Kinetics and Product Branching Ratio Study of the CH3O2 Self-Reaction in the Highly Instrumented Reactor for Atmospheric Chemistry

The fluorescence assay by gas expansion (FAGE) method for the measurement of the methyl peroxy radical (CH3O2) using the conversion of CH3O2 into methoxy radicals (CH3O) by excess NO, followed by the detection of CH3O, has been used to study the kinetics of the self-reaction of CH3O2. Fourier transform infrared (FTIR) spectroscopy has been employed to determine the products methanol and formaldehyde of the self-reaction. The kinetics and product studies were performed in the Highly Instrumented Reactor for Atmospheric Chemistry (HIRAC) in the temperature range 268–344 K at 1000 mbar of air. The product measurements were used to determine the branching ratio of the reaction channel forming methoxy radicals, rCH3O. A value of 0.34 ± 0.05 (errors at 2σ level) was determined for rCH3O at 295 K. The temperature dependence of rCH3O can be parametrized as rCH3O = 1/{1 + [exp(600 ± 85)/T]/(3.9 ± 1.1)}. An overall rate coefficient of the self-reaction of (2.0 ± 0.9) × 10–13 cm3 molecule–1 s–1 at 295 K was obtained by the kinetic analysis of the observed second-order decays of CH3O2. The temperature dependence of the overall rate coefficient can be characterized by koverall = (9.1 ± 5.3) × 10–14 × exp((252 ± 174)/T) cm3 molecule–1 s–1. The found values of koverall in the range 268–344 K are ∼40% lower than the values calculated using the recommendations of the Jet Propulsion Laboratory and IUPAC, which are based on the previous studies, all of them utilizing time-resolved UV–absorption spectroscopy to monitor CH3O2. A modeling study using a complex chemical mechanism to describe the reaction system showed that unaccounted secondary chemistry involving Cl species increased the values of koverall in the previous studies using flash photolysis to initiate the chemistry. The overestimation of the koverall values by the kinetic studies using molecular modulation to generate CH3O2 can be rationalized by a combination of underestimated optical absorbance of CH3O2 and unaccounted CH3O2 losses to the walls of the reaction cells employed.


■ INTRODUCTION
Methyl peroxy (CH 3 O 2 ) radicals are key species in atmospheric oxidation 1 and the combustion of volatile organic compounds. 2,3 The chemistry of CH 3 O 2 in the troposphere is typically dominated by the reaction with NO, particularly in environments influenced by anthropogenic NO x emissions (reaction R1). The reaction is a critical step in the tropospheric production of ozone in the presence of NO and converts NO into NO 2 The methoxy radicals generated by channel R4.b subsequently react with oxygen (reaction R2) to form CH 2 O and HO 2 . Despite its importance, the reported values for the rate coefficient of reaction R4, k 4 , at room temperature lie in a wide range from (2.7−5.2) × 10 −13 cm 3 molecule −1 s −1 , 5 with IUPAC 5 and the Jet Propulsion Laboratory (JPL) 6 giving 20− 40% and 40−50%, respectively, uncertainties at the 2σ level for k 4 in the temperature range 270−350 K. The previous kinetic studies used either the flash photolysis (FP) technique 7−11 or the molecular modulation (MM) method 12−14 to generate CH 3 O 2 radicals, which were coupled to time-resolved UV− absorption spectroscopy to detect CH 3 O 2 at fixed wavelengths in the range ∼210−270 nm (typically at 250 nm). As UVabsorption is a relatively insensitive technique, the detection limits of CH 3 O 2 were high, for example around 4 × 10 12 molecules cm −3 , 8,11 and the UV-absorption studies used high initial concentrations of CH 3 O 2 , 10 13 −10 14 molecules cm −3 orders of magnitude. 7−11 The kinetic studies of the CH 3 O 2 self-reaction used photolytic mixtures of CH 4 /Cl 2 /O 2 to generate CH 3 O 2 . 7−14 CH 3 O formed by the reaction R4.b is rapidly removed via the reaction with O 2 (reaction R2) in high concentrations (10 17 − 10 18 molecules cm −3 orders of magnitude) 7−14 to generate HO 2 , which quickly reacts further with another CH 3  As each HO 2 radical consumes rapidly one CH 3 O 2 species on the time scale of the CH 3 O 2 self-reaction, the determination of the overall rate coefficient of the reaction R4, k 4 , requires knowledge of the branching ratios for reaction R4 (eq 1) 7,11 k k r (1 ) where k obs is the second-order observed rate coefficient and r CH3O is the branching ratio of the reaction channel producing CH 3 O (reaction R4.b). The branching ratios in the CH 3 O 2 self-reaction were the subject of a number of experimental studies performed from the mid 1970s to 1990 inclusive, which were followed by the study of . 15 The studies used photolysis of CH 4 /Cl 2 /O 2 or (CH 3 ) 2 N 2 /O 2 and either end product detection employing mass spectrometry (MS), 16 GC-MS, 17 and infrared spectroscopy, 15,18−20 or kinetic measurements using time-resolved UV−absorption spectroscopy. 7 Some of the early studies reported a third channel of the selfreaction leading to CH 3 OOCH 3 (reaction R4.c) with an insignificant contribution to the overall reaction rate coefficient at all temperatures�such as ≤0.08 at 297 K 19 and ≤0.07 at 298 K 18 �with other studies finding no evidence for any contribution of peroxide formation. 15 IUPAC 5 and JPL 6 use the evaluation of Tyndall et al. 21 that recommends considering reactions R4.a and R4.b as the sole reaction channels of the self-reaction. The majority of the branching ratios studies were carried out at room temperature 15 21 revised the room temperature results to obtain r CH3O = 0.37 ± 0.06 at 298 K, which is recommended by IUPAC 5 and JPL 6 . There have been two experimental studies of the temperature dependence of the branching ratios, which were conducted over different temperature ranges. 7,20 Lightfoot et al. 7 found a positive temperature dependence for r CH3O between 388 and 573 K using flash photolysis in combination with time-resolved UV−absorption spectroscopy. Horie et al. 20 performed the only branching ratio experimental study covering temperatures below room temperature using matrix isolation Fourier transform infrared spectroscopy. The results were obtained from 223−333 K to show a positive temperature dependence for r CH3O . Tyndall et al. 21 combined the results of Lightfoot et al. 7 and Horie et al. 20 with their recommended value at 298 K, r CH3O = 0.37 ± 0.06, and results published prior to 1990 to describe the ratio of the rate coefficients of the t w o r e a c t i o n c h a n n e l s R 4 . a a n d R 4 . b a s T (26. 4. . The evaluation of Tyndall et al. 21 is recommended by JPL. 6 This work reports on the determination of the branching ratios and the overall rate coefficient of the CH 3 O 2 selfreaction in the temperature range of 268−344 K at 1000 mbar of synthetic air. The kinetic and branching ratio measurements were performed in the Highly Instrumented Reactor for Atmospheric Chemistry (HIRAC). Fourier Transform Infrared (FTIR) spectroscopy was employed to monitor the time profiles of the concentrations of CH 2 O and CH 3 OH produced by the CH 3 O 2 self-reaction to determine the branching ratios of the two reaction channels R4.a and R4.b.
The fluorescence assay by gas expansion (FAGE) method for the selective and sensitive detection of CH 3 O 2 radicals was used to study the kinetics of the self-reaction. 22,23 The method involves the titration of CH 3 O 2 to CH 3 O by reaction with added NO, followed by the detection of the resultant CH 3 O by off-resonant LIF with laser excitation at ca. 298 nm. 22 The FAGE instrument was calibrated for CH 3 O 2 using the 184.9 nm photolysis of water vapor in air to generate OH followed by the conversion of OH to known concentrations of CH 3 O 2 by reaction with CH 4 and O 2 . 22,23 The 184.9 nm photolysis of water vapor is a well-established method of FAGE calibration for OH and HO 2 . 24−26 The FAGE method for the CH 3 O 2 detection has been validated previously using the direct and absolute near-IR Cavity Ring Down Spectroscopy (CRDS) method to detect CH 3 O 2 . 23 FAGE measurements were carried out in the HIRAC chamber to determine the observed rate coefficient of the CH 3 O 2 self-reaction, k obs , at 1000 mbar and temperatures in the range of 268−344 K. Using k obs and the branching ratio of the reaction channel leading to methoxy radicals, r CH3O , the overall rate coefficient of the self-reaction, k 4 , was derived. This is the first kinetic study of the CH 3 O 2 self-reaction using a different detection method to that of UV−absorption spectroscopy.
The photochemistry is initiated by eight UV lamps, each of them housed in a quartz tube. The quartz tubes are mounted radially inside the chamber (aligned parallel to the chamber longitudinal axis). In order to perform experiments at temperatures different to the room temperature a thermofluid (HUBE6479 DW-therm oil) is circulated from a high capacity thermoregulator (Huber Unistat 390W) through a series of stainless steel channels welded to the outside of the chamber. To ensure that the temperature is homogeneous within the chamber, the layout of these channels is evenly distributed on the chamber outer surface and HIRAC is lagged in a 20-mmthick expanded neoprene.
The experiments were carried out at 268, 284, 295, 323, and 344 K and 1000 mbar of synthetic air obtained by mixing high purity oxygen (BOC, > 99.999%) and nitrogen (BOC, > 99.998%) in the ratio of O 2 :N 2 = 1:4. CH 4 (BOC, CP grade, 99.5%) and Cl 2 (Sigma-Aldrich, ≥ 99.5%) were delivered to the chamber. Initial reagent concentrations in HIRAC were [CH 4 ] = (2.0−3.0) × 10 17 molecules cm −3 and [Cl 2 ] = (0.3− 5.5) × 10 14 molecules cm −3 . After adding the reagents into the chamber, the lamps (Phillips, TL-D36W/BLB, λ = 350−400 nm) were turned on to generate CH 3 O 2 by Cl 2 photolysis at ∼365 nm (reaction R6) followed by reactions R7 and R8. In the kinetic experiments the lamps were turned on for about 5 min, and then they were turned off to record the generated CH 3 O 2 kinetic decay. In the experiments performed to determine the product branching ratio the lamps were turned on to measure CH 3 OH and CH 2 O using FTIR spectroscopy for a typical time of 20 min.
Fourier Transform Infrared (FTIR) Measurements. An in situ multipass FTIR (Bruker IFS66) arrangement along the long axis of HIRAC was used to measure the concentrations of CH 2 O and CH 3 OH produced during the times with the lamps turned on. The multipass Chernin arrangement within the chamber was optimized for 72 internal reflections giving an approximate total path length of 128.5 m. 27,29 IR spectra were recorded every 30−60 s as the average of 30−100 scans at 1 cm −1 resolution. The concentration−time profiles for CH 2 O and CH 3 OH were obtained using the absorption at around 1740 cm −1 due to the stretch of the C�O bond of CH 2 O and at around 1030 cm −1 due to the C−O stretch of CH 3 OH and using reference spectra taken of formaldehyde and methanol. Reference spectra were taken delivering CH 2 O and CH 3 OH in known concentrations to the chamber under the same conditions as those used for the CH 3 O 2 self-reaction experiments. The reference compound, either CH 2 O or CH 3 OH, was delivered in the vapor phase by direct heating of either liquid CH 3 OH or para-formaldehyde powder in a glass finger connected to a one liter stainless steel cylinder to achieve a gas pressure of a few mbar. Then the gas was delivered from the cylinder to the chamber using a flow of N 2 (p N2 = 2000 mbar).
The reference spectra of CH 2 O and CH 3 OH were fitted to the observed IR absorbance recorded as a function of wavelength (λ) and time (t), A obsd λ,t , between ∼1600−1900 cm −1 and ∼900−1120 cm −1 , respectively, at each time point to determine the changes to concentrations of CH 2 O and CH 3 OH vs time during the period of time with the lamps switched on (eq 2).
Here  22,23,30 The instrument sampled gas through a 1mm-diameter pinhole mounted on one end of a 50-mm-i.d. flow tube at a rate of ∼3 slpm. The pressure inside the sampling tube was maintained at 3.3 mbar for a chamber pressure of 1000 mbar of synthetic air. A CH 3 O fluorescence detection cell was integrated in the tube at ∼600 mm distance from the pinhole. About 25 mm prior to the detection cell, high purity NO (BOC, N2.5 nitric oxide) was injected at 2.5 sccm using a mass flow controller (Brooks 5850S) into the center of the gas flow to convert CH 3 O 2 radicals into CH 3 O. CH 3 O radicals were subsequently detected by LIF spectroscopy, directing laser light at λ online ≅ 297.79 nm to excite the A 2 A 1 (ν 3 ′ = 3) ← X 2 E(ν 3 ″ = 0) transition of CH 3 O with a 5 kHz pulse repetition frequency through the cell at a right angle to the gas flow. The off-resonant red-shifted LIF (320−430 nm) was monitored using photon counting. The laser background was measured at a wavelength of λ online + 2.5 nm and then subtracted to obtain the fluorescence signal.
The FAGE technique requires calibration to convert the measured fluorescence signal into CH 3 O 2 concentration. The calibration procedure has been described in detail previously 22,23 and hence only the important points are presented here. OH radicals were generated photolyzing water vapor in synthetic air at 184.9 nm to react with methane in excess (BOC, CP grade, 99.5%) to generate CH 3 O 2 . The produced air/radical mixture was then sampled by the FAGE instrument. The concentration of CH 3 O 2 was determined using eq 4.
Here σ is the absorption cross section of water vapor at 184.9 nm, (7.2 ± 0.2) × 10 −20 cm 2 molecule −1 ; 31,32 Φ is the photodissociation quantum yield of OH at 184.9 nm (unity), t is the photolysis time, and F is the lamp flux at 184.9 nm, which was varied to generate a range of CH 3 O 2 radical concentrations. The product F × t was determined employing chemical actinometry. 28 The FAGE calibration factor was utilized to determine [CH 3 O 2 ] in the HIRAC experiments: where S CHd 3 Od 2 (counts s −1 mW −1 ) is the recorded signal.
Previous studies have shown that the FAGE sensitivity toward OH does not depend on the chamber temperature in the range 263−344 K, 33 and thus the calibration factor determined at a The Journal of Physical Chemistry A pubs.acs.org/JPCA Article room temperature of 295 K, C CHd 3 Od 2 = (5.0 ± 1.7) × 10 −10 counts cm 3 molecule −1 s −1 mW −1 , was used in the kinetic analysis at all temperatures.

Product Branching Ratios in the CH 3 O 2 Self-Reaction.
To determine the branching ratios in the CH 3 O 2 self-reaction (reactions R4.a and R4.b) the time profiles of the concentrations of the self-reaction products CH 3 OH and CH 2 O generated by turning the lamps on were recorded employing FTIR spectroscopy at a time intervals of 30−60 s. Over the first few minutes of the reaction [CH 3 Figure S6).
Using the integrating rate ratio for the two parallel reactions R4.a and R4.b, eq 6 is obtained. Using eq 7 the branching ratio of the reaction channel R4.b, which produces CH 3 O, is given by The branching ratio r CH3O was determined using [CH 2 O] overall and [CH 3 OH] measured at early reaction times, when the product concentrations increased linearly in time ( Figure S5 in the Supporting Information). A number (6−20) of values of r CH3O were obtained at each temperature, 268, 284, 295, 323, and 344 K. Figure S7 (Supporting Information) shows that there was no trend with time in the extracted values over the initial few minutes used to determine r CH3O , and thus the secondary reactions of CH 2 O and CH 3 OH can be neglected in the analysis.
The mean values of r CH3O are shown in Figure 1 and Table  S2 (Supporting Information). The results show a positive temperature dependence which can be characterized by r CH3O = 1/{1 + [exp(600 ± 85)/T]/(3.9 ± 1.1)}, i.e., k 4.b /k 4.a = (3.9 ± 1.1) × exp(−600 ± 85)/T. There have been two temperature dependence studies of the branching ratios previously, in the range 388−573 K 7 and between 223−333 K. 20 The result obtained by Horie et al., 20 Figure 1, as the temperature range used by these authors overlaps with the range of temperatures where the measurements reported in the present work were carried out. In addition, Figure 1 shows r CH3O derived from the evaluation of Tyndall et al., 21 which is recommended by the Jet Propulsion Laboratory (JPL) evaluation. 6 The values found by this work have overlapping error limits with the results reported by Horie et al. 20 and the values given by the recommendation of Tyndall et al. 21 at the 2σ level. However, the temperature dependence measured by Horie et al. 20 and the temperature dependence recommended by Tyndall et al. 21 are steeper than the increase in the value of r CH3O with the temperature found in this study. The result at 295 K, r CH3O(this work) = 0.34 ± 0.05, is between the result of Horie et al., 20 15 The study of Horie et al. 20 is the single experimental study performed at temperatures in a range around room temperature, i.e., (293 −70 +40 ) K. The results of the present study agree well with the values reported by Horie et al. 20 at 323 and 344 K ( Figure 1). However, going down in temperature r CH3O(this work) is increasingly higher than r CH3O (Horie et al.) . 20 Horie et al. 20   The Journal of Physical Chemistry A pubs.acs.org/JPCA Article found no evidence for the formation of CH 3 OOCH 3 by the reaction channel R4.c. However, two sets of numerical simulations were performed: assuming a branching ratio of 0.1 for reaction R4.c and excluding reaction R4.c from the chemical mechanism used in the numerical simulations. Figure  1 shows the reported temperature dependence derived averaging the results generated by the two sets of simulations. 20 20 trapped the reaction products outside the reaction cell in a CO 2 matrix at 50 K to analyze them by IR spectroscopy to obtain [CH 2 O]/ [CH 3 OH], which was then used in the determination of r CH3O . The concentrations of CH 2 O produced by the self-reaction were corrected taking into account CH 2 O formed in the matrix using a correction factor less than 10%. The lower values obtained for r CH3O(Horie et al.) relative to r CH3O(this work) at T ≤ 295 K can be explained by a process leading to the CH 2 O removal enhanced by reducing the reaction temperature which was not included in the reaction mechanism employed in the analysis performed by the authors. 20 Horie et al. 20 reported evidence of aerosol formation at 213 K resulting in unaccounted removal of CH 2 O leading to a value of [CH 2 O]/[CH 3 OH] lower than unity, a result which was not expected based on the reaction mechanism; the results obtained at 213 K were thus excluded from the analysis. The experiments below room temperature used in the determination of r CH3O �i.e., in the range 223−298 K�were reported "free" of aerosols. However, [CH 2 O] and [CH 3 OH] in the experiments were relatively large, a few times higher than in the present work, increasing the potential of oligomers/particle formation at low temperatures.
Kinetics of the CH 3 O 2 Self-Reaction. Figure 2 shows examples of CH 3 O 2 decay generated by turning the HIRAC lamps off following the production of CH 3  The observed rate coefficient is larger than the second-order rate coefficient of just the CH 3 O 2 recombination reaction (R4), k 4 , as the methoxy radicals generated by channel R4.b react rapidly with molecular oxygen, which is present in large excess, 5 × 10 18 molecules cm −3 , to produce HO 2 (reaction R2), which in turn reacts with another CH 3 O 2 radical (reaction R5). As each HO 2 radical consumes one CH 3 O 2 species (reaction R5) on the time scale of reaction R4, k 4 is derived from k obs as follows: 7,11 k k r (1 ) where r CH3O is the branching ratio for the reaction channel R4.b. The applicability of eq 1 in the analysis of the kinetic data generated by the HIRAC experiments was demonstrated by modeling the observed temporal decays using a variety of The results showed that the removal of HO 2 by wall loss is negligible and thus can be excluded from the model. 22 Figure 3 shows the determined temperature dependence for k 4 . At all temperatures employed, the values of k 4 obtained using both k obs and r CH3O determined in this work are practically the same as the k 4 values obtained using k obs determined in this work and r CH3O given by the evaluation of Tyndall et al., 21 which is recommended by JPL. 6 Using the value of r CH3O = 0.34 ± 0.05 determined at 295 K in this work the rate coefficient of the overall reaction k 4 (295 K) = (2.0 ± 0.7) × 10 −13 cm 3 molecule −1 s −1 (uncertainties at 2σ level). Using the value of r CH3O (295 K) = 0.36 ± 0.12 recommended by Tyndall et al. 21 does not change the result at this level of precision: k 4 (295 K) = (2.0 ± 0.9) × 10 −13 cm 3 molecule −1 s −1 (uncertainties quoted at 2σ level). The negative temperature   The previous studies upon which the JPL 6 and IUPAC 5 recommendations are based utilized the UV-absorption of CH 3 O 2 at fixed wavelengths in the range ∼210−270 nm (typically 250 nm) usually to determine the ratio between the observed rate coefficient and the absorption cross-section of CH 3 O 2 , k obs /σ CH3O2 . 5,6 The values used for σ CH3O2 by the previous UV-absorption studies vary significantly, between (2.5−4.8) × 10 −18 cm 2 molecule −1 at 250 nm, leading to a large variation in k obs across the studies, which at 298 K ranges between k obs = (3.0−5.9) × 10 −13 cm 3 molecule −1 s −1 . 7−14 The 2020 JPL 6 evaluation report recommends the cross sections obtained by the re-evaluation of Tyndall et al. in 2001 21 of the previous reported UV-absorption spectra. At 250 nm Tyndall et al. 21 recommend σ CH3O2(250 nm) = 3.8 × 10 −18 cm 2 molecule −1 . Our calculations show that using σ CH3O2 reported by Tyndall et al. 21 and the ratios k obs /σ CH3O2 found in the previous kinetic studies, [7][8][9][10][11]13,14 the range of the values of k obs at 298 K is reduced to (4.1−5.1) × 10 −13 cm 3 molecule −1 s −1 . The present result at 298 K, k obs = 2.9 × 10 −13 cm 3 molecule −1 s −1 , is 30% smaller than the lowest value of 4.1 × 10 −13 cm 3 molecule −1 s −1 and 40% lower than the JPL 6 and IUPAC 5 recommendations of k obs = 4.8 × 10 −13 cm 3 molecule −1 s −1 .
There is a significant difference between the results obtained here and the recommendations, 5,6 and here we explore possible reasons for this discrepancy. The previous studies used either the flash photolysis (FP) technique 7−11 or the molecular modulation (MM) method 12−14 to generate CH 3 O 2 radicals employing photolytic mixtures of CH 4 /Cl 2 /O 2 . The discrepancy between k obs determined in here and the results reported in the FP studies 7−11 could be due to unaccounted secondary chemistry of CH 3 O 2 due to the high radical concentrations, on the order of [CH 3 O 2 ] of 10 13 −10 14 molecules cm −3 , and/or unaccounted spectral interferences. The MM experiments 12−14 used 1−2 orders of magnitude lower concentrations of CH 3 O 2 than [CH 3 O 2 ] in the FP studies, 7−11 i.e., concentrations on the order 10 12 molecules cm −3 , to minimize the impact of the secondary chemistry on k obs . However, as this method consisted of modulating the photolysis and hence the production of radicals by alternating the time with the lamps switched on and the time with the lamps turned off, a potential important source of error was the buildup of products absorbing in the UV range of the measurements (see below). The contributions of the absorbing products (see below) were subtracted from the overall absorbance measured by the MM experiments 12,14 to extract the absorbance of CH 3 O 2 , and thus the extracted absorbance depended on the concentrations and the cross sections attributed to the products. Note that the LIF method is selective and more sensitive, with a limit of detection for CH 3 O 2 of 2.0 × 10 9 molecules cm −3 for a signalto-noise ratio of 2, 1 s averaging time of the online data points measured during the kinetic decay, and 60 s averaging period for the offline data points recorded at the end of the experiment. Therefore, the LIF method requires significantly lower radical concentrations than the FP and MM studies; here, [CH 3 O 2 ] 0 = (0.1−1) × 10 12 molecules cm −3 , which helps to minimize potential secondary chemistry.
The FP studies 7−11 typically derived the k obs /σ CH3O2 ratio fitting either eq 10 or eq 11 to the measured optical absorbance (A t ) or absorption coefficient (α t ) at/around 250 nm.
Here A 0 and α 0 are the absorbance and the absorption coefficient at the time zero of the reaction, respectively, and l is the total optical path length.
To investigate the potential impact of the secondary chemistry on the CH 3 O 2 kinetic decays in the FP studies 7−11 numerical simulations were performed using a reaction system at 298 K described in the Supporting Information (Table S1). The previous kinetic studies of the CH 3 O 2 recombination reaction 7−14 did not investigate the impact of the CH 3 O 2 reaction with the ClO radicals produced by the CH 3 O 2 + Cl reaction on the CH 3 O 2 kinetic decay. The kinetics of the CH 3 O 2 + Cl reaction 34,35 was studied in the years around 1995 and thus after all the previous kinetic studies of the CH 3 O 2 self-reaction (1980−1990). 5,6 The CH 3 O 2 + Cl reaction is fast producing ClO (rate coefficient of 7.7 × 10 −11 cm 3 molecule −1 s −1 at 298 K). 34 The generated ClO radicals predominantly react with CH 3 O 2 with an overall rate coefficient of 2.4 × 10 −12 cm 3 molecule −1 s −1 at 298 K. 6  Figure  S3). The fit of eq 9 to the CH 3 O 2 decay provided a value of k obs (fit) = (3.0 ± 0.1) × 10 −13 cm 3 molecule −1 s −1 at 298 K, which results in k 4 (fit) = 2.2 × 10 −13 cm 3 molecule −1 s −1 , i.e., practically the same as the value measured by the present experiments k 4 = 2.1 × 10 −13 cm 3 molecule −1 s −1 . Therefore, no impact by the secondary chemistry of CH 3 O 2 included in the simulations (Supporting Information) on the value determined for k 4 was found in the present study. Hence in the absence of secondary chemistry the value of k 4 , as determined in this work, is considerably lower than k 4 when the secondary chemistry is present as a result of much higher initial [Cl] concentrations.
We now consider potential spectral interferences in previous work which monitored CH 3 O 2 concentrations using UV absorption. At the typical λ = 250 nm used to monitor the CH 3 ) where α i, t is the absorption coefficient of species i (CH 3 O 2 or ClO) at reaction time t, σ i is the absorption cross-section of species i at 250 nm and [i] t is the concentration of species i at time t. Figure S4 in the Supporting Information shows the generated α CH3O2, t and α ClO, t and their sum, Σα i, t = α CH3O2, t + α ClO, t . The results ( Figure S4) show that there is a minor contribution of α ClO, t to Σα i, t at the start of the reaction, i.e., ∼8% in the first millisecond of the reaction, which drops to 3% at t = 10 ms. The fit of the eq 11 (see above) to Σα i, t vs time resulted in k obs /σ CH3O2 . Using σ CH3O2 = 3.8 × 10 −18 cm 2 molecule −121 a value of k obs = (3.9 ± 0.1) × 10 −13 cm 3 molecule −1 s −1 is obtained, which is 3% higher than k obs given fitting eq 9 to the temporal decay of [CH 3 O 2 ] generated by the same numerical simulations ( Figure S1). The result suggests that there was no significant optical interference due to the ClO absorption which impacted the previous determinations of k obs using the FP technique. 7−11 The molecular modulation (MM) studies 12−14 used the time-resolved modulated UV-absorption (absorption waveform) generated in the range 210−270 nm via switching the photolysis lamps on and off with a typical frequency of 10 −1 Hz (order of magnitude) to determine k obs and σ CH3O2 . The absorption waveform consisted of an initial rise followed by a pseudo-steady-state and then a decay during the dark phase of the modulation period. 12−14 The modulated absorption components depended on a relatively large number of parameters: illumination time, rate of Cl 2 photolysis, kinetic parameters, and absorbing species cross sections. 12−14 Photolytic mixtures of CH 4 /Cl 2 /O 2 were flowed through the reactor to minimize the buildup of the products. However, the residence times of the gases in the reaction cell were relatively long, for example, 35 and 60 s. 13,14 The contributions of the absorbing products accumulating over the photolysis cycles was calculated and subtracted from the observed absorption to derive the modulated absorption of CH 3 O 2 . 12,14 The contribution of the products were important in the range 210−250 nm where the cross sections of HO 2 (formed by reaction R2 and the reactions of Cl with the products CH 3 OH The Journal of Physical Chemistry A pubs.acs.org/JPCA Article and CH 2 O) and CH 3 OOH (produced by the CH 3 O 2 + HO 2 reaction) increases rapidly with decreasing λ. 6 However, the cross sections of HO 2 and/or CH 3 OOH were significantly overestimated in the MM studies. 12−14 The references cited for the cross-section of HO 2 by Cox and Tyndall 12 reported σ HO2 larger than the JPL recommendations 6 by 60−70% in the range 210−250 nm 37 and by ∼10% between 210−220 nm. 38 The cross-section of HO 2 39 used by Simon et al. 14 is ∼30% larger than the JPL recommendation 6 between 210−240 nm and by ∼60% at 250 nm. Both the studies of Cox and Tyndall 12 and Jenkin et al. 13 employed values for σ CH3OOH at least 40−50% higher than the JPL recommendation in the range 210−250 nm. 6 An overestimation of the contributions of these species to the measured absorbance results in an underestimation of the CH 3 O 2 absorbance, A CH3O2 , which in turn results in an overestimation of k obs /σ CH3O2 (eq 10).
In addition, the CH 3 O 2 loss to the walls of the reaction cell were not accounted for by neither the FP studies 7−11 nor the MM studies 12−14 and could also result in an overestimation of k obs . As the CH 3 O 2 self-reaction is slow, the wall-loss could significantly contribute to the overall CH 3 O 2 removal in the previous studies.

CALIBRATION
As LIF is not an absolute detection method, the FAGE instrument required calibration and a calibration factor, C CHd 3 Od 2 , is used to convert the measured signal, S CHd 3 Od 2 , to the CH 3 O 2 concentration: To calibrate FAGE, CH 3 O 2 radicals were generated in known concentrations employing the 184.9 nm photolysis of water vapor in synthetic air followed by the complete conversion of the generated OH radicals to CH 3 O 2 by reaction with CH 4 in a large excess in the presence of O 2 . Previous work described in detail the water vapor method of calibration and the uncertainties in the calibration factor, C CHd 3 Od 2 (water vapor method). 22 As seen previously, 22 in this work an overall 34% error at 2σ level was obtained for C CHd 3 Od 2 (water vapor method) combining the systematic and statistical uncertainties. Similar overall errors, 31% and 36%, were reported for C HOd 2 (water vapor method) previously. 28,30 The photolysis of water vapor at 184.9 nm represents the most common method used to generate accurate concentrations of OH and HO 2 . The method has been applied for many years for the calibration of FAGE instruments. 24−26 The reliability of the method has been confirmed by intercomparisons with alternative methods of calibration for OH and HO 2 . The calibration of OH using the water vapor photolysis and the OH calibration based on the generation of OH by ozone reactions with alkenes have been found to agree within their experimental uncertainties. 40 Very good agreement (difference within 1−13%) has been obtained by comparing the OH measurements in the SAPHIR atmospheric simulation chamber using a number of FAGE instruments and instruments employing differential optical laser absorption spectroscopy (DOAS) and chemical ionization mass spectrometry (CIMS). 41,42 In the case of HO 2 , C HOd 2 (water vapor method) and the calibration factor obtained analyzing the kinetic decay of HO 2 by its self-reaction generated in HIRAC, C HOd 2 (kinetic method), were found in a very good agreement (difference within 8%). 28,30 However, a discrepancy within ∼40% was found between C CHd 3 Od 2 (water vapor method) and the CH 3 O 2 calibration factor determined using the kinetics of the secondorder recombination of CH 3 O 2 observed in HIRAC and k obs (298 K) = 4.8 × 10 −13 cm 3 molecule −1 s −1 , 5,6 C CHd 3 Od 2 (kinetic method). 22,23 As the error in the fraction of OH which is converted to CH 3 O 2 upon the addition of methane in the water vapor method is minor (4% at 2σ level), 22 the discrepancy between the two calibration methods can be attributed to an overestimation of the reported value of k obs for the CH 3 O 2 self-reaction at 298 K. 5,6 The ∼40% difference in C CHd 3 Od 2 (kinetic method) and C CHd 3 Od 2 (water vapor method) resulted in different values for the gradient of the correlation plot of [CH 3 O 2 ] measured by FAGE (y-axis) as a function of [CH 3 O 2 ] measured by near-infrared cavity ring down spectroscopy, CRDS (x-axis), at 1000 mbar of synthetic air using the sensitivities from the two methods of calibration of FAGE: 1.35 ± 0.07 (water vapor calibration) and 0.92 ± 0.05 (kinetic method of calibration). 23 The results show a significantly better agreement with the kinetic method than with the water vapor method. A very good level of agreement between the FAGE and CRDS measurements of CH 3 O 2 was also obtained using the kinetic method for FAGE calibration at 100 mbar of synthetic air and 80 mbar of 3:1 He:O 2 mixture. The very good agreement achieved under all conditions when the kinetic method was employed for the FAGE calibration was expected as the kinetic method was also used to determine the absorption cross section of CH 3 O 2 from the temporal decays of the optical absorption coefficient of CH 3 O 2 and hence calibrate the CRDS method. Therefore, with both FAGE and CRDS calibrated using the same method, the intercomparison was not subject to any error in the rate coefficient, k obs for the CH 3 O 2 self-reaction, and the obtained very good agreement provides a validation of the FAGE (water vapor) method to determine concentrations of CH 3 O 2 . The present result at 298 K, k obs = 2.9 × 10 −13 cm 3 molecule −1 s −1 , shows a 40% reduction in the reported value of k obs = 4.8 × 10 −13 cm 3 molecule −1 s −1 . 5,6 A reduction of 40% in the reported k obs would bring [CH 3 O 2 ] CRDS generated using the kinetic method of calibration into agreement with [CH 3 O 2 ] FAGE determined using the water vapor method. 23 Therefore, the FAGE−CRDS intercomparison also suggests that k obs = 2.9 × 10 −13 cm 3 molecule −1 s −1 and thus k 4 = 2.1 × 10 −13 cm 3 molecule −1 s −1 at 298 K, consistent with the values found in the present study.

■ CONCLUSIONS
Experiments were carried out in the range 268−344 K and at 1000 mbar to measure CH 3 OH and CH 2 O generated by the CH 3 O 2 self-reaction in HIRAC using in situ FTIR detection to determine the product branching ratios in the self-reaction. The chemistry was initiated using photolysis of Cl 2 /CH 4 /N 2 / O 2 mixtures (photolysis range: λ = 350−400 nm). The temperature dependence of the product branching ratio of the reaction channel producing CH 3 O can be described as r CH3O  17. This is the second experimental study of the temperature dependence of the product branching ratios in a range including temperatures relevant for atmospheric chemistry. The positive temperature dependence found for r CH3O is less marked than the increase in r CH3O with temperature measured by the previous study performed in a temperature range around 298 K (223−333 K), which used matrix isolation FTIR. 20 The results of the present work for product branching agree well with the values obtained by Horie et al. 20 at 295, 323, and 344 K. By decreasing the temperature, the present results are increasingly higher than the results reported by Horie et al. 20 with a positive deviation of 22% at 268 K. The kinetics of the CH 3 O 2 self-reaction has been studied coupling a FAGE instrument to HIRAC to carry out timeresolved measurements of the CH 3 O 2 concentrations during the reaction. Second-order decays of CH 3 O 2 were generated by turning the chamber lamps off. The observed rate coefficient at 295 K and 1000 mbar was k obs = (2.7 ± 0.9) × 10 −13 cm 3 molecule −1 s −1 . Using k obs (295 K) = k 4 (1 + r CH3O ) with r CH3O(this work, 295 K) = 0.34 ± 0.05 the second-order rate coefficient for the self-reaction at 295 K and 1000 mbar is k 4 = (2.0 ± 0.7) × 10 −13 cm 3 molecule −1 s −1 ; employing the recommended value of 0.36 for r CH3O at 295 K by JPL 6 and IUPAC 5 does not change the result. The result at 295 K is is ∼40% lower than the IUPAC and JPL recommendations: k 4 = 3.5 × 10 −13 cm 3 molecule −1 s −1 . The temperature dependence of the overall rate coefficient can be parametrized as k 4 = (9.1 ± 5.3) × 10 −14 × exp((252 ± 174)/T) cm 3 molecule −1 s −1 . The present results have overlapping error limits at the 2σ level with both JPL and IUPAC recommendations. However, on average the results of this work are ∼40% lower than the values calculated using the JPL 6 recommendation for the temperature dependence (k 4 = 9.5 × 10 −14 × exp(390/T) cm 3 molecule −1 s −1 ) and the values obtained employing the temperature dependence recommended by IUPAC, 5 k 4 = 1.03 × 10 −13 × exp((365 ± 200)/T) cm 3 molecule −1 s −1 .
The previous kinetic studies utilized UV−absorption spectroscopy and may be impacted by secondary chemistry owing to the high radical concentrations generated in the reaction mixtures. Chemical modeling using the conditions of the previous studies and which included secondary chemistry of Cl species showed that the secondary chemistry increases the value of k 4 obtained significantly. The FAGE method detects CH 3 O 2 sensitively, with a limit of detection of 2.0 × 10 9 molecules cm −3 for a signal-to-noise ratio of 2, 1 s online averaging time, and 60 s offline averaging period. Therefore, the experiments reported here required a few orders of magnitude lower concentrations of CH 3 O 2 than [CH 3 O 2 ] used in the UV−absorption studies and the impact of the secondary chemistry on the kinetic decays obtained by this work is negligible. In addition, the FAGE method probes CH 3 O 2 selectively, in the absence of any interference from other species.
Numerical models predict that CH 3 O 2 is the most abundant RO 2 species in the atmosphere. Even though CH 3 O 2 has not been selectively measured in the atmosphere so far, its concentration at daytime has been estimated to peak in the range of daytime peak [HO 2 ], i.e., at 10 7 −10 8 molecules cm −3 . 43−45 The atmospheric fate of CH 3 O 2 is typically dominated by the reaction with NO, with the CH 3 O 2 + HO 2 reaction becoming the main daytime loss of CH 3 O 2 under low NO x levels. As CH 3 O 2 and HO 2 reach similar levels at daytime and 298K, k CH3O2+HO2 (5.2 × 10 −12 cm 3 molecule −1 s −1 ) is ∼15 times faster than the JPL and IUPAC recommendations for k 4 (3.5 × 10 −13 cm 3 molecule −1 s −1 ) 5,6 and ∼25 times higher than k 4(CH3O2+CH3O2) determined in this work (2.1 × 10 −13 cm 3 molecule −1 s −1 ) the inclusion of the present k obs(CH3O2+CH3O2) value in the atmospheric models might not impact significantly the daytime radical budget predicted by the models. However, at night-time it has been predicted that the self-reaction is the dominant removal of CH 3 O 2 due to a rapid loss of HO 2 under dark conditions 46