Coded Substrate Identifier for IP-Safe AI Recipe Tuning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Intellectual property restrictions prevent the sharing of substrate context information with machine learning (ML) and artificial intelligence (AI) modules in semiconductor processing, leading to suboptimal use of these modules.
Innovation Solution
A coded substrate material identifier system using a matrix instead of proprietary information is implemented, allowing ML/AI modules to process substrates without knowing underlying substrate properties by associating matrix identifiers with substrate properties and storing sensor data in a database.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If substrate context information is provided to ML/AI modules, then processing performance and uniformity are improved, but intellectual property restrictions are violated and sensitive information is exposed
Solution Approach 1:
The patent introduces a coded substrate material identifier as an intermediary that bridges the need for substrate information and IP protection. The identifier system allows ML/AI modules to access substrate properties through coded representations without exposing actual substrate information, thus resolving the contradiction between performance improvement and IP protection
Solution Approach 2:
The patent creates coded copies of substrate information that can be used by ML/AI modules without exposing the original sensitive data. The matrix identifiers serve as informational copies that enable processing while maintaining IP restrictions, allowing the system to use substrate context without accessing actual substrate properties
2Reliability
If proprietary substrate information is used, then accurate processing control is achieved, but information sharing restrictions prevent optimal ML/AI module utilization
Solution Approach 1:
The patent transforms substrate information from proprietary detailed parameters into coded matrix identifier parameters. This parameter transformation maintains the essential information needed for process control while changing the form to enable sharing with ML/AI modules, thus improving both reliability and adaptability
3Object-affected harmful factors
If no substrate context information is provided, then intellectual property restrictions are maintained, but ML/AI module performance is suboptimal
Solution Approach 1:
The coded substrate material identifier acts as a mediator that enables ML/AI modules to receive substrate context information without violating IP restrictions. This intermediary system resolves the contradiction by allowing information flow necessary for high productivity while maintaining IP protection through coded representations
Data Source
AI summary
Embodiments disclosed herein include methods of processing substrates in a tool. In an embodiment, the method of processing the substrate in the tool, comprises receiving an augmented recipe with a machine learning (ML) and/or an artificial intelligence (AI) module. In an embodiment, the augmented recipe comprises, a recipe for processing the substrate in the tool, and a matrix identifier that corresponds to one or more substrate properties. In an embodiment the method further comprises using the ML and/or AI module to retrieve a data set from a database, where the data set is associated with the matrix identifier, and using the ML and/or AI module to modify the augmented recipe to form a modified recipe, where the modification is dependent on the data set.