You Can Choose What You Add, Not What Is in the Bale: Europe's PFAS Total-Fluorine Limit

Somewhere in Germany this month a converter will open a bale of post-consumer board, run a total-fluorine screen on it, and discover a number they did not put there. The number will be legal. It may still fail. That, in one sentence, is the PFAS compliance problem in European food contact packaging — and it is not the problem most suppliers spent the last two years preparing for.

Regulation (EU) 2025/40 applies from 12 August 2026. It replaced Directive 94/62/EC, and alongside its recyclability and recycled-content duties it sets three limits on per- and polyfluoroalkyl substances in food contact packaging: 25 parts per billion for any individual PFAS by targeted analysis, 250 parts per billion for the sum of PFAS, and 50 parts per million for total fluorine including polymeric PFAS. The industry has talked mostly about the first two. They are the easy ones.

Three numbers, and only one of them is about what you did

The 25 ppb and 250 ppb limits are about intent. They measure named molecules that somebody either added or did not add, and a supply-chain declaration can in principle settle the question without anyone owning a laboratory. A formulator who has removed the fluoro treatment can say so, and the document satisfies the file.

The 50 ppm total-fluorine limit is a different kind of test. It combusts the sample and measures the fluoride released. It does not know what molecule the fluorine was in, and it does not care whether the fluorine was deliberate. It captures polymeric PFAS that targeted methods are not looking for, and it captures fluorine that arrived in the material through mineral fillers, process chemicals, or the recycled fibre the converter bought on the merchant market at a price that reflected nobody’s analytical obligation.

A producer can comply with 25 ppb by choosing what to add. Nobody chooses what is already inside the recycled fibre they buy.

That is why the constraint in this market turns out to be analytical rather than chemical. Workable non-fluoro barriers exist for most applications, and the ones that do not exist yet are a known and bounded list. What does not exist is a cheap, fast, defensible way to certify a 50 ppm total-fluorine ceiling on a recycled substrate whose fluorine content the buyer cannot see. The investment that fixes that is at-line measurement and fibre traceability. The supplier who sells it is not a chemical company.

What actually leaves, and where it goes

Pheonix’s Europe PFAS-Free Food Contact Packaging Market Report measures the EU-27 pool of food contact fibre packaging that functionally requires a grease, oil or water barrier at 1,520 kilotonnes and €3,404.6 million in 2026, reaching 2,645 kilotonnes and €6,924.6 million by 2033 — 8.24 per cent volume, 10.67 per cent value, 2.25 per cent price.

Inside that, 334.4 kilotonnes — 22 per cent of the 2026 pool — is still fluorochemical-treated. By 2033 it is 52.9 kilotonnes. The transfer is 281.5 kilotonnes in seven years, which is 10.6 per cent of the 2033 pool and 25.0 per cent of the pool’s entire growth. A conversion of that size is not a subplot inside this market’s growth rate. It sets it.

But the destination is the finding, not the exit. The two fastest-growing places for that tonnage are uncoated high-caliper fibre, up 136.2 per cent to 502.6 kilotonnes, and sol-gel and silicate chemistry, from 60.8 to 238.1 kilotonnes. Neither is a conventional water-based dispersion. Where a wet coating cannot hold hot grease at temperature, the market does not coat harder — it thickens the fibre until the barrier is the substrate, or it changes the chemistry to something inorganic. The obvious answer, a better coating, is only the fourth or fifth most common answer.

The tonne changes format more often than it changes coating

Strip a fluoro barrier off a wrap and the replacement is frequently not another wrap. A large share of applications end up in a moulded fibre bowl, tray or clamshell — a different format, at a different price, on a different machine, from a supplier who may not be the one who lost the business.

This is why the format cut explains more of this market than the chemistry cut. Moulded fibre ware is 26 per cent of the pool’s tonnage and 42.8 per cent of its value, 1.65 times the pool average. Realisation runs from €1,075 a tonne for liner and interleaving papers to €3,684 for moulded fibre — a 3.4-times spread that is the mechanism by which the market grows. Value compounds 2.4 points faster than volume; price inflation contributes 2.25 per cent and the rest is mix.

The corollary is a warning for anyone modelling this as a straight substitution. If the chemistry swap happened in place, this would be an 8 per cent volume story with flat prices. It is a 10.67 per cent value story precisely because in the hardest applications the swap is not possible in place, and the market routes around the gap by changing the product. That puts thermoformed and moulded capacity and fibre supply in front of the same investment decision, which is not where a packaging analyst would normally look for a barrier story.

48 per cent of the pool is asserted, not measured

Pheonix’s compliance-route cut is the least comfortable table in the report. Thirty-four per cent of the pool sits on total-fluorine screening with targeted confirmation, and 18 per cent on targeted LC-MS/MS alone. That is 52 per cent measured. The other 48 per cent runs on 31 per cent supplier declaration or chain of custody, and 17 per cent no documented verification at all.

A declaration is only as good as the declarer’s visibility, and for a converter buying board through a merchant, visibility frequently stops one tier up. This is where the first failures will appear — not in the plants that reformulated early, but in the ones holding a document that says the question has been answered. Enforcement is national, so the effective deadline is set by market surveillance capacity rather than by the Official Journal. The obligation is identical across all 27 member states on 12 August 2026. The probability of being asked to prove it is not, and Germany, France and Italy — 48 per cent of the pool between them — are where that probability is concentrated.

€12 against €95, and the percentage you choose

The cost argument is usually made with the wrong denominator. On Pheonix’s model, a fluorochemical treatment added roughly €12 to a tonne of board. A dispersion barrier adds about €95. That is 7.9 times as much — and also 8.7 per cent of a €950 substrate, and 3.8 per cent of a €2,180 finished pack. The whole commercial negotiation between converter and customer is which of those percentages is treated as the reference point, and nothing about the chemistry decides it.

Capital is smaller still. Converting the full 281.5 kilotonnes at €144 of installed capital per annual tonne for coating, against €505 for moulded fibre, is roughly €81 million against a €3.4 billion pool — 2.4 per cent. Money is not the barrier. Twelve to twenty-four months of migration testing and customer re-qualification per SKU is the barrier, and a calendar does not compress because a deadline was legislated.

Which explains the structure of the supply side. No participant holds a double-digit share, and the reason is not that the market is young. Only the chemistry layer was ever concentrated; the moat has moved to fibre access plus the qualification file — the accumulated evidence that one specific pack, on one specific line, held one specific food at one specific temperature. That file is why Michelman positions its VaporCoat range on repulpability rather than on grease resistance alone: it is selling into the PFAS limit and the recyclability duty that arrives beside it. It is why Huhtamaki competes across 35 countries on breadth of format rather than on a molecule.

What would have to be true

The base case needs three things. Recycled fibre has to stay available at 638.4 kilotonnes — 42 per cent of the pool — while being simultaneously claimed by recycled-content obligations elsewhere, which means the same molecule is being counted twice by two pieces of legislation. Enforcement has to arrive at something close to the statutory date. And the 48 per cent asserted share has to convert to measurement without stalling qualification queues for longer than the forecast allows.

The three scenarios bracket €6,540 million to €7,290 million against the €6,924.6 million base — a −5.6 to +5.3 per cent band. That looks like a narrow band and is not: the low case is slower enforcement plus a harder substitution shift to polypropylene, CPET and aluminium, and the high case is faster conversion with recycled-content premiums surviving. Both are policy outcomes, not demand outcomes.

The date is real. The enforcement date is a national budget question. Between those two facts sits the entire first two years of this market, and the companies that will do well are not the ones with the best coating — they are the ones who can produce a number for a bale they did not make.

Basis of estimate: the regulatory framework, the application date and the three limits are taken from the adopted instruments. Every volume, share, realisation price and forecast path is a Pheonix model, built bottom-up from format volumes and priced at format-level realisation; the six segmentation cuts are independent marginals, not a matrix. The full report states which figures are published and which are modelled.