Kinetics of the Methyl–Vinyl Radical + O2 Reactions Associated with Propene Oxidation

The bimolecular rate coefficients of reactions CH3CCH2 + O2 (1) and cis/trans-CH3CHCH + O2 (2a/3a) have been measured using a tubular laminar flow reactor coupled with a photoionization mass spectrometer (PIMS). These reactions are relevant in the combustion of propene. Pulsed excimer laser photolysis of a ketone or a bromide precursor molecule at 193 or 248 nm wavelength was used to produce radicals of interest homogeneously along the reactor. Time-resolved experiments were performed under pseudo-first-order conditions at low pressure (0.3–1.5 Torr) over the temperature range 220–660 K. The measured bimolecular rate coefficients were found to be independent of bath gas concentration. The bimolecular rate coefficients possess negative temperature dependence at low temperatures (T < 420 K) and appear to be independent of temperature at high temperatures (T > 420 K). Observed products of the reaction CH3CCH2 + O2 were CH3 and H2CO, while for the reaction cis/trans-CH3CHCH + O2, observed products were CH3CHO and HCO. Current results indicate that the reaction mechanism of both reactions is analogous to that of C2H3 + O2. Methyl substitution of the vinyl radical changes its reactivity toward O2 upward by ca. 50% if it involves the α-position and downward by ca. 30% if the methyl group takes either of the β-positions, respectively.


■ INTRODUCTION
Propene is the second simplest unsaturated hydrocarbon after ethene and serves as a key compound in the petrochemical industry. Propene is also an important intermediate species in the combustion of propane and larger hydrocarbons and has a significant impact on combustion processes. 1−5 The presence of a double bond in the propene molecule gives it a relatively high thermal stability. Abstraction of the weakest H atom from propene molecule produces a resonance-stabilized allyl radical, which is highly stable against unimolecular decomposition. The allyl radical reacts very slowly with O 2 and can thus reach high concentrations under high temperature conditions. Consequently, the allyl radical is an important precursor in the formation of aromatic structures which may fuse together and finally lead to the soot formation. 6−8 As shown in Scheme 1, four different radicals can be produced from propene via hydrogen abstraction reactions. Three of the four radicals (CH 3 CCH 2 , cis-CH 3 CHCH, and trans-CH 3 CHCH) are vinyltype and CH 2 CHCH 2 is the resonance stabilized allyl radical.
Considering that the bond dissociation energy (BDE, D 0 ) of vinylic hydrogens is ∼94 kJ/mol higher than that of the allylic hydrogens of propene 9−11 it is obvious that allyl radicals are produced in significantly higher abundance than methyl−vinyl radicals. Burke et al. 12 performed a reaction flux analysis of the propene oxidation experiments conducted in a jet-stirredreactor (JSR) and showed that in a mixture of 1.67% propene and 3.38% O 2 in helium (T = 970 K, equivalence ratio ϕ = 2. 19), oxidation resulted in 54% formation of allylic radical and only 9% formation of the combined other three methyl−vinyl radicals. While they are formed with a relatively low abundance, methyl−vinyl radicals are highly reactive in comparison to the very unreactive allyl radical. If one considers specifically the kinetics of the methyl−vinyl + O 2 reactions, their impact on combustion chemistry turns out to be rather important and needs to be included in accurate combustion chemistry models. The Burke et al. 12 reaction flux analysis of Scheme 1. Production of Methyl−Vinyl and Allyl Radicals from Propene via Hydrogen Abstraction Reactions a a The C−H bond dissociation energies associated with the formation of allyl radical CH 2 CHCH 2 , methyl-vinyl radicals CH 3 CCH 2 , trans-CH 3 CHCH, and cis-CH 3 CHCH are 360.7 kJ mol −1 , 443.9 kJ mol −1 , 461.1 kJ mol −1 , and 459.0 kJ mol −1 , respectively. 9−11 propene oxidation also showed that 47% of the CH 2 CHCH 2 radicals produced were consumed by the CH 2 CHCH 2 + HO 2 reaction and only 14% were consumed by the CH 2 CHCH 2 + O 2 reaction. Under the same conditions, however, almost all of the methyl−vinyl radicals produced were consumed via reaction with O 2 . 12 Slagle et al. 13 carried out an experimental kinetic study of the CH 3 CHCH + O 2 reaction employing the laser photolysis− photoionization mass spectrometry (PIMS) method to produce and detect CH 3 CHCH radicals. The bimolecular reaction rate coefficient was measured over the temperature range 296−600 K, and no dependence on temperature or bath gas concentration (pressure) was observed. However, a recent theoretical and computational study of trans-CH 3 CHCH + O 2 reaction by Chen and Goldsmith 14 show an explicit temperature dependence. Slagle et al. 13 did not report whether they were investigating a single isomer or a mixture of both isomers (cis-CH 3 CHCH or/and trans-CH 3 CHCH).
The purpose of the present work is to perform direct kinetic experiments of reactions 1−3 individually: The aim is to better understand the temperature dependence of the bimolecular rate coefficients of these oxidation reactions and compare the experimental results with the computational results of Chen and Goldsmith, 14 who combined high level ab initio calculations with master equation modeling to study the kinetics of reactions 1 and 3. To our knowledge, there are no previous measurements available on the CH 3 CCH 2 + O 2 reaction. The reaction of vinyl radical, CH 2 CH + O 2 , has key importance in the modeling of combustion processes and has been studied extensively. 15−22 The current study also aims to quantify the effect of an H atom substitution in vinyl radical by a methyl group on the oxidation kinetics.

■ EXPERIMENTAL SECTION
Pulsed excimer laser photolysis of a relevant precursor molecule was used to produce the radical of interest, and its decay was followed in real-time using a PIMS for which ionizing radiation was provided by a microwave-powered gasdischarge lamp. The experimental apparatus and procedures employed have already been described previously, 22 hence only a brief description of experimental details specific to this work are presented in this section. The reaction gas mixture flowing through the temperature controlled tubular reactor contained mostly the inert helium bath-gas (>99%) and molecular oxygen. The mixture contained a relatively small amount of precursor (<0.03%), which was introduced into the reactor by constantly bubbling helium through a temperature-controlled precursor cavity. The gas mixture containing the precursor was photolyzed using 193 nm (ArF) or 248 nm (KrF) radiation from an exciplex laser (ASX-750) along a reactor axis to achieve a homogeneous concentration of radicals along the reactor. All measurements were done under pseudo-first-order conditions, that is, [R]/ [O 2 ] ≤ 0.05. For the production of methyl−vinyl radicals, the following precursors were used: 3-Methyl-3-buten-2-one precursor was used to produce CH 3 CCH 2 radicals: Since CH 3 CO can further undergo chemically activated decomposition to give CO and CH 3 , and methyl radical is one potential product of the reaction CH 3 CCH 2 + O 2 , another precursor, 2-bromopropene, was used in product detection measurements.
As shown in Figure 1, the isomerization barrier separating the cis-CH 3 CHCH and trans-CH 3 CHCH isomers is very small (<20 kJ/mol) and can be easily surpassed even if the newly formed methyl−vinyl radical loses some part of the excess energy (>140 kJ/mol) available after laser photolysis at 193 or 248 nm. Hence it is possible that although we have used cis-1bromopropene and trans-1-bromopropene to produce cis-CH 3 CHCH and trans-CH 3 CHCH radicals, respectively, both these radicals may have isomerized to each other to some extent before being thermalized by collisions with the buffer gas molecules. Consequently, photolyzing isomerically pure precursors at either wavelength may lead to an isomeric mixture of cis-CH 3 CHCH and trans-CH 3 CHCH. The fact that we observed different decay rates depending on which precursor isomer was used indicates that equilibration of the two radical isomers was not achieved. Hereafter, the following notation is used to take this into account: reactions 2a and 3a represent the conditions in which radicals were produced using cis-1-bromopropene and trans-1-bromopropene precursors, respectively, and isomerization took place to an unknown extent. The measurements were carried out under low pressure (0.3− 1.5 Torr) conditions using a quartz or stainless steel reactor that had an inner diameter of 17 mm. The flow velocity in the reactor was maintained at about 5 ms −1 in order to ensure that the gas-mixture was completely replaced between laser pulses with the employed repetition rate of 5 Hz. The reactor tube employed was either uncoated or coated with halocarbon wax or polydimethylsiloxane. The reaction mixture was continuously sampled via a 0.4 mm diameter hole on the reactor wall forming a beam expanding into high-vacuum. A conical skimmer selected the center portion of the beam, producing a collimated beam before it entered the ionization chamber The Journal of Physical Chemistry A Article containing the PIMS. Ionization radiation (8.9−9.1 eV) was provided by a microwave-discharge of Cl 2 -helium mixture (Cllamp) with a CaF 2 window and was used for the detection of all three vinyl-type radicals, which have an m/z value of 41 (C 3 H 5 + ). The observed reaction products were ionized with a H 2 -lamp with a MgF 2 window (10.2 eV) and an Ar-lamp with a LiF window (11.6−11.8 eV). The ions produced were massselected by the quadrupole mass-spectrometer followed by their detection using an off-axis electron multiplier. The temporal ion signal (see insets in Figures 2 and 3) was recorded using a multichannel scaler from 10 ms before and up to 70 ms after each laser pulse and transferred to a computer for the subsequent analysis. Typically the R decay signal profile was accumulated from 3000−8000 repetitions before the nonlinear least-squares method was used to fit an exponential function, proportional to the radical concentration at time t, k′ is the pseudo-first-order decay rate coefficient, and A is the background signal.
Initial radical concentrations were kept low, at, or below about 2 × 10 11 cm −3 , in order to suppress any radical−radical reactions. Under pseudo-first-order experimental conditions with low initial radical concentrations, only two significant reactions, A and B, contribute to the loss rate of radical R: Reaction A is the bimolecular gas-phase reaction between R and the reactant (O 2 ) and is the primary interest in the current work. Reaction B is denoted as k wall and involves radical losses on the reactor wall as well as other slow first-order reactions (e.g., with radical precursor, with impurities in buffer gas, etc.) without added reactant. The first-order decay rate without added reactant, k wall , was determined in the beginning and end of each bimolecular rate coefficient measurement. The precursors 2-bromopropene (Sigma-Aldrich, > 99%), cis-1-bromopropene (Sigma-Aldrich, > 97%), trans-1-bromopropene (Sigma-Aldrich, > 99%), and 3-methyl-3-buten-2-one (TCI, > 95%) were degassed by several freeze−pump−thaw cycles before use. Helium (Messer-Griesheim, purity of 99.9996%) and O 2 (Aga, purity of 99.998%) were used as supplied. The gas pressures were monitored by capacitance pressure gauges. Estimated temperature uniformity of the reaction zone is ±2 K. . Relative enthalpies at 298 K of the trans-and cis-methylvinyl isomers and allyl radical along with the isomerization saddle points in kJ/mol, calculated at the CCSD(T)/cc-pVTZ//CCSD/cc-pVTZ level using the CCSD/cc-pVDZ//CCSD/cc-pVDZ vibrational frequencies. The details of electronic structure calculations are described in the Supporting Information, and Figure S1 shows calculated bond distances for the H atom shift in the transition structure associated with cis-CH 3 CHCH ↔ CH 2 CHCH 2 isomerization. The colored horizontal lines represent the maximum energy available for the isomerization of the nascent cis-CH 3 CHCH and trans-CH 3 CHCH radicals produced by the photolysis of cis-1bromopropene and trans-1-bromopropene at 193 and 248 nm, respectively, and were calculated by subtracting the bond dissociation energy (BDE) of C−Br bond [BDE(cis-1-bromopropene) = 337.6 kJ mol −1 and BDE(trans-1-bromopropene) = 334.7 kJ mol −1 ] 23 from the energy of the 193 and 248 nm photon, assuming that the accompanying Br-atom is formed in the ground state.  Table 1 along with the corresponding experimental conditions. The estimated uncertainty of the bimolecular rate coefficients is ±20%. This originates from uncertainties in fitting processes, in flow rate measurements, and slight temperature fluctuations (±2 K) during measurements.
The radical decay signals along with the corresponding product formation signals are shown in Figures 2 and 3. Formaldehyde (H 2 CO, Ar/LiF) and methyl radical (CH 3 , H 2 / MgF 2 ) were observed as products of reaction 1. For reactions 2a and 3a, the observed products were formyl radical (HCO, H 2 /MgF 2 ) and acetaldehyde (CH 3 CHO, H 2 /MgF 2 ). Assuming an analogy with the reaction C 2 H 3 + O 2 → CHO + H 2 CO, the expected products are H 2 CO and acetyl radical (CH 3 CO) for reaction 1 and HCO and CH 3 CHO for reactions 2a and 3a. The calculations of Chen and Goldsmith 14 for reactions 1 and 3 at low pressures and temperatures support our expectations. We did not directly observe any signal for the formation of CH 3 CO at m/z = 43 for reaction 1; we believe that the formed CH 3 CO underwent chemically activated decomposition to produce CH 3 and CO, of which CH 3 was observed at m/z = 15. This is likely since the products of the CH 3 CCH 2 + O 2 → H 2 CO + CH 3 CO reaction channel are formed with ∼90 kcal/ mol (377 kJ/mol) of excess energy 14 and the threshold energy for decomposition CH 3 CO → CH 3 + CO is only about 15.6 kcal/mol (65.3 kJ/mol). 24 None of the product signals were detected in the absence of O 2 . Hence, we conclude that under the experimental conditions of our measurements, the major product channels are the formation of H 2 CO and CH 3 CO for reaction 1 and the formation of HCO and CH 3 CHO for reactions 2a and 3a. The nature of the products formed for reactions 1, 2a, and 3a indicate that the reaction mechanism involves the formation of a cyclic intermediate containing an O−O bond, decomposition of which is a highly exothermic process.
The nonlinear least-squares method was used to fit the equation I t (P + ) = B + I 0 (P + )(1 − exp(−k′t)) to the formation signals of the closed-shell products CH 3 CHO and H 2 CO. Here parameters B, I 0 (P + ), and k′ are the non-time-resolved background signal, the signal value at t → ∞, corresponding to channel yield × [R vinyl ] 0 , and the pseudo-first-order decay rate coefficient. Since the other two products, CH 3 and HCO, are open-shell species and react with O 2 , although the CH 3 + O 2 reaction is very slow (k′ ≤ 0.03 s −1 ) under our experimental conditions, the following expression was fitted to the signal profiles: where k p ′ is a pseudo-first-order rate coefficient of the reaction of the open-shell species with O 2 and is treated as an individual fitting parameter. Experiments were performed at different temperatures and pressures to determine the bath gas concentration dependence and temperature dependence of the measured bimolecular rate coefficients. No bath gas concentration dependence (pressure dependence) was observed for the rate coefficients of reactions 1−3, an observation that is in agreement with the computational results of Chen and Goldsmith 14 under low-pressure conditions of the current work. Similarly, no pressure dependence has been observed for the C 2 H 3 + O 2 reaction. 15−22 Arrhenius plots of the bimolecular rate  Figure 4. Also shown are the results of previous measurements of the CH 3 CHCH + O 2 and CH 2 CH + O 2 reactions as well as the results of recent calculations of Chen and Goldsmith for the CH 3 CCH 2 + O 2 and trans-CH 3 CHCH + O 2 reactions. As can be seen from Figure 4, the measured bimolecular rate coefficient of reaction 1 has negative temperature dependence up to about 414 K and becomes temperature independent at higher temperatures. Likewise, the bimolecular rate coefficients of reactions 2a and 3a measured in this work behave similarly to reaction 1 and have negative temperature dependence up to about 363 K and become temperature independent at higher temperatures. The thermal stability of the employed precursors limited the upper bound of our experimental temperature range. The Arrhenius , where k is the bimolecular rate coefficient, A is the pre-exponential factor, E a is the activation energy, and R is the gas constant, was fitted to the measured bimolecular reaction rate coefficients of reactions 1, 2a, and 3a over the above-mentioned negative temperature dependence range. The following expressions were obtained: k 1 = (6.51 ± 0.4) × 10 −12 exp(1.69 ± 0.14 kJ mol −1 /RT) cm 3 s −1 (T = The shown error for the bimolecular rate coefficients k are 1σ fitting uncertainties + estimate of other uncertainties. The error limits in k wall are 1σ fitting uncertainties only. a Chlorine lamp with a CaF 2 window was used for detection; ArF laser (193 nm), stainless steel with halocarbon wax coated reactor with inner diameter (id) = 1.7 cm, unless otherwise stated. b Average of measured wall rates. c Wall rate determined from the linear fit y-axis intercept. d 3-Methyl-3-buten-2-one was used as the precursor. e Clean quartz reactor with id = 1.7 cm. f 2-Bromopropene was used as the precursor. g trans-1-Bromopropene was used as the precursor. h Quartz reactor with polydimethylsiloxane (PDMS) coating, id = 1.7 cm. i KrF laser (248 nm) was used for photolysis. j cis-1-Bromopropene was used as the precursor.  In Figure 4, our experimental results are compared with the computational results of Chen and Goldsmith. 14 By comparing the temperature dependencies of the CH 3 CCH 2 + O 2 and trans-CH 3 CHCH + O 2 reactions, one observes that both processes possess similar temperature dependence. In addition, by multiplying the computational results with constant factors [about 1.1 for reaction 1 and about 0.7 for reaction 3a], a good agreement can be obtained over the investigated temperature range. The Chen and Goldsmith 14 results, which are compared in Figure 4, are the total capture rate constants of CH 3 CCH 2 + O 2 and trans-CH 3 CHCH + O 2 reactions. The similarity of temperature dependencies between experiments and theory may suggest that the negative temperature dependency observed at low temperatures originates from the capture rate, which is an interesting observation.
As discussed in the Experimental Section, CH 3 CHCH radical can undergo cis ↔ trans isomerization with the surplus energy it gains in the photolysis process of the precursor molecule using 193 or 248 nm radiation. In addition, photolysis at 193 nm provides sufficient energy for further isomerization to the allyl radical, see Figure 1. The relative enthalpies of the isomers and their isomerization barriers are shown in Figure 1. However, we observed that the bimolecular rate coefficients of reactions 2a and 3a were independent of photolysis wavelength. This observation shows that any (small) amount of allyl radical possibly produced in measurements using 193 nm photolysis had a negligible effect on kinetic results, because no allyl can be formed in 248 nm photolysis, see Figure 1. Dibble et al. 25 carried out an RRKM/ Master Equation analysis of the trans ↔ cis isomerization of chemically activated 1-methylallyl (CH 3 CHCHCH 2 ) radical produced in the OH + trans-CH 3 CHCHCH 3 reaction. The analysis was performed at different levels of initial excitation as a function of total pressure and temperature. Their results suggest that an initial excitation of 5 kcal/mol (21 kJ/mol) or more above the trans ↔ cis isomerization barrier results in constant cis/trans population ratio under low-pressure conditions. The calculated cis/trans population ratio, however, depended on the employed density functional theory method. The results of Dibble et al. 25 may explain why the bimolecular rate coefficients of reactions 2a and 3a measured in this work do not depend on photolysis wavelength employed (193 or 248 nm), although the initial excitation of CH 3 CHCH radicals was significantly above the cis ↔ trans isomerization barrier.
However, we also observed that when the trans-CH 3 CHCHBr precursor was used to produce the CH 3 CHCH radicals, we obtained a slightly but systematically larger rate coefficient for the CH 3 CHCH + O 2 reaction compared with the rate coefficient obtained when the cis-CH 3 CHCHBr precursor was used. Because it is very likely that isomerization did take place to some, but unknown extent, the trans(cis)-CH 3 CHCH + O 2 and cis(trans)-CH 3 CHCH + O 2 notation was employed, as described in the Experimental Section. The current observation that the trans(cis)-CH 3 CHCH + O 2 reaction is slightly faster than the cis(trans)-CH 3 CHCH + O 2 reaction is in line with the Chen and Goldsmith 14 calculations which show that the potential energy surface (PES) of the trans-CH 3 CHCH + O 2 reaction is slightly more attractive than the PES of the cis-CH 3 CHCH + O 2 reaction. We conclude that the photolysis of the trans-CH 3 CHCHBr or cis-CH 3 CHCHBr precursor at 193 or 248 nm results in a mixture of CH 3 CHCH isomers for which the initial (nascent) isomer is still in excess after thermalization.
Current results may also be used to interpret the OH + CH 3 CHCH 2 → cis/trans-CH 3 CHCH + H 2 O (Δ r H 298.15 K ≈ −35 kJ/mol) and the subsequent cis/trans-CH 3 CHCH + O 2 reaction system. Since the cis/trans-isomerization barrier is about 18 kJ/mol (see Figure 1) at room temperature, the excitation energy of nascent cis/trans-CH 3 CHCH radical may exceed the barrier by as much as 17 kJ/mol, close to the abovediscussed 5 kcal/mol (21 kJ/mol), which in the case of 1methylallyl, was enough to lead significant isomerization before thermalization. However, in this case the H 2 O molecule takes a portion of the 35 kJ/mol energy released in the radical-forming reaction, so that the chemical excitation of the cis/trans-CH 3 CHCH radical is lower than the possible maximum, but might still be more than enough to lead to significant isomerization.
To the best of our knowledge, the kinetic measurements carried out by Slagle et al. 13 employing PIMS is the only study available in the literature regarding the cis/trans-CH 3 CHCH + O 2 reaction, and their results are presented in Figure 4. The experimental technique used by Slagle et al. 13 is similar to ours and in the temperature range 296 K−600 K, they observed that the rate coefficient of the cis/trans-CH 3 CHCH + O 2 reaction is independent of temperature and bath gas concentration. The Journal of Physical Chemistry A Article Clearly, this observation is in disparity with our results. We observe explicit negative temperature dependence for the cis/ trans-CH 3 CHCH + O 2 reactions and the temperature dependence is in good agreement with the computational results of Chen and Goldsmith. 14 Additionally, the bimolecular reaction rate coefficient Slagle et al. 13 measured is on average about 40% faster than those measured in the present work for the cis/trans-CH 3 CHCH + O 2 reactions. Such an offset associated with these two apparatuses has already been observed previously. 22,26 Finally, comparing the bimolecular rate coefficients of reactions 1, 2a, 3a, and k(CH 2 CH + O 2 ) 22 shown in Figure  4 manifest that substituting the α-H-atom in vinyl radical with a methyl group increases the rate of the reaction by almost 50%, whereas substituting either β-H-atom of the vinyl radical with a methyl group decreases the rate of reaction by around 30%. The increase of reactivity upon substituting an α-H-atom of a radical with a methyl group has been observed many times before 26 but the reduction of reactivity upon substituting a ß-H-atom with a methyl group is a new observation. 27

■ CONCLUSION
In this study, we have measured the bimolecular rate coefficients of the CH 3 CCH 2 + O 2 and cis/trans-CH 3 CHCH + O 2 reactions over wide temperature ranges at low pressures. The bimolecular rate coefficients of all the three reactions were found to have a negative temperature dependence at temperatures below about 380 K and appear to become independent of temperature at somewhat higher temperatures, in good agreement with the recent computational study of Chen and Goldsmith. 14 On the other hand, current results show that the CH 3 CCH 2 + O 2 reaction is faster and trans-CH 3 CHCH + O 2 reaction slower than their computational study suggests. As expected, no bath gas concentration dependence was observed for any of the reactions studied in the employed pressure range (0.3−1.5 Torr). The observed products of CH 3 CCH 2 + O 2 and cis/trans-CH 3 CHCH + O 2 reactions show that their reaction mechanism is analogous to the reaction of vinyl radicals with molecular oxygen.