Composition-Space Screening
Generate candidates under elemental, density, cost, critical-material, and processing constraints. Physics-based descriptors and uncertainty-aware surrogate models prioritize regions for higher-fidelity calculation.
Overview
Navigate alloy complexity with linked physics and data
High-entropy alloys (HEAs), multi-principal-element alloys, and advanced superalloys offer broad composition and processing freedom, but performance depends on more than nominal chemistry. Competing solid solutions and intermetallics, segregation during solidification, γ/γ′ balance, topologically close-packed phase formation, grain-boundary chemistry, oxidation, creep, and fatigue can change across temperature and time.
Our AI for Advanced Metals and Ceramic Materials workflow connects curated evidence, CALPHAD and first-principles descriptors, process-aware machine learning, microstructure models, and validation planning. Predictions are reported with applicability limits and uncertainty so that candidates can be down-selected for targeted experiments rather than treated as verified materials.
Core Services
Decision-focused modeling for HEAs and superalloys
Phase Stability and Precipitation
Assess equilibrium and metastable phase tendencies, solid-solution stability, γ/γ′ fractions, solvus behavior, TCP risk, and precipitation windows using fit-for-purpose CALPHAD, DFT, and kinetic inputs.
Solidification and Segregation
Evaluate Scheil-type solidification paths, partitioning, hot-cracking indicators, homogenization requirements, and additive-manufacturing sensitivity while distinguishing model assumptions from process-specific reality.
Microstructure–Property Modeling
Link composition and heat treatment to grain size, precipitate fraction and scale, strengthening, elastic response, hardness, creep, fatigue, and fracture-relevant indicators with calibrated uncertainty.
High-Temperature Environmental Risk
Rank oxidation, hot-corrosion, interdiffusion, coating compatibility, and phase-instability risks for defined temperature, atmosphere, stress, and exposure-time envelopes.
Active-Learning Validation Design
Select informative compositions, heat treatments, and measurements that reduce model uncertainty, test competing mechanisms, and support an efficient design–make–test–learn cycle.
Integrated Workflow
From service conditions to validation-ready candidates
| Stage | Key Activities | Decision Output |
|---|---|---|
| 1. Project scoping | Define alloy family, product form, process route, temperature–stress–environment envelope, target properties, restricted elements, and experimental budget. | Design brief and success criteria |
| 2. Evidence and data audit | Harmonize composition basis, heat treatment, test conditions, censoring, provenance, duplicates, and measurement uncertainty; map the domain represented by available data. | Analysis-ready dataset and gap map |
| 3. Physics-informed screening | Combine thermodynamic descriptors, CALPHAD/DFT results, empirical constraints, and interpretable ML; use grouped or leave-family-out validation where relevant. | Candidate regions with confidence tiers |
| 4. Process and microstructure assessment | Evaluate solidification, segregation, homogenization, precipitation, grain structure, additive or casting constraints, and likely failure mechanisms. | Feasible composition–process windows |
| 5. Multi-objective down-selection | Balance strength, ductility, creep, oxidation, density, manufacturability, cost, and supply risk using Pareto analysis and sensitivity testing. | Ranked shortlist and trade-off map |
| 6. Validation and model update | Specify witness alloys, heat treatments, characterization, test conditions, acceptance logic, and feedback needed to recalibrate the model. | Executable validation plan |
Deliverables
Traceable outputs for alloy and process decisions
Curated Data Package
Normalized compositions, process histories, test conditions, provenance, exclusion rules, and documented data limitations.
Candidate Ranking
Shortlisted chemistries and confidence tiers with the drivers, constraints, and applicability domain behind each recommendation.
Phase and Risk Maps
Phase-fraction, solvus, segregation, precipitation, TCP, oxidation, or failure-risk maps over relevant composition and temperature windows.
Process Windows
Recommended casting, additive, homogenization, solution, aging, and cooling ranges with sensitivities and model assumptions.
Model and Audit Report
Methods, feature definitions, validation splits, residuals, uncertainty, sensitivity, extrapolation checks, and reproducible settings.
Experimental Validation Plan
Prioritized compositions, characterization, mechanical and environmental tests, acceptance criteria, and learning-loop updates.
Applications
Alloy development under demanding service conditions
Gas Turbines and Propulsion
γ′-strengthened alloys, turbine components, combustor hardware, coatings, and oxidation-aware high-temperature design.
Energy and Nuclear Systems
Refractory and radiation-tolerant MPEAs, heat-exchanger materials, corrosion-resistant alloys, and long-duration thermal exposure.
Additive Manufacturing
Printable alloy selection, solidification and hot-cracking risk, heat-treatment design, and process–microstructure consistency.
Cryogenic and Structural Alloys
Strength–ductility trade-offs, phase stability, deformation mechanisms, fatigue, and damage-tolerant candidate selection.
Wear and Corrosion Protection
Bulk alloys and coatings designed for coupled hardness, chemical stability, interdiffusion, wear, and environmental resistance.
Compositionally Complex Coatings
Bond-coat and overlay concepts assessed for oxidation, phase evolution, thermal mismatch, and substrate compatibility.
Scientific Evidence
Published studies supporting the modeling strategy
Peer-reviewed studies demonstrate that machine learning can classify phase constitution, predict phase fractions, and identify stable-phase regions across the large composition space of high-entropy alloys. These results support the integration of thermodynamic descriptors, composition-aware models, and targeted experimental validation for accelerated alloy development.1–3
Combining machine learning with thermodynamic and processing information enables rapid candidate screening, phase-stability assessment, and focused selection of compositions for experimental evaluation.
1 Machaka, R. Machine Learning-Based Prediction of Phases in High-Entropy Alloys. Computational Materials Science 2021, 188, 110244. https://doi.org/10.1016/j.commatsci.2020.110244. Distributed under the Creative Commons CC BY 4.0 license.
2 Liu, S.; Li, X.; et al. A Comparative Study of Predicting High Entropy Alloy Phase Fractions with Traditional Machine Learning and Deep Neural Networks. npj Computational Materials 2024, 10, 128. https://doi.org/10.1038/s41524-024-01335-1. Distributed under the Creative Commons CC BY 4.0 license.
3 Peivaste, I.; et al. Data-Driven Analysis and Prediction of Stable Phases for High-Entropy Alloy Design. Scientific Reports 2023, 13, 22942. https://doi.org/10.1038/s41598-023-50044-0. Distributed under the Creative Commons CC BY 4.0 license.
Project Strategy
Models built for the next experimental decision
We keep composition basis, processing history, service conditions, model version, uncertainty, and validation handoffs visible throughout the project. To discuss an HEA, MPEA, superalloy, coating, or internal alloy dataset, please Contact Us or submit the Online Inquiry below.
Related Services
