Atmospheric Formation of Trifluoroacetic Acid (TFA) from Fluorinated Refrigerants: Current Evidence and Remaining Uncertainties

1. Introduction

The Kigali Amendment (2016), the EU Mobile Air Conditioning (MAC) Directive 2006/40/EC (global warming potential, GWP, < 150 in new vehicles) and the revised F-gas Regulation (EU) 2024/573 are phasing down hydrofluorocarbons (HFCs) such as HFC-134a in refrigeration and air conditioning. Their main replacements are hydrofluoroolefins (HFOs), hydrochlorofluoroolefins (HCFOs) and HFO/HFC blends, whose atmospheric lifetimes are days to weeks rather than years.

Many of these fluorinated gases (F-gases) carry a carbon-bound trifluoromethyl group (C–CF₃). Since the strong C–F bonds make the CF₃ group resistant to further oxidation, their atmospheric degradation can end in trifluoroacetic acid (TFA, CF₃C(O)OH) (Section 3, Figure 1), which is removed from the air within days by wet and dry deposition and is highly persistent and mobile in water and soil [1,2,3]. Climate metrics such as the GWP describe the radiative effect of the emitted gas, but not the fate of its fluorine, which for several of these molecules ends up as TFA deposited at the surface. TFA is now also on the regulatory agenda: the 2023 proposal to restrict per- and polyfluoroalkyl substances (PFAS) as a class under the Registration, Evaluation, Authorisation and Restriction of Chemicals (REACH) Regulation includes F-gases [4,5], and in June 2026 the Risk Assessment Committee (RAC) of the European Chemicals Agency (ECHA) recommended a harmonised classification for TFA as toxic for reproduction, persistent, mobile and toxic (PMT), and very persistent and very mobile (vPvM) [6].

Recent reviews and assessments have examined TFA mainly from the perspective of its environmental persistence, exposure and risk [1,2,3,7]. This short review takes an atmospheric perspective, focusing on the low-GWP F-gases that are replacing HFCs: how their chemistry determines whether TFA is formed (Section 3), how their emissions translate into TFA formation and deposition (Section 4), and what this implies for refrigerant choices, regulation and assessment metrics (Section 5). It closes with the questions that current evidence leaves open (Section 6).

2. Methodology

This review is a critical narrative synthesis rather than a systematic review. The evidence on TFA formation from F-gases spans laboratory kinetic and product studies, chemical transport modelling and field measurements, which differ too much in design and metrics for a systematic protocol such as PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) to be applied meaningfully. The literature was identified through searches of Scopus and through backward citation tracking and was complemented by regulatory documents and technical reports. Sources that have not been peer-reviewed are identified as such in the text.

3. F-Gases as TFA Precursors

The fraction of an F-gas converted to TFA depends on which carbonyl compound its oxidation produces, not on its fluorine content alone. When this carbonyl is CF₃C(O)F, the yield is essentially a property of the molecule; when it is CF₃CHO, it also depends on the atmospheric fate of the carbonyl. All the refrigerants considered here are removed mainly by reaction with the hydroxyl radical OH; for HFO-1234ze(E), for example, OH accounts for about 99.6% of the atmospheric loss [8]. In practice, the precursors fall into two groups, depending on whether oxidation yields trifluoroacetyl fluoride, CF₃C(O)F, or trifluoroacetaldehyde, CF₃CHO (Table 1, Figure 1).

Precursors forming CF₃C(O)F. HFC-134a is oxidised by H-atom abstraction to the CF₃CHFO radical, which either reacts with O₂ to give CF₃C(O)F or breaks apart; this branching, still uncertain, limits its conversion to TFA to about 7–20% [10,11]. In HFO-1234yf, OH adds to the double bond and the resulting radicals lead almost entirely to CF₃C(O)F [12]. CF₃C(O)F is not significantly photolysed in the troposphere; it is taken up by cloud water and hydrolysed to TFA within days [12,15,16]. A recent theoretical study questioned whether the previously proposed gas-phase mechanism can form CF₃C(O)F and TFA from HFC-134a under atmospheric conditions [17]; a subsequent comment attributed this result to the omission of key reaction steps, including the hydrolysis of CF₃C(O)F, and reaffirmed a conversion to TFA of about 7–20% [18].

Precursors forming CF₃CHO. HFO-1234ze(E), HCFO-1233zd(E) and HFO-1336mzz(Z) are oxidised to CF₃CHO with near-unit yield [8,14,19]. The fate of CF₃CHO then determines how much TFA is formed. Photolysis does not give TFA, whereas reaction with OH leads, through the trifluoroacetylperoxy radical (CF₃C(O)O₂), to TFA [20]; degradation of trifluoroacetyl peroxynitrate (CF₃C(O)O₂NO₂) in the presence of water may add a further minor route [21]. Model studies that treat CF₃CHO as removed mainly by photolysis obtain a low conversion to TFA, for example about 2% for HCFO-1233zd(E) despite a lifetime of about one month [14]. A recent global model study of HFO-1234ze(E) that includes wet and dry deposition of CF₃CHO finds instead that, under upper-bound assumptions for its solubility, deposition may account for about half of its loss, photolysis for one third and reaction with OH for about 15%; if the wet-deposited CF₃CHO were fully converted, current emissions of about 15 Gg/yr could form up to 4.5 Gg/yr of TFA, depending on the assumed solubility [8], corresponding to a conversion of up to about 30% (upper bound rather than a central estimate). The conversion to TFA of CF₃CHO-forming precursors is therefore uncertain and depends on aqueous-phase processes that remain poorly constrained. Figure 1 summarises these pathways, including CF₃CHO photolysis, hydration, uptake and deposition, HFC-23 formation and the deposition of TFA; ozonolysis, a minor sink of these olefins [8], is not shown.

By-products. Photolysis of CF₃CHO produces small amounts of fluoroform (CHF₃, HFC-23), a very potent and long-lived greenhouse gas [22], and ozonolysis of HFOs and HCFOs can yield HFC-23, PFC-14 and CFC-13 [23]. In a box-model study, the HFC-23 channel accounted for 1.3–4.4% of CF₃CHO removal, depending on the assumed photolysis quantum yield [24]; quantum yields since measured at atmospheric pressure are lower [22]. Because HFC-23 has a 100-year GWP (GWP₁₀₀) of 14,600 [9], even such small conversions raise the climate impact of CF₃CHO-forming precursors well above their direct GWP [25]: a global model study puts the total (direct plus indirect) GWP₁₀₀ of HFO-1234ze(E) at about 11, compared with a direct value of 1.37 [9], of which about 8 is due to HFC-23 formed from CF₃CHO [8].

Blends. In stationary refrigeration and air conditioning, HFOs are mostly used in blends with HFCs. The amount of TFA a blend can form depends on the mass fraction of each component and on its conversion to TFA (Table 1). Assuming that the whole charge is eventually emitted and oxidised, and because HFO-1234yf and TFA have almost the same molar mass, the mass of TFA that a blend based on HFO-1234yf can form is roughly proportional to its HFO-1234yf mass fraction. Blends with the same components can therefore differ markedly: HFO-1234yf makes up 78.5% of R-454C but 31.1% of R-454B [26], while HFC-32, the other component of both, contains no CF₃ group and is not a TFA precursor. Servicing extends the use of these blends to existing equipment. Under Article 13 of Regulation (EU) 2024/573, F-gases with a GWP of 2500 or more may no longer be used to service refrigeration equipment from 2025, except for reclaimed or recycled gases until 2030, and the threshold falls to 750 for stationary refrigeration other than chillers from 2032. Existing systems can be retrofitted with HFC/HFO blends such as R-448A and R-449A, which combine 20–25% HFO-1234yf with 21–26% HFC-134a by mass [26]; leakage during operation and servicing of the installed bank therefore remains a diffuse and continuing source of TFA precursors.

4. From Emission to Deposition

For a given precursor, TFA formation scales with the mass emitted, while where it is deposited depends on the spatial distribution of emissions, OH levels, cloud and precipitation regimes; emissions, and hence how F-gases are used, are as important as their chemistry. Charge size, leakage, servicing and end-of-life recovery determine emissions. Mobile air conditioning is the main source of HFO-1234yf in Europe; stationary systems use HFOs mainly in blends; and HFO-1234ze(E) has been proposed as a propellant for pressurised metered-dose inhalers (pMDIs) [27]. Atmospheric observations show that emissions are already significant: in north-western Europe they reached about 1.50 Gg/yr for HFO-1234yf, 0.96 Gg/yr for HFO-1234ze(E) and 1.04 Gg/yr for HCFO-1233zd(E) in 2023 [28].

Chemical transport models translate emissions into TFA formation and deposition at global and regional scales; Table 2 summarises selected studies, their scenarios and main results.

These results depend on assumptions that differ between studies: emission inventories and scenarios, the conversion of each precursor to TFA, the treatment of CF₃CHO, the representation of clouds, aerosols and dry and wet deposition, and the meteorological year. Direct measurements of TFA in air, which could validate them, remain scarce (Section 6). Despite variations among studies, they collectively indicate that the deposition of TFA from HFO-1234yf corresponds to emitted mass, with about 60–70% deposited outside the emitting region in Europe and 86–89% in East Asia, while areas with low rainfall exhibit the highest concentrations in rainwater. [36,37]. In the European study (full conversion of the 2020 car fleet, single meteorological year), the higher-resolution model placed the deposition maximum over the Po Valley, where emission hot spots coincide with high dry deposition; the two models gave European maxima of 1.9–2.5 kg/km²/yr [30]. Although based on a single scenario, this result makes Northern Italy a priority region for national emission estimates and deposition measurements.

A projection of the European passenger-car fleet, published as a white paper by a supplier of CO₂ and propane air-conditioning systems and not peer-reviewed, estimates that annual TFA formation will keep rising until about 2036 without further restrictions, and notes that larger refrigerant charges in heat pumps of battery electric vehicles may make this an underestimate [32]. The direction of this trend is consistent with peer-reviewed studies showing that HFO-1234yf may already exceed HFC-134a as a source of TFA over Europe [31] and that emissions will increase substantially once the fleet has fully converted [30]. The installed bank of HFC-134a will also keep producing TFA for years after its phase-down.

Substituting HFC-134a with HFO-1234yf alters the formation and deposition patterns of TFA, as the former disperses globally before degradation while the latter is transformed within days, resulting in TFA being generated closer to its sources and concentrated in deposition areas at a continental scale rather than hemispheric, especially near emission hotspots and their downwind regions. [30,33,36]. Once formed, TFA has a tropospheric lifetime of only a few days [29]; how it partitions between the gas phase, cloud water and particles controls how far it travels and whether it is removed by rain or by dry deposition [38,39].

The overall monitoring of emissions from precursors to TFA deposition remains sparse. The Advanced Global Atmospheric Gases Experiment (AGAGE) network now measures precursor levels in ambient air [28,40], and between 2021 and 2023, TFA concentrations in Swiss precipitation have surged four to six times since 1996/1997, reaching mean values of 0.30–0.96 µg/L across 14 monitoring sites [41]. At the sites of Swiss Plateau, modelled degradation of known precursors explains about 63% of the observed deposition, 48% from HFOs alone, and leaves more than one third unexplained, with summertime deposition underestimated about five-fold [41]. This gap may reflect unaccounted sources, underestimated emissions, or uncertainties in atmospheric processing and deposition, and cannot be attributed to F-gases without further evidence. Where the atmospheric sources of TFA have been examined, the picture varies: HFC-134a alone could explain 56–90% of rainwater TFA in Guangzhou [42], whereas in Beijing it accounted for only 14% (6–33%) of TFA deposition [43], a share later revised to about 6% (2–13%) with an updated TFA yield [44]. Other sources, such contr4ibute to atmospheric TFA, such as inhalation anaesthetics [45] and fluoropolymer thermolysis, as found to contribute little to rainwater TFA [46], and early studies already found TFA levels difficult to explain with the degradation of HFCs and hydrochlorofluorocarbons (HCFCs) alone [47]. Direct measurements of TFA in air, in the gas and particulate phases, remain scarce [39,43].

5. Implications for Use

5.1. Refrigerant Choices

Non-fluorinated refrigerants do not form TFA, but they involve other trade-offs. In mobile air conditioning, CO₂ heat-pump systems are already in series production, while propane systems are under development [32]; alternatives for hybrid and combustion-engine vehicles are less mature. In stationary applications, CO₂, ammonia and hydrocarbons are established options; a recent review concludes that only slight technical barriers remain for most new systems, with possible trade-offs in initial cost, operating pressure or capacity, but not in energy efficiency [48]. Where F-gases remain in use, TFA formation and by-products are relevant alongside GWP when comparing them (Section 3).

5.2. Regulation

Regulation of TFA-forming F-gases has so far been driven by ozone and climate objectives rather than by TFA. The Montreal Protocol, the MAC Directive and the F-gas Regulation select refrigerants on the basis of their ozone depletion potential and GWP, and the GWP limit of the MAC Directive was the main driver of the shift to HFO-1234yf in Europe. Only recently have instruments begun to address TFA directly: the proposed REACH restriction covers F-gases as PFAS, and the harmonised classification of TFA would formally recognise its hazards, although classification does not by itself limit emissions (Table 3).

In the same industry projection, cumulative atmospheric TFA formation over 2028–2058 ranges from about 102 Gg with a full restriction to 143 Gg with time-limited derogations and 240 Gg without restriction [32]; these figures illustrate the effect of timing rather than providing quantitative estimates. Because TFA from HFO-1234yf is formed within days of emission, reductions in emissions would lower TFA formation almost immediately, although part of the benefit would be felt downwind of the regions that reduce emissions. Outside the EU, approaches diverge. The structural definition of PFAS adopted by the US EPA for reporting under the Toxic Substances Control Act does not appear to capture TFA or most F-gases used as refrigerants, which lack the required fluorinated substructures [51]. By contrast, Maine and Minnesota define PFAS as fluorinated organic chemicals containing at least one fully fluorinated carbon atom, a definition that includes TFA and its precursors; Maine will prohibit the sale of refrigerants containing intentionally added PFAS from 2040 [52], and Minnesota prohibits products with intentionally added PFAS from 2032 unless their use is deemed currently unavoidable [53].

5.3. Assessment Metrics

Refrigerant selection is influenced by the GWP of the parent molecule, while life-cycle metrics like total equivalent warming impact (TEWI) and life-cycle climate performance (LCCP) account for leakage and energy use. reflect the atmospheric behaviour of refrigerants only in part: the degradation-product term is small for HFO-1234yf and is set to zero when data are lacking [54], and it could at most capture climatically active by-products such as HFC-23. The widely used LCCP analysis of HFO-1234yf dates back to 2009 [55], and more recent assessments, for instance of heat pumps in electric vehicles, still consider climate impacts only [54]. Non-climatic consequences, such as the formation and deposition of persistent TFA, therefore remain outside these metrics.

6. Open Questions

For HFO-1234yf the picture is fairly clear, but for CF₃CHO-forming precursors, and for the TFA already observed in the environment, several questions remain open. The first concerns the fate of CF₃CHO. For the short-lived HFOs and HCFOs that form it, the conversion to TFA depends on how CF₃CHO partitions between photolysis, reaction with OH, and uptake into cloud water, aerosol and surfaces followed by deposition, and on how much HFC-23 it forms. These branchings remain poorly constrained.

A second set of questions concerns the TFA already observed. Refrigerants, anaesthetics and other fluorinated sources overlap in time and space, and attribution to F-gases rests on few studies: in Switzerland, known precursors explain about 63% of the observed deposition, leaving more than one third unexplained [41]. Where TFA is deposited, and by which route, is also uncertain, since modelled deposition patterns differ between studies and the dry deposition and gas/particle partitioning of TFA are poorly constrained. Measurements of TFA in air are sparse and not harmonised, and national emission estimates are lacking; harmonised methods for gas- and particle-phase TFA, deposition networks and inverse modelling of precursor emissions would allow model results to be tested against observations.

Finally, HFO-1234yf may already exceed HFC-134a as a source of TFA over Europe [31], and its contribution is likely to keep increasing as the vehicle fleet converts to it [30,31]. The largest atmospheric source of TFA may therefore peak before any restriction takes full effect.

7. Conclusions

Low-GWP F-gases reduce the direct climate impact of refrigerants, but several of them are efficient atmospheric precursors of TFA. For HFO-1234yf, whose OH-initiated oxidation leads near-quantitatively to CF₃C(O)F and hence to TFA, the atmospheric consequence is a rapid formation of TFA within days of emission, deposited at continental rather than global scale and concentrated in and downwind of emission regions. For CF₃CHO-forming molecules, the conversion to TFA is lower but uncertain, and a substantial part of their climate impact may come from by-products such as HFC-23. In both cases, the metrics currently used to select refrigerants do not account for their atmospheric fate, and the open questions above would need to be addressed to include it. These findings support closer monitoring and the prevention of emissions of TFA-forming F-gases, both in view of the progressively stricter requirements of Regulation (EU) 2024/573 and in line with the precautionary and polluter-pays principles on which EU environmental policy is based (Article 191(2) TFEU). In Italy, the same principles underpin a joint call by water utilities and environmental and consumer associations for a rapid and complete restriction of PFAS, with strictly time-limited exemptions [56].