Processing Chamber Recovery Using OES and Machine Learning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional methods for determining chamber conditions in semiconductor manufacturing systems are inefficient, often requiring unnecessary quantities of seasoning wafers and causing delays in returning chambers to production mode, as they rely on test samples and do not provide in-situ monitoring, leading to reduced productivity and potential yield decreases due to abnormal chamber conditions.
Innovation Solution
A method using machine learning models that leverage historical sensor data, including optical emission spectra, to predict chamber conditions and recovery status, allowing for real-time adjustments and reducing the need for test samples, thereby enabling faster chamber recovery and improved process control.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional test sample methods are used to determine chamber conditions, then chamber health can be assessed, but unnecessary quantities of seasoning wafers are required and delays occur in returning chambers to production mode
Solution Approach 1:
The patent replaces the mechanical/physical test sample method with an optical sensing system using OES (Optical Emission Spectroscopy) to monitor plasma emissions. Sensors detect spectral data during substrate processing, enabling non-contact, real-time chamber condition assessment without consuming test wafers, thus resolving the contradiction between measurement accuracy and productivity
Solution Approach 2:
The patent introduces spectral data as an intermediary medium between the chamber condition state and the assessment result. By analyzing plasma emission spectra during normal substrate processing, the system indirectly measures chamber health (such as chamber wall deposition levels) without directly testing with seasoning wafers, enabling faster recovery while maintaining assessment accuracy
2Reliability
If chamber seasoning is performed to restore chamber conditions after preventative maintenance, then etch performance is improved, but chamber downtime increases
Solution Approach 1:
The patent performs preliminary chamber condition assessment using OES spectral analysis during the first substrate processing after preventative maintenance. By detecting chamber wall deposition levels through plasma emission spectra before completing the full seasoning process, the system can determine when chamber conditions are sufficient for production, reducing unnecessary seasoning time and chamber downtime while maintaining etch performance consistency
Solution Approach 2:
The patent implements a feedback mechanism where spectral data from ongoing substrate processing is continuously analyzed to monitor chamber condition changes during and after seasoning. This real-time feedback allows dynamic adjustment of the seasoning process, enabling early termination when chamber conditions reach acceptable levels, thus reducing chamber downtime while ensuring etch performance reliability
3Extent of automation
If in-situ monitoring is implemented using machine learning models, then real-time chamber condition detection is achieved, but system complexity increases
Solution Approach 1:
The patent makes the OES spectral analysis system multi-functional by using the same optical emission spectroscopy sensors and machine learning models for both process monitoring and chamber condition assessment. The spectral data processing pipeline serves dual purposes: monitoring substrate processing parameters and detecting chamber wall deposition levels, thereby achieving real-time monitoring without proportionally increasing system complexity
Solution Approach 2:
The patent implements self-service through automated machine learning models that continuously analyze spectral data and automatically determine chamber condition status. The system self-adjusts by comparing real-time spectral patterns against trained models to detect chamber health changes, reducing the need for manual intervention and complex control systems while achieving high-level automation
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 in-situ monitoring and real-time control of chamber conditions, reducing downtime and increasing productivity by accurately determining chamber health and recovery status, allowing for quicker return to production and consistent process results.
Implementation Method 1
spectral data indicating optical emission spectra (OES) measurement of plasma disposed within the process chamber
Data Source
AI summary
A method for determining processing chamber conditions using sensor data and a machine learning model is provided. The method includes receiving, by a processing device, sensor data that include chamber data indicating a state of an environment of a processing chamber processing a substrate according to a set of process parameters of a current process. The sensor data further include spectral data indicating optical emission spectra (OES) measurements of a plasma disposed within the processing chamber. The method further includes using the sensor data as input to a machine learning model and obtaining one or more outputs that indicate one or more chamber condition metrics. The method further includes determining a recovery status of a processing chamber based on the one or more chamber condition metrics. The method further includes causing a modification to a performance of the processing chamber based on the recovery status of the processing chamber.