Contact over Active Gate
Contact-over-active-gate, universally abbreviated COAG, is an advanced integration technique that allows metal contacts to land directly over the gate electrode inside the active transistor region itself, rather than requiring every gate contact to land only over field isolation outside the active area — a constraint that previously forced standard cell layouts to reserve extra space purely to give gate contacts somewhere safe to go. Once a sufficiently robust self-aligned contact cap exists to guarantee no short occurs even directly over the active channel, COAG removes that reserved space entirely, which is exactly why it became essential for competitive standard cell area at 7nm and below.
Traditional gate-contact placement rules force every standard cell to reserve dedicated space over shallow trench isolation specifically so the gate contact has a landing zone where a short to the active channel is geometrically impossible — space that contributes nothing to the transistor's actual electrical function and exists purely as a placement safety margin. This isolation-only placement rule, while safe, directly inflates standard cell area at every node that follows it, because that reserved isolation margin scales far less favorably than the active transistor region itself does as dimensions shrink — making it an increasingly disproportionate fraction of total cell area at each successive node.
COAG eliminates that reserved isolation margin by placing the gate contact directly over the active channel region instead, which is only safe because the same self-aligned contact cap that already protects against shorts at the edge of the gate provides equivalent protection directly above the gate as well — COAG doesn't introduce a new protection mechanism, it simply extends trust in the existing one to a new location. This is exactly why COAG is correctly understood as building on [self-aligned contact](self-aligned-contact-process) technology rather than as an independent innovation: the SAC cap's demonstrated reliability at the gate edge is the entire justification for trusting it directly over the gate's active region as well.
Standard cell height reduction of roughly one to two fin pitches is COAG's direct, quantifiable payoff, and because fin pitch is itself one of the most tightly scaled dimensions in an advanced node's design rules, reclaiming even one or two of them from every single standard cell in a design compounds into a substantial overall die-area reduction across a design containing many millions of cell instances. This area reduction is exactly why COAG became a required technique rather than an optional one at 7nm and below — a design that can't use COAG simply can't match the standard cell area that a design using it achieves, putting any foundry process without mature COAG capability at a direct competitive disadvantage on cell density.
COAG also gives place-and-route tools meaningfully more contact-placement freedom, since contacts are no longer artificially restricted to isolation-only landing zones, which in turn gives the automated design flow more flexibility to optimize routing congestion and cell-internal connectivity without being constrained by contact-placement rules that have nothing to do with the actual circuit being implemented. This additional placement freedom is a secondary but genuinely valuable COAG benefit beyond the direct area reduction itself — it removes an artificial geometric constraint that previously had nothing to do with circuit function, letting the place-and-route tool optimize purely against actual design objectives instead.
COAG's robustness requirement is uncompromising in a way few other integration techniques are, because the SAC cap protecting against a gate-to-contact short now has to perform reliably in the single worst-case geometry possible — a contact landing directly centered over the gate — rather than merely at the gate's edge, which is exactly why COAG's production readiness at any given node depends entirely on how mature and well-controlled that node's self-aligned contact process has actually become. A foundry attempting COAG before its SAC process is sufficiently mature and well-controlled risks a meaningfully higher gate-to-contact short rate than a traditional isolation-only contact scheme would ever produce — which is exactly why COAG adoption timing at each node tracks so closely with SAC process maturity rather than being an independent design-rule decision.