High Entropy Phase (HEP) Insights

• High Entropy Phase (HEP) is defined as a material phase stabilized by configurational entropy and free-energy competition among multiple principal components.
• HEP manifests in various structures including disordered alloys, ordered intermetallics with high entropy mixing, and defect-rich oxides, with stability influenced by processing and observation scale.
• Thermodynamic analyses and predictive frameworks show that entropy alone does not ensure phase stability, but must counterbalance enthalpy, lattice frustration, and kinetic constraints.

High Entropy Phase (HEP) denotes a phase whose stability or persistence is enabled by configurational entropy associated with multiple principal components, but the term has acquired a broader technical scope than the classical single-phase solid-solution picture. In current arXiv literature, HEP can refer to a disordered metallic solid solution, an ordered intermetallic with high-entropy mixing on one sublattice, an amorphous or nanocrystalline state stabilized against recrystallization by multi-principal-element disorder, a defect-rich oxide stabilized by coupled cation and vacancy disorder, or a finite-temperature metastable state selected by entropy near frustrated phase boundaries. Across these usages, the common criterion is not compositional complexity alone, but free-energy competition in which disorder, lattice frustration, vacancy disorder, or near-degenerate microstates prevent a simpler ordered alternative from dominating (Zhou et al., 2019, Sharma et al., 2024, Yang et al., 3 Dec 2025, Timsina et al., 19 May 2025).

1. Conceptual scope and definitional boundaries

In the conventional high-entropy-alloy literature, a high-entropy material is often defined as an alloy containing five or more elements with configurational entropy satisfying $\Delta S_{\text{config}} > 1.5R$, and the historical emphasis has been on identifying a single-phase solid solution, typically FCC, BCC, or occasionally HEX/HCP (Kaufmann et al., 2024). A closely related thermodynamic definition appears in high-throughput screening work, where a single-phase high-entropy alloy is an equimolar multicomponent alloy whose random solid solution lies on the thermodynamic convex hull relative to competing phases (Chen et al., 2022).

The term HEP, however, is not confined to chemically random substitutional alloys. High-entropy intermetallic compounds (HEICs) extend the idea to ordered B2 or D022 structures in which several principal elements occupy one sublattice in nearly equal fractions while Al occupies the other; these are treated as high-entropy phases because the phase itself is entropy-stabilized even though long-range order is present (Zhou et al., 2019). A still broader interpretation appears in mechanically milled Mg-based alloy powder, where the amorphous/nanocrystalline final state is regarded as a HEP because multi-principal-element disorder, high configurational entropy, and strong lattice frustration stabilize the disordered structure against return to a simpler ordered crystal (Sharma et al., 2024).

A central misconception corrected by multiple studies is that “high entropy” does not automatically imply a true single phase. Gas-atomized AlCoCr$_{0.75}$Cu$_{0.5}$FeNi powder appears single-phase by XRD and EBSD, yet aberration-corrected STEM and atom probe tomography reveal nanoscale B2/A2 decomposition, Cu clustering, and grain-boundary segregation. The resulting conclusion is explicit: HEP character can be scale-dependent rather than absolute (Peter et al., 2020).

2. Thermodynamic basis: free-energy competition rather than entropy alone

The standard thermodynamic relation underlying HEP stability is

$\Delta G = \Delta H - T\Delta S,$

with configurational entropy of mixing written as

$\Delta S_{\mathrm{conf}} = -R \sum_i c_i \ln c_i$

per mole, or equivalently in $k_B$ units per atom (Sharma et al., 2024, Evans et al., 2020). In every materials system surveyed here, entropy enters as one term in a competition, not as a sufficient criterion.

This point is made explicitly in temperature-dependent stability analyses of high-entropy alloys. Inverse Hull Webs replace barycentric composition axes by formation energy and inverse hull energy, permitting direct visualization of when a multicomponent solid solution is stable, when it becomes metastable on quenching, and which competing intermetallics define the decomposition pathway. The method introduces a critical solid-solution temperature and a critical adjacent phase temperature to quantify the temperature window over which a HEP remains thermodynamically favored (Evans et al., 2020).

The same thermodynamic logic is formalized more directly in the dominant-pair free-energy framework for multicomponent alloys. There, phase selection is posed as an explicit minimum Gibbs free-energy problem over FCC solid solution, BCC solid solution, B2 ordered phase, and Laves intermetallic competitors. The central conclusion is that high configurational entropy does not by itself guarantee a single-phase HEP; rather, a disordered phase survives only when no lower-free-energy ordered competitor exists (Boakye et al., 30 Jun 2026).

Large-scale screening reinforces the rarity of genuine single-phase stability. High-throughput calculations over $\binom{40}{5}=658{,}008$ equimolar quinary alloys identify 30,201 potentially stable single-phase HEAs at $0.9T_m$, about 4.6% of the total space, with the majority forming in BCC structures (Chen et al., 2022). This does not weaken the HEP concept; it sharpens it by locating HEP stability within a constrained thermodynamic landscape rather than an unrestricted composition class.

3. Structural embodiments of HEPs

The most familiar HEP embodiment is the single-phase solid solution. In AO high-entropy oxides built from five equimolar cations chosen from $\{\mathrm{Ca, Co, Cu, Fe, Mg, Mn, Ni, Zn}\}$, descriptor-based first-principles screening identifies $(\mathrm{CoCuMgNiZn})\mathrm{O}$ as the best candidate for a homogeneous rock-salt entropy-stabilized oxide because it combines low mean local mixing enthalpy with low spread in local mixing enthalpies (Pitike et al., 2020). In lanthanide high-entropy oxides of composition $_{0.75}$0, the relevant HEP is the disordered defective fluorite phase, where disorder exists on both the cation sublattice and the oxygen-vacancy sublattice (Yang et al., 3 Dec 2025).

Ordered HEPs are exemplified by B2 and D022 HEICs. Single-phase or nearly single-phase B2 aluminides such as $_{0.75}$1 and $_{0.75}$2 show that a high-entropy phase may retain long-range intermetallic order while deriving entropy stabilization from near-equimolar disorder on one sublattice. These materials also exhibit significant antisite disorder, estimated at about 10% (Zhou et al., 2019).

Amorphous and nanocrystalline HEPs represent a different structural class. In Mg$_{0.75}$3Ti$_{0.75}$4Cu$_{0.75}$5Zn$_{0.75}$6Fe$_{0.75}$7 powder, high-energy mechanical milling first produces a Mg–Ti-rich hexagonal Laves-type intermetallic precursor, with early formation of TiFe$_{0.75}$8 and later disordered/intermixed Laves-type phases involving TiZn$_{0.75}$9 and MgZn$_{0.5}$0. Continued milling generates dislocations, stacking faults, twins, grain refinement, antisite disorder, and nanograin boundaries until the ordered intermetallic lattice loses coherence and transforms into an amorphous phase. The authors explicitly interpret this as type III amorphization. The final structure contains nanograins smaller than 5 nm, exhibits disorder-driven lattice expansion from 15.83 Å to 16.11 Å in the FCC-derived amorphous supercell, and resists reversion after annealing at 500 °C for 6 hours (Sharma et al., 2024).

The concept also extends beyond crystalline materials and alloys. In frustrated pyrochlore iridates $_{0.5}$1 with $_{0.5}$2, Monte Carlo simulations reveal two finite-temperature metastable HEPs, $_{0.5}$3 and $_{0.5}$4, located between stable 2I2O, fragmented 3I1O/1I3O, and AIAO phases. These states lack simple long-range order, have high susceptibility, and are stabilized because the entropy term in $_{0.5}$5 dominates near phase boundaries (Timsina et al., 19 May 2025).

4. Characterization, hidden heterogeneity, and length-scale dependence

HEP identification is inseparable from characterization length scale. The AlCoCr$_{0.5}$6Cu$_{0.5}$7FeNi case is a canonical demonstration. XRD showed peaks consistent with a B2 ordered structure, and EBSD phase mapping also indicated a single B2 phase. Yet aberration-corrected STEM and APT revealed nanoscale phase separation into Ni-Al-rich B2 regions, Fe-Cr-rich A2 regions, and Cu-rich clusters confined to the B2 matrix. The Cu-rich cluster number density was reported as $_{0.5}$8, characteristic domain sizes were about 5–9 nm, the A2/B2 interfaces were perfectly coherent, and the effective lattice mismatch was only $_{0.5}$9. The resulting interpretation is that the material is single-phase only in an average, coarse-probe sense (Peter et al., 2020).

A similar multi-length-scale requirement appears in lanthanide high-entropy oxides. For $\Delta G = \Delta H - T\Delta S,$0, XRD and TEM delineate bixbyite, transition, and fluorite regions, but Raman spectroscopy detects local bixbyite order persisting to about 50% Ce under equilibrium synthesis conditions even where XRD and TEM indicate fluorite-like long-range structure. SAED at 32.5% Ce shows satellite reflections and diffuse scattering, demonstrating that the structural transition regime contains local ordering invisible to coarse probes (Yang et al., 3 Dec 2025).

The general consequence is methodological rather than merely descriptive. Any claim that a HEP is “single-phase” must specify the technique, the probed length scale, and whether the relevant notion of phase identity is long-range crystallographic, nanoscale chemical, or local defect-structural. This suggests that HEP should often be treated as a multiscale designation rather than a binary label.

5. Processing routes and far-from-equilibrium stabilization

Many HEPs are processing-dependent. High-energy mechanical milling drives the Mg-based alloy through repeated cold welding, fracturing, and severe plastic deformation; the resulting antisite disorder, lattice distortion, and nanograin-boundary accumulation are not incidental defects but the mechanism by which the precursor intermetallic transforms into an amorphous HEP (Sharma et al., 2024). Mechanical alloying followed by spark plasma sintering and homogenization annealing similarly enabled the fabrication of B2 and D022 HEICs that bridge conventional HEAs and high-entropy ceramics (Zhou et al., 2019).

Kinetic trapping can be equally decisive in oxides. In equilibrium bulk ceramics, the fluorite HEP in $\Delta G = \Delta H - T\Delta S,$1 appears only above a threshold Ce content and/or temperature, whereas pulsed-laser-deposited thin films can metastably trap fluorite at only 20% Ce. The charge-balance estimate for that fluorite-like composition requires an oxygen-vacancy fraction of about 16.7%, underscoring that the metastable HEP is a highly vacancy-rich disordered fluorite rather than an ideal fluorite crystal (Yang et al., 3 Dec 2025).

High pressure introduces a separate axis of control through the $\Delta G = \Delta H - T\Delta S,$2 term and oxygen activity. In high-entropy oxides, pressure can either stabilize or destabilize the target phase. At 15 GPa, the spinel HEO $\Delta G = \Delta H - T\Delta S,$3 transforms into a metastable modified ludwigite-type structure with nominal formula $\Delta G = \Delta H - T\Delta S,$4, while the rock-salt HEO may instead show competing tenorite, wurtzite, or a layered rock-salt-like transformation (Aamlid et al., 2024).

Dynamic compression highlights an important limiting case: a high-entropy alloy need not preserve a distinct HEP under extreme loading. In additively manufactured eutectic AlCoCrFeNi$\Delta G = \Delta H - T\Delta S,$5, in-situ XRD under shock reveals transformation from the initial fcc+bcc eutectic to a pure fcc region from roughly 99 to 198 GPa, a pure bcc region above about 210 GPa, and liquid-like scattering from 308 GPa upward. The alloy’s high-entropy character therefore manifests as a rich phase landscape and delayed transformation behavior rather than a single persistent shock-stabilized HEP (Parsons et al., 17 Jul 2025). A related XFEL study on CuPdAgPtAu reports a transient compressed phase with 5.1% compression along the (111) plane lasting about 0.3 ns, but explicitly stops short of claiming direct proof of a distinct generalizable HEP (Huang et al., 22 Mar 2026).

6. Predictive frameworks and design methodologies

HEP research has generated a dense ecosystem of predictive frameworks, all motivated by the same difficulty: multicomponent phase stability cannot be inferred reliably from entropy alone. One route mines binary phase diagrams. In a phenomenological machine-learning model trained on progressively richer HEA phase databases, the central descriptors are phase field parameters $\Delta G = \Delta H - T\Delta S,$6 and the phase separation parameter PSP, evaluated at an optimized phase-formation temperature $\Delta G = \Delta H - T\Delta S,$7. Using WEKA and a Random Forest model with 300 trees, the method achieves a single-phase HEA prediction rate greater than 80% and a 77% success rate on 30 newly synthesized validation alloys (Jie et al., 2019).

Another route relies on explicit free-energy ranking. For 269 experimentally characterized HEAs, the dominant-pair free-energy classifier reaches 54.6% accuracy and macro-F1 0.532 in a four-class problem, and 77.9% accuracy with macro-F1 0.763 in the well-posed three-class FCC/BCC/duplex task. Its significance lies less in raw classification than in replacing single-scalar heuristics with continuous phase-stability maps over composition and temperature (Boakye et al., 30 Jun 2026).

High-throughput DFT screening provides a complementary map of accessible HEP space. Over 650,000 equimolar quinary alloys, more than 30,000 candidate single-phase alloys were identified, mainly BCC, and two new HEAs—AlCoMnNiV and CoFeMnNiZn—were successfully synthesized as single-phase BCC and FCC materials, respectively (Chen et al., 2022). Additional screening rules invoke the role of allotropy: analysis of 434 unique known single-phase HEAs suggests that, once high-entropy conditions are met, the majority crystal structure among non-allotrope-forming elements predicts the final phase with overall accuracy of about 71% (Kaufmann et al., 2024). In cobalt-free HEAs, a GAN-augmented Gaussian process classifier based on $\Delta G = \Delta H - T\Delta S,$8, $\Delta G = \Delta H - T\Delta S,$9, $\Delta S_{\mathrm{conf}} = -R \sum_i c_i \ln c_i$0, VEC, $\Delta S_{\mathrm{conf}} = -R \sum_i c_i \ln c_i$1, and $\Delta S_{\mathrm{conf}} = -R \sum_i c_i \ln c_i$2 reaches 84% accuracy for BCC prediction after reduction to five principal components, with $\Delta S_{\mathrm{conf}} = -R \sum_i c_i \ln c_i$3 and $\Delta S_{\mathrm{conf}} = -R \sum_i c_i \ln c_i$4 identified as the most important descriptors (Luo et al., 11 May 2026).

The convergent lesson from these methods is consistent. HEP design is tractable, but only when thermodynamic competition, defect chemistry, structural preference, and processing pathway are all admitted as first-class variables.

7. Generalizations, misconceptions, and terminological reuse

Three general conclusions recur across the literature. First, entropy stabilization is real but conditional: a HEP exists only within a temperature, pressure, redox, or kinetic window, and it can become metastable or disappear outside that window (Evans et al., 2020, Yang et al., 3 Dec 2025). Second, high entropy does not require complete chemical randomness. Ordered B2 and D022 intermetallics, vacancy-rich fluorites, and amorphous/nanocrystalline alloys can all qualify as HEPs when configurational disorder remains central to phase stability (Zhou et al., 2019, Sharma et al., 2024). Third, “single-phase” is often an operational statement tied to measurement scale rather than a universal structural truth (Peter et al., 2020).

The phrase has also begun to migrate beyond materials thermodynamics. In test-time scaling for LLMs, “High Entropy Phase” has been formalized as a variable-length segment of a reasoning trajectory that begins at a high-entropy token and ends only when consecutive low-entropy tokens appear. There, HEP is not a thermodynamic phase but a segment-level representation of uncertainty used to define an Entropy Centroid and a Lowest Centroid selection rule (Zhao et al., 28 Apr 2026). This reuse confirms that the phrase now has a wider technical life, but it does not alter its core materials-science meaning: a HEP is a phase whose stability, persistence, or observability depends critically on entropy-mediated competition among alternative ordered states.