FinFET scaling has long depended on making the transistor’s channel narrower while maintaining control over current flow. A reported atomic-layer FinFET takes that goal to a physical limit: its vertical channel is a single layer of two-dimensional material, about 0.6 nanometers thick. The experimental devices demonstrate a route beyond the few-nanometer fin widths that conventional patterning can readily produce, while also highlighting the performance challenges that remain before the approach can be considered for practical electronics.
Why pursue an atomic-layer FinFET?
A field-effect transistor controls current through a semiconductor channel using an electrically insulated gate. As devices shrink, a planar channel becomes harder to control effectively. At dimensions below roughly 10 nanometers, quantum confinement and short-channel effects can complicate transistor behavior. A FinFET addresses this challenge by extending the channel vertically and placing gate material around multiple sides, improving the gate’s ability to control the channel.
Conventional FinFETs are generally made by etching fins from a semiconductor wafer. That top-down approach is constrained by lithography precision; the paper describes a narrowest line width of approximately 6 nanometers for the lithography tools discussed. The new work takes a different route: instead of etching a very narrow fin, it grows a two-dimensional crystal so that the crystal’s single-layer thickness defines the fin width.
The distinction matters for semiconductor scaling. A thin fin can reduce the channel dimension across the device, while its vertical height remains substantial. In the reported structure, the fin is approximately 0.6 nm wide and supported along a step about 300 nm tall. The research therefore demonstrates a channel geometry that would be difficult to define through conventional fin etching alone.
How the vertical channel is fabricated
The researchers used a bottom-up template-growth process. They first prepared silicon-on-insulator substrates with sharp, approximately 300 nm-high step edges. A 10 nm hafnium oxide (HfO₂) coating was applied to the sidewall to support and protect the material during subsequent processing. Two-dimensional crystals were then grown across the step, forming a thin layer on the vertical surface.
The process used wet-spray chemical vapor deposition to grow monolayer transition metal dichalcogenides (TMDs), including molybdenum disulfide (MoS₂) and tungsten disulfide (WS₂). The researchers also demonstrated carbon nanotube films in related structures. To retain the vertical channel, they removed the material deposited on planar areas, then etched away the silicon step. The remaining supported channel was positioned between source and drain electrodes.
A further HfO₂ coating served as the gate dielectric, after which gate electrodes were added. The gates could be made from deposited metal or carbon nanotube film. This sequence illustrates a central fabrication tradeoff: the monolayer defines an exceptionally narrow channel, but the device still depends on carefully controlled growth, support, etching, dielectric deposition, and electrode formation.
Measured performance and remaining limitations
The demonstrated MoS₂ monolayer FinFETs showed on/off current ratios ranging from 10² to 10⁷ across tested devices. The best reported subthreshold swing was 300 mV per decade. Extracted carrier mobilities were between 1 and 6 cm²/V·s, which the authors identify as relatively low and an area requiring improvement. They attribute the limited mobility in part to defective scattering centers associated with the wet-spray growth process.
These measurements are important to interpret alongside the device’s remarkable geometry. An atomic-scale fin width does not by itself establish that a transistor meets the requirements of a production process. Mobility, switching behavior, reproducibility, and integration all matter. The paper presents the devices as a fabrication demonstration and notes that electrical performance must improve for future applications.
The researchers also used COMSOL Multiphysics to model an idealized device with a 4 nm gate length and a 0.65 nm fin. The simulations estimated strong gate control and reduced short-channel effects under the modeled conditions. Those results are projections, not measurements from a fabricated 4 nm-gate device, so they should not be confused with the experimental on/off ratios or mobility data.
What the result means for electronics development
The work offers a different way to think about transistor scaling: use the thickness of an atomically thin material to set the channel width, rather than relying exclusively on finer lithographic patterning. The study demonstrated multiple fin materials and a fin-array structure, with a minimum pitch of 50 nm under the reported experimental conditions. That is a research result, not evidence of a ready-to-deploy manufacturing process, but it establishes a possible direction for future nanoelectronics.
For engineers and technical decision-makers, the practical takeaway is to separate geometric innovation from system readiness. The single-layer channel addresses one scaling constraint, while the reported mobility and fabrication process identify areas that need further development. Any assessment of the approach should consider both the potential for compact, gate-controlled devices and the work required to improve material quality and electrical performance.
Conclusion
The atomic-layer FinFET demonstrates that a vertically grown two-dimensional crystal can serve as a channel just one atomic layer thick. Its high measured on/off ratios show promising switching behavior, but relatively low mobility and the process’s research-stage status remain significant limitations. The result expands the options for transistor architecture; realizing practical benefits will depend on improving growth and device performance, not simply shrinking the fin further.