arxiv_ai 85% Match Research Paper Materials Scientists,Computational Physicists,Metallurgists,AI Researchers in Materials Science 2 weeks ago

Clarifying the Ti-V Phase Diagram Using First-Principles Calculations and Bayesian Learning

graph-neural-networks › molecular-modeling

📄 Abstract

Abstract: Conflicting experiments disagree on whether the titanium-vanadium (Ti-V) binary alloy exhibits a body-centred cubic (BCC) miscibility gap or remains completely soluble. A leading hypothesis attributes the miscibility gap to oxygen contamination during alloy preparation. To resolve this disagreement, we use an ab initio + machine-learning workflow that couples an actively-trained Moment Tensor Potential with Bayesian inference of free energy surface. This workflow enables construction of the Ti-V phase diagram across the full composition range with systematically reduced statistical and finite-size errors. The resulting diagram reproduces all experimental features, demonstrating the robustness of our approach, and clearly favors the variant with a BCC miscibility gap terminating at T = 980 K and c = 0.67. Because our simulations model a perfectly oxygen-free Ti-V system, the observed gap cannot originate from impurity effects, in contrast to recent CALPHAD reassessments.

Authors (3)

Timofei Miryashkin

Olga Klimanova

Alexander Shapeev

Submitted

June 21, 2025

arXiv Category

cond-mat.mtrl-sci

arXiv PDF

Key Contributions

This work resolves the conflicting experimental data on the Ti-V phase diagram by employing a novel workflow coupling ab initio calculations with Bayesian learning. It constructs a robust phase diagram using an actively-trained Moment Tensor Potential and Bayesian inference of the free energy surface, clearly favoring a BCC miscibility gap and demonstrating that impurity effects are not the cause.

Business Value

Provides accurate and reliable phase diagrams for alloy systems, which is crucial for designing new materials with desired properties, optimizing manufacturing processes, and ensuring material reliability in various applications.

Paper Metadata

Innovation Type

Methodological/Algorithmic

Deployment Feasibility

High for research and development. The workflow is computationally intensive but provides high-fidelity results for materials design.

Limitations Addressed

Addresses conflicting experimental results regarding the Ti-V binary alloy phase diagram and the hypothesis that oxygen contamination is the cause of observed discrepancies.

Technical Tags

phase diagramfirst-principles calculationsBayesian learningab initioMoment Tensor Potentialfree energy surfacealloy systemsmaterials science

Research Topics

Materials ScienceComputational Materials DesignPhase DiagramsFirst-Principles CalculationsMachine Learning in MaterialsThermodynamics

Methods & Architectures

Ab initio calculationsActively-trained Moment Tensor Potential (MTP)Bayesian inference of free energy surfaceWorkflow coupling Moment Tensor Potential (MTP)Bayesian Inference Models

Applications & Tasks

Materials Science Metallurgy Alloy Design Resolving conflicting experimental dataPredicting phase diagramsUnderstanding alloy behavior Phase diagram predictionMaterials property predictionAlloy characterization

Related Fields

PhysicsChemistryComputational ScienceMachine Learning

Keywords

Phase DiagramTitanium-VanadiumFirst-PrinciplesBayesian LearningMoment Tensor PotentialFree EnergyAlloyMaterials ScienceComputational MaterialsThermodynamicsAb Initio

Academic Context

#Materials Science#Computational Materials Design#Phase Diagrams#First-Principles Calculations#Machine Learning in Materials#Thermodynamics

Commercial Potential

Potential Products

Materials design softwareDatabases of material properties and phase diagrams

Target Industries

AerospaceAutomotiveManufacturingEnergyMaterials Research

Use Case Examples

Designing high-strength titanium alloys for aerospacePredicting the behavior of vanadium alloys in high-temperature applications

Competitive Edge

Offers a more rigorous and accurate method for phase diagram prediction compared to traditional CALPHAD methods, especially when experimental data is conflicting or limited.

Resource Requirements

Compute Needs

Very High, due to ab initio calculations and extensive simulations.

Data Requirements

Requires experimental data for validation and potentially initial training data for MTP.

Deployment Constraints

Computational cost and time required for simulations.

Scalability

Scalability is limited by the computational cost of ab initio methods and MTP training.

Production Readiness

Maturity Level

Research

Time to Market

Long, for integration into materials design workflows.

Patent Potential

Low, primarily methodological.

View Full Paper Back to Papers