Forever,
by Accident

In Switzerland’s national groundwater monitoring pilot, trifluoroacetate was at or above the limit of quantification at 508 of 517 monitoring sites; 146 sites exceeded 1 µg/L and two exceeded 10 µg/L.

Caveat Sampled 2022–2023. The workbook reports counts against thresholds for the maximum value per site, not a national median.

The Swiss environment office reports that TFA concentrations in groundwater run 100 to 1,000 times higher than those of any other PFAS detected there to date.

Caveat An agency summary comparison; the page does not give compound-by-compound distributions.

At Swiss monitoring sites influenced by arable land, TFA averages around 1.2 µg/L and reaches 5 µg/L; two peaks of 14 and 23 µg/L were measured near the River Ergolz, where contaminated river water infiltrates groundwater.

Caveat Stratified results by land-use context, not a national range.

In France’s 2024–2025 national drinking water campaign, TFA was quantified in 595 of 647 raw water samples, with a median of 810 ng/L and a quantified range of 100 to 20,000 ng/L.

Caveat The limit of quantification was 100 ng/L. The 92% detection frequency is conditional on that limit — anything below it is invisible, not absent.

In the same campaign, TFA was quantified in 578 of 627 treated water samples, with a median of 780 ng/L and a range of 100 to 25,000 ng/L.

Caveat Same 100 ng/L limit of quantification applies.

Two treated water samples exceeded 10 µg/L, at 18 and 25 µg/L. Both came from plants downstream of the same TFA-producing industrial site.

Caveat ANSES could not identify sources for most sites. These two are the exception, not the pattern.

The EU water directive records that PFAS have been detected at more than 70 per cent of groundwater measuring points in the Union, with existing national threshold values clearly exceeded at a considerable number of locations.

Caveat Recital 15. Covers PFAS generally, not TFA specifically.

The dossier’s table of identified uses covers manufacture, use as an intermediate, formulation, industrial site use and professional laboratory use. It records no consumer uses and no article service life.

Caveat This is the direct-use footprint only.

Some per- and polyfluoroalkyl substances containing a single trifluoromethyl group are potential precursors to TFA. The dossier names fluorinated gases and active ingredients in biocides, plant protection products and pharmaceuticals carrying that group bound to an aromatic ring.

The EU water directive states that TFA has many sources, including the use of PFAS pesticides and fluorine-containing refrigerant gases, and that it is extremely important to address its presence in both surface water and groundwater.

Caveat Recital 18. The recital identifies source categories but does not quantify their relative contributions.

Modelling for 2023 estimates global TFA generation of 11.1 Gg per year from HFO-1234yf, against 18.9 to 53.9 Gg per year from HFC-134a, whose emissions are around 22 times higher.

Caveat Model output from assumed emissions and yields, not a measured mass balance. HFC-134a still generates more TFA globally.

The same modelling finds HFO-1234yf produces higher maximum TFA surface concentrations over Europe than HFC-134a, with deposition up to 3.6 times higher under the conservative high-yield scenario and 10.3 times higher under the low-yield scenario, because its atmospheric lifetime of around 12 days causes TFA to form and deposit close to source regions while HFC-134a disperses globally first.

Caveat Regional deposition and global generation are different quantities. HFO-1234yf dominates the first, not the second.

German federal and university researchers found that more than 80 per cent of the PFAS active pharmaceutical ingredients they examined could potentially degrade to TFA.

Caveat The release states the proportion and study scope but does not list individual ingredients.

Diflufenican, a cereal herbicide, is itself identified as meeting the PFAS definition on the basis of its chemical structure, and TFA is among its degradation products.

The classification dossier drew its main study sources from the registration dossiers for trifluoroacetic acid and its sodium and potassium salts. Literature searches across four databases returned 682 records; nine contained toxicological information and none were considered to add anything relevant.

Caveat The toxicological case rests substantially on studies commissioned by registrants.

Archived Swiss precipitation samples show the precipitation-weighted mean TFA concentration rising from 0.085 µg/L in 1986–87 to 0.307 µg/L in 2018 and 0.496 µg/L in 2023.

Caveat Early archived periods have small sample counts; the authors note some are too sparse to infer a smooth annual trend.

Average TFA in major Swiss rivers is estimated to have risen from 0.087 µg/L in 1996–97 to 0.521 µg/L in 2021–2023, excluding industrially impacted Rhone observations.

Caveat A comparison across historical and recent datasets, not one continuously operated sampling design.

Danish groundwater dated by recharge age shows TFA below quantification in all three samples recharged before 1960; 81 per cent detection and a median of 0.06 µg/L for 1960–1980 recharge; 100 per cent detection and 0.34 µg/L for 1980–2000; and 100 per cent detection and 0.50 µg/L after 2000.

Caveat Not an archived-sample time series. The trend is reconstructed from recharge ages estimated by tritium-helium dating, and some older samples may include mixing with tritium-free water.

German monitoring of monthly wet-deposition composites at seven weather stations found a median TFA concentration of 0.250 µg/L in 2018–19 and 0.244 µg/L in 2019–20.

Caveat One station excluded because its samples combined wet and dry deposition.

Germany-wide TFA deposition derived from those measurements was 187 g/km², or 67 tonnes, in 2018–19 and 276 g/km², or 99 tonnes, in 2019–20.

Caveat Deposition estimates derived from monitored concentrations and precipitation amounts, not direct measurements in water bodies.

TFA was detected in all 32 surface snow samples taken on Spitsbergen in the Norwegian Arctic between January and August 2019, at 5.6 to 270 ng/L.

No degradation of trifluoroacetate was observed in any of the four reliable studies assessed. No degradation half-life could be derived in fresh water, sediment or soil, because the rate constant in all three compartments is not significantly different from zero.

The studies ran 120 days in a laboratory water-sediment system, 365 days in a field aquatic microcosm, and 120 days in soil. The regulatory thresholds they exceed are 40 days for fresh water and 120 days for sediment and soil.

The two studies used as key evidence for mobility produced twenty log Koc values between minus 2.02 and 0.19 and five between 1.27 and 2.49. The threshold for very mobile under CLP is a log Koc below 2.

In activated carbon filtration testing, more than 90 per cent of the TFA feed concentration was detected in the filtrate.

Caveat A result for the conditions studied, not a universal performance figure for every carbon, contact time or water matrix.

At three full-scale waterworks, TFA was not removed by flocculation, ozonation, biologically active filtration or adsorptive filter media.

Across six municipal wastewater treatment plants, no decrease in TFA was observed after biological treatment. At one plant, mean TFA rose from 0.8 µg/L in the influent to 4.2 µg/L in the effluent.

Caveat One plant, and about formation from precursors rather than a failure of removal. Do not generalise the increase to all plants.

Shortening a full-scale ion-exchange filter cycle from 16 hours to about six would remove TFA, but would raise carbon dioxide regeneration costs from 0.04 to 0.12 € per cubic metre and produce regeneration wastewater equal to 67 per cent of the water produced.

Caveat Specific to the waterworks studied, and covering regeneration cost only, not total treatment cost.

Reverse osmosis does work: TFA in the permeate was below the detection limit in every test phase.

Caveat Because permeate concentrations were below detection, the study could not quantify how operating conditions affect rejection.

The European water services association states that reverse osmosis for TFA requires up to 20 per cent more water abstraction than conventional treatment, is considerably more energy-intensive, requires remineralisation, and produces a concentrated waste stream needing further treatment or disposal.

Caveat A water utilities position paper, not a peer-reviewed plant-scale measurement. Utilities have a direct interest in emphasising treatment cost.

On 7 January 2021, Bayer — as notifier of the REACH registration dossier on TFA — together with the REACH lead registrant and producers of TFA, notified EFSA, the European Commission and Member States under Article 56 of the plant protection products regulation of adverse developmental effects in rabbits observed in a new developmental toxicity study.

Caveat Article 56 is a mandatory disclosure obligation covering information on potentially harmful or unacceptable effects.

The notifying group later became the TFA task force, consisting of BASF, Bayer, Corteva and Syngenta.

In May 2024 the TFA task force submitted an update to the Article 56 notification, providing the remaining studies to EFSA, the Commission and Member States.

Caveat Reconstructed from EFSA material referring to the notification. The notifications themselves are not published.

The proposal to classify TFA as toxic for reproduction category 1B was sent to ECHA on the basis of that new developmental toxicity study in the REACH registration dossier.

The dossier’s entire stated justification for action at EU level is that Delegated Regulation (EU) 2023/707 introduced new hazard classes into CLP, creating a need for action because the classification criteria had changed.

Caveat The proposal was not triggered by new monitoring or rising concentrations. Do not let the piece imply otherwise.

TFA was already registered under REACH and already carried a harmonised classification — skin corrosion category 1A, acute toxicity category 4 and aquatic chronic category 3. What changed was the set of hazard classes against which it could be assessed.

Caveat The acute toxicity entry carried an asterisk denoting a minimum classification open to revision.

PMT and vPvM were introduced into CLP by Delegated Regulation (EU) 2023/707. A PMT substance is persistent, mobile and toxic; a vPvM substance is very persistent and very mobile, and requires no toxicity criterion at all.

Caveat These are EU-specific classes and are not part of the UN Globally Harmonised System, so the classification does not automatically travel to other jurisdictions.

The two statements read: can cause long-lasting and diffuse contamination of water resources, and can cause very long-lasting and diffuse contamination of water resources.

Mobility is defined by the organic carbon partition coefficient: a log Koc below 3 meets the mobility criterion and below 2 the very mobile criterion, applying the lowest value across pH 4 to 9 for ionisable substances.

The developmental case rests on two prenatal developmental toxicity studies in New Zealand White rabbits conducted to OECD test guideline 414.

The clear evidence for category 1B was malformation of the eye in rabbit foetuses: folding of the retina, and failure to form aqueous or vitreous humour.

Caveat Findings from animal studies at administered doses. Not evidence about environmental exposure, and must not be placed next to a drinking water concentration without that distinction stated.

Multiple folded retina appeared in none of 150 control foetuses, one of 158 at 180 mg per kg body weight per day, five of 173 at 375 mg/kg, and nine of 140 at 750 mg/kg. By litter, incidence rose from zero in controls to 4.8, 17 and 35 per cent across the three dose groups.

Caveat Absent aqueous or vitreous humour follows the same pattern: 0, 4.8, 17 and 26 per cent of litters.

The laboratory’s historical control range for each of these malformations was zero to one in 5,687 foetuses, and zero to one in 744 litters.

The dossier records 180 mg per kg body weight per day as the lowest observed adverse effect level and states that a no observed adverse effect level cannot be determined, because lower doses were not tested.

Caveat No margin of exposure can be calculated from this study.

The developmental effects occurred alongside maternal toxicity, mainly reduced body weight gain. The dossier applied guidance holding that developmental effects occurring even in the presence of maternal toxicity count as evidence of developmental toxicity unless they can be unequivocally shown to be secondary, and concluded that they could not be.

Category 1A was rejected because no human data exist: the dossier records no information supporting a known adverse effect of TFA on reproduction in humans.

At its 77th meeting on 5 June 2026, ECHA’s Committee for Risk Assessment adopted by consensus harmonised classification opinions for trifluoroacetic acid and for its inorganic salts.

Caveat Adopted by consensus — meaning no sustained objection — not by a recorded unanimous vote. Several secondary sources say unanimously; the minutes do not.

Observers present and commenting included Cefic, CropLife Europe with experts from Bayer and from BASF acting on behalf of the TFA task force, PAN Europe, CHEM Trust, EACL, EFPIA, Animal Health Europe with an expert from Zoetis, EUROPUR and Ramboll, alongside EFSA observers.

Caveat Evidence of who considered themselves exposed, not of any position taken. Do not characterise individual organisations’ views beyond what the minutes record.

Germany’s dossier proposed adding reproductive toxicity category 1B, PMT and vPvM, modifying the acute toxicity entry to category 3 with an inhalation acute toxicity estimate of 5 mg/L, and adding a supplemental statement for respiratory tract corrosivity.

The committee’s opinion adds reproductive toxicity category 1B, acute toxicity category 4 by the oral route, PMT and vPvM; modifies the acute toxicity entry to category 3; and sets acute toxicity estimates of 3 mg/L for vapour inhalation and 500 mg per kg body weight orally.

Caveat This row records what the committee changed, not the full label. Skin corrosion and aquatic chronic toxicity were not at issue and do not appear.

The dossier had assessed acute oral toxicity and rejected classification: it recorded an oral LD50 between 500 and 1,000 mg per kg body weight, acknowledged this would warrant category 4, but argued no classification was warranted because the mode of action indicated corrosivity and the sodium salt returned an LD50 above 2,000 mg/kg. The committee rejected that argument and classified anyway, setting the estimate at 500 mg/kg — the lower bound of the dossier’s own range.

Caveat The committee’s reasoning for rejecting the corrosivity argument is not available. Present the disagreement, not an account of how it was resolved.

The hazard statement is a compound one. The capital D carries the category 1B conclusion that the substance may damage the unborn child; the lower-case f carries the weaker conclusion that it is suspected of damaging fertility.

Caveat Widely reported summaries flatten this to "category 1B toxic to reproduction", which overstates the fertility finding.

For the inorganic salts the opinion is narrower — reproductive toxicity category 1B, PMT and vPvM, with no acute toxicity classification — and the group entry covers only hazardous properties common to all members, so individual salts may require separate evaluation.

Caveat Commercially significant: a salt portfolio cannot be read as fully covered by the group entry.

The toxicity criterion for the PMT classification is satisfied by way of the reproductive toxicity classification. The PMT designation therefore traces back to the rabbit eye malformations. The vPvM designation does not: it stands on persistence and mobility alone and would hold regardless.

The dossier assessed specific target organ toxicity on repeated exposure and concluded that classification for that class is not appropriate. It appears in neither adopted classification.

Several hazard classes were not assessed in the dossier at all, including carcinogenicity, germ cell mutagenicity, sensitisation, endocrine disruption and aquatic environmental hazard.

Caveat Not assessed is different from assessed and rejected. Both appear in the same summary table and must not be conflated.

The opinion does not amend Annex VI to CLP and does not create a binding harmonised classification. Binding effect would require the European Commission to adopt an adaptation to technical progress and publish it in the Official Journal.

The opinion is a hazard classification. It does not conclude that measured environmental or drinking water concentrations of TFA are causing harm.

The opinion is not a REACH restriction, a plant protection product authorisation decision, an F-gas measure or a drinking water measure, though it can inform each of them.

On 22 July 2026, EFSA lowered the acceptable daily intake for TFA to 0.014 mg per kg body weight per day, down from 0.05 — a reduction by a factor of around 3.6.

Caveat The two values were derived on different bases. The 0.05 figure was set during the 2017 pesticide peer review of flurtamone, from a 90-day rat study with a no observed adverse effect level of 9.9 mg/kg bw/day, a standard factor of 100 and an extra factor of 2 for extrapolating from sub-chronic to long-term exposure.

The point of departure is a benchmark dose lower confidence limit of 8.6 mg per kg body weight per day, at a 20% benchmark response, for decreased thyroxine in adult male rats of the F1 generation in the extended one-generation reproductive toxicity study.

Caveat Benchmark dose limits for the other critical effects were higher — 18.6 mg/kg bw/day for hepatocyte hypertrophy and 41.1 mg/kg bw/day for the rabbit eye malformations. The thyroid endpoint was the most sensitive, so it set the value.

A standard uncertainty factor of 100 was applied for inter- and intra-species differences, together with an additional overall factor of 5 addressing two specific gaps: the absence of a long-term toxicity and carcinogenicity study, and possible developmental immunotoxicity, given that no functional immunotoxicity testing was carried out during the developmental phase.

Caveat The additional factor of 5 rests on expert judgement. EFSA graded the carcinogenicity gap as medium uncertainty and the developmental immunotoxicity gap as medium to high.

The available 52-week rat study did not reach the maximum tolerated dose, because the highest dose tested — 37.8 mg per kg body weight per day — was not properly selected. EFSA concluded it may have limited sensitivity to detect preneoplastic findings and does not satisfy the requirements of a full carcinogenicity assessment.

A dose-related decrease in the total number of splenic immune cells was seen in both sexes at all dose levels in the F1 generation of the extended one-generation study. The results were not statistically analysed, no functional immunotoxicity test or developmental cohort is available for TFA, and no consistent effects on immunological endpoints were seen in adult animals across the data package.

Caveat EFSA graded this uncertainty medium to high. It is an unfollowed-up signal, not a demonstrated effect — the finding is a cell count without a functional test behind it.

An acute reference dose was set for the first time, at 0.07 mg per kg body weight, derived from the same benchmark dose lower confidence limit of 8.6 mg/kg bw/day for decreased thyroxine, with a standard uncertainty factor of 100 and a conversion factor of 0.84 to express the result as TFA rather than as its sodium salt.

Caveat No additional factor of 5 was applied here — that factor addresses long-term data gaps, which do not bear on a single-exposure value. The reasoning is that a transient fall in thyroxine during a critical window of pregnancy may affect brain development.

On the weight of the available evidence, EFSA concluded that TFA and its sodium salt are unlikely to be genotoxic. All bacterial reverse mutation tests and all five mammalian cell gene mutation tests were negative, with or without metabolic activation.

Caveat A conclusion reached without in vivo testing. EFSA judged in vivo tests unnecessary given consistently negative in vitro results, in line with its own genotoxicity testing strategy.

The chromosomal damage tests did not all agree. One highly relevant chromosome aberration test using the acid was positive, while two in vitro micronucleus tests were negative and three studies using the sodium salt were negative. EFSA gave greater weight to the salt results, reasoning that both forms dissociate at physiological pH so exposure to the acid form is unlikely in the body, and that the acid produced highly variable cytotoxicity attributable to its strength as an acid.

Caveat The reasons for the inconsistency remain unclear. The point was raised in the public consultation and referred to EFSA’s cross-cutting working group on genotoxicity, which the TFA group then concurred with.

EFSA drew on previous assessments of pesticide active substances that form TFA and on the studies ECHA considered for the classification proposal, ran a public consultation that drew 177 comments, and worked through a group including Member State and academic experts.

The report contains no dietary exposure assessment. EFSA was asked to collect the evidence and, if possible, derive the guidance values — not to estimate how much TFA people actually consume. No occurrence data, no intake estimate and no comparison against the new acceptable daily intake appear anywhere in its 75 pages.

Caveat The terms of reference run to hazard identification and characterisation only. The contents pass from the guidance values directly to uncertainty analysis and conclusions.

EFSA does record one comparison: serum concentrations at the no- and lowest-observed-adverse-effect levels in the animal studies are several orders of magnitude higher than TFA levels reported in human populations, indicating a substantial margin of exposure between experimental doses and background human exposure.

Caveat This is a toxicokinetic comparison of concentrations in blood, not a dietary intake assessment. It does not tell you what proportion of the acceptable daily intake is used, nor how intake splits between drinking water and food.

The registration dossier reports an extended one-generation reproductive toxicity study in Han Wistar rats concluding no adverse effects on reproductive performance, fertility or offspring development, with a no observed adverse effect level of roughly 203 to 222 mg per kg body weight per day.

Caveat Registrant-supplied data. The dossier submitter drew on the same registration dossiers and reached a different conclusion — the disagreement is about interpretation, not access to data.

Directive (EU) 2026/805, amending the Water Framework, Groundwater and Environmental Quality Standards Directives, was adopted on 30 March 2026, published in the Official Journal on 20 April, and entered into force on 10 May 2026.

Caveat Some secondary sources give 11 May. The directive’s own delegation provision runs from 10 May 2026.

TFA is included in a sum of 25 PFAS with an environmental quality standard for surface water, at entry 65 of Annex I. At the next review the Commission is to consider establishing a separate standard for TFA.

The standards are an annual average of 0.0044 µg/L in both inland and other surface waters, and 0.077 µg/kg wet weight in biota, each expressed as a sum of PFOA equivalents. No maximum allowable concentration is set. The entry is flagged as a priority hazardous substance, as ubiquitous, persistent, bioaccumulative and toxic, and as tending to accumulate in sediment or biota.

Caveat Read from the Annex I table. The maximum allowable concentration columns read “not applicable”, so secondary coverage citing one is wrong.

The directive states that for this group the biota standard is the critical one and must be complied with, and that the annual average standards are not equivalently protective.

Caveat From footnote 27 to Annex I. The operative test is therefore concentration in fish, not concentration in water.

Within that sum, each substance carries a relative potency factor scaling it to PFOA. TFA — named by CAS number 76-05-1 — is assigned 0.002, the second lowest of the 25. The factors range from 10 for perfluorononanoic acid down to 0.001 for perfluorobutane sulfonic acid, so TFA counts for five hundred times less than PFOA and five thousand times less than the most heavily weighted substance in the group.

Caveat A potency weighting, not a measurement. It reflects relative toxicity per unit mass, and says nothing about how much TFA is present — which is far more than any other PFAS in the group.

The Commission’s original 2022 proposal covered 24 PFAS. TFA was added during negotiation, in the provisional Council and Parliament agreement of September 2025, making it 25.

Caveat Confirm from the legislative history rather than press coverage.

For groundwater, no TFA standard was set. The Commission is instead to consider establishing one at the next review — whether separate or within a sum — taking into account the most recent scientific knowledge on TFA, including work carried out by ECHA, EFSA and the World Health Organization.

That next review falls due by 11 May 2032, and every six years thereafter.

Member States must establish supplementary monitoring programmes by 22 December 2027 and preliminary programmes of measures by 22 December 2030, with final measures in the 2033 river basin management plans and good groundwater chemical status to be achieved by 22 December 2039.

Caveat Time extensions are available and limited to one further update of the river basin management plan except where natural conditions prevent achievement.

The directive gives ECHA a permanent role in prioritising substances for the watch lists and pollutant annexes, and in deriving quality standards, with its Risk Assessment and Socio-Economic Analysis Committees providing opinions.

EFSA states that, considering the toxicological properties of TFA, its presence above the parametric drinking water limit of 0.1 µg/L would have consequences for decision making under the approval criteria for pesticide active substances.

Caveat A procedural trigger within the groundwater metabolite relevance test, not an automatic ban.

The 2026 peer review of diflufenican records an outstanding data gap, relevant to all representative uses, for investigations sufficient to exclude that TFA is formed in soil and surface water systems and present as a residue in plants.

That peer review was approved in November 2025 without the Article 56 data, because the follow-up investigations triggered by the 2021 notification were still running. It cites the acceptable daily intake of 0.05 mg per kg body weight per day that EFSA replaced on 22 July 2026.

In summer 2025 Denmark withdrew 23 plant protection products based on six TFA-forming active substances: diflufenican, fluazinam, fluazifop-P-butyl, fluopyram, mefentrifluconazole and tau-fluvalinate.

Caveat Reported via the Dutch impact analysis. Confirm against the Danish EPA decision directly.

In December 2025 the Dutch authorisation board informed its ministry that it would reassess 46 PFAS pesticides based on six active substances, in light of the Danish evidence.

Caveat Described in an advocacy organisation’s legal filing. Confirm against the board’s own communication.

In June 2026 the Dutch parliament published an impact assessment of withdrawing the six TFA-forming actives, framing the trade-off between groundwater protection and crop protection dependence.

In June 2026 PAN-Netherlands filed a formal request under Article 44 of the plant protection products regulation, asking the Dutch board to withdraw authorisations for PFAS and TFA-forming products, citing the January 2021 and May 2024 Article 56 notifications as part of its toxicological basis.

Caveat An advocacy filing, not a decision. Attribute by name.

A joint EFSA and ECHA mandate on the fate and behaviour of TFA in soil and water, covering formation from biocides and pesticides and methods for predicting concentrations in groundwater and surface water, is due to report by June 2027.

Caveat A scientific assistance mandate, not a classification procedure.