CMP In-Situ Signal Filtering for Accurate 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 slurry distribution, polishing pad condition, relative speed, and load, leading to inconsistent material removal rates and uncertainty in achieving desired flatness or thickness.
Innovation Solution
An in-situ monitoring system generates a signal that is filtered using a linear prediction filter to model periodic disturbances and underlying signals, allowing for reliable determination of the polishing endpoint or adjustment of the polishing rate by applying a Kalman filter to reduce signal noise and minimize delay.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If in-situ monitoring is used to determine polishing endpoint, then polishing precision is improved, but measurement accuracy deteriorates due to periodic disturbances in the signal
Solution Approach 1:
The patent extracts and removes periodic disturbance components from the monitoring signal using filtering techniques. The filter separates the useful polishing endpoint information from harmful periodic disturbances caused by platen rotation and carrier head sweeping, thereby improving measurement accuracy while maintaining endpoint detection capability
Solution Approach 2:
The patent introduces a filter as an intermediary component between the sensor and the endpoint determination system. This filter acts as a mediator that processes the raw signal, removing periodic disturbances while preserving the underlying polishing endpoint information, thus resolving the contradiction between maintaining signal fidelity and achieving accurate measurements
2Measurement precision
If filtering is applied to reduce signal noise, then measurement precision is improved, but response time deteriorates due to filter delay
Solution Approach 1:
The patent applies partial filtering by selectively removing only the specific periodic disturbance frequencies related to platen rotation and carrier head sweeping, rather than applying aggressive broad-spectrum filtering. This approach achieves sufficient noise reduction for accurate endpoint detection while minimizing the introduction of filter delay
Solution Approach 2:
The patent optimizes filter parameters (such as cutoff frequencies and filter order) to achieve the best compromise between noise reduction and response time. By carefully selecting filter parameters that match the specific disturbance frequencies in the system, the patent minimizes filter delay while maintaining measurement precision
Data Source
AI summary
A method of controlling polishing includes polishing a substrate, monitoring the substrate during polishing with an in-situ monitoring system, filtering a signal from the monitoring system to generate a filtered signal, and determining at least one of a polishing endpoint or an adjustment for a polishing rate from the filtered signal. The filtering includes modelling a plurality of periodic disturbances at a plurality of different frequencies using a plurality of disturbance states, modelling an underlying signal using a plant state, and applying a linear prediction filter to the plant state and the plurality of disturbance states to generate a filtered signal representing the underlying signal.