Reference Spectra Library for CMP Endpoint Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Chemical mechanical polishing (CMP) processes face challenges in determining the polishing endpoint due to variations in substrate layer thickness, slurry composition, polishing pad conditions, and load, making it difficult to accurately assess when a desired flatness or thickness is achieved, and existing optical monitoring techniques fail to meet the demands of semiconductor manufacturers.
Innovation Solution
Generating a library of reference spectra by modeling refractive index and extinction coefficient variations using Cauchy equations, allowing for a range of refractive index and extinction coefficient values to account for customer-specific and lot-to-lot variations, and using these spectra to compare with in-situ measured spectra during polishing to determine the polishing endpoint.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If a single reference spectrum is generated from an optical model with fixed n and k values, then the optical monitoring system is simple to operate, but the endpoint detection reliability deteriorates due to variations in refractive index and extinction coefficient across different customers and lots
Solution Approach 1:
The reference spectrum library is segmented into multiple discrete spectra, each corresponding to a specific combination of refractive index (n) and extinction coefficient (k) values. This segmentation allows the system to cover a broad range of material variations without requiring a continuous, overly complex model. By dividing the parameter space into manageable discrete steps, the system achieves comprehensive coverage while maintaining operational simplicity.
Solution Approach 2:
The optical model parameters (n and k values) are systematically varied across a defined range to generate multiple reference spectra. This parameter variation approach ensures that the reference library accounts for customer-to-customer and lot-to-lot variations in film optical properties. The system uses user-defined margin parameters to control the extent of variation, allowing flexibility in adapting to different application requirements.
2Measurement precision
If the refractive index and extinction coefficient values are fixed in the optical model, then the reference spectrum calculation is simple and fast, but the accuracy of endpoint detection deteriorates due to material variations
Solution Approach 1:
The system performs preliminary generation of a comprehensive reference spectrum library before the actual polishing process. By pre-calculating reference spectra for all expected variations in n and k values, the system eliminates the need for real-time parameter adjustment during polishing. This preliminary action ensures that when endpoint detection is performed, the system can quickly match the measured spectrum against the pre-prepared library without computational delays.
Solution Approach 2:
The reference spectrum library is designed to be universal, serving multiple functions: it accommodates variations in refractive index, extinction coefficient, and different material compositions across different customers and lots. The user-defined margin parameters allow the same library structure to be applied universally across different applications by simply adjusting the parameter ranges, rather than creating separate systems for each case.
3Manufacturing precision
If optical monitoring is not performed during polishing, then the polishing process is simpler to control, but the manufacturing precision of substrate flatness and thickness deteriorates
Solution Approach 1:
The optical monitoring system provides real-time feedback during the polishing process by continuously measuring the substrate surface characteristics and comparing them against the reference spectrum library. This feedback mechanism enables dynamic adjustment of the polishing process to achieve the desired endpoint, ensuring high manufacturing precision for substrate flatness and thickness. The system translates optical measurements into actionable process control information.
Solution Approach 2:
The system uses optical copying techniques to create a virtual representation of the substrate surface characteristics through spectral measurements. Instead of directly measuring physical dimensions, the system captures optical signatures that represent the substrate state and compares these copies against reference spectra. This indirect measurement approach simplifies the physical measurement system while maintaining high precision through sophisticated spectral analysis.
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 improves the reliability of endpoint detection and reduces within-wafer and wafer-to-wafer thickness non-uniformity by generating a comprehensive library of reference spectra that accounts for likely variations in optical film properties, enabling more accurate monitoring of the polishing process.
Implementation Method 1
calculating a reference spectrum using the optical model based on the refractive index function, the extinction coefficient function and a first thickness of the first layer
Implementation Method 2
dispersion of n and k in the visible spectrum can be modeled using Cauchy equations for many dielectric materials
Data Source
AI summary
A method of generating a library of reference spectra includes storing an optical model for a layer stack having at a plurality of layers, receiving user input identifying a set of one or more refractive index functions and a set of one or more extinction coefficient functions a first layer from the plurality of layers, wherein the set of one or more refractive index functions includes a plurality of different refractive index functions or the set of one or more extinction coefficient functions includes a plurality of different extinction coefficient functions, and for each combination of a refractive index function from the set of refractive index functions and an extinction coefficient function from the set of extinction coefficient functions, calculating a reference spectrum using the optical model based on the refractive index function, the extinction coefficient function and a first thickness of the first layer.