Selectivity in Metrology refers to the ability to measure target parameters in the presence of interfering materials or signals — isolating the desired measurement from confounding factors like underlayers, adjacent films, or process variations, critical for accurate characterization of complex multi-layer stacks at advanced semiconductor nodes.
What Is Selectivity in Metrology?
- Definition: Ability to measure target parameter without interference from other sources.
- Quantification: Ratio of sensitivity to target vs. sensitivity to interferents.
- Goal: Isolate desired measurement from confounding factors.
- Challenge: Complex stacks have many overlapping signals.
Why Selectivity Matters
- Complex Stacks: Advanced nodes have 10+ layers contributing to signal.
- Accurate Measurement: Must isolate target layer from others.
- Process Control: Incorrect measurements lead to wrong process adjustments.
- Yield: Poor selectivity causes mischaracterization and yield loss.
- Advanced Nodes: Increasingly critical as stacks become more complex.
Selectivity Challenges
Thin Film Thickness Measurement:
- Problem: Underlayers contribute to optical signal.
- Example: Measuring 5nm film on top of 100nm film.
- Interference: Both films affect reflectance spectrum.
- Solution: Multi-wavelength measurement, modeling both layers.
Composition vs. Density:
- Problem: XRF (X-ray fluorescence) signal depends on both.
- Example: Measuring copper concentration in alloy.
- Interference: Density variations mimic composition changes.
- Solution: Combine XRF with XRR (X-ray reflectometry) for density.
Process Variation vs. Metrology Noise:
- Problem: Distinguish real process variation from measurement noise.
- Example: CD variation across wafer.
- Interference: Metrology precision limits detection of small variations.
- Solution: High-precision metrology, statistical analysis.
Enhancement Techniques
Multiple Wavelengths:
- Method: Measure at wavelengths with different penetration depths.
- Benefit: Separate surface from bulk contributions.
- Example: UV for surface, IR for bulk in optical metrology.
- Application: Thin film thickness, composition profiling.
Angular Resolution:
- Method: Measure at multiple angles of incidence.
- Benefit: Separate surface scattering from bulk reflection.
- Example: Ellipsometry at multiple angles.
- Application: Surface roughness, interface characterization.
Reference Measurements:
- Method: Measure reference sample, subtract background.
- Benefit: Remove systematic contributions.
- Example: Blank wafer measurement for background subtraction.
- Application: Defect detection, contamination monitoring.
Model-Based Separation:
- Method: Physical model separates contributions.
- Benefit: Leverages known physics to isolate target.
- Example: OCD modeling of multi-layer stack.
- Application: Complex structure characterization.
Polarization Control:
- Method: Use specific polarization states.
- Benefit: Different materials respond differently to polarization.
- Example: Ellipsometry separates film properties.
- Application: Anisotropic materials, stress measurement.
Techniques by Metrology Type
Optical Metrology (OCD, Ellipsometry):
- Challenge: Multiple films contribute to spectrum.
- Selectivity: Model all layers, fit simultaneously.
- Enhancement: Multiple angles, wavelengths, polarizations.
- Limitation: Model accuracy critical.
X-Ray Metrology (XRF, XRR, XRD):
- Challenge: Overlapping elemental peaks, substrate signal.
- Selectivity: Energy-resolved detection, grazing incidence.
- Enhancement: Synchrotron sources, high-resolution detectors.
- Limitation: Penetration depth limits surface sensitivity.
Electron Microscopy (SEM, TEM):
- Challenge: Charging, material contrast, depth information.
- Selectivity: Energy-filtered imaging, backscatter detection.
- Enhancement: Low voltage, multiple detectors.
- Limitation: Surface-sensitive, sample prep artifacts.
AFM (Atomic Force Microscopy):
- Challenge: Tip convolution, adhesion forces.
- Selectivity: Mode selection (contact, tapping, non-contact).
- Enhancement: Sharp tips, force spectroscopy.
- Limitation: Slow, limited to surface.
Applications at Advanced Nodes
High-k/Metal Gate Stacks:
- Challenge: Measure 1nm high-k layer under metal gate.
- Selectivity: XRR for thickness, XPS for composition.
- Requirement: Sub-angstrom thickness precision.
Multi-Layer Interconnects:
- Challenge: Measure barrier layer between copper and dielectric.
- Selectivity: TEM for cross-section, XRF for composition.
- Requirement: Distinguish 2nm barrier from adjacent layers.
FinFET/GAA Structures:
- Challenge: Measure fin dimensions in 3D structure.
- Selectivity: CD-SEM for top, OCD for profile, TEM for validation.
- Requirement: Separate fin width from spacer thickness.
EUV Resist Characterization:
- Challenge: Measure resist thickness on complex underlayers.
- Selectivity: Ellipsometry with modeling of full stack.
- Requirement: <1nm thickness precision.
Quantifying Selectivity
Sensitivity Ratio:
``
Selectivity = (∂Signal/∂Target) / (∂Signal/∂Interferent)
- High Selectivity: Large ratio, target dominates signal.
- Low Selectivity: Small ratio, interferent affects measurement.
- Goal: Maximize selectivity for accurate measurement.
Signal-to-Noise Ratio:
`
SNR = Signal_target / Noise_total
- Includes: Measurement noise, interferent contributions.
- Requirement: SNR > 10 for reliable measurement.
Uncertainty Budget:
- Target Uncertainty: Desired measurement precision.
- Interferent Contribution: Uncertainty from confounding factors.
- Total Uncertainty: Quadrature sum of all sources.
- Goal: Minimize interferent contribution.
Improving Selectivity
Measurement Optimization:
- Parameter Selection: Choose wavelengths, angles for maximum selectivity.
- Multi-Modal: Combine techniques with complementary selectivity.
- Calibration: Use reference samples to characterize interferents.
Sample Preparation:
- Isolation: Remove or mask interfering layers when possible.
- Reference Structures: Fabricate structures with isolated target.
- Blanket Films: Use blanket wafers for calibration.
Data Analysis:
- Modeling: Accurate physical models separate contributions.
- Statistical Methods: PCA, ICA to separate signal components.
- Machine Learning: Train models to recognize target vs. interferent patterns.
Validation:
- Cross-Check: Compare with orthogonal metrology technique.
- Reference Metrology: Validate against TEM, AFM, or other gold standard.
- Correlation: Correlate to electrical or functional measurements.
Tools & Approaches
- Multi-Technique: KLA, Onto Innovation integrated metrology.
- Advanced Modeling: Rigorous simulation (RCWA, FEM) for selectivity.
- Machine Learning: AI-enhanced metrology for complex stacks.
- Reference Labs: NIST, PTB for traceable standards.
Selectivity in Metrology is essential for advanced semiconductor manufacturing — as material stacks become increasingly complex with 10+ layers and sub-nanometer critical dimensions, the ability to isolate target measurements from interfering signals determines whether metrology can provide the accuracy needed for process control, making selectivity enhancement a critical focus for metrology development.