Multi-patterning is the family of manufacturing techniques that print circuit features at a finer pitch than a single lithographic exposure can resolve, by splitting the pattern across two or more exposure-and-etch steps. When the target pitch drops below what one exposure can cleanly image, adjacent features blur together; multi-patterning sidesteps this by decomposing the layer so that each individual exposure only ever prints a relaxed, printable pitch, and the steps combine on the wafer into the dense final pattern. It was the workhorse that carried 193nm immersion lithography from roughly 40nm pitch down toward 20nm and below before EUV, and it remains essential even in the EUV era for the tightest layers. The two families are litho-etch (LELE) and self-aligned spacer (SADP/SAQP).\n\nLitho-etch multi-patterning (LELE) splits the pattern across colored masks. The layout is decomposed — 'colored' — into two or more sub-masks, each holding a subset of the features spaced far enough apart to expose cleanly. The wafer is patterned and etched with mask A, then the process repeats with mask B interleaving between A's features (litho-etch-litho-etch), and adding more colors (triple patterning, LELELE) pushes the pitch finer still. LELE is flexible about geometry, but its critical weakness is overlay: because the final spacing between an A feature and a B feature is set by how accurately mask B aligns to mask A, any misalignment becomes pitch variation and edge-placement error — and each added mask multiplies cost and cycle time.\n\nSelf-aligned patterning (SADP/SAQP) uses spacers so the pitch comes from deposition, not alignment. SADP prints a relaxed-pitch sacrificial pattern of mandrels (cores), conformally deposits a thin film over them, then etches it anisotropically so the film survives only as sidewall spacers on each mandrel. Removing the mandrels leaves two spacers per original line — doubling the feature density — and the new pitch is set by the deposited film thickness, which is controlled to the angstrom and is identical everywhere, so there is no mask-to-mask overlay error. Self-aligned quadruple patterning (SAQP) repeats the spacer step to quadruple density. The trade is rigidity: spacers naturally form closed loops of uniform lines, so SADP needs additional cut and block masks to carve those lines into real circuit shapes.\n\n| | LELE (litho-etch) | SADP / SAQP (self-aligned) |\n|---|---|---|\n| How | color layout into N masks, expose+etch each | spacers on mandrel sidewalls |\n| Pitch set by | mask-to-mask overlay | deposited spacer thickness |\n| Density gain | ÷2 per color (2×, 3×…) | 2× (SADP), 4× (SAQP) |\n| Strength | flexible geometry | no overlay, uniform pitch |\n| Weakness | overlay error, cost per mask | regular lines only, needs cut mask |\n| Eased by | one EUV exposure at tightest layers | still used even with EUV |\n\n``svg\n\n``\n\nMulti-patterning trades exposures, masks, and cost for resolution. Every extra patterning step adds masks, deposition and etch operations, metrology, and yield risk, so multi-patterning is expensive in both dollars and cycle time — a single triple- or quadruple-patterned layer can dominate a mask set's cost. This is precisely the economic pressure that justified EUV: one EUV exposure can replace several 193i multi-patterning steps at the tightest layers, simplifying the flow. But EUV itself now needs multi-patterning at the very densest layers of leading nodes, so the technique never went away — it moved up the stack. It also feeds back into design: the coloring must be possible, which imposes multi-patterning-aware design rules (no odd cycles in the conflict graph) on the layout itself.\n\nRead multi-patterning through a quant lens rather than a 'do lithography twice' lens: the number it moves is pitch, driven below the single-exposure limit by paying in exposures — and the two families spend that payment differently. LELE buys arbitrary geometry but makes the final pitch a function of overlay, so its error budget is really an alignment budget that worsens with every added color. SADP buys a pitch defined by film thickness, essentially removing overlay from the equation, but constrains you to regular gratings that a cut mask must then edit. The design question at each node is which is cheaper: more colored masks whose yield falls with overlay, or a self-aligned flow plus the cut masks to make it useful — until one EUV exposure undercuts both.