Optical Emission Waveform Classification for Etching Monitoring
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing etching technologies face challenges in classifying and monitoring optical emission waveforms during plasma etching treatments, particularly in selecting representative wavelengths without prior information on substances or chemical reactions, leading to inefficiencies in data analysis and potential errors in judging etching performance.
Innovation Solution
A system that acquires and analyzes optical emission intensity waveforms, calculates correlation matrices, classifies waveforms based on similarity, and selects representative wavelengths for monitoring, allowing for efficient etching process control and anomaly detection without requiring information on substances or chemical reactions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If optical emission spectrum data is collected at multiple wavelengths to monitor plasma state, then measurement precision is improved, but device complexity increases due to the need to process and analyze multiple wavelength data
Solution Approach 1:
The patent extracts only the essential waveform characteristics (optical emission intensity over time) from the multi-wavelength spectrum data, selecting representative wavelengths that provide sufficient monitoring precision without requiring processing of all wavelength data, thus reducing device complexity while maintaining measurement precision
Solution Approach 2:
The patent creates simplified waveform representations (time-series intensity data) that copy the essential plasma state information from complex multi-wavelength spectrum data, enabling monitoring with reduced data processing requirements while preserving the critical plasma state characteristics
2Productivity
If waveform classification is performed based on shape similarity to identify representative wavelengths, then productivity is improved by automating wavelength selection, but measurement precision may be affected by loss of detailed spectral information
Solution Approach 1:
The patent transforms the wavelength selection problem from a manual process to an automated one by changing the parameter evaluation criterion to waveform shape similarity, enabling systematic classification and automatic identification of representative wavelengths based on quantitative similarity metrics rather than subjective judgment
Solution Approach 2:
The patent performs preliminary waveform classification and similarity assessment on historical or reference data before actual plasma monitoring, pre-identifying representative wavelengths that can be used for subsequent real-time monitoring, thus improving productivity without compromising measurement precision during operation
3Reliability
If all optical emission wavelengths are monitored to ensure comprehensive plasma state detection, then reliability is improved, but loss of time increases due to extensive data analysis requirements
Solution Approach 1:
The patent segments the comprehensive multi-wavelength monitoring task into focused monitoring at representative wavelengths identified through waveform classification, dividing the data analysis workload into manageable portions that maintain reliability by concentrating on the most informative wavelength channels rather than processing all wavelengths equally
Solution Approach 2:
The patent applies partial monitoring action by selecting and monitoring only the representative wavelengths that provide sufficient plasma state information, using fewer wavelength channels than the full spectrum while maintaining adequate monitoring reliability through intelligent selection based on waveform similarity and plasma chemistry relevance
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 automatic selection of monitoring wavelengths, reduces manual errors, and improves the efficiency of etching process control by quantitatively evaluating waveform changes, thereby enhancing the accuracy of etching treatment results.
Implementation Method 1
The ionization and dissociation phenomenon caused by the plasma results in a luminous phenomenon, and therefore, an optical emission spectroscopy (OES) is mounted on an etching apparatus
Data Source
AI summary
There is provided an etching apparatus in which, without setting the information of the substance and the chemical reaction, a small number of representative wavelengths can be selected from a waveform at a lot of wavelengths, and an analysis process of etching data which needs large man-hours can be eliminated to efficiently set the monitoring of the etching. The etching apparatus includes: a lot/wafer/step-depending OES-data searching/acquiring function 511 for acquiring a plurality of optical emission intensity waveforms along an etching-treatment time axis; a waveform-change-existence judgment function 521 for judging the existence of the change of the plurality of light emission intensity waveforms; a waveform-correlation-matrix calculating function 522 for calculating a correlation matrix between the optical emission intensity waveforms; a waveform classifying function 523 for classifying the optical emission intensity waveforms into groups; and a representative-waveform selecting function 524 for selecting a representative optical emission intensity waveform from the group.