Imec finds contact formation caps 2D GAA CFET pitch at 36 nm for the A2 node
Imec's July 2026 arXiv paper shows 2D GAA CFETs offer no contacted poly pitch advantage over silicon at A2, with contact resistance and parasitic capacitance as the binding constraints.

Researchers at imec published a paper on 14 July 2026 concluding that gate-all-around complementary field-effect transistors built on atomically thin two-dimensional semiconductors do not deliver the pitch scaling benefit that had been widely anticipated for the A2 logic node[1]. The key number: contact formation constraints impose a minimum contacted poly pitch (CPP) of 36 nm for 2D GAA CFETs - identical to the floor already set by silicon GAA CFETs[1].
What the paper examined
The imec team, led by Fengben Xi and 17 co-authors, developed an A2-oriented 2D CFET integration flow targeting a CPP of 36 nm and a gate length (Lg) of 10 nm, and demonstrated several key process modules[1]. To evaluate the architecture rigorously, they built a multiscale power-performance-area (PPA) framework that spans:
- Quantum transport simulations
- Compact-model generation
- A2-targeted GAA integration-flow definition
- Parasitic extraction
- Circuit-level benchmarking[1]
The channel materials under study are transition-metal dichalcogenides (TMDs) - semiconductors such as MoS₂, WS₂, and WSe₂ that can be deposited in layers only a few atoms thick. Their atomically thin body is attractive because it suppresses short-channel effects at gate lengths below 10 nm, where conventional silicon channels begin to lose electrostatic control.
Where the expected gains evaporate
The analysis found that two non-idealities dominate device behavior and erode the theoretical PPA advantage:
- High contact resistance. The interface between a three-dimensional metal contact and an atomically thin 2D channel creates current-crowding effects that drive up resistance, a problem the field has tracked for years but has not yet solved at production-relevant pitches.
- Layout-induced parasitic capacitances. At A2 ground rules, parasitic capacitances arising from the cell layout become the dominant capacitive load, not the intrinsic gate capacitance.
Architectural optimization can improve the effective current-to-capacitance ratio (Ieff/Ceff), but the associated rise in absolute capacitance limits circuit-level gains[1]. In other words, tuning the device geometry to push one metric tends to worsen another, leaving net circuit performance below what the intrinsic channel properties would suggest.
The paper's conclusion is direct: meaningful progress will require co-optimization of contacts, transport, and parasitics, together with 2D-specific CFET architectures[1].
Context in imec's broader roadmap
Imec's logic roadmap places CFET introduction in manufacturing at the A7 node, with silicon-channel devices as the baseline[1]. The 2D-channel variant is positioned as a potential route to extend scaling further - but the July paper makes clear that the path is not straightforward. A separate June 2026 result from imec, ASML, and TSMC demonstrated scaled MoS₂ nFETs and WSe₂ pFETs with a 50 nm CPP on 300 mm wafers, described as a step toward fab readiness - but that CPP is still wider than the 36 nm A2 target, underscoring how much contact engineering remains to be done.
The imec paper sets a concrete benchmark for what the field must achieve: contact resistance and parasitic capacitance co-optimization that is specific to 2D CFET geometries, not borrowed from silicon integration flows. Progress on those two fronts - and whether 2D-specific cell architectures can be defined that avoid the layout parasitics seen here - will determine whether 2D GAA CFETs become a viable option at A2 or remain a post-A2 candidate.
Written by Electronics Insider's automated desk from the sources above and published automatically. How we work.
Related
Design & EDAETH Zürich and NYU show JEDEC test loops miss worst-case HBM2 power by 23.6%
Researchers at ETH Zürich and NYU tested 36 real HBM2 chips and found JEDEC measurement loops understate peak power by 23.6%, with read ops and data-bus toggling as the main drivers.
15 Oct 2026
Design & EDAFlowAccel's Joel Jorgensen argues chip teams must design their workflows before AI can help them hit tape-out
Joel Jorgensen of FlowAccel argues that AI tools alone won't save a tape-out schedule - teams must first design the cross-team workflows that let those tools compound into real gains.
15 Oct 2026
SemiconductorsETH Zürich and NYU find JEDEC test loops underestimate HBM2 power by up to 35.4%
Researchers at ETH Zürich and NYU tested 36 real HBM2 chips and found JEDEC-specified current loops can underestimate peak power by up to 35.4%, with implications for AI accelerator thermal design.
15 Oct 2026