A Low Temperature Study of the Reactions of Atomic Chlorine with

Sep 30, 2009 - Julien Daranlot , Kevin M. Hickson , Jean-Christophe Loison , Raphaël Méreau , Françoise Caralp ... Ian W. M. Smith , Peter W. Barne...
0 downloads 0 Views 219KB Size
Download PDF
3038

J. Phys. Chem. A 2010, 114, 3038–3044

A Low Temperature Study of the Reactions of Atomic Chlorine with Simple Alkanes† Kevin M. Hickson, Astrid Bergeat,* and Michel Costes UniVersite´ de Bordeaux, Institut des Sciences Mole´culaires, 351 Cours de la Libe´ration, 33405 Talence Cedex, France, and CNRS UMR 5255, ISM, 33405 Talence Cedex, France ReceiVed: June 30, 2009; ReVised Manuscript ReceiVed: September 9, 2009

The kinetics of the reactions of atomic chlorine with ethane and propane have been studied in a continuous supersonic flow reactor over the range 48 K e T e 167 K. Chlorine atoms were produced by microwave discharge upstream of the Laval nozzle and were probed in the vacuum ultraviolet wavelength range around 138 nm by resonance fluorescence. The reaction of Cl with ethane has been found to exhibit a positive temperature dependence, with a rate coefficient decreasing from (4.3 ( 0.6) × 10-11 cm3 molecule-1 s-1 at 167 K to (2.9 ( 0.3) × 10-11 cm3 molecule-1 s-1 at 48 K and deviates from true Arrhenius behavior below 120 K. In contrast, the rate coefficient for the reaction of Cl with propane has been found to have a constant value of (1.4 ( 0.2) × 10-10 cm3 molecule-1 s-1 over the same temperature range. The expressed uncertainties are the combined statistical (a single standard deviation from the mean) and systematic (estimated at 10%) uncertainties. Introduction The reactions of ground electronic state atomic chlorine with alkanes play a significant role in the chemistry of the Earth’s atmosphere.1,2 These processes all proceed through an abstraction of atomic hydrogen to yield HCl and a hydrocarbon radical species as the coproduct as in the specific cases of Cl with ethane and propane:

Cl + C2H6 f HCl + C2H5

(1)

Cl + C3H8 f HCl + C3H7

(2)

In particular, HCl formed through the reaction of Cl with CH4 in the stratosphere acts as a sink for atomic chlorine which would otherwise participate in the catalytic destruction of ozone.1 The reactions of Cl with non-methane hydrocarbons (NMHCs) are less important to atmospheric chemistry, although it has been reported that the reactions of atomic chlorine with ethane and propane could account for 26 and 15% of their losses, respectively, in the marine boundary layer.2 The reactions of atomic Cl with alkanes are among the most thoroughly studied processes in gas phase chemistry, with an extensive catalogue of previous work including experimental kinetic studies of the Cl + ethane reaction by absolute3-21 and relative methods,22-33 the Cl + propane reaction by absolute7,12,14,15,21,34 and relative methods,12,13,23-25,28,29,31,32,35 experimental measurements of the site specificity of the H atom abstraction for Cl + propane,21,35 and theoretical predictions of the rate coefficients for Cl + ethane by variational transition state methods36,37 or through structure activity relationships.38 Despite the numerous earlier investigations of the kinetics of these reactions, they have been studied only to temperatures as low as 180 K. The most recent absolute study of the temperature dependence of the Cl + ethane reaction20 extended the temperature range of kinetic measurements down to 177 K and in conjunction with earlier work at †

Part of the “Benoît Soep Festschrift”. * Corresponding author. Phone: (33) 5 40 00 63 41. Fax: (33) 5 40 00 66 45. E-mail: [email protected].

higher temperatures found that it exhibited straightforward Arrhenius type behavior from 177 to 600 K with a very small activation barrier of the order of 600 J mol-1 (70 K). It would be interesting to extend these measurements to much lower temperatures to examine the behavior of the rate coefficient at mean collision energies corresponding to this estimated barrier height, a barrier which is only poorly reproduced theoretically.36,37,39 Furthermore, extending the range of the currently available data should help to increase the precision of the currently used Arrhenius parameters for this reaction. Previous studies of the reaction of Cl with propane covering the range 195 K e T e 700 K report either temperature-independent values14,21 or a very slight negative temperature dependence,7 although current evaluations40,41 have adopted a temperatureindependent value. However, the value of the rate coefficient for this reaction at low temperatures is currently based on a single absolute measurement of the rate coefficient.7 Consequently, a reinvestigation of the Cl + propane reaction to lower temperatures would allow the entire range of potential atmospheric temperatures to be covered and would provide a more accurate description of the reaction’s temperature dependence over a much larger range. To this end, we report the results of experimental measurements of the Cl + C2H6 and Cl + C3H8 reactions over the range 48 K e T e 167 K using a continuous supersonic flow reactor. Experimental Section Reactor. The experimental apparatus, which has been described previously,42 utilizes the CRESU method (Cine´tique de Re´action en Ecoulement Supersonique Uniforme or reaction kinetics in a uniform supersonic flow), thus building on the pioneering work of Rowe et al.43 Briefly, the main reactor was constructed from stainless steel with an internal diameter of 180 mm. The Laval nozzle which was mounted on a piston and backed by a gas reservoir could be displaced 350 mm along the chamber axis. The reagent and precursor molecules and the carrier gas were passed into the reservoir and were isentropically expanded through the Laval nozzle into the reactor. Five home designed Laval nozzles allow us to perform kinetic measure-

10.1021/jp9061253  2010 American Chemical Society Published on Web 09/30/2009

Reactions of Atomic Chlorine with Simple Alkanes ments at specified temperatures of 48, 72, 124, 167, and 211 K, although for this study the 211 K nozzle was not used. The pumping speed (up to 1100 m3 h-1) was adjusted by a throttling valve. The detection region was located 380 mm from the upstream end of the reactor and consisted of four ports perpendicular to the chamber axis at 90° separations around the circumference. The background pressure in the reactor and the stagnation pressure within the reservoir were measured Via capacitance manometers (Leybold CERAVAC CTR90) which were either attached in the proximity of the detection region for the reactor or to the convergent region of the nozzle for the reservoir. An 11 mm inner diameter quartz tube in the form of a “Y” was mounted upstream of the nozzle reservoir. A McCarroll type microwave discharge cavity44 (Opthos Instruments) was mounted on one of the arms of the quartz tube, approximately 350 mm from the Laval nozzle entrance. This distance ensured that all of the electronically excited atoms produced in the discharge were collisionally quenched. Traces of a dilute mixture of 5.15% Cl2 in He were passed through the cavity using Ar as the carrier gas. The discharge was operated at 80 W and 2450 MHz (SAIREM, GMP 03 K/SM generator) to produce atomic chlorine in both spin-orbit levels of the ground electronic state (Cl(2P°3/2) and Cl(2P°1/2), hereafter referred to as Cl and Cl*, respectively). Argon was always used as the carrier gas through the cavity, even for the nozzles which were designed for operation with N2 as the carrier gas so that the production of atomic nitrogen was avoided. As such, this Ar flow was maintained as a small percentage of the total gas flow (typically 4-6%). The other arm of the quartz tube carried the main gas flow, either argon or nitrogen, with the two arms of the quartz tube recombining 150 mm upstream of the Laval nozzle. To reduce the heterogeneous recombination of atomic chlorine, the quartz tube was coated with halocarbon wax (Halocarbon Corp, grade 1500) and the reservoir itself was fabricated from polytetrafluoroethylene. In the absence of the coreagent species, the Cl atom density in the supersonic flow was estimated to be approximately 3% of the Cl2 density. The loss of atomic chlorine upstream of the Laval nozzle (and thus at room temperature) was minimized by introducing the coreagent molecules at the last possible moment upstream of the nozzle, as described in a previous publication.42 Reagent chlorine atoms were detected by resonance fluorescence around 138 nm, where several transitions of Cl and Cl* originate.45 Atom excitation was achieved using a microwave discharge lamp powered at 85 W and 2450 MHz (SAIREM, GMP 03 K/SM generator). It consisted of a Vidal type cavity46 mounted on a quartz tube isolated from the reactor by a MgF2 window and pumped with a mechanical pump (Edwards E2M5) backed with a cryogenic trap. The total flow in the lamp consisted of around 0.167 kPa of He carrying trace amounts of Cl2 (∼0.1%) to obtain optimal emission intensities from excited Cl atoms: a typical emission spectrum of the lamp is presented in Figure 1. Atomic emission from the lamp was passed into the reactor, perpendicular to the supersonic flow, whereupon the cold atomic chlorine atoms remaining within the flow were illuminated, fluorescing on-resonance with the exciting radiation. This fluorescence was collected at right angles to both the supersonic flow and the lamp light such that the background light intensity was minimized. The fluorescence was passed through a CaF2 window to eliminate atomic emission primarily from the Lyman-R transition of atomic hydrogen at 121.6 nm and was focused by a LiF biconvex lens and onto the photocathode of a solar blind channel photomultiplier in photon counting mode (Perkin-Elmer MP1911-RS232). It can be seen in Figure 1 that

J. Phys. Chem. A, Vol. 114, No. 9, 2010 3039

Figure 1. VUV lamp emission spectrum when traces of Cl2 are added. The dashed line delimits the approximate lower end of the observable wavelength range during fluorescence measurements. The observable wavelength range extends to approximately 200 nm.

emission from atomic oxygen at 130.4 nm which was transmitted by the CaF2 window was negligible with respect to that of atomic chlorine. The linear dependence of the chlorine atom fluorescence signal vs the lamp emission intensity was verified by changing the Cl2 flow added to the main He flow in the lamp while keeping the chlorine atom concentration constant in the supersonic flow. Cl2 in the lamp was kept to a minimum to avoid self-absorption. The discharge lamp was also used in an absorption configuration, i.e., with the lamp at 180° to a VUV monochromator (ARC VM502) to estimate the atomic chlorine concentration and also to verify the attenuation of lamp radiation by Cl, Cl2, and the alkane coreagent in the reactor. The Cl atom concentration in the reactor was always less than 1 × 1011 atom cm-3 and was less than 5 × 1010 atom cm-3 for most of the experiments. At these levels, it was found not to attenuate the incident radiation. Attenuation of the incident radiation by Cl2 and the alkane coreagent was observed however, although this did not interfere with the measurements, as both of these concentrations were fixed for each individual experiment. Gas Flows. All gases were flowed directly from cylinders (stated gas purities: Sigma Aldrich: C3H8 99.99%; Linde: Ar 99.999%, C2H6 99.5%; Air Liquide: N2 99.9999%, Cl2/He 5.15%, and He 99.9999%), with no further purification prior to usage. The carrier gas, reagent, and precursor flows were all passed into the reservoir via digital mass flow controllers (Brooks) attached via either Teflon or stainless steel tubing. The controllers were calibrated using the pressure rise at constant volume method for the specific gas or gas mixture used. The velocity, temperature, and total density of the supersonic flows were calculated from separate measurements of the impact pressure using a Pitot tube and the stagnation pressure within the reservoir as previously described.47 These measurements, coupled with the individual gas flows, allowed us to determine the coreagent concentrations and estimate the minor reagent concentrations for each experiment. The relative uncertainty in the alkane concentration was determined from statistical errors in the calculated density and flow rates and was found to be as large as 45% for the lowest alkane flow rates used, but was less than 10% for most of the experiments, as the magnitude of the uncertainty was approximately constant for all concentrations of the coreagent species used.

3040

J. Phys. Chem. A, Vol. 114, No. 9, 2010

Figure 2. Plot of the pseudo-first-order rate coefficient versus C3H8 concentration at 72 K for the loss of Cl atoms (solid squares, data with their statistical uncertainties). The solid line is the least-squares fit of the Cl data. The inset shows a typical exponential decay fit (line) of the Cl atom fluorescence signal (crosses) for 1.07 × 1014 molecules cm-3 of C3H8 and a flow velocity of 482 m s-1.

Results Methods. The evolution of the chlorine atom concentration was followed through its fluorescence intensity by varying the distance between the end of the Laval nozzle and the detection zone. As the flow velocity is constant over the effective length of the supersonic flow, this distance is thus proportional to the reaction time. The alkane coreagent was introduced in excess. A typical trace is shown in the inset of Figure 2. This distance was varied with a precision of (0.1 mm. As atomic chlorine was created upstream of the Laval nozzle where the gas mixing occurs, a loss of Cl atoms was observed through the reaction of Cl with the respective alkane at room temperature despite the introduction of the coreagent molecules at the last possible moment. This amounted to as much as 98% of the initial chlorine atom concentration for the highest alkane concentrations used and was estimated from the chlorine atom fluorescence intensity with and without added coreagent in the supersonic flow at short distances from the Laval nozzle. However, as the upstream Cl atom concentration was always considerably less than 1% of the alkane concentration, loss of the coreagent alkane before the supersonic expansion was negligible. Thus, the alkane concentration was simply determined from its partial pressure. Several first-order decay profiles of the atomic chlorine fluorescence signal were recorded at different alkane concentrations for each nozzle. Subsequently, a nonlinear least-squares analysis was used to fit the data at a minimum distance of 15 mm to avoid the possible detection of scattered light from the nozzle surface. Baseline intensities were measured and compared with the resulting baseline values of the nonlinear fits in order to minimize the error when low partial pressures of hydrocarbon reagents were used. Finally, secondorder rate coefficients were determined from a weighted linear least-squares analysis of the slopes of the plot of the pseudo-

Hickson et al.

Figure 3. Temperature dependence of the Cl + C2H6 reaction rate displayed on a log-log scale: 9, this work with statistical uncertainties at the 1σ level combined with systematic uncertainties of 10%; s, best fit of all of the data with the exception of Bryukov et al.;19 b, measurements of Bryukov et al.;19 [, measurements of Lewis et al.;7 0, measurements of Pilgrim et al.;14 O, measurements of Hickson et al.;20 4, measurements of Manning and Kurylo;5 ], measurements of Dobis and Benson;9 - -, VTST calculations of Fernandez-Ramos et al.;37 · · · · , Sander et al.40 evaluation with recommended uncertainties ( · - · ). Inset: expanded view of the region from 180 to 800 K and k(Cl+C2H6) from (4.5-8.5) × 10-11 cm3 molecule-1 s-1.

first-order decay rates versus the alkane concentration, the weights being given by the reciprocal of the uncertainty for each respective point. Figure 2 represents the linear dependence of the pseudo-first-order rate coefficient versus propane concentration for the 72 K nozzle. This procedure was repeated with both alkane coreagents for each Laval nozzle to determine the temperature dependence of the rate coefficient for both reactions. Temperature-Dependent Rate Coefficients. The secondorder rate coefficients obtained at each temperature are summarized in Table 1, with the respective alkane concentration ranges used for each set of experiments. The uncertainties displayed are the combined statistical and systematic uncertainties. Statistical uncertainties were calculated from the weighted fitting procedure described in the Methods section and are included at the level of a single standard deviation, whereas the systematic uncertainties which were thought to result from errors in the measured flow velocities, densities, temperatures, and flow rates are included in our uncertainty estimate at the level of 10%. Our results for the Cl + ethane reaction over the range 48 K e T e 167 K are presented and compared to previous work performed over a range of temperatures from 177 to 1002 K in Figure 3. Only the earlier studies reporting absolute rate coefficients covering a range of temperatures are displayed for clarity. The current Sander et al.40 recommendation with its upper and lower error limits and valid over the range 177 K e T e 353 K is shown alongside the experimental data and extended to encompass the entire range of current and previous values. It can be seen that although the temperature ranges for the previously determined values for the Cl + ethane rate

TABLE 1: Measured Rate Coefficients for the Cl + C2H6 and Cl + C3H8 Reactions T/K

[C2H6]/1014 molecules cm-3

kC2H6/cm3 molecule-1 s-1

[C3H8]/1014 molecules cm-3

kC3H8/cm3 molecule-1 s-1

167 ( 2 124 ( 7 72 ( 3 48 ( 1

(0.28-2.67) (0.09-4.55) (0.36-2.32) (0.22-3.69)

(4.34 ( 0.61) × 10-11 (4.16 ( 0.43) × 10-11 (3.39 ( 0.44) × 10-11 (2.86 ( 0.30) × 10-11

(0.07-0. 70) (0.02-1.07) (0.06-1.84) (0.19-0.91)

(1.34 ( 0.14) × 10-10 (1.45 ( 0.18) × 10-10 (1.33 ( 0.20) × 10-10 (1.37 ( 0.20) × 10-10

Reactions of Atomic Chlorine with Simple Alkanes

J. Phys. Chem. A, Vol. 114, No. 9, 2010 3041 Discussion Numerical Simulations. Simulations were performed to verify the effect of potential secondary chemistry on the measured rate coefficients. To a first approximation, possible reactions which could have interfered with the measurement of reaction 1 under our experimental conditions are

Figure 4. Temperature dependence of the Cl + C3H8 reaction rate displayed on a log-log scale: 0, this work with statistical uncertainties at the 1σ level combined with systematic uncertainties of 10%; [, measurement of Mellouki;15 9, measurements of Pilgrim et al.;14 O, measurements of Lewis et al.;7 ], measurements of Choi et al.;21 g, measurement of Hitsuda et al.;34 f, measurement of Beichert et al.;12 s, Sander et al.40 evaluation for the combined rate coefficient. Vertical dashed lines delimit the recommended temperature range. - -, Sander et al.40 evaluation for the primary abstraction channel; · · · · , Sander et al.40 evaluation for the secondary abstraction channel.

coefficient5,7,9,14,19,20 and the presently measured values do not overlap, the results are very consistent in terms of their temperature dependence and overall magnitude. However, it is strikingly apparent from Figure 3 that a straightforward Arrhenius fit is inadequate to describe the behavior of the Cl + ethane reaction over the entire temperature range. Consequently, rather than reporting the parameters of the fit to our data alone, we have instead undertaken a global fit to the current and previous temperature-dependent data shown in Figure 3 with the exception of the Bryukov et al.19 results (which are characterized by a much stronger temperature dependence between 300 and 1000 K than observed in other work) with a Tx dependence of the pre-exponential factor (Kooij expression):

k(Cl+C2H6) ) (1.27 ( 0.27) × 10-11 × T(0.27(0.03) × exp(-(14 ( 8)/T) cm3 molecule-1 s-1

(3)

The parameters reported in expression 3 are the result of an unweighted nonlinear least-squares fit to the data over the range 48 K e T e 800 K with the uncertainties cited at the level of a single standard deviation from the mean. No weighting was applied in the fitting procedure due to the much higher concentration of data points around room temperature. Our results for the Cl + propane reaction over the range 48 K e T e 167 K are presented and compared to previous work performed over a range of temperatures from 195 to 700 K in Figure 4. Earlier studies reporting absolute rate coefficients are displayed alongside the present work. The current Sander et al.40 recommendation is also shown. This recommendation which is valid for the range 300 K e T e 400 K comprises two components due to primary and secondary H atom abstraction (and the corresponding formation of n-propyl and isopropyl radicals, respectively), and these are also plotted in Figure 4. Previous work has shown that the overall reaction has little or no temperature dependence over the range 195 K e T e 700 K, and the present results reinforce this observation in extending the lower limit of currently available temperatureindependent kinetic data to 48 K.

C2H5 + Cl2 f Cl + C2H5Cl

(4)

Cl + C2H5 f HCl + C2H4

(5)

Cl + C2H4 + M f C2H4Cl + M

(6)

Reaction 4 has been measured to have a rate coefficient of 1.5 × 10-11 cm3 molecule-1 s-1 at 300 K48 and is seen to increase with decreasing temperature to 3.2 × 10-11 cm3 molecule-1 s-1 at 190 K. Reaction 5 has been measured to have a rate coefficient of 2.9 × 10-10 cm3 molecule-1 s-1 at 300 K and is seen to decrease to 2.0 × 10-10 cm3 molecule-1 s-1 at 220 K.49 Reaction 6, with an evaluated40 rate coefficient of 1.3 × 10-29 cm6 molecule-2 s-1 from 200 to 300 K is unlikely to play a significant role under our low pressure experimental conditions, but it is nevertheless included for completeness. Reactions 1 and reactions 4-6 were modeled using the differential integration software package FACSIMILE (MCPA Software, UK) at 127 K by extrapolating the experimental values obtained for reactions 4-6 at higher temperatures. Inputs to the model included the Cl2 and C2H6 concentrations, calculated from their flow rates and the estimated Cl and C2H5 concentrations, which were determined from the observed loss of atomic chlorine upon addition of the coreagent C2H6 and the estimated Cl2 discharge efficiency. C2H5 concentrations at the outset were nonzero due to the requirement to mix the two reagents upstream of the supersonic flow at room temperature. Input C2H4 concentrations were set to zero. The model was allowed to evolve for a fixed time period corresponding to the duration of the supersonic flow, and the resulting output chlorine atom concentrations were fit using the same nonlinear least-squares fitting method as that used in the experiments. Subsequently, these computed pseudo-first-order decay rates were plotted against the input C2H6 concentration to recover the second-order rate coefficient. It was found that, when the modeled molecular chlorine concentrations were kept to the values used during the experiments, the pseudo-first-order rate coefficients for the loss of atomic chlorine were seen to increase slightly above the expected value for low ethane concentrations. However, for high ethane concentrations, the modeled interference from reactions 4-6 was found to be negligible when the modeled molecular chlorine concentrations were kept to the values used during the experiments. Nevertheless, when [Cl2] was increased again by a factor of 10-100 at high concentrations of ethane, the pseudo-firstorder rate coefficients for the loss of atomic chlorine were seen to decrease with respect to the expected values. As a result of this “pivot” effect of the second-order plot, we would effectively observe a decrease in the measured second-order rate coefficient if the Cl2 concentration was maintained at too high a value during the experiments. Although the measured first-order decay rate was observed to increase by approximately 10% when the molecular chlorine concentration was increased by a factor of 50 for low ethane concentrations at 127 K, the resulting secondorder rate coefficient only decreased by approximately 2% when these points were included in the fit. As this difference is much less than the estimated systematic error, these points were finally

3042

J. Phys. Chem. A, Vol. 114, No. 9, 2010

included in the analysis. No decrease of the pseudo-first-order decay rates was ever observed experimentally at high concentrations of ethane. Molecular chlorine concentrations were kept to a minimum in the experiments, and no further correlation between the Cl2 concentration and the decay rates in the presence of ethane was observed at other temperatures. For the Cl + propane reaction, secondary reactions should not interfere to a significant extent due to the fast rate of the primary process. As a result, simulations were not performed on this system. The Cl + Ethane Reaction. The kinetics of the Cl + ethane reaction has been the subject of several theoretical studies.36,37 Roberto-Neto et al.36 using a range of computational methods found that the reaction was exothermic by 10 kJ mol-1 with a small activation barrier of 1.3 ( 0.8 kJ mol-1. In addition, they calculated thermal rate coefficients for this process using a relatively simple level of theory, zeroth-order variational transition state theory, with tunneling corrections giving rise to rate coefficients which were in good agreement with the experimentally determined ones at 800 K, but up to 5 times smaller at room temperature. Fernandez-Ramos et al.37 studied reaction 1 at the MP2/aug-cc-pVDZ and MP2/cc-pVTZ levels of theory for geometry optimization, and at the CCSD(T)/IB level for single point energies (with extrapolation to the infinite basis set limit). They found that the transition state (TS) was “loose” and below the reactants after correcting for zero point energy (ZPE). Moreover, they determined the presence of a post-TS van der Waals minimum, indicating that the reaction is in fact stepwise, which might play a role in the dynamics at low temperatures. They used microcanonical variational transition state theory with multidimensional corrections for tunneling in order to calculate reaction rate coefficients over the range 200 K e T e 1000 K. These values are plotted in Figure 3 alongside the experimental ones and can be seen to overestimate the high temperature values as well as overestimate the deviation from linearity of the temperature dependence. More recently, Greaves et al.39 using geometries and harmonic frequencies obtained at the MP2/aug-cc-pVTZ level and with the CCSD(T) method extrapolated to the complete basis limit calculated the reaction to be exothermic by 15.5 kJ mol-1 with a “submerged” activation barrier and therefore below the separated reactants by -11.6 kJ mol-1 after ZPE corrections. In addition to the post-TS well, they identified a shallower pre-TS well but concluded that this should have little influence on the reaction dynamics. Only the calculations by Roberto-Neto et al.36 took into consideration spin-orbit splitting in atomic chlorine. Experimentally, the reported kinetic measurements of the Cl + ethane reaction fall into two categories. The first category comprises competitive chlorination experiments22-33 where the temperature-dependent rate coefficients for the Cl + ethane reaction were primarily determined relative to the Cl + methane reaction over the range 198 K e T e 700 K. The second category comprises measurements by absolute methods3-21 using a wide variety of radical production and detection techniques. Six of these studies5,7,9,14,19,20 report the temperature dependence of the rate coefficient covering the combined range 177 K e T e 1002 K. The reaction is slightly exothermic (-8.9 ( 1.7 kJ mol-1) from tabulated gas phase enthalpy values,40 in good agreement with theoretically determined energies. However, in contrast to the theoretical studies, all of the earlier temperaturedependent experimental studies estimate the presence of an activation barrier on going from reactants to products. The current work is no exception, although, with the use of a temperature-dependent pre-exponential factor in the fit, the

Hickson et al. barrier height is now so significantly reduced as to be almost thermoneutral with a value of Ea ) 0.11 ( 0.06 kJ mol-1. The conventional Arrhenius expression recommended by Sander et al.40 and based upon a combination of absolute measurements5,7,9,12-14,18,19 seems to adequately describe the temperature dependence of the Cl + ethane reaction over the range 120 K e T e 500 K; however, it significantly underestimates both the high (>500 K) and low temperature (48 K e T e 120 K) values where curvature is observed. Similar behavior at high temperature has also been observed for the Cl + CH4 reaction. In this instance, Michelsen and Simpson50 argued that the deviation from Arrhenius behavior at high temperature might be ascribed to the higher Boltzmann population in C-H stretching modes, which have been shown to enhance the reaction probability by a factor of 30 with respect to ground vibrational state CH4.51 However, the reactivity of Cl with C2H6 with 1 quantum in the ν5 asymmetric stretch has been shown to be enhanced with respect to that of ground state C2H6 by only 5-10%.52 Given the difference in height of the activation barriers for these two reactions (∼11 kJ mol-1 for Cl + CH4 versus