SDS Section 3: Composition / information on ingredients

The only section that says what is actually in the container

Every other heading on a safety data sheet makes a statement about the product. Section 3 makes a statement about its parts, and everything else is derived from it. The classification in section 2 is calculated from this composition. The exposure limits in section 8 attach to the substances named here, not to the trade name on the front page. The candidate-list and authorisation entries in section 15 apply to those same substances. If section 3 is wrong, nothing downstream of it can be right, and no amount of care in the other fifteen sections repairs it.

Which is why most of the regulatory text under this heading answers one question that has nothing to do with chemistry: which ingredients must be named, and at what concentration does the duty to name them begin. That question has three different answers in the three legal texts, and they are further apart here than in almost any other section.

3.1 — a substance is rarely just one substance

The EU text opens by asking for the chemical identity of the main constituent, given by at least the product identifier or one of the other means of identification already used in subsection 1.1. That much is uncontroversial. The interesting part is what follows: the identity of any impurity, stabilising additive or other constituent that is itself classified and contributes to the classification of the substance.

Both conditions have to hold. A 0,3 % impurity that carries no classification of its own stays out of section 3. A 0,3 % impurity classified as a respiratory sensitiser, and which is the reason the substance as a whole carries a sensitisation classification, goes in — with its own identity. The same two-part test appears word for word in OSHA’s Appendix D and in A4.3.3.1.4 of GHS Annex 4. It is one of the few places in section 3 where all three regimes agree exactly.

The consequence for a reader is that a “pure” substance sheet with a populated impurity line is not a worse product. It is usually a better-characterised one.

Three numbers that are not properties

Annex II 3.1 also asks for the specific concentration limit, the M-factor and the acute toxicity estimate, where those exist for substances in Part 3 of Annex VI to CLP or as determined under Annex I. None of the three is a property of the substance in the way a boiling point is. Each is a classification input — a value that someone further down the chain needs in order to work out how the substance behaves when it becomes one ingredient among several.

That is why they live in section 3 rather than in section 11 or section 12 alongside the underlying study data. The acute toxicity estimate in particular exists to be consumed by a mixture calculation — see the ATE mixture calculator — and a substance sheet that omits an ATE it has forces every downstream formulator to reconstruct one.

3.2 — the thresholds that decide who gets named

A classified mixture: the lowest threshold wins

Subsection 3.2.1(a) is the single most misapplied provision in section 3, because it looks like a threshold and is in fact a comparison. A substance presenting a health or environmental hazard must be listed when it is present at or above the lowest of a set of values: the generic cut-off values in Table 1.1 of Annex I to CLP, the generic concentration limits in Parts 3 to 5 of that Annex, a specific concentration limit from Part 3 of Annex VI, a cut-off adjusted by an M-factor, a limit notified to the classification and labelling inventory, or one tenth of the specific concentration limit of a sensitiser.

An author who applies “one per cent” as a working rule will therefore be right about a great many ingredients and wrong about precisely the ones with the sharpest hazards. Two further routes sit alongside it: 3.2.1(b) catches any substance with a Union workplace exposure limit that (a) missed, and 3.2.1(c) catches PBT and vPvB substances, candidate-list substances and endocrine disruptors at 0,1 % by weight.

A mixture that is not classified still has a section 3

This is the provision most often absent from sheets that are otherwise competent. Subsection 3.2.2 applies to mixtures that meet no CLP classification criterion, and it still sets six separate thresholds — running from 1 % by weight down through 0,1 % to 0,01 % for a category 1A sensitiser, and to one tenth of a specific concentration limit where a sensitiser carries one.

A product can therefore be entirely unclassified, carry no pictogram and no signal word, and still owe the reader a populated section 3. The lowest of those numbers is one part in ten thousand.

Exact percentages, ranges, and what a range is allowed to hide

All three texts accept ranges; they disagree on when. REACH Annex II 3.2 gives a free choice between exact percentages and ranges, both in descending order by mass or volume, where technically possible. OSHA’s Appendix D does not: exact percentages are the default, and a range is permissible only under a trade-secret claim, where there is genuine batch-to-batch variability, or for a group of substantially similar mixtures with similar composition.

UN GHS Annex 4 adds the safeguard that keeps a range honest. Under A4.3.3.2.3, where a proportion range is used the health and environmental hazard effects described should be those of the highest concentration of each ingredient, unless data for the mixture as a whole are available. A range is a permitted imprecision about quantity, not a licence to describe the mildest end of it.

Withholding an ingredient: two systems that do not resemble each other

EU — an alternative name, and nothing else

Annex II 3.2 offers exactly one accommodation: where the use of an alternative chemical name is permitted under Article 24 of Regulation (EC) No 1272/2008, that name may be used. Everything else survives. The concentration or range is still declared, and under 3.2.3 the substance’s hazard class, category codes and hazard statements are still printed. A European reader who meets an alternative chemical name loses the ability to look the substance up — and loses nothing else.

US — a trade-secret claim, with a prescribed range ladder

The US section 3 opens “Except as provided for in paragraph (i) of § 1910.1200 on trade secrets”, which tells you how tightly the two are bound. Under paragraph (i)(1) the claim must be supportable, the properties and effects of the chemical must still be disclosed, and the sheet must state that the specific chemical identity and/or the exact concentration has been withheld as a trade secret. Where the concentration is withheld, § 1910.1200(i)(1)(iv) supplies thirteen fixed ranges, lettered (A) to (M) — from 0.1 % to 1 %, from 0.5 % to 1.5 %, from 1 % to 5 %, and so on up to from 80 % to 100 %. The author picks one; the author does not design one. And the identity itself must be released to a treating health professional where a medical emergency exists.

GHS Annex 4 states the underlying principle for both regimes in its note to A4.3.3: competent authority rules for confidential business information take priority over the rules for product identification, and where they apply, the sheet must indicate that confidential composition information was omitted. In none of the three systems is an unexplained gap a compliant answer.

How section 3 connects to the rest of the sheet

• The identity used here is the identity established in section 1 — 3.1 sends the author back to subsection 1.1 for the means of identification, so a synonym that appears only in section 3 is already an inconsistency.
• The classification in section 2 is the arithmetic result of this composition. Where the two disagree, one of them has not been updated after a reformulation.
• Every exposure limit in section 8 attaches to a substance named here, and 3.2.1(b) exists so that a substance with a Union limit cannot be invisible in section 3 while its limit is quoted in section 8.
• The acute toxicity estimates and M-factors declared here are the mixture-level counterparts of the study data in section 11 and section 12.
• The candidate-list, authorisation and restriction entries in section 15 hang off the CAS and EC numbers listed here; a missing identifier makes the regulatory section unverifiable.
• The hazard statements that 3.2.3 requires per ingredient are abbreviated to codes; their full text belongs in section 16.

What a good section 3 looks like

1. Can you follow every identifier out of the section? Take each CAS or EC number and find the same substance in sections 8, 11, 12 and 15. Where an identifier appears here and nowhere else, either the rest of the sheet is incomplete or this line is decorative.
2. Does each named ingredient carry its classification and its values? Hazard class and category codes, hazard statements, and the specific concentration limit, M-factor and acute toxicity estimate wherever one exists. A table of names, numbers and percentages with no classification column meets the US text and fails the EU one.
3. Where something is absent, does the sheet say why? An Article 24 alternative chemical name, a trade-secret statement with a prescribed range, or an explicit note that confidential composition information was omitted. Absence with an explanation is compliant; absence without one is a defect you can see from across the room.