Two German studies report genuine breakdown of a “forever chemical”, not just capture. But the measure that shows carbon–fluorine bonds actually breaking lags far behind the removal figure — and in the best-performing runs, most of the fluorine could not be traced at all.
Researchers at the Helmholtz-Zentrum Dresden-Rossendorf published two papers this year on breaking down perfluorooctane sulfonate, or PFOS — among the most studied of the “forever chemicals”, and one used for decades in firefighting foams, textiles and non-stick coatings.1 One uses cold atmospheric plasma with air bubbled through the water.1 The other forces water through a narrow orifice at 48 bar until vapour bubbles form and collapse.2 Both work. Neither is a filter.
The nickname is earned by a single feature of the chemistry. PFOS is a chain of eight carbon atoms sheathed in fluorine, and the bond between carbon and fluorine is the strongest in organic chemistry — strong enough that neither weathering, nor bacteria, nor conventional water treatment can break it.1 That is why the compounds accumulate: not because they are produced in vast quantities, but because almost nothing takes them apart again.
That distinction is the point. Activated carbon, ion-exchange resin and high-pressure membranes — the established options — strip more than 99% of PFOS out of drinking water.3 They do it by moving the molecule onto a solid, intact. The carbon then becomes contaminated waste, usually bound for an incinerator.1 Nothing is destroyed; the problem is relocated.
So the papers report two different numbers, and the distance between them is the story. Degradation is how much PFOS is left: the instrument searches for that one molecule and reports what it can no longer find. Defluorination is how much fluorine has come away as fluoride — a simple inorganic ion that is no longer part of any fluorinated molecule at all. Only the second is direct evidence that a carbon–fluorine bond actually broke.
A rough analogy: measuring plastic pollution by counting intact bottles. Smash the bottles and the count falls to zero, but the plastic is all still present as fragments. Degradation is the bottle count. Defluorination is the measure that tells you material genuinely left the system — and in these experiments the two numbers are a long way apart.
A single 20-minute plasma run makes the divergence plain. PFOS collapses — 92% gone in five minutes, 99% in ten — a half-life of 1.6 minutes, meaning half of whatever is present clears every minute and a half.1 Fluoride release does not track it: 33% at five minutes, 27% at ten, 35% at twenty.
The authors are explicit that the parent molecule is not simply vanishing. They detect a chain of shorter perfluorinated acids forming as PFOS breaks down — PFOA first, then progressively shorter.1 Those are still PFAS, and the shorter ones adsorb less readily than the compound they came from.4
Cavitation looks far worse on the headline measure: 37% degradation against the plasma’s 99.99%.2 But ask a different question — of the PFOS each method did break down, how much was taken apart all the way? — and the ranking inverts. Cavitation’s 20% defluorination against 37% degradation works out at 54%. The plasma’s 35 against 99.99 works out at 35%. Per unit of PFOS destroyed, the slower method freed roughly one and a half times as much fluorine.
A second pattern in the plasma paper runs the same way. Bubbling air through the water is what lifts degradation from 76% to 99.99%. It also makes the fluorine accounting markedly worse: without gas dispersion the authors recover 68–106% of the fluorine they started with, and with it, 31–50%.1 The configuration that produces the impressive removal figure is the one that loses track of most of the fluorine.
One reading — interpretation, not a finding either paper states — is that this is the same effect twice: fast removal of the parent compound outruns the slower work of cleaving C–F bonds, leaving more material in intermediate states nobody has measured. A peer-reviewed review of the wider field this year, unconnected to the Dresden group, names incomplete defluorination as the central unsolved bottleneck.6 Consistent with the pattern — not proof of it.
The Dresden group’s own stated next step is to combine the two processes, on the expectation that plasma’s reactive species and cavitation’s collapsing bubbles will do together what neither manages alone.7 That work has not been published.
Where most of the fluorine went. In the gas-dispersed runs, between 50% and 69% of the fluorine present at the start could not be found at the end — not as fluoride, not as any of the intermediates the team screened for. The authors list unidentified fluorinated products, escape into the gas or foam phase, and adsorption onto reactor walls as possibilities, and state that the fate is unresolved.1 This is disclosed in the paper, not uncovered here.
Whether either method scales. Both were run on laboratory volumes — 10 to 40 millilitres in the plasma work — and at PFOS concentrations of 1 to 5 milligrams per litre. UK drinking water is regulated against a guideline of 0.1 micrograms per litre for the sum of 48 PFAS.5 The experiments ran at 1 to 5 milligrams per litre — ten thousand to fifty thousand times more concentrated than that guideline. Dilute solutions are harder to treat, not easier, because the reactive species have to find far fewer target molecules.1
What the intermediates do. The shorter-chain acids that form during treatment are quantified, but their toxicity and persistence are not addressed by either study. Ultra-short-chain products such as trifluoroacetic acid were looked for and not detected — which the authors note may mean they were absent, or below the detection threshold.1