Planar, FinFET and GAA: what actually changes
Planar MOSFETs, FinFETs and gate-all-around (GAA) transistors differ in how the gate controls the channel geometry. A conventional planar device controls a channel near a flat surface. A tri-gate FinFET places the gate along the top and sides of a fin. GAA surrounds the channel cross-section, as in a nanowire or nanosheet. The gate remains separated from the semiconductor by a dielectric.
This comparison is about geometry and electrostatic control. It is not a ranking of finished chips: contacts, materials, interconnect, operating conditions and circuit design also affect performance, power and cost.
Process map
14nm FinFET
Locate fin-related and gate modules in the 14nm FinFET overview as one manufacturing reference; this destination is not a GAA flow.
Compare the same cross-section
The diagram cuts across the channel, perpendicular to the source-to-drain direction. Current flows into or out of the page; the source and drain are outside this section. Using the same viewing direction prevents a fin's height from being mistaken for gate length.
| Structure | Gate relationship to the channel | What to look for in a section |
|---|---|---|
| Conventional planar MOSFET | Gate above the surface channel | A gate dielectric between gate and substrate surface |
| Tri-gate FinFET | Gate on the fin's top and two sides | A narrow raised semiconductor fin |
| GAA nanosheet | Gate surrounds each sheet's perimeter | Semiconductor sheets separated from the gate by dielectric |
The FinFET drawing represents the common tri-gate form; the broader FinFET family includes other gate configurations. “GAA” describes gate coverage, while “nanosheet” describes a channel shape. GAA does not require a circular channel.
Conceptual teaching diagram; independently drawn from the cited principles, not a course process section.
Why additional gate control helps
As a transistor's channel becomes short, the drain can increasingly influence the potential barrier that the gate is intended to control. Bringing the gate around more of a thin channel gives it stronger electrostatic influence, helping manage short-channel behavior. Channel dimensions and gate-dielectric properties still matter; the architecture label alone does not establish leakage or switching performance.
Imagine trying to infer a device's behavior from a colored section without dimensions, materials or bias conditions. You can identify which surfaces face the gate, but you cannot calculate a credible speed or power advantage. Numerical comparisons need matched assumptions and measured or appropriately validated device data.
What changes in manufacturing
Planar fabrication defines devices around a wafer surface. FinFET integration additionally has to form and preserve a narrow raised fin, then build a gate structure around its accessible surfaces. Fin geometry and variation become part of the integration problem.
For a common stacked-silicon-nanosheet approach, alternating channel and sacrificial layers enable the channel sheets to be released before the surrounding gate stack is completed. Selective removal must preserve the channel, and the spaces around the sheets must accommodate the dielectric and gate materials. Inner spacers, source/drain connections and contacts add related constraints. The ordering here explains the integration concept; implementations vary.
The extra gate coverage therefore creates opportunities and fabrication obligations together. A visually complete gate wrap does not prove low contact resistance, uniform sheets, good interfaces or acceptable yield.
Keep three comparisons separate
Geometry is not gate material. High-k metal gate describes a dielectric-and-electrode material system. Planar versus FinFET versus GAA describes channel and gate arrangement. These are related design choices, not mutually exclusive alternatives.
A node name is not a dimension measurement. Use actual gate length, fin or sheet dimensions and the stated device structure when interpreting technical results. Do not infer every dimension from a process-node label.
An architecture is not a chip benchmark. Better electrostatic control can help device scaling, but a chip's behavior also includes wiring, loading, memory, circuit choices and operating conditions. This article makes no fixed percentage claim for speed, power or area.
Connect the comparison to a reference process
Read FinFET fundamentals and fin-process physics, then open the 14nm FinFET flow overview. Locate the fin-related modules and gate formation to see how a geometric concept becomes a sequence of manufacturing tasks.
The linked course is a FinFET reference. It does not provide a GAA manufacturing flow. The GAA comparison is completed here through the conceptual section, the manufacturing distinctions and the cited literature. After reading, you should be able to identify the channel, gate and dielectric in all three structures, explain why the viewing direction matters, and state what further evidence would be needed to compare performance.
References
Silicon VLSI Technology - Full
James D. Plummer, Michael D. Deal, Peter B. Griffin
Silicon VLSI Technology · ISBN 978-0130850379
FinFET to GAA MBCFET: A Review and Insights
Rinku Rani Das, T. R. Rajalekshmi, Alex Pappachen James · IEEE Access