Self-Aligned Contact

Self-aligned contact, almost universally called SAC, is a process integration technique where the source/drain contact opening is defined by the gate's own dielectric spacers and cap rather than by a separate, independently aligned lithography step — which lets the contact sit immediately adjacent to, or even partially overlapping, the gate without risking an electrical short between them. SAC has been standard at every advanced node since roughly 14/10nm, precisely because it solves a problem that lithographic overlay control alone genuinely cannot solve at these dimensions.

Self-Aligned Contact: Letting the Spacer Do the Work The gate cap and spacer, not a separate mask, define exactly where the contact etch stops. WITHOUT SAC — LITHOGRAPHY MUST GUESS gate contact overlay drift → short risk WITH SAC — SPACER DEFINES THE EDGE SAC cap (SiN) gate contact etch stops on cap/spacer — contact can even overlap safely THE ETCH-SELECTIVITY REQUIREMENT The contact etch must remove the surrounding interlayer dielectric aggressively while barely touching the SAC cap and spacer — a large etch-rate contrast between two dielectrics (typically oxide versus nitride) is what makes the entire scheme work at all. Self-alignment replaces lithographic precision with etch-chemistry precision — exactly what scaling below ~14nm demands.

The SAC cap — a dielectric layer, typically silicon nitride, deposited directly on top of the metal gate stack — is what makes self-alignment physically possible, because it gives the subsequent contact etch something it can stop on reliably even if the contact pattern itself drifts slightly off its nominal target position. Without this cap, a contact landing even slightly off-target risks etching straight into the gate itself, creating a direct short; with the cap in place, the etch chemistry — not the lithography alignment — becomes the thing actually responsible for keeping the contact electrically isolated from the gate.

Etch selectivity between the surrounding interlayer dielectric (typically silicon dioxide) and the SAC cap and spacer (typically silicon nitride) is the single physical property the entire self-aligned contact scheme depends on, because the contact etch has to aggressively remove the oxide to open the contact hole while barely touching the nitride cap and spacer protecting the gate underneath. A given etch chemistry's oxide-to-nitride selectivity ratio — how much faster it removes one material than the other — directly sets the margin for error the contact etch has before it starts eating into the gate protection it's supposed to respect, which is exactly why etch-chemistry development, not lithography improvement, is the limiting discipline behind pushing SAC to ever-tighter gate-to-contact spacing.

Eliminating the need for tight lithographic overlay between the contact layer and the gate layer is SAC's most direct practical payoff, because at advanced nodes the gate-to-contact spacing required by device scaling is routinely tighter than what overlay control between two independently patterned lithography layers can reliably guarantee. Rather than asking lithography to hit an overlay target that's simply below what the tool and process are capable of achieving reliably at volume, SAC sidesteps the overlay requirement entirely by making the contact's final position self-defining relative to the gate it needs to avoid — converting an overlay problem that lithography alone cannot solve into an etch-selectivity problem that process chemistry can.

Contact fill, using a metal such as tungsten, cobalt, or increasingly ruthenium at the most advanced nodes, completes the self-aligned contact by filling the etched opening and making the actual electrical connection to the source or drain region beneath it — a step that has to accommodate the narrower, more aspect-ratio-constrained openings that the whole self-aligned approach tends to produce. The specific metal choice matters here because narrower, taller contact openings created by aggressive scaling increasingly favor metals with better gap-fill characteristics and lower resistivity at very small cross-sections, which is exactly why contact-fill metal selection has itself evolved alongside SAC scaling rather than staying fixed across node generations.

SAC is the specific process foundation that [contact-over-active-gate](contact-over-active-gate-integration) integration directly builds on, since COAG's defining move — placing contacts directly over the active gate region rather than only over field isolation — is only safe to attempt once the SAC cap and spacer can be trusted to reliably prevent a short even when the contact deliberately overlaps the gate, rather than merely sitting adjacent to it. This dependency is exactly why COAG only became a viable, production-worthy technique after SAC itself had matured into a well-controlled, reliable process — COAG is best understood as pushing SAC's existing self-alignment guarantee to its logical limit, not as an independent technique layered on top of it.

Take self-aligned contact further

Ask the copilot about this term, or have our engineers assess it against your process.

Ask CFSGPT about this → Request a technical assessment → Open the CFS app