EU PFHxA Limit: A Clean Formula Can Still Fail at 25 ppb
Since 10 October 2026, the EU has capped a chemical called PFHxA at 25 ppb in every cosmetic product sold on its market. PFHxA belongs to the PFAS family, often called "forever chemicals". A second cap of 1,000 ppb covers related chemicals that can break down into PFHxA. The rule never asks whether the chemical was added on purpose, and that is the catch: a formula sheet cannot answer it. What decides the outcome is the impurity statement from the raw material supplier, and most suppliers were never asked for numbers this precise.
What exactly changed on 10 October 2026?
The EU adopted Commission Regulation 2024/2462 on 19 September 2024, and it took effect on 10 October 2024. It added a new rule, known as entry 79, to REACH, the EU's main chemicals law. The rule covers PFHxA (a PFAS chemical with a six-carbon chain), its salts, and a defined group of related chemicals that can break down into PFHxA.
The rule says these substances may not be sold or used at a concentration of 25 ppb or more (PFHxA and its salts combined), or 1,000 ppb or more (the related chemicals combined). The amount is measured in the finished material itself, not the whole packed product. A ppb is one part in a billion. In a 50 ml jar of cream, 25 ppb is about 1.25 micrograms.
The rule covers five product groups: clothing textiles, leather, furs and hides for the general public (and their accessories); footwear; paper and cardboard that touches food; mixtures sold to the general public; and cosmetics, as defined in EU cosmetics law (Regulation (EC) No 1223/2009). The same entry has other start dates too. Firefighting foams for training and public fire services started on 10 April 2026. Non-clothing textiles follow on 10 October 2027, and civil aviation foams on 10 October 2029. Cosmetics got a two-year grace period, which ended this month.
Why does it matter that the rule ignores intent?
Most chemical rules are about a choice. A banned-ingredient list asks what the maker decided to put in, and the answer is on the formula sheet. Entry 79 is about a measurement instead. It sets a concentration limit and stops there. It makes no difference whether the chemical was added on purpose, arrived as a contaminant in a pigment, or formed while the batch sat in a warehouse.
The rule decides which chemicals count by their molecular shape, not by what they do in the product. A "PFHxA-related substance" is any chemical that could break down into PFHxA because of its structure. In practice, that means chemicals built around a chain of five or six fully fluorinated carbon atoms (written C5F11- or C6F13-). A few specific forms are carved out, including C6F14 and C6F13-C(=O)OH. So whether a material is covered depends on its chemical skeleton, not on its job in the product.
The closest neighbouring rule works differently. France's Law No. 2025-188 was adopted by the National Assembly on 20 February 2025 and published a week later. From 1 January 2026 it bans making, importing and selling cosmetics with intentionally added PFAS, with an exemption for products below a residue limit that a later decree will set. So the French rule turns on intent and allows a small leftover amount. The EU rule turns on a number, with no intent test and no allowance for unavoidable traces.
How can PFHxA end up in a formula nobody added it to?
PFHxA is not a cosmetic ingredient. It does nothing useful in a cream and never appears on an ingredient list under its own name. It reaches a finished product in one of two ways: as an impurity carried in a raw material, or as a breakdown product of a larger fluorinated chemical that was added deliberately.
Some fluorinated materials do appear on cosmetic ingredient lists, and the list is short. It includes PTFE (the non-stick coating chemical), perfluorodecalin, perfluorohexane and a handful of silicone and ether compounds with similarly long names. They help products last longer, resist water and hold colour, so they are mostly found in powders, mascaras, foundations and long-wear treatments. Only some of them fall inside the structural boundary the rule draws, and the name on the label is a weak guide to which ones.
That is why checking the ingredient list is the wrong first step. A product with none of those names can still carry more than 25 ppb, because the residue can travel in the pigment, the silicone or a processing aid rather than in the functional ingredient. A product that does contain one of them can still sit well under the limit. You cannot predict either result from the formula alone.
What does "measured in the finished material" mean for a cosmetic?
For a layered product, the rule looks at the smallest part that can be separated on its own, such as a coating measured apart from what it sits on. For a cosmetic, that part is the formula itself: the cream, the pressed powder or the mascara. The weight of the bottle, the cap and the carton is not counted.
That cuts two ways. You cannot dilute a result by spreading it across the whole packed unit, which some exporters hoped for. But the limit applies to the finished batch, not to each incoming material, so a pigment with a high residue can still pass once it is mixed in at a small share. The maths is doable, but it has to be done material by material, with a number for each one, and redone whenever a supplier changes.
The rule also has no "treat it as zero" threshold for very low test results. The explanatory notes at the start of the law say that testing methods are available and reliable, but they do not create a safe harbour in the rule itself. There is no clause for results below the limit of quantitation (the lowest amount a test can reliably measure), and no exemption for unavoidable traces.
Which stock is covered, and which is exempt?
Paragraph 7 is the transition rule, and it is one sentence: the limits do not apply to articles and mixtures that were already placed on the market before 10 October 2026. Paragraph 8 does the same job for the 2027 textile date, but covers articles only.
Everything turns on what "placed on the market" means. Under REACH, it means supplying a product or making it available to someone else, whether paid or free, and importing counts as placing on the market. So the key event is customs clearance into the EU. A consignment cleared on 9 October 2026 was already on the market and stays outside the restriction for the rest of its shelf life. Identical product sitting in a Korean warehouse that same day had not been placed on the EU market, and every unit of it will be placed on the market after the cut-off.
Recital 25 puts the burden of proof where you would expect: the restriction should not apply where the supplier can show the product was placed on the market earlier. "Show" is the key word. Stock that crossed in time is exempt only if there is a record proving it crossed in time.
Why is that exemption worth far less to an exporter than to a distributor?
For an EU distributor holding stock, paragraph 7 is a lasting asset. The stock in the warehouse was placed on the market before the date, so it stays outside the restriction and can be sold down at its own pace. The problem has a limited size and an end date.
For a manufacturer outside the EU, paragraph 7 is almost worthless. Nothing in the home warehouse was ever placed on the EU market. Every future shipment counts as a new placing and is judged against the limit at the moment it clears customs. An importer can bank this transition relief, but the exporter never holds it. That is why the two sides of the same supply contract read the same paragraph so differently.
The practical cost comes from that gap, and it is not a reformulation cost. Removing a named ingredient is a project with a clear scope. Proving that an incoming lot of titanium dioxide, a silicone elastomer or a wax carries less than 25 ppb of a six-carbon acid is different. It is a paperwork problem spread across every material in the recipe, and suppliers have rarely been asked that question at that level of detail. The real constraint is data from upstream suppliers, and reformulating does not fix it.
What does Korea's own PFAS rule cover, and what does it leave open?
Korea treats these chemicals as banned cosmetic ingredients. In May 2024 the Ministry of Food and Drug Safety (MFDS) updated its testing guide for banned ingredients to add one method that tests for 12 PFAS at once, the first time the group appeared in that guide. No numerical limit comes with it, because a ban does not need one: the permitted amount is none, and the question is whether the substance was used.
That is a coherent rule, and it asks a different question from the EU one. A manufacturer who can certify that no PFAS were used has fully satisfied the Korean rule, but has produced nothing that answers entry 79, which asks what is in the material rather than what went into it. Two documents, two questions, and the first does not answer the second.
Whether Korean cosmetics actually carry PFHxA above the EU limits is not known from public data. No survey of PFHxA in Korean products or raw materials has been published. The screening study most often cited when the guide was revised was Canadian and measured total fluorine, not PFHxA. It found fluorine in 52 per cent of the cosmetics tested, including 82 per cent of waterproof mascaras and 63 per cent of foundations. Total fluorine shows that the wider chemical family may be present, but it is not a measurement of the restricted substance. What we can see here is how the two rules are built, not the result of tests that have not been done.
How long did this take, and what comes next?
Germany submitted the restriction proposal to the European Chemicals Agency in December 2019. The agency's risk committee backed a broad restriction in June 2021, and its socio-economic committee agreed in December 2021. The European Commission adopted the regulation in September 2024, it took effect that October, and cosmetics got two more years. From proposal to enforceable number took close to seven years.
A much broader EU-wide PFAS restriction is moving through the same process and is now past its midpoint. The risk committee published its final opinion, and the socio-economic committee a draft opinion, in March 2026, with public consultation running to 25 May 2026. The socio-economic committee is expected to adopt its final opinion by the end of 2026. After that the Commission prepares a proposal for the REACH Committee, and the European Parliament and Council review it with a power of veto but no further required consultation.
The lists of exemptions are the part worth reading. The risk committee supported exemptions for protective equipment and related waterproofing agents, and rejected a proposed exemption for fully degradable PFAS groups because of a lack of evidence. The socio-economic committee recommended time-limited exemptions for eight more sectors: printing, sealing, mechanical applications, other medical uses, military, explosives, technical textiles and broader industrial uses. Cosmetics appears on neither list.
Where does this leave suppliers, brands and buyers?
For the raw material supplier, an impurity profile stops being internal know-how and becomes a product specification. A pigment maker that can put a PFHxA number on a certificate of analysis (a lab report for a specific batch) is selling something that a maker without one is not, and that difference now has a price.
For the brand and its contract manufacturer (an ODM: a company that develops and makes a formula that the client sells under its own name), the key document changes. The formula sheet used to prove compliance under an ingredient-based rule. Under a concentration-based rule it proves very little, and the file of incoming-material test results takes its place. Those are handled by different departments, with different budgets and different levels of seniority.
For buyers and investors, the due-diligence question changes in the same way. Asking whether an ingredient list is clean was a fair test while the rules were about ingredients. Against a number measured in the finished material, the useful question is narrower and harder to answer well: can this supplier produce a figure, for this material, from this lot?
Seven years of process ended in a limit that nobody puts in the jar on purpose. The formula was never the control point. The certificate of analysis is.
Comments