Chemical ID Systems — Research Summary

🔢 Chemical Identifier Systems

CAS Registry Number Commercial

American Chemical Society (ACS) • Since 1965 • 290M+ substances

Universal standard — everyone uses it
Unique, unambiguous identifier
Works across languages

Commercial — $6+ per lookup
CAS RN is trademarked IP
Can't verify without paying
Errors propagated because no one checks
Doesn't cover mixtures/undefined compositions
ACS told Wikipedia not to verify numbers (!)

PubChem CID Free

NIH / National Library of Medicine • Since 2004 • 115M+ compounds

Completely free and open
Government-backed (NIH mandate)
Structure data freely available
Actively maintained
Deliberately moved away from CAS

Only covers compounds with structures
Multiple CIDs can map to same CAS
Less universal recognition than CAS

InChI / InChIKey Open Standard

IUPAC • Since 2005 • Computable from structure

Free, non-proprietary
Can be computed — doesn't need authority
Structure information encoded in ID
InChIKey is web-searchable (27 chars)

Not human-readable
Can't handle all chemistry (clays, polymers)
Tautomers cause issues
InChIKey collisions possible (rare)

EC Number (ECHA) Free

European Chemicals Agency • EU regulation (REACH)

Free to access
Required for EU market
Good regulatory data

EU-only coverage
Not all substances have EC numbers
Less global recognition

"We (including Wikipedia) should now switch from using CAS numbers to using PubChem IDs wherever possible... PubChem has deliberately moved away from CAS because CAS numbers are IP." — Peter Murray-Rust, Cambridge chemist, 2008

"~120,000 of the 350,000+ chemicals in commercial products were too poorly described to link to a CAS number or their identities were withheld as trade secrets." — Environ. Sci. Technol. research

🛒 Consumer Safety Databases

EWG Skin Deep Biased

Environmental Working Group • 88K+ products

Free consumer access
Large product database
Raised awareness of ingredient safety
Easy-to-understand ratings (1-10)

80% of toxicologists say EWG overstates risks
Ratings for "data: none" ingredients
Chemically identical compounds rated differently
Natural bias — natural ingredients rated better
Doesn't update with new research
Pay-to-play "EWG Verified" program
Amazon affiliate links on "dangerous" products

Safety Data Sheets (SDS) Free

Manufacturer-required • GHS standard (16 sections)

Legally required for hazardous chemicals
Standardized format (since 2012)
Comprehensive information
Manufacturer-specific data

Hard to find — buried in manufacturer sites
Written for industrial/workplace use
Not consumer-friendly language
41% of SDSs in one study didn't mention combustibility
Many consumer products don't have SDS
Not designed for pets/children contexts

PubChem Free

NIH • Comprehensive compound data

Free, authoritative, government-backed
Chemical/physical properties
Hazard information
Literature citations

Designed for researchers, not consumers
Technical language
No product-level data
No pet-specific information

"A decade ago, George Mason University surveyed ~1000 members of the Society of Toxicology. 80% felt that EWG overstated the risks of chemicals." — The Eco Well, citing toxicologist survey

📊 Comparison Matrix

System	Free?	Authority	Consumer-friendly	Pet data	Products
CAS	No ($6/lookup)	High	No	No	No
PubChem	Yes	High	Somewhat	No	No
InChI	Yes	High	No	No	No
ECHA	Yes	High	Somewhat	No	No
EWG	Yes	Low (biased)	Yes	No	Yes
SDS/MSDS	Yes	High	No	No	Some
HQ Safety DB	Yes	High (cited)	Yes (goal)	Yes	Yes

🕳️ The Gap HQ Fills

            What exists:
            High authority, bad UX: PubChem, ECHA, SDS — accurate but not consumer-friendly
Consumer-friendly, bad authority: EWG — accessible but biased/inaccurate
Universal identifier, paywalled: CAS — everyone uses it, but verification costs money

        

            What doesn't exist:
            Free, accurate, consumer-friendly safety data
Multi-context data (human, pet, environmental) in one place
Compound → Material → Product linkage
Non-extractive (no pay-to-play verification programs)
Citable, ALCOA+ compliant, conflict-acknowledged

        

🔧 HQ Nomenclature Approach

Layer	HQ ID	Cross-references
Compound	`hq-c-0001`	CAS, PubChem CID, InChIKey, ECHA EC
Material	`hq-m-0001`	Resin codes, ASTM standards
Product	`hq-p-0001`	None (our category, our ID)

            Why our own IDs + crossrefs?
            Independence: We don't depend on CAS (commercial) or any single system
Verification: Anyone can check our data against official sources
Coverage: We can ID things that don't have CAS numbers (mixtures, products)
Stability: Our IDs won't change if external systems change
Non-extractive: Our system is free, theirs may not be

        

Proposed Schema Addition:

{
  "hq_id": "hq-c-0001",
  "name": "Glyphosate",
  
  "crossrefs": {
    "cas": "1071-83-6",
    "pubchem_cid": "3496",
    "inchi_key": "XDDAORKBJWWYJS-UHFFFAOYSA-N",
    "echa_ec": "213-997-4"
  },
  
  "identity": { ... },
  "safety": { ... }
}

✓ Resolved Decisions

Q: Should HQ IDs be in filenames? RESOLVED

✓ Option A: hq-c-0001.json — Machine-friendly, unambiguous
Option B: glyphosate.json — Human-friendly, HQ ID inside file only

Human-readable name lives inside the file, not in filename.

Q: Sequential vs structured IDs? RESOLVED

✓ Option A: hq-c-0001 — Simple sequence, just order of entry
Option B: hq-c-herb-0001 — Category embedded
Option C: hq-c-2025-0001 — Year embedded

Simple sequential. Categories change, years complicate lookups. Registry tracks assignments.

Q: How to handle "same compound, different form"? RESOLVED

Glyphosate acid vs glyphosate isopropylamine salt — same thing?
Option A: Same HQ ID, different CAS in crossrefs array
✓ Option B: Separate HQ IDs, linked via hierarchy

Parent compound (hq-c-0001 glyphosate) → children reference via hierarchy.parent.
Children have own CAS numbers but inherits_safety: true from parent.

📊 Current Registry State

As of March 2026

888

Active Compounds

+ 295 aliases = 1,183 files

Next: hq-c-1226

262

Materials

hq-m-0001–0262

Next: hq-m-0263

237

Products

237 active (with gaps)

Next: hq-p-0362

Key Coverage Milestones

Field	Coverage	Notes
hazard_profile, found_in, alternatives	100%	All 888 active compounds
regulatory.classifications	95%	~43 at ceiling (no classifiable language)
GHS hazard data	80%	Ceiling: mixtures/classes lack single SDS
identity (formula + SMILES)	78%	Ceiling: classes/mixtures can't have structure
dose_response.ld50	68%	PubChem exhausted; CTX API pending

First Family: Glyphosate (still canonical example)

hq-c-0001  Glyphosate (acid)           CAS 1071-83-6      [parent]
    ├── hq-c-0002  IPA salt            CAS 38641-94-0     [alias → hq-c-0001]
    ├── hq-c-0003  Potassium salt      CAS 70901-12-1     [alias → hq-c-0001]
    └── hq-c-0004  Ammonium salt       CAS 114370-14-8    [alias → hq-c-0001]

Chemical Identification Systems — Research Summary