Kinetics of the reaction of CF₃CHO with OH between 204 K and 361 K

Abstract

Trifluoroacetaldehyde (CF₃CHO) is formed in the atmosphere by the oxidation of a number of fluorinated, organic compounds of anthropogenic origin. The reaction of CF₃CHO with the OH radical is a potential source of atmospheric trifluoroacetic acid (TFA) which is a highly persistent, water-soluble compound that may accumulate in aquatic ecosystems and for which uncertainty about its sources, fate, and potential ecological impact persists. In light of growing concerns about the impact of TFA, we present the first study of the temperature dependence of the rate coefficient for the title reaction over the atmospherically relevant temperature range of 204 K to 361 K. Rate coefficients were determined using pulsed laser photolysis–pulsed laser induced fluorescence (PLP–PLIF) and direct concentration measurements via Fourier Transform Infrared (FTIR) spectroscopy, as well as relative rate experiments in an atmospheric simulation chamber using ethane (C₂H₆) as a reference compound. The rate coefficient (k₁) obtained with PLP–PLIF at room temperature is (5.8 ± 0.5) × 10⁻¹³ cm³ molecule⁻¹ s⁻¹. The temperature dependence is described by the expression k₁(T) = (3.8 ± 0.2) × 10⁻¹³ × (T/300)² × exp[(131 ± 16)/T]. The relative-rate experiments showed that the rate coefficient obtained can be significantly biased by reactions of the CF₃O radical with CF₃CHO and/or C₂H₆ and also reactions of CF₃CHO with HO₂. Based on the expression of k₁ given above, the lifetime of CF₃CHO with respect to reaction with the OH radical varies from 22 days at the surface (T ∼ 300 K) to 30 days in the upper troposphere (T ∼ 220 K).

---

1. Introduction

Fluorinated chemicals are produced at a worldwide annual rate of multiple million tons, with a rising trend. Owing to their chemical stability and unique physicochemical properties,⁸ they are widely used as refrigerants, propellants, specialty solvents, and high-performance lubricants. It is estimated that ∼20% of marketed drugs and ∼50% of agrichemicals contain at least one F-atom underlining their importance.^9,10 The generation and usage of fluorinated trace gases including hydrochlorofluorocarbons (HCFCs), hydrofluorocarbons (HFCs), fluorotelomer alcohols and hydrofluoroolefins (HFOs)^11–13 results in their release to the atmosphere where they may undergo chemical transformation. While HCFCs and HFCs are long-lived (up to decades) and are transported to the stratosphere, HFOs have lifetimes of the order of days to weeks and are degraded in the lower atmosphere (troposphere). HFOs are of topical interest as they are gaining importance as replacements for traditional refrigerants such as chlorofluorocarbons (CFCs) and hydrofluorocarbons (HFCs)^14,15 and are now widely used in a variety of industrial applications, including refrigerants, foam blowing agents and propellants in medical devices.^16–18

Fluorinated chemicals have come under scrutiny owing to their potential risk to the environment as greenhouse gases or precursors to toxic and persistent chemicals. CF₃CHO, the object of this study, is one of the products that can result from the breakdown of some fluorinated chemicals in the atmosphere.

The atmospheric degradation of CF₃CHO proceeds via both photolysis and reaction with OH radicals. Based on the available literature, photolysis is expected to contribute ∼80% and reaction with OH ∼20%.¹⁹ However, there is considerable uncertainty associated with the relative contribution of these loss pathways at e.g. different altitudes, as temperature-dependent rate coefficients for (R1) or temperature dependent quantum yields for (R2) have not been published (IUPAC, 2025).⁵ The reaction with OH is thought to proceed entirely via abstraction of the aldehydic H-atom to generate the trifluoro acetyl radical (CF₃CO) as written in (R1). The predominant products of photodissociation at actinic wavelengths (λ > 300 nm) are CF₃ and HCO (R2).

CF₃CHO + OH → CF₃CO + H₂O (R1)

CF₃CHO + hν → CF₃ + HCO (R2)

The CF₃CO product of (R1), will react with O₂ in air to form the CF₃C(O)O₂ radical (R3), which, via subsequent reaction with atmospheric peroxy radicals or reactive nitrogen trace-gases represents a pathway to the formation of stable and semi-stable trace-gases such as CF₃C(O)OH (TFA, (R4a)), CF₃C(O)OOH (trifluoroperacetic acid, (R4b)) and CF₃C(O)O₂NO₂ (trifluoroperoxyacetyl nitric anhydride, FPAN, (R5)).⁵

CF₃CO + O₂ + M → CF₃C(O)O₂ + M (R3)

CF₃C(O)O₂ + HO₂ → CF₃C(O)OH + O₃ (R4a)

→ CF₃C(O)OOH + O₂ (R4b)

→ CF₃CO₂ + O₂ + OH (R4c)

CF₃C(O)O₂ + NO₂ + M → CF₃C(O)O₂NO₂ + M (R5)

Thus, while the photolysis of CF₃CHO leads to radical fragments that further degrade to e.g. CO and F₂CO, reaction with OH maintains the C-2 chain and can thus result in the formation of TFA which is a highly persistent, water-soluble compound that is primarily removed from the atmosphere through wet deposition.²⁰ TFA accumulates in aquatic ecosystems, particularly in oceans and terminal basins, with no known significant degradation pathway.⁶ Although current and predicted environmental concentrations are well below toxic thresholds for aquatic life, debates and uncertainties about the sources, fate, and potential ecological impact of TFA persist.⁷ The formation of FPAN in (R5) may also represent a chemical pathway to TFA via hydrolysis on aqueous hydrometeors or wet-surfaces or even via reaction with water vapour.²¹

Several experimental studies of the reaction of CF₃CHO with the OH radical have been published and these have been summarised and evaluated by the IUPAC panel which, in the absence of published, temperature-dependent rate coefficients, makes a recommendation for the rate coefficient at room temperature only with k₁ = 5.5 × 10⁻¹³ cm³ molecule⁻¹ s⁻¹ at 298 K.⁵ This number is reported to be associated with large uncertainty (58%), reflecting the scatter in individual determinations of k₁. Rate coefficients derived using absolute methods vary from 1.1 × 10⁻¹² cm³ molecule⁻¹ s⁻¹ (Dóbé et al. 1989) to 6.5 × 10⁻¹³ cm³ molecule⁻¹ s⁻¹ (Scollard et al. 1993).^1,2 Relative rate studies also report very different rate coefficients with values of 4.4 × 10⁻¹³ cm³ molecule⁻¹ s⁻¹ (Scollard et al. 1993), 4.8 × 10⁻¹³ cm³ molecule⁻¹ s⁻¹ (Sellevåg 2004), 6.15 × 10⁻¹³ cm³ molecule⁻¹ s⁻¹ and 6.93 × 10⁻¹³ cm³ molecule⁻¹ s⁻¹ (Sulbaek-Andersen 2004).^2,3,22 These studies are discussed in detail in Section 3.2 where we compare them with our results.

For the equivalent non-fluorinated acetaldehyde (CH₃CHO) a negative temperature dependence for reaction with OH has been reported by Sivakumaran et al. 2003 and IUPAC over the temperature range 200–380 K.^5,23 Literature studies have shown that the rate coefficient for OH with CH₃CHO is orders of magnitude larger than that for OH + CF₃CHO indicating a strong effect of the fluorine atoms on the reactions potential energy surface and thus its temperature dependence. The lack of published, temperature-dependent kinetic data for OH + CF₃CHO thus makes prediction of its lifetime at various altitudes in the troposphere (where temperatures may vary between ∼300 and 230 K) precarious and precludes accurate assessment of the lifetime of CF₃CHO and its potential to lead to TFA formation.

The objective of this research is to experimentally determine the temperature-dependent rate coefficient for the reaction of CF₃CHO with OH. To achieve this, we have used pulsed laser photolysis coupled with pulsed laser-induced fluorescence (PLP–PLIF) and online gas-phase FTIR measurements of reactant concentrations, which enable precise determination of rate coefficients across a large temperature range. Additionally, we conducted room-temperature relative-rate experiments in a photolytic reactor with in situ FTIR analysis (RR-FTIR).

2. Experimental

Absolute, temperature-dependent rate coefficients, k₁(T), for the title reaction were obtained using pulsed laser photolytic (PLP) generation of OH coupled to pulsed laser induced fluorescence (PLIF) detection with [CF₃CHO] ≫ [OH] (pseudo-first-order conditions). The concentration of CF₃CHO was measured in situ via Fourier-transform infra-red absorption (FTIR) spectroscopy. In further, relative-rate experiments using a photolytic chamber with FTIR absorption spectroscopy (RR-FTIR), the room temperature (298 ± 2 K) rate coefficient was measured using C₂H₆ as reference reactant. The experimental setups are briefly described below; further details for the PLP–PLIF and RR-FTIR set-ups are found in Wollenhaupt et al. 2000 and Groß et al. 2014 (PLP–PLIF) and in Crowley et al. 1999 and Berasategui et al. 2020 (RR-FTIR).^24–27

2.1. PLP–PLIF experiments

Experiments were carried out in a double-jacketed quartz reactor (volume ∼500 cm³) thermostatted by circulating ethanol (204 K to 300 K) or water (300 K to 364 K). The reactor is equipped with a K-type thermocouple, which was used to probe the temperature in the centre of the reactor (where the photolysis and excitation laser pulses overlap) before each experiment. The pressure in the reactor and FTIR-absorption cell (see below) was measured using 100 and 1000 Torr capacitance manometers. Gas flows were set up so that a linear velocity of ∼70 cm s⁻¹ was achieved, ensuring complete gas exchange for each photolysis pulse (10 Hz), preventing the accumulation of products. Gases were pre-mixed in a glass manifold before flowing into the reactor. A distance of ∼15 cm between the reactor inlet and the photolysis zone ensured a thermal equilibration with the reactor walls.

2.1.1. Generation of OH. Pulsed radiation for the in situ OH-radical generation from a suitable gas-phase precursor (see below) was provided by a KrF excimer laser (COMPex pro 201F, Coherent), which delivered pulses (∼20 ns) at 248 nm perpendicular to the gas flow direction. The laser fluence was measured with a Joule–meter positioned behind the exit window of the reactor and, based on the precursor absorption cross-section and quantum yield at 248 nm, was used to calculate approximate, initial OH concentrations. A total of four schemes ((R6), (R7), (R8) + (R9), (R8) + (R10)) for generating OH were tested.

H₂O₂ + hν (248 nm) → 2 OH (R6)

HNO₃ + hν (248 nm) → OH + NO₂ (R7)

O₃ + hν (248 nm) → O (¹D) + O₂ (R8)

O(¹D) + H₂O → 2 OH (R9)

O(¹D) + CH₄ → OH + CH₃ (R10)

Apart from photolysis of H₂O₂ all other schemes suffered from problems related to secondary chemistry or (for HNO₃) surface reaction with CF₃CHO. These issues are highlighted and discussed in the SI. All the PLP–PLIF rate coefficients we report were thus obtained using H₂O₂ as OH source, whereby typical H₂O₂ concentrations of ∼5 × 10¹⁴ molecule cm⁻³ resulted in initial OH concentrations of (1–5) × 10¹⁰ molecule cm⁻³.

2.1.2. Detection of OH and CF₃CHO. OH radicals were excited (A² Σ(ν = 1) ← X² Π (ν = 0) transition) at ∼282 nm using a YAG-pumped dye laser (Quantel Brilliant B, Lambda Physik Scanmate). The subsequent OH fluorescence was detected using a photomultiplier screened by an interference filter (309 ± 5 nm). The triggering of the photolysis and probe lasers was controlled by a digital delay generator. Transient OH profiles consisting of 20 pre-excimer pulses and 50 post-excimer pulses were obtained by accumulating the fluorescence signal using a boxcar averager. To achieve an improved signal-to-noise ratio, ∼20 OH profiles were accumulated for the derivation of decay kinetics.

The CF₃CHO concentration was monitored by its infra-red absorption in a 45 cm long cell coupled to a FTIR spectrometer (Bruker-Vector 22 with an external, L_N2-cooled MCT detector). The absorption cell could be located either upstream or downstream of the reactor. Infrared absorption spectra were recorded with a resolution of 1 cm⁻¹ by coaddition of 64 interferograms (background: 128 interferograms) and phase-corrected and Boxcar-apodized (Mertz) with a zero-filling factor of 4. The measured CF₃CHO IR-absorbance was converted to a concentration by spectral deconvolution/fitting to a calibrated, CF₃CHO reference spectrum as well as other relevant absorbing species. There were no strong, unassigned bands in the spectrum and later we set upper limits to the concentrations of IR-absorbing impurities such as C₂H₄. The statistical uncertainty in the CF₃CHO concentration was generally <1% and always <3%. The CF₃CHO reference spectrum was obtained relative to the well-known UV-absorption spectrum of CF₃CHO by simultaneously recording optical density between 250 and 400 nm and absorbance in the 44 cm FTIR cell. For this purpose, the UV-absorption of CF₃CHO was measured in a multipass absorption cell with 880 cm optical path-length coupled to a UV-vis spectrometer with CCD-camera (Andor DU420A-OE).

The very good agreement in the absorption cross-section of CF₃CHO at λ_max = 300 nm between several studies (IUPAC: 3.06 × 10⁻²⁰, Hashikawa 2006: 3.1 × 10⁻²⁰, Chiappero 2006: 2.92 × 10⁻²⁰, Meller 1993: 3.0 × 10⁻²⁰, Sellevag 2004: 3.17 × 10⁻²⁰ in cm² molecule⁻¹) indicate that the uncertainty associated with the concentrations obtained in this manner (and thus the IR-cross-sections for CF₃CHO) are <5%.^3,5,28,29 Cross-sections obtained in independent experiments in which the IR absorption of CF₃CHO was determined using a short absorption cell (10 cm, with the CF₃CHO taken from the headspace of the pure liquid) were used to derive integrated band strengths that agreed to within 3% with those obtained when concentrations were derived by UV-absorption. These results are presented in detail in the SI.

2.2. Relative rate experiments, RR-FTIR

The relative rate experiments were carried out in a 44.39 L cylindrical quartz chamber equipped with in situ, infra-red analysis of CF₃CHO and the reference reactants. The same FTIR spectrometer was used as described in the PLP–PLIF experiments, in this case coupled to an internal, white-type multi pass mirror system resulting in an infra-red optical pathlength of 38.4 m. Infrared absorption spectra (600–3600 cm⁻¹) were recorded at 1 cm⁻¹ resolution by coaddition of between 64 and 128 interferograms.

Experiments were carried out on static samples at room temperature and a pressure of around 750 Torr using synthetic air or N₂ as bath gases. Gases (C₂H₆, CF₃CHO, H₂ and O₃) were sequentially dosed via a glass vacuum line and flushed into the chamber with bath gas. The pressures in the chamber and the vacuum line were measured using capacitance manometers (reactor: 1000 mbar, vacuum line: 1000 mbar, 100 Torr, 10 Torr).

Photochemical OH generation was initiated by the photolysis of O₃ with eight radially-mounted, low-pressure Hg lamps providing a homogeneous light flux (∼253.65 nm) in the chamber. O(¹D) was converted to OH via its reaction with H₂ as described previously.^27,30

O₃ + hν (253.65 nm) → O(¹D₂) + O₂ (R11)

O(¹D₂) + H₂ → OH + H (R12)

H + O₂ + M → HO₂ + M (R13)

Typically, initial concentrations of 1 × 10¹⁶ molecule cm⁻³ of O₃ and 0.5–2 × 10¹⁷ molecule cm⁻³ of H₂ were used. The generation of OH in this manner also produces HO₂ which further reacts (e.g. with O₃) to recycle OH.²⁷ Simulations of the chemistry indicate an [HO₂]/[OH]₀ ratio between 44 and 1900 depending on the initial H₂ and O₃ concentrations (see SI for further details). The use of H₂O₂ and HNO₃ as OH precursors was not possible in the relative-rate method as their rate coefficients for reaction with OH are much larger than for CF₃CHO. In addition, neither O₃ nor H₂ has strong absorption features in the relevant IR regions, which facilitates the analysis of the spectra. The initial concentrations of CF₃CHO and C₂H₆ were typically 8.5 × 10¹³ molecule cm⁻³ and 1.6 × 10¹⁴ molecule cm⁻³, respectively. As CF₃CHO absorbs dissociatively at 253.65 nm (σ ∼ 5 × 10⁻²¹ cm² molecule⁻¹), its slow loss due to photolysis needs to be taken into account for the relative-rate analysis.^3,31 The first-order loss rate constant for CF₃CHO in this set-up (k_phot) was obtained by measuring its depletion in the chamber in different bath gases (N₂, N₂/H₂ and synthetic air/H₂) with the photolysis lamps on, but without adding O₃.

2.3. Chemicals

N₂ (Air Liquide, 99.999%) and synthetic air (Air Liquide, 99.999%) were used without further purification. Anhydrous HNO₃ was prepared by mixing H₂SO₄ (Roth, 98%) and KNO₃ (Roth, >99%) and collecting the HNO₃ vapour in a liquid-nitrogen-cooled trap. The liquid HNO₃ was stored at −30 °C. H₂O₂ (AppliChem, 50 wt%) was concentrated by vacuum distillation to >90 wt%. H₂ (Westfalen, 99.999%) was used without further purification. O₃ was photochemically generated from O₂ (Westfalen, 99.999%) and stored on silica gel at −78 °C prior to each experiment. CF₃CHO was prepared by drying the monohydrate (Sigma Aldrich, 75% in water), passing it over P₂O₅ (Thermo Scientific, >99%) and collecting the vapour in a liquid-nitrogen-cooled trap. Diluted samples of gaseous CF₃CHO in N₂ (5–10 wt%) were prepared by manometric methods and stored in a darkened, 20 L glass bulb. C₂H₆ (Air Liquide, 99.95%) was used without further purification.

3. Results & discussion

3.1. Rate coefficients (204–364 K) for k₁via PLP–PLIF

Fig. 1 displays a representative set of OH decays for a range of [CF₃CHO] obtained in N₂ bath gas at 296 K. The IR-absorbance by CF₃CHO for each [CF₃CHO] is shown in the lower panel of this Figure.

Fig. 1 (a) Exponential decay of the OH-LIF signal over two orders of magnitude versus reaction time for five different CF₃CHO concentrations at 296 K. H₂O₂ was used as photolytic OH source. (b) Corresponding FTIR spectra for the five different CF₃CHO concentrations recorded simultaneously with the PLP–PLIF measurements.

The time-dependent OH-LIF signals exhibit strictly mono-exponential behaviour, characterized by their respective decay constants k′. Under pseudo-first-order conditions ([CF₃CHO] ≫ [OH]₀), the time-dependent OH concentration [OH]_t is given by:

[OH]_t = [OH]₀ × exp(−k′t) (1)

where [OH]₀ is the initial concentration after the excimer-laser pulse. The first-order decay constant k′ includes loss of OH due to reaction with CF₃CHO (k₁[CF₃CHO]) as well as reaction with H₂O₂ and diffusive losses. As, for a given experiment, the pressure and temperature diffusive losses are constant and H₂O₂ is unchanged, the latter two terms can be combined as a single term d:

k′ = k₁[CF₃CHO] + d (2)

The rate coefficient k₁ can thus be obtained by plotting k′ versus [CF₃CHO] as shown in Fig. 2 for experiments conducted at different temperatures. As expected from (1), k′ increases linearly with [CF₃CHO]. The term d (the y-axis intercept) varied from one experiment to the next according to the temperature, pressure and the H₂O₂ concentration, with its values in the absence of CF₃CHO (not plotted) consistent with those derived from linear fits as shown in Fig. 2. All determinations of k₁ (a total of 17 at various temperatures), obtained using H₂O₂ photolysis as OH source, are listed in Table 1.

Fig. 2 OH decay constants k′ versus the concentration of CF₃CHO at different temperatures. The y-axis intercept varies from 70 s⁻¹ to 190 s⁻¹ due to OH loss processes such as the reaction with H₂O₂. The horizontal error-bars combine the 2σ uncertainties obtained from spectral deconvolution, and an estimated systematic uncertainty of 5% related to the IR-cross-section. Uncertainties in the decay constants k′ correspond to the 2σ confidence intervals derived from mono-exponential fits to the OH-LIF signals.

Table 1 Experimental rate coefficients k₁ obtained using PLP–PLIF

T (K)	p (Torr)	k 1 (10^−13 cm^3 molecule^−1 s^−1)	[CF3CHO] (10^15 molecule cm^−3)
204	103	3.4 ± 0.1	0.8–4.2
213	103	3.5 ± 0.2	1.0–3.8
222	103	3.9 ± 0.1	0.9–4.6
233	103	4.3 ± 0.1	0.8–4.0
244	104	4.0 ± 0.2	0.6–5.5
249	103	4.5 ± 0.3	0.8–4.2
268	96	4.8 ± 0.1	0.4–4.7
278	96	5.2 ± 0.1	0.3–4.8
283	97	5.3 ± 0.1	0.3–5.4
288	97	5.4 ± 0.2	0.3–5.1
294	108	5.6 ± 0.3	1.4–8.0
295	101	5.5 ± 0.1	1.4–5.4
296	107	5.6 ± 0.1	1.1–5.7
296	108	5.5 ± 0.1	1.1–5.5
299	104	6.0 ± 0.3	0.5–4.5
333	106	7.2 ± 0.1	0.6–3.2
361	106	8.3 ± 0.1	0.6–2.9

---

As the rate coefficient for OH + CF₃CHO is not large, there is a potential for fast secondary reactions (i.e. reaction of OH with radical products) biasing the OH decay constant to high values. As an example, if we assume that CF₃CO radical product of the title reaction reacts with collision frequency (∼2 × 10⁻¹⁰ cm³ molecule⁻¹ s⁻¹) with OH to form CF₃C(O)OH, then a ∼0.1% conversion of CF₃CHO to CF₃CO would result in an enhancement of the OH decay rate (by up to a factor 2 at longer reaction times) and a distortion of the OH decay from mono-exponential behaviour. Such secondary effects can be avoided by working at low initial OH concentrations and thus minimal conversion of CF₃CHO to reactive products. In our experiments the ratio of [CF₃CHO]/[OH]₀ was >1 × 10⁴ so such effects should be negligible. In order to confirm this, a series of experiments was conducted in which the laser fluence (and thus the initial OH concentration) was varied for a fixed H₂O₂ and CF₃CHO concentration. As shown in Fig. S2.1, k′ was independent of the laser fluence and thus independent of [OH]₀ over the approximate range 1 × 10¹⁰–5 × 10¹¹ molecule cm⁻³. This confirms that reactions of OH with products did not influence our rate coefficients and also confirms that radical fragments formed in the photolysis of CF₃CHO did not play a significant role. This is to be expected as the absorption cross-section of CF₃CHO at 248 nm (2.60 × 10⁻²¹ cm² molecule⁻¹) is rather low.²⁹ The temperature dependent rate coefficients obtained are listed in Table 1 and also plotted in Fig. 3 where an overall positive dependence of k₁ on temperature is observed as well some curvature in the Arrhenius plot. The rate coefficients are well reproduced by the expression:

k₁(T) = (3.8 ± 0.2) × 10⁻¹³ × (T/300)² × exp[(131 ± 16)/T] (3)

with all data within the 2σ confidence interval. The error bars represent statistical uncertainty only (2σ) as returned by weighted, least-squares fits to the data as shown in Fig. 2 and thus also include statistical uncertainty in deriving concentrations by least-squares fitting of the FTIR spectra to a reference spectrum. A potential source of systematic bias in absolute rate coefficients is related to the measurement of the concentration of the excess reactant, i.e. CF₃CHO. In this study, on-line measurement of [CF₃CHO] via IR absorption spectroscopy ensures that this parameter is known to better than 5% (see Section 2.1.3). A further potential source of error when measuring relatively low rate coefficients is the presence of reactive impurities which can lead to a positive bias. C₂H₄ is a known impurity in samples of CF₃CHO that reacts with OH (k = 6.6 × 10⁻¹² cm³ molecule⁻¹ s⁻¹ at 298 K and 100 Torr N₂).^32,33 C₂H₄ was indeed found as an impurity in our flowing CF₃CHO samples but at concentrations that were ≤0.1% of the CF₃CHO concentration. This would bias the result by ∼1%, which is much lower than our overall uncertainty and thus negligible.

Fig. 3 Temperature dependent rate coefficients from this study (PLP–PLIF) along with literature values.^1–5 Uncertainties in k₁ correspond to the 2σ confidence interval derived from the linear fit of k′ versus [CF₃CHO]. RR = relative rate method, DF-RF = discharge flow-resonance-fluorescence method.

Our rate coefficient at room temperature is a factor ∼270 lower than the value of 1.5 × 10⁻¹¹ cm³ molecule⁻¹ s⁻¹ for the non-fluorinated analogue, CH₃CHO.⁵

CH₃CHO + OH → CH₃CO + H₂O (R14)

The positive dependence of k₁ on temperature also strongly contrasts k₁₄, which decreases with increasing temperature over the same range (i.e. from 2.6 × 10⁻¹¹ at 205 K to ∼1.0 × 10⁻¹¹ at 355 K). As, in both cases, the reaction proceeds via abstraction of the aldehydic hydrogen atom, it is clearly the substitution of the CH₃ group by CF₃ that reduces the rate coefficient by reducing the electron density around the carbonyl group. In the case of CH₃CHO, OH is believed to form a pre-reaction complex (PRC) with a five-membered ring in which the H-atom of OH is attached to the carbonyl oxygen.³⁴ From the PRC, the reaction may proceed via either passing over a barrier to products (i.e. a transition state in which the C–H bond lengthens compared to the PRC) or by tunnelling through the barrier, which leads to the observed negative dependence of k(OH + CH₃CHO) on temperature. The lower electron density around the carbonyl group for CF₃CHO results in a much less stable PRC, for which the decomposition back to reactants is faster. This results in a reduced net flux through the transition state to products and thus to a lower rate constant. For CF₃CHO, the contribution of “direct” abstraction (with an associated positive dependence on temperature) becomes relatively more important. The different temperature dependencies observed for CF₃CHO and CH₃CHO may also result from a difference in the barrier heights through the transition states relative to reactants.a

3.2. Comparison with literature

The 298 K rate coefficient for the reaction of OH with CF₃CHO from this study is 5.8 × 10⁻¹³ cm³ molecule⁻¹ s¹ which can be compared to room-temperature literature values, also plotted in Fig. 3. Our rate coefficient at 298 K agrees well with the IUPAC evaluation of previous datasets for this reaction.⁵ This agreement may however be somewhat fortuitous as the previous values cover a range of 1.1 × 10⁻¹² to 4.4 × 10⁻¹³ cm³ molecule⁻¹ s¹. The rate coefficient published by Scollard et al. was obtained by a similar method to that used here (pulsed laser photolysis to generate OH) and returned a value of k₁ = 6.5 × 10⁻¹³ cm³ molecule⁻¹ s¹ which is ∼12% larger than our value.² We note that Scollard et al. did not measure CF₃CHO in-line but relied on manometric measurements (i.e. partial pressures and partial flows) which are potential sources of systematic bias, and also do not discuss or rule out the presence of reactive impurities in their CF₃CHO samples (which were only stated to have >99% purity). In addition, the authors state that the initial CF₃CHO/OH ratio was greater than 100, which, given that they used a less sensitive technique for OH detection (resonance fluorescence rather than laser-induced-fluorescence), may not have been sufficient to rule out reactions of OH with secondary products. This in turn would have led to a high bias to their derived rate coefficient as described above. We cannot prove if any of these potential factors may have biased the Scollard et al. rate coefficient to slightly higher values than ours, but simply indicate that their slightly larger rate coefficient is consistent with such effects. Scollard et al. reported a value for k₁ at room-temperature only, however the same experimental set-up was used to derive rate coefficients between 230 and 355 K, which were only reported as conference proceedings with k₁ = 3.1 × 10⁻¹² exp[−(488 ± 57)/T].³⁵ As these data have not been rigorously peer-reviewed, the IUPAC panel has not considered them in their evaluations.⁵ The rate coefficients of Laverdet et al.³⁵ all lie ∼10% higher than our values and the same potential reasons as described for Scollard et al. apply. In further absolute measurements of k₁, Dóbé et al.¹ used a low-pressure flow tube with resonance-fluorescence detection of OH at 299 K. In their experiments the CF₃CHO/OH ratio was between ∼50 and 280 which would almost certainly have biased their rate coefficients to high values. Indeed, their rate coefficient is the highest of all literature values (k₁(299 K) = 1.1 × 10⁻¹² cm³ molecule⁻¹ s¹) and was not considered by IUPAC in their evaluation.

In addition to the absolute studies, there are several relative rate experiments which report values of k₁. As in the absolute studies, the values obtained also show considerable scatter (between (6.9 ± 0.1) and (4.4 ± 1.0) × 10⁻¹³ cm³ molecule⁻¹ s⁻¹) depending on the choice of OH-source (RONO photolysis in air or O₃ photolysis in the presence of H₂) and reference reactant including CH₃C(O)CH₃ (k_OH = 1.8 × 10⁻¹³ cm³ molecule⁻¹ s⁻¹), C₂H₆ (k_OH = 2.4 × 10⁻¹³ cm³ molecule⁻¹ s⁻¹), C₂H₄ (k_OH = 7.8 × 10⁻¹² cm³ molecule⁻¹ s⁻¹) and C₂H₂ (k_OH = 7.5 × 10⁻¹³ cm³ molecule⁻¹ s⁻¹).⁵ With the exception of C₂H₄, the other reference reactants all react with rate coefficients which are similar (within a factor of ∼two) to that of OH + CF₃CHO and are thus suitable for accurate determination of k₁. The remaining three determinations yield rate coefficients (10⁻¹³ cm³ molecule⁻¹ s⁻¹) of 4.4 ± 1.0 (Scollard et al. 1993), 4.8 ± 0.3 and 5.44 ± 0.71 (Sellevag et al. 2004), which are in reasonable agreement.^2,3

In order to better understand the differences between the various relative rate studies we conducted our own set of experiments in which the relative loss rates of CF₃CHO and C₂H₆ in the presence of an OH-source were determined.

3.3. Relative rate study of OH + CF₃CHO

The relative rate of reaction of OH with CF₃CHO (R1) and C₂H₆ (R15) was measured in the RR-FTIR set-up with the photolysis of O₃ providing the source of OH (R11).

C₂H₆ + OH → products (R15)

As CF₃CHO is lost by both reaction with OH and photolysis the relative change in concentration of CF₃CHO and reference reactant is given by:

where [CF₃CHO]₀ and [C₂H₆]₀ are initial concentrations of CF₃CHO and C₂H₆, [CF₃CHO]_t and [C₂H₆]_t are the concentrations after reaction time t. k₁ and k₁₅ are the bimolecular rate coefficients of the reaction of CF₃CHO and C₂H₆ with OH and J_phot is the photolysis frequency of CF₃CHO. J_phot = (8.3 ± 0.5) × 10⁻⁵ s⁻¹ was determined by monitoring the depletion of CF₃CHO absorption features during the lights-on phase (but in the absence of O₃). More details are provided in the SI (Fig. S4.1).

In the relative-rate experiments, the relative change in concentration of both reactants was monitored in situ with FTIR using the infra-red absorption feature at 3030–2957 cm⁻¹ for C₂H₆ and three different features of CF₃CHO at 1335–1280 cm⁻¹, 1210–1165 cm⁻¹ and 840–860 cm⁻¹. Exemplary FTIR spectra at reaction times of 0, 40 and 170 s are displayed in Fig. 4 along with reference spectra for CF₃CHO and C₂H₆. The strong absorption feature at around 2100 cm⁻¹ is from O₃. The spectra were baseline corrected and the absorption features of H₂O were removed as far as possible by fitting the spectra (2100–1300 cm⁻¹) to a reference spectrum of H₂O. In each composite spectrum, the concentrations of CF₃CHO and C₂H₆ were extracted by orthogonal distance regression to the reference spectra and also the spectra of any overlapping products that could be identified during the experiments. Since all the absorption features used exhibited Beer–Lambert linearity over the concentration range studied (see Fig. S1.2), the ratio of the fitting coefficients is equivalent to the concentration ratio, i.e. [CF₃CHO]₀/[CF₃CHO]_t and [C₂H₆]₀/[C₂H₆]_t.

Fig. 4 Plot of the FTIR-spectra measured during a relative-rate experiment at different reaction times (0 s, 40 s and 170 s, upper three traces) along with reference spectra of C₂H₆ and CF₃CHO. The sharp features around 1800 cm⁻¹ are due to H₂O vapour. The traces have been vertically offset for clarity of display.

As is apparent from Fig. 4, the concentration of O₃ decreases very rapidly, which is related to not only its photolysis but also secondary reactions with H-atoms and both OH and HO₂ radicals (see the SI for more details of the chemical scheme). Both CO₂ and CO (absorption features at ∼2350 and 2140 cm⁻¹) are clearly observed products from the degradation of the organic reactants.

Fig. 5 displays the depletion factors for CF₃CHO and C₂H₆, the slope of which gives the relative rate coefficient (k₁/k₁₅) for an experiment conducted with 1.1 × 10¹⁶ molecule cm⁻³ of H₂. As discussed previously by Sellevåg et al., H₂ serves two purposes; converting O(¹D) to OH and also scavenging CF₃O in order to prevent it from reacting with either CF₃CHO or C₂H₆.³

Fig. 5 Depletion factors (logarithmic ratio of initial concentration to concentration at reaction time t) for CF₃CHO and C₂H₆ obtained in a relative-rate experiment in N₂ with 1.1 × 10¹⁶ molecule cm⁻³ of H₂. The CF₃CHO depletion factor contains the term Jt which takes into account its slow loss owing to photolysis. The different colours represent different IR-absorption features that were used to derive the depletion of CF₃CHO and the linear fit (black line) represents an average.

Values of (k₁/k₁₅) obtained using the relative-rate method range from (1.39 ± 0.03) to (3.06 ± 0.09) depending on the experimental conditions (see Table 2). Using the IUPAC-recommended rate coefficient for OH + C₂H₆ (k₁₅ = 2.4 × 10⁻¹³ molecule cm⁻³), the rate coefficient for CF₃CHO + OH varies between (3.3 ± 0.3) to (7.4 ± 0.7) × 10⁻¹³ cm³ molecule⁻¹ s⁻¹.⁵ The variation in the rate coefficient is attributed to (1) the formation of CF₃O radicals and their reaction with CF₃CHO or C₂H₆ when [H₂] is low ((R16) and (R17)) and (2) the removal of CF₃CHO via reaction with HO₂ (R18), when [H₂] is high.³⁶

CF₃O + CF₃CHO → CF₃CO + CF₃OH (R16)

CF₃O + C₂H₆ → C₂H₅ + CF₃OH (R17)

CF₃O + H₂ (+ O₂) → HO₂ + CF₃OH (R18)

Table 2 Summary of relative rate experiments

Relative rate^a	Absolute k1^b (cm^3 molecule^−1 s^−1)	Initial [H2] (molecule cm^−3)	Bath gas
a k(OH + CF3CHO)/k(OH + C2H6). b Based on k(OH + C2H6) = 2.4 × 10^−13 cm^3 molecule^−1 s^−1.
1.39 ± 0.03	(3.3 ± 0.3) × 10^−13	4.3 × 10^15	Air
1.30 ± 0.03	(3.1 ± 0.3) × 10^−13	1.1 × 10^16	Air
1.71 ± 0.04	(4.1 ± 0.4) × 10^−13	1.1 × 10^16	N2
1.95 ± 0.10	(4.7 ± 0.5) × 10^−13	1.4 × 10^17	Air
3.07 ± 0.09	(7.4 ± 0.7) × 10^−13	1.9 × 10^17	Air

---

The source of the CF₃O radicals in these experiments is secondary reactions of CF₃O₂, themselves formed in e.g. the self-reaction of the CF₃C(O)O₂ radical. The rate coefficient for the reaction of CF₃O with C₂H₆ is 1.3 × 10⁻¹² cm³ molecule⁻¹ s⁻¹ but the rate coefficient for CF₃O with CF₃CHO (k₁₆) is not known.⁵ The rate coefficient k₁₆ was therefore estimated by assuming the same rate coefficient ratio of k(OH + CF₃CHO)/k(CF₃O + CF₃CHO) as that for k(OH + C₂H₆)/k(CF₃O + C₂H₆), resulting in k₁₆ = 3 × 10⁻¹² cm³ molecule⁻¹ s⁻¹. In a similar manner, by analogy with the rate coefficients for OH with H₂ and CH₄, the rate coefficient for the reaction of CF₃O with H₂ was estimated to be 2 × 10⁻¹² cm³ molecule⁻¹ s⁻¹. A concentration of H₂ of 2.3 × 10¹⁶ molecule cm⁻³ is thus needed to reduce the loss of C₂H₆ and CF₃CHO with CF₃O to <1% of their total (OH-dominated) loss.

To assess the impact of these reactions, we carried out numerical simulations of the chemistry (FACSIMILE³⁷) using an assumed reaction scheme (see Table S5.1). The key reactions in the scheme are listed in Table 3. Increasing the H₂ concentration reduces the potential impact of the unwanted reactions of CF₃O with CF₃CHO and C₂H₆ so one would expect the experiments at the highest H₂ concentrations to result in more accurate results. However, the use of very high H₂ also means that not only the CF₃O radical but also a significant fraction of the initially formed OH reacts with H₂ (in air) to form the HO₂ radical ((R19) and (R20)).

OH + H₂ → H₂O + H (R19)

H + O₂+ M → HO₂ + M (R20)

Table 3 Key reactions considered in the numerical simulations

Reaction	k(298 K) (cm^3 molecule^−1 s^−1)
OH reactions
OH + CF3CHO → H2O + CF3CO	5.8 × 10^−13
OH + C2H6 → H2O + CH2CH3	2.4 × 10^−13 (IUPAC)^5
Photolysis of TFAA
CF3CHO → CF3 + HCO	3.28 × 10^−5 s^−1 (IUPAC)^5
CF3CHO → CF3H + CO	3.04 × 10^−5 s^−1 (IUPAC)^5
CF3O reactions:
CF3O + CF3CHO → CF3OH + CF3CO	3.0 × 10^−12 (Assumed)
CF3O + C2H6 → CF3OH + CH2CH3	1.3 × 10^−12 (IUPAC)^5
CF3O + H2 → CF3OH + H	2.0 × 10^−12 (Assumed)
HO2 reactions
HO2 + CF3CHO → CF3CH(O)HOO	2.3 × 10^−13 (Long et al., 2022)^36
CF3CH(O)HOO → HO2 + CF3CHO	2.0 × 10^3 (Long et al., 2022)^36
CF3CH(O)HOO + HO2 → CF3CH(OH)OOH + O2	1.0 × 10^−11 (Assumed)

---

The use of high concentrations of H₂ thus results in a shift in the [HO₂]/[OH]₀ ratio towards HO₂ with the simulated ratios varying from 47 at [H₂] = 10¹⁶ molecule cm⁻³ to 789 at [H₂] = 2 × 10¹⁷ molecule cm⁻³ ([O₃] = 10¹⁶ molecule cm⁻³). According to a theoretical study,³⁶ the HO₂ radical may react with CF₃CHO to form the CF₃CH(OH)OO peroxy radical (R21) with a rate coefficient of k₂₀ = 2.34 × 10⁻¹³ cm³ molecule⁻¹ s⁻¹.

CF₃CHO + HO₂ → CF₃CH(OH)OO (R21)

CF₃CH(OH)OO → CF₃CHO + HO₂ (R22)

CF₃CH(OH)OO is unstable at room temperature and decomposes back to reactants on a ms timescale (k₂₁ = 1.72 × 10³ s⁻¹), which reduces the impact of the reaction between HO₂ with CF₃CHO (R22). However, given the high radical densities in these experiments, some fraction of the CF₃CH(OH)OO radical may react with OH or HO₂ ((R23) and (R24)). This would bias the relative rate coefficient to high values by providing an unwanted, extra sink of CF₃CHO.

CF₃CH(OH)OO + OH → products (R23)

CF₃CH(OH)OO + HO₂ → products (R24)

The rate coefficients for (R23) and (R24) are unknown, but a reasonable estimate can be obtained by drawing analogy to other HO₂ + RO₂ (k ∼1 × 10⁻¹¹ cm³ molecule⁻¹ s⁻¹) and OH + RO₂ reactions (k ∼1 × 10⁻¹⁰ cm³ molecule⁻¹ s⁻¹).

Using the theoretical rate coefficients from Long et al.³⁶ for the reactions (R20) and (R21) and the estimated rate coefficients for CF₃O with CF₃CHO and H₂, the results of the numerical simulations (see SI for a detailed discussion and Figures) show that, as observed, the relative rate coefficient obtained in the O₃/H₂ photolysis system suffer from systematic bias owing to unwanted reactions of CF₃O with both reactants (when [H₂] is too low) and also reactions of HO₂ with CF₃CHO (when [H₂] is too large). There thus appears to be a narrow window of H₂ concentrations in which these biases in the relative-rate measurements are minimised, and the value obtained approaches that derived from the PLP–PLIF experiments. We conclude that the relative rate studies of this reaction using the OH generation scheme outlined are unlikely to lead to accurate results for k₁. We note that the use of other OH source in relative rate experiments (e.g. the photolysis of CH₃ONO/NO mixtures) could conceivably deliver accurate rate coefficients if sufficient NO (or CH₃ONO) is present to scavenge any CF₃O formed and thus prevent its reaction with CF₃CHO or the chosen reference reactant. Such experiments were carried out by Sulbaek-Andersen et al. and Scollard et al., though the authors did not consider the possible interfering role of CF₃O.^2,22 In the experiments of Sulbaek-Andersen et al., a large concentration of CH₃ONO was used (a factor 5–10 more than CF₃CHO) which may plausibly have been sufficient to scavenge most CF₃O. In contrast, Scollard et al. used almost identical amounts and it is not possible to rule out that some loss of CF₃CHO and/or reference reactant biased their results. The excellent agreement between our PLP–PLIF rate coefficient at room temperature and that obtained by Sulbaek-Andersen et al. using the relative-rate technique likely reflects avoidance of potential pitfalls when determining k₁ using absolute and relative methods.

4. Conclusions and atmospheric lifetime of CF₃CHO

We have investigated the reaction between OH and CF₃CHO using PLP–PLIF over an atmospherically relevant temperature range of 204 K to 361 K using a number of OH sources whereby only the photolysis of H₂O₂ generated reproducible, accurate rate coefficients. Our results are summarised by the expression k₁(T) = (3.8 ± 0.2) × 10⁻¹³ × (T/300)² × exp[(131 ± 16)/T] cm³ molecule⁻¹ s⁻¹, reflecting a weak positive dependence on temperature over this temperature range. This is the opposite trend to that observed for the non-fluorinated analogue which, (along with the much lower absolute values of the rate coefficient) is related to destabilisation of the pre-reaction complex by the substitution of H- with F-atoms on the methyl-group. A series of relative rate experiments revealed sources of potential bias involving formation of the CF₃O radical and its reaction with CF₃CHO and the reference reactant, as well as reactive loss of CF₃CHO through reaction with the HO₂ radical.

For the reasons listed above, we consider our PLP–PLIF experiments, conducted using very high [CF₃CHO]/[OH]₀ ratios and also better control of impurities, to be more reliable than the previously published absolute rate studies, which were conducted at room-temperature only. In addition, our room temperature result (k₁ = (5.8 ± 0.5) × 10⁻¹³ cm³ molecule⁻¹ s⁻¹) is in excellent agreement with the previous relative rate study by Sulbaek-Andersen et al. (k₁ = (5.4 ± 0.7) × 10⁻¹³ cm³ molecule⁻¹ s⁻¹), which is likely the least affected by secondary reactions involving the CF₃O radical.²² The tropospheric lifetime of CF₃CHO with respect to OH can be estimated using our temperature dependent rate coefficient and an approximate OH concentration (independent of altitude) of 10⁶ molecule cm⁻³. Based on a standard temperature profile,³⁸ the lifetime in the lower three kilometres of the troposphere is estimated to be 22 days, increasing to 30 days at altitudes between 8 and 11 kilometres due to the lower temperatures. This is longer than the CF₃CHO lifetime with respect to photolysis, which is presently thought to be <1 week in the boundary layer for which photolysis quantum yields close to room temperature are available.⁵

Conflicts of interest

There are no conflicts to declare.

Data availability

IR characterization of reactants, additional details on the measurements (PLP–PLIF and RR-FTIR), and simulations (reaction scheme) supporting this article are included in the SI. See DOI: https://doi.org/10.1039/d5cp02871j

Acknowledgements

The authors thank AstraZeneca for partial financial support of this investigation. Open Access funding provided by the Max Planck Society.

References

1. S. Dóbé, L. A. Khachatryan and T. Bérces, Z. Phys. Chem., 1989, 93, 847.
2. D. J. Scollard, J. J. Treacy, H. W. Sidebottom, C. Balestra-Garcia, G. Laverdet, G. Lebras, H. Macleod and S. Teton, J. Phys. Chem., 1993, 97, 4683.
3. S. R. Sellevåg, T. Kelly, H. Sidebottom and C. J. Nielsen, Phys. Chem. Chem. Phys., 2004, 6, 1243.
4. M. P. Sulbaek Andersen, C. Stenby, O. J. Nielsen, M. D. Hurley, J. C. Ball, T. J. Wallington, J. W. Martin, D. A. Ellis and S. A. Mabury, J. Phys. Chem. A, 2004, 108, 6325.
5. M. Ammann, R. A. Cox, J. N. Crowley, H. Herrmann, M. E. Jenkin, V. F. McNeill, A. Mellouki, M. J. Rossi, J. Troe and T. J. Wallington, Evaluated kinetic and photochemical data for atmospheric chemistry, 2025, https://iupac.aeris-data.fr/.
6. J. C. Boutonnet, P. Bingham, D. Calamari, C. D. Rooij, J. Franklin, T. Kawano, J.-M. Libre, A. McCul-loch, G. Malinverno, J. M. Odom, G. M. Rusch, K. Smythe, I. Sobolev, R. Thompson and J. M. Tiedje, Hum. Ecol. Risk Assess., 1999, 5, 59.
7. K. R. Solomon, G. J. M. Velders, S. R. Wilson, S. Madronich, J. Longstreth and P. J. Aucamp, J. Toxicol. Environ. Health, Part B, 2016, 19, 289.
8. R. Britton, V. Gouverneur, J.-H. Lin, M. Meanwell, C. Ni, G. Pupo, J.-C. Xiao and J. Hu, Nat. Rev. Methods Primers, 2021, 1, 47.
9. S. Caron, Org. Process Res. Dev., 2020, 24, 470.
10. Y. Ogawa, E. Tokunaga, O. Kobayashi, K. Hirai and N. Shibata, iScience, 2020, 23, 101467.
11. M. S. Javadi, R. Søndergaard, O. J. Nielsen, M. D. Hurley and T. J. Wallington, Atmos. Chem. Phys., 2008, 8, 3141.
12. M. D. Hurley, T. J. Wallington, M. P. Sulbaek Andersen, D. A. Ellis, J. W. Martin and S. A. Mabury, J. Phys. Chem. A, 2004, 108, 1973.
13. W. M. O. (WMO), State of the Global Climate 2022, Geneva, 2023.
14. J. B. Burkholder, R. A. Cox and A. R. Ravishankara, Chem. Rev., 2015, 115, 3704.
15. T. J. Wallington, M. P. S. Andersen and O. J. Nielsen, Adv. Atmos. Chem., 2017, 305 DOI:10.1142/9789813147355_0005.
16. X. Zhang and Y. Li, Energy, 2024, 311, 133423.
17. A. Tewari, A. Parashar, W. Ahmad, A. Kumar, R. K. Vishwkarma, M. A. Ansari and P. Argulles, Compressors and Blowers: Maintenance, Practical Guidance, Energy-Efficient Technologies, 2025, 13, 179.
18. Y. Wang, Z. Wang, M. Sun, J. Guo and J. Zhang, Sci. Total Environ., 2021, 780, 146631.
19. M. P. Pérez-Peña, J. A. Fisher, C. Hansen and S. H. Kable, Environ. Sci.: Atmos., 2023, 3, 1767.
20. M. L. Hanson, S. Madronich, K. Solomon, M. P. Sulbaek Andersen and T. J. Wallington, Environ. Toxicol. Chem., 2024, 43, 2091.
21. J. Salas, A. L. Cardona, M. A. Burgos Paci and F. E. Malanca, Atmos. Environ., 2024, 337, 120780.
22. M. P. Sulbaek Andersen, O. J. Nielsen, M. D. Hurley, J. C. Ball, T. J. Wallington, J. E. Stevens, J. W. Martin, D. A. Ellis and S. A. Mabury, J. Phys. Chem. A, 2004, 108, 5189.
23. V. Sivakumaran and J. N. Crowley, Phys. Chem. Chem. Phys., 2003, 5, 106.
24. M. Wollenhaupt, S. A. Carl, A. Horowitz and J. N. Crowley, J. Phys. Chem., 2000, 104, 2695.
25. C. B. M. Groß, T. J. Dillon, G. Schuster, J. Lelieveld and J. N. Crowley, J. Phys. Chem. A, 2014, 118, 974.
26. J. N. Crowley, G. Saueressig, P. Bergamaschi, H. Fischer and G. W. Harris, Chem. Phys. Lett., 1999, 303, 268.
27. M. Berasategui, D. Amedro, A. Edtbauer, J. Williams, J. Lelieveld and J. N. Crowley, Atmos. Chem. Phys., 2020, 20, 2695.
28. R. Atkinson, D. L. Baulch, R. A. Cox, J. N. Crowley, R. F. Hampson, R. G. Hynes, M. E. Jenkin, M. J. Rossi, J. Troe and T. J. Wallington, Atmos. Chem. Phys., 2008, 8, 4144.
29. M. S. Chiappero, F. E. Malanca, G. A. Argüello, S. T. Wooldridge, M. D. Hurley, J. C. Ball, T. J. Wallington, R. L. Waterland and R. C. Buck, J. Phys. Chem. A, 2006, 110, 11944.
30. A. J. C. Bunkan, G. Srinivasulu, D. Amedro, L. Vereecken, T. J. Wallington and J. N. Crowley, Phys. Chem. Chem. Phys., 2018, 20, 11306.
31. Y. Hashikawa, M. Kawasaki, R. L. Waterland, M. D. Hurley, J. C. Ball, T. J. Wallington, M. P. S. Andersen and O. J. Nielsen, J. Fluorine Chem., 2004, 125, 1925.
32. R. Atkinson, D. L. Baulch, R. A. Cox, J. N. Crowley, R. F. Hampson, R. G. Hynes, M. E. Jenkin, M. J. Rossi and J. Troe, Atmos. Chem. Phys., 2006, 3625,  DOI:10.5194/acp-6-3625-2006.
33. M. P. Sulbaek Andersen, M. D. Hurley, T. J. Wallington, J. C. Ball, J. W. Martin, D. A. Ellis, S. A. Mabury and O. J. Nielsen, Chem. Phys. Lett., 2003, 379, 28.
34. I. W. M. Smith and A. R. Ravishankara, J. Phys. Chem. A, 2002, 106, 4798.
35. G. Laverdet, G. Le Bras, H. MacLeod, G. Poulet, S. Teton, D. Scollard, J. Treacy and H. Sidebottom, Laser photolysis-resonance fluorescence investigation of the reactions of hydroxyl radicals with CCl3CHO and CF3CHO as a function of temperature, SPIE, 1993.
36. B. Long, Y. Xia and D. G. Truhlar, J. Am. Chem. Soc., 2022, 144, 19910.
37. A. R. Curtis and W. P. Sweetenham, Facsimile, Atomic Energy Research Establishment, Report R-12805, 1987.
38. U. S. Atmosphere, US standard atmosphere, National Oceanic and Atmospheric Administration, 1976.

---