Semiconductor Process Modeling for Equipment State Changes
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing semiconductor process models struggle to accurately predict process parameters during state changes due to regular inspection and maintenance of equipment.
Innovation Solution
A semiconductor process modeling method using a Q-learning model that incorporates process parameters as states and process recipes as actions to improve prediction consistency, particularly during equipment state changes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a model predicts FDC data based on previously processed wafers, then the model predicts change over time in FDC data well, but the model is poor at predicting FDC data at a point at which a state of semiconductor process equipment changes due to regular inspection and maintenance
Solution Approach 1:
The patent applies preliminary action by incorporating maintenance schedule information and equipment state change predictions into the modeling process before actual maintenance occurs. The system proactively identifies when equipment state changes are likely to happen based on maintenance schedules, and pre-adjusts the prediction model to account for these anticipated changes, thereby improving prediction accuracy during transitions.
Solution Approach 2:
The patent implements dynamics by making the prediction model adaptive to changing equipment states. The system dynamically adjusts prediction parameters based on detected equipment state changes and maintenance schedules, allowing the model to transition from static predictions to dynamic predictions that respond to real-time equipment conditions and scheduled maintenance events.
2Reliability
If process parameters are predicted using traditional models, then the models can handle steady-state conditions, but the models fail to maintain prediction consistency when equipment undergoes regular inspection and maintenance
Solution Approach 1:
The patent uses maintenance schedule information and equipment state indicators as intermediary variables that bridge the gap between steady-state process parameters and changing equipment conditions. These intermediaries carry information about upcoming maintenance events and current equipment states, allowing the prediction model to adjust its behavior without requiring complete redesign of the underlying prediction architecture.
Solution Approach 2:
The patent applies parameter changes by modifying prediction model parameters based on equipment state and maintenance schedule information. When maintenance is scheduled or equipment state changes are detected, the system adjusts key prediction parameters such as time windows, weighting factors, and reference data selection criteria to maintain prediction reliability during transitions.
Data Source
AI summary
A semiconductor process modeling method is performed by a computing device, and includes obtaining a plurality of first raw data including values of a process parameter on a plurality of first wafers and a plurality of second raw data including values of a process recipe on the plurality of first wafers; preprocessing the plurality of first raw data and the plurality of second raw data to generate a plurality of first tensor data corresponding to the plurality of first raw data and a plurality of second tensor data corresponding to the plurality of second raw data; and inputting the plurality of first tensor data and the plurality of second tensor data into a predictive model, and thus, outputting, from the predictive model, a plurality of output data including values of a process parameter on a second wafer.