Toolchest · Datasheet

Accumath Chemicals Dataset — Notes

Dataset: accumath-chemicals-dataset.csv — 151 chemicals, 20 columns per row. Built for the Accumath Toolchest (planned feature). Every value was read from a page opened during the build; anything that could not be verified from an opened page was left empty — never estimated, never recalled from memory.

Sources

• Primary: NIST Chemistry WebBook (webbook.nist.gov) — molar masses, phase-change temperatures, enthalpies of formation, standard entropies, heat capacities, vapor pressures. 139 of 151 rows cite NIST.
• Secondary: PubChem (pubchem.ncbi.nlm.nih.gov) — density, flash point, autoignition, solubility, GHS hazard summaries, pKa. 52 rows cite PubChem (every one of them also cites NIST — there are no PubChem-only rows).
• Fallback fact sources (named per row in the CSV sources column): used only where the two primary routes were unreachable in this build environment (see "Access notes" below) or where neither primary source carries the entry at all (mixtures). These are CAMEO Chemicals / NIOSH (NOAA / CDC — US government), ICSC/ILO cards, an ACS Organic Division solvent table, and individual manufacturer SDS facts. All are individual published facts, attributed per row.
• 12 rows cite neither NIST nor PubChem: the 9 mixture rows (gasoline, diesel, kerosene, Jet A, crude oil, LNG, LPG, R-410A, sodium silicate), iron(III) oxide, and rows whose NIST page is spectrum-only. One row — ammonium hydroxide — has an empty sources field because no usable page for CAS 1336-21-6 was obtained; its property cells are empty accordingly.

Legal basis

• NIST Chemistry WebBook is a work of the National Institute of Standards and Technology, a US federal agency. US government works are public domain (not subject to copyright under 17 U.S.C. § 105). NIST asks for attribution as a courtesy; the sources column provides it on every row.
• PubChem is a work of the National Center for Biotechnology Information / National Institutes of Health, likewise a US federal work in the public domain. Its GHS hazard data is aggregated from regulatory and supplier submissions and is publicly published data.
• Per Jeff's ruling (2026-10-20): publicly published data is public; facts do not require licensure. No commercial compilation source was used as a data source, and no expressive text was copied — values are facts, recorded with attribution.

Conventions

• Standard state column = physical state of the pure compound at 25 °C, 1 atm. LNG/LPG are gas (stored liquefied under pressure; noted in-row). Formaldehyde and acetaldehyde are gases at 25 °C (formaldehyde is normally handled as formalin solution — noted in-row). tert-Butanol is solid (mp 25.7 °C).
• Temperature basis travels inside the value: densities state their temperature (typically 20 or 25 °C, as published); vapor pressures state their temperature; melting/boiling points are at 1 atm unless the row's notes say otherwise.
• Phase basis travels inside the value for phase-specific quantities: specific heat, ΔfH°, and S° carry a (gas) / (liquid) / (solid) tag. For the elements H₂, N₂, O₂, etc., ΔfH° is 0 by definition of the reference state.
• Specific heat is J/g·K (per gram). Where a source gave only a molar Cp, the value was divided by the molar mass — arithmetic on read values, flagged in the row's notes. Where NIST offers only a Shomate equation, the value was evaluated at 298.15 K where stated, or left blank (chloroform, dichloromethane, tetrachloroethylene — blank).
• Unit conversions (K→°C, kcal→kJ, cal→J, °F→°C) are arithmetic on values read from the source page and are flagged in row notes where they occur.
• Mixtures (gasoline, diesel, kerosene, Jet A, crude oil, LNG, LPG, R-410A, dry air, sodium silicate solution): category mixture; molar mass and all thermodynamic fields (ΔfH°, S°, Cp) are empty by design; only genuinely published representative values (density, flash point, boiling range) appear, labeled "representative". R-410A has no single CAS number — the field is empty with a note. Dry ice is the CO₂ species in solid form; its sublimation temperature is recorded in its row.
• Hydrates: salts are on the anhydrous basis of the listed CAS unless a row's notes say otherwise (borax is the decahydrate by name; pentahydrate/solution values for NaOCl are labeled as such). Hydrate and anhydrous values were never silently mixed.
• Decomposition: where a compound decomposes/sublimes instead of melting or boiling, the field is empty and the note says so (e.g., KMnO₄, CuSO₄, CO₂, NH₄Cl).
• Conflicts: where sources disagreed (e.g., propylene oxide and hydrazine autoignition, hydrogen mp/bp transposition in one PubChem table), the value was either resolved to the concordant figure with the conflict disclosed in notes, or left empty. Nothing was averaged or guessed.
• Merge-time category fills: four group-3 rows (phenol, aniline, nitrobenzene, carbon disulfide) were delivered with blank category + explanation because the batch taxonomy had no fit; categories were assigned at merge ("aromatic compound", "aromatic amine", "aromatic compound", "industrial solvent"). Categories are editorial labels, not measured values.

Coverage (of 151 rows)

Field	Filled
Molar mass	141
Density	124
Melting point	121
Solubility in water	117
ΔfH°	108
Hazard summary	107
Boiling point	106
Entropy S°	86
Vapor pressure	77
Flash point	73
Specific heat	66
Autoignition	57
pKa / pH note	41

Biggest gaps

1. Group 4 (acids/bases/salts) PubChem fields are systematically blank — density, solubility, pKa, and hazard summary for all 25 rows. PubChem's compound pages were blocked by a reCAPTCHA interstitial in this environment (see Access notes), and the group's researcher declined to backfill from search snippets. A follow-up pass with working PubChem access fills most of these — pKa especially (0 of the group's 11 acids have pKa values in this pass).
2. Specific heat (66/151) — NIST tabulates Cp for gases and common liquids but not for most solids/salts; PubChem rarely carries it. This is a source-availability gap, not an access gap.
3. Autoignition (57/151) — published values are sparse and sometimes wildly discordant (hydrazine: 24–270 °C across sources); discordant values were left empty rather than picked.
4. Ammonium hydroxide row is nearly empty — no NIST species page exists for CAS 1336-21-6 (it is aqueous ammonia, not an isolable compound).
5. Benzene's PubChem fields are blank (density, flash, solubility, GHS) — PubChem was unreachable by every route during its batch; the row is NIST-only. 1,3-Butadiene's hazard summary is likewise blank (a 429 rate limit, not retried).
6. Mixtures carry no thermodynamic data by design — representative physical values only. Crude oil is the sparsest: its CAS-specific SDS reported "no data available" for nearly everything; only a labeled representative density is in.
7. Xylene (mixed isomers) has no NIST mixture page: its molar mass, mp, bp, ΔfH°, and Cp are the NIST p-xylene values, flagged as single-isomer in the row's notes; its PubChem mixture record carries no experimental properties, so density/flash/solubility are blank.

Access notes (why some rows are NIST-only)

• PubChem's HTML compound pages return a "Checking your browser" reCAPTCHA page to plain text fetching. Two working routes were found during the build: (a) PubChem PUG View JSON records (used successfully for the gases and aromatics batches — every CID verified against the record title before use), and (b) search-indexed PubChem excerpts (used sparingly, and only where the excerpt displayed the value). Batches that found neither route left PubChem fields empty and said so per row.
• NIST WebBook rate-limited (HTTP 429) three page opens (KMnO₄, AgNO₃, SiO₂); those molar masses were read from WebBook search-result pages instead, noted in-row.
• Several NIST pages are spectrum-only (IR/THz) with no thermochemistry: NaHCO₃, CaCO₃, CaSO₄, NaNO₃, KNO₃, NaNO₂, NaClO₃, KClO₃, (NH₄)₂SO₄, Fe₂O₃. For those, ΔfH°/S° are blank or came from a named public fact source, per row.
• Group 6's "NIST WebBook" links were reached through the WebBook's own index; R-22's NIST phase table contains an anomalous boiling entry (313.3 K) inconsistent with its own vaporization data — the concordant SDS value was used and the anomaly flagged in the row's notes.

Fragment files and per-batch working notes are preserved in ~/workspace/your_files/.chemfrag/ (group1.json … group6.json, group3-notes.md) for provenance.