1

Executive Summary

700K+
Polymorphs analyzed
7
Anion families
2
Novel descriptors (β, T)
3
Landscape classes
17–28%
β change after filtering

The organization of polymorph energy landscapes — not merely the count of structures — determines phase accessibility and synthesis windows in materials design. This framework quantifies landscape topology using high-throughput analysis of 700,000+ polymorphs from the AFLOW repository across seven anion families: oxides, sulfides, carbides, nitrides, phosphides, borides, and halides.

Two topology descriptors are introduced: (1) the tailness index T quantifying the extension of high-energy states, and (2) the compression exponent β from power-law scaling ΔEspan ∝ (1 + N50)^−β. Analysis reveals three distinct landscape organizations with clear predictive implications for synthesis accessibility.

2

Novel Descriptors

β
Compression Exponent
ΔEspan ∝ (1 + N50)^−β

Quantifies how energetic dispersion evolves with increasing near-ground-state crowding. High β (oxides, nitrides ≈ 3) indicates strong compression — low-energy polymorphs suppress the global energy span. Low β (borides < 1) indicates rigid, incompressible landscapes.

T
Tailness Index
T = ΔE_tail / ΔE_core

Captures the relative extension of high-energy polymorphs beyond the near-ground-state core. T > 1 indicates an extended metastable tail. Minimal sensitivity to compositional filtering — reflects intrinsic topological property of polymorph landscapes.

Together, β and T encode complementary and non-redundant aspects of landscape structure. β describes compression efficiency of the core; T describes the extension of the metastable periphery. Families with comparable β may differ substantially in T, as observed in the oxide–nitride comparison.

3

Compression Exponent β — Interactive

Hover each bar to see the compression exponent and fit quality (R²) for each anion family. The hierarchy is monotonic: Oxides ≈ Nitrides > Carbides > Halides > Sulfides > Phosphides > Borides.

β Compression Exponent by Anion Family
Power-law regression: ΔEspan ∝ (1 + N50)^−β · Hover for details
Interactive
Anion Family β (compression exponent) Landscape Type Interpretation
Oxides≈ 3.0≈ 0.40ContinuousStrong crowding–flattening behavior
Nitrides≈ 3.0≈ 0.38ContinuousComparable to oxides, ionic coordination
Carbides≈ 2.4≈ 0.28IntermediateHigh β but low T — rigid core
Halides≈ 2.0≈ 0.22FragmentedSensitive to mixed-anion filtering
Sulfides≈ 1.8≈ 0.18FragmentedWeak, incoherent scaling
Phosphides≈ 1.6≈ 0.15FragmentedLow compression, decoupled topology
Borides< 1.0< 0.10RigidNo power-law scaling — static landscape
4

Tailness Index T

The tailness index T is robust to compositional filtering — unlike β, which decreases 17–28% after strict single-anion filtering, T changes less than 2% for phosphides, borides, and halides. This indicates that tail organization is an intrinsic topological property, not a compositional artifact.

β vs T — Two-Dimensional Landscape Space
Each point is an anion family · Hover for details · Three distinct regimes visible
Interactive
5

Landscape Classification

Combining β and T reveals three characteristic landscape types. This classification is robust with respect to dataset size and filtering strategy.

Continuous
Coherent Landscapes
Oxides · Nitrides

High β, moderate-to-high T, and strong scaling coherence. Smooth energetic gradients connect polymorphs. Proliferating low-energy states drive coordinated compression of the global energy span. Core and tail are mechanically coupled.

β ≈ 3.0 T ≈ 1.5–2.5 R² ≈ 0.40
Fragmented
Decoupled Landscapes
Sulfides · Phosphides · Halides

Intermediate β, moderate T, and reduced scaling coherence. Clustered low-energy states separated from sparse high-energy configurations. Phase purity intrinsically difficult to maintain — "leaky" core landscape.

β ≈ 1.6–2.0 T ≈ 1.2–1.5 R² ≈ 0.15–0.22
Rigid
Static Landscapes
Borides · (Carbides)

Low β, low T, and absence of clear scaling. Structurally constrained, strongly bonded frameworks. Minimal crowding–flattening response. Directional covalent bonding prevents tail formation.

β < 1.0 T ≈ 1.1 R² < 0.10
6

Effect of Strict Single-Anion Filtering

Strict compositional filtering (removing mixed-anion compounds) reveals that mixed-anion chemistry contributes disproportionately to apparent compression efficiency. Despite representing only 10–20% of polymorph entries, mixed-anion compounds substantially inflate β in phosphides, borides, and halides.

β Before vs After Strict Single-Anion Filtering
Hover to see the change · Oxides and nitrides are robust; others decrease 17–28%
Interactive
Family β (Unfiltered) β (Strict) Change T Change Interpretation
Oxides≈ 3.0≈ 2.95−2%<2%Dominated by intrinsic anion properties
Nitrides≈ 3.0≈ 2.92−3%<2%Robust — ionic landscape character
Carbides≈ 2.4≈ 2.1−13%<2%Moderate mixed-anion contribution
Sulfides≈ 1.8≈ 1.4−22%<2%Mixed-anion enhances flexibility
Phosphides≈ 1.6≈ 1.2−25%<2%Compositional mixing critical
Halides≈ 2.0≈ 1.5−25%<2%Mixed-anion drives 25–28% of β
Borides< 1.0< 0.85−17%<2%Rigid regardless of filtering
7

Implications for Materials Discovery

The β–T framework provides a practical synthesis heuristic. Families with high β exhibit efficient energy compression — controllable phase selection via temperature or pressure tuning is feasible. Low-β families show limited tunability through polymorph crowding alone.

Landscape Type β–T Signature Synthesis Implication Key Families
Coherent High β, moderate-high T Metastability accessible but competitive — precise energy control required Oxides, Nitrides
Decoupled Low β, moderate T Phase purity intrinsically difficult — high-energy metastable phases unexpectedly synthesizable Sulfides, Phosphides
Rigid Low β, low T Limited structural diversity — phase selection constrained by covalent framework Borides, Carbides

All analysis codes and processed datasets are released open-source to enable community validation and extension to additional material families.

Data & Code Availability

Framework code and processed datasets available on GitHub. Cross-database validation using Materials Project (MD-6K sulfide validation set) confirms robustness of topological classification.