Caller context

The brief is motivation. It is not a cited source.

Caller brief (context, not evidence)

This file is motivation and constraint for the v0.3 working line. It is not evidence. Discovery must be able to contradict it. Do not cite this file as an external source.

Why this run exists

Apex Material Semantic Core v0.2 parks solid-state phase in microstructure and lets β- and γ-Ca₂SiO₄ share a composition kind_id. That encoding is an open item on the v0.2 unresolved table, not a published result. This run asks whether published identity practice agrees.

In scope

How published materials-identity systems (standards, ontologies, data models, handbooks) treat solid-state polymorphs and microstructure relative to chemical composition when deciding whether two records are the same kind.

Out of scope (do not expand)

• F9: a published necessary-and-sufficient identity model for all materials.
• Live ASTM / ISO 15510 / MatML XML corpora (H1/H2), unless such files are already in this repository.
• Rewriting runs/v02-material-identity-model/ or runs/v01-material-semantic-foundation/.
• Bumping material_semantic_core to 0.3.0 during Discovery. A package revision is allowed only after Scientist conclusions that actually require a semantic change.
• Construction classification (IFC, Uniclass, EPD) except as surveyed context if sources themselves raise it.

Desired Scientist question (after handoff)

Whether the v0.2 encoding (shared composition kind; phase outside kind) matches published kind-splits for polymorphs. The hypothesis must be able to fail.

Perspectives

1. p00-basic-facts. Basic facts

   Basic facts: broadly covering established facts about the topic, including definitions, scope, and commonly cited claims.

   STORM default basic-facts perspective (always included). Not derived from survey findings.

2. p01-mineral-species-nomenclature. Mineral species and crystallographic nomenclature

   How mineralogical nomenclature bodies and mineral encyclopedia treatments decide whether two solids of the same ideal chemical formula are the same species or different kinds, including topological versus merely metric or order–disorder structural differences, polytypes, and named polymorph pairs.

   Opened Nickel and Grice 1998 (IMA CNMMN procedures), Hatert et al. 2023 (CNMNC guidelines for the nomenclature of polymorphs and polysomes), and Wikipedia Mineral all organise identity as chemical composition plus crystallographic properties, with explicit species splits for topologically different structures of one formula (pyrite/marcasite, calcite/aragonite) and explicit non-splits for polytypes and structurally similar polymorphs. Wikipedia Belite/Larnite and the fetched Ca2SiO4 structure papers sit in the same mineral-naming cluster.

3. p02-substance-registry-identifiers. Chemical-substance and registry identifiers

   How chemical-registry and molecular-identifier systems, and pharmacopoeial solid-form notes, treat physical phase, allotropy, and crystal polymorphism relative to a single substance or compound identity.

   Opened CAS Registration Criteria PDF, Wikipedia CAS Registry Number, Heller et al. 2015 (InChI), the InChI Trust Technical FAQ, Wikipedia International Chemical Identifier, Wikipedia Allotropy, and British Pharmacopoeia Appendix I F form a cluster about substance/compound identifiers: one RN or InChI for a compound, with documented exceptions (minerals; graphite vs diamond) and pharmacopoeial language that polymorphs share composition while differing in the solid state.

4. p03-record-and-exchange-models. Structure records, grades, and exchange schemas

   How materials databases, APIs, and engineering-information models key a record (structure instance, grouped polymorph, grade, or material element) and where they store formula, crystal structure, named phases, grain-scale microstructure, and delivery/heat-treatment condition relative to that key.

   Opened OPTIMADE v1.3.0 specification and structures schema, Materials Project FAQ and identifier-systems pages, Gražulis et al. COD papers, OASIS MatML / NIST IR 6939 Characterization (Formula, ChemicalComposition, PhaseComposition, DimensionalDetails), Wikipedia Steel grades plus ISO 683-2 and ISO 630-4 samples (grade vs delivery condition), Wikipedia Heat treating / Microstructure, and PMDco 3.0.0 (separate chemical-composition, microstructure, and mineral classes) organise the topic as how information systems attach structure and processing fields to a record rather than as mineral-species or CAS-registry rules.

Report

Summary

Published materials-identity systems do not use one rule for whether two records are the same kind when they share a chemical formula but differ in crystal form or grain-scale structure. They decide sameness for different units: a chemical or drug substance, a mineral species, a crystal-structure record, or an engineering grade. src-cas-registration-criteria src-nickel-grice-1998 src-optimade-v13 src-matml

Three definitional splits recur. A crystal polymorph is a crystalline phase of a given substance (same composition, different crystal structure). src-iupac-pac-1994 src-iupac-pac-2011 Allotropy is restricted, in encyclopedia use, to elements in the same physical state, and does not fully overlap polymorphism (graphite is treated as chemically distinct from diamond). src-wiki-allotropy src-wiki-polymorphism-ms Microstructure is the nm–cm arrangement of phases and defects, not a formula and not a synonym for a polymorph. src-cambridge-cp1 src-wiki-microstructure

On identity, opened systems disagree. IMA/CNMNC split same-formula solids into species when bonding topology differs (pyrite/marcasite; calcite/aragonite) and keep polytypes as suffixes on one species. src-hatert-2023 src-nickel-grice-1998 CAS assigns one Registry Number across physical forms except naturally occurring minerals (water/ice/vapor share 7732-18-5); encyclopedia and ChEBI show distinct numbers for carbon, graphite, and diamond. src-cas-registration-criteria src-wiki-cas-rn Standard InChI cannot represent polymorphs; pharmacopoeia and ICH Q6A treat forms as the same substance, with optional solid-state specifications. src-inchi-trust-faq src-ich-q6a OPTIMADE, the Materials Project, and COD key a structure or determination instance, not a formula. src-optimade-v13 src-mp-identifiers src-grazulis-2012-cod Steel grades attach +N/+QT delivery conditions; pearlite and martensite are microstructures of a parent Fe–C chemistry. src-iso-683-1 src-bhadeshia-steel-micro

Prominent disagreements are whether composition is a sufficient kind key, whether a crystal polymorph mints a new kind, how carbon allotropes relate to ice/water, and how Ca2SiO4 names (belite, larnite, calcio-olivine) map across cement and mineral systems. Wikipedia states that the definition of polymorphism remains under discussion. src-wiki-polymorphism-ms No fetched source states a necessary-and-sufficient identity model for all materials.

Caller brief.md (v0.2 parking of solid-state phase in microstructure, shared composition kind for β- and γ-Ca2SiO4) is context, not evidence.

Published identity units and the matching decision

Opened sources do not share one primitive named “material kind.” They decide sameness for different units: a chemical substance, a mineral species, a crystal-structure record, or an engineering grade. src-nickel-grice-1998 src-cas-registration-criteria src-optimade-v13 src-matml

Substance, species, structure instance, and grade

A CAS Registry Number identifies a registered chemical substance. The official criteria assign one number across physical forms, with an explicit exception for naturally occurring minerals. src-cas-registration-criteria Standard InChI identifies a compound from a connection table; the current version cannot represent polymorphs. src-inchi-trust-faq src-heller-2015-inchi

An IMA mineral species is a mineral substance with well-defined chemical composition and crystallographic properties, meriting a unique name. src-nickel-grice-1998

An OPTIMADE structures entry is data belonging to a single structure; chemical formula is an entry property, not the identifier. src-optimade-v13 A Materials Project material_id is defined so that a specific polymorph can be referenced. src-mp-identifiers A COD number identifies a particular instance of a structure determination. src-grazulis-2012-cod

MatML keys a named Material; formula, phase composition, and grain-scale descriptors sit under characterization. src-matml Steel delivery standards attach heat-treatment condition to a grade (for example C45E+N+BC), rather than minting a new chemistry table. src-iso-683-1 src-wiki-steel-grades

These units are not interchangeable. Calling two records “the same material” can mean the same substance, the same species, the same structure cluster, or the same grade-plus-condition call-out.

Composition as a field versus as a sufficient key

Every opened system uses chemical composition somewhere. None of the fetched texts states that formula is a universal unique key.

CAS uses a substance key that collapses physical form except for minerals. src-cas-registration-criteria InChI is a molecular-structure string with no crystal-form layer in production. src-inchi-trust-faq IMA species require crystallography as well as composition; same ideal formula can be two species when bonding schemes differ. src-hatert-2023 Structure databases store formula as a search field or property while keying the record by structure or determination instance. src-optimade-v13 src-mp-identifiers src-grazulis-2009-cod MatML requires Formula inside Characterization, not as the document primary key. src-matml

The IUCr Working Group on Crystal Phase Identifiers treated InChI as identifying composition and molecular topology, then proposed extra layers specifically to distinguish crystalline forms of a given compound; formula alone was said to suffice only when a compound is known in one form. src-iucr-2005-phase-id That report is a recommendation, not production Standard InChI. src-iucr-2005-phase-id src-inchi-trust-faq

No opened source in this run is a universal materials-kind identifier that ranks those units into one matching rule. That gap is recorded under Open issues.

Definitions of polymorphism, allotropy, and microstructure

Fetched definitions distinguish three objects that share a surface resemblance in ordinary speech: a crystal polymorph, an allotrope, and a microstructure. They are not synonyms, and they do not all decide record identity.

Same composition with a different crystal structure

IUPAC Recommendations 1994 define a polymorphic transition as a reversible transition of a solid crystalline phase, at a stated temperature and pressure (the inversion point), to another phase of the same chemical composition with a different crystal structure. src-iupac-pac-1994 A monotropic transition is the irreversible change from a metastable polymorph to the stable one. src-iupac-pac-1994

IUPAC Recommendations 2011 define a crystal polymorph as one of the different crystalline phases in which a given substance is able to crystallize. Polymer notes add that the difference may be packing only, or packing plus chain conformation. src-iupac-pac-2011

Ph. Eur. method 5.9 / BP Appendix I F restates polymorphism as the ability of a compound in the solid state to exist in different crystalline forms having the same chemical composition. src-bp-pheur-5-9 ICH Q6A’s glossary: polymorphism is different crystalline forms of the same drug substance. src-ich-q6a

Wikipedia’s materials-science polymorphism article restates the IUPAC transition definition, adds McCrone’s claim that polymorphic phases are identical in the liquid or vapour, and states that a polymorphic transition cannot include chemical change. src-wiki-polymorphism-ms Classic pairs named there include calcite/aragonite (CaCO3) and TiO2 rutile/anatase/brookite. src-wiki-polymorphism-ms Crystal structure itself is the ordered atomic arrangement specified by unit-cell geometry and space group. src-wiki-crystal-structure

The same encyclopedia page says the definition of polymorphism has evolved and, as of the fetched 2023-era text, is still under discussion. src-wiki-polymorphism-ms IUPAC Gold Book HTML pages were not fetched (Cloudflare); PAC 1994 and 2011 supply the corresponding recommendation text. src-iupac-pac-1994 src-iupac-pac-2011

Allotropy as an element-only term

Wikipedia Allotropy restricts allotropy to chemical elements, in the same physical state; different states of matter alone are not allotropes. Polymorphism is described as the more general term for compounds (usually solids). src-wiki-allotropy Ostwald (1912) treated allotropy as a special case of polymorphism and proposed dropping the element-only term; the article states that IUPAC and most texts still keep it. src-wiki-allotropy

PAC 1994 names an allotropic transition as a change to another structure containing the same atoms but different properties, with examples including fcc iron and orthorhombic→monoclinic sulfur. src-iupac-pac-1994

Wikipedia Crystal polymorphism states that not all allotropes are polymorphs. Graphite, diamond, and lonsdaleite are allotropes of carbon, but graphite is treated as chemically distinct (sp2) from diamond/lonsdaleite (sp3); diamond and lonsdaleite are called polymorphs of each other. Iron’s α/γ/δ forms are given as allotropes that are also polymorphs. src-wiki-polymorphism-ms

Britannica allotropy pages were not fetched (Cloudflare).

Microstructure as a different length scale from crystal structure

Wikipedia Microstructure: structure revealed by an optical microscope above 25×; crystal structure is reserved for the arrangement of individual atoms; finer-than-optical structure is often called nanostructure. Grain size is said to be controlled by processing conditions and composition; multiple phases may exist at once. src-wiki-microstructure

A Cambridge materials-science handout draws the distinction in one sentence: crystal structure = lattice + atomic coordinates (Å scale); microstructure = appearance on the nm–cm scale, working definition “the arrangement of phases and defects within a material.” src-cambridge-cp1 A phase has a distinct crystal structure and/or chemical composition; a multi-component material can still be one phase if mixed at atomic scale (solid solution). src-cambridge-cp1 Microstructures form by phase transformation, deformation/processing, or making a composite. src-cambridge-cp1

Rhines (fetched SciELO extract): crystal structure = atoms on lattices; microstructures = space-filling distributions of phases and their boundaries; a qualitative microstructural state lists phases with their compositions and crystal structures plus 3-/2-/1-/0-dimensional features. src-scielo-rhines

These definitions do not treat microstructure as another name for chemical composition, nor as a synonym for a polymorph. Two solids can share composition and still differ in crystal structure (polymorphs) or share a crystal-structure type and still differ in grain, defect, or phase-mixture arrangement (microstructure). src-cambridge-cp1 src-iupac-pac-1994 No opened source in this definitional cluster states a single rule for when two materials-database records are the same kind.

Mineral-species nomenclature

IMA/CNMNC texts decide whether two naturally occurring solids are one species or two. The unit is not a CAS substance, an OPTIMADE structure, or an engineering grade. src-nickel-grice-1998

Composition plus crystallography

Nickel and Grice 1998: a mineral species is a mineral substance with well-defined chemical composition and crystallographic properties, meriting a unique name. A new species is possible if composition or crystallography (or both) is substantially different. src-nickel-grice-1998 Compositional splits usually require a different dominant occupant of at least one structural site. src-nickel-grice-1998

Hatert, Mills, Pasero, Miyawaki and Bosi 2023 keep that species rule and add naming prefixes for crystal system or space group. They also state that names with “polymorph” suffixes but different compositions are not true polymorphs. src-hatert-2023

Species status is a Commission decision, not an automatically computed formula key. Borderline cases are judged on their own merits. src-nickel-grice-1998

Topology versus distortion, order–disorder, and polytypes

Polymorphic minerals have essentially the same chemical composition but different crystal structures. They are different species if topologies (bonding schemes) differ, and not different species if only distortion or order–disorder differs. src-nickel-grice-1998 src-hatert-2023 Graphite versus diamond is the topology-split example; analcime symmetry variants are the similar-topology non-split. src-nickel-grice-1998 Hatert et al. suggest renaming “topologically similar” as “structurally similar” polymorphs. src-hatert-2023

Polytypes (layer stacking of nearly identical layers) and polytypoids (same topology, somewhat different composition; pyrrhotite vacancy-order forms) are not separate species; they may take a crystallographic suffix that is not part of the species name (molybdenite-2H/-3R; muscovite-1M/-2M/-3T). src-nickel-grice-1998 src-hatert-2023 Those suffixes can also label structurally similar polymorphs. src-hatert-2023 Regular interstratifications and some polysomes may be species; modulated variants of an existing species (antigorite waves) may not. src-nickel-grice-1998 src-hatert-2023

Historical names can persist even when topology is similar (orthoclase and microcline). src-nickel-grice-1998 Mindat lists wollastonite as one grandfathered IMA species with several polytypes (-1A, -2M, …) and separately lists breyite, davemaoite, and pseudowollastonite as polymorphs of that species. src-mindat-wollastonite

Bosi 2022 PDF timed out; any later proposal to replace “topology” with a stricter structural-similarity test is not evidenced from fetched text in this run.

Same-formula species pairs

Hatert et al. 2023 cite pyrite/marcasite, calcite/aragonite, and andalusite/kyanite/sillimanite as same ideal formula, distinct bonding schemes, therefore separate species. src-hatert-2023 Mindat: marcasite is a grandfathered IMA species, the orthorhombic polymorph of isometric pyrite, formula FeS2. src-wiki-marcasite Handbook of Mineralogy: pyrite is dimorphous with marcasite. src-hom-pyrite

Wikipedia Mineral: if a compound occurs with different crystal structures, each structure is a different mineral species; quartz and stishovite are both SiO2. src-wiki-mineral Stishovite is a separate IMA-named tetragonal species with six-coordinate Si (rutile-type), not a quartz variety. src-wiki-mineral Rutile and anatase are separate mineral articles for TiO2. src-wiki-rutile-anatase

This is a structure-based kind split in mineralogy. It contradicts any claim that same ideal formula is sufficient for one mineral species. src-hatert-2023 src-nickel-grice-1998

Ca2SiO4 names across mineral and cement systems

Same formula Ca2SiO4 is used in at least two naming systems, which the opened sources do not equate.

Wikipedia Belite: belite is a cement-industry name, not a recognised mineral name; the natural mineral named is larnite; clinker belite is a substituted solid solution, not pure Ca2SiO4. src-wiki-belite PCA and TRB cement guides use belite/alite as portland-cement phase labels for impure C2S/C3S. src-cement-belite

Wikipedia Larnite: IMA-symbolled mineral Ca2SiO4, monoclinic P21/n; cement industry “usually referred to as belite.” src-wiki-larnite The same page calls larnite the calcium member of the olivine group. src-wiki-larnite

Mindat’s Olivine Group member table lists Calcio-olivine, Ca2SiO4, orthorhombic, and does not list larnite. src-mindat-olivine-group Tzoumerkas et al. 2022: five Ca2SiO4 polymorphs (α, α′H, α′L, β, γ); γ is Calcio-Olivine and non-hydraulic; β “does not belong to the olivine group.” src-crystals-2022-belite

Wikipedia Larnite’s olivine-group sentence therefore conflicts with the opened Mindat group table and with the 2022 crystal-chemistry paper. src-wiki-larnite src-mindat-olivine-group src-crystals-2022-belite A dedicated species-page fetch for calcio-olivine failed (Wikipedia 404; Mindat and Zadov timeouts), so IMA approval details for that name are not recovered from fetched text here.

“Belite” is not an IMA species competing with larnite or calcio-olivine. Cement literature keys a substituted clinker phase with several thermal polymorphs; mineral nomenclature keys natural species. Same formula, different identity systems. src-wiki-belite src-wiki-larnite src-crystals-2022-belite

Chemical-substance registries and identifiers

CAS, InChI, and pharmacopoeial texts decide sameness for a substance or drug substance. That unit is not an IMA species vote and not an OPTIMADE structure key. src-cas-registration-criteria src-inchi-trust-faq src-ich-q6a

CAS physical form, minerals, and carbon allotropes

Official CAS overview: a single Registry Number for a substance regardless of varying physical forms, except for naturally occurring minerals. Worked example: water, ice, and water vapor share 7732-18-5. Stereoisomers and salt-ratio forms are split; conformational isomers are not. src-cas-registration-criteria

Wikipedia’s granularity section: “distinct” excludes different phases of the same substance (water/ice) but includes narrowing such as carbon versus graphite and diamond; it restates separate RNs 7440-44-0 / 7782-42-5 / 7782-40-3 and generalizes that different crystal structures receive different RNs. src-wiki-cas-rn Wikipedia Carbon lists those three CAS numbers in the infobox and describes graphite and diamond as well-known allotropes. src-wiki-carbon ChEBI treats graphite as both an allotrope of carbon and a native-element mineral, lists CAS 7782-42-5, and further splits hexagonal versus rhombohedral graphite as children. src-chebi-graphite

The official PDF therefore states a mineral exception, not a blanket “every crystal structure gets an RN” rule. src-cas-registration-criteria The encyclopedia generalization to all crystal structures is stronger than the fetched CAS criteria text. src-wiki-cas-rn No opened CAS table in this run assigns or withholds separate RNs for pyrite versus marcasite.

Physical-form collapse and crystal-structure split are both present in the same registry corpus once encyclopedia and ChEBI are included. The RN identifies a CAS “substance,” which may or may not coincide with a crystal-structure kind. src-cas-registration-criteria src-wiki-cas-rn

Standard InChI and the IUCr crystal-phase proposal

Production InChI is a molecular-structure string (formula, connectivity, charge, tautomer H, stereo, isotopes). The InChI Trust FAQ lists Polymorphs among things the current version cannot represent, alongside conformers, mixtures, and polymers. src-inchi-trust-faq

Heller, McNaught, Pletnev, Stein and Tchekhovskoi 2015: chemists expect steam, ice, and liquid water to share a chemical identifier; an identifier’s “identifying power is inherently limited” for aggregate states and polymorphs. src-heller-2015-inchi No crystal-structure or space-group layer is described as part of production Standard InChI. src-heller-2015-inchi

The IUCr 2005 Working Group on Crystal Phase Identifiers proposed extra InChI-compatible layers (state of matter PH:, space-group number SG:, optional Wyckoff sequence WS:) so that, for example, rutile could be written TiO2 plus xtl plus SG 136. src-iucr-2005-phase-id Version 1.0 is restricted to finite-molecule topology and is not designed for infinite-structure connectivity. src-iucr-2005-phase-id Remaining collisions are noted (two bcc iron phases sharing SG 229). src-iucr-2005-phase-id Those layers are recommendations, not the live Standard InChI layer set. src-iucr-2005-phase-id src-inchi-trust-faq

Pharmacopoeial and ICH treatment of solid forms

Ph. Eur. 5.9 / BP Appendix I F: polymorphism is different crystalline forms of a compound with the same chemical composition. All forms of a species share solution or melt chemistry; solid-state properties (solubility, melting point, bioavailability) may differ. src-bp-pheur-5-9 Monograph language “shows polymorphism” can mean true polymorphs, solvates, allotropy, or amorphous material. src-bp-pheur-5-9 The text prefers “solvates” and “hydrates” over “pseudopolymorphism.” src-bp-pheur-5-9 This chapter was inspected via a third-party HTML mirror, not the publisher site.

USP 〈197〉: IR mismatch may be polymorphism; the default identity procedure is to recrystallize sample and reference standard together and re-run, so identity is not the as-received crystal form unless the monograph specifies a form. src-usp-197-941 USP 〈941〉: several polymorphs or solvates of one compound, illustrated with four solid phases of ampicillin. src-usp-197-941 These chapters were inspected via unofficial HTML mirrors.

ICH Q6A glossary: polymorphism = different crystalline forms of the same drug substance, including solvation or hydration products and amorphous forms. src-ich-q6a Solid state is specified only when performance, bioavailability, or stability is affected; decision trees decide whether to set acceptance criteria for polymorph content. src-ich-q6a The forms remain forms of the same drug substance, not separate new substances by default. src-ich-q6a

Opened pharmacopoeial and ICH texts therefore treat crystal form as subordinate to substance identity, with optional solid-state specifications. They do not state that each polymorph is a separate official substance.

Structure-record databases

OPTIMADE, the Materials Project, and the Crystallography Open Database key a structure (or a grouped set of similar calculations, or a structure-determination instance). Formula is a property or a duplicate-check field, not the unique key. src-optimade-v13 src-mp-identifiers src-grazulis-2012-cod These systems do not claim to be mill-grade or CAS-substance identity.

OPTIMADE structure entries

OPTIMADE v1.3.0: an entry is a single instance of a resource type; a structures entry is data belonging to a single structure. src-optimade-v13 Chemical formula is an entry property (for example chemical_formula_descriptive, chemical_formula_reduced). src-optimade-v13 Taken together, ID and entry type must uniquely identify the entry; IDs may change over time. src-optimade-v13 dimension_types encodes periodicity (molecule versus slab versus bulk). src-optimade-v13

Formula is therefore not the unique key of an OPTIMADE structure record. Same formula can correspond to many structure entries.

Materials Project polymorph identifiers

The Materials Project FAQ states the purpose of material_id as allowing “a specific polymorph of a given material to be referenced,” with wurtzite GaN = mp-804 and zinc-blende GaN = mp-830. src-mp-identifiers A task_id is one calculation; a material_id aggregates tasks. src-mp-identifiers

Identifier-systems documentation: tasks are grouped by structures similar enough to be identified as the same material (diamond-cubic Si at 5 Å versus 5.1 Å would match) but not different structure types of the same composition (diamond-cubic Si versus simple-cubic Si would not). src-mp-identifiers

The identifier is a grouped computational material/polymorph, not a chemical-substance RN and not an IMA species vote. Project documentation is not a peer-reviewed identity standard; class remains unknown after inspection. src-mp-identifiers

COD structure-determination instances

Gražulis et al. 2009: each structure deposited in the COD or PCOD gets a unique seven-digit identifier; if a structure of a compound is redetermined, with higher precision or under different conditions, it is deposited under a new number. src-grazulis-2009-cod

Gražulis et al. 2012: a COD number identifies a particular instance of a structure determination. Two structures of the same compound published in two journals can both be deposited and receive distinct numbers; a higher-resolution re-solution likewise gets a new number. src-grazulis-2012-cod Duplicate detection compares formula plus unit cell (and optional T/P/history), not formula alone; the algorithm is not intended to find similar compounds in different cell settings. src-grazulis-2012-cod

The COD CIF dictionary: _cod_database_code_structure is a persistent unique identifier assigned upon deposition and is the preferred way of referencing individual COD entries. src-cod-cif-dict

Opened structure databases therefore contradict any claim that chemical formula is the unique record key. Same formula can have many IDs (polymorphs; near-identical lattices grouped or not; experimental redeterminations). src-mp-identifiers src-grazulis-2012-cod

Engineering exchange, grades, and heat-treatment products

Engineering information models and steel-delivery texts key a named material or a grade, then attach composition, phase, grain-scale structure, and heat-treatment condition as characterization or delivery fields. They do not, in the opened passages, treat grain-scale microstructure as a new bulk chemical formula. src-matml src-iso-683-1 src-bhadeshia-steel-micro

MatML characterization slots

OASIS MatML / NIST IR 6939: MatML_Doc contains one or more Material elements. Material has an optional id. Required BulkDetails includes Name (once) and optional Class, Subclass, Specification, Source, Form, processing, and Characterization. src-matml

If Characterization is present, Formula must occur once: a string representation of the chemical formula for the bulk material or component. ChemicalComposition may occur once or not at all. PhaseComposition may occur zero or more times (phase Name plus optional concentration). DimensionalDetails (grain size, porosity, precipitate size and distribution, and similar) may occur zero or more times. src-matml

Formula is therefore a required string inside characterization, not the document primary key. Phase and grain-scale descriptors are optional siblings of Formula under the same named Material. src-matml

Grade plus delivery or heat-treatment condition

ISO 683-1 sample text specifies technical delivery requirements for named non-alloy steels supplied in one of the heat-treatment conditions in Table 1 and one of the surface conditions in Table 2. src-iso-683-1 Cast analysis must comply with the grade chemistry table; mechanical properties apply as appropriate for the particular heat-treatment condition. src-iso-683-1 Ordering example: steel grade C45E in heat-treatment condition +N, surface +BC, written ISO 683-1 – C45E+N+BC. src-iso-683-1 These passages come from preview/sample PDFs, not a purchased full text.

ISO 630-3 landing text lists as-rolled, normalized/normalized-rolled, and thermomechanical processed delivery conditions for structural grades; not all grades exist in all qualities. src-iso-630-3 The full ISO 630-3 standard was not opened.

Wikipedia Steel grades / EN 10027: grades classify by composition and/or properties. Category 1 names are specified by purpose and mechanical properties (for example S355); category 2 by chemical composition. Additional symbols after + mark delivery condition (+N, +QT, annealed, untreated). src-wiki-steel-grades SAE AIR 4127A sample: chemical composition indicates only one aspect of a steel product and is not a complete guide to quality and performance, yet composition is the most common basis for classification and designation. src-sae-air-4127a

Two pieces with the same grade chemistry but different + conditions are the same grade designation, not the same delivered product state. src-iso-683-1 src-wiki-steel-grades Opened texts do not treat grain-scale microstructure as the primary key. Official EN 10027 PDF was not opened. Mill-certificate practice stating whether quenched and annealed pieces of one grade are “the same material kind” was not fetched.

Pearlite and martensite at parent chemistry

Bhadeshia: in Fe–C systems, the average chemical composition of pearlite is identical to that of the parent austenite, so austenite can transform completely to pearlite; martensite is a diffusionless lattice deformation of austenite. src-bhadeshia-steel-micro

Wikipedia Pearlite: pearlite is a two-phased lamellar structure of ferrite and cementite; it is a microstructure occurring in structural steels. src-wiki-pearlite Wikipedia Heat treating: annealing versus quenching of steel is presented as changing microstructure (ferrite+cementite layers versus trapped-carbon martensite) by cooling rate. src-wiki-heat-treating

This pair is not introduced as two mineral species or two CAS Registry Numbers. It is two transformation products / heat-treatment outcomes of one alloy chemistry. src-bhadeshia-steel-micro src-wiki-heat-treating Named products (pearlite, martensite) look like material names but are classified in these texts as microstructures of a parent phase. src-wiki-pearlite src-bhadeshia-steel-micro

Ontologies that separate composition, crystal phase, and microstructure

PMDco lists mineral, chemical composition, and microstructure as separate entities. A mineral is a naturally occurring material characterized by a defined chemical composition and a specific crystal structure. Chemical composition is an intensive quality describing types and proportions of elements. Microstructure is the small-scale structure of a material, including grains, phases, and defects, visible under a microscope. src-pmdco Competency questions ask independently for identifier, composition, specifications, phases, and how microstructure changes after processing. src-pmdco The ontology files are community artifacts; they are not classified here as a peer-reviewed identity standard.

Cambridge CP1, already used for definitions, likewise keeps crystal structure (unit-cell atomic positions) distinct from microstructure (nm–cm arrangement of phases and defects). src-cambridge-cp1 A phase has distinct structure and/or composition; a solid solution can be one phase with several chemical components. src-cambridge-cp1

These sources describe annotation axes. They do not, by themselves, decide “same kind” for two records across CAS, IMA, OPTIMADE, and mill grades. They do contradict collapsing composition, crystal polymorph, and grain-scale microstructure into one field. src-pmdco src-cambridge-cp1 src-matml

When systems split or merge same-formula records

The same chemical formula is split or merged depending on the system’s unit. Opened sources do not agree on a single matching rule.

Mineral species (split when bonding topology differs). FeS2 is two IMA species (pyrite, marcasite). src-hatert-2023 src-wiki-marcasite src-hom-pyrite CaCO3 calcite/aragonite and Al2SiO5 andalusite/kyanite/sillimanite are the same pattern. src-hatert-2023 SiO2 quartz and stishovite are separate species. src-wiki-mineral TiO2 rutile and anatase are separate mineral articles. src-wiki-rutile-anatase Graphite and diamond are separate species under the topology rule. src-nickel-grice-1998

Mineral species (merge stacking and similar-topology differences). Polytypes and polytypoids are not species; suffixes are not the species name. src-nickel-grice-1998 src-hatert-2023 Wollastonite polytypes share one species. src-mindat-wollastonite Distortion or order–disorder with the same topology is not a new species (analcime). src-nickel-grice-1998 Exceptions: historical names; some polysomes and regular interstratifications. src-nickel-grice-1998

CAS substance (merge physical form; split some minerals/allotropes). Water, ice, and vapor share 7732-18-5. src-cas-registration-criteria Naturally occurring minerals are an official exception to the single-RN rule. src-cas-registration-criteria Encyclopedia and ChEBI show distinct RNs for carbon, graphite, and diamond. src-wiki-cas-rn src-wiki-carbon src-chebi-graphite

InChI and pharmacopoeia (merge crystal forms into one substance). Standard InChI cannot represent polymorphs. src-inchi-trust-faq Heller et al. expect ice/steam/liquid water to share an identifier. src-heller-2015-inchi ICH Q6A and Ph. Eur. 5.9 treat polymorphs as forms of one substance; USP identity IR may recrystallize away form differences. src-ich-q6a src-bp-pheur-5-9 src-usp-197-941 IUCr 2005 proposed extra layers to split crystalline forms; those layers are not production InChI. src-iucr-2005-phase-id

Structure databases (split by structure or determination instance). GaN wurtzite versus zinc blende: two Materials Project material_ids, one formula. src-mp-identifiers OPTIMADE keys a structure; formula is a property. src-optimade-v13 COD keys a determination instance; redeterminations get new numbers. src-grazulis-2009-cod src-grazulis-2012-cod

Engineering grades (merge chemistry; attach condition). ISO 683-1 and EN-style +N/+QT keep the grade and add delivery/heat-treatment. src-iso-683-1 src-wiki-steel-grades Pearlite versus martensite is a microstructural pair at one Fe–C chemistry, not two bulk formulae. src-bhadeshia-steel-micro src-wiki-pearlite

Ca2SiO4 (mixed systems, not one merge). Belite is a cement-phase label, not an IMA name. src-wiki-belite Larnite is a named mineral; γ-Ca2SiO4 appears as calcio-olivine in an olivine-group table; cement literature lists five thermal polymorphs. src-wiki-larnite src-mindat-olivine-group src-crystals-2022-belite Those names are not shown to be one species or one industrial kind.

No opened source in this run states when two mill certificates are the same kind, nor a necessary-and-sufficient identity model covering all of the units above.

Open issues

The isolated outline asked several questions that fetched sources do not answer, or answer only inside one unit.

No cross-system necessary-and-sufficient kind model. Opened IMA, CAS, InChI, pharmacopoeia, OPTIMADE, COD, MP, MatML, PMDco, and steel-grade texts each identify a scoped entity. None of them states a published minimum model that decides “same material kind” across minerals, chemical substances, computed structures, drug substances, and engineering grades. That claim is left unresolved. Constraints on what must not be collapsed (formula is not a universal key; microstructure is not a crystal polymorph; topology splits mineral species; InChI does not encode polymorphs) are the positive result, not a substitute for such a model.

Official CAS tables for non-carbon polymorphs. Encyclopedia text generalizes “different crystal structures receive different RNs.” src-wiki-cas-rn The official criteria PDF states a mineral exception and a water/ice merge, not a blanket crystal-structure rule. src-cas-registration-criteria Whether pyrite and marcasite, or rutile and anatase, have distinct CAS RNs was not recovered from a CAS table in this run.

Calcio-olivine IMA status. Dedicated species pages failed (Wikipedia 404; Mindat and Zadov 2009 DOI timeouts). Wikipedia Larnite’s olivine-group claim conflicts with the opened Mindat olivine-group member table and with Tzoumerkas et al. 2022. src-wiki-larnite src-mindat-olivine-group src-crystals-2022-belite Current IMA list status of calcio-olivine is therefore not evidenced here.

IUCr 2005 layers are not shown as implemented. Opened InChI sources say polymorphs are out of scope of the current identifier. src-inchi-trust-faq src-heller-2015-inchi Production Standard InChI with PH/SG/WS layers was not fetched.

Mill certificates and quenched versus annealed product. ISO 683-1 and EN-style delivery symbols distinguish grade from condition. src-iso-683-1 src-wiki-steel-grades They do not answer a universal kind-id question for two heats of one grade. Full purchased ISO 683-1 / EN 10027 / ISO 630-3 bodies were not opened.

Fetch failures. IUPAC Gold Book HTML (Cloudflare). Bosi 2022 PDF (timeout). Britannica allotropy pages (Cloudflare). Official pharmacopoeia publisher HTML (mirrors used). src-iupac-pac-1994 src-bp-pheur-5-9 src-usp-197-941

Headings from the isolated draft with no fetched body. Nanostructure and size-dependent identity, hydrates/co-crystals as a general identity system beyond ICH/Ph. Eur. notes, incommensurate crystals, and a single identity model covering metals, ceramics, polymers, and molecular crystals were not evidenced as published matching rules in this corpus. They are not filled from unaided memory.

Caller brief.md (v0.2 parking of solid-state phase in microstructure, shared composition kind_id for β- and γ-Ca2SiO4) is context, not a source. It is not cited as evidence.

Aggregation

Aggregation

Topic: How do published materials-identity systems treat solid-state polymorphs and microstructure relative to chemical composition when deciding whether two records are the same material kind?

Caller brief.md (v0.3 motivation: v0.2 parks solid-state phase in microstructure and lets β- and γ-Ca₂SiO₄ share a composition kind_id) is caller assumption, not evidence. It is not cited below.

Agreed across opened sources

1. Polymorphism is defined as same chemical composition, different crystal structure. IUPAC PAC 1994 (polymorphic transition) and PAC 2011 (crystal polymorph of a given substance) state this. Pharmacopoeial Ph. Eur. 5.9 restates “different crystalline forms having the same chemical composition.” Encyclopedia pages use calcite/aragonite and TiO₂ rutile/anatase as textbook pairs.

1. Microstructure is not a synonym for composition or for a crystal polymorph. Cambridge CP1, Wikipedia Microstructure, Rhines, and PMDco treat microstructure as nm–cm arrangement of phases, grains, and defects. Crystal structure is Å-scale lattice + coordinates. A phase has distinct structure and/or composition; a solid solution can change composition without becoming a second phase.

1. Chemical formula is not a universal record key. Opened identifier systems all use composition as part of the description, then diverge on whether it is sufficient to say two records are the same kind.

1. Heat-treatment products at essentially one bulk chemistry are not presented as new formulae. Bhadeshia: pearlite’s average composition equals parent austenite. Wikipedia Heat treating / Pearlite: annealing vs quenching changes microstructure (pearlite vs martensite). ISO 683-1 and EN 10027-style +N/+QT attach delivery/heat-treatment to a grade, not a new chemistry table.

1. Structure databases key a structure (or grouped isomorphic calculations), not a formula. OPTIMADE: formula is an entry property. Materials Project material_id names a specific polymorph (GaN mp-804 vs mp-830). COD numbers a structure-determination instance; redeterminations get new IDs; duplicate checks use formula and cell.

1. MatML and PMDco keep formula / phase / grain-scale descriptors in different slots. MatML: Name is the Material key; Formula, PhaseComposition, and DimensionalDetails are characterization. PMDco: mineral (composition + crystal structure), chemical composition, and microstructure are distinct entities.

Conflicts (do not collapse)

Missing

• A published necessary-and-sufficient identity model that decides “same material kind” across minerals, CAS substances, computed structures, pharmacopoeial substances, and engineering grades (caller F9; not recovered here).
• Official CAS Registry table proving whether pyrite vs marcasite (or rutile vs anatase) receive distinct RNs. Encyclopedia states a crystal-structure split for carbon; the official criteria PDF only names the mineral exception.
• Fetched IMA list / Mindat species pages for calcio-olivine (Wikipedia 404; Mindat and Zadov 2009 DOI timeouts). Current IMA status of that name is therefore not recovered from fetched text.
• IUPAC Gold Book HTML (Cloudflare). Bosi 2022 PDF (timeout). Britannica allotropy pages (Cloudflare).
• Production implementation of IUCr 2005 InChI crystal-phase layers (PH/SG/WS). Opened InChI sources say those layers are not Standard InChI.
• Mill-certificate or product-specification practice stating whether quenched and annealed pieces of one grade are “the same material kind.” Opened steel texts distinguish grade vs delivery condition; they do not answer a universal kind-id question.
• Full purchased ISO 683-1 / EN 10027 / ISO 630-3 bodies (previews and encyclopedia used).
• Official pharmacopoeia publisher HTML (Ph. Eur. 5.9 and USP chapters were inspected via mirrors).

Caller assumptions (not evidence)

The brief assumes v0.2 encoding: solid-state phase parked in microstructure, β- and γ-Ca₂SiO₄ sharing a composition kind_id. Opened sources support treating grain-scale heat-treatment products (pearlite/martensite) as microstructure of a parent chemistry, not new formulae. They contest treating crystal polymorphs with different bonding topology as the same mineral species, and contest treating MP/COD/OPTIMADE records as formula-keyed. They split Ca₂SiO₄ across systems: cement “belite” is not an IMA name; larnite vs γ-calcio-olivine are not shown to be one mineral species. They do not supply a single published rule that phase must live inside or outside a composition kind for all materials. F9 remains unanswered by the opened corpus.

Handoff findings

Polarity is supporting / opposing / context for the literal claim.

Experiments

Scientist conclusions — v03-polymorph-microstructure

Question (H2 vs H3 vs H7, then H10, then H8 from observed results):

On fixtures reconstructed from fetched identity rules, does a composition-only kind encoder match published polymorph splits, or do IMA topology/polytype and structure-record keys split pairs that composition collides?

Discovery files were not edited. Caller brief.md is not experimental evidence. F9 was not resolved as a positive universal model.

What to preserve (experimental)

Composition-only collides same-formula polymorphs that IMA topology and MP/COD/OPTIMADE structure-record keys split (SH1 supported as collision; SH2 and SH3 supported as published splits).

Composition-only matching IMA or MP equalities on the dimorph pairs is false on these fixtures. That is evidence against F2 as a cross-system sufficient kind key. It is not a rewrite of F2.

Chain

E001  draft   H2 / H3 / H7 polymorph keys
      SH1, SH2, SH3 supported
   ↓  composition_same on FeS2 and GaN; ima_same false on pyrite/marcasite;
      structure_same false on mp-804/mp-830
E002  branch  H10 Ca2SiO4 name classes
      SH4, SH5, SH6 supported
   ↓  belite industrial_phase_label ≠ larnite ima_mineral;
      composition still hill:Ca2O4Si
E003  branch  H8 grade+condition / microstructure
      SH7, SH8, SH9 supported
   ↓  stop (3/8 nodes; selected question answered)

Literature vs experiment

None rejected. None invalid_formulation. None experiment_failed. H9 remains untestable (F9 not claimed as a positive universal model).

Stop reason

journal.json. The selected H2/H3/H7 questions, then H10, then H8, each have a discriminating node caused by the previous result. Further fixtures would repeat these encoders. Caps not exhausted (3/8 nodes, revision depth 2). F9 remains unresolved in discovery and is not an experimental necessary-and-sufficient model. No material_semantic_core version bump is implied by this method.

Weaknesses

• Single seed=0; encodings are deterministic (no Monte Carlo).
• Topology, MP/COD ids, name classes, and mass fractions are reconstructed fixtures, not live APIs or mill certificates.
• Wikipedia Larnite olivine-group wording was deliberately not used as the membership table; current IMA-list status of calcio-olivine was not fetched.
• The experiment shows which tokens match pre-registered equalities. It does not produce a published ontology or a ranked minimum tuple.
• Same agent ran and recorded the loop.

Conversations

p00-basic-facts

Turn 01

Q. What published definitions exist for solid-state polymorphism, allotropy, and microstructure, and how do those definitions relate each term to chemical composition? Under which stated conditions are two solids with the same composition still distinguished, and under which are they not?

IUPAC 1994: a polymorphic transition is reversible, same chemical composition, different crystal structure. IUPAC 2011: a crystal polymorph is a crystalline phase of a given substance. Allotropy is element-only and same physical state; not all allotropes are polymorphs (graphite vs diamond). Microstructure is nm–cm arrangement of phases and defects, distinct from crystal structure and from composition. Gold Book HTML was not fetched.

Turn 02

Q. In published identifier or numbering systems for substances, minerals, or computed materials, is a chemical formula (or composition) treated as enough to say that two records are the same kind? What additional attributes, if any, are required or used to keep records distinct?

Formula is not a universal key. CAS: one RN across physical forms except naturally occurring minerals; water/ice share a RN; graphite/diamond do not. IMA: same composition yields two species when bonding topology differs, not when only order–disorder differs. Materials Project material_id names a specific polymorph (GaN mp-804 vs mp-830). IUCr 2005 proposed adding state, space group, and Wyckoff layers to InChI; that is a recommendation, not production InChI.

Turn 03

Q. What commonly cited pairs (or sets) of solids share a chemical formula, and how do published sources treat those pairs: as the same kind, as different kinds, or as the same kind with an extra form/condition label? Include at least one case that is about crystal structure and one that is about processing or grain-scale structure if sources provide it.

Mineral encyclopedia and species pages split FeS2 (pyrite/marcasite), TiO2 (rutile/anatase), SiO2 (quartz/stishovite). Carbon allotropes are one element but three CAS numbers. Pearlite and martensite are transformation products of austenite at a given Fe–C chemistry, not presented as new bulk formulae. Same-formula pairs are therefore split or merged depending on the system’s unit (species, RN, heat-treatment product).

p01-mineral-species-nomenclature

Turn 01

Q. How do published IMA/CNMNC mineral-nomenclature guidelines decide whether two naturally occurring solids that share (essentially) the same chemical composition are one mineral species or two? What role is assigned to crystal structure, and what role to composition?

Species = well-defined composition plus crystallographic properties. Same formula can be two species if bonding topology differs (pyrite/marcasite; quartz/stishovite). Distortion or order–disorder with the same topology is not a new species. Compositional splits usually need a different dominant occupant of a structural site.

Turn 02

Q. How do published IMA/CNMNC nomenclature texts treat polytypes, polytypoids, and polymorphs whose structures are described as similar (for example, only a metric distortion or a difference of cation order) relative to polymorphs whose bonding topology differs? Are those cases given separate species names, suffixes, or other labels, or are they kept as one species?

Polytypes, polytypoids, and structurally/topologically similar polymorphs are not separate species; suffixes are not part of the species name (molybdenite-2H/-3R; wollastonite-1A/-2M). Topologically different structures are separate species. Polysomes and regular interstratifications may be species. Bosi 2022 PDF was not fetched.

Turn 03

Q. For named solids that share an ideal formula (including Ca2SiO4 phases such as larnite, calcio-olivine, and belite, and other published same-formula mineral pairs), how do mineral-species lists and related encyclopedic or crystallographic pages treat those names: as separate species, as polytypes or varieties of one species, as industrial/cement labels rather than species, or as something else?

Pyrite/marcasite and calcite/aragonite are separate IMA species at one formula. Belite is a portland-cement phase name, not a recognised mineral name; larnite is the named natural β-Ca2SiO4 mineral. Calcio-olivine appears as orthorhombic Ca2SiO4 in the olivine group; dedicated species-page fetches failed. Cement literature lists five thermal Ca2SiO4 polymorphs; γ is non-hydraulic.

p02-substance-registry-identifiers

Turn 01

Q. How do published CAS Registry (or closely related chemical-substance registry) criteria treat physical form, phase, allotropy, and crystal polymorphism when deciding whether two records receive the same Registry Number? Is composition treated as enough, and what published exceptions exist?

CAS assigns one RN regardless of physical form except naturally occurring minerals; water/ice/vapor share 7732-18-5. Wikipedia and ChEBI show distinct RNs for carbon vs graphite vs diamond. The official PDF does not state a blanket ‘every crystal structure gets an RN’ rule.

Turn 02

Q. How do published InChI specifications or FAQs treat solid-state polymorphism, crystal structure, or physical phase relative to a single identifier string? What, if anything, is encoded for condensed-phase or crystal form, and what is explicitly out of scope?

Standard InChI cannot represent polymorphs. Heller et al. expect ice/steam/liquid water to share an identifier. IUCr 2005 proposed extra PH/SG/WS layers (e.g. rutile) but those are recommendations, not production Standard InChI.

Turn 03

Q. How do published pharmacopoeial or other official substance-monograph texts treat solid-state polymorphic forms relative to a named drug substance or chemical? Are different crystal forms presented as the same substance with form notes, as different substances, or as unspecified?

Ph. Eur. 5.9: same composition, different crystalline forms. USP identity IR may recrystallize away form differences. ICH Q6A: forms of the same drug substance; specify solid state only if performance is affected.

p03-record-and-exchange-models

Turn 01

Q. In published materials-structure databases or APIs, what is the primary key of a record (formula, grouped material, specific crystal structure, or something else), and where are chemical composition, crystal structure, and named polymorphs stored relative to that key?

OPTIMADE keys a single structure; formula is a property. MP material_id names a specific polymorph (GaN mp-804 vs mp-830) and groups near-identical lattices. COD numbers a structure-determination instance; redeterminations get new IDs.

Turn 02

Q. In published engineering or materials-information exchange models (for example MatML-style characterization schemas, steel-grade standards, or related grade/condition documents), is the identity of a material record the chemical composition, a named grade, a delivery/heat-treatment condition, a microstructure description, or a combination? Where are phase, grain-scale structure, and processing stored relative to that identity?

MatML keys a named Material; Formula/PhaseComposition/DimensionalDetails are characterization children. ISO 683-1 orders grade plus heat-treatment condition (C45E+N). EN grades may be property-based (S355) or chemistry-based; +N/+QT are delivery symbols.

Turn 03

Q. How do published materials ontologies or processing encyclopedias place chemical composition, named mineral/crystal phases, and grain-scale microstructure (including heat-treatment products) relative to one another? Are they the same class of identity, or distinct classes/attributes?

PMDco and Cambridge CP1 keep composition, crystal structure, and microstructure distinct. Pearlite and martensite are transformation products at essentially the same bulk Fe–C chemistry, not new formulae.

Evidence

1. ev-iupac-1994-poly-transition src-iupac-pac-1994. polymorphic [transition]: A reversible transition of a solid crystalline phase at a certain temperature and pressure (the inversion point) to another phase of the same chemical composition with a different crystal structure.

   Published definition of solid-state polymorphism as same composition, different crystal structure.

2. ev-iupac-1994-monotropic src-iupac-pac-1994. monotropic transition: The irreversible transition from a metastable polymorphic form to the stable polymorph.

   Polymorphs of one composition can be related by irreversible as well as reversible transitions.

3. ev-iupac-1994-allotropic src-iupac-pac-1994. allotropic transition: [transition of] structure to another which contains the same atoms but which has different properties. Examples named in the fetched text include face-centred cubic iron, and the transition of orthorhombic sulfur to monoclinic sulfur.

   IUPAC allotropic transition is an element/same-atoms structural change, not a composition change.

4. ev-iupac-2011-crystal-polymorph src-iupac-pac-2011. crystal polymorph / crystal modification / crystalline form: One of the different crystalline phases in which a given substance is able to crystallize. Notes: polymer polymorphs may differ by packing only, or by packing plus chain conformation.

   A crystal polymorph is a crystalline phase of a given substance, not a new chemical composition.

5. ev-wiki-poly-definition-open src-wiki-polymorphism-ms. In crystallography, polymorphism is a phenomenon where a compound or element can crystallize into more than one crystal structure. The definition of polymorphism has evolved over the years and as of 2023 is still under discussion.

   Encyclopedia restatement of the structure-based definition, with an explicit caveat that the definition remains under discussion.

6. ev-wiki-poly-no-chemical-change src-wiki-polymorphism-ms. These defining facts imply that polymorphism involves changes in physical properties but cannot include chemical change. … allotropes of an element are not always polymorphs. A common example is the allotropes of carbon … graphite is not a polymorph of diamond and lonsdaleite, since it is chemically distinct, having sp2 hybridized bonding. Diamond and lonsdaleite are chemically identical … making them polymorphs.

   Incomplete overlap of allotropy and polymorphism; graphite vs diamond treated as chemically distinct despite being carbon allotropes.

7. ev-wiki-poly-classic-pairs src-wiki-polymorphism-ms. A classical example of polymorphism is the pair of minerals calcite, which is rhombohedral, and aragonite, which is orthorhombic. Both are forms of calcium carbonate. … TiO2: Rutile (equilibrium phase), Anatase (metastable), Brookite (metastable).

   Same-formula mineral pairs used as textbook polymorphism examples.

8. ev-wiki-allotropy-elements src-wiki-allotropy. The term allotropy is used for elements only, not for compounds. The more general term, used for any compound, is polymorphism, although its use is usually restricted to solid materials such as crystals. Allotropy refers only to different forms of an element within the same physical phase … The differences between these states of matter would not alone constitute examples of allotropy.

   Allotropy is element-only and same physical state; ice/steam vs liquid water is not allotropy by this definition.

9. ev-wiki-microstructure-scale src-wiki-microstructure. Wikipedia Microstructure: structure revealed by an optical microscope above 25×; crystal structure is reserved for the arrangement of individual atoms; finer-than-optical structure is often called nanostructure. Grain size is said to be controlled by processing conditions and composition; multiple phases may exist at once.

   Microstructure is a different length-scale object from crystal structure and from bulk composition.

10. ev-wiki-crystal-structure src-wiki-crystal-structure. Crystal structure is the ordered atomic arrangement specified by unit-cell geometry and space group.

    Crystal structure is an atomic-arrangement descriptor, not a chemical-formula descriptor.

11. ev-cambridge-cp1-split src-cambridge-cp1. When describing the structure of a material, we make a clear distinction between its crystal structure and its microstructure. The term ‘crystal structure’ is used to describe the average positions of atoms within the unit cell … on an atomic (or Å) length scale. The term ‘microstructure’ is used to describe the appearance of the material on the nm-cm length scale. Working definition: “The arrangement of phases and defects within a material.” A ‘phase’ is taken to be any part of a material with a distinct crystal structure and/or chemical composition. … a multi-component material can exist as a single phase if the different chemical components are intimately mixed at the atomic length scale. In the solid state, such mixtures are called ‘solid solutions’.

    Composition, crystal structure, phase, and microstructure occupy distinct slots; a solid solution can change composition without becoming a second phase.

12. ev-rhines-microstructures src-scielo-rhines. Crystal structure = atoms on lattices; microstructures = space-filling distributions of phases and their boundaries; a qualitative microstructural state lists phases with their compositions and crystal structures plus 3-/2-/1-/0-dimensional features.

    A microstructural description presupposes phases that already have compositions and crystal structures; it is not itself the composition key.

13. ev-cas-one-rn-except-minerals src-cas-registration-criteria. CAS will assign a single CAS RN to a substance regardless of varying physical forms, except for naturally occurring minerals. For example, water, ice, and water vapor are all represented by CAS RN 7732-18-5.

    Official CAS default: physical form/phase does not mint a new RN; minerals are an explicit exception. Composition/substance identity, not crystal form, is the default key.

14. ev-wiki-cas-granularity src-wiki-cas-rn. The definition of “distinct” excludes different phases of the same substance (e.g. water and ice), but includes narrowing of existing concepts (e.g. carbon vs. graphite and diamond …). Different phases do not receive different CAS RNs (liquid water and ice both have 7732-18-5), but different crystal structures do (carbon in general is 7440-44-0, graphite is 7782-42-5 and diamond is 7782-40-3).

    Encyclopedia generalizes carbon allotropes to a ‘different crystal structures get different RNs’ rule, which is stronger than the official ‘minerals’ exception.

15. ev-chebi-graphite src-chebi-graphite. ChEBI treats graphite as both an allotrope of carbon and a native element mineral, lists CAS 7782-42-5, and further splits hexagonal vs rhombohedral graphite as children of graphite.

    A curated chemical ontology records graphite as a distinct substance/mineral from elemental carbon, with further polytype-like children.

16. ev-wiki-carbon-three-cas src-wiki-carbon. Infobox CAS numbers listed separately: atomic carbon 7440-44-0; graphite 7782-42-5; diamond 7782-40-3. The three relatively well-known allotropes of carbon are amorphous carbon, graphite, and diamond.

    Same element, three CAS numbers; carbon is a worked exception to a single-substance reading of allotropy.

17. ev-nickel-grice-species-def src-nickel-grice-1998. A mineral species is a mineral substance with well defined chemical composition and crystallographic properties, and which merits a unique name. … If a mineral is found whose composition or crystallographic properties (or both) are substantially different from those of any existing mineral species, there is a possibility that it may be a new species.

    IMA species identity is composition AND crystallography, not formula alone.

18. ev-nickel-grice-topology src-nickel-grice-1998. Polymorphic minerals are those that have essentially the same chemical composition, but different crystal structures. The polymorphic forms of a mineral are regarded as different species if their structures are topologically different. However, if the crystal structures of the polymorphs have essentially the same topology, differing only in terms of a structural distortion or in the order–disorder relationship of some of the atoms comprising the structure, such polymorphs are not regarded as separate species.

    Same-formula solids split into species only when bonding topology differs; distortion/OD is not a new species.

19. ev-nickel-grice-polytypes src-nickel-grice-1998. Polytypes are substances that occur in several different structural modifications, each of which may be regarded as being built up by the stacking of layers of (nearly) identical structure and composition … Polytypoids are substances that do not fit the strict definition of a polytype, and include minerals with the same topology and with somewhat different compositions. Polytypes and polytypoids are not regarded as separate species and, like topologically similar polymorphs, they can be distinguished by the addition of a crystallographic suffix to the mineral name.

    Stacking variants and polytypoids are labels on one species, not new species names.

20. ev-hatert-bonding-scheme src-hatert-2023. Two substances with the same composition may also be defined as separate mineral species if their crystal structures are substantially different. This means that their bonding schemes must show significant differences. Many mineral species are known that show the same ideal formula but distinct crystal structures, such as pyrite/marcasite, calcite/aragonite, or andalusite/kyanite/sillimanite.

    CNMNC 2023 keeps the same-formula / different-bonding-scheme species split and names classic pairs.

21. ev-hatert-similar-not-species src-hatert-2023. … space group and/or unit-cell parameters variations may also be induced by some order/disorder variations in crystal structures or structural distortion; in these cases, the two polymorphs show essentially the same bonding scheme. Such substances are named topologically similar polymorphs (we suggest here using the term ‘structurally similar polymorphs’), and are not considered as distinct mineral species (Nickel and Grice, 1998). A well-known example is given by analcime.

    Metric/OD/distortion differences with the same bonding scheme do not mint IMA species.

22. ev-hatert-polytype-suffix src-hatert-2023. Such polytypes are not considered as separate mineral species as all layers constituting their structures have essentially the same basic topologies. … distinguished by the addition of a polytype suffix that is not part of the mineral species name … e.g.: molybdenite-3R and molybdenite-2H … Those polytype suffixes can also be used to distinguish structurally similar polymorphs. Polytypoids are not considered as separate mineral species. Polysomes … may be considered as separate valid mineral species.

    A crystallographic suffix is not equivalent to a new species; polysomes may still be species.

23. ev-wiki-mineral-each-structure src-wiki-mineral. Wikipedia Mineral: if a compound occurs with different crystal structures, each structure is a different mineral species; quartz and stishovite are both SiO2. Stishovite is a separate IMA-named tetragonal species with six-coordinate Si (rutile-type), not a quartz variety.

    Encyclopedia illustration of composition-insufficient species identity for SiO2.

24. ev-marcasite-species src-wiki-marcasite. Mindat Marcasite: A valid IMA mineral species - grandfathered. Formula: FeS2. The orthorhombic polymorph of isometric (cubic) pyrite.

    FeS2 is two IMA species (pyrite vs marcasite), not one formula-keyed kind.

25. ev-hom-pyrite-dimorphous src-hom-pyrite. Handbook of Mineralogy: pyrite is dimorphous with marcasite.

    Reference-handbook confirmation of a same-formula dimorph pair treated as two minerals.

26. ev-rutile-anatase src-wiki-rutile-anatase. Rutile and anatase are separate mineral articles for TiO2, with rutile called the common form and anatase a metastable mineral form; brookite and akaogiite are named as further TiO2 polymorphs.

    TiO2 is split by crystal form in mineral encyclopedia practice.

27. ev-mindat-wollastonite-polytypes src-mindat-wollastonite. Mindat lists wollastonite as one grandfathered IMA species with several polytypes (-1A, -2M, …) and separately lists breyite, davemaoite, and pseudowollastonite as polymorphs of that species.

    Worked example: stacking polytypes share one species name; some other CaSiO3 structures are listed as distinct polymorphs/species.

28. ev-wiki-belite-not-mineral src-wiki-belite. Belite is a name in common use in the cement industry, but is not a recognised mineral name. It occurs naturally as the mineral larnite… The belite found in Portland cement differs in composition from pure di-calcium silicate. It is a solid solution and contains minor amounts of other oxides besides CaO and SiO2.

    ‘Belite’ is an industrial/clinker-phase label, not an IMA species competing with larnite.

29. ev-wiki-larnite-mineral src-wiki-larnite. Larnite is a calcium silicate mineral with the formula Ca2SiO4. Infobox: IMA symbol Lrn; crystal system monoclinic; space group P21/n. When used in the cement industry, the mineral is usually referred to as belite. This page also states: ‘It is the calcium member of the olivine group of minerals.’

    Larnite is treated as the named natural β-Ca2SiO4 mineral. The olivine-group membership sentence conflicts with other opened sources.

30. ev-mindat-calcio-olivine src-mindat-olivine-group. Olivine Group Members table includes: Calcio-olivine | Ca2SiO4 | Orth. mmm (2/m 2/m 2/m). Larnite does not appear in that member table.

    γ-Ca2SiO4 is listed as olivine-group calcio-olivine, distinct from larnite in the opened group table.

31. ev-tzoumerkas-ca2sio4 src-crystals-2022-belite. The belite phase presents five polymorphs … α-Ca2SiO4 … α'H-Ca2SiO4 … α'L-Ca2SiO4 … β-Ca2SiO4 … and γ-Ca2SiO4 (γ-C2S, stable at room temperature). … γ-C2S does not exhibit hydraulic properties. β-C2S is the high-temperature monoclinic polymorph of the Calcio-Olivine γ-C2S polymorph but does not belong to the olivine group.

    Cement crystal-chemistry treats α/α'/β/γ as polymorphs of one Ca2SiO4 family; β is not olivine-group; γ is non-hydraulic. The paper does not itself issue IMA species decisions.

32. ev-cement-belite-phase-label src-cement-belite. PCA 2013 and TRB SR 127 use belite/alite as portland-cement phase labels for impure C2S/C3S.

    Industry guides key a substituted clinker phase, not a pure-formula mineral species.

33. ev-inchi-cannot-polymorphs src-inchi-trust-faq. Section 4.15 “What can the current version of InChI not represent?” lists, among others: Polymers; Complex organometallics; Markush structures; Mixtures; Conformers; … Cluster molecules; Polymorphs; Unspecific isotopic enrichment; Reactions.

    Production InChI explicitly does not encode solid-state polymorphism; crystal form is out of scope of the identifier string.

34. ev-heller-water-ice-same-id src-heller-2015-inchi. This consideration is also somewhat counterintuitive, for example, as concerns aggregate states, polymorphs, etc. Thus, most chemists would agree that “water” is a chemical substance, that may appear as steam, ice and liquid water, and that all three should have the same chemical identifier—despite the fact that each may be isolated as a different state of matter … In other words, the “identifying power” of a chemical identifier is inherently limited.

    IUPAC InChI design intent: phase and polymorphs are outside chemical-substance identity.

35. ev-iucr-2005-extra-layers src-iucr-2005-phase-id. Version 1.0 will therefore be restricted to describing the topology of finite molecules and is not being designed to describe the connectivity of infinite structures. Recommendation: add layers for state of matter (PH: gas/liq/sol/xtl/qxl/ams/lxl), space-group number 1–230 (SG:), and optional Wyckoff sequence (WS:). Example: Rutile 1.02/TiO2/PH:xtl/SG:136/WS:af2. Remaining collisions: two bcc iron phases sharing SG:229.

    A published proposal to distinguish crystalline forms of a given compound by extra InChI layers; explicitly not production Standard InChI. Formula alone is treated as insufficient for a crystal-phase identifier except when only one form is known.

36. ev-pheur-same-composition src-bp-pheur-5-9. Polymorphism (or crystal polymorphism) is a phenomenon related to the solid state; it is the ability of a compound in the solid state to exist in different crystalline forms having the same chemical composition. … The identity of chemical composition implies that all crystalline and amorphous forms of a given species have the same chemical behaviour in solution or as a melt; in contrast, their physico-chemical and physical characteristics … and therefore their reactivity and bioavailability may be different at the solid state.

    Pharmacopoeial definition: polymorphs are forms of one composition/species; solid-state properties may still be specified.

37. ev-usp-recrystallize-identity src-usp-197-941. USP 〈197〉: If a difference appears in the IR spectra of the analyte and the standard, dissolve equal portions of the test specimen and the Reference Standard … evaporate … and repeat the test on the residues. USP 〈941〉: it is not uncommon to find several polymorphs of crystalline pharmaceutical compounds … Figure caption: Typical Powder Patterns Obtained for Four Solid Phases of Ampicillin.

    Default identity testing can recrystallize away form differences; multiple solid phases belong to one named compound unless a monograph restricts form.

38. ev-ich-q6a-same-substance src-ich-q6a. Polymorphism: The occurrence of different crystalline forms of the same drug substance. This may include solvation or hydration products (also known as pseudopolymorphs) and amorphous forms. … In cases where differences exist which have been shown to affect drug product performance, bioavailability or stability, then the appropriate solid state should be specified.

    ICH treats polymorphs as forms of one drug substance; solid state is a specification attribute, not a new substance key.

39. ev-optimade-structure-entry src-optimade-v13. Entry: A single instance of a specific type of resource served by the API implementation. For example, a structures entry is comprised by data that belong to a single structure. Entry property: One data item which belongs to an entry, e.g., the chemical formula of a structure. Taken together, ID and entry type MUST uniquely identify the entry.

    OPTIMADE keys a structure instance; chemical formula is a property, not the unique key.

40. ev-mp-material-id-polymorph src-mp-identifiers. In the Materials Project, each unique material is given a material_id … This allows a specific polymorph of a given material to be referenced. For example, wurtzite GaN is assigned the material_id of mp-804, while zinc blende GaN is assigned a material_id of mp-830. Tasks are grouped by structures which are similar enough to identified as the same material (e.g., diamond-cubic Si with lattice constant 5 Å and 5.1 Å would be matched, but diamond-cubic Si and simple-cubic Si would not).

    Computational materials identity is a grouped crystal-structure/polymorph, not a formula. Near-identical lattices merge; different structure types of the same composition do not.

41. ev-cod-2009-new-number src-grazulis-2009-cod. Each structure deposited in the COD and the PCOD gets a unique seven-digit number, a (P)COD identifier. If a structure of a compound is redetermined, with higher precision or under different conditions, it will be deposited in the (P)COD under a new (P)COD number.

    COD identity is a structure-determination instance, not a compound/formula kind.

42. ev-cod-2012-instance src-grazulis-2012-cod. A COD number identifies a particular instance of a structure determination. As a rule, COD does not accept duplicate structures. If, however, two structures of the same compound are published in two different peer-reviewed journals, both can be deposited to COD and receive distinct COD numbers. Duplicate check compares formula plus unit cell (and optional T/P/history), not formula alone.

    Same compound can have many COD numbers; duplicate detection is formula+cell, not formula alone.

43. ev-cod-dict-persistent src-cod-cif-dict. _cod_database_code_structure — A persistent unique identifier assigned to each crystal structure upon successful deposition to the COD. The identifier is guaranteed not to change and is the preferred way of referencing individual COD entries.

    Preferred COD reference is a per-structure database code, not a formula.

44. ev-matml-formula-characterization src-matml. Material has optional id. Required child BulkDetails includes Name (once) … optional Characterization. Formula contains a string representation of the chemical formula for the bulk material or component and must occur once and only once within the Characterization element. … ChemicalComposition … may occur once or not at all … PhaseComposition … may occur zero or more times … DimensionalDetails contains information relating to … grain size, porosity, precipitate size and distribution, etc., and may occur zero or more times.

    MatML keys a named Material; formula, phase composition, and grain-scale descriptors are characterization children, not alternative primary keys.

45. ev-iso-683-grade-plus-condition src-iso-683-1. ISO 683-1 specifies technical delivery requirements for named non-alloy steels supplied in one of the heat-treatment conditions given in Table 1 and in one of the surface conditions given in Table 2. Chemical composition of the cast analysis shall comply with Table 3 (grade chemistry). Mechanical properties apply as appropriate for the particular heat-treatment condition. Ordering example: ISO 683-1 – C45E+N+BC.

    Engineering steel identity at order is grade plus heat-treatment/surface condition; condition is not a new chemical grade.

46. ev-iso-630-delivery src-iso-630-3. ISO 630-3:2021 landing: fine-grain structural steels in as-rolled, normalized/normalized-rolled, and thermomechanical processed delivery conditions; not all grades exist in all qualities.

    Structural-steel delivery condition is attached to a grade, not treated as a new composition kind.

47. ev-sae-air-composition-not-complete src-sae-air-4127a. Chemical composition indicates only one aspect of a steel product and is not to be considered as a complete guide to quality and performance. Chemical composition, however, is the most common basis for the classification and designation of steels.

    Composition is the common designation basis but is explicitly not complete product identity.

48. ev-en-10027-plus-symbols src-wiki-steel-grades. Steel grades are grades used to classify various steels by their composition and physical properties. EN 10027-1: two categories — (1) specified by purpose and mechanical properties (e.g. S355); (2) specified by chemical composition. Additional symbols after + include delivery conditions: A annealed, QT quenched and tempered, N normalised, U untreated. Example: S355J2+N.

    Not all grade names are composition keys; +N/+QT encode delivery/heat-treatment on the same grade designation.

49. ev-pmdco-three-classes src-pmdco. mineral — A mineral is a naturally occurring material characterized by a defined chemical composition and a specific crystal structure … chemical composition — The chemical composition is an intensive quality of a portion of matter which describes the types and proportions of pure chemical elements … microstructure — A microstructure is a portion of matter that represents the small-scale structure of a material, including grains, phases, and defects, visible under a microscope. Competency questions ask independently for identifier, composition, specifications, phases, and how microstructure changes after processing.

    A materials ontology keeps mineral (composition+structure), chemical composition, and microstructure as distinct entities/queries, not one identity class.

50. ev-wiki-heat-treating src-wiki-heat-treating. Heat treating … processes used to alter the physical, and sometimes chemical properties of a material. Metallic materials consist of a microstructure of small crystals called “grains” … Heat treatment provides an efficient way to manipulate the properties of the metal by controlling the rate of diffusion and the rate of cooling within the microstructure. Slow cool from austenite → pearlite; quench → martensite.

    Heat treatment is presented as changing microstructure of a metal, not as minting a new bulk chemical formula.

51. ev-wiki-pearlite src-wiki-pearlite. Pearlite is a two-phased, lamellar structure composed of alternating layers of ferrite and cementite. Pearlite is a microstructure occurring in the vast majority of structural steels.

    Pearlite is classified as a microstructure (two-phase mixture), not as a new chemical kind.

52. ev-bhadeshia-pearlite-martensite src-bhadeshia-steel-micro. In Fe-C systems, the average chemical composition of the pearlite is identical to that of the austenite; the latter can therefore completely transform into pearlite. Martensite: a diffusionless transformation achieved by the deformation of the parent lattice into that of the product.

    Pearlite and martensite are transformation products at the parent Fe–C chemistry, not new bulk formulae.

Sources

1. src-iupac-pac-1994 IUPAC Recommendations 1994 — definitions of terms relating to phase transitions of the solid state. doi:10.1351/pac199666030577 · https://doi.org/10.1351/pac199666030577 · class A · fetched
2. src-iupac-pac-2011 IUPAC Recommendations 2011 — terminology of polymers and crystallization (crystal polymorph). doi:10.1351/pac-rec-10-11-13 · https://doi.org/10.1351/pac-rec-10-11-13 · class A · fetched
3. src-nickel-grice-1998 Nickel and Grice 1998 — IMA CNMMN procedures and guidelines on mineral nomenclature. https://www.eps.mcgill.ca/~courses/c644/Biomineralization%20(2011)/Mineral_Nomenclature_IMA_Nickel_and_Grice_1998.pdf · class A · fetched
4. src-hatert-2023 Hatert, Mills, Pasero, Miyawaki and Bosi 2023 — CNMNC guidelines for the nomenclature of polymorphs and polysomes. doi:10.1180/mgm.2023.13 · https://doi.org/10.1180/mgm.2023.13 · class A · fetched
5. src-cas-registration-criteria CAS Registration Criteria-Overview. https://web.cas.org/marketing/pdf/CAS-Registration-Criteria-Overview.pdf · class B · fetched
6. src-cas-registry-page CAS REGISTRY product/FAQ page. https://www.cas.org/cas-data/cas-registry · class unknown · fetched
7. src-wiki-cas-rn CAS Registry Number (Wikipedia). https://en.wikipedia.org/wiki/CAS_Registry_Number · class unknown · fetched
8. src-inchi-trust-faq InChI Trust Technical FAQ. https://www.inchi-trust.org/technical-faq/ · class B · fetched
9. src-heller-2015-inchi Heller, McNaught, Pletnev, Stein and Tchekhovskoi 2015 — InChI, the IUPAC International Chemical Identifier. doi:10.1186/s13321-015-0068-4 · https://doi.org/10.1186/s13321-015-0068-4 · class A · fetched
10. src-iucr-2005-phase-id Brown et al. 2005 — IUCr Working Group report on crystal phase identifiers. doi:10.1107/S010876730503179X · https://doi.org/10.1107/S010876730503179X · class A · fetched
11. src-bp-pheur-5-9 Ph. Eur. 5.9 / BP Appendix I F — Polymorphism. https://www.drugfuture.com/Pharmacopoeia/BP2013/data/1005.html · class unknown · fetched
12. src-usp-197-941 USP general chapters 〈197〉 and 〈941〉. https://www.drugfuture.com/Pharmacopoeia/usp38/data/v38332/usp38nf33s2_c197.html · class unknown · fetched
13. src-ich-q6a ICH Q6A Specifications: Test Procedures and Acceptance Criteria for New Drug Substances and New Drug Products. https://database.ich.org/sites/default/files/Q6A_Guideline.pdf · class B · fetched
14. src-optimade-v13 OPTIMADE API specification v1.3.0. https://www.optimade.org/specification/latest/ · class B · fetched
15. src-mp-identifiers Materials Project FAQ and identifier-systems documentation. https://docs.materialsproject.org/frequently-asked-questions · class unknown · fetched
16. src-grazulis-2009-cod Gražulis et al. 2009 — Crystallography Open Database. doi:10.1107/S0021889809016690 · https://doi.org/10.1107/S0021889809016690 · class A · fetched
17. src-grazulis-2012-cod Gražulis et al. 2012 — COD as open-access collection and collaboration platform. doi:10.1093/nar/gkr900 · https://academic.oup.com/nar/article/40/D1/D420/2903497 · class A · fetched
18. src-cod-cif-dict COD CIF dictionary — _cod_database_code_structure. https://wiki.crystallography.net/cif/dictionaries/cif_cod/ · class unknown · fetched
19. src-matml OASIS MatML PR-01 and NIST IR 6939 MatML 3.0 schema. doi:10.6028/nist.ir.6939 · https://docs.oasis-open.org/materials/materials-matml-spec-pr-01.htm · class B · fetched
20. src-iso-683-1 ISO 683-1:2016 Heat-treatable steels — Part 1 (sample text). https://www.iso.org/standard/67747.html · class unknown · fetched
21. src-iso-630-3 ISO 630-3:2021 Structural steels — Part 3 landing page. https://www.iso.org/standard/73843.html · class unknown · fetched
22. src-sae-air-4127a SAE AIR 4127A — steel designation catalog sample. https://img.antpedia.com/standard/files/pdfs_ora/20230614/SAE/AIR/SAE%20AIR%204127A-2015.pdf · class unknown · fetched
23. src-wiki-steel-grades Steel grades (Wikipedia). https://en.wikipedia.org/wiki/Steel_grades · class unknown · fetched
24. src-pmdco PMDco (Platform MaterialDigital Core Ontology) v3 documentation and pmdco-base.owl. https://materialdigital.github.io/core-ontology/docs/intro.html · class unknown · fetched
25. src-cambridge-cp1 Cambridge materials practical CP1 — Introduction to microstructure. https://www.inference.org.uk/prlw1/minp/CourseC/CP1.pdf · class unknown · fetched
26. src-wiki-heat-treating Heat treating (Wikipedia). https://en.wikipedia.org/wiki/Heat_treating · class unknown · fetched
27. src-wiki-pearlite Pearlite (Wikipedia). https://en.wikipedia.org/wiki/Pearlite · class unknown · fetched
28. src-bhadeshia-steel-micro Bhadeshia — Interpretation of the Microstructure of Steels. http://www.phase-trans.msm.cam.ac.uk/2008/Steel_Microstructure/SM.html · class unknown · fetched
29. src-wiki-polymorphism-ms Polymorphism (materials science) (Wikipedia). https://en.wikipedia.org/wiki/Polymorphism_(materials_science) · class unknown · fetched
30. src-wiki-allotropy Allotropy (Wikipedia). https://en.wikipedia.org/wiki/Allotropy · class unknown · fetched
31. src-wiki-microstructure Microstructure (Wikipedia). https://en.wikipedia.org/wiki/Microstructure · class unknown · fetched
32. src-wiki-crystal-structure Crystal structure (Wikipedia). https://en.wikipedia.org/wiki/Crystal_structure · class unknown · fetched
33. src-scielo-rhines Rhines — Microstructures (SciELO reprint extract). https://www.scielo.br/ · class unknown · fetched
34. src-wiki-mineral Mineral (Wikipedia) and related silica pages. https://en.wikipedia.org/wiki/Mineral · class unknown · fetched
35. src-wiki-marcasite Marcasite (Mindat / Wikipedia cluster). https://www.mindat.org/min-2571.html · class unknown · fetched
36. src-hom-pyrite Handbook of Mineralogy — Pyrite. https://www.handbookofmineralogy.org/pdfs/pyrite.pdf · class unknown · fetched
37. src-wiki-carbon Carbon (Wikipedia). https://en.wikipedia.org/wiki/Carbon · class unknown · fetched
38. src-wiki-rutile-anatase Rutile / anatase encyclopedia pages. https://en.wikipedia.org/wiki/Rutile · class unknown · fetched
39. src-chebi-graphite ChEBI CHEBI:33418 graphite. https://www.ebi.ac.uk/chebi/CHEBI:33418 · class unknown · fetched
40. src-mindat-wollastonite Mindat Wollastonite. https://www.mindat.org/min-4323.html · class unknown · fetched
41. src-mindat-olivine-group Mindat Olivine Group. https://www.mindat.org/min-29264.html · class unknown · fetched
42. src-wiki-larnite Larnite (Wikipedia). https://en.wikipedia.org/wiki/Larnite · class unknown · fetched
43. src-wiki-belite Belite (Wikipedia). https://en.wikipedia.org/wiki/Belite · class unknown · fetched
44. src-cement-belite PCA/MIT SN3228 and TRB SR 127 cement compound guides. http://www2.cement.org/exec2/pdfs/sn3228.pdf · class unknown · fetched
45. src-crystals-2022-belite Tzoumerkas et al. 2022 — Structure–superstructure inter-relations in Ca2SiO4 belite phase. doi:10.3390/cryst12121692 · https://doi.org/10.3390/cryst12121692 · class A · fetched

Audit

Verdict. STORM_FOUNDATION_READY

STORM audit — runs/v03-polymorph-microstructure

Independent Discovery boundary audit (atomic_action: audit_discovery). This file inspects research behavior, not merely file presence. Scientist was not run (stages.scientist* remain pending). brief.md is caller context, not evidence.

python3 scripts/check_run.py runs/v03-polymorph-microstructure → OK (referential integrity only: stage files, source/evidence id links, report [src-*] ids).

Bounded conversation sample: turn 01 only for p00-basic-facts, p01-mineral-species-nomenclature, p02-substance-registry-identifiers, p03-record-and-exchange-models (lexicographic). Follow-up structure is taken from each conversation.json index, not from later turn bodies.

Faithfulness to STORM

1. Topic-driven, not planner→judge. PASS. run.json topic and discovery/handoff.json research_question are the same string: how published materials-identity systems treat solid-state polymorphs and microstructure relative to chemical composition when deciding whether two records are the same material kind. Stages are survey → perspectives → conversations → aggregate → isolated outline_draft → outline_refine → synthesis → polish → handoff. No voting swarm, no critic-everywhere loop, no Scientist experiments folded into Discovery.

1. Survey of related information before perspectives. PASS. discovery/survey/related_topics.md lists 12 searches, opened pages (not snippet-only), fetch failures, and nearby named topics. discovery/survey/survey_notes.md is a coverage map with disputes left uncollapsed. run.json stages.survey.ended_at 2026-09-18T10:48:18+00:00 precedes stages.perspectives.started_at 2026-09-18T10:48:25+00:00.

1. Perspectives discovered from that survey, plus basic-facts. PASS. discovery/perspectives.json: discovered_from = discovery/survey/related_topics.md; basic_facts_included true; emergency_fallback_used false. p00-basic-facts is the STORM default (why_created: not derived from survey findings). p01–p03 match survey clusters (mineral-species nomenclature; substance/registry identifiers; structure records / grades / exchange schemas). Survey notes listed a fourth crystallographic-record lens and an engineering-exchange lens; those were combined into p03 rather than invented from a hard-coded persona menu.

1. Independent Writer/Asker vs Expert/Answerer. PASS. Each perspective directory separates turns/01-question.md (asker, with confirmation / contradiction / exception / reframing) from turns/01-queries.json and turns/01-answer.md (answers cite fetched turns/01-sources/ files). discovery/outline_draft.isolation.md records that the outline writer received only the topic string and format instructions. No merged planner/judge transcript.

1. One question per turn; follow-ups use prior answers. PASS. Four perspectives × three turns. Turn-01 questions are single-topic (definitions; IMA species rule; CAS RN rule; structure-record key), each with a clarifying clause rather than a second topic. Follow-ups in conversation.json use prior answers: p01 T1 reserves polytypes and T2 asks them; p02 T2 moves from CAS to InChI after the RN rule; p03 T2 asks engineering-exchange keys after structure-database keys; p00 T2–T3 move from definitions to identifier disagreement to named pairs.

1. Retrieval before factual answering; expert refuses when information is missing. PASS. Turn-01 answers cite local fetched extracts, not unaided claims. p00 T1 records Gold Book HTML and Britannica allotropy as not fetched and uses PAC 1994/2011 instead; it also refuses a single materials-database kind rule from definitional sources. Survey and later conversation indexes record further refusals (Bosi 2022 PDF timeout; calcio-olivine species-page 404/timeouts; paywalled full ISO/EN bodies). Those gaps are not treated as evidence bodies.

1. Shared information base after conversations. PASS. stages.conversations.ended_at 2026-09-18T11:08:21+00:00; stages.aggregate 11:08:21–11:14:16. discovery/information_base/sources.json has 45 merged records (all fetched: true); evidence.json has 52 excerpts with source_id; aggregation.md has agreed / conflicts / missing and labels brief.md as caller assumption, not evidence.

1. Draft outline generated in isolation. PASS. discovery/outline_draft.isolation.md: received only the topic string and format instructions; no survey, brief, conversations, or other research. discovery/outline_draft.md headings are parametric (composition as default kind key; allotropes; polymorphs as distinct vs as variants; microstructure as processing; identifier schemes). They do not name CAS, InChI, OPTIMADE, IMA/CNMNC, MatML, belite, larnite, or Ca2SiO4. Compare with research-aware outline_refined.md. Residual: outline_draft ran after aggregate completed (11:14:29 vs 11:14:16); isolation is evidenced by the note plus the draft’s lack of survey leakage, not by overlapping clocks.

1. Refined outline uses research. PASS. discovery/outline_refined.md replaces the isolated spine with published identity units (substance / species / structure instance / grade), IMA topology vs polytypes, Ca2SiO4 names, CAS/InChI/pharmacopoeia, OPTIMADE/MP/COD, MatML and steel delivery condition, PMDco, and open issues. Isolated-draft headings without fetched bodies (nanostructure identity; a single metals/ceramics/polymers model) are dropped or parked under Open issues, matching outline_refine notes.

1. Sections written from retrieved information with citations. PASS. discovery/sections/.md and discovery/report.md attach [src-] to supported sentences. check_run.py resolves those ids to sources.json. Open issues record unfetched Gold Book HTML, calcio-olivine species pages, official CAS tables for non-carbon pairs, IUCr 2005 layers not shown as implemented, and mill-certificate kind language — not filled from unaided memory.

1. Global polish with a lead/summary. PASS. discovery/report.md opens with # Summary (units of identity, three definitional splits, identity disagreements, caller brief as context), then the refined section bodies. discovery/citations.md lists cited src-* ids with class, URL/DOI, and fetched status. src-cas-registry-page is opened but not cited in the report, and is labeled as such.

Apex evidence rules

1. Project brief not cited as external evidence. PASS. No src-brief. citations.md has no brief row. Report, aggregation, survey notes, and related_topics name brief.md only as motivation/context and say it is not a source. Findings are allowed to contest the caller (F2, F6, F10; F9 left unresolved).

1. Foundational claims not based only on snippets when fetch was possible. PASS. Foundational identity rules are tied to fetched PAC 1994/2011, Nickel–Grice 1998, Hatert 2023, CAS criteria PDF, InChI Trust FAQ, Heller 2015, OPTIMADE v1.3, MP identifier docs, COD papers, MatML/NIST IR 6939, ISO 683-1 sample, and named encyclopedia/ontology pages (sources.json fetched: true with fetch_path). Gold Book HTML, PubChem calcio-olivine, and Bosi 2022 remain flagged as unfetched; encyclopedia/search hits are not used as those bodies. Pharmacopoeia chapters are inspected via mirrors and labeled as such.

1. Duplicate URLs/DOIs merged when practical. PASS. 45 canonical URLs, 0 duplicate URL groups, 9 unique DOIs, 0 duplicate DOI groups. Identity notes record merges (Hatert DOI + booklet + Cambridge HTML; two CAS criteria PDF URLs; p00/p01 Nickel–Grice extracts; p00/p02 CAS wiki).

1. Questions are non-leading. PASS. Each sampled turns/01-question.md lists confirmation / contradiction / exception / reframing. p00 T1 allows microstructure to be equated with formula (contradiction) or a different object (reframing). p01 T1 allows composition-only species or structure that never splits. p02 T1 allows every crystal form to get its own RN or no solid-state distinction. p03 T1 allows formula as unique identity. No “what evidence proves the v0.2 encoding is wrong?” form.

1. Handoff polarity and fail-able hypotheses. PASS. discovery/handoff.json: F1/F3/F4/F5/F7/F8 supported; F2/F6/F10 contested; F9 unresolved. All evidence ids resolve into evidence.json. Supporting ids argue for the literal claim; opposing ids argue against it (F2 opposing IMA topology / OPTIMADE / MP / COD / IUCr extra layers / wiki CAS granularity; F6 opposing official CAS one-RN-except-minerals; F10 opposing distinct larnite / calcio-olivine / belite-as-not-mineral treatments). F9 has empty supporting and opposing, context only. Each finding has a hypothesis with could_fail_if and a predicted outcome. Residual (does not fail the item): F10 is contested with empty supporting_evidence_ids; cement “one industrial family” lives in alternative_explanations / context rather than supporting the universal single-kind claim.

Independence

1. No hard-coded downstream domain pack. PASS. The topic is published materials identity (polymorph vs microstructure vs composition). Perspectives are mineral nomenclature, chemical-substance registries, and record/exchange models — from the survey TOCs, not a construction/BIM/medicine menu. Cement and steel appear because opened sources raise Ca2SiO4 naming and heat-treatment microstructure; the report does not import Uniclass, OmniClass, IFC, bSDD, or a construction-engineer persona.

1. At least one finding that is not a restatement of brief.md. PASS. The brief assumes v0.2 parking of solid-state phase in microstructure and a shared composition kind for β- and γ-Ca2SiO4. F1 is IUPAC/pharmacopoeial polymorph definitions. F3/F4 are IMA topology vs polytype rules. F5 is Standard InChI’s lack of a polymorph layer. F7 is OPTIMADE/MP/COD structure keys. F8 is pearlite/martensite and grade+condition. F10 contests treating β/γ-Ca2SiO4 as one kind. F9 (no published necessary-and-sufficient cross-domain model) is left unresolved, matching the brief’s out-of-scope item rather than filling it.

Causal stage order

survey (ended 10:48:18) → perspectives (ended 10:48:58) → conversations (ended 11:08:21) → aggregate (ended 11:14:16) → isolated outline_draft (ended 11:17:38) → outline_refine (ended 11:18:04) → synthesis (ended 11:21:03) → polish / handoff (ended 11:23:44). stages.scientist* pending.

Residual limits (do not hide)

• IUPAC Gold Book HTML, Bosi 2022, Britannica allotropy, dedicated calcio-olivine IMA/Mindat species pages, official CAS tables for non-carbon same-formula pairs, and full purchased ISO 683-1 / EN 10027 / ISO 630-3 bodies were not opened.
• Ph. Eur. 5.9 and USP 〈197〉/〈941〉 were inspected via unofficial HTML mirrors.
• Those limits constrain which later experiments are cheap to run locally; they do not make the STORM behavior non-credible.

Verdict

The discovery pipeline behaved as STORM+Apex require: survey then discovered perspectives, retrieve-and-fetch conversations with refusals on missing pages, an isolated then refined outline, cited synthesis, and a polarity-aware handoff that can contradict the caller. Materials-identity content is the topic, not an injected industry persona pack.

STORM_FOUNDATION_READY

Scientist audit

Scientist audit — runs/v03-polymorph-microstructure

Run: v03-polymorph-microstructure Auditor: independent audit_scientist (did not author this Scientist tree) Question: On fixtures reconstructed from fetched identity rules (IMA topology vs polytype; composition-only token; OPTIMADE/MP/COD keys), does a composition-only kind encoder match published polymorph splits, or do IMA topology/polytype and structure-record keys split pairs that composition collides? Mechanical (referential): python3 scripts/check_scientist.py runs/v03-polymorph-microstructure → OK. python3 scripts/check_run.py runs/v03-polymorph-microstructure → OK. The checker banner Reasoning audit remains required is not a missing-file flag and is not a scientific-quality score.

This file is a reasoning audit. Mechanical pass means hashes, parents, and verdict enums exist. It does not by itself mean hypotheses could fail or that conclusions follow results.

Default read set used: discovery/handoff.json; scientist/{hypotheses,journal,conclusions}.{json,md}; scientist/discovery_snapshot.json; each node node.json / criteria.json / evaluation.md; results.json fields cited by the evaluations. stdout/stderr not opened (no experiment_failed; results present). Discovery survey/conversations not opened. Architecture/source-method docs not loaded.

Independence: materials-identity (polymorph / microstructure vs composition) is the topic of this run. That is allowed. No industry engine plugin was required to pass.

---

Faithfulness to AI-Scientist-v2 (adapted)

Apex rules

Independence

---

Evaluation vs cited results.json fields

Evaluations applied the pre-registered success conditions. Cited fields match:

E001 (assay.ok true). pairs.pyrite_marcasite.composition_same true (hill:FeS2); ima_same false (topo:pyrite vs topo:marcasite); structure undecidable (null keys). pairs.gan_wz_zb.composition_same true (hill:GaN); structure_same false (struct:mp-804 vs struct:mp-830); ima undecidable. pairs.molybdenite_2h_3r.ima_same true (hill:MoS2\|topo:molybdenite). pairs.cod_redetermination.structure_same false. summary.optimade_structure_id_equals_formula_reduced false. Assay: pyrite_duplicates.composition_same true; fes2_vs_gan.composition_same false.

E002 (assay.ok true). summary.belite_name_class industrial_phase_label; larnite_name_class ima_mineral; calcio_olivine_name_class olivine_group_member; belite_larnite_name_class_same false; larnite_in_olivine_group false; calcio_olivine_in_olivine_group true; beta_olivine_group false. Ca2SiO4 composition pairs all true (hill:Ca2O4Si). pairs.larnite_calcio_olivine.ima_undecidable true (both ima keys null). Assay: belite_duplicates.composition_same true; ca2sio4_vs_fes2.composition_same false.

E003 (assay.ok true). pairs.c45e_n_qt.grade_same true (grade:C45E); condition_same false (cond:N vs cond:QT); composition_same true. pairs.pearlite_parent.composition_same true and pairs.pearlite_martensite.composition_same true (mf:C=0.008,Fe=0.992); microstructure_same false. Assay: c45e_n_duplicates.grade_same true; c45e_vs_c40e.grade_same false.

---

Gaps (do not fail this gate; they bound how far the POC generalizes)

• Same agent as the Scientist run. This audit is a sibling pass over files, not an independent lab replication.
• All nine tested hypotheses supported. Failure conditions existed and were not hit. Unlike a run that rejects composition-only on a scored pair, this tree never lands rejected. That is consistent with the observed fields; it is not a demonstration of the reject path.
• No crash node. Item 4 is vacuously passed; debug-from-experiment_failed was not shown.
• Reconstructed fixtures, not live APIs. Topology strings, MP/COD/OPTIMADE ids, name classes, and mass fractions are in-node reconstructions of opened rules. Criteria and evaluations say so. They do not parse live IMA/MP/COD HTTP or mill certificates.
• Held-fixed chemistry. E003 C45E+N/+QT and pearlite/martensite mass fractions are identical by fixture construction. That is a valid falsifier of “condition/microstructure leaks into composition”; it is not a measured mill recipe.
• Single seed (0). Encodings are deterministic; no second seed.
• underdetermined unused. Criteria name the third Apex verdict; no node used it.
• H9 left untestable. Correct relative to F9; not a hidden positive universal model.

---

Verdict

The Scientist tree is a credible executable loop on this handoff: competing encoders were implemented and run; evaluations cite result fields that exist; follow-ups quote those prior results; literature and experiment stay in separate fields; discovery hashes are unchanged. Remaining limits are fixture reconstruction, the unused reject/crash paths, and untestable F9 — not a fake critic swarm or Sakana KEEP-as-verdict.

SCIENTIST_FOUNDATION_READY