Substrate Process Modeling Across Multiple Manufacturing Chambers
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing manufacturing systems face challenges in efficiently predicting and optimizing the performance of substrate processing operations across multiple process chambers, leading to difficulties in maintaining quality, identifying faults, and optimizing equipment performance.
Innovation Solution
A comprehensive process modeling platform that integrates multiple process models, allowing for the prediction of substrate performance across various process operations and chambers. This platform utilizes input data from sensors and metrology tools to provide predictive outputs, enabling corrective actions and optimizing manufacturing processes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If multiple separate process models are used for different process operations, then each model can be specialized and accurate, but the system complexity increases and integration becomes difficult
Solution Approach 1:
The patent combines multiple separate process models into a unified process modeling platform that can handle multiple process operations. The platform integrates different process models (e.g., for deposition, etching, cleaning) into a single cohesive system that manages them collectively, reducing integration complexity while preserving the specialized predictive capabilities of each individual model through modular architecture.
Solution Approach 2:
The process modeling platform is designed as a universal system that can accommodate multiple types of process models and operations through standardized interfaces. It provides multi-functional capabilities to manage diverse process operations (deposition, etching, cleaning, etc.) within a single platform, eliminating the need for separate specialized systems while maintaining predictive accuracy through adaptable model structures.
2Measurement precision
If process models are trained using extensive historical data, then predictive accuracy improves, but the time and computational resources required for training and updating increase
Solution Approach 1:
The system performs preliminary actions by pre-training process models using historical data before actual production use. Models are trained in advance on extensive historical process data to establish baseline predictive capabilities, then deployed for real-time predictions without requiring continuous retraining, thus reducing ongoing time requirements while maintaining high accuracy.
Solution Approach 2:
The system enables continuous useful action through real-time predictive monitoring and automated corrective actions. Once models are trained, they continuously provide predictions and trigger corrective actions without interruption to production, maximizing the utility of the initial training investment while minimizing the need for repeated training cycles.
3Reliability
If real-time monitoring and predictive analysis are implemented across all process chambers, then quality control and fault identification improve, but the computational load and processing time increase
Solution Approach 1:
The system segments the monitoring and predictive analysis functions into chamber-specific process models that operate independently. Each process chamber has its own dedicated model that processes data locally, avoiding the need to analyze all chamber data centrally. This segmentation reduces computational load while maintaining comprehensive quality control across all chambers through distributed processing.
Solution Approach 2:
The system introduces an intermediary layer (the process modeling platform) that sits between raw sensor data and quality control decisions. This intermediary performs predictive analysis and translates complex sensor data into actionable predictions, reducing the computational burden on the overall system while improving the speed and accuracy of quality control responses.
4Reliability
If comprehensive process modeling is implemented to predict performance across multiple operations, then the ability to identify and correct faults improves, but the initial setup time and resource investment increase
Solution Approach 1:
The system uses copying by creating virtual models (digital twins) of physical process chambers and operations. These virtual copies replicate the behavior and characteristics of actual chambers, allowing comprehensive fault identification and predictive analysis to be performed on the models without requiring physical intervention or extensive setup in the actual production environment. Once modeled, the same framework can be copied across multiple chambers with minimal additional setup.
Data Source
AI summary
In one aspect of the present disclosure, a method includes obtaining, by a processing device, input data indicative of a first set of process parameters. The method further includes providing the input data to a first process model. The method further includes obtaining, from the first process model, first predictive output indicative of performance of a first process operation in accordance with the first set of process parameters. The method further includes providing the first predictive output to a second process model. The method further includes obtaining, from the second process model, second predictive output indicative of performance of a second process operation, different than the first process operation or a repetition of the first process operation, in accordance with the first set of process parameters. The method further includes performing a corrective action in view of the second predictive output.