When lithography cannot resolve dense pitch, sacrificial cores guide spacer growth. Mandrel spacer patterning removes cores to transfer tighter lines.
How the mandrel becomes a pattern
Mandrel-spacer patterning begins with a lithographically defined sacrificial core, or mandrel. A conformal film coats its sides; directional etching removes much of that film from horizontal surfaces, leaving sidewall spacers. After the core is removed, the remaining spacer pattern can transfer a denser line pattern into a hardmask or underlying layer. A published spacer-based self-aligned double-patterning (SADP) example describes a core, deposited spacers, core removal and pattern transfer in that order . This is the sequence behind pitch division, not a claim that one material or deposition method is mandatory.
The public 7nm FinFET flow overview locates self-aligned pitch division within a broader integration flow. The mandrel removal sequence is explained in this article. Individual course Steps have separate access rules; the overview does not promise their unlocked text.
Process map
7nm FinFET
Use the public 7nm FinFET overview to locate self-aligned pitch division in a full flow; individual Step access is separate.
Public flow overview
Double and quadruple patterning
In SADP, spacers formed on both sides of the original core can create a line density beyond a single printed core pattern, subject to subsequent cutting and transfer. A second spacer-defined cycle can be used for self-aligned quadruple patterning (SAQP). A published SAQP study demonstrates this second cycle for a particular hardmask integration and identifies possible STI and gate-pattern transfer applications . Neither “double” nor “quadruple” means that every final feature is automatically usable: cuts, line ends and layout constraints remain.
Spacer deposition can use different methods. The cited SADP demonstration used CVD spacers , so an article that requires ALD for every mandrel-spacer process would be inaccurate. The final dimensions reflect core profile, deposited-film uniformity, directional etch and pattern transfer together. Spacer thickness alone is not an unconditional final critical-dimension rule.
Where variation enters
Core critical-dimension variation, spacer asymmetry and etch differences can produce alternating spacing often called pitch walking. Pattern transfer can add further roughness or bias. These mechanisms must be checked at the relevant process stage; seeing a regular mandrel array does not prove a regular final fin array. The SADP and SAQP examples establish workable integration sequences, not universal yield or node-wide superiority .
The same distinction matters when reading the course flow. The 7nm overview locates self-aligned patterning, while this article supplies the mandrel sequence; it does not quantify a particular spacer profile, etch selectivity or defect rate. If the reader only needed the definition and sequence, this article may be sufficient; the overview adds the position of that sequence in a complete FinFET flow.
Does pitch division also specify final line width?
No. Pattern repeat distance and the width of one feature are different quantities, and the transfer bias discussed above can change final edges. Practice marking both in the center-spacing and edge-dimension comparison, then apply the distinction to the mandrel, spacer and transferred pattern. The gate-line example teaches how to read dimensions; it does not equate mandrel pitch with final gate pitch or fin width.
References
22nm half-pitch patterning by CVD spacer self alignment double patterning (SADP)
C. Bencher, Yongmei Chen, H. Dai, W. Montgomery, L. Huli · SPIE Advanced Lithography
Sidewall spacer quadruple patterning for 15nm half-pitch
P. Xu, Yongmei Chen, Yijian Chen, Liyan Miao, Shiyu Sun, Sung-Woo Kim et al. · Advanced Lithography
Silicon VLSI Technology - Full
James D. Plummer, Michael D. Deal, Peter B. Griffin
Silicon VLSI Technology · ISBN 978-0130850379
Modern Semiconductor Devices for Integrated Circuits - MOS Transistor
Chenming Hu
Modern Semiconductor Devices for Integrated Circuits · Ch6 MOS Transistor