Computational tuning of the elastic properties of low- and high-entropy ultra-high temperature ceramics

Ultra-high temperature ceramics (UHTCs) represent a class of crystalline materials for extreme environments. They can withstand extremely high temperatures but are mechanically difficult to work with due to their inherent brittleness. Mixture compounds, in particular high-entropy mixtures, offer a pathway to tune the physical properties of UHTCs such as their elastic constants. Here we fine-tune the MACE-MPA-0 universal machine-learning potential on rocksalt carbide UHTCs containing group IV-V metals and demonstrate that not only do the elastic constants deviate from the rule of mixtures approximation in the high-entropy limit, but also in the low-entropy limit of binary and ternary mixtures. We find that this is caused by distortion imposed by the lattice mismatch, enabling the tuning of the physical properties of UHTC mixtures in both low- and high-entropy compounds. We identify a three-component mixture compound, HfCVCZrC, as the best balance between synthesizability and toughness, and apply our developed MACE-UHTC model to identify a range of non-equimolar candidate compositions of this compound which may enable the synthesis of a mixture UHTC with a Young’s modulus up to 40 GPa below that of ZrC.

💡 Research Summary

This paper presents a computational strategy for tuning the elastic properties of Ultra-High Temperature Ceramics (UHTCs) using machine learning, focusing on both low- and high-entropy mixture compounds. UHTCs, such as rocksalt carbides of group IV-V metals, exhibit exceptional thermal stability but suffer from high brittleness due to their large Young’s modulus. The authors address this by exploring the design space of multi-component mixtures, where configurational entropy can stabilize novel phases and enable property tailoring.

The core methodological advance is the development of the “MACE-UHTC” interatomic potential. This model was created by fine-tuning the pre-trained, universal MACE-MPA-0 machine learning potential on a dataset specific to rocksalt carbide UHTCs. This approach combines near-density functional theory (DFT) accuracy with the computational efficiency needed to perform ensemble-averaged calculations of elastic constants for configurationally disordered mixtures—a task prohibitively expensive for direct DFT.

Using this model, the authors systematically investigated all possible equimolar mixtures derived from six parent carbides (TiC, ZrC, HfC, VC, NbC, TaC), totaling 63 compositions. Key findings include: 1) Elastic constants deviate from the simple rule of mixtures not only in the high-entropy limit but also for binary and ternary (low-entropy) mixtures. 2) This deviation is primarily driven by local lattice distortion caused by mismatch between the lattice parameters of the constituent carbides. The degree of lattice mismatch (∆a/ā) strongly correlates with both a reduction in Young’s modulus and an increase in the “effective stabilization temperature” (T*), a metric introduced to estimate synthesizability. T* represents the temperature at which the configurational entropy contribution balances the energy penalty of mixing.

By constructing a Pareto plot of Young’s modulus versus T*, the study identifies the ternary equimolar compound HfCVCZrC as the most promising candidate, offering a balance between reduced stiffness (E = 380 GPa, 13 GPa lower than ZrC) and reasonable synthesizability (T* ~2000 K, comparable to the known binary TiCZrC).

Leveraging the efficiency of the MACE-UHTC model, the authors then perform a high-resolution scan of the full non-equimolar composition space for the HfC-VC-ZrC system. Ternary diagrams map the continuous variation of Young’s modulus and T*. This analysis reveals a Pareto front of non-equimolar compositions that offer further reductions in Young’s modulus, down to a minimum of 354 GPa—nearly 40 GPa below that of ZrC. However, these compositions, often with very low Hf content, trade improved stiffness for a significantly increased T*, indicating greater synthetic challenge.

In conclusion, this work demonstrates that lattice distortion, accessible even in low-entropy mixtures, is a powerful lever for tuning UHTC properties. It establishes a machine learning-aided framework for efficiently navigating the vast composition space of complex ceramics, identifying HfCVCZrC as a prime target for experimental synthesis and providing a roadmap for its composition optimization to achieve tougher, more damage-tolerant ultra-high temperature materials.