Active learning design of bcc solid solution alloys with gigapascal strength and elemental metal-level ductility

Area of Science:

• Materials Science
• Metallurgy
• Computational Materials Science

Background:

• Body-centered cubic (bcc) alloys exhibit high yield strength but limited tensile ductility.
• Multi-principal-element alloys offer potential for enhanced strength-ductility combinations.
• Traditional alloy exploration is costly, and machine learning (ML) faces data scarcity challenges.

Purpose of the Study:

• To develop an ML-guided framework for rapid alloy composition optimization.
• To discover novel alloys with synergistic strength and ductility.
• To overcome data scarcity limitations in ML for materials discovery.

Main Methods:

• Integration of active learning with physics-informed Bayesian optimization.
• Development of an ML-guided framework for accelerated alloy design.
• Characterization of alloy mechanical properties and microstructural features.

Main Results:

• Discovery of a Ti-V-Nb-Hf-Zr-Al alloy (Ti36V14Nb22Hf22Zr1Al5) with 953 MPa yield strength and 42% tensile ductility.
• Identification of ~1-nm local chemical fluctuations (LCFs) contributing to high strength and ductility.
• Demonstration of enhanced dislocation multiplication and strain hardening due to LCFs.

Conclusions:

• The ML-guided framework effectively accelerates the discovery of high-performance alloys.
• LCFs in concentrated bcc solid solutions are crucial for achieving superior strength-ductility synergy.
• This approach shows significant promise for future materials design and discovery.