Atmospheric sink of methyl chlorodi ﬂ uoroacetate and ethyl chlorodi ﬂ uoroacetate: temperature dependent rate coe ﬃ cients, product distribution of their reactions with Cl atoms and CF 2 ClC(O)OH formation †

a Rate coe ﬃ cients as a function of temperature have been measured for the ﬁ rst time for the gas-phase reactions of chlorine atoms with methyl chlorodi ﬂ uoracetate ( k 1 ) and ethyl chlorodi ﬂ uoroacetate ( k 2 ) using the relative rate technique. The experiments were carried out in a 1080 L photoreactor over the temperature range 287 – 313 K at a total pressure of 1000 (cid:1) 10 mbar of synthetic air using in situ FTIR spectroscopy to monitor reactants and products. The following Arrhenius expressions were obtained: k (MCDFA+Cl) ¼ (9.6 (cid:1) 5.1) (cid:3) 10 (cid:4) 12 exp[ (cid:4) (1363 (cid:1) 79)/ T ] and k (ECDFA+Cl) ¼ (64.4 (cid:1) 29.7) (cid:3) 10 (cid:4) 12 exp [ (cid:4) (1110 (cid:1) 68)/ T ]. The kinetic results are compared with previous experimental and theoretical studies. In addition, a product study of the reactions of Cl with methyl chlorodi ﬂ uoracetate and ethyl chlorodi ﬂ uoroacetate is reported. The results indicate that in the absence of NO x the main fate of the alkoxy radicals formed after H-atom abstraction by Cl from the – CH 3 group in methyl chlorodi ﬂ uoroacetate is reaction of the radical with O 2 to form the mixed anhydride CF 2 ClC(O)OC(O)H. In the case of ethyl chlorodi ﬂ uoroacetate the main fate of the alkoxy formed via H-atom abstraction by Cl from the – CH 2 – entity in the ethyl group is a -ester rearrangement to produce chlorodi ﬂ uoroacetic acid and the corresponding radical. The yields of chloro ﬂ uoracetic acid (CF 2 ClC(O)OH) obtained were as follows: (34 (cid:1) 5)% and (86 (cid:1) 8)% for the reactions of Cl with CF 2 ClC(O)OCH 3 and CF 2 ClC(O)OCH 2 CH 3 , respectively. The measured yields are rationalized in terms of mechanisms consisting of competitive reaction channels for the formed products in the oxidation, i.e. reaction with O 2 , a -ester rearrangement and a decomposition pathway. Atmospheric implications are discussed according to the rate coe ﬃ cients obtained as a function of temperature and altitude, and regarding the formation of chloro ﬂ uorocarboxylic acid. uoroacetic acid (CF 2 ClC(O)OH), trace spectrum monoxide


Introduction
Oxygenated Volatile Organic Compounds (OVOCs) such as hydrouoroethers (HFEs) and hydrochlorouoroethers (HCFEs) have an oxygen linkage which makes the hydrogen atoms of the ether molecule more labile and increases their reactivity toward OH radicals and Cl atoms in the lower atmosphere compared to chlorouorocarbons (CFCs) and hydrochlorouorocarbons (HCFCs). This increased atmospheric photooxidation potential of HFEs and HCFEs over CFCs and HCFCs makes them good replacement materials in many applications such as solvents, heat-transfer uids in refrigeration systems, cleaning agents in the electronic industry, foam-blowing, lubricant deposition, and dry etching agents and as anesthetics. 1,2 The hydroxyl radical (OH) is the primary cleaning agent of the lower atmosphere and is thus the dominant sink for many greenhouse gases and pollutants. In coastal and marine regions and in inland areas near to pollutant sources the concentration of Cl can attain concentration levels which make reactions with Volatile Organic Compounds (VOCs) competitive with the OHradical initiated degradation of these VOCs. [3][4][5][6] In recent years, evidence has been presented for Cl-atom chemistry in continental regions far removed from coastal and marine regions where observations of Cl-atom precursors, have been made. [7][8][9][10] Additionally, urban sources of Cl atoms due to industrial activities can exist, and Cl concentrations higher than the concentration levels observed in marine regions have been reported, for example, in the vicinity of ceramic industries. 11 Chlorinated uoroesters (CFESs) are emitted to the atmosphere mainly by anthropogenic sources. Methyl chlorodi-uoroacetate (MCDFA) is used as a building block in organic synthesis 12 and ethyl chlorodiuoroacetate (ECDFA) is employed extensively in re retardants and also nds use in the production of pesticides, dyestuffs, phosphates and medicine. 13 MCDFA and ECDFA can be produced directly in the troposphere through the photooxidation of hydrochlorouoroethers analogous to the formation of uoroesters from HFEs. 14,15 In order to evaluate if HCFEs are environmentally acceptable replacements for CFCs and HCFCs, rate coefficients and product distributions for the reactions of the CFESs with the main tropospheric oxidants are necessary.
Hence, as part of a systematic study on the kinetics and oxidation products of the atmospheric reactions of haloesters, we report in this work, rate coefficients for the reactions of Cl atoms with methyl chlorodiuoroacetate and ethyl chlorodi-uoracetate measured over the temperature range 287-313 K using a relative kinetic technique: This study extends our previous kinetic database for the title reactions at 298 K 16 through provision of the temperature dependence for the reactions.
To the best of our knowledge, the temperature dependence of the reactions has not been previously reported. Although, there are many kinetic and mechanistic studies of the reactions of Cl atoms with a variety of hydrogenated esters, the database for haloesters is very limited. Furthermore, there is no information available on the degradation mechanisms of chloro-uorinated esters in the atmosphere and the resulting products.
Therefore, the main reaction products of the haloesters have been identied and their yields quantied. This information has been used to propose atmospheric degradation mechanisms for the Cl-atom mediated oxidation of the compounds. Formation of chlorodiuoracetic acid has been observed in high yields. Field measurements of the acid in precipitation suggest that it may accumulate in the aquatic enviroment. 17 The aim of the present work is to determine the fate of the chlorouoroalkoxy radical formed in the Cl-initiated photodegradation of the chlorouoroacetates studied, and appraise its possible contribution to the accumulation in the environment of chlorouorinated compounds such as chlorodiuoracetic acid (CDFA).

Experimental
The experiments were performed in a 1080 L photoreactor a detailed description of which can be found elsewhere thus only a brief general overall description is given here. 18 A diagram of the reactor is presented below (Scheme 1). The reactor has a total length of 6.2 m and an internal diameter of 0.47 m, it consists of two quartz glass cylinders which are connected by a central enamel ange ring and silicon rubber seals and is closed at both ends by aluminum anges which contain inlet ports for reactants etc. The reactor can be temperature regulated within the range 287 to 313 K with a precision of AE2 K through a regulation system using three PT-100 thermo-elements. The chamber can be evacuated by a pumping system consisting of a turbo-molecular pump backed by a double stage rotary fore pump to 10 À3 Torr. The photolysis system consists of 32 superactinic uorescent lamps (Philips TL05 40 W), which are evenly spaced around the reaction vessel. The lamps emit light with a maximum intensity at 360 nm. The chamber is equipped with a White type multiple-reection mirror system for sensitive in situ long path absorption monitoring of reactants and products in the IR spectral range 4000-700 cm À1 and was operated at total optical path length of 484.7 AE 0.8 m. IR spectra were recorded with a spectral resolution of 1 cm À1 using a Nicolet Nexus FT-IR spectrometer equipped with a KBr beam splitter and a liquid nitrogen cooled mercury-cadmium-telluride (MCT) detector.
The rate coefficients for the reaction of Cl atoms with the chlorouoroacetates were determined using the relative kinetic technique. Chlorine atoms were generated by photolysis of molecular Cl 2 : In the presence of Cl atoms, the chlorouoroacetates and the reference compounds decay through the following reactions: Cl + reference / products, k reference (5) Provided that the reference compound and the reactant are lost only by reactions (4) and (5), then it can be shown that: where, [chlorouoroacetate] 0 , [reference] 0 , [chlorouoroacetate] t and [reference] t are the concentrations of the halogenated esters and reference compound at times t ¼ 0 and t, respectively and k chlorouoroacetate and k reference are the rate coefficients of reactions (4) and (5), respectively. The relative rate technique relies on the assumption that both the chlorouoroacetates and reference organics are removed solely by reaction with Cl atoms. To verify this assumption, mixtures of chlorine and both organics (chloro-uoroacetate and reference hydrocarbon) in air were prepared and allowed to stand in the dark for 30 minutes the typical duration of an experiment. In all cases, reaction of the organic species with the Cl-atom precursor Cl 2 in the absence of UV light was of negligible over this time period. Additionally, to test for possible photolysis of the reactants mixtures of the chloro-uoroesters and reference compounds in air in the absence of molecular chlorine were irradiated using the output from all the uorescent lamps surroundings the chamber for 30 min. Photolysis was not observed for any of the reactants.
For the kinetic investigations chloromethane was used as the reference compound and experiments were performed on Cl 2 /chlorouoroester/chloromethane/air mixtures which were irradiated for periods of 15-20 minutes during the course of which infrared spectra were recorded with the FTIR spectrometer. Typically 64 interferograms were co-added per spectrum over a period of approximately 1 minute. Similar experiments were performed for the product studies without the reference compound in the reaction mixtures.

Results and discussion
Temperature dependence of the reactions of Cl atoms with MCDFA and ECDFA Rate coefficients for the reactions of Cl atoms with MCDFA and ECDFA were determined at atmospheric pressure over the temperature range of 287-313 K relative to the reaction of Cl atoms with chloromethane. Fig. S1 and S2 † show examples of the kinetic data obtained at three temperatures (287, 303 and 313 K) for both methyl chlorodiuoroacetate and ethyl chlor-odiuoroacetate plotted according to eqn (I). The rate coefficient ratios k CFES /k reference obtained from linear least squares analyses of such plots are listed in Table 1. The rate coefficient ratios are each from a minimum of at least three or four experiments for each temperature and CFES and the errors are the 2s statistical errors from the linear regression analyses.
The rate coefficients for the reaction of Cl atoms with the chlorouoroacetates listed in Table 1 have been put on an absolute basis using rate coefficients for Cl with chloromethane at the respective temperatures calculated from the following Arrhenius expression (with k in cm 3 per molecule per s units): k(Cl + chloromethane) ¼ 3.3 Â 10 À11 e À(1250)/T cm 3 per molecule per s. 19 Scheme 1 Schematic diagram of the 1080 L quartz-glass reactor. The errors given for the rate coefficients are a combination of the 2s statistical errors from the linear regression analyses plus an added uncertainty to take into account the error associated with the values of the reference rate coefficients.
The values of the rate coefficients for the reactions of Cl atoms with methyl chlorodiuoroacetate and ethyl chlorodi-uoroacetate as a function of temperature are plotted in Arrhenius form in Fig. 1. Linear least-squares analysis of the plots leads to the following Arrhenius expressions in units of cm 3 per molecule per s: The errors in the activation term and the pre-exponential factor are the 2s random statistical errors from ts to the data presented in Table 1 and plotted in Fig. 1 (panels a and b). To the best of our knowledge there are no prior experimental determinations of the Arrhenius parameters for these reactions, hence, this study represents the rst temperature dependence study of the reactions of Cl atoms with MCDFA and ECDFA.
The positive E a /R values of (1363 AE 79) K and (1110 AE 68) K for the reaction of Cl atoms with MCDFA and ECDFA, respectively, indicates the dominance of a hydrogen abstraction pathway under the experimental conditions i.e. the rate coefficients increase slightly with temperature.
The E a /R values of 1363 K and 1110 K reported in this work for the reactions of MCDFA + Cl and ECDFA + Cl, respectively are consistent with the available E a /R literature data for the reactions of Cl atoms with hydrouoroethers and uoroesters where E a /R values of 852, 1190, 2254, 968, 1216, 1637 and 1368 K have been reported for the reactions of Cl atoms with HFE-7200 (C 4 F 9 OC 2 H 5 ), HFE-7500 (n-C 3 F 7 CF(OC 2 H 5 )CF(CF 3 ) 2 ), HFE-7100 (C 4 F 9 OCH 3 ), HFE-7300 (n- The employment of temperature dependent kinetic data in model calculations increases the accuracy of the models since the temperature gradient of the troposphere and the seasonal temperature variations can be taken into consideration.

Comparison with previous kinetic determinations
The following rate coefficients have been obtained at 298 K for the reaction of Cl with MCDFA and ECDFA: 3

per molecule per s
The values of k 1 and k 2 at room temperature are in very good agreement with our previously reported values 16 3 per molecule per s 16 obtained at 298 K and 750 Torr total pressure in a 80 liter Teon bag using the relative kinetic method with gas chromatography/ame ionization detection (RR-GC/FID) for the analyses. These two measurements are represented in Fig. 1 by lled symbols, for the reaction of Cl with MCDFA ( Fig. 1a) and ECDFA (Fig. 1b), respectively. Mishra et al. 24 have calculated the rate coefficient for H-atom abstraction from CF 2 ClC(O)OCH 3 by Cl atoms using canonical transition state theory (CTST) and employing Wigner 25 and Eckart 26 symmetrical methods which gave values of 0.96 Â 10 À13 and 1.20 Â 10 À13 cm 3 per molecule per s, respectively, at 298 K for the reaction. These values are in very good agreement with the experimental value of 0.98 Â 10 À13 cm 3 per molecule per s reported in this work and the previous experimental value reported by our group, using a different experimental setup, of (1.0 AE 0.2) Â 10 À13 cm 3 per molecule per s. 19 In another study Mishra et al. 27   OCHCH 3 + HCl, respectively. Mishra et al. 27 reported a value of 1.86 Â 10 À12 cm 3 per molecule per s for the overall rate coefficient of the reaction (i.e. the sum of the rate coefficients for each reaction channel) which is in very good agreement with our previous experimental value of (1.5 AE 0.3) Â 10 À13 cm 3 per molecule per s 16 and the rate coefficient of (1.55 AE 0.44) Â 10 À13 cm 3 per molecule per s reported in this work. The RC(O)O-group is known to have a deactivating effect on C-H bonds positioned both a and also b to the group. 28 It is interesting to compare the rate coefficients for Cl with the two chlorodiuoroesters with those of methyl acetate (MA) and ethyl acetate (EA). The reactions of Cl with MA and EA have rate coefficients of 2.2 Â 10 À12 and 1.76 Â 10 À11 cm 3 per molecule per s, respectively, whereby the contribution from H-atom abstraction from the acetyl group CH 3 C(O)O is <1 Â 10 À13 . 28 The rate coefficients for the reactions of Cl atoms with MCDFA and ECDFA are factors of approximately 22 and 11 slower than those of Cl with MA and EA, respectively, indicating the strong deactivating effect of the ClF 2 C-group on the reactivity of the chlorodiuoroesters toward Cl. It is also interesting to note when comparing rate coefficients for halogenated esters 23 supporting an expected stronger negative inductive effect for F 3 C-over ClF 2 C-.
Product distribution studies Fig. 2 shows typical product spectra recorded in the Cl-atom initiated oxidation of CF 2 ClC(O)OCH 3 . Trace (A) shows a product spectrum obtained aer irradiation of a Cl 2 / chlorouoroester/air mixture and subtraction of residual MCDFA bands. It can be seen from a comparison of the spectrum with reference spectra of chlorodiuoroacetic acid (trace (B)) and carbon monoxide (trace (C)) that these two compounds are products formed in the Cl + MCDFA reaction. Trace (D) is the residual spectrum that is obtained aer subtraction of known products from the spectrum in trace (A).
Concentration-time proles of MCDFA and the identied products are shown in Fig. S3. † Plots of the concentration of the identied products as a function of reacted MCDFA (Fig. S4 in ESI †) are linear supporting that they are primary products. From a minimum of 3 experiments averaged molar formation yields of (34 AE 5)% and (26 AE 3)% were obtained for chlorodi-uoracetic acid and carbon monoxide, respectively.
In the reaction of Cl atoms with ethyl chlorodiuoroacetate, chlorodiuoracetic acid and carbon monoxide have been also identied as products. Fig. S5, † trace (A), shows an IR spectra acquired aer irradiation of a mixture of ECDFA and Cl 2 in air where residual absorptions due to ECDFA have been subtracted. Fig. S5, † trace (B), shows a reference spectra of chlorodiuoroacetic acid (CF 2 ClC(O)OH), trace (C) a reference spectra of carbon monoxide and trace (D) the residual spectrum that is obtained aer subtraction of known products from the spectrum in trace (A).
Concentration-time proles of ECDFA and the identied products chlorodiuoroacetic acid and carbon monoxide are shown in Fig. S6 in the ESI. † The concentration-time contour of chlorodiuoroacetic acid and the linearity of the plot of its concentrations as a function of reacted ECDFA support that it is a primary product. Carbon monoxide shows the same behavior, however, as discussed later the CO is thought to result from the fast reaction of Cl atoms with HCHO. Analysis of the slopes of the plots shown in Fig. S7 † give product yields of (86 AE 8)% for chlorodiuoroacetic acid and (34 AE 6)% for carbon monoxide in the reaction of Cl with ethyl chlorodiuoroacetate. Table 2 lists the yields of the products formed in the reactions of MCDFA and ECDFA with Cl atoms. The errors quoted are a combination of the 2s statistical errors from the regression analysis plus a contribution to take into account the uncertainty in the spectral subtraction procedure.
In the atmosphere the oxidation of the saturated esters is mainly initiated by reaction with OH radicals and proceeds via H-atom abstraction from the alkyl groups (-CH-, -CH 2 -, -CH 3 ) in the ester forming alkyl radicals. The reactions of the haloesters studied with Cl atoms are less selective than the analogous OH reactions but the subsequent chemistry is similar. Thus, although reactions of Cl-atoms with the esters are not so atmospherically important as the OH reactions, studies of their mechanisms are still necessary and are a convenient way to emulate OH-radical initiated chemistry.
Under atmospheric conditions alkyl radicals react with oxygen to give peroxy radicals. The peroxy radicals formed can react with NO, NO 2 , HO 2 , and other peroxy radicals forming, to a large extent, the corresponding alkoxy radicals. 29 In the present NO x free reaction systems the alkoxy radicals will be formed by reaction of the peroxy radicals with RO 2 /HO 2 and those shown in Schemes 2 and 3 for MCDFA and ECDFA, respectively, are expected to be dominant.
Alkoxy radicals are known to react by three different pathways: (i) reaction with oxygen forming a carbonyl compound and HO 2 radicals, (ii) decomposition via C-C bond cleavage producing a stable product and a radical and, (iii) if possible, isomerization via a six-membered transition state producing polyfunctional compounds. In addition in the case of esters, the alkoxy radical, formed by H-atom abstraction from the carbon attached to the non-carbonyl oxygen of the ester entity, can undergo what is known as an a-ester rearrangement 30 to form an acid and a radical. The products identied in the irradiation of MCDFA/Cl 2 /air and ECDFA/Cl 2 /air mixtures allow the elucidation of the fate of the alkoxy radicals.
a Atmospheric fate of CF 2 ClC(O)OCH 2 Oc radicals. Fig. 3 shows the possible fate of the CF 2 ClC(O)OCH 2 Oc radicals formed in the reaction of MCDFA + Cl. The observation of the formation of near equivalent amounts of chlorodiuoracetic acid and carbon monoxide around 30% (see Table 2) supports that the CF 2 ClC(O)OCH 2 Oc radicals undergoes an a-ester rearrangement (Fig. 3, channel A) forming the corresponding acid and HCO radicals which will react further with O 2 to give CO and HO 2 radicals. This is comparable to the acetic acid yield of around 30% obtained by Christensen et al. in their study on the reaction of Cl with methyl acetate in the absence of NO 31 and a triuoroacetic acid yield of $25% obtained by our group in a previous study on the reaction of Cl with methyl tri-uoroacetate under similar conditions to this work. 32 The possible C-O bond cleavage decomposition channel of the CF 2 ClC(O)OCH 2 Oc radical to form the CF 2 ClC(O)Oc radical and HCHO has been omitted in Fig. 3 since formation of formaldehyde was not observed in the reaction system. The CF 2 ClC(O)Oc radical can decompose to form CO 2 and the CF 2 Clc radical which can react with O 2 to form carbonyl uoride (CF 2 O). 32 Carbonyl uoride was observed in the system but its formation was clearly from secondary reactions. In addition, a previous study on the fate of the analogous alkoxy radical CH 3 C(O)OCH 2 Oc formed in the oxidation of methyl acetate has shown that the decomposition channel was also negligible. 31 The reaction of CF 2 ClC(O)OCH 2 Oc radicals with O 2 will lead to the formation of the mixed acid anhydride CF 2 ClC(O)OC(O)H (Fig. 3, channel B). Since a reference spectrum for CF 2 ClC   and alkyl haloacetates. 30,31 The residual product spectrum bares many similarities with that obtained in the study of Cl atoms with CF 3 C(O)OCH 3 where the main absorptions bands were assigned to formic triuoroacetic anhydride CF 3 C(O)OCHO. 32 The unique structural feature of an anhydride is the presence of two carbonyl groups linked to each other via an oxygen atom. This structure gives rise to two carbonyl absorptions in the infrared arising from symmetrical and asymmetrical stretching of the carbonyl groups. The residual product spectrum from Cl + MCDFA contains two carbonyl absorptions at 1831 and 1805 cm À1 separated by 26 cm À1 whereby the absorption at 1805 cm À1 is stronger than that at 1831 cm À1 . The small carbonyl band splitting is expected for anhydrides of structure RC(O)OCHO. 34 In the ngerprint region of the product spectrum band maxima occur at $1279, $1193, $1115, $999, $966 cm À1 whereby those at 1115 and 966 cm À1 are the strongest and equally intense. Interpretation of anhydride spectra in the ngerprint region is complicated because of a large number of C-O and C-C stretchings and various bending modes, many of which are strongly mixed. The strong band at 1115 cm À1 is an indication for the presence of an anhydride and is generally associated with C-O and C-C vibrations. However, although many anhydrides have strong bands in this region it is not a unique marker for the presence of an anhydride since many other compounds absorb in this region. 35,36 The bands at 1279 and 1193 cm À1 are consistent with vibrations from C-F and C-Cl in the ClF 2 C-group: similar bands occur in the infrared spectra of formic triuoroacetic anhydride and tri-uoroacetic anhydride. [35][36][37] The absorption bands at $999 and $960 cm À1 are most probably also from C-F and C-Cl vibrations, halogenated compounds such as 1,2-dichloro-tetrauoroethane show fairly strong bands in this region.
An infrared spectrum has been simulated for CF 2 ClC(O) OC(O)H using the using Gaussian 03 ab initio quantum chemistry package. 38 The spectrum was calculated using the B3LYP/6 31+G level of theory. [39][40][41] The simulated infrared spectrum is compared in Fig. S8, † panel B, with the residual product spectrum (panel A) from the Cl + MCDFA reaction. Although the computed spectrum captures the main spectral features of the experimental absorption bands reasonably well, there are large discrepancies in the absorption frequencies especially for the carbonyl bands. The computed spectrum, however, does lend additional support to the other spectral experimental evidence that the residual product spectrum is very probably mainly due to formation of CF 2 ClC b Atmospheric fate of CF 2 ClC(O)OCHO$CH 3 radicals. Mishra et al. 27 have reported that H-atom abstraction from the CH 3group in CF 2 ClC(O)OCH 2 CH 3 at 298 K to give the CF 2 -ClC(O)OCH 2 CH 2 c radical is very slow (k ¼ 3.88 Â 10 À14 cm 3 per molecule per s) compared to H-atom abstraction from the -CH 2group to give the CF 2 ClC(O)OCH$CH 3 radical (k ¼ 1.82 Â 10 À12 cm 3 per molecule per s). The authors reported that the barrier heights for hydrogen abstraction from the -CH 2group by Cl at both the MPWB1K and G2(MP2) levels of theory are lower than the corresponding values for the -CH 3 site indicating that the formation of CF 2 ClC(O)OCHO$CH 3 radicals is the main fate for the reaction of Cl with ECDFA (see Scheme 3). However, they concluded that some H-abstraction from the -CH 3 group in ECDFA could occur and could not be completely excluded. Orlando and Tyndall 42 have studied the reaction of Cl with ethyl acetate and found that Cl atom attack at the -CH 2group dominates for this non-halogenated ester.
The experimental evidence in this work, i.e. the high formation yield of chlorodiuoroacetic acid, also supports that in the reaction of Cl with ECDFA attack of the Cl atom at the -CH 2group dominates. The CF 2 ClC(O)OCH$CH 3 radical thus formed will add O 2 to form a peroxy radical further reactions of which with RO 2 /HO 2 will yield to a large extent the alkoxy radical CF 2 ClC(O)OCHO$CH 3 . A simplied mechanism showing the further possible reaction pathways of the CF 2 -ClC(O)OCHO$CH 3 radical is presented in Fig. 5.
The observation of a high chlorodiuoroacetic acid yield ($85%) supports that the a-ester rearrangement (Fig. 4, channel A) is the main reaction channel for the CF 2 ClC(O)OCHO$CH 3 radical, assuming that no other source of the acid exists. Blanco et al. 32 in their study on the Cl-atom oxidation of uoroacetates report a value of 78 AE 5% for the formation of triuoroacetic acid in the reaction of Cl with ethyl triuoroacetate. Orlando and Tyndall 42 have reported a yield of 53% for the formation of acetic acid in the reaction of Cl with ethyl acetate and concluded from their system analysis that other routes to the formation of the acid were very minor. In addition, Tuazón et al. 30 and Picquet-Varrault et al. 43 have reported yields of 96% and 75% for the formation of acetic acid in the reaction of OH radicals with CH 3 C(O)OCH 2 CH 3 , respectively. All of these studies are consistent with the main reaction pathway being abstraction of an H-atom from the -CH 2group with subsequent chemistry via an a-ester rearrangement leading to the corresponding acid compound.
The co-product of CF 2 ClC(O)OH, the CH 3 C(O) radical, will add O 2 to form the CH 3 C(O)OO peroxy radical. This radical can react with HO 2 to form peroxy acetic acid (CH 3 C(O)OOH), or react with other peroxy radicals to form CO 2 and CH 3 radicals. No evidence for formation of peroxy acetic acid was been observed in the residual product spectrum, however, an OH absorption was observed at $3585 cm À1 suggesting formation of a product containing either an OH or OOH group. Although CO 2 cannot be quantitatively determined in the system because of high background levels, there where indications that its concentration increased during the course of the reaction. The CH 3 radical will be oxidized to formaldehyde (HCHO) the further reaction of which will give CO. Formaldehyde could not be detected in the reaction system, which is not surprising given the high reactivity of HCHO with Cl. Orlando and Tyndall 42 in their study on the reaction of Cl with ethyl acetate could only detect traces of HCHO. Interestingly, the acid to CO ratio in this study on Cl with CF 2 ClC(O)OCH 2 CH 3 and in the studies on Cl with CF 3 C(O)OCH 2 CH 3 32 and Cl with CH 3 C(O)OCH 2 CH 3 42 are all $0.4 indicating other fates for the CH 3 C(O) radical rather than oxidation through HCHO to CO. It was not possible to elucidate the other reaction pathways from the present FTIR data.
The CF 2 ClC(O)OCHO$CH 3 alkoxy radical could potentially decompose via a C-O-C bond cleavage forming CF 2 ClC(O)Oc radicals and acetaldehyde. The CF 2 ClC(O)Oc radicals would be expected to decompose to form CO 2 and cCF 2 Cl radicals and the subsequent reactions of which would be expected to produce CF 2 O. Neither the formation of CF 2 O nor formation of the decomposition co-product acetaldehyde was observed in the product spectra suggesting that the contribution from this channel was negligible. The absence of CF 2 O also shows that no other reaction product was oxidized through to CF 2 O in the reaction system.
The absorption bands in the residual product spectrum from the reaction of Cl with ECDFA are not as distinct as in the residual product spectrum for Cl with MCDFA with more overlapping of bands which suggests more than one compound is present. Indeed, a comparison of the residual product spectrum from Cl with MCDFA, which has been identied as belonging mainly to the anhydride ClCF 2 C(O)OCHO, with that of Cl with ECDFA shows that ClCF 2 C(O)OCHO is formed in the reaction of Cl with ECDFA. Aer subtraction of the absorptions due to ClCF 2 C(O)OCHO, the following absorptions remain at $1864, $1819, $1302, $1169, $1125, $1093 (shoulder) and $978 cm À1 . The two carbonyl absorptions at 1864 and 819 cm À1 are separated by 45 cm À1 , whereby 1819 cm À1 is the stronger of the two. As discussed for the assignment of CClF 2 C(O)OC(O)H as a product in the reaction of Cl with MCDFA, bands at $1302, $1169 and $978 cm À1 are consistent with C-F and C-Cl vibrations from the ClF 2 C-group and a strong absorption at $1125 cm À1 would t with a C-O-C type vibration. These observed absorptions are consistent with the formation of the anhydride CF 2 ClC(O)OC(O)CH 3 from the reaction of CF 2 ClC(O) OCHO$CH 3 radical with O 2 (Fig. 4, channel B). This anhydride is not commercially available for quantication purposes, however, the high yield of CF 2 ClC(O)OH and evidence for the formation of CF 2 ClC(O)OC(O)H suggest that its yield is probably <10%.

Atmospheric implications
Methyl chlorodiuoracetate and ethyl chlorodiuoroacetate are removed from the troposphere mainly by their rapid reactions with OH radicals 16 which results in atmospheric lifetimes of the order of 10-50 days. However, in some marine and continental regions close to industrial emissions, peak values of 1 Â 10 5 atoms per cm 3 for Cl atoms concentrations have been observed. [3][4][5][6][7][8][9][10][11] Under such circumstances the reaction of Cl with the chlorouoroacetates can be a signicant if not dominant This journal is © The Royal Society of Chemistry 2016 loss process for the compounds. Photolysis is not an important atmospheric loss process for these chlorouorinated compounds since they do not absorb radiation at actinic wavelengths. 16 As reported for other haloesters 44,45 we expect relatively low values of the Henry laws constants at 298 K for both compounds, indicating that wet deposition of these chloro-uoro acetates will not be signicant compared to the OH-radical or Cl-atom initiated degradation. The volatility of these compounds will render dry deposition an unlikely removal mechanism.
The temperature dependences of the Cl-rate coefficients reported in this work were used to calculate the rate of loss of the chlorouoroacetates studied by their tropospheric reaction with Cl atoms. At a given temperature (T) corresponding to a given altitude in the troposphere, the rate of loss of the chloro-uoracetates is dened as the product between the Cl rate coefficient (T) and the [Cl] at this altitude.
Using a 24 h daytime average global tropospheric Cl atom concentration of [Cl] ¼ 1 Â 10 4 atoms per cm 3 , 46 the Arrhenius parameters reported in this work and considering a lapse rate in the troposphere of À6.5 K km À1 , 47 we have calculated the temperature prole between 0 and 10 km. Table S1 † shows the Cl removal rates for the chlorouoroacetates studied as a function of altitude in the troposphere considering temperature of 298.15 K at 0 km. The results showed in Table S1 † are plotted in the Fig. 5.
The positive temperature dependence of the reactions of Cl atoms with both MCDFA and ECDFA is consistent with the dominance of a hydrogen atom abstraction mechanism.
The product investigations presented in this work have shown that the major product of the Cl-atom initiated oxidation of methyl chlorodiuoroacetate under NO x -free conditions will be formic dichlorouoroacetic anhydride (CF 2 ClC(O)OC(O)H). We expect that chlorouoro anhydride will be further oxidized fairly rapidly by reaction with OH radicals and may also photolyze since the compound contains the -C(O)H chromophore. The other product observed in the reaction of Cl with MCDFA was chlorodiuoracetic acid ($30%). Both, anhydrides and acids are highly soluble compounds and may be rapidly incorporated into cloud droplets (where hydration of the anhydrides leads to acids) and rain out, contributing to the precipitation acidity.
For the reaction of Cl with ETFA the main product observed was chlorodiuoroacetic acid ($88%). Haloacetic acids are ubiquitous environmental pollutants and their presence in the environment is due both to natural processes and to human activities. They partition into water because of their large Henry's law constants. 48,49 They could be accumulated to relatively high concentrations in aqueous reservoirs and would likely accumulate to toxic levels. Furthermore, haloacetic acids have been detected in many places such as soil, conifer needles and lichens, snow and ice in Antarctica, fog and rainwater, sea water, lake water, wastewater and drinking water. 50,51 Haloacetic acids, as chlorodiuoracetic acid are generally phytotoxic and toxic to green algae, 52 while some haloacetic acids, e.g., dichloroacetic acid and trichloracetic acid, have hepatocarcinogenic potential. 53 In this sense, studies of the tropospheric degradation of chlorouoroacetates are important and necessary, since these compounds break down relatively rapidly to several products including persistent substances such as chlorodiuoroacetic acid. Concentrations of CDFA with a maximum of 0.0002 ng L À1 were reported in water samples taken in Canada. 48 However, a continuing monitoring of the environmental concentrations of CDFA is recommended to conrm trends in surface and rainwater.
In addition, further studies on the product distributions from the OH-radical and Cl-atom initiated oxidation of haloesters need to be performed over the range of temperatures and NO x levels typically prevailing in the troposphere in order to gain a better quantitative understanding of their contribution to photooxidant formation and precipitation acidity.