FinFET 2010s Pattern Fin with Sidewall Image Transfer Double Patterning
# FinFET (2010s): Step 4 — Pattern the Fin With Sidewall-Image-Transfer Double Patterning
## 1. Why the Scanner Alone Cannot Print the Fin Pitch
Step 3 fixed the fin's governing dimensions; actually placing dense arrays of 8-10nm-wide fins at the roughly 40-60nm pitches early FinFET nodes required went beyond what contemporary 193nm immersion scanners could resolve directly as a single exposure. Rather than wait for a shorter-wavelength scanner, the industry adopted a patterning technique that multiplies the pitch a given scanner can achieve without changing the scanner itself.
## 2. Sidewall-Image-Transfer, Step by Step
A sacrificial mandrel line is first patterned by the scanner at a pitch twice as coarse as the final fin pitch. A conformal film is then deposited over the mandrel and etched back, leaving only sidewall spacers standing on either side of each mandrel line. The mandrel itself is removed, leaving twice as many fin-defining lines, at half the original pitch, as the scanner's own resolution ever printed — a spacer deposition and an etch-back step, not the lithography, is what actually sets the fin pitch.
## 3. Self-Aligned Double and Quadruple Patterning
Running the sidewall-spacer process once (self-aligned double patterning, SADP) doubles the mandrel's pitch; running it twice in sequence (self-aligned quadruple patterning, SAQP) quadruples it, letting later FinFET generations reach sub-30nm fin pitches from scanners that could never print that pitch unassisted. Each additional spacer cycle buys pitch reduction at the cost of tighter overlay and line-edge-roughness budgets, since errors from every prior cycle accumulate into the final fin pattern.
## 4. What This Step Does and Does Not Cover
This step covers how the fin pitch is defined; it does not yet address how the fin's own height and width are made uniform once patterned and etched, which Step 5 takes up as a distinct variability concern.