Crystal Polymorphs and Observability as a Persistence Filter: Separating Formation from Survival in the Inorganic Structural Record

Crystal structure prediction returns hundreds or thousands of nearly degenerate candidate structures per composition, yet experiment records only a few polymorphs. The usual explanation is energetic: the observed forms are those closest to the energy minimum. This paper instead treats observability as a filter. A structure enters the record only if it is formed and then survives until it is observed. Thermally activated survival works as a sharp, Gumbel-shaped filter in the escape barrier. Its threshold is E* = k_BT ln(ν t_obs), and it moves by only k_BT per e-fold of observation time. From a single observed population and its energies alone, formation and survival cannot be told apart; this is the crystallographic instance of the Retrospective Drift Theorem. Under a Bell–Evans–Polanyi coupling, survival alone produces what looks like a thermodynamic ceiling on metastability. The empirical part uses open data: 154,387 Materials Project entries (via OPTIMADE), ICSD tags from JARVIS-DFT, and Crystallography Open Database metadata on pressure, temperature and mineral names, giving 10,583 metastable polymorphs in 5,417 compositions. The fraction of computed polymorphs that are observed is non-monotonic in energy, which rules out energy ranking as the selection rule. The sign structure of the record follows remnant metastability, and a coordination-based barrier proxy fails. Two tests diverge from formation-only explanations: Among phases documented above 1 GPa, retention at ambient conditions falls with energy (ρ = −0.29, p = 0.005). Natural minerals are less metastable than synthetic-only polymorphs (median 18.2 vs 30.0 meV/atom, p = 2.6×10⁻⁴). The record is therefore formation times persistence. The full analysis code is included in the appendix. Keywords: polymorphism; crystal structure prediction; overprediction; metastability; remnant metastability; Ostwald’s rule of stages; disappearing polymorphs; observation selection; persistence conditioning; retrospective drift; Materials Project; Crystallography Open Database Related identifiers: Is supplement to / References — https://doi.org/10.13140/RG.2.2.26532.64648 (The Retrospective Drift Theorem) Creator: Kriger, Boris — Information Physics Institute; Institute of Integrative and Interdisciplinary Research — ORCID 0009-0001-0034-2903 License: CC BY 4.0

Authors

Publication Details

Journal
Zenodo (CERN European Organization for Nuclear Research)
Published
2026-11-16
DOI
https://doi.org/10.5281/zenodo.24278666
Primary Topic
Machine Learning in Materials Science
Type
article
Field-Weighted Citation Impact
0.00
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
article

Crystal Polymorphs and Observability as a Persistence Filter: Separating Formation from Survival in the Inorganic Structural Record

Boris Kriger
Zenodo (CERN European Organization for Nuclear Research)
Machine Learning in Materials Science
article

Crystal Polymorphs and Observability as a Persistence Filter: Separating Formation from Survival in the Inorganic Structural Record

Boris Kriger
article en

Abstract

Crystal structure prediction returns hundreds or thousands of nearly degenerate candidate structures per composition, yet experiment records only a few polymorphs. The usual explanation is energetic: the observed forms are those closest to the energy minimum. This paper instead treats observability as a filter. A structure enters the record only if it is formed and then survives until it is observed. Thermally activated survival works as a sharp, Gumbel-shaped filter in the escape barrier. Its threshold is E* = k_BT ln(ν t_obs), and it moves by only k_BT per e-fold of observation time. From a single observed population and its energies alone, formation and survival cannot be told apart; this is the crystallographic instance of the Retrospective Drift Theorem. Under a Bell–Evans–Polanyi coupling, survival alone produces what looks like a thermodynamic ceiling on metastability. The empirical part uses open data: 154,387 Materials Project entries (via OPTIMADE), ICSD tags from JARVIS-DFT, and Crystallography Open Database metadata on pressure, temperature and mineral names, giving 10,583 metastable polymorphs in 5,417 compositions. The fraction of computed polymorphs that are observed is non-monotonic in energy, which rules out energy ranking as the selection rule. The sign structure of the record follows remnant metastability, and a coordination-based barrier proxy fails. Two tests diverge from formation-only explanations: Among phases documented above 1 GPa, retention at ambient conditions falls with energy (ρ = −0.29, p = 0.005). Natural minerals are less metastable than synthetic-only polymorphs (median 18.2 vs 30.0 meV/atom, p = 2.6×10⁻⁴). The record is therefore formation times persistence. The full analysis code is included in the appendix. Keywords: polymorphism; crystal structure prediction; overprediction; metastability; remnant metastability; Ostwald’s rule of stages; disappearing polymorphs; observation selection; persistence conditioning; retrospective drift; Materials Project; Crystallography Open Database Related identifiers: Is supplement to / References — https://doi.org/10.13140/RG.2.2.26532.64648 (The Retrospective Drift Theorem) Creator: Kriger, Boris — Information Physics Institute; Institute of Integrative and Interdisciplinary Research — ORCID 0009-0001-0034-2903 License: CC BY 4.0

Zenodo (CERN European Organization for Nuclear Research)
Machine Learning in Materials Science
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon