Increasing and Widespread Deposition of Trifluoroacetic Acid from Global Emissions of HFO-1234yf in Mobile-Air Conditioning

Hart, Lucy; Hossaini, Ryan; Wild, Oliver; Halsall, Crispin; Vollmer, Martin K.; Sherry, David

doi:10.5194/egusphere-2026-4613

Preprints

https://doi.org/10.5194/egusphere-2026-4613

Preprints

07 Aug 2026

| 07 Aug 2026

Increasing and Widespread Deposition of Trifluoroacetic Acid from Global Emissions of HFO-1234yf in Mobile-Air Conditioning

Lucy Hart, Ryan Hossaini, Oliver Wild, Crispin Halsall, Martin K. Vollmer, and David Sherry

Abstract. HFO-1234yf is an emerging low global warming potential refrigerant with increasing emissions, especially from mobile air-conditioning (MACs), that are poorly constrained. It is oxidised in the atmosphere to form trifluoroacetic acid (TFA), a persistent, mobile, and widespread pollutant. Despite growing concerns around TFA accumulation in the environment, the contribution of HFO-1234yf to its global deposition is not known. Here, we present the first global bottom-up emissions inventory for HFO-1234yf from passenger vehicles. We use a chemical transport model to calculate the global TFA budget from these emissions and quantify source–receptor relationships from idealised emission scenarios. The magnitude and trend in HFO-1234yf emissions from North-West Europe are in close agreement with top-down assessments. Using our derived emissions, the model reproduces surface HFO-1234yf observations from the AGAGE network reasonably well. A sensitivity study using estimates of fugitive emissions from HFO-1234yf production improves agreement, particularly at Gosan, South Korea, where the model underestimates observations. We find that global TFA deposition from HFO-1234yf in MACs has increased from 0.10 (0.05–0.17) Gg yr^-1 in 2014 to 3.7 (1.6–6.4) Gg yr^-1 in 2024, with 75 % of cumulative HFO-1234yf emissions coming from Europe. A large proportion of TFA from HFO-1234yf (57 %) is deposited in oceans. Source–receptor analysis shows that most (86–97 %) TFA deposition occurs outside of the original HFO-1234yf emission region. All regions except Europe are net atmospheric importers of TFA, underscoring the importance of international efforts to reduce future TFA pollution.

Received: 31 Jul 2026 – Discussion started: 07 Aug 2026

Publisher's note: Copernicus Publications remains neutral with regard to jurisdictional claims made in the text, published maps, institutional affiliations, or any other geographical representation in this paper. While Copernicus Publications makes every effort to include appropriate place names, the final responsibility lies with the authors. Views expressed in the text are those of the authors and do not necessarily reflect the views of the publisher.

Download & links

Preprint (PDF, 2147 KB)

Supplement (1665 KB)

Download & links

Lucy Hart, Ryan Hossaini, Oliver Wild, Crispin Halsall, Martin K. Vollmer, and David Sherry

Status: final response (author comments only)

RC1:
'Comment on egusphere-2026-4613', Anonymous Referee #1, 01 Sep 2026
This manuscript, 'Increasing and Widespread Deposition of Trifluoroacetic Acid from Global Emissions of HFO-1234yf in Mobile-Air Conditioning' by Hart and co-authors, presents the first global, country resolved bottom-up emissions inventory for HFO-1234yf from passenger vehicle mobile air conditioning, and uses a global chemical transport model to quantify the resulting TFA deposition and its source receptor relationship. The inventory is a genuine contribution and is well validated against an independent top down estimate for North West Europe and against AGAGE surface observations. The topic is timely and within ACPs scope, and the work is an advance on the existing literature. I consider it suitable for publication after the points below are addressed.

Specific comments:
Operational emissions account for 98% of the inventory (line 277), so the headline emissions and all downstream deposition estimates are strongly dependent on the choice of EF_OP = 4 (2-6)% /yr. The authors acknowledgethat this parameter is not well constrained, and the low/middle/high scenario design is a sensible way to propagate the uncertainty. However, the derivation of the central value and its range (lines 164-169) is currently quite brief, relying on 1990s-2000s estimates of 2-10%/yr and a statement that leakage rates have since halved. I think that the manuscript could be improved if the authors a) set out the reasoning behind the central 4% more explicitly, b) comment on whether a single global EF_OP is appropriate, and c) discuss the choice of 6% as the upper bound vs the 10%/yr used in the German national inventory (line 292). If the authors consider 10% unrealistic, a justification should be included.

The high OH bias (CH4 lifetime of 10.2 yr versus 11.2 ± 1.3 yr, lines 248-253) is handled through the BASE-OH sensitivity run. Since the correction lengthens the HFO-1234yf lifetime by ~20% (line 518) and the time-varying simulations in Sect. 4.3 use the uncorrected field, it would be useful to state briefly in the main text whether the regional deposition shares reported there (56%, 43%, 39%) are affected.

The model is run at 2.8°x 2.8 ° and the difficulties this creates for comparison at coastal and mountain regions are discussed in lines 339-345. A similar discussion should be included for the SR analysis. The partition between within-region and extra-regional deposition is a central result of this study, and for a species with a lifetime of days to weeks, numerical mixing through large grid boxes may dilute near-source concentrations and affect transport across regional boundaries, potentially altering the within-region and extra-regional deposition fractions. Please comment on the likely direction of resolution effects on the SR partition, and acknowledge this as a source of uncertainty in the within-region and extra-regional deposition fractions reported in Sect 5.1.

Fig S4 is hard to compare with Fig 5, so it is not very useful for judging the sensitivity runs. The two figures use different units (g kg^-1 in Fig 5, kg kg^-1 in S4) and the rows run in opposite order. Please present S4 in the same format as Fig. 5 so that the two can be read side by side. I also noticed that the PAN row in panel (e) is identical to the NAM row – please check whether this is a plotting error.

Technical corrections:
Lines 175, 365 and 551: HFO-1234yf is written as ‘HFO-123yf’.

Line 279: ‘Morrocco’ should be changed to ‘Morocco’

Inconsistent SAF definition – on line 220 SAF is defined as Sub-Saharan Africa, but on line 399 it is defined as South Africa.

Line 364: 2014 lower bound (0.078) is inconsistent with the one given in the abstract (0.05)

Lines 157, 260: Both equations are labelled (9)

Lines 503, 511 and 520: The acronym ‘DIFR’ doesn't appear to be defined, please clarify whether it is a typo for DFFR/DTFR or a separate metric, and if the latter, include a definition.
Citation: https://doi.org/10.5194/egusphere-2026-4613-RC1
RC2:
'Comment on egusphere-2026-4613', Anonymous Referee #2, 27 Sep 2026
Review of egusphere-2026-4613
The manuscript describes the development of a global HFO-1234yf emission inventory from mobile air conditioning and its application in a global chemistry and transport model to simulate TFA production and deposition on the global scale. The manuscript puts a focus on constructing so-called source-receptor relationships between different emission regions and highlights the role of Europe as the current main exporter of TFA. The study is of relevance because only few global simulations of TFA formation/deposition carried out with even fewer model systems exist. However, there are several shortcomings that need to be addressed before publication.
Major comments
Validation/discussion of obtained deposition values against observations: In section 4.2 the manuscript presents comparisons of simulated HFO-1234yf concentrations against atmospheric observations. However, there is very little or no discussion of the absolute values of simulated TFA deposition rates. These are shown in Fig. 4 and briefly mentioned but not put into any context with other modelling or observational studies. Comparisons versus observed and simulated values presented in Khan et al. 2026, Freeling et al. 2020, Henne et al. 2026 and elsewhere would be helpful to support the suitability of the model system to reasonably describe the transformation to and deposition of TFA.

Source-receptor attribution seems to contradict previous findings. The current study reports very small fractions of "domestic deposition". Most TFA formed from HFO-1234yf emissions in a given region is deposited elsewhere. This is not in line with other studies and given the short lifetime of HFO-1234yf a bit surprising. Further discussion and/or re-evaluation of model results is required (see details below).

Detailed comments
Equation 2 and others: When first reading I stumbled over the question what values were used for the emission factors, here EF_FF? It would be good to mention Section 2.6 and Table 1 from the beginning, hence it is clear where to look for these values.
Equation 4: I don't understand how the equation works and what the meaning of Z and Deff really is. In my simple mind, I would have thought that you would have reports of charges in each vehicle class and also the number of vehicles in each class. Together this could be used to construct a weighted mean and the spread in each class could also be propagated to the total spread. 25 % sounds like a rather large uncertainty for the mean. A bit of additional explanation of the equation would be helpful.
L 115f: The 539 g are also in close agreement with the assumptions made by Papasavva et al. 2009 and Henne et al., 2012, when constructing HFO-1234yf inventories.
Table 1: In current HFC emission reporting to UNFCCC EF_OP varies strongly between different countries (e.g. UK 3.6 %, Germany 11 % in 2024 reporting tables). Strangely bottom-up estimates in both countries agree fairly well with top-down estimates (DeLongueville et al., 2026), which makes the representativeness of this emission factor a bit tricky to understand. I suppose the range is reflected in your different scenarios. However, the ranges slightly differ. Given the discussion of improvements with newer vehicles, it might also have been reasonable to assume different values/ranges for different world regions. Maybe this could be commented on.
Equation 6: I suppose the N in equation 6 is not the same as in equation 3 so the penetration rate is not needed, right? But then it should be the number of cars sold that are using HFO-1234yf, which for most areas is not the same as the total number of cars sold. Please clarify and change the variable name (N) to something else, so it is not confused with the N from equation 3.
L 145: EF_rec: Shouldn't this be the fraction NOT recovered when you are calculating the emissions to the atmosphere?
Table 2: Could you comment on the fact that Europe is the main consumer of HFO-1234yf but there seems to be no production. A bit surprising.
Section 3: I suggest an alternative section title "Atmospheric model and methods" (or similar) to distinguish it from the previous section which was 'bottom-up model'.
Section 3.1: If I understand correctly, you are using a full chemistry simulation and not a prescribed OH field. Please mention what chemistry scheme is used and if you can how the simulated OH compares to other studies. Are all the used emissions time variable or only those of HFO-1234yf?
L197f: Here, ozonolysis of HFO-1234yf was not treated because it is considered of minor importance. What about chlorine radicals? Previous studies (e.g., Luecken et al. 2010) considered the chlorine pathway in addition to OH and reported on contribution in the order of 10 % of HFO-1234yf degradation through Cl radicals.
Section 3.2: Somehow, I feel that the part describing the HFO runs should go to the previous sub-section and the current sub-section should focus on the long-lived compounds.
L 204: I am getting confused with the simulation periods. Maybe the BASE run should be included in Table 3 as well. Are the BASE simulations using variable meteorology and emissions from 2014 to 2024? What was the rational behind choosing 2017 for the sensitivity simulations. Wouldn't a later year be more relevant because of a deeper penetration of HFO usage in more world regions?
L 206: Please give the rational for using traffic NOx emissions as a proxy. Did you assume constant HFO emissions throughout the year? Some studies have suggested seasonal variable emissions from MACs depending on usage (summer vs. winter).
L 211: Why was HFC-125 not included in this list? Assumed low TFA yield?
L 211: How do the HCFC/HFC simulations compare to direct atmospheric observations? I did not find such a direct comparison in the previous publication.
Section 3.3: It is unclear for which year the sensitivity study was run? This is relevant because of the different timing of market introduction in the different regions.
L233ff: This and the following paragraphs would fit better to the previous section. It has little to do with the source-receptor calculations. Or it should become its own subsection (Sensitivity runs)?
L252f: What about HFO-1234yf lifetimes? Are they affected in the same way and how to they compare to previous estimates? Later (L412) an average lifetime of 18 days for European HFO-2134yf is mentioned. This seems rather large compared to other studies (e.g., Khan et al. 2026 give 12 days as global average). What is the global average lifetime in your simulations?
L273: "… banned the use of refrigerants …" This is not sufficiently specific. High GWP refrigerants were banned for all vehicle type approvals after 2017. So, the existing fleet was still using HFC-134a and even new vehicles that received their type approval before 2017 were still delivered with HFC-134a. Similar in L 290: "full substitution".
L285: Please mention which countries are part of NW-Europe.
L 298f: The report by Warncke and Gschrey mentions ~70 t from all other uses of HFO-1234yf compared to 1298 t from MAC in Germany for 2024. This supports your assumption of minor contributions and could be cited.
Figure 2: There is no discussion of regions other than NW-Europe and Germany in the text. Either mention/discuss the other regions in the text as well or drop because there seems little additional information.
L318: The underestimation in the model is much more pronounced for the two high altitude sides JFJ and CMN. Could this be related to the height at which the model output was sampled? At the coarse horizontal resolution, it is difficult to represent mountain sites properly. Usually sampling the model output at a much lower altitude above sea level than the station altitude is required. I suggest mentioning this problem. For coastal sites (like MHD) the coarse resolution may cause similar problems in that coastal wind regimes cannot be captured. The good agreement for TAC and TOB seems to be the best indicator that European emissions are well captured by the inventory as these sites are close to emissions. However, the close proximity may also introduce representativeness problems as well at the given grid resolution. A similar comparison to AGAGE measurements is also part of Henne et al. 2026 and could be compared to as well.
L319: The other obvious problem would be inconsistencies in the MAC emission inventory due to wrong assumptions about market penetration and leakage rates in different areas.
L374: For which years are these regional contributions given. Henne et al. 2026 report HFOs (upscaled from explicit simulations of HFO-1234yf) to be responsible for 40 to 50 % of observed TFA deposition in Switzerland for the years 2021-2023, whereas HFC/HCFCs contributed only 10 to 20 %. So, in their case HFOs seem much more dominant as reported here.
Figures 5 and 6: If I read Fig. 5 correctly your results suggest that only 5 % of European emissions are deposited in Europe again, right? Comparing to Fig 6 I have a hard time believing this number. Also, according to Figure 5, a larger fraction of European emissions 7 % is deposited in North America than in Europe itself. Again, hard to believe from the figure and the relatively short lifetimes. It's clear that Europe is rather a small area and you crop to some kind of land mask that is potentially influenced by the coarse model resolution. Maybe you could show the regions in figure 1 at the resolution of the model, which would probably look quite different. The presented findings of very small domestic deposition from European emissions are in contradiction to previous studies. Henne et al. 2012 using two different transport models estimated a domestic European fraction of 34 - 42 %. Wang et al. 2018 give 31 % for European domestic TFA deposition. Khan et al. 2026, who (using one of the models already used in Henne et al. 2012) come to similar numbers: "We found 31%, 32%, and 30% TFA deposition from HFO-1234yf for separate North America, Europe and Asia emission scenarios." There is either a misunderstanding in how to interpret the numbers in the current study or something fundamentally different in the attribution that needs to be analysed and discussed.
Figure 5: PAN is really a strange abbreviation for Australia and New Zealand.
Figure 6: Deposition distribution in Luecken et al for NAM looks quite different. Please comment.
L411: "transatlantic easterly flows": These could happen over the southern North Atlantic in the tropical easterlies with short atmospheric lifetimes followed by rapid washout in convection. I don't think this can be a relevant pathway for European TFA. In the northern North-Atlantic direct easterly transport is very unlikely!
L416f: I disagree, the comparison to other studies should still be valid even if somewhat different region definitions are used. Currently, there is an order of magnitude difference between the domestic fractions reported here and in previous studies. This cannot just be because of different region definitions.
L421f: Emissions from these regions have not been overlooked but considered minor given the absence of HFC regulation in most of the southern hemisphere. I am surprised that your bottom-up inventory results in considerable emissions in these regions. Could you provide a table with the emissions by region? I don't think there is one included in the main text or the supplement, yet. The first sentence in section 4.1 suggests that it is in the order of 1 % for the southern hemisphere, which would prove my point of rather negligible contributions to global TFA formation.
L439: Maybe your point needs to be spelled out more clearly. Historically, the ice core or firn record was representing global TFA production because TFA was mostly produced from long-lived compounds. With the switch to short-lived TFA precursors, this global representativeness will be lost especially when different regions develop with different speed.
Table 4: Again, these numbers surprise. If Europe is the largest source of HFO (68 % globally in 2024) and we have relatively short lifetimes, how can 63 % of the deposition in Europe be imported?
L477: "European emissions reaches all receptor regions". I find this statement misleading. A compound that has an atmospheric lifetime of 18 days (your value) will not make it from the northern hemisphere midlatitudes to Australia or Antarctica. At least not in any significant amount. That's also what we see in Fig 7 where the contributions to PAN and ANT are well below 1 %.
Table 5 and Fig 7: The numbers of Table 5 and Figure 7 don't seem to align. In Table 5 Europe exports 3600 t of TFA, but if I sum the numbers in the lowest column of Fig 7 I get something closer to 10000 t.

Technical comments
L 205: Missing word:" … within 'each' country …"
Caption Figure 6: Twice "of TFA per Mg of HFO-1234yf emitted".
Citation: https://doi.org/10.5194/egusphere-2026-4613-RC2

Lucy Hart, Ryan Hossaini, Oliver Wild, Crispin Halsall, Martin K. Vollmer, and David Sherry

Supplement

https://doi.org/10.5194/egusphere-2026-4613-supplement

Lucy Hart, Ryan Hossaini, Oliver Wild, Crispin Halsall, Martin K. Vollmer, and David Sherry

Viewed

Total article views: 440 (including HTML, PDF, and XML)

HTML	PDF	XML	Total	Supplement	BibTeX	EndNote
249	166	25	440	47	28	27

HTML: 249
PDF: 166
XML: 25
Total: 440
Supplement: 47
BibTeX: 28
EndNote: 27

Views and downloads (calculated since 07 Aug 2026)

Month	HTML	PDF	XML	Total
Aug 2026	94	45	2	141
Sep 2026	134	99	19	252
Oct 2026	21	22	4	47

Cumulative views and downloads (calculated since 07 Aug 2026)

Month	HTML	PDF	XML	Total
Aug 2026	94	45	2	141
Sep 2026	134	99	19	252
Oct 2026	21	22	4	47

Viewed (geographical distribution)

Total article views: 396 (including HTML, PDF, and XML) Thereof 396 with geography defined and 0 with unknown origin.

Country	#	Views	%

Latest update: 07 Oct 2026

Short summary

HFO-1234yf is increasingly used in cars because of its low climate impact but once emitted to the atmosphere it forms trifluoroacetic acid (TFA), a chemical that remains on Earth permanently. We present the first global emission inventory of HFO-1234yf from cars and compare our results to atmospheric measurements. Using an atmospheric chemistry model, we found that TFA produced from this source is rapidly growing and most TFA is deposited outside of the region HFO-1234yf was emitted.


Total:	0
HTML:	0
PDF:	0
XML:	0