Document XpNVzqMNJpdGKVGYQJ4R6XgR

Environ. Sci. Technol. 2001, 35, 114-120 CI Atom-Initiated Oxidation of Three Homologous Methyl Perfluoroalkyl Ethers KAYO NOHARA,* , t M ITSU H IRO TOM A,t SHUZO KUTSUNA,t KOJ I TAKEUCHI ,t AND TAKASHI IBUSUKI t New Refrigerants Development Group, Research Institute of Innovative Technology for the Earth (RITE), and National Institute for Resources and Environment (NIRE), 16-3 Onogawa, Tsukuba, I baraki 305-8569, Japan Chlorine atom-initiated photooxidations of three homologous methyl perfluoroalkyl ethers (HFEs), n-CnF2n+10CH3 (n = 2, 3, and 5), in air in the absence of NO, were investigated with a long path FTIR/photochemical reaction system to elucidate the degradation mechanisms. The environmental removal processes of these three ethers in the troposphere were estimated. For oxidation of the three ethers, perfluoroalkyl formates (CnF2n-pi.0CH0; n = 2, 3 and 5) as relatively stable intermediates were produced at unity of the production ratio, which was independent of the perfluoroalkyl length. The rate constants for the reaction of CI atoms with C2F50CH0, C3F70CH0, and C5F110CH0 were (1.2 0.5) x 10-14, (1.2 0.5) x 10--N, and (1.8 0.7) x 10-14 cm3 molecule-1s-1, respectively. The rate constants of the reaction of CI with produced perfluoroalkyl formates were larger than these of perfluoroalkyl ethers. The formyl group of the perfluoroalkyl formates was finally converted to carbon dioxide. The --CF2-- of the perfluoroalkyl groups for the three ethers was mainly converted to C0F2 through the C --C cleavage; the conversion ratios from the carbons of the perfluoroalkyl group to C0F2 were 48 10, 76 10, and 60 10% for C2F50CH3, n-C3F70CH3, and n-05F110CH3, respectively. Sixteen percent of the perfluoroalkyl group for n-C3F70CH3 was converted to C2F5C0F. Similarly, the perfluoroalkyl group of n-05F110CH3 was converted to CnF2n-pi.00F (n = 2, 3, and/or 4) with the yield of 15-30%, while for C2F50CH3, the formation of CF3C0F was not confirmed. As an oxidation product of the terminal CF3-- group, 20, 22, and 16% of the CF3 group for C2F50CH3, n-C3F70CH3, and n-05F110CH3, respectively, were converted to CF3000CF3. Introduction Development of alternatives to hydrofluorocarbons (HFCs) and hydrochlorofluorocarbons (HCFCs) has been extensively conducted throughout the world, becauseHFCs and HCFCs will be regulated due to their global warming potentials (GWPs) and/or stratospheric ozone depletion potentials (ODPs) (1,2). Hydrofluorinated ether compounds (H FEs) (3, 4) are considered to be promising materials as new alterna- *Correspondivat ior phone: ; fax: e-mail: g -lire.go.jp. t Research InstituteofInnovativeTechnologyfor theEarth (RITE). National Institute for Resources and Environment (NIRE). 114 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 35, NO. 1, 2001 tives. Their ODP is zero, because they contain no chlorine atoms. RITE (Research Institute of Innovative Technology for theEarth) has been developingthree homologousmethyl perfluoroalkyl ethers (n-CriF2n+1OCH3; n = 2, 3, 5) (5) as refrigerants and/or blowing agents. The rate constants for C2F5OCH3 and n-C3F7OCH3 for the reaction with the OH radical at 273 K are 0.819 0.025 x 10- N cm3 molecule-1s-1 and 0.741 0.023 x 1O-14 cm3 molecule-1s-1(6) which were measured bythe laser induced fluorescencetechnique(LIF). Their tropospheric Iifetimesestimated by usingthe 7.5 x 1O5 molecule cm-3 of the concentration of [OH] (7) is thus less than 6 years. If they are degraded in the troposphere to form relativestablebyproducts,it would inducetheenvironmental impacts in addition to global warming. The research of degradation processes for H FEs in troposphere is important for the environmental assessment of HFEs. I n this study, the oxidations of the three ethers were initiated by CI atoms. The changes in the concentration of the ethers and their degradation products were monitored by an FTIR spectrometer. It was revealed that the perfluoroalkyl formate was formed as a major intermediate in the reaction. The final products observed were CO2 (produced from CH3 group), COF2, CF3OOOCF3, and CnF2n-A.COF (n = 2, 3, and/or 4) which were produced from the perfluoro alkyl group. The concentrations of all the final products were determined, to check the mass balance of the reactions and to elucidate the reaction mechanisms. The kind and distribution of the products from the perfluoro alkyl groups were found to be dependent on the chain length. The fate of the degradation products in the environment was discussed in connection with their environmental impact. Experimental Section The experimental apparatus used here has been previously described in detail (8). Experiments were carried out using a 25-L evacuable quartz cylinder (14cm in diameter, 150 cm long), surrounded with four 40-W black lamps, equipped with an in situ multiple reflection optical system (30-m path length) interfaced to a J EOL J I R100 FTIR spectrometer. The IR spectra of the reactants and products were measured over the frequency range from 700 to 3500 cm-1 at 0.5 cm-1 resolution by 50 scans/ interferograms. Liquid fluorinated ethers were injected into the chamber with a microsyringe. The prescribed amount of Cl2 (2.46 x 1014 molecule cm-3) was taken in a gas cell and then introduced into the reaction chamber. Each of the HFEs--C12 mixtures was diluted to atmospheric pressure (1.01x 1O5 Pa) with synthetic air (80% N 2 + 20%O2) at 298 3 K. Theinitial concentrations ofHFEs was 1.65 x 1O14, 1.23 x 1O14, and 2.16 x 1O14 molecule cm-3 for C2F5OCH3, n-C3F7OCH3, and n-05FuOCH3, respectively. Theoxidation of HFEs was initiated by CI atoms, which were generated by photolysis of Cl2. Three kinds of methyl perfluoroalkyl ethers (C2F5OCH3, 99.9%; n-C3F7OCH3, 99.8%; n-05FEOCH3, 99.8%) were supplied from RITE and used without further purification. Results and Discussion The Identification of the Degradation Products by the IR Spectra. IR spectra of n-05FuOCH3 (2.16 x 1O14 molecule cm-3) and Cl2 (2.46 x 1014 molecule cm-3) in 1.01 x 105 Pa of air are shown in Figure 1(a) (before irradiation) and (b) (after 30 min irradiation). Figure 1(c) is the subtracted spectrum (1(b) -- 1(a)). After 30-min irradiation, the peaks in the range of 1875-1978 cm-1and split peaks about 2300 cm-1 appeared, which were attributed to COF2 and CO2, 10.1021/es000895f CCC: $20.00 2001 American Chemical Society Published on Web 11/28/2000 TABLE 1. Comparison of Calculated and Observed Frequencies (cm-1) for the IR Spectrum of Perfluoroalkyl Formates CHdO (c alc d)a CHdO (exptl) C2F5OCHO anti syn C3F7OCHO anti syn C5F11OCHO anti syn 1827.4 1806.0 1827.0 1805.0 1827.3 1805.0 1822 1805 1822 1805 1822 1805 a Ab initio MO calculation with RHF/6-31G(d, p) frequencies in cm-1, scaled by 0.873. FIGURE 1. IR spectra in the frequency region of 700-3500 cm-1 obtained from the photolysis of a mixture containing C5F11OCH3 (2.16 1014 molecule cm-3) and Cl2 (2.46 1014 molecule cm-3) in 1.01 105 Pa of air: (a) before irradiation; (b) after 30-min irradiation; and (c) differential spectrum. FIGURE 2. Comparison w ith the three subtracted IR features (17002100 cm-1) obtained from the photolysis of the gas mixture: (a1) C5F11OCH3 (2.16 1014 molecule cm-3) after 180-min irradiation, (b1) C3F7OCH3 (1.23 1014 molecule cm-3) after 120-min irradiation, and (c1) C2F5OCH3 (1.65 1014 molecule cm-3) after 60-min irradiation w ith Cl2 (2.46 1014 molecule cm-3) in 1.01 105 Pa of air. (a2), (b2), and (c2) are the subtracted IR features from each of the (a1), (b1), and (c1) from the IR spectra of COF2. respectively. Carbon monoxide was not detected in this experiment.Theabsorption bandsof HCl wereobserved from 2726 to 3014 cm-1. The IR absorption of intermediates appeared at 1805 and 1822 cm-1. Similar experiments were conducted for C2F5OCH3 (1.65 1014 molecule cm-3) or C3F7OCH3(1.23 1014 molecule cm-3) mixed with Cl2(2.46 1014moleculecm-3).IRspectrum of C5F11OCH3 in the range of 1700 to 2050 cm-1 at 180 min illumination, that of C3F7OCH3 after 120-min, and that of C2F5OCH3 after 60-min irradiation are shown in Figure 2 [parts (a1), (b1), and (c1), respectively]. The peaks in the rangeof 1875-1978cm-1 for COF2 weredetected as thefinal product. Two IR peaks at 1805 and 1822 cm-1 for intermediates also appeared for the oxidation of both C2F5OCH3 and n-C3F7OCH3, as similarly to n-C5F11OCH3. The absorption bands of the intermediates other than these two peaks were not observed in the oxidation of three ethers. Wallington et al. (7) calculated thevibrational frequencies of syn and anti structuresof CF3OCHO byab initio molecular orbital calculation with RHF/ 6-31G(d,p) to confirm the carbonyl stretchingbands(CHdO) at1800cm-1and 1823cm-1 for syn and anti structures, respectively. Therefore, the observed two peaks at 1805 cm-1 and 1822 cm-1 in the oxidation of three ethers can be attributed to the perfluoroalkyl formates. To estimate the vibrational frequencies of the carbonyl stretching bands (CHdO) for the three perfluoroalkyl formates (C2F5OCHO, C3F7OCHO, and C5F11OCHO), the syn and anti structures of these three perfluoroalkyl formates were calculated by the ab initio molecular orbital calculation with restricted Hartree-Fock (RHF), basis set of 6-31G-(d,p), by using the Gaussian 94 program. The calculated spectra are corrected by the scale factor of 0.873 which has been proposed for CH3OCHO (7). The results of thiscalculation aresummarized in Table1.Calculated spectra for the CHdO vibrational stretching bands for syn and anti of the three kinds of perfluoroalkyl formates show the frequenciessimilar to theexperimental ones. Thevibrational frequencies of CHdO for syn and anti of three perfluoroalkyl formates are almost same. For the oxidation of C2F5OCH3, n-C3F7OCH3, and n-C5F11OCH3, theobserved spectraat 1805 and 1822 cm-1 are thus attributed to the CHdO bands of syn and anti of C2F5OCHO, C3F7OCHO, and C5F11OCHO, respectively. Parts (a2), (b2), and (c2) of Figure 2 are obtained by subtracting the spectrum around 1900 cm-1 of COF2 from each of the IR spectra of (a1), (b1), and (c1). An absorption at1889cm-1wasconfirmed for both n-C3F7OCH3and n-C5F11OCH3 degradation. The peak at 1889 cm-1 was the CdO stretching band of C2F5COF, which was confirmed by using the authentic sample. Although the CdO stretching band of CF3COF has been reported to appear at 1898 cm-1 (9), the absorption at 1898cm-1(CF3COF) wasnot detected in Figure 2 [parts (a2), (b2), and (c2)]. The absorption at 1889cm-1 for n-C3F7OCH3 can be attributed to C2F5COF. The peak at 1889 cm-1 for n-C5F11OCH3 may be attributed to C2F5COF, C3F7COF, and/ or C4F9COF, taking into account the degradation processes as described later, if the CdO stretching band of CnF2n+1COF (n g 2) is independent of the perfluoroalkyl length. The Quantification of the Intermediates and Final Products.Theproductsother than theperfluoroalkyl formate were not observed at the initial reaction period for the oxidation of thethreeethers. Assumingthat CO2 isproduced from the oxidation of -CHO group for the perfluoroalkyl formates, the values which are derived by subtracting the CO2 concentration formed from the loss of n-C5F11OCH3 ([HFEs]-[CO2]) would be same as the concentration of the VOL. 35, NO. 1, 2001 / ENVIRONM ENTAL SCIENCE & TECHNOLOGY 9 115 FIGURE 3. The plots of IR peak area in the range of 1798-1812 cm-1 for syn-C5F11OCHO (]) at arbitrary unit and the concentration of [C5F11OCH3]-[CO2] ([), and the concentration of CO2 (b) versus the loss of C5F11OCH3 in the photolysis of the gas mixture of C5F11OCH3 (2.16 1014 molecule cm-3) w ith Cl2 (2.46 1014 molecule cm-3) in 1.01 105 Pa of air. perfluoroalkyl formates. For the oxidation of n-C5F11OCH3, the concentrations of ([n-C5F11OCH3]-[CO2]) are plotted against the loss of n-C5F11OCH3 in Figure 3. Since the absorption coefficient of the perfluoroalkyl formate has not been reported and evaluated, the concentration of the perfluoroalkyl formate cannot be calculated. Therefore, the peak area in the range of 1798-1812 cm-1 which has been assigned as the CHdO stretching band of syn-C5F11OCHO is also plotted against the[n-C5F11OCH3]in Figure3. Theplots of the peak area at an arbitrary unit could be adjusted to those of ([HFEs]-[CO2]). The profile of plots of [n-C5F11OCH3]-[CO2] is good agreement with that of the peak area in the range of 1798-1812 cm-1. For the oxidation of both C2F5OCH3 and n-C3F7OCH3, the profile of plots of [HFEs][CO2] were also similar to that of the peak area in the range of 1798-1812 cm-1. This result supports that the perfluoroalkyl formate is the sole primary intermediate for the oxidation of the three ethers, and the CO2 is quantitatively produced by thedegradation of methyl perfluoroalkyl ether. It has been reported that syn-CF3OCHO is more stable than anti byabout1kcal which impliestherelativepopulation of 80% syn and 20% anti (7). The C5F11OCHO may exist at the relative population of 80%syn and 20%anti on the basis of reference about CF3OCHO. Given that the 80% of ([nC5F11OCH3]-[CO2]) corresponds to theconcentration of synC5F11OCHO, theabsorption coefficient of IR absorption band in the range of 1798-1812 cm-1 for the CHdO stretching band of syn-C5F11OCHO is calculated to be -(base e) ) 3.9 10-17 cm2 molecule-1. To confirm whether this absorption coefficient isreasonablefor theCHdO vibration band of synC5F11OCHO, the infrared absorption coefficient in the range of 1798-1812 cm-1 was estimated by the following steps. IR absorption coefficient in the range of 1680-1820 cm-1 for CHdO vibration band of CH3OCHO wasmeasured as-(base e) ) 4.97 10-17 cm2 molecule-1 (a). The IR intensity for CdO vibration band of CH3OCHO calculated byusingtheab initio MO calculation with RHF/ 6-31G-(d,p) was 452 km/ mol (b). The ratio of (a)/ (b), the experimental IR absorption coefficient to the calculated IR intensity for CH3OCHO, was utilized to evaluatetheIR absorption coefficient of syn-C5F11OCHO as follows; the IR intensity for the CHdO vibration band (1798-1812 cm-1) of syn-C5F11OCHO calculated with the ab initio MO was multiplied by (a)/ (b). The calculated IR absorption coefficient for syn-C5F11OCHO was obtained 116 9 ENVIRONM ENTAL SCIENCE & TECHNOLOGY / VOL. 35, NO. 1, 2001 FIGURE 4. Change in the ratio of the concentrations of CnF2n+1OCHO (2, 3, and 5) to the initial HFEs concentrations versus the fractional loss of CnF2n+1OCH3 (2, 3, and 5) in the photolysis of HFEs w ith Cl2 (2.46 1014 molecule cm-3) in 1.01 105 Pa of air: [, C2F5OCH3, b, C3F7OCH3 and 4, C5F11OCH3. The initial concentrations of C2F5OCH3, C3F7OCH3, and C5F11OCH3 are 1.65 1014, 1.23 1014, and 2.16 1014 molecule cm-3, respectively. to be -(base e) ) 3.1 10-17 cm2 molecule-1, which is not so different from the experimental derived value of -(base e) ) 3.9 10-17 cm2 molecule-1. The rate constants of the perfluoroalkyl formates with the Cl atom can be evaluated using eq I (10) [formate]t/ [HFE]t0 ) [R/ (1 - k2/ k1)](1 - x)[(1 - x)(k2/k1-1) -1] (I) x ) 1 - [HFE]t/ [HFE]t0 (II) where R is the yield of formate from perfluoroalkyl methyl ether (0 e R e 1), and k1 and k2 are the secondary rate constants of reactions 1 and 2, respectively. The relevant reactions are Cl + CnF2n+1OCH3 k1 98 R[CnF2n+1OCHO] + other product (1) Cl + CnF2n+1OCHO k2 98 HCl + CnF2n+1OC(O) (2) Ratios of the concentration of CnF2n+1OCHO (n ) 2, 3, and 5), which are used by the value of [CnF2n+1OCH3] - [CO2] (n ) 2, 3, and 5), to theinitial HFEsconcentration areplotted against the fractional loss of HFEs in Figure 4. Each rate constant (k1) of theH abstractions for theC2F5OCH3, n-C3F7OCH3, and n-C5F11OCH3 by Cl atom was measured by the method of relative rate constant study (11). The kinetics of (HFEs + Cl) were measured relative to reaction (CH4 + Cl). Theinitial concentrationswere1.2-2.2 1014moleculecm-3 of the three ethers, 3.2-4.5 1014 molecule cm-3 of Cl, and 4.9 1014 molecule cm-3 of CH4 in the 1.01 105 Pa of air diluent. The observed loss of three ethers versus those of reference compounds in the presence of Cl atoms is shown in Figure 5. Absorption bands used were at 1462.2 cm-1 for C2F5OCH3, 1461.7 cm-1 for n-C3F7OCH3, 1462.0 cm-1 for n-C5F11OCH3, and 3017 cm-1 for CH4. The k1 values for three ethers were obtained by linear least-squares analysis of the data in Figure 5. Using the rate constant for CH4 + Cl as 1.0 10-13cm3molecule-1s-1(12), thesek1for C2F5OCH3, n-C3F7OCH3 and n-C5F11OCH3 are (11.0 ( 1.4) 10-14 cm3 FIGURE 5. The loss of three ethers, C2F5OCH3 ([), C3F7OCH3 (b), and C5F11OCH3 (4), versus the loss of CH4 compounds in the presence of Cl atoms in the 1.01 105 Pa of air diluent. TABLE 2. k2/k1, k1, and k2 Values for the Degradation of the Three Ethers ( 10-14 c m3 molecule-1 s-1) CF3OCH3 C2F5OCH3 C3F7OCH3 C4F9OCH3 C5F11OCH3 k2/k1 0.11 ( 0.02 0.10 ( 0.02 0.17 ( 0.07c 0.17 ( 0.06 0.09 ( 0.02b k1 14 ( 0.2a 11.0 ( 1.4 11.8 ( 1.4 9.7 ( 1.4c 10.3 ( 1.4 9.1 ( 1.3b k2 0.98 ( 0.12a 1.2 ( 0.5 1.2 ( 0.5 1.6 ( 0.7c 1.8 ( 0.7 0.82 ( 0.22b a The data for the Cl atom initiated oxidation of CF3OCH3 by L. K. Christensen et al. (11). b The data for the Cl atom initiated oxidation of C3F7OCH3 by Y. Ninomiya et al. (13). c The data for the Cl atom initiated oxidation of C4F5OCH3 by T. J . Wallington et al. (7). molecule-1 s-1, (11.8 ( 1.4) 10-14 cm3 molecule-1 s-1, and (10.3 ( 1.4) 10-14 cm3 molecule-1 s-1 (see Table 2), respectively. The data in Figure 4 were fitted nonlinearly by eq I to obtain the k2/ k1 which are summarized in Table 2. Using k1 values which was obtained in our experiments, the rate constants (k2) for reaction of Cl atoms with the formates C2F5OCHO, C3F7OCHO, and C5F11OCHO were calculated to be (1.2 ( 0.5) 10-14, (1.2 ( 0.5) 10-14, and (1.8 ( 0.7) 10-14 cm3 molecule-1 s-1, respectively (see Table 2). The rate constants of the reaction of Cl with produced perfluoroalkyl formateswerelarger than theseof perfluoroalkyl ethers. The k2/ k1, k1, and k2 for the Cl atom initiated degradations of CF3OCH3 (11), n-C4F9OCH3 (7), and n-C3F7OCH3 (13) which have been reported by other groups were also depicted in Table 2. Measured k1 values for three ethers were similar to that of n-C4F9OCH3. The values of k2 (Cl + C5F11OCHO) were also similar to that of k2 (Cl + C4F9OCHO) and were larger than that of k2 (Cl + C2F5OCHO) and k2 (Cl + C3F7OCHO). Thevalueof k2 (Cl + C3F7OCHO) observed in our experiment indicated thesimilar valueof k2 (Cl + C3F7OCHO) which was reported by Ninomiya et al. in the range of experimental error. Figure 6(a),(b) shows the change in the concentration of CO2 and perfluoroacetyl fluoride, respectively, with the loss of HFEs (HFEs) for the degradation of C2F5OCH3 (1.65 1014moleculecm-3),n-C3F7OCH3(1.23 1014moleculecm-3), and n-C5F11OCH3(2.16 1014moleculecm-3) under thesame conditions as in Figure 3. For the oxidation of three ethers, FIGURE 6. The plots of the production yields of CO2 (a) and perfluoroalkylacetyl fluoride (b) versus the fractional loss of HFEs (HFEs) for the photolysis of C2F5OCH3 ([), C3F7OCH3 (b), and C5F11OCH3 (4). The initial concentrations of C2F5OCH3, C3F7OCH3, and C5F11OCH3 are 1.65 1014, 1.23 1014, and 2.16 1014 molecule cm-3, respec tively. the formations of CO2 were not observed in the initial loss of HFEs. CO2 was produced after approximately 20% loss of HFEs. When the loss of HFEs reached 100%, the production of CO2 continues to increase. The final concentration of CO2 for C2F5OCH3, n-C3F7OCH3, and n-C5F11OCH3 were 1.59 1014 molecule cm-3, 1.18 1014 molecule cm-3, and 2.10 1014 molecule cm-3, respectively, which was the same as the initial concentration of each ether. The production of perfluoroalkylacetyl fluoride was not observed at the initial reaction period. For n-C3F7OCH3, perfluoroalkyl acetyl fluorideappeared after a60%lossof n-C3F7OCH3.Concerning n-C5F11OCH3, perfluoroalkyl acetyl fluoride began to be produced after a 32% loss of n-C5F11OCH3. The final concentrations of perfluoroalkyl acetyl fluoride for n-C3F7OCH3 and n-C5F11OCH3 were 2.16 1013 molecule cm-3 and 8.11 1013 molecule cm-3, respectively. For the degradation of n-C3F7OCH3, only C2F5COF is produced as a perfluoroalkyl acetyl fluoride. Theconversion ratio of the perfluoroalkyl group to C2F5COF was 11.7%. For n-C5F11OCH3, theproduced perfluoro acetyl fluoridemaybe C2F5COF, C3F7COF, and/ or C4F9COF as mentioned in the previous section. The actual conversion ratio cannot be calculated, but the conversion ratio would be 15%, if only C2F5COF would beproduced. Similarly, given that onlyC4F9COF was produced, the ratio would be 30%. Therefore, the conversion ratio is estimated to be in the range of 15-30%. Figure 7(a),(b) plots the COF2 and bis(trifluoromethyl)trioxide (CF3OOOCF3) formed, respectively, as a function of the loss of HFEs (HFEs) for the degradation of C2F5OCH3 (1.65 1014 molecule cm-3), n-C3F7OCH3 (1.23 1014 moleculecm-3),and n-C5F11OCH3(2.16 1014moleculecm-3) under the same conditions as in Figure 3. The subtracted IR feature(700-1400cm-1) for thereaction gasmixtureof C2F5OCH3 degraded after 90 min is shown in Figure 8. The absorption of CF3OOOCF3 appeared at 1292, 1252, and 1169 cm-1 (14, 15). The concentration of CF3OOOCF3 was calculated by using the absorption coefficient at 1292 cm-1 ( (basee) ) 4.1 10-18 cm2 molecule-1) (16). For theoxidation of three ethers, the formation of COF2 seems to have the same induction periods. The production of COF2 occurred after the approximately 50-60% loss of HFEs. The concentration of COF2 reached a constant value at 1.77 1014, 2.77 1014, and 6.81 1014 molecule cm-3 for C2F5OCH3, n-C3F7- VOL. 35, NO. 1, 2001 / ENVIRONM ENTAL SCIENCE & TECHNOLOGY 9 117 TABLE 3. Molar Conversion Ratio of the Perfluoroalkyl Group in the Ethers to the Products COF2 (% ) CnF2n+1COFa CF3OOOCF3 (% ) (% ) total conversion C2F5OCH3 48 ( 10 C3F7OCH3 76 ( 10 C5F11OCH3 60 ( 10 0 12 15-30 20 68 ( 10 22 110 ( 10 16 91 ( 10-106 ( 10 a CF3COF for C2F5OCH3, C2F5COF for C3F7OCH3, CnF2n+1COF (n ) 2, 3, and/or 4) for C5F11OCH3fhjki. SCHEME 1 FIGURE 7. The plots of the concentrations of COF2 (a) and CF3OOOCF3 (b) versus the fractional loss of HFEs (HFEs) obtained from the photolysis of C2F5OCH3 ([), C3F7OCH3 (b), and C5F11OCH3 (4). The initial concentrations of C2F5OCH3, C3F7OCH3, and C5F11OCH3 are 1.65 1014, 1.23 1014, and 2.16 1014 molecule cm-3, respec tively. FIGURE 8. The subtracted IR feature (700-1400 cm-1) of the gas mixture of C2F5OCH3 degraded after 60-min irradiation. OCH3, and n-C5F11OCH3, respectively. On the other hand, the production of CF3OOOCF3 was observed after approximately 90% loss of HFEs. The final concentrations of CF3OOOCF3 were 1.69 1013, 1.99 1013, and 1.92 1013 moleculecm-3for C2F5OCH3, n-C3F7OCH3, and n-C5F11OCH3, respectively. Concerning C2F5OCH3, the molar conversion for the carbon of the perfluoroalkyl group to COF2 was 48 ( 10%, meaningthat the-CF2group wascompletelyconverted to COF2, while the -CF3- group was not oxidized to COF2. For thedegradation of n-C3F7OCH3, theconversion from the perfluoroalkyl group to COF2 was 76 ( 10%. For the degradation of n-C5F11OCH3, theconversion from C5F11to COF2 was 60 ( 10%. Probably, a part of the CF3- moiety is transformed into theCF3OOOCF3.Theconversion ratiosfrom CF3- group to CF3OOOCF3 wereestimated as20%, 22%, and 16% for C2F5OCH3, n-C3F7OCH3, and n-C5F11OCH3, in that order. The conversion ratios of the perfluoroalkyl group for the ethers to COF2, perfluoroalkyl acetyl fluoride, and CF3OOOCF3 after 100% loss of the perfluoroalkyl formates are summarized in Table 3. For n-C3F7OCH3 and n-C5F11OCH3, 118 9 ENVIRONM ENTAL SCIENCE & TECHNOLOGY / VOL. 35, NO. 1, 2001 the mass balance of the degradation of the perfluoroalkyl group is about 100%. Scheme 1 of the Cl-initiated reaction of the methyl perfluoroalkyl ether (n-CF3CF2CF2OCH3) is estimated from the identified degradation products. The perfluoroalkyl ether iscompletelyconverted to perfluoroalkyl formate (n-CF3CF2CF2OCHO). The O-C bond cleavage of perfluoroalkyl formate takes place to produce the CO2 and perfluoroalkoxy radical (CF3CF2CF2O). The degradation of perfluoroalkoxy radical proceeds through the two different pathways to mainly produce the COF2 and has the minor production of the perfluoroacetyl fluoride (CF3CF2COF). Degradation Mechanism of the Methyl Perfluoroalkyl Ethers. The initial stages of oxidation are considered as follows: H-abstraction with aCl atom from themethyl group occursto producethealkyl radical (17), and, then, thealkoxy radical isformed viaO2addition and radical-radical reactions 3-5. Actually, the detection of C4F9OCH2 and C4F9OCH2O2 radicals for the degradation of HFE-7100 (C4F9OCH3) was reported (7). C2F5OCH3 + Cl f C2F5OCH2 + HCl (3) C2F5OCH2 + O2 f C2F5OCH2O2 (4) 2C2F5OCH2O2 f 2C2F5OCH2O + O2 (5) C2F5OCH2O + O2 f C2F5OCHO + HO2 (6) Hydrogen of C2F5OCH2O is abstracted by O2 to produce pentafluoroethyl formate (C2F5OCHO) as reaction 6. Relatively stable C2F5OCHO is oxidized through reactions 7-10 to giveC2F5O. TheC-C cleavageof C2F5O leadsto formation of the CF3 radical and COF2. For C3F7OCH3 and C5F11OCH3, C3F7O, and C5F11O are produced by a route analogous to the degradation processes of C2F5OCH3. Subsequently, the cleavage of C-C bond occurs to form C2F5 and C4F9 along with COF2. The C-C bond cleavage of C2F5 and C4F9 occurs gradually through such reactions as 12a-d and 14a-d, and the (n-1)COF2 (n ) 3and 5) and CF3 were produced finally. In another pathway,theperfluoro acetyl fluorideisproduced. The possible process of formation of it is that C3F7O may react with HO2 (18) to produce C3F7OH, which is thermally transformed to C2F5COF (19). CnF2n+1O (n ) 5, 4, and 3) also may react with HO2 to produce Cn-1F2n-1COF (n ) 5, 4, and 3) as 15a-b. However, the processes of formation of the perfluoroacetyl fluoride have not been revealed. Further discussion of these formation processes will be needed. C2F5OCHO + Cl f C2F5OCO + HCl (7) C2F5OCO + O2 f C2F5OC(O)OO (8) 2C2F5OC(O)OO f 2C2F5OCOO + O2 (9) C2F5OCO2 f C2F5O + CO2 (10) C2F5O f CF3 + COF2 In the case of C3F7OCH3 (11) C3F7O f C2F5 + COF2 (12a) C2F5 +O2 f C2F5O2 (12b) 2C2F5O2 f 2C2F5O + O2 (12c) C2F5O f CF3 + COF2 (12d) C3F7O + HOO f C3F7OH + O2 C3F7OH f C2F5COF + HF In the case of C5F11OCH3 (13a) (13b) C5F11O f C4F9 + COF2 (14a) C4F9 + O2 f C4F9O2 (14b) 2C4F9O2 f 2C4F9O + O2 (14c) C4F9O f f f CF3 + 3COF2 (14d) CnF2n+1O + HO2 f CnF2n+1OH + O2 (15a) (n ) 5, 4, and 3) CnF2n+1OH f Cn-1F2n-1COF + HF (15b) CF3 radicals react with oxygen to produce the CF3O2 radicals (20) as shown in reaction 16, and, subsequently, the bimolecular reaction of CF3O2 (reaction 17) takes place to produce the CF3O (21-23). The CF3O radical reacts with CF3O2 to produce CF3OOOCF3 as shown in reaction 18. CF3 + O2 f CF3O2 (16) 2CF3O2 f CF3O + O2 (17) CF3O2 + CF3O f CF3OOOCF3 (18) FIGURE 9. 1H NM R spectrum of the photolysis mixture of C5F11OCH3 (12.3 1014 molecule cm-3) and Cl2 (49.2 1014 molecule cm-3) in 1.01 105 Pa of air sampled in CDCl3 after 30-min irradiation. Estimation of Environmental Impacts of the DegradationProducts.Formation of variouskindsof relativelystable intermediatesand degradation productshasbeen confirmed in this study. The Cl-initiated oxidation of three methyl perfluoroalkyl ethers (n-CnF2n+1OCH3; n ) 2, 3, and 5) produces the corresponding perfluoroalkyl formates (nCnF2n+1OCHO; n ) 2, 3, and 5) as the sole product. This formation processof theperfluoroalkyl formatesisinsensitive to the difference in the perfluoroalkyl length of the compounds. In general, the reaction rate of organic compounds with Cl atoms is faster than that with OH radicals (12). Therefore, the lifetime of the perfluoroalkyl formates in the tropospherewasestimated byassumingthek2valuesin Table 2 as an upper limit of the reaction rate constant of OH + CnF2n+1OCHO. Using the average tropospheric OH concentration of 7.5 105 molecules cm-3 (7), the corresponding lifetimes were calculated to be longer than 3.5 ( 0.10, 3.6 ( 0.10, and 2.4 ( 0.3 years for C2F5OCHO, C3F7OCHO, and C5F11OCHO, respectively, assuming a temperature of 298 K. Since the lifetimes of the formates seem to be relatively long with respect to reaction with OH radicals, tropospheric removal processes by water droplets such as clouds and/ or rain will be important. A reaction gas mixture of n-C5F11OCH3(1.23 1015moleculescm-3)/ Cl2(4.92 1015molecules cm-3)/ air (1.01 105 Pa) which was irradiated for 30 min was taken by a stainless steel cylinder cooled at 203 K. The collected samplewastransferred to CDCl3solution in aNMR sample tube and was analyzed by a 1H and 19F-NMR spectrometer. The spectrum of 1H NMR shows that water is present in thesample(seeFigure9).In the1H NMR spectrum, the peak of the formyl proton appeared at 8.03 ppm. The peak at 119ppm in 19F NMR wasassigned asF atom of -CF2COOH. It is thus implied that the C5F11OCHO produced in the gas-phase reaction is hydrolyzed to CF3(CF2)4OH and formic acid in the water which contained HCl in the steel cylinder.NoNMRpeakswhich couldbeassignedtoCF3(CF2)4OH wereobserved. CF3(CF2)4OH, which isunstable(23), may be readily converted to CF3(CF2)3COF, which may be further hydrolyzed to produce a perfluoro carboxylic acid. Further studies on an uptake and a chemical reaction of the perfluoroalkyl formates by water will be desired to estimate their environmental impacts including their GWP values. The formyl group of perfluoroalkyl formate is oxidized to CO2 via C-C cleavage in the presence of oxygen. The perfluoroalkyl group mainly reacts with O2 to produce RfCF2O, followed by C-C bond cleavage to form COF2. However, when the carbon number of the perfluoroalkyl group is larger than 2, e.g. n-C3F7- and n-C5F11-, their correspondinga few perfluoro alkoxy radicals areconverted to perfluoroacetyl fluorides. The terminal CF3- group is oxidized to produce CF3OOOCF3 through the CF3O in the chamber experiment. However, CF3OOOCF3 is unlikely to VOL. 35, NO. 1, 2001 / ENVIRONM ENTAL SCIENCE & TECHNOLOGY 9 119 form in the troposphere because the concentration of the CF3O and CF3O2 is considered to be very low. As has been reported, the COF2 and CF3COF have the values larger than 0.5 of Hk, where H is the Henry's law constant and k is the hydrolysis rate constant and their tropospheric lifetimesbyclouds/ rain areestimated lessthan 30 days (24). For the perfluoroalkyl halides (CnF2n+1COF, n > 2), the values of Hk have not been determined but may be similar to those for COF2 and CF3COF. It is thus inferred that the perfluoroalkyl halides may not contribute to the global warming, because their short lifetimes will be estimated. However, it is to be noted that the perfluoroalkyl halides will be converted to perfluoroalkyl carboxylic acids and HF, which are suspected to cause ecological impacts. In summary, wehavethepotential degradation products of the three ethers in troposphere. All the products except CO2 are likely to be removed by clouds/ rainwater, taking into account theratesof their gas-phasereactionsestimated. The products themselves may not cause human health and ecological systems, but more attention should be paid to perfluoroalkyl carboxylic acids and HF. Acknowledgments The author (K. Nohara) thanks H. Yamamoto of Asahi Glass Company for the advice on computational calculations. Literature Cited (1) Molina, M. B.; Rowland, F. S. Nature 1974, 249, 810. (2) Fisher, D. A.; Hales, C. H.; Filkin, D. L.; Ko, M. K. W.; Sze, N. D.; Connel, P. S.;Wuebbles, D. J.;Isaksen, I. S. A.;Strordal, F. Nature 1990, 344, 508. (3) Sekiya, A.; Misaki, S. Chemtech 1996, 26, 44. (4) Cavalli, F.; Glasius, M.; Hjorth, J.; Rindone, B.; Jensen, N. R. Atmos. Environ. 1998, 32, 3767. (5) Heathfield,A.E.;Anastasi,C.;Pagsberg,P.;McCulloch,A,Atmos. Environ. 1998, 32, 711. (6) Tokuhashi, K.; Takahashi, A.; Kaise, M.; Kondo, S.; Sekiya, A.; Yamashita, S.; Ito, H. Int. J. Chem. Kinet. 1999, 31, 846. (7) Wallington, T. J.; Schneider, W. F.; Sehested, J.; Bilde, M.; Platz, J.; Nielsen, O. J.; Christensen, L. K.; Molina, M. J.; Molina, L. T.; Wooldridge, P. W. J. Phys. Chem. A 1997, 101, 8264. (8) Itoh, K.; Kato, J.; Nakayama, Y.; Kutsuna, S.; Koike, K.; Ibusuki, T. Chemosphere 1994, 29, 1701. (9) Kutsuna, S.; Toma, M.; Takeuchi, K.; Ibusuki, T. Environ. Sci. Technol. 1999, 33, 1071. (10) Meagher, R. J.; McIntosh, M. E.; Hurly, M. D.; Wallington, T. J. Int. J. Chem. Kinet. 1997, 29, 691. (11) Christensen, L. K.; Wallington, T. J.; Guschin, A.; Hurley, M. D. J. Phys. Chem. A 1999, 103, 4202. (12) DeMore, W. B.; Sander, S. P.; Golden, D. M.; Hampson, R. F.; Kurylo, K. J.; Howard, C. J.; Ravishankara, A. R.; Kolb, C. E.; Molina, M. J. JPL Publication No 97-4; NASA Propulsion Laboratory: Pasadena, CA, 1997. (13) Ninomiya, Y.; Kawasaki, M.; Guschin, A.; Molina, L. T.; Molina, M. J.; Wallington, T. J. Environ. Sci. Technol. 2000, 34, 2973. (14) Tompson, P. G. J. Am. Chem. Soc. 1967, 89, 4316. (15) Hirschmann, R. P.; Fox, W. B.; Anderson, L. R. J. Phys. Chem. A 1969, 25A, 811. (16) Nielsen, O.J.; Ellermann, T.; Sehested, J.; Bartkiewicz, E.; Wallington, T. J.; Hurley, M. D. Int. J. Chem. Kinet. 1992, 24, 1009. (17) Tuazon, E. C.;Atkinson, R. Environ. Sci. Technol. 1994, 24, 2306. (18) Kormann, C. W.; Bahnemann, D.; Hoffmann, M. R. Environ. Sci. Technol. 1991, 25, 494. (19) Ishikawa, N.; Kobayashi, Y. Fluorine Chemistry; published in Koudansha, printed in Japan, 1979; p 101. (20) Nielsen, O. J.; Ellermann, T.; Sehested, J.; Bartkiewicz, E.; Wallington, T. J.; Hurley, M. D. Int J. Chem. Kinet. 1992, 24, 1009. (21) Wallington. T. J.; Hurley, M. D.; Schneider, W. F.; Sehested, J.; Nielsen, O. J. J. Phys. Chem. A 1993, 97, 7606. (22) Chen, J.; Young, V.; Niki, H.; Magid, H. J. Phys. Chem. A 1997, 101, 2648. (23) Mogelberg, T. E.;Sehested, J.;Bilde, M.;Wallington T. J.;Nielsen, O. J. J. Phys. Chem. A 1996, 100, 8882. (24) De Bruyn, W. J.; Shorter J. A.; Davidovits P.; Worsnop D. R.; Zahniser, M. S.;Kolb, C. E. Environ. Sci. Technol. 1995, 29, 1179. Received for review January 12, 2000. Revised manuscript received September 26, 2000. Accepted October 12, 2000. ES000895F 120 9 ENVIRONM ENTAL SCIENCE & TECHNOLOGY / VOL. 35, NO. 1, 2001