claim space.

The assignee set in this corpus includes fab equipment vendors, major foundries, and semiconductor manufacturers, with Tokyo Electron showing the most co-assignee filing activity of any tracked entity. Notably, every major assignee tracked for recent-year momentum — including Tokyo Electron, Applied Materials, TSMC, AMD, and SMIC — shows zero filings in the most recent tracked year, which likely reflects publication lag rather than an actual halt in filing. Ranking by total family count, rather than recent momentum, gives a clearer picture of long-run investment.

US6556884B1 claims the specific interface between a statistical process control (SPC) system and a manufacturing process control framework, where both SPC and APC analyses feed back into modifications of the manufacturing facility itself. Anyone building a system that runs both SPC and APC analysis on the same processed lot and uses both outputs to adjust the facility needs to check this claim's scope carefully, since it sits close to that exact architecture. Designing around it typically means separating the SPC and APC feedback paths, or routing the facility modification through a different intermediate model than the one this patent specifies.

The thinnest IPC coverage in this corpus sits in G06N (AI-based computing) and G01R (electrical measurement), each appearing in only 3 of 82 families, despite the search terms explicitly targeting sensor-fusion and virtual-metrology language. That gap suggests AI-driven virtual metrology models and sensor-fusion fault detection are more discussed than claimed, leaving room for a well-drafted first claim that ties a specific machine-learning model architecture to a specific fab sensor input. Cross-tool yield correlation is a related area with limited dedicated coverage in this dataset.