Rate coefficients and Arrhenius parameters for the reactions between chloroethenes and the nitrate radical Ingse M. W. Noremsaune,a* Sarka Langer,b Evert Ljungstro mb and Claus J. Nielsena* a Department of Chemistry, University of Oslo, P. O. Box 1033 Blindern, N-0315 Oslo, Norway b Department of Inorganic Chemistry, University of Go teborg and Chalmers University of T echnology, S-412 96 Go teborg, Sweden Rate coefficients for the reactions between NO and the chloroethenes have been determined by the fast-Ñow-discharge technique 3 and by the relative rate method in a static reactor employing FTIR detection. The relative rate experiments were performed at 298 ^ 2 K and 1013 ^ 3 mbar in a nitrogen atmosphere with excess ethane as a chlorine scavenger. The Ñow tube experiments were carried out under pseudo-Ðrst-order conditions in NO . The temperature dependence of the NO reactions with chloroe3 3 thene and trichloroethene was investigated over a temperature range of ca. 100 K, and their rate coefficients were Ðtted to Arrhenius expressions. A comparison between our measured rate coefficient data for the NO reaction with chloroethenes reveals 3 that the relative rate results are up to 20% lower than those from the fast-Ñow-discharge experiments. Possible secondary reactions in the Ñow tube are discussed. The nitrate radical and its night-time chemistry in the atmosphere have received much attention during the past two decades. Indeed, the nitrate radical is able to oxidize a wide variety of organic compounds, thereby providing a chemical scavenging mechanism for atmospheric constituents of both natural and antropogenic origins. However, these oxidation reactions can lead to the formation of reactive species that may inÑuence the local environment. In other cases long-lived reaction products may have an impact on atmospheric processes by long-range transport. Consequently, it is relevant to know reaction rates and oxidation mechanisms, and extensive studies have been performed in many laboratories to gain this knowledge.1,2 The chloroethenes are widely used as solvents and intermediates in the manufacture of polymers. They are toxic and volatile and may be of major concern in local industrial areas. The chloroethenes react with the nitrate radical, and the mechanism for these oxidation reactions has recently been investigated.3 The major products in these reactions are substituted acid chlorides, formaldehyde, formyl chloride, phosgene, dinitrates and in addition, sometimes chlorine atoms. The only long-lived reaction product is phosgene, which may be transported to the stratosphere where photolysis will lead to the release of both chlorine atoms.4 The room-temperature rate coefficients for the reactions between the nitrate radical and the chloroethenes have previously been presented by Atkinson et al.5 who used a relative rate method with ethene as the reference compound. The published rate coefficients may be subject to signiÐcant uncertainties due to the low reactivity and the potential for production of chlorine atoms from the more chlorinated haloalkenes.1 Anderson and Ljungstrom have determined the rate coefficient for chloroethene (vinyl chloride) by measuring the enhanced decay of N O in the presence of vinyl chlo2 5 ride.6 Their value for the rate coefficient is rather uncertain due to the N O ¢ NO ] NO equilibrium constant which 2 5 3 2 is strongly temperature dependent and has an estimated uncertainty of 1.3 at 296 K.7 Recently, the discharge-Ñow technique was employed to determine the Arrhenius parameters for the reactions of the nitrate radical with 1,1-dichloroethene and (E)-1,2-dichloroethene.8 The agreement was good between rate coefficients obtained by the discharge Ñow and relative rate methods at room temperature. This was inter- preted as an indication of Cl-atom elimination being unimportant in these reactions. The present work is the Ðrst investigation of the temperature dependence of the nitrate radical reactions with vinyl chloride and trichloroethene. The work also includes determination of the rate coefficients for all the chloroethenes by both an absolute rate and a relative rate method at room temperature. For both methods the problems connected with possible secondary reactions are addressed. Experimental The absolute rate coefficients and their temperature dependence were measured under pseudo-Ðrst-order conditions using a fast-Ñow-discharge (FFD) technique where NO was 3 detected by optical absorption in a multi-path cell (Goteborg). This conventional FFD system was discussed in detail in a previous publication.9 Nitrate radicals were produced according to reaction (1) by mixing nitric acid with Ñuorine atoms, generated by passing F in He through a microwave dis2 charge. The dissociation of F was close to 100%, and a 2 massive excess of HNO was immediately added to remove all 3 atomic Ñuorine. F ] HNO ] NO ] HF 3 3 (1) The nitrate radicals were added at the upstream end of the Ñow tube with helium as the carrier gas. The radicals were detected by optical absorption at 662 nm using a multi-pass cuvette with a total optical path length of 72 cm. The practical nitrate radical detection limit was ca. 5 ] 1011 molecules cm~3. All the chloroethenes, except vinyl chloride, are liquids at STP, and they were introduced as gases from a low pressure distillation set-up. Vinyl chloride was taken directly from the lecture bottle. The gases were introduced into the Ñow tube through a sliding injector, the position of which is proportional to the contact time. The contact time could also be changed by a throttling stopcock between the pump and the Ñow tube, and contact times of 10 to 400 ms were achieved. The Ñow tube could be temperature regulated from 266 to 367 K by passing a waterÈethylene glycol mixture through the Ñow tube jacket. J. Chem. Soc., Faraday T rans., 1997, 93(4), 525È531 525 The relative rate (RR) experiments were carried out in a 250 l stainless-steel reactor (Oslo). The reactor was equipped with a multiple reÑection White type mirror system with an optical base path of 2 m and adjusted to give a total optical path of 120 m. The optical system was connected to a Bruker IFS 88 FTIR spectrometer, allowing in situ analysis of intermediates and end products. Spectra were generally recorded from the time of mixing and to a maximum of 2 h by coadding 50 scans (collection time ca. 45 s) at 0.5 cm~1 instrumental resolution. The RR experiments were performed at 298 ^ 2 K and 1013 ^ 3 mbar in a nitrogen atmosphere. Nitrate radicals were generated by thermal dissociation of N O , that was 2 5 added to the chamber by evaporation of solid N O . Mixing 2 5 ratios of the organic reactants were in the range 12È30 ppm in the presence of ca. 15 ppm N O . The decrease in reactant 2 5 concentration during the reactions was either determined by subtraction of reference spectra of the pure compounds or by subtraction employing the Ðrst recorded spectrum as a reference. A linear relationship between absorption and concentration was conÐrmed by separate calibrations of each reactant. Ethane was sometimes added to the reaction chamber in sufficient excess (300È450 ppm) to scavenge any Cl atoms generated and thereby counteract the e†ect a possible elimination of Cl atoms might have on the total loss of organics.3 Additional experiments were carried out at low pressure and an oxygen content of less than 100 ppmv to mimic the Ñow tube conditions where the oxygen content is estimated to be around 5 ppmv or less. All compounds employed in the FFD and RR experiments had a stated purity of 98% or better and were used without further puriÐcation : vinyl chloride (Norsk Hydro), ethene, 1,1dichloroethene, trichloroethene, (Z)-1,2-dichloroethene and tetrachloroethane (Fluka), (E)-1,2-dichloroethene (Aldrich), N 2 bath gas (grade 2.6, AGA). Helium (grade 5.5, AGA) was passed through a liquid-N trap with 4 Ó molecular sieves to 2 remove traces of oxygen. Fluorine, (5% in He, MesserGriesheim) was used as supplied. Anhydrous, gaseous HNO 3 in a He carrier was prepared by bubbling He through anhydrous HNO , that was distilled from a 2 : 1 mixture of conc. 3 H SO and HNO . 2 4 3 valid : ) \ k@t (I) /* NO3 NO3`org The Ðrst-order rate coefficients, k@, for the loss of NO were 3 obtained from least-squares adjusted straight lines according ) vs. to eqn. (I). Fig. 1(a) and (b) show plots of ln(* /* NO3 NO3`org time for various concentrations of vinyl chloride and trichloroethene, respectively. The plots are linear but with non-zero intercept since it is not easy (nor necessary) to deÐne time zero. However, the chloroethene reactions are fairly slow, and a small concentration of a fast reacting impurity will also give intercepts without altering the slope. The slopes from unit-weighted least-squares plots of k@ vs. [chloroethene] yield the apparent bimolecular rate coefficients, k, for the reactions : ln(* k products NO ] CClX1 \ CX2X3 ÈÈÕ 3 X1h3 \ H or Cl (2) Examples of second-order plots are shown in Fig. 2(a) and (b) and Table 1 gives a survey of the determined rate coefficients at room temperature together with some additional experi- Results Fast Ñow discharge measurements The reaction rates between the nitrate radical and the chloroethenes were investigated under pseudo-Ðrst-order conditions, that is, with the organic reactant being in sufficient excess. The experiments consisted of measuring values for * and * as a function of contact time. Here, * is NO3`org NO3 NO3 the change in transmittance signal between the case when NO is produced in the system and the case when no NO is 3 3 produced. Similarly, * is the change in signal between NO3`org the case when both NO and the organic substance are added 3 as compared to the case when no NO is present. For small 3 absorbances and pseudo-Ðrst-order conditions eqn. (I) is Fig. 1 (a) Pseudo-Ðrst-order plots for the reaction of NO with 3 1015, various concentrations of vinyl chloride at 295 K. (…) 1.20 ] (L) 2.91 ] 1015, (=) 4.78 ] 1015, (K) 7.62 ] 1015. (b) Pseudo-Ðrstorder plots for the reaction of NO with various concentrations of 3 ] 1014, (L) 1.46 ] 1015, (=) trichloroethene at 295 K. (…) 5.48 2.68 ] 1015, (K) 4.90 ] 1015. Table 1 Experimental conditions employed in measuring the rate coefficients for the reactions between chloroethenes and NO by fast Ñow 3 discharge at 295 K ; [NO ] O 1.5 ] 1013 molecule cm~3 3 0 compound p/mbar [org]/1015 molecule cm~3 v/cm s~1 vinyl chloride 1,1-dichloroethene (Z)-1,2-dichloroethene (E)-1,2-dichloroethene trichloroethene tetrachloroethene 5 3.6È5 1.2È8.0 0.84È7.3 547È630 615È972 6 5 8 1.2È28 0.55È4.9 0.64È2.1 262È390 514È652 280È308 a Errors quoted with 95% conÐdence in the slope of the plots k@ vs. [org]. b Not determined. c Determined at 292 K. 526 J. Chem. Soc., Faraday T rans., 1997, V ol. 93 k /10~16 cm3 org molecule~1 s~1 a 4.26 ^ 0.39 18.8 ^ 1.2 b 0.90 ^ 0.14c 4.43 ^ 0.32 0.96 ^ 0.81 Fig. 3 Second-order plots for the reaction of NO with tetra3 chloroethene at 295 K. The error bars show the 95% conÐdence in the ) vs. time. slope of the plots ln(* /* NO3 NO3`org Fig. 2 (a) Second-order plots for the reaction of NO with vinyl 3 295 K, (=) chloride at various temperatures. (…) T \ 280 K, (L) T \ T \ 325 K, (K) T \ 367 K. (b) Second-order plots for the reaction of NO with trichloroethene at various temperatures. (…) T \ 295 K, (L) 3T \ 328 K, (=) T \ 343 K, (K) T \ 367 K. mental information. The rate coefficient for (Z)-1,2-dichloroethene was not studied for costÈbeneÐt reasons. The positive intercepts that were observed in all the second-order plots, are not easily explained. Mathematically, the rate coefficient for wall loss does not appear in eqn. (I), because the loss of NO 3 with reactant is normalized to that without reactant. Thus the intercept of the second-order plots cannot be interpreted as wall loss of NO . (The wall loss rate coefficient of NO in the 3 3 Ñow tube was measured to be 0.5 s~1 or less.10) The requirements for a process causing the intercepts in the second-order plots are quite demanding. The process has to be quite insensitive to the concentration of organic compound over the range where measurements are made. However, it has to be initiated by the addition of organics and cannot be active when only NO is observed. The process must also be 3 time dependent so that more NO is lost at longer contact 3 times. In fact, it has to be close to Ðrst order in NO . 3 One possible process causing an intercept could be adsorption of the organic compound on the wall, followed by NO 3 reaction. The surface concentration must not change with the gas-phase concentration of the organic compound, and thus the available sites have to be saturated even at the lowest concentration of organic compound used. The limited axial di†usion would restrict the surface covered approximately to the surface downstream of the injector. This would fulÐl the requirements of a contact time dependent process. From our available information, we cannot say whether such an adsorption/surface reaction actually takes place. It should be noted, however, that such a process would not a†ect the slope of the second-order plot. The rate coefficient for the tetrachloroetheneÈNO reaction 3 was quite difficult to determine, as is clearly displayed in Fig. 3. The reaction is slow, and the rate could not be increased the normal way by increasing the concentration of tetrachloroethene due to its low vapour pressure. The pressure in the Ñow tube was increased to lengthen the contact time, Table 1, but it was still difficult to get reliable results. In addition, the uncertainty in measuring the organic concentration was much higher, about 10%, compared with 1È2% for the other chloroethenes. The unit-weighted plot of k@ vs. tetrachloroethene, Fig. 3, yielded a rate coefficient of (9.6 ^ 8.1) ] 10~17 cm3 molecule~1 s~1. The temperature dependence of the reactions with vinyl chloride and trichloroethene were investigated over a temperature range of about 100 K. Both reactions are clearly temperature dependent as the detailed results listed in Table 2 and as Fig. 4(a) and (b) show. The following Arrhenius expressions were obtained from a non-linear, unit-weighted leastsquares Ðtting procedure : (T ) \ 1.8 ] 10~13 (vinyl chloride) ] exp([1770/T ) cm3 molecule~1 s~1 k k (T ) \ 4.0 ] 10~13 (trichloroethene) ] exp([2030/T ) cm3 molecule~1 s~1 as compared to the parameter values 1.8 ] 10~13/1780 for vinyl chloride and 4.3 ] 10~13/2060 for trichloroethene when the Arrhenius expressions are obtained by linear Ðtting of the usual lnk vs. 1/T . Table 2 Arrhenius data for the reaction of NO with vinyl chloride 3 and trichloroethene k/10~16 cm3 molecule~1 s~1 a temperature/K vinyl chlorideb 2.43 ^ 0.57 3.57 ^ 0.60 4.26 ^ 0.39 7.52 ^ 1.2 12.1 ^ 1.5 14.5 ^ 1.5 266 280 295 325 348 367 trichloroethenec 278 295 313c 328 343.5 353 367.5 2.75 ^ 0.67 4.43 ^ 0.32 6.14 ^ 2.14 7.36 ^ 0.40 10.4 ^ 1.0 13.9 ^ 1.9 14.9 ^ 2.3 a Errors quoted with 95% conÐdence in the slope of the plots k@ vs. [org]. b E \ 14.7 ^ 0.9 kJ mol~1, A \ (1.84 ^ 0.55) ] 10~13 cm3 molecule~1a s~1 ; error quoted at 1p. c E \ 16.9 ^ 1.8 kJ mol~1, A \ (4.0 ^ 2.5) ] 10~13/cm3 molecule~1 s~1a ; error quoted at 1p. J. Chem. Soc., Faraday T rans., 1997, V ol. 93 527 Fig. 4 (a) Arrhenius plot of the reaction between NO and vinyl chloride. X denotes the rate coefficient determined by the 3relative rate method. (b) Arrhenius plot of the reaction between NO and trichlo3 relative rate roethene. Z denotes the rate coefficient determined by the method. Attempts to observe any temperature dependence of the NO reaction with tetrachloroethene were futile. 3 Relative rate measurements The rate coefficients for the reactions between the chloroethenes and the nitrate radical were determined by the relative rate technique where the organic of interest competes with a reference compound for the available nitrate radicals.11 The degradation of both compounds are measured simultaneously, and the relative rate coefficient, k /k , is determined from org ref the slope of straight lines according to the following expression : k [reference] [organic] 0 0 \ org ] ln (II) k [reference] [organic] ref t t The lines were constrained through the origin. Ethene, with a recommended literature rate coefficient of k(NO ) \ (2.01 3 ^ 0.30) ] 10~16 cm3 molecule~1 s~1 at 298 K, was employed as the reference compound for all the chloroethenes.1 Fig. 5(a) and (b) show plots of the data for the individual chloroethene reactions, except for tetrachloroethene, where it was not possible to get an accurate measure of the change in concentration at all times during the experiment. The derived relative rate coefficients for the individual tetrachloroethene experiments were determined as 0.29, 0.38 and 0.50, but the uncertainty in each value is quite high. The relative rate data are summarized in Table 3. The rate coefficients for the chloroethenes span about two orders of magnitude, Table 3, and ethene was chosen as the reference compound because it has a rate coefficient that lies approximately in the middle of this span. The ethene and chloroethene reactions are fairly slow and 10% or less of the organic reactant had normally reacted during the time of the experiment, except for 1,1-dichloroethene where about one ln 528 J. Chem. Soc., Faraday T rans., 1997, V ol. 93 Fig. 5 (a) Plots of eqn. (II) for 1,1-dichloroethene (A) and trichloroethene (B) with ethene as the reference compound. For each compound the di†erent symbols represent di†erent experiments. In particular, Ðlled symbols denote experiments with ethane as the chlorine scavenger. (b) Plots of eqn. (II) for vinyl chloride (C), (Z)-1,2dichloroethene (D) and (E)-1,2-dichloroethene (E) with ethene as the reference compound. For each compound the di†erent symbols represent di†erent experiments. third had reacted. Due to the low amounts consumed of the reactants and the problem of Ðnding suitable absorption bands to measure the small changes in concentration, the uncertainties in the rate coefficients are fairly large, Table 3. The rate coefficient for the 1,1-dichloroethene reaction with NO is almost a factor 10 faster than the analogue reaction 3 with ethene, and the rate coefficient for 1,1-dichloroethene was also measured with acetaldehyde as the reference compound, see Table 3. Additional RR experiments with vinyl chloride as the reference compound were performed, and the resulting rate coefficient ratios, k /k , are displayed in Table 3. Again, org vinyl chloride the rate coefficient for the reaction with tetrachloroethene was very difficult to determine. Ethane was sometimes added as a Cl-atom scavenger, but the rate coefficients derived in these experiments seemed unaffected, see Fig. 5(a), as was also reported in a previous study.5 In addition, we did not observe HCl, which should be a product in the reaction. However, we do have indications that Cl atoms were eliminated, since we observe nitrooxyacetyl Table 3 Data from relative rate measurements of the chloroethenes compound vinyl chloride 1,1-dichloroethene (Z)-1,2-dichloroethene (E)-1,2-dichloroethene trichloroethene tetrachloroethene temperature/K k /k a org ethene 296 296 299 299 298 298 299 1.78 ^ 0.27 8.72 ^ 0.47 101 ^ 0.16 0.46 ^ 0.13 1.79 ^ 0.18 0.39 ^ 0.11 k /k b org vinyl chloride 4.24 ^ 0.14 0.530 ^ 0.028a,c 0.47 ^ 0.03 0.29 ^ 0.01 1.3 ^ 0.04 0.1 a Two standard errors of the average of individual experiments. b Errors quoted with 95% conÐdence. c Relative to acetaldehyde. chloride (NAC, O NOwCH wCOCl), which is formed by 2 2 elimination of Cl-atoms from the nitrooxy alkoxy radical intermediate in the degradation of 1,1-dichloroethene.3 We also observe ClNO, but we do not know if this is an elimination product or an eventual result of the released Cl atoms. pseudo-Ðrst-order rate coefficients were varied until agreement was reached between the NO concentrations simulated with 3 and without secondary reactions at di†erent concentrations of the organics. The following two simple models were employed : Discussion Model 1 A summary of the rate coefficients determined in this work and by previous investigators is listed in Table 4. As can be seen, there is some disagreement between the obtained rate coefficients. If one compares the rate coefficients determined by the FFD and RR methods in this work, the FFD rate coefÐcients are as much as 20% higher for the faster reacting chloroethenes. This could indicate that secondary reactions are signiÐcant in the Ñow tube, but it may also just reÑect the general uncertainties in the FFD and RR methods. Since the reactions were measured to be faster at low pressure, the chloroethene reactions are probably not pressure dependent within the pressure range applied. This is also in agreement with previous investigations where no pressure dependence was noted.5,8 Secondary reactions in the fast Ñow system With the FFD system we have measured changes in the NO 3 concentration when the chloroethenes were being added in sufficient excess. It is not possible to investigate in situ whether the nitrate radical reacts with the chloroethenes only or also with one or several products from the initial reaction. Consequently, the measured rate coefficients must be considered as upper limits. The same also applies to the subsequently derived apparent bimolecular rate coefficients. Several products that are able to initiate secondary reactions may be formed in the Ñow tube. As mentioned before, it is possible that Cl atoms are eliminated, and these Cl atoms may react with either of the initial reactants.3 Furthermore, oxiranes which have been recognized as major products in reactions with unsaturated hydrocarbons and NO at low 3 pressure,12h14 may also be formed here. Oxirane formation in Ñow tube experiments with NO will a†ect the rate coefficient 3 determination because the released NO enters a pressure 2 dependent equilibrium with NO forming N O . The e†ect 3 2 5 that a possible elimination of Cl-atoms and NO may have on 2 the rate coefficients was tested by separate simulations with the computer program FACSIMILE.15 In the simulations the NO ] chloroethene ] product ] Cl 3 Cl ] chloroethene ] product (4a) (Cl ] chloroethene ] product ] Cl) (4b) (3) Cl ] NO ] ClO ] NO 3 2 (5) Model 2 NO ] chloroethene ] oxirane ] NO (6) 3 2 NO ] NO ] N O (7) 2 3 2 5 N O ] NO ] NO (8) 2 5 2 3 In model 1 the Cl-atom reactions with the chloroethenes are pressure dependent (ref. 16 and references therein), and the rate coefficients at 5 mbar total pressure are not known. We roughly estimate the ClÈchloroethene rate coefficients to be the same as that of the ClÈethene reaction,8 ca. 2 ] 10~12 cm3 molecule~1 s~1 at 5 mbar total pressure.17 Employing eqn. (3), (4a) and (5) and a rate coefficient of 4.5 ] 10~11 cm3 molecule~1 s~1 for eqn. (5),2 then the simulations show that elimination of Cl atoms will have no signiÐcant e†ect on the rate coefficients determined under our experimental conditions. However, previous investigations show that the Cl atoms initiate a chain reaction because Cl atoms are also released from the Cl atom adduct. If this chain reaction is included in the model of the Ñow tube reactions [eqn. (4b) instead of eqn. (4a)], the simulations indicate that the apparent bimolecular rate coefficients could be as much as a factor of two larger than the true rate coefficients. It is unlikely that the total yield of Cl atoms will be unity. In fact, the simulations show that the yield of Cl atoms in both eqn. (3) and (4b) must be greater than 95 and 80% for 1,1-dichloroethene and trichloroethene, respectively, in order to have any e†ect on the measured rate coefficients. Thus, Cl atom elimination is not considered to be a problem in the FFD experiments. Table 4 Room-temperature rate coefficients for the reactions of chloroethenes with NO from this work compared with literature values 3 k/10~16 cm3 molecule~1 s~1 this work compound vinyl chloride 1,1-dichloroethene FFD RRa RRb other 4.3 ^ 0.4 19 ^ 1 3.6 ^ 0.8c 18 ^ 3c 16 ^ 4f 15 ^ 1g 2.0 ^ 0.4c 1.8 ^ 0.4f 0.9 ^ 0.3c 1.1 ^ 0.2f 3.6 ^ 0.7c 4.7 ^ 1.1f 0.8 ^ 0.3c 0.4f 4.2 ^ 0.7 12 ^ 2 2.2 ^ 1.1d 12 ^ 3e (Z)1,2-dichloroethene n.d. (E)-1,2-dichloroethene 1.0 ^ 0.2h trichloroethene 4.4 ^ 0.3 tetrachloroethene lit. values 1 ^ 0.8 1.4 ^ 0.2 1.0 ^ 0.2 1.2 ^ 0.5e 2.8 ^ 0.4 \0.5 a Errors quoted combining the errors in k /k and k . b Presented relative to k \ (2.01 ^ 0.30) ] 10~16 cm3 molecule s~1, ref. 5. c k ethene`NO3ref. 7. e Determined by FFD, ref. 8. f Presented relative ref as in footnote b, ref. 1. d Presented relativeorgto ref K (N Oref) \ 2.90 ] 10~11 cm3 molecule~1, eq 2 5 \ (2.74 ^ 0.07) ] 10~15 to k (298) \ (3.73 ^ 0.82) ] 10~16 cm3 molecule~1 s~1, this work. g Presented relative to k 3 acetaldehyde`NO3 chloride`NOs~1, cm3 vinyl molecule~1 ref. 12. h Corrected to 295 K, ref. 8. J. Chem. Soc., Faraday T rans., 1997, V ol. 93 529 Oxirane formation has been proposed as a major product in the chloroethene reactions in Ñow-tube systems.8 To investigate this, the chloroetheneÈNO reactions were studied in 3 the static reactor applying the same total pressure as in the Ñow tube. Although not immediately comparable due to differences in oxygen and reactant concentrations and in the surface to volume ratio, such experiments may give an indication as to what could take place in the Ñow tube. Generally, the same products were observed at low pressure and at atmospheric pressure. However, for 1,1-dichloroethene new bands that were not observed at normal pressures, appeared at 974 cm~1 and in the region 1040È1070 cm~1. These bands may originate from 1,1-dichlorooxirane, and infrared spectra of other halogenated oxiranes show that some of the characteristic absorption bands lie in these regions.18 Unfortunately, there is no published data on 1,1-dichlorooxirane to conÐrm or rule out the formation of this compound. Relying on IR literature data on oxiranes,18 we could not observe these in the reactions of NO with (Z)- and (E)-1,2-dichloroethene and 3 with tetrachloroethene. On the other hand, oxiranes are known to have weak absorption bands, and, in addition, the concentration of the oxiranes should be low due to the low conversion of the reactants. There are many uncertainties concerning oxirane formation in the chloroethene reactions. We have investigated the possible e†ect NO may have on the measured rate coefficients 2 assuming that oxirane and NO are formed with unit yield, 2 model 2. When simulating the reaction system, the forward and reverse reactions in the equilibrium between NO , NO 2 3 and N O were included [eqn. (7) and (8)]2 together with eqn. 2 5 (6). The model shows that the resulting bimolecular rate coefficients for the faster reacting compounds are less inÑuenced by the formation of NO . Accordingly, the rate coefficient for 2 (E)-1,2-dichloroethene is a†ected the most (ca. 25%), which is contrary to the experimental RR and FFD results, Table 4. However, as mentioned before, the uncertainty in the RR rate coefficients for (E)-1,2-dichloroethene is fairly large. For vinyl chloride, 1,1-dichloroethene and trichloroethene the discrepancy in the rate coefficients determined by the FFD and RR methods may very well be explained by the formation of NO . 2 Arrhenius parameters Recently, a theoretical study of the reactivity of the NO 3 radical towards a series of alkenes and halogenoalkenes was presented.19 In Table 5 the measured and predicted19 Arrhenius parameters are compared for the chloroalkeneÈNO 3 reactions. As can be seen, the agreement is quite good between the measured and predicted activation energies, and the preexponential factors di†er by only a factor of ca. two. For vinyl chloride, however, the predicted activation energy is ca. 25% higher than the measured value, and the pre-exponential factors di†er by ca. a factor two the opposite way as compared to the other chloroethenes. We o†er at present no explanation for this. Relative rate The validity of the FFD results may be checked against results from the RR method, although secondary reactions may also inÑuence the RR coefficient determination. This may happen if one or several of the intermediates or products react with the initial organic reactants at di†erent rates than with the standard compounds. Previous studies show that this should neither be a problem for vinyl chloride, nor for (Z)and (E)-1,2-dichloroethene.3 Tetrachloroethene reacts so slowly that it is difficult to observe any reaction at all, and we have no evidence for secondary reactions here. In contrast, it is evident that Cl atoms are eliminated during the NO initi3 ated oxidation of 1,1-dichloroethene and trichloroethene,3 and Cl atoms react fast with their organic precursors as well as with the reference compound ethene.16 However, as is clearly displayed in Fig. 5(a), the addition of excess ethane as a Cl atom scavenger in experiments with 1,1-dichloroethene and trichloroethene had little or no e†ect on the loss of the other reactants. This could suggest that the Cl atoms react with the chloroethenes and ethene with the same relative rate coefficients as in the NO radical reactions. However, the measured 3 rate coefficients for the chloroetheneÈCl atom reactions show that this is not so.16 Accordingly, the RR rate coefficients seem to be una†ected by secondary reactions. As mentioned before, the same was observed by Atkinson et al. in their study of the reactivity of the chloroethenes.5 The experiments with excess ethane added are somewhat puzzling ; we did not observe any HCl formation but we did observe nitrooxyacetyl chloride (NAC), which can only be formed by elimination of Cl atoms.3 The absence of HCl may in part be explained by a low yield of NAC. Unfortunately, we cannot quantify the yield, but it appears to be low because the characteristic CxO stretching band at 1825 cm~1 is weak. The absence of HCl may also be explained by the Cl atoms having other loss channels, of which we can only guess. Frontier orbital theory has been used to rationalize the reactivity of the chloroethenes towards NO .19,20 In 3 summary, this theoretical approach predicts the following order of reactivity : 1,1-dichloroethene [ vinyl chloride [ tetrachloroethene[trichloroethene[(Z)-1,2-dichloroethene[ (E)-1,2-dichloroethene. In view of the approximations involved in this theoretical approach, the agreement with the experimental results is quite remarkable, except for tetrachloroethene, Table 4. Our experimental results show that the reactivity of the chloroethenes falls into three groups : (i) 1,1dichloroethene [ (ii) vinyl chloride and trichloroethene [ (iii) (Z)-1,2-dichloroethene, (E)-1,2-dichloroethene and tetrachloroethene. The objectives of the RR experiments with vinyl chloride as the reference compound were to conÐrm the order of reactivity within group (ii) where the ratio found in the present work di†ers from that of the previous investigation, Table 4,5 and group (iii) where the compounds react so slowly that it is difficult to obtain reliable data. The rather superÐcial analysis of these RR experiments results in trichloroethene Table 5 Arrhenius parameters measured compound vinyl chloride 1,1-dichloroethene (Z)-1,2-dichloroethene (E)-1,2-dichloroethene trichloroethene tetrachloroethene E /kJ a mol~1 A/10~13 cm3 molecule~1 s~1 E /kJ a mol~1 A/10~13 cm3 molecule~1 s~1 15 ^ 1b 15 ^ 2c 1.8 ^ 0.6b 5 ^ 2c 20.0 ^ 0.3c 17 ^ 2b 4.5 ^ 0.5c 4 ^ 2b 18.7 14.0 18.9 19.4 17.5 15.4 4.7 3.9 1.9 2.0 1.4 0.90 a Ref. 19. b This work. c Ref. 8. 530 predicteda J. Chem. Soc., Faraday T rans., 1997, V ol. 93 reacting faster, not slower, than vinyl chloride. This observation, combined with the other RR experiments and the FFD temperature studies, strongly supports the statement that the rate coefficients for vinyl chloride and trichloroethene are approximately equal at room temperature. From the amount consumed of each reactant in group (iii), we must conclude that the NO reaction with tetrachloroethene deÐnitely is 3 slower than the (E)-1,2-dichloroethene reaction, thus supporting the data in Table 4. Conclusion The FFD and RR rate coefficients determined in this work are in fair agreement with di†erences of 20% or less, Table 4. These di†erences may be explained by uncertainties in the two methods employed, but simulations show that they may also be explained by secondary reactions involving NO , although 2 this has not been conÐrmed experimentally. Secondary reactions due to Cl atom elimination are not likely to a†ect the FFD rate coefficients, because simulations indicate that the Cl atoms then must be formed with a total yield near unity. No secondary reactions were observed to signiÐcantly interfere with the RR rate coefficient determinations. Several of the previously reported rate coefficients lie about 20 to 30% below the RR rate cofficients determined in this work, Table 4. Except for 1,1-dichloroethene and (Z)-1,2dichloroethene most of the previous data are still within the uncertainty limits of the RR data in this work. For (Z)-1,2dichloroethene this discrepancy can readily be explained by the small difficulty of Ðnding a suitable absorption band for measuring changes in the reactant concentration. In contrast, we cannot explain the di†ering values for 1,1-dichloroethene, but Ðnd the results puzzling since essentially the same methods and the same reference compound have been employed. Another point to note about the previously published rate coefficients is that all our experimental data indicate that the rate coefficient for vinyl chloride is approximately equal to the rate coefficient for trichloroethene at room temperature, in contrast to the previous result.5 Financial support from the Norwegian Research Council and Norsk Hydro is gratefully acknowledged. Part of this work was carried out with the support from NorFA enabling IMWN to spend a training and research period at the Uni- versity of Goteborg and Chalmers University of Technology. The authors also wish to thank Anastasia Pagou, an IAESTE trainee, for her assistance in the laboratory. References 1 R. Atkinson, J. Phys. Chem. Ref. Data, 1991, 20, 459. 2 R. P. Wayne, I. Barnes, P. Briggs, J. B. Burrows, C. E. CanosaMas, J. Hjorth, G. LeBras, G. Moortgat, D. Perner, G. Poulet, G. Restelli and H. Sidebottom, Atmos. Environ., 1991, 25A, 1. 3 I. M. W. Noremsaune, J. Hjorth and C. J. Nielsen, J. Atmos. Chem., 1995, 21, 223. 4 S. R. Wilson, P. J. Crutzen, G. Schuster, D. W. T. Griffith and G. Helas, Nature (L ondon), 1988, 334, 689. 5 R. Atkinson, S. M. Aschmann and M. Goodman, Int. J. Chem. Kinet., 1987, 19, 299. 6 Y. Andersson and E. Ljungstrom, Atmos. Environ., 1989, 23, 1153. 7 W. B. DeMore, S. P. Sander, D. M. Golden, R. F. Hampson, M. J. Kurylo, C. J. Howard, A. R. Ravishankara, C. E. Kolb and M. J. Molina, T he NASA data evaluation, 1994, 11, 107. 8 B. C. Galan, G. Marston and R. P. Wayne, J. Chem. Soc., Faraday T rans., 1995, 91, 1185. 9 S. Langer, E. Ljungstrom and I. Wangberg, J. Chem. Soc., Faraday T rans., 1993, 89, 425. 10 E. Ljungstrom and S. Langer, personal communication, 1991. 11 R. Atkinson, C. N. Plum, W. P. L. Carter, A. M. Winer and J. N. Pitts Jr., J. Phys. Chem., 1984, 88, 1210. 12 E. J. Dlugokencky and C. J. Howard, J. Phys. Chem., 1989, 93, 1091. 13 U. Wille and R. N. Schindler, Ber. Bunsen-Ges. Phys. Chem., 1993, 97, 1447. 14 Th. Benter, M. Liesner, R. N. Schindler, H. Skov, J. Hjorth and G. Restelli, J. Phys. Chem., 1994, 98, 10492. 15 E. M. Chance, A. R. Curtis, I. P. Jones and C. R. Kirby, FACSIMILE, Report R 8775, United Kingdom Atom Energy Authority, Harwell, 1977. 16 R. Atkinson and S. M. Aschmann, Int. J. Chem. Kinet., 1987, 19, 1097. 17 T. J. Willington, J. M. Andino, I. M. Lorkovic and E. W. Kaiser, J. Phys. Chem., 1990, 94, 3644. 18 J. Derkosch, E. Ernstbrunner, E. G. Ho†mann and F. Osterreicher, Monatsh. Chem., 1967, 98, 956. 19 G. Marston, P. S. Monks, C. E. Canosa-Mas and R. P. Wayne, J. Chem. Soc., Faraday T rans., 1993, 89, 3899. 20 R. W. S. Aird, C. E. Canosa-Mas, D. J. Cook, G. Marston, P. S. Monks and R. P. Wayne, J. Chem. Soc., Faraday T rans., 1992, 88, 1093. Paper 6/06008K ; Received 30th August, 1996 J. Chem. Soc., Faraday T rans., 1997, V ol. 93 531