Refractory Complex Concentrated Alloys (RCCA)

• RCCA are multi-principal-element alloys composed of refractory metals that achieve high melting points and exceptional thermal stability.
• They utilize configurational entropy, DFT, CALPHAD, and machine-learning methods to predict phase stability and tailor mechanical properties.
• Experimental and computational strategies guide RCCA design to mitigate challenges such as impurity embrittlement and poor oxidation resistance.

Refractory Complex Concentrated Alloys (RCCA) are a class of multi-principal-element alloys predominantly composed of refractory metals, distinguished by their high melting points, exceptional thermal stability, and strong resistance to mechanical degradation at elevated temperatures. Characterized by equiatomic or near-equiatomic incorporation of four or more refractory elements—such as Nb, Ta, Ti, Hf, Mo, W, Cr, V, Zr—RCCAs are synthesized to exploit maximized configurational entropy, moderate mixing enthalpy, and minimized atomic-size mismatch to stabilize single-phase, typically bcc, solid solutions. Their vast compositional space and rich defect chemistry enable complex property tuning for extreme-environment applications, but also introduce challenges in phase stability, embrittlement by interstitial impurities, and oxidation resistance (Huang et al., 8 Dec 2025, Byggmästar et al., 2024, Bejjipurapu et al., 2 Nov 2025).

1. Thermodynamic Stabilization, Phase Formation, and Modeling

The formation and stability of RCCA solid solutions are dictated by a combination of configurational entropy (ΔSₘᵢₓ), mixing enthalpy (ΔHₘᵢₓ), atomic-size mismatch (δ), and interaction parameters (Ω{ij}). RCCAs are engineered to maximize –R∑ᵢcᵢ ln cᵢ (entropy) and keep ΔHₘᵢₓ within a slightly negative or weakly positive regime such that the free energy of mixing ΔGₘᵢₓ = ΔHₘᵢₓ – TΔSₘᵢₓ remains negative at processing/operating temperatures (Byggmästar et al., 2024). The bcc solid solution is stabilized when atomic-size mismatch is δ < ~0.06, Ω{ij} is modest, and the valence electron concentration (VEC) matches the stability field of the bcc (A2) phase (VEC ≈ 5–6 for Mo–W–Ta–Ti–Zr alloys) (Singh et al., 2017).

Thermodynamic models for phase stability employ high-throughput DFT-derived binary mixing enthalpy databases, regular-solution models for multicomponent ΔHₘᵢₓ and ΔSₘᵢₓ, and convex-hull analyses versus intermetallic competitor phases (Zhang et al., 2022). Limitations in empirical descriptors (e.g., Ω, δ, VEC) and semiempirical rules for high-order systems necessitate physics-based CALPHAD, DFT, and machine-learning–assisted simulations for robust prediction of phase content and stability (Kozlík et al., 2024). CALPHAD's practical accuracy is currently restricted by incomplete quaternary/quinary intermetallic assessments; extensions into non-equimolar and metastable regions require further data-driven refinement (Kozlík et al., 2024, Zhang et al., 2022).

2. Local Chemical Environments, Defect Chemistry, and Microstructure

In RCCAs, the random substitutional disorder generates spatially heterogeneous local chemistries, leading to the existence of a continuum of "local composition vectors" (c_local), which deviate from the global composition (c_global) (McCarthy et al., 2023). The full distribution of c_local, quantified via deviation metrics (λ) and convex-hull fractions (V_f), governs the spread of local properties (e.g., vacancy formation energies), as confirmed in large-scale hybrid MC/MD and DFT simulation frameworks for MoNbTaTi and related systems (McCarthy et al., 2023). Short-range order (SRO) emerges from pairwise attractions (negative Warren–Cowley α parameters) and can distort the local-composition point cloud, broadening property distributions beyond the mean values predicted by rule-of-mixtures.

RCCA microstructures evolve through solidification, annealing, and deformation-induced phase transformations. In dual-phase MoNbTaTiZr, cold rolling to 30% reduction fragments grains, induces sub-5 µm substructure, nucleates a C15 Zr₂Ta-type Laves phase, and stimulates {110} twinning. Despite strain localization and eventual cracking, slip remains operative (high Schmid and Taylor factors) and twinning enhances ductility (Skolakova et al., 10 Dec 2025). The interplay between BCC1/BCC2 partitioning, Laves fraction, orientation relationships (rare <110> and <100> rotations), and twinning collectively determine the work-hardening rate, anisotropy, and mechanical response.

3. Interstitial Impurity Control and Grain Boundary Embrittlement

Interstitial impurities—primarily O, N, and C—are major factors limiting RCCA mechanical performance. Oxygen, for example, resides uniformly in the bcc lattice up to a solubility limit of 0.8–1.0 at. % in NbTiHfTa; above this, grain-boundary segregation and thermodynamic precipitation of monoclinic HfO₂ precipitates destabilize the matrix and produce catastrophic embrittlement (Huang et al., 8 Dec 2025). The Gibbs free energy of solution for O is described as

$$\Delta G_{sol}(c_O,T) = \Delta H_{mix}(c_O) - T\,\Delta S_{mix}(c_O),$$

with ΔH_{mix}(c_O) determined by machine-learning interatomic potentials (MLIP, e.g., ORB-v2) integrated into finite-temperature MC sampling. Microstructural and mechanical characterization demonstrates the transition from ductile fracture (O < 0.8 at. %) to brittle failure (O > 1.0 at. %) concomitant with the appearance of boundary oxides. Alloying strategies that diminish local Hf–O coordination, ultra-high-purity processing, and post-solidification vacuum annealing are effective for mitigating interstitial-induced embrittlement.

4. Plasticity, Dislocation Physics, and Fracture Mechanisms

Plasticity mechanisms in RCCAs are governed by a complex interplay of lattice friction, stacking-fault/twinning-fault energies, local chemical disorder, and dislocation character. In MoTaW, stacking-fault ($\gamma_{usf} = 1.361\,\mathrm{J/m}^2$) and twinning-fault ($\gamma_{utf} = 1.683\,\mathrm{J/m}^2$) energies significantly exceed rule-of-mixtures predictions, leading to preferential activation of {110}<111> slip and suppression of twinning during nanoindentation (Dominguez-Gutierrez et al., 11 Jan 2026). The resultant plasticity exhibits strong orientation dependence—symmetric, multi-lobed slip for [001], anisotropic and junction-enhanced strain for [011]—and is characterized by a network of dislocations and local structure transformations.

High-fidelity atomistic simulations, employing both MLIPs (tabGAP, ACE) and first-principles DFT, have revealed that compositional complexity enhances fracture resistance by promoting dislocation nucleation at crack tips. In Nb₄₅Ta₂₅Ti₁₅Hf₁₅, dislocation pile-up at the tip induces a rising R-curve, transitioning from emission-limited to slip-limited crack growth, substantially elevating toughness compared to NbMoTaW or pure Nb (Wang et al., 25 Feb 2025). These effects are not fully captured by classical Rice–Thomson theory, emphasizing the essential role of atomic-scale disorder and group IV alloying in enabling ductile fracture (via $\gamma_{usf}$ reduction and $\gamma_{surf}$ enhancement).

In NbTaTiV, systematic V addition increases atomic misfit (δ) and raises the edge-dislocation glide barrier, thereby effecting a temperature-dependent transition from screw- to edge-dislocation control. This transition enhances high-temperature strength, resulting in a pronounced yield-strength plateau (e.g., 600 MPa at 1173 K, V25) not attainable in pure or low-V alloys (Zakia et al., 14 Nov 2025). STEM and neutron analysis confirm the mechanism switch and highlight compositional tuning as a principal design handle.

5. Oxidation Resistance and Active-Learning Design Paradigms

RCCAs are generally deficient in long-term oxidation resistance (>3 h at 1000 °C) without protective coatings, owing to rapid scale volatilization or non-protective oxide formation (Bejjipurapu et al., 2 Nov 2025). Physics-informed machine-learning (Gaussian process regression, GPR) models using alloy and oxide descriptors (oxygen chemical potential, weighted Pilling–Bedworth ratio, oxide solidus) accurately predict specific mass gain (mg/cm²) after 24 h oxidation, achieving $\mathrm{MAE}=5.78$ mg/cm² (Bejjipurapu et al., 2 Nov 2025). By high-throughput screening in Al–Cr–Nb–Ti–X quaternaries, compositions combining high Cr (≥25 at.%), moderate Al (15–35 at.%), and limited Zr/V are found to promote the formation of continuous Cr₂O₃ and/or Al₂O₃ scales with low mass gain (<5 mg/cm² at 1000 °C).

Active-learning frameworks, iterating between GPR surrogate modeling and batch Bayesian optimization, drastically reduce experimental campaigns to converge upon optimal chemistries (e.g., Al₃₀Mo₅Ti₁₅Cr₅₀ and Al₄₀Mo₅Ti₃₀Cr₂₅ with <1 mg/cm² gain at 1000 °C (Bejjipurapu et al., 17 Dec 2025)). These alloys form adherent external α-Al₂O₃ scales with parabolic kinetics, high specific hardness (>0.12 HV₀.₅m³/kg), and CTE matching for TBC compatibility, thus providing efficient entrypoints for multiobjective, property-balanced RCCA development.

6. Computational Design, Data-Driven Discovery, and Alloy Optimization

The stiffness, strength, and ductility landscape of RCCAs is efficiently mapped through DFT, MLIP-driven MD simulations, and deep-learning property regressors. High-throughput workflows (∼10⁵+ candidates) leverage descriptors such as ΔHₘᵢₓ, δ, VEC, Δχ, Tₘ, elastic moduli, and oxidation thermodynamics to enable rapid screening and Pareto analysis of strength–plasticity–oxidation tradeoffs (Giles et al., 20 Nov 2025). Two-shot experimental/computational cycles informed by Bayesian optimization can deliver >50% improvement in simultaneous specific modulus and hardness in under six months (Paramore et al., 2024). These computational loops are now generalizable to any alloy system and are critical for autonomous RCCA discovery.

7. Practical Guidelines and Future Directions

Practical design of RCCAs integrates: (1) compositional tuning to position the Fermi level in the d-state pseudogap and minimize ΔH_form (Singh et al., 2017); (2) control of atomic-size mismatch for targeted defect physics (Zakia et al., 14 Nov 2025); (3) process and alloying regimes maintaining interstitial content below critical solubility thresholds (Huang et al., 8 Dec 2025); (4) multi-objective selection for oxidation, mechanical, and thermomechanical compatibility using active learning (Bejjipurapu et al., 17 Dec 2025, Bejjipurapu et al., 2 Nov 2025). Future challenges include extension of CALPHAD databases to full quinary phases, integration of vibrational and electronic entropy in thermodynamic stability, coupling of defect kinetics to mechanical simulations, and systematic incorporation of local chemical/structural metrics into multi-scale models (Huang et al., 8 Dec 2025, McCarthy et al., 2023).

RCCAs exemplify how complex, high-dimensional design spaces require data-driven, physics-based, and experimentally integrated approaches for property optimization. Analogous strategies are applicable for next-generation superalloy development, extreme environment structural components, and beyond.