Kinetics of a Criegee intermediate that would survive high humidity and may oxidize atmospheric SO₂

Significance

Ozonolysis of alkenes produces highly reactive Criegee intermediates. Whereas water dimer efficiently scavenges the simplest Criegee intermediate CH₂OO in the troposphere, this study clear demonstrates that water vapor does not react with dimethyl substituted Criegee intermediate (CH₃)₂COO, at least not fast enough to significantly consume (CH₃)₂COO in the troposphere. On the other hand, (CH₃)₂COO reacts with SO₂ three times faster than CH₂OO does, indicating Criegee intermediates of a structure similar to (CH₃)₂COO are potential candidates for an efficient oxidant in the atmospheric SO₂ oxidation.

Abstract

Criegee intermediates are thought to play a role in atmospheric chemistry, in particular, the oxidation of SO₂, which produces SO₃ and subsequently H₂SO₄, an important constituent of aerosols and acid rain. However, the impact of such oxidation reactions is affected by the reactions of Criegee intermediates with water vapor, because of high water concentrations in the troposphere. In this work, the kinetics of the reactions of dimethyl substituted Criegee intermediate (CH₃)₂COO with water vapor and with SO₂ were directly measured via UV absorption of (CH₃)₂COO under near-atmospheric conditions. The results indicate that (i) the water reaction with (CH₃)₂COO is not fast enough (k_H2O < 1.5 × 10⁻¹⁶ cm³s⁻¹) to consume atmospheric (CH₃)₂COO significantly and (ii) (CH₃)₂COO reacts with SO₂ at a near–gas-kinetic-limit rate (k_SO2 = 1.3 × 10⁻¹⁰ cm³s⁻¹). These observations imply a significant fraction of atmospheric (CH₃)₂COO may survive under humid conditions and react with SO₂, very different from the case of the simplest Criegee intermediate CH₂OO, in which the reaction with water dimer predominates in the CH₂OO decay under typical tropospheric conditions. In addition, a significant pressure dependence was observed for the reaction of (CH₃)₂COO with SO₂, suggesting the use of low pressure rate may underestimate the impact of this reaction. This work demonstrates that the reactivity of a Criegee intermediate toward water vapor strongly depends on its structure, which will influence the main decay pathways and steady-state concentrations for various Criegee intermediates in the atmosphere.

Unsaturated hydrocarbons are emitted into the atmosphere in large quantities from either human or natural sources. Ozonolysis of unsaturated hydrocarbons produces highly reactive Criegee intermediates (CIs) (1), which may (i) decompose to radical species like OH radicals or (ii) react with a number of atmospheric species, for example, with SO₂ to form SO₃ and with NO₂ to form NO₃ (2, 3). The SO₂ oxidation by CIs has gained special attentions because the SO₃ product would be converted into H₂SO₄, an important constituent of aerosols and acid rain (4–8). For example, Mauldin et al. (4) have speculated that Criegee intermediate reactions with SO₂ may account for the discrepancy between the observed and modeled concentrations of H₂SO₄ in a boreal forest region, where various alkenes are emitted by trees.

Recently, Welz et al. (2) demonstrated an efficient method to prepare a CI in a laboratory by the reaction of iodoalkyl radical with O₂ (for example, CH₂I + O₂ → CH₂OO + I). This method can produce a CI of high enough concentration that allows direct detection. With photoionization mass spectrometry (PIMS) detection, Welz et al. (2) measured the rate coefficients of the simplest CI (CH₂OO) reactions with SO₂ and NO₂. Notably, these new rate coefficients, confirmed by a few later investigations (9–11), are orders of magnitude larger than those previously used (12, 13) in atmospheric models (e.g., MCM v3.3, available at mcm.leeds.ac.uk/MCM/browse.htt?species=CH2OO), suggesting a greater role of CIs in atmospheric chemistry. This result also indicates previous ozonolysis analyses may be affected by complicated and partly unknown side reactions and may contain errors in some of the reported rate coefficients.

Typical water concentration in the troposphere (1.3 × 10¹⁷ to 8.3 × 10¹⁷ cm⁻³ at the dew point of 0–27 °C) is orders of magnitude higher than those of atmospheric trace gases like SO₂, NO₂, and volatile organic compounds (VOC) (on the order of 10¹² cm⁻³ or less). Although it has been shown that CIs may react very fast with SO₂, NO₂, and organic acids (2, 3, 14), the reactions of CIs with atmospheric water vapor would still strongly influence the fates and concentrations of atmospheric CIs (see Fig. 1 for a simplified schematic). As expected, the reactivity of CIs toward water vapor would govern the modeling results of atmospheric H₂SO₄ formation from CIs (8, 15, 16).

Fig. 1.

Reaction scheme showing competitions for CIs between reactions with water (monomer and dimer) and with SO₂.

However, there had been discrepancies about the reactivity of CIs toward water. Whereas studies (17–20) using C₂H₄ ozonolysis as a CH₂OO source show substantial reactivity of CH₂OO toward water vapor, despite a large scatter (10⁻¹⁷ to 10⁻¹² cm³s⁻¹) in the reported rate coefficient, other studies (2, 10, 21) using the CH₂I+O₂ reaction as a CH₂OO source reported negative observation for the CH₂OO reaction with water vapor.

More recently, Chao et al. (22) and Berndt et al. (23) investigated the reaction of CH₂OO with water vapor using the CH₂I+O₂ reaction and the C₂H₄ ozonolysis as their CH₂OO sources, respectively. Both groups observed clear second-order kinetics with respect to the concentration of water and concluded that reaction with water dimer predominates in the decay of CH₂OO under atmospheric conditions and that previous studies may require some reinterpretations. The reported rate coefficient of the CH₂OO reaction with water dimer is large, about 7 × 10⁻¹² cm³s⁻¹ (22), leading to extremely fast decay rate of CH₂OO under typical tropospheric conditions (Table 1).

Table 1.

Reported bimolecular rate coefficients k and effective first-order rate coefficients (k_eff = k[Coreactant]) for simple CI reactions with H₂O, (H₂O)₂, and SO₂ at a given atmospheric condition

CI	Coreactant	[Coreactant]/cm−3	k/cm3s−1	keff/s−1	Reference
CH2OO	H2O	5.4 × 1017	<1.5 × 10−15	<810	(22)
	(H2O)2	6.0 × 1014	6.5 × 10−12	3,900	(22)
	SO2	1.2 × 1012	3.9 × 10−11	47	(2)
anti-CH3CHOO	H2O	5.4 × 1017	1.0 × 10−14	5,400	(3)
			2.4 × 10−14	13,000	(24)
	SO2	1.2 × 1012	6.7 × 10−11	80	(3)
			2.2 × 10−10	260	(24)
syn-CH3CHOO	H2O	5.4 × 1017	<4 × 10−15	<2,200	(3)
			<2 × 10−16	<110	(24)
	SO2	1.2 × 1012	2.4 × 10−11	29	(3)
			2.9 × 10−11	35	(24)
(CH3)2COO	H2O	5.4 × 1017	<1.5 × 10−16*	<81*	This work
	(H2O)2	6.0 × 1014	<1.3 × 10−13*	<78*	This work
	SO2	1.2 × 1012	1.3 × 10−10	160	This work

The assumed concentrations of H₂O and SO₂ correspond to a relative humidity RH = 70% and a SO₂ mixing ratio of 50 ppb at 298 K and 1 atm. Only data from direct kinetic measurements are selected.

^*The rate constant of the H₂O reaction with (CH₃)₂COO is obtained by assuming the rate constant of the (H₂O)₂ reaction with (CH₃)₂COO is zero and vice versa (SI Appendix, Table S2); thus, these two effective decay rates should not be added together.

Taatjes et al. (3) and Sheps et al. (24) have reported that the anti- form of methyl-substituted CI (CH₃CHOO, R¹ = H in Fig. 1) reacts with water vapor much faster than the syn- form (R¹ = CH₃ in Fig. 1) does. Quantum-chemical investigations (25–27) as well as studies of alkene ozonolysis (20, 28) also indicate that the structure of a CI strongly influences its reactivity toward water vapor. If one type of CI reacts slowly with water vapor but reacts quickly with SO₂, these CIs may accumulate to higher concentrations and have higher probability to oxidize atmospheric SO₂. Table 1 shows selected rate coefficients for relevant CI reactions and the effective first-order decay rate coefficients (k_eff) of small CIs under an atmospheric condition.

As will be discussed in detail in Discussion and Conclusions, the steady-state concentration of a particular CI would depend on its formation rate and effective decay rate coefficient; its impact on the SO₂ oxidation would further depend on its concentration and reaction rate coefficient with SO₂. Experimental results (Table 1) show that CH₂OO and anti-CH₃CHOO react with water vapor very quickly (3, 22–24). Thus, their steady-state concentrations would be too low to have a significant impact in SO₂ oxidation under typical atmospheric conditions, as shown in modeling results (15, 16). On the other hand, previous experimental data for syn-CH₃CHOO (3, 24) are not precise enough to determine its main decay pathways in the atmosphere.

Quantum-chemistry (25–27) calculations predicted that the anti- form of CIs (CIs with R¹ = H in Fig. 1, including CH₂OO) react with water vapor very quickly and that the syn- form of CIs (CIs with R¹ ≠ H in Fig. 1, including dialkyl-substituted CIs) react slowly with water vapor. Here, steric hindrance of the alkyl group may account for the structure dependence in the reactivity. However, due to uncertainty in the calculated rate coefficients, it is unclear about the main decay channels of the syn-CIs in the atmosphere. For example, some theoretical investigation (29) shows that the reactions of water vapor with syn-CIs may still be fast enough (with a large uncertainty) to efficiently scavenge atmospheric syn-CIs, whereas some other calculations (25) suggest that these reactions are too slow to consume syn-CIs significantly. If the latter is the case, syn-CIs may have higher steady-state concentrations in the troposphere and may still play an important role in the SO₂ oxidation; otherwise the steady-state concentrations of syn-CIs would still be low due to their fast consumption by reactions with water vapor (unless their sources are significantly larger than current estimation) and we might need to find another candidate for the unknown oxidant [oxidant X in the work by Mauldin et al. (4)] in the SO₂ atmospheric chemistry.

To shed some light on this important issue, we performed direct kinetic measurements of the reactions of dimethyl-substituted CI, (CH₃)₂COO, with water vapor and with SO₂ under near-atmospheric conditions. By introducing the water reactant at high concentrations, the rate coefficient of (CH₃)₂COO reaction with water can be better constrained. In contrast with the fast reaction of CH₂OO with water dimer, this result shows that the relative probabilities of (CH₃)₂COO reactions with SO₂ and with water vapor are comparable in the troposphere, so water alone would not completely scavenge (CH₃)₂COO, suggesting CIs of similar structures may play a more important role in the atmospheric oxidation of SO₂.

Results

Fig. 2A shows time-resolved difference absorption spectra recorded in the (CH₃)₂CI₂/O₂ pulsed photolysis system, in which the laser pulse defines delay time t = 0. The photodissociation of the (CH₃)₂CI₂ precursor produces (CH₃)₂CI radicals; (CH₃)₂COO was formed through the reaction of (CH₃)₂CI + O₂ → (CH₃)₂COO + I (30). At t = 50 μs, a strong absorption band peaked at 330 nm was observed, which decays with the delay time. At very long delay time (e.g., 2,000 μs), a negative difference absorption signal peaked at 296 nm was observed, which corresponds to the depletion of the (CH₃)₂CI₂ precursor. Fig. 2B shows the spectra recorded at a very similar condition but adding SO₂ gas to scavenge CIs. The absorption signal in Fig. 2B consists mainly of the depletion of (CH₃)₂CI₂ and a small amount of IO [a byproduct, likely from I + (CH₃)₂COO → IO + (CH₃)₂CO]. Other possible species in the absorption cell are SO₃ and acetone, which absorb rather weakly in the studied wavelength region [the absorption cross-sections σ for the relevant species are σ(SO₃) < 1 × 10⁻²¹ cm², σ(acetone) < 5 × 10⁻²⁰ cm², and σ ∼ 10⁻¹⁷ cm² for a CI] (30, 31). Following the established method of spectral analysis in similar systems (22, 31–33), the absorption band of (CH₃)₂COO can be obtained from the difference between the spectra in Fig. 2 A and B and is shown in Fig. 2C. The resultant spectrum of (CH₃)₂COO is slightly broader than but still consistent with the jet-cooled spectrum reported by Liu et al. (30). See SI Appendix, Fig. S2.

Fig. 2.

Typical transient absorption spectra recorded in the (CH₃)₂CI₂/O₂ photolysis system at selected delay times after the photolysis laser pulse (t = 0). Total pressure was 100.1 torr (P_N2 = 89.9 torr, P_O2 = 10.2 torr, [(CH₃)₂CI₂]₀ = 1.1 × 10¹⁴ cm⁻³). (A) Spectra recorded without adding SO₂. (B) Spectra recorded at [SO₂] = 5.3 × 10¹⁴ cm⁻³; (CH₃)₂COO was consumed by its reaction with SO₂. (C) Difference spectra between the two sets of spectra in A and B. The main spectral carrier is assigned to (CH₃)₂COO because the reaction of (CH₃)₂COO with SO₂ is fast. The residue contributions of the IO byproduct and the (CH₃)₂CI₂ precursor are minor and have been removed in C.

In the kinetic measurements, we chose a detection window of 335–345 nm to monitor the change in (CH₃)₂COO concentrations. In this detection window, the absorption signal of IO is negligible [σ(IO) = 2.7 × 10⁻¹⁹ cm² << σ((CH₃)₂COO) ∼ 10⁻¹⁷ cm²] (22); the absorption of the precursor (CH₃)₂CI₂ is small but not negligible. Fortunately, Fig. 2B shows that the depletion of the precursor is a constant after the photolysis laser pulse, which would not affect our kinetic analysis for (CH₃)₂COO.

Representative difference transient absorption traces recorded at various SO₂ concentrations are shown in Fig. 3; a more complete set of the experimental data can be found in SI Appendix, Figs. S3–S6 and Table S1. As mentioned above, (CH₃)₂COO is the main spectral carrier for the absorption at 340 nm. The rapid rise of the signal after the photolysis laser pulse was due to (CH₃)₂COO formation. When SO₂ is added, the reaction of (CH₃)₂COO with SO₂ dominates in the observed decay of (CH₃)₂COO. When no SO₂ is added, the (CH₃)₂COO decay is due mainly to reactions of (CH₃)₂COO with radical species, including I atoms, OH radicals [possibly from decomposition of (CH₃)₂COO (30)], and (CH₃)₂COO itself, similar to the case of CH₂OO (33). Because the concentration of (CH₃)₂COO in our experiment was low (on the order of 10¹¹ cm⁻³), the self-reaction between two (CH₃)₂COO molecules did not dominate in the decay of (CH₃)₂COO, leading to the observed first-order–like decay.

Fig. 3.

Typical temporal profiles of the absorbance change at 340 nm. The change in absorbance was mainly due to (CH₃)₂COO and was monitored in real time with a balanced photodiode detector at various SO₂ concentrations. The smooth lines are single exponential fit to the data.

The decay of the (CH₃)₂COO signals can be well described by Eqs. 1 and 2. Under the high O₂ pressure (∼10 torr) used in this study, the formation time of (CH₃)₂COO is within 1 μs. This formation time is relatively short in comparison with the decay time of (CH₃)₂COO. Thus, the fitting for the decay typically started at t = 10 μs to decouple the formation kinetics.

$$\left[{\left({\text{CH}}_3\right)}_2\text{COO}\right]={\left[{\left({\text{CH}}_3\right)}_2\text{COO}\right]}_0{e}^{-k^{\prime }t},$$ [1]

$$-\frac{d\left[{\left({\text{CH}}_3\right)}_2\text{COO}\right]}{dt}=k^{\prime }\left[{\left({\text{CH}}_3\right)}_2\text{COO}\right]=(k_0+k_{\text{SO}2}\left[{\text{SO}}_2\right])\left[{\left({\text{CH}}_3\right)}_2\text{COO}\right].$$ [2]

Fig. 4 shows the observed pseudo–first-order rate coefficient (k′− k₀) as a function of SO₂ concentration, where the k₀ term (Eq. 2) accounts for the (CH₃)₂COO decay when no SO₂ was added (see SI Appendix, Table S1 for the values of k₀). A linear relationship between k′ and [SO₂] is found with the slope corresponding to the rate coefficient k_SO2 for the (CH₃)₂COO reaction with SO₂. In Fig. 4, it is obvious that the slope is different at different total pressure.

Fig. 4.

Pseudo–first-order rate coefficient of (CH₃)₂COO reaction with SO₂ plotted as a function of [SO₂]. k₀ is the rate coefficient at zero [SO₂]. The measurements were performed at 298 K with N₂ as the buffer gas at the indicated total pressure. The error bars indicate the SDs; the number of independent data can be found in SI Appendix, Table S1 for each experiment. Solid lines are linear fit to the data at the corresponding pressure. Data at higher and lower pressure ranges are shown in SI Appendix, Figs. S7 and S8.

Fig. 5 plots the reaction rate coefficient k_SO2 as a function of the total pressure with N₂ as the buffer gas. At pressure higher than 100 torr ([M] = 3.2 × 10¹⁸ cm⁻³), the rate coefficient is leveled. This high-pressure-limit rate coefficient is 1.3 × 10⁻¹⁰ cm³s⁻¹, about 3 times the reported rate coefficient of CH₂OO reaction with SO₂ [k_CH2OO+SO2 = (3.9 ± 0.7) × 10⁻¹¹ cm³s⁻¹] (2). At low pressure, the rate coefficient drops. A simple model in Fig. 6 is used to analyze the experimental data.

Fig. 5.

Observed second-order rate coefficient of (CH₃)₂COO reaction with SO₂ plotted as a function of the total pressure and number density. The fitting is plotted as solid lines. The error bars of the data indicate the SDs. The errors in the fitted equation are the SDs resulting from the fitting only, not including any possible systematic uncertainty yet. There are 18 independent measurements.

Fig. 6.

Proposed mechanism for CI reaction with SO₂.

Quantum-chemistry studies (29) suggested that CI reaction with SO₂ would first go through a barrierless formation of a cyclic intermediate, INT* (R1). In addition, because the formation of the cyclic intermediate is quite exothermic, its stabilization requires collision with a third body (R2). Furthermore, the energized intermediate INT* may dissociate back to the reactants or form products directly (likely SO₃ + carbonyl); (R4) connects the stabilized intermediate INT to the final products.

Applying steady-state approximation to INT* leads to the following:

$$-\frac{d\left[\text{CI}\right]}{dt}=k_1\left[\text{CI}\right]\left[{\text{SO}}_2\right]-k_{-1}\left[{\text{INT}}^{\ast }\right]=k_{\text{SO}2}\left[\text{CI}\right]\left[{\text{SO}}_2\right],$$ [3]

$$k_{\text{SO}2}=\frac{\hspace{0.17em}\hspace{0.17em}\hspace{0.17em}\hspace{0.17em}\hspace{0.17em}\hspace{0.17em}\hspace{0.17em}\hspace{0.17em}{k}_1(k_3+k_2\left[M\right])\hspace{0.17em}\hspace{0.17em}\hspace{0.17em}\hspace{0.17em}\hspace{0.17em}\hspace{0.17em}\hspace{0.17em}\hspace{0.17em}}{\hspace{0.17em}{k}_{-1}+k_3+k_2\left[M\right]\hspace{0.17em}}.$$ [4]

Theoretical analysis (29) suggests that (R2) is the main pathway (97%) for the (CH₃)₂COO reaction with SO₂ under atmospheric pressure. Thus, we may simplify the mechanism by assuming k₂[M] >> k₃, unless the pressure is low. (R4) may be a slow reaction, but if there are no other competing processes, the yield of R4 may still be high, as indicated by previous ozonolysis studies (20, 28, 34).

The best fit of Eq. 4 to the data of Fig. 5 is shown as the red line (the best fit gives k₃ =0). At the high-pressure limit, k_SO2 = k₁ = 1.32 × 10⁻¹⁰ cm³s⁻¹. Nonzero k₃ leads to a nonzero low-pressure-limit rate constant k₁k₃/(k_-1+k₃) but a value larger than 6 × 10⁻¹¹ cm³s⁻¹ for this term cannot fit the data satisfactorily (green line in Fig. 5). The error in k₁ mostly comes from the uncertainty in the absolute concentrations of SO₂, which is less than 10%. Thus, we report k₁ = (1.32 ± 0.13)×10⁻¹⁰ cm³s⁻¹.

To further elucidate the role of the buffer gas, we changed the buffer gas from N₂ to CO₂ or Ne and performed similar experiments at 50 torr. SI Appendix, Fig. S9 shows that CO₂ is a more efficient collider than N₂ and Ne is a less efficient collider, as one may expect. Furthermore, we also investigated the pressure effect in the reaction of CH₂OO with SO₂. However, the reaction rate coefficient of CH₂OO with SO₂ does not exhibit significant pressure dependence (see SI Appendix, Figs. S13–S15 and Table S3 and ref. 10).

The pressure dependence of the reaction of (CH₃)₂COO with SO₂ has implications in atmospheric chemistry. Applying low-pressure rates in a model may underestimate the impact of the (CH₃)₂COO reaction with SO₂. Similarly, the pressure dependence of the reactions of SO₂ with other CIs needs to be investigated because quite a number of rate coefficients of CI reactions are measured under low-pressure conditions [∼4 torr for PIMS experiments (2, 3, 14)].

The kinetics of the (CH₃)₂COO reaction with water vapor were also measured. In sharp contrast with the fast reaction of CH₂OO with water vapor, the decay of (CH₃)₂COO barely depends on [H₂O], even for high water concentration up to 7 × 10¹⁷ cm⁻³ (RH = 90% at 298 K). Fig. 7 shows the first-order rate coefficients k′ of (CH₃)₂COO decay as a function of [H₂O]. We have varied the experimental conditions including the total pressure, precursor concentration, photolysis laser power, etc. (SI Appendix, Table S2) and the results show that the effect of water in the decay of (CH₃)₂COO is extremely weak. More detailed analysis indicates the decay rate coefficient of (CH₃)₂COO depends mainly on the radical concentration (SI Appendix, Fig. S12). We estimate an upper limit of 1.5 × 10⁻¹⁶ cm³s⁻¹ for the rate coefficient of the (CH₃)₂COO reaction with H₂O, based on the fluctuation of the experimental data (SI Appendix, Table S2).

Fig. 7.

Pseudo–first-order rate coefficient of (CH₃)₂COO as a function of [H₂O]. Solid lines are linear fit to the data of each experiment. There are 223 data points (see SI Appendix, Table S2 for the experimental conditions).

Discussion and Conclusions

So far, there is no method available to detect the concentration of any CI in the atmosphere. The estimation of the CI concentration can only be done by knowing its formation rate and consumption rate. The impact of a given CI (e.g., to the SO₂ oxidation) depends on its concentration and the rate coefficient of its reaction with SO₂, as shown in the following (Eqs. 5–9):

$${\text{O}}_3+\text{alkene}\to \text{CI}\hspace{1em}{k}_{\text{form}},$$ [5]

$$\text{CI}\to \text{Products}\hspace{1em}{k}_{\text{decay}},$$ [6]

$${\left[\text{CI}\right]}_{\text{ss}}=\frac{k_{\text{form}}{[\text{O}}_3\left]\right[\text{alkene}]}{k_{\text{decay}}},$$ [7]

$$k_{\text{decay}}=k_{\text{therm}}+k_{\text{H}2\text{O}}\left[{\text{H}}_2\text{O}\right]+k_{\text{w}2}\left[{\left({\text{H}}_2\text{O}\right)}_2\right]+k_{\text{SO}2}\left[{\text{SO}}_2\right]+k_{\text{NO}2}\left[{\text{NO}}_2\right]+\mathrm{\dots },$$ [8]

$$\frac{d\left[{\text{SO}}_2\right]}{dt}=-k_{\text{SO}2}{\left[\text{CI}\right]}_{\text{ss}}\left[{\text{SO}}_2\right]=-k_{\text{SO}2}\frac{k_{\text{form}}{[\text{O}}_3\left]\right[\text{alkene}]}{k_{\text{decay}}}\left[{\text{SO}}_2\right],$$ [9]

where [CI]_ss is the steady-state concentration of the CI; k_form is the formation rate constant of the CI from its precursors (e.g., ozone and an alkene); k_decay would be the summation of the effective decay rate constants k_eff of all possible decay pathways; k_therm is the thermal decomposition rate coefficient; k_H2O, k_w2, k_SO2, and k_NO2 are the bimolecular rate coefficients of CI reactions with H₂O, (H₂O)₂, SO₂, and NO₂, respectively. The k_therm for (CH₃)₂COO has been reported recently to be 3 ± 0.4 s⁻¹ at 293 K (34) with a significant temperature dependence (28).

Considering the possible concentrations of H₂O and SO₂ in the troposphere, we can evaluate the effective decay rate constants of simple CIs by water reaction and by SO₂ reaction. As shown in Table 1, the consumptions of CH₂OO and anti-CH₃CHOO by water (including monomer and dimer) are extremely fast, >10³ s⁻¹, predominating in their k_decay. The very large k_decay would result in very low steady-state concentrations for these CIs, limiting their roles in oxidizing other atmospheric species like SO₂ and NO₂ (see Eqs. 7–9).

On the other hand, the reaction of (CH₃)₂COO with H₂O is much slower, such that the water reaction would not limit the (CH₃)₂COO concentration in the troposphere anymore. In addition, the rate coefficient of (CH₃)₂COO reaction with SO₂ is larger than those of other known CI reactions with SO₂, indicating a greater role of (CH₃)₂COO in the atmospheric SO₂ oxidation. In other words, there is a strong structure dependence in the CI oxidation of SO₂. For (CH₃)₂COO, its k_decay is smaller and its k_SO2 is larger, such that the SO₂ oxidation rate by (CH₃)₂COO is faster in comparison with those for CH₂OO and anti-CH₃CHOO, assuming similar formation rates (see Eq. 9). In addition, for a CI with small k_H2O and k_w2, its k_therm would be another important factor that also influences its [CI]_ss and oxidation capacity. A related issue is that the thermal decomposition of syn-CIs [including (CH₃)₂COO] may form OH radicals through a 1,4 H-migration process (30), which may be an important nonphotolytic OH source in the troposphere. Again, structure dependence of k_therm for various CIs would need further investigation.

Previous studies of ozonolysis of 2,3-dimethyl-2-butene (tetramethyl ethylene, R¹ = R² = R³ = R⁴ = CH₃) (20, 28, 34) have provided some information about the (CH₃)₂COO reactions with H₂O and with SO₂. But, the results are far from consistency. First, the reported k_SO2 are on the order of 10⁻¹³ cm³s⁻¹ (34), much smaller than the value determined from the direct kinetic measurement (Table 1). A similar situation also happens for smaller CIs like CH₂OO and CH₃CHOO (12, 13, 35), suggesting there might be some systematic issues in the ozonolysis experiments. Very recently, Newland et al. (20) measured the removal of SO₂ in the presence of alkene–ozone systems and concluded that the SO₂ removal displays a clear dependence on relative humidity for all four alkene ozonolysis systems [ethene (to form CH₂OO), cis-2-butene (to form CH₃CHOO), trans-2-butene (to form CH₃CHOO), and 2,3-dimethyl-2-butene (to form (CH₃)₂COO)], confirming a significant reaction for stabilized CIs with H₂O. However, Berndt et al. (28) investigated the H₂SO₄ formation as a function of [H₂O] in similar ozonolysis reactions and reported quite different values. The results for (CH₃)₂COO are summarized in Table 2. The result by Newland et al. is not consistent with the results of Berndt et al. and this work. The reason for this discrepancy may originate from the complexity of the ozonolysis experiments, in which it is difficult to fully quantify the side reactions during the long reaction time.

Table 2.

Ratio of the rate coefficients of (CH₃)₂COO reactions with H₂O and with SO₂, k_H2O/k_SO2

kH2O/kSO2	Method	Reference
<1.2 × 10−6	Absolute direct kinetic measurements of (CH3)2COO	This work
<4 × 10−6	Relative measurements of the H2SO4 product	Berndt et al. (28)
(8.7 ± 2.5) × 10−5	Relative measurements of the SO2 and O3 consumptions	Newland et al. (20)

It is important to note that the reactivity of a particular CI toward water vapor strongly depends on its structure. Table 1 indicates that a methyl group substitution for R¹ may alter the rate coefficient by orders of magnitude. Although ozonolysis experiments (20, 28, 34) have also shown a trend for the structure dependence, the magnitude of the reactivity difference is much smaller than what is observed in the direct kinetic measurements (Table 1). In addition, the ozonolysis experiments cannot distinguish the anti- and syn- conformers of CIs, which have quite different reactivity.

The results summarized in Table 1 give direct evidence that the reaction of water with (CH₃)₂COO is greatly hindered by the methyl group at the R¹ position. Interestingly, methyl substitution seems to enhance the reactivity of CIs toward SO₂. One can imagine that CIs with more complicated substitution groups may also react with water slowly but react with SO₂ quickly, similar to (CH₃)₂COO. Such CIs may be the candidate for the oxidant X in the SO₂ oxidation, an important issue raised recently (4).

Considering that much of the VOC emissions consist of a large variety of alkenes, ranging from simple alkenes like C₂H₄ and C₃H₆ to bigger alkenes like isoprene, monoterpenes, sesquiterpenes, etc., various CIs are expected to form in the atmospheric ozonolysis reactions. To assess the impact of the CI+SO₂ reaction class on the atmosphere, it is critical to know the atmospheric concentration of CIs. As mentioned above, it requires the relevant rate coefficients to estimate the atmospheric concentration of a CI. Water is the third most abundant molecule in the air. The reactions of CIs with water are crucial in determining the concentrations and fate of the CIs. The slow rates of water reactions with (CH₃)₂COO or with similar CIs are very difficult to measure, but very important in estimating the concentrations of those CIs and thus the oxidizing capacity of the atmosphere.

Materials and Methods

The experimental setup has been described in detail elsewhere (22, 32). In brief, (CH₃)₂COO was generated from photolysis of a gaseous mixture consisting of (CH₃)₂CI₂, O₂, and buffer gas (N₂) at 248 nm (KrF excimer laser) via the established preparation method: (CH₃)₂CI₂ + hν → (CH₃)₂CI + I; (CH₃)₂CI +O₂ → (CH₃)₂COO + I (2, 30). (CH₃)₂COO was monitored via its strong UV absorption (30). Continuous probe light went through the photolysis reactor (25 or 20 mm inner diameter, 76 cm long) six or eight times to enhance the absorption signal.

Spectral Measurements.

Transient absorption spectra of the reaction system were measured with a continuous broadband light source (Energetiq, EQ-99) and a time-gated iCCD spectrometer (Andor, SR303i and DH320T-18F-03). The light source was projected to the entrance of the absorption cell by an achromatic lens (Thorlabs ACA254-100-UV). To enhance the absorption signal, the probe light was reflected eight times through the photolysis reactor by a spherical mirror (R = 1 m, Thorlabs, CM750-500-F01) and a SiO₂ prism. The probe beam and the photolysis beam were overlapped collinearly in the photolysis reactor. For the iCCD measurement, the reference spectrum was recorded 200 μs before the photolysis pulse. Transient spectra at delay times 50, 100, 150, 200, 250, 300, 500, 1,000, and 2,000 μs were recorded.

Kinetics Measurements.

Absorption signal at 340 nm was measured in real time by using a balanced photodiode detector (Thorlabs, PDB450A) and a bandpass filter (Edmund Optics, 65129, 10-nm OD4 band pass filter) and the same light source. A time-dependent transmittance change (<1%) was observed after the photolysis pulse even without adding any sample. This background did not depend on [H₂O], or on [SO₂]. It can be subtracted by performing a background run under the same experimental condition except adding (CH₃)₂CI₂. The presented data are after the background subtraction.

[SO₂] Measurements.

The SO₂ concentration was adjusted by controlling the amount of SO₂ in buffer gas with mass flow controllers (Brooks, 5850E or 5800E) and monitored by its UV absorption (190–330 nm) with a D₂ lamp (Ocean Optics, D-2000) and a spectrometer (Ocean Optics, Maya2000 Pro).

Relative Humidity Measurements.

The relative humidity was adjusted by controlling the mixing ratio of dry and moisturized buffer gases with mass flow controllers (Brooks, 5850E or 5800E) and monitored with a humidity sensor (Rotronic, HC2-S). The temperature for measuring the water reaction was controlled at 298.2 ± 0.5 K.

Precursor Preparation.

The precursor, (CH₃)₂CI₂, was synthesized following a reported method (36). In brief, acetone was added to hydrazine monohydrate (80 °C, >1 h) to obtain acetone hydrazone (CH₃)₂C=NNH₂. Saturated solution of iodine in ethyl ether was added to acetone hydrazone (at room temperature), which is mixed with ethyl ether and triethylamine, to obtain the final product. The structure of the synthesized (CH₃)₂CI₂ was checked with H-NMR spectroscopy [(CH₃)₂CI₂: 3.00 ppm (6H, s, Me) in CDCl₃] (37).