Interpretable Machine Learning Framework for Nb─Si Based Alloy Design with Enhanced Fracture Toughness.

Area of Science:

• Materials Science
• Metallurgical Engineering
• Aerospace Engineering

Background:

• High-temperature alloys are critical for improving aero-engine efficiency and reducing aviation emissions.
• Niobium-silicon (Nb-Si) alloys offer potential for higher operating temperatures and lower densities than nickel-based superalloys.
• A key challenge for Nb-Si alloys is overcoming the fracture toughness (K_Q) limit of 20 MPa·m^1/2.

Purpose of the Study:

• To develop a machine learning-driven design framework for Nb-Si ultra-high temperature alloys.
• To identify strategies for surpassing the 20 MPa·m^1/2 fracture toughness barrier.
• To guide the design of novel Nb-Si alloys with improved high-temperature performance.

Main Methods:

• A three-step feature screening strategy was employed within a machine learning framework.
• Predictive modeling for fracture toughness (K_Q) was performed with an error below 7%.
• SHAP (SHapley Additive exPlanations) and PDP (Partial Dependence Plot) analyses were used for model interpretation and alloy design guidance.

Main Results:

• Five Nb-Si alloys were synthesized and tested, validating the predictive model.
• Sample #5 (Nb_38.5Ti_38.5Si₃Zr₁₈V₂) achieved a fracture toughness (K_Q) of 22.791 MPa·m^1/2, exceeding the typical 20 MPa·m^1/2 threshold.
• Microstructural analysis revealed enhanced toughness due to the transformation of brittle silicide phases to ductile Nbss and crack-bridging mechanisms.

Conclusions:

• The machine learning framework effectively guided the design of Nb-Si alloys with superior fracture toughness.
• Solid solution strengthening was identified as the dominant strengthening mechanism (68%-84%), contributing to an excellent strength-toughness balance.
• The developed alloys show promise for next-generation aerospace materials, supporting emission reduction targets.