High-Entropy Alloy Phase Design With ML-CALPHAD Screening
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for predicting and designing high-entropy alloys (HEAs) face challenges in accurately distinguishing between specific phases and efficiently exploring the vast compositional space for alloys with superior structural and functional properties, particularly due to limitations in machine learning datasets and feature selection.
Innovation Solution
A synergistic approach combining phenomenological and adaptive features, utilizing a database management system with machine learning algorithms, including phase diagram scanning and active learning, to predict thermodynamic phases of HEAs, enabling efficient design of high-entropy alloys with improved accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If machine learning methods are used to predict HEA phases, then productivity is improved, but measurement precision deteriorates due to inability to distinguish specific phases
Solution Approach 1:
The patent divides the phase prediction task into two distinct stages: first using machine learning for rapid screening of compositional regions, then applying CALPHAD thermodynamic calculations for precise phase identification. This segmentation allows each method to operate in its optimal performance zone while compensating for the other's limitations.
Solution Approach 2:
The patent introduces CALPHAD thermodynamic calculations as an intermediary step between machine learning prediction and final phase identification. This intermediary refines the machine learning outputs by providing detailed thermodynamic analysis that enables accurate distinction between specific phases like BCC, FCC, and HCP structures.
2Manufacturing precision
If the compositional space of HEAs is explored thoroughly, then manufacturing precision is improved, but loss of time increases due to the exponentially large compositional space
Solution Approach 1:
The patent performs preliminary machine learning-based screening of the compositional space before conducting detailed analysis. By pre-identifying promising compositional regions using ML models trained on existing HEA data, the system reduces the vast compositional space to a manageable subset that requires thorough exploration, thereby saving significant time.
Solution Approach 2:
The patent applies different levels of analysis depth to different regions of the compositional space. Promising regions identified by machine learning receive detailed CALPHAD analysis and precise phase prediction, while other regions receive minimal or no analysis. This localized quality approach ensures manufacturing precision where needed while minimizing time loss in less critical areas.
3Productivity
If existing machine learning models are used for HEA prediction, then productivity is improved, but measurement precision worsens due to limited datasets and feature selection
Solution Approach 1:
The patent merges machine learning methods with CALPHAD thermodynamic calculations into a hybrid predictive system. This combination leverages the speed and pattern recognition capabilities of ML while incorporating the thermodynamic rigor of CALPHAD, resulting in both improved productivity and enhanced measurement precision that neither method could achieve alone.
Solution Approach 2:
The patent implements a feedback mechanism where CALPHAD results are used to validate and refine machine learning predictions. The thermodynamic calculations provide ground truth data that feeds back into the ML model, improving its accuracy over time while maintaining the efficiency benefits of machine learning.
Data Source
AI summary
Embodiments relate to system and methods involving use of a technique for managing a database for producing a material composition having a thermodynamic phase. The technique can include: receiving a binary phase diagram for each material to be used as a component of a high-entropy alloy (HEA); using one or more active learning machine learning techniques for generating a feature, the feature including: a primary feature that is representative of a probability that an HEA will exhibit a solid solution phase and/or an intermetallic phase, and a physics-based feature that is representative of a factor related to formation of a desired intermetallic HEA phase; encoding the primary feature and the physics-based feature; generating an output representation of a HEA alloy composition and phase of a predicted materials composition; and selecting a HEA composition and phase that will meet a material design criterion.