Optical Model Library for CMP Endpoint Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing optical monitoring techniques for chemical mechanical polishing (CMP) are unreliable due to variations in substrate layer thickness and slurry composition, making it difficult to determine the polishing endpoint accurately.
Innovation Solution
A method is developed to generate a library of reference spectra using an optical model that accounts for variations in layer thickness and metal contribution, allowing for the calculation of multiple reference spectra that span the likely range of incoming substrates, improving the reliability of the matching algorithm.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If multiple substrates with different underlying layer thicknesses are measured to create a larger library, then the matching reliability is improved, but the time and labor required increases significantly
Solution Approach 1:
The patent uses an optical model to calculate and generate reference spectra that copy the characteristics of actual substrate reflections without requiring physical measurement of multiple substrates. This virtual copying approach creates a comprehensive library spanning the full range of thickness variations while avoiding the time-consuming process of measuring numerous physical samples.
Solution Approach 2:
The optical model systematically varies key parameters such as layer thickness, refractive index, and absorption coefficient to generate reference spectra across the entire expected range of substrate variations. This parametric approach efficiently creates a robust library that accounts for wafer-to-wafer and within-wafer thickness non-uniformity without requiring physical samples for each parameter combination.
2Productivity
If theoretically calculated reference spectra are used, then the library generation is quick and cost-effective, but the spectra often diverge from empirically determined spectra reducing reliability
Solution Approach 1:
The patent performs preliminary calibration by measuring reference spectra from substrates with known thicknesses before production use. These empirically obtained spectra are used to adjust and validate the optical model parameters, ensuring that subsequent theoretically calculated reference spectra accurately reflect real substrate behavior while maintaining the efficiency of model-based library generation.
Solution Approach 2:
The system incorporates feedback mechanisms where initial empirical measurements inform and refine the optical model parameters. The model is continuously validated against actual measurement data, and discrepancies are used to adjust model parameters, creating a closed-loop system that maintains high reliability while preserving the speed advantages of theoretical calculation.
3Measurement precision
If variations in layer thickness and slurry composition are accounted for, then the polishing endpoint detection accuracy is improved, but the complexity of the optical model increases
Solution Approach 1:
The optical model is segmented into distinct modular components, each handling a specific aspect of the substrate structure or optical property. This modular architecture allows the model to account for multiple varying parameters such as layer thickness, refractive index, and slurry composition without becoming unmanageably complex, as each component can be independently configured and validated.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
This approach enables quick and cost-effective generation of a robust library of reference spectra, enhancing the reliability of the polishing endpoint detection and reducing within-wafer and wafer-to-wafer thickness non-uniformity.
Implementation Method 1
calculating a reference spectrum using an optical model based on the first thickness value, second thickness value and third thickness value
Implementation Method 2
the optical properties thin film stacks
Data Source
AI summary
A method of controlling polishing includes storing a library having a plurality of reference spectra, polishing a substrate, measuring a sequence of spectra of light from the substrate during polishing, for each measured spectrum of the sequence of spectra, finding a best matching reference spectrum using a matching technique other than sum of squared differences to generate a sequence of best matching reference spectra, and determining at least one of a polishing endpoint or an adjustment for a polishing rate based on the sequence of best matching reference spectra. Finding a best matching reference spectrum may include performing a cross-correlation of the measured spectrum with each of two or more of the plurality of reference spectra from the library and selecting a reference spectrum with the greatest correlation to the measured spectrum as a best matching reference spectrum.