Feature-scale simulation models the evolution of individual device features (trenches, vias, lines, contact holes) during fabrication processes — capturing the detailed geometry development that determines device dimensions, profiles, and structural characteristics at the nanometer scale.

What Feature-Scale Simulation Covers

• Etch Profile Evolution: How a trench or via shape develops during reactive ion etching — sidewall angle, bottom rounding, notching, bowing, micro-trenching, and ARDE (aspect-ratio dependent etch).
• Deposition Conformality: How thin films deposit inside high-aspect-ratio structures — step coverage, void formation, seam issues, overhang, and keyhole development.
• Lithography/Patterning: How resist profiles develop during exposure and development — footing, rounding, scumming.
• CMP Surface Evolution: How planarization evolves across feature topography — dishing in wide trenches, erosion of dense arrays.

Physics Involved

• Ion Transport: In plasma etch, ions travel through the sheath and arrive at the wafer surface with angular and energy distributions. Feature walls shadow ions, creating directional effects.
• Neutral Transport: Reactive neutrals (radicals) enter features through random walk / Knudsen transport — aspect ratio affects how many neutrals reach the bottom.
• Surface Chemistry: Etch rates, deposition rates, and selectivity depend on local flux of ions, neutrals, surface temperature, and surface composition.
• Redeposition: Etch byproducts can redeposit on feature sidewalls — affecting profile shape and CD.

Simulation Methods

• Level Set Method: Tracks the evolving surface as the zero-contour of a higher-dimensional function. Handles topological changes (merging, splitting) naturally. Widely used in commercial tools.
• String/Segment Method: Represents the surface as connected segments that move according to local etch/deposition rates. Simple and fast for 2D.
• Monte Carlo (Particle Tracking): Simulates individual ion and neutral trajectories — captures angular distributions and multiple reflections inside features. Most physically accurate but computationally expensive.
• Cell-Based (Voxel): Divides space into cells and evolves each based on local conditions. Good for 3D simulations.

Applications

• High-Aspect-Ratio Etch: Predict profile shape for deep trenches (capacitor trenches in DRAM, TSVs, deep STI) — identify conditions that prevent bowing, twisting, or non-opening.
• Contact/Via Fill: Simulate metal fill of high-AR contact holes — predict void-free fill conditions.
• Gate Spacer: Model spacer deposition and etch to predict final spacer width and shape.
• Dual Damascene: Simulate the trench-via integration sequence.

Feature-scale simulation is essential for process development at advanced nodes — it predicts whether a process recipe will produce acceptable feature profiles before committing expensive silicon experiments.