Thermodynamic phase-field model for microstructure with multiple components and phases: the possibility of metastable phases

A diffuse-interface model for microstructure with an arbitrary number of components and phases was developed from basic thermodynamic and kinetic principles and formalized within a variational framework. The model includes a composition gradient energy to capture solute trapping, and is therefore suited for studying phenomena where the width of the interface plays an important role. Derivation of the inhomogeneous free energy functional from a Taylor expansion of homogeneous free energy reveals how the interfacial properties of each component and phase may be specified under a mass constraint. A diffusion potential for components was defined away from the dilute solution limit, and a multi-obstacle barrier function was used to constrain phase fractions. The model was used to simulate solidification via nucleation, premelting at phase boundaries and triple junctions, the intrinsic instability of small particles, and solutal melting resulting from differing diffusivities in solid and liquid. The shape of metastable free energy surfaces is found to play an important role in microstructure evolution and may explain why some systems premelt at phase boundaries and phase triple junctions while others do not.

💡 Research Summary

The paper presents a thermodynamically consistent phase‑field framework capable of handling an arbitrary number of chemical components and phases, with particular emphasis on phenomena where the physical width of the interface matters. Starting from the Cahn‑Hilliard functional for a binary alloy, the authors systematically extend the free‑energy expansion to a multicomponent (M) and multiphase (N) system. By retaining only isotropic second‑order gradient terms, the functional includes composition‑gradient energy coefficients κᵢⱼ and phase‑gradient coefficients λᵅᵝ, while cross‑terms ξᵢᵅ are set to zero for simplicity. The total free energy is written as a weighted sum of homogeneous free‑energy densities of each phase, multiplied by phase fractions φᵅ, subject to the constraints Σφᵅ = 1 and Σcᵢ = 1. To prevent unphysical mixing of phases, a multi‑obstacle barrier function is introduced, which enforces φᵅ·φᵝ = 0 for any pair of distinct phases unless a metastable phase is deliberately allowed to appear at an interface.

A key conceptual advance is the careful distinction between chemical potential μᵢ and diffusional potential used as the driving force for mass transport. The authors argue that the common definition δF/δcᵢ violates the mole‑fraction constraint in multicomponent systems. Instead, they define the diffusion potential as the derivative of the Gibbs free energy with respect to the number of moles of component i (μᵢᵉ = ∂G/∂nᵢ), ensuring compliance with the Gibbs‑Duhem relation and the Nernst‑Einstein equation. This formulation yields diffusion equations that naturally incorporate differing diffusivities in solid and liquid phases.

The model is applied to four representative problems. First, solidification with nucleation demonstrates that the presence of a metastable free‑energy basin can alter nucleation pathways and growth kinetics. Second, premelting at phase boundaries and triple junctions is captured: when the metastable surface lies sufficiently above the common tangent, a thin liquid film forms spontaneously, reproducing experimentally observed interfacial premelting. Third, the intrinsic instability of small particles is reproduced; particles below a critical radius dissolve because surface‑energy contributions outweigh bulk stability. Fourth, solutal melting caused by disparate solid‑liquid diffusivities is simulated, showing enrichment of solute in the liquid and localized melting.

Across all cases, the shape of the metastable free‑energy landscape emerges as the decisive factor governing whether premelting, transient phase formation, or dissolution occurs. The authors contrast their approach with the Wheeler‑Boettinger‑McFadden (WBM) model, which assumes identical composition in both phases across an interface, and the Access/Steinbach models, which place interfacial compositions on the common tangent and thus suppress metastable behavior. By retaining the composition‑gradient term (∇c)², the present framework naturally accounts for solute trapping during rapid solidification, a feature absent in many thin‑interface models.

In summary, the work delivers a versatile, thermodynamically rigorous phase‑field model that unifies the treatment of multiple components, multiple phases, solute trapping, and metastable phase dynamics. It bridges the gap between fully physical Cahn‑Hilliard descriptions (which are computationally prohibitive for large microstructures) and thin‑interface models (which often neglect important energetic contributions). The authors suggest future extensions to anisotropic gradient coefficients, elastic effects, and direct coupling with CALPHAD databases to enable quantitative predictions for real alloy systems.