NIST WebBook Thermodynamic Properties: Evaluated Constants, Delivered as Rows

The phrase maps to one named product: the NIST Chemistry WebBook, Standard Reference Database SRD 69, compiled under NIST's Standard Reference Data Program with named evaluators credited block by block - Afeefy, Liebman, Stein and Burgess on thermochemistry, Lias on ion energetics, Jacox on spectra, Huber, Bell and McLinden on fluid properties. It is where engineers go when a number has to be defensible rather than merely findable.

Its thermodynamic holdings split into two families. Evaluated thermochemistry covers more than 7,000 organic and small inorganic compounds - formation and combustion enthalpies, heat capacities, phase-change temperatures, critical constants - alongside reaction thermochemistry for over 8,000 reactions and ion energetics for over 16,000 compounds. Thermophysical property surfaces cover 74 fluids: density, Cp/Cv, enthalpy, entropy, viscosity and thermal conductivity as functions of temperature and pressure.

Spectral and chromatographic coverage rides along on the same species records: over 16,000 IR spectra, more than 33,000 mass spectra, 1,600-plus UV/Vis spectra and gas-chromatography retention data for over 27,000 compounds. Within Datadory's catalog it scores 9 out of 10 - one of just 534 records at that mark among the 1,744 datasets we track - and ranks fourth among the best commodity chemicals datasets. The full field-by-field breakdown lives on the NIST Chemistry WebBook dataset page.

Five rows for one workhorse solvent - methanol, CAS Registry Number 67-56-1, formula CH4O, molecular weight 32.0419 g/mol - exactly as they arrive:

species   : Methyl Alcohol (methanol)   formula : CH4O   MW : 32.0419
quantity  : ΔfH°gas      value : -205. ± 10.      units : kJ/mol
method    : AVG          comment : Average of 9 values

species   : Methyl Alcohol (methanol)
quantity  : ΔcH°gas      value : -763.68 ± 0.20   units : kJ/mol
method    : Cm           reference : Rossini, 1932
comment   : Flame Calorimetry; corresponding ΔfH°gas = -201.49 kJ/mol

species   : Methyl Alcohol (methanol)
quantity  : Tboil        value : 337.8 ± 0.3      units : K
method    : AVG          comment : Average of 154 out of 171 values

species   : Methyl Alcohol (methanol)
quantity  : Tc           value : 513. ± 1.        units : K
method    : AVG          comment : Average of 27 out of 31 values

species   : Methyl Alcohol (methanol)
quantity  : Pc           value : 81. ± 1.         units : bar
method    : AVG          comment : Average of 17 out of 20 values

Read the block as one argument about provenance. The method column separates evaluated averages (AVG - NIST weighed the field and chose) from named single-source experiments (Cm - flame calorimetry, and here is the paper). The comment column quantifies disagreement: when a boiling point keeps 154 of 171 published determinations, you know exactly how contested that constant was before you build on it. And the reference column carries the lineage down to Rossini's 1932 combustion measurement, which survives because nothing better has beaten it in ninety-four years.

Uncertainty travels with every value rather than living in an appendix, so error propagation becomes a query-time calculation. A boiling point that arrives as a bare number makes sensitivity analysis a matter of opinion; this one arrives with ±0.3 K attached.

Ten core fields ride on every measurement record: the quantity name, the evaluated value with its experimental uncertainty, the units (SI or calorie-based), the method code, the literature reference, the provenance comment counting how many determinations entered each average - plus species identity carried four ways as formula, molecular weight, CAS Registry Number and IUPAC Standard InChI/InChIKey (methanol resolves to OKKJLVBELUTLKV-UHFFFAOYSA-N).

That identity pair is why the CAS registry number glossary entry treats CAS lookups as the join spine of this industry. Species records also carry synonyms, isotopologue listings, 2-D Mol files, computed 3-D SD files and per-block compiler credits; spectral records carry their own axes and metadata conventions. Those fold under additional fields on request, confirmed and populated when you scope a sample.

Every definition above is verified against arriving rows rather than inferred from column naming. Get a sample of this dataset scoped to your substance list and the dictionary travels with it.

Process design. Evaluated constants feed heat and material balances directly: methanol's ΔfH°gas of -205 ± 10 kJ/mol anchors reaction-enthalpy estimates, while critical properties (Tc 513 K, Pc 81 bar) parameterize equations of state for flash and VLE work - the classic load behind process-modeling and safety-analysis workflows.

Emergency-response playbooks. Constants describe behavior; response decisions need operational thresholds beside them. The pairing of choice puts NIST's evaluated surfaces next to NOAA CAMEO Chemicals: methanol's datasheet there carries flash point 52°F (NTP, 1992), flammability limits of 6 to 36.5 percent, autoignition 867°F (USCG, 1999) and IDLH 6,000 ppm (NIOSH, 2024), plus a mixing reactivity predictor that screens collected substance sets for dangerous combinations before a pilot batch is charged. Datadory delivers both sides joined on CAS number, so the screening step is a query rather than a reconciliation project.

Property models and feature stores. For modeling teams the WebBook functions as ground truth: per-species rows with method and reference columns make clean labels for property-prediction models, in contrast to secondhand aggregations of unknown vintage, where a copied copy is indistinguishable from a measurement. The data scientists use cases page maps where this fits in the industry's modeling shelf.

Supply context. Constants describe molecules, not markets. When the question shifts from what a compound does to who makes how much of it, the inventories take over - CDR's roughly 8,650 substances per cycle with site-level volumes, TSCA's 86,741-substance universe - and flow questions belong to the EIA ledger. That boundary is drawn explicitly in the NIST Chemistry WebBook vs EIA petrochemical feedstock comparison.

Three habits separate a defensible constant from a plausible-looking one, and all three are structural in this record rather than bolted on.

Counting beats quoting. Because each quantity stores one row per determination behind it, you can count how much independent evidence supports a value before using it. Methanol's critical temperature averages 27 of 31 published measurements - a sentence a handbook cannot make about itself.

Method codes beat silence. An AVG flag says an evaluator weighed the field; Cm says one experiment type produced the figure and names it. Aggregated tables that flatten both into a single column destroy exactly the information that decides whether a value survives peer review or a HAZOP.

Reference ages differently than data. The current edition still cites a 1932 combustion measurement, not because the compilation is stale but because the measurement was never beaten. Revisions here track better experiments, not calendar years - which is why delivery tracks the evaluation, not a publication schedule.

None of this requires heroics to preserve once the rows arrive typed; it only requires that nobody upstream flattened the columns first. That flattening is the part Datadory handles.

The reference lane sits inside a five-family industry, and three neighbors bracket it:

Identity is the hinge that makes the four compose. Every one of them keys on CAS Registry Numbers, so a four-step lookup - physics, hazard, regulation, market - runs as joins on one identifier instead of four separate research projects. The commodity chemicals data hub lays out all 16 primary records and 11 cross-listed neighbors by family.

Process modeling and simulation teams populate property backends for flowsheet simulators - heat capacities, enthalpies of formation and critical constants keyed on CAS number, uncertainties attached so sensitivity analysis runs on real error bars instead of guessed ones.

Process safety engineers pull phase-change and flammability-relevant thermodynamics for relief sizing and consequence models, then check the substance against CAMEO's reactivity groups before writing the playbook.

Data scientists and ML engineers get a cleanly-keyed identity spine plus property targets whose uncertainties are explicit - rare supervised ground truth for property predictors, delivered as typed rows that join onto internal substance masters without fuzzy name matching.

Battery, refrigerant and energy-storage researchers work from the 74 fluid surfaces - viscosity and thermal conductivity across temperature and pressure are exactly what cycle models consume.

Regulatory, ESG and diligence analysts cite evaluated constants in submissions where a number needs a citation chain rather than a screenshot - and investors screening process-efficiency claims sanity-check them against thermodynamic ceilings, because no term sheet fixes a pitch that violates ΔfH°.

Name your compounds - by CAS number, formula family or InChIKey - when you request the sample, and it returns cut to them with the field dictionary intact. Changing scope afterward is a settings conversation, not a re-integration project.

Start with the NIST Chemistry WebBook dataset page - sample rows, the ten-field dictionary, coverage chips and delivery options on one page - then read the pillar, the commodity chemicals data guide, which situates the WebBook among all 27 pooled records in this industry. The scored shortlist sits at best commodity chemicals datasets.