Plasma Etching Endpoint Detection Using Empirical Mode Decomposition
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Endpoint detection in plasma etching processes is challenging, especially as the open area of substrates decreases and aspect ratios increase, leading to difficulties in signal magnitude and noise introduction from complex subsystems like RF power pulsing and gas pressure pulsing.
Innovation Solution
The implementation of empirical mode decomposition (EMD) to decompose signals into components containing endpoint information and noise, followed by normalization and filtering to enhance the signal-to-noise ratio, allowing for more reliable endpoint detection.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If optical emission spectroscopy is used for endpoint detection in plasma etching, then process monitoring capability is improved, but measurement precision deteriorates due to noise from complex subsystems like RF power pulsing and gas pressure pulsing
Solution Approach 1:
The patent extracts the endpoint detection signal from the noisy plasma emission spectrum by identifying and isolating specific spectral features that correlate with endpoint conditions. This separation allows the endpoint information to be extracted independently from the harmful noise components generated by RF power pulsing and gas pressure pulsing subsystems.
Solution Approach 2:
The system implements feedback by continuously monitoring the plasma emission spectrum and adjusting the endpoint detection algorithm based on real-time spectral characteristics. This feedback mechanism allows the system to adapt to changing noise conditions and maintain precise endpoint detection despite variations in RF power and gas pressure pulsing.
2Manufacturing precision
If substrate open area decreases and aspect ratio increases, then manufacturing capability is improved, but endpoint detection reliability deteriorates due to reduced signal magnitude
Solution Approach 1:
The patent changes the detection parameters by shifting from monitoring overall plasma emission intensity to analyzing specific spectral feature ratios and patterns. This parameter transformation allows endpoint detection to remain reliable even when substrate open area decreases and aspect ratio increases, as the spectral feature-based approach is less dependent on total signal magnitude.
3Measurement precision
If signal filtering is applied to remove noise, then measurement precision is improved, but loss of endpoint information may occur
Solution Approach 1:
The patent segments the plasma emission spectrum into multiple wavelength channels and analyzes each channel independently for specific spectral features. This segmentation allows selective filtering of noise in each wavelength region while preserving endpoint information carried by specific spectral features, thereby maintaining both signal-to-noise ratio and endpoint information integrity.
Data Source
AI summary
A substrate etching system includes an etching control module, a filtering module, and an endpoint module. The etching control module selectively begins plasma etching of a substrate within an etching chamber. The filtering module, during the plasma etching of the substrate: receives a signal including endpoint information; decomposes the signal using empirical mode decomposition (EMD); and generates a filtered signal based on results of the EMD. The endpoint module indicates when an endpoint of the plasma etching of the substrate has been reached based on the filtered signal. The etching control module ends the plasma etching of the substrate in response to the indication that the endpoint of the plasma etching of the substrate has been reached.