The Evolution of CD-SEM: 3D Nanometrology in the 2nm Era

Why 2D CD-SEM Has Reached Its Limit — And Why 3D Metrology Becomes Mandatory at 2nm

As leading foundries move from FinFET to Gate-All-Around (GAA) nanosheet transistors at 2nm and beyond, dimensional control shifts from a mostly top-down problem to a fully three-dimensional one. Traditional 2D (top-down) CD-SEM still excels at fast in-line linewidth and space measurements, but it can only see a surface projection. For nanosheets, many yield-driving parameters are buried or profile-dependent, which is why the industry is converging on 3D-capable CD-SEM workflows as a practical necessity for advanced-node process control.

---

CD-SEM in One Page: What It Measures Well (and Why It Became the Workhorse)

CD-SEM is a production-optimized SEM configured for automated, high-throughput critical-dimension measurement on patterned wafers. It scans a finely focused electron beam and collects secondary electrons to form images, then uses measurement algorithms to extract CDs such as line width, space, and contact-hole size.

Why it dominated for decades:

• Non-destructive and in-line compatible
• High throughput with recipe-driven automation
• Strong repeatability for top-down CDs

These strengths carried planar CMOS and early FinFET generations, where many control parameters were accessible from a top view.

---

Why 2D CD-SEM Breaks Down for Nanosheets at 2nm

Nanosheets turn CDs into a 3D parameter set

GAA nanosheet devices stack multiple horizontal Si channels surrounded by a wrap-around gate. That creates critical parameters that a top-down image cannot directly observe, including:

• Individual nanosheet thickness and width per sheet
• Inter-sheet spacing after sacrificial layer removal
• Inner spacer recess depth and profile
• Sidewall angle and full 3D profile shape

A top-down measurement can report an “outer” contour, but it cannot reliably tell whether every sheet in the stack is within spec or whether buried cavities and spacers are correctly formed.

---

How CD-SEM Is Evolving into 3D (Without Sacrificing the Fab)

The most practical path is not “one perfect 3D tool,” but CD-SEM platforms adding 3D sensitivity through a few complementary approaches.

1) Multi-angle and tilt imaging

By imaging at multiple tilt angles, CD-SEM can capture parallax information that helps infer sidewall angles and profile-related parameters.

Key trade-offs:

• Reduced throughput per site
• Higher cumulative dose and potential material sensitivity concerns
• Limited access to deeply buried features

2) Physics-based, model-driven 3D reconstruction

Model-based measurement fits a simulated SEM signal to measured images by adjusting a 3D geometry model (widths, heights, angles, layer thicknesses) until the prediction matches the data. This can extract profile parameters that simple edge detection cannot.

The main challenge for nanosheets is complexity: more stacked layers means more parameters, more correlations, and more calibration work.

3) Higher-voltage operation for sub-surface contrast (where applicable)

Higher accelerating voltages can increase interaction volume and support backscattered-electron contrast that is sometimes more sensitive to buried layers and material changes. In practice, this is used as a complementary mode rather than a universal replacement for low-voltage surface-sensitive CD work.

---

Why Hybrid Metrology Matters (CD-SEM plus complementary physics)

Nanosheet process control increasingly relies on hybrid metrology, where CD-SEM data is combined with other techniques to resolve parameter ambiguities:

• OCD scatterometry for fast, model-based extraction of periodic structure parameters, including some buried sensitivity
• X-ray methods for deeper penetration and layer or interface information in selected use cases
• AFM and TEM primarily as reference and calibration methods rather than in-line workhorses

---

What Engineers and Equipment Buyers Should Validate Now

1. Nanosheet-relevant 3D modes, not just “higher resolution” claims.
2. A clear throughput vs. accuracy plan for multi-angle or model-based recipes.
3. A realistic calibration strategy using reference metrology and process splits.
4. Strong data infrastructure for integrating CD-SEM with OCD or other signals in hybrid workflows.

---

Conclusion

At 2nm, metrology success is defined by the ability to control 3D nanosheet geometry, not just top-down linewidth. CD-SEM remains central because it is production-proven and scalable, but it must evolve through multi-angle imaging, model-based 3D reconstruction, and hybrid metrology integration to stay yield-relevant in the nanosheet era.