Kinetics of stabilised Criegee intermediates derived from alkene ozonolysis: reactions with SO₂, H₂O and decomposition under boundary layer conditions

Abstract

The removal of SO₂ in the presence of alkene–ozone systems has been studied for ethene, cis-but-2-ene, trans-but-2-ene and 2,3-dimethyl-but-2-ene, as a function of humidity, under atmospheric boundary layer conditions. The SO₂ removal displays a clear dependence on relative humidity for all four alkene–ozone systems confirming a significant reaction for stabilised Criegee intermediates (SCI) with H₂O. The observed SO₂ removal kinetics are consistent with relative rate constants, k(SCI + H₂O)/k(SCI + SO₂), of 3.3 (±1.1) × 10⁻⁵ for CH₂OO, 26 (±10) × 10⁻⁵ for CH₃CHOO derived from cis-but-2-ene, 33 (±10) × 10⁻⁵ for CH₃CHOO derived from trans-but-2-ene, and 8.7 (±2.5) × 10⁻⁵ for (CH₃)₂COO derived from 2,3-dimethyl-but-2-ene. The relative rate constants for k(SCI decomposition)/k(SCI + SO₂) are −2.3 (±3.5) × 10¹¹ cm⁻³ for CH₂OO, 13 (±43) × 10¹¹ cm⁻³ for CH₃CHOO derived from cis-but-2-ene, −14 (±31) × 10¹¹ cm⁻³ for CH₃CHOO derived from trans-but-2-ene and 63 (±14) × 10¹¹ cm⁻³ for (CH₃)₂COO. Uncertainties are ±2σ and represent combined systematic and precision components. These values are derived following the approximation that a single SCI is present for each system; a more comprehensive interpretation, explicitly considering the differing reactivity for syn- and anti-SCI conformers, is also presented. This yields values of 3.5 (±3.1) × 10⁻⁴ for k(SCI + H₂O)/k(SCI + SO₂) of anti-CH₃CHOO and 1.2 (±1.1) × 10¹³ for k(SCI decomposition)/k(SCI + SO₂) of syn-CH₃CHOO. The reaction of the water dimer with CH₂OO is also considered, with a derived value for k(CH₂OO + (H₂O)₂)/k(CH₂OO + SO₂) of 1.4 (±1.8) × 10⁻². The observed SO₂ removal rate constants, which technically represent upper limits, are consistent with decomposition being a significant, structure dependent, sink in the atmosphere for syn-SCI.

---

1. Introduction

Atmospheric oxidation processes are central to understanding trace gas atmospheric composition, the abundance of air pollutants harmful to human health, crops and ecosystems, and the removal of reactive greenhouse gases such as methane. The principal atmospheric oxidants responsible for initiating the gas-phase degradation of volatile organic compounds (VOCs), NO_x and SO₂ are OH, NO₃ and O₃, with additional contributions from other species such as halogen atoms. Recent field measurements¹ in a boreal forest have identified the presence of an additional oxidant species, removing SO₂ and producing H₂SO₄. The additional SO₂ oxidation observed was substantial (comparable to that due to OH radicals alone), and attributed to a product of alkene ozonolysis – the boreal forest environment being one in which substantial biogenic alkene emissions occur – suggested to be the stabilised Criegee intermediate (SCI). The gas-phase oxidation of SO₂ in the atmosphere is of interest to climate research as it leads to formation of H₂SO₄, contributing to new particle formation and sulphate aerosol loading, in competition with condensed phase oxidation.² A missing mechanism for the conversion of SO₂ to H₂SO₄ could lead to model underestimation of the sulphate aerosol burden and affect radiative forcing calculations,³ with corresponding implications for climate predictions. Enhanced SO₂ oxidation in alkene–ozone systems was first reported by Cox & Penkett⁴ over forty years ago, however the precise reaction mechanism giving rise to this effect remains elusive. The “Criegee” ozonolysis reaction mechanism was first postulated in the 1940s.⁵ It is now accepted that the ozone molecule adds to the C=C double bond via a concerted cycloaddition to form a primary ozonide, followed by cleavage of the C–C bond and one of the O–O bonds forming a carbonyl molecule and a carbonyl oxide, or ‘Criegee intermediate’ (CI).⁶ Ozonolysis derived SCIs are formed with a broad internal energy distribution, to yield chemically activated and stabilised SCIs. SCIs can have sufficiently long lifetimes to undergo bimolecular reactions with H₂O and SO₂ amongst other species. Chemically activated SCIs may also undergo collisional stabilisation, unimolecular decomposition or isomerisation (Scheme 1). For substituted alkenes, SCIs can undergo a 1,4-H-shift rearrangement through a vinyl-hydroperoxide (VHP) via the so-called “hydroperoxide channel” and decompose to yield OH and a vinoxy radical – a substantial non-photolytic source of atmospheric oxidants.^7,8 This is the favoured channel for SCIs formed in the syn-configuration. Time resolved studies⁹ show that the VHP may persist for appreciable timescales under boundary layer conditions, giving rise to the observed pressure dependence of OH radical yields,¹⁰ and opening the possibility for bimolecular reactions of this species to occur. SCIs formed in the anti-configuration are thought to primarily undergo rearrangement and decomposition via a dioxirane intermediate (“the acid/ester channel”), producing a range of daughter products and contributing to the observed overall HO_x radical yield.^6,11

Scheme 1 Simplified mechanism for the reaction of Criegee Intermediates (SCIs) formed from alkene ozonolysis.

Until recently, it has been thought that the predominant atmospheric fate for SCIs was reaction with water vapour^12,13 – leading to a significant source of organic acids and hydroperoxides, suggesting that bimolecular reaction with SCIs is an unimportant oxidation mechanism for trace gas species. This view was recently challenged by direct observation¹⁴ and kinetic studies^15–17 of the CH₂OO and CH₃CHOO SCIs.

Taatjes and co-workers,¹⁵ directly observing CI kinetics for the first time, found reactions of CH₂OO with SO₂ and NO₂ to be substantially faster than previously thought, pointing to a potentially important role for this species in atmospheric SO₂ and NO₂ oxidation; subsequent measurements have identified SO₃ (ref. 16) and NO₃ (ref. 18) as products of these reactions. The key to whether SCIs are indeed significant contributors to gas-phase atmospheric SO₂ oxidation is the ratio of the rate constants for reaction of the SCI with SO₂ (k₂), to that with H₂O (k₃) and decomposition (k_d). In laboratory studies where SCIs were produced by the 248 nm laser photolysis of alkyl iodide precursors at 4 Torr total pressure, with the SCI decay monitored by VUV photoionisation in the presence of excess co-reactants, this ratio has recently been reported to be 10³–10⁴ for the smallest two SCIs (CH₂OO^15,17 and CH₃CHOO¹⁶), with k₂ on the order of 10⁻¹¹ cm³ s⁻¹ and k₃ on the order of 10⁻¹⁵ cm³ s⁻¹. In contrast, alternative studies of SO₂ oxidation in alkene–ozone systems, performed at atmospheric pressure through detection of the H₂SO₄ product, find much smaller SCI + SO₂ rate coefficients (by ca. two orders of magnitude).¹⁹ Explanations for this apparent discrepancy may include the lifetime of the secondary ozonide (SOZ) adduct formed from the SCI + SO₂ reaction, collisionally stabilised at atmospheric pressure,²⁰ effects of the presence of multiple SCI conformers with differing reactivity, or contributions from other oxidant species, formed within the ozonolysis system, to SO₂ reaction. Understanding the behaviour of the ozone–alkene–SO₂ system in the presence of water vapour is critical to quantifying the impact of SCI chemistry upon atmospheric SO₂ oxidation.²¹

An additional, potentially important, fate of SCI under atmospheric conditions is unimolecular decomposition (denoted k_d in (R4)). For CH₂OO, rearrangement via a ‘hot’ acid species represents the lowest accessible decomposition channel, but due to lack of alkyl substituents, the theoretically predicted 298 K rate constant is rather low, 0.3 s⁻¹.²² Previous studies have identified the hydroperoxide rearrangement as dominant for SCIs with a syn configuration, determining their overall unimolecular decomposition rate.^7,8 For syn-CH₃CHOO recent experimental²³ work yielded a decomposition rate of 3–30 s⁻¹. Theoretical work²⁴ has predicted a decomposition rate coefficient of 24.2 s⁻¹ for syn-CH₃CHOO and 67.2 s⁻¹ for anti-CH₃CHOO (for which only the ester channel exists), owing to the potential energy release from the higher energy anti-conformer.²³ An upper limit to total CH₃CHOO loss (decomposition and heterogeneous wall losses combined) of <250 s⁻¹ has been reported by Taatjes et al.¹⁶ Earlier experimental work reported decomposition rate constants of ≤20 s⁻¹ for CH₃CHOO derived from cis-but-2-ene ozonolysis,²⁵ and 76 s⁻¹ (accurate to within a factor of three) for CH₃CHOO derived from trans-but-2-ene ozonolysis.²⁶ For (CH₃)₂COO (derived from 2,3-dimethyl-but-2-ene ozonolysis) a total loss rate of 3.0 s⁻¹ has recently been determined experimentally.¹⁹ This value (which represents an upper limit for k_d) is somewhat smaller than but comparable in magnitude to an earlier measurement of 6.4 s⁻¹ (determined at 100 Torr).²⁷ Theoretical estimates of (CH₃)₂COO decomposition rates are higher, at up to 250 s⁻¹.²² Photolysis loss rates have also recently been reported for CH₂OO²⁸ and CH₃CHOO²⁹ of 1 s⁻¹ and 0.2 s⁻¹ respectively, calculated for actinic flux values at midday, SZA = 0°.

Here, we present results of a series of experiments in which the oxidation of SO₂ during the ozonolysis of ethene, cis-but-2-ene, trans-but-2-ene and 2,3-dimethyl-but-2-ene (tetramethylethylene, TME), was investigated in the presence of varying amounts of water in the European Photochemical Reactor facility (EUPHORE), Valencia, Spain.

2. Experimental

2.1 EUPHORE

EUPHORE is a 200 m³ simulation chamber used for studying reaction mechanisms under atmospheric boundary layer conditions. In general, experiments comprised time-resolved measurement of the removal of SO₂ in the presence of an alkene–ozone system, as a function of humidity. SO₂ and O₃ abundance were measured using conventional fluorescence and UV absorption monitors, respectively; alkene abundance was determined via FTIR spectroscopy. The precision of the SO₂ and O₃ monitors were 0.25 and 0.47 ppbv respectively (evaluated as 2 standard deviations of the measured value prior to SO₂ or O₃ addition). The chamber is fitted with large horizontal and vertical fans to ensure rapid mixing (three minutes). Further details of the chamber setup and instrumentation are available elsewhere.^30,31 Experiments were performed in the dark (i.e. with the chamber housing closed; j(NO₂) ≤ 10⁻⁶ s⁻¹), at atmospheric pressure (ca. 1000 mbar) and temperatures between 296 and 303 K, on timescales of ca. 20–30 minutes. Chamber dilution was monitored via the first order decay of an aliquot of SF₆, added prior to each experiment. Cyclohexane (ca. 75 ppmv) was added at the beginning of each experiment to act as an OH scavenger, such that SO₂ reaction with OH was calculated to be ≤1% of the total chemical SO₂ removal in all experiments.

2.2 Experimental approach

Experimental procedure, starting with the chamber filled with clean air, comprised addition of SF₆ and cyclohexane, followed by water vapour, O₃ (ca. 500 ppbv) and SO₂ (ca. 50 ppbv). A gap of five minutes was left prior to addition of the alkene, to allow complete mixing. The reaction was then initiated by addition of the alkene (ca. 500 ppbv for ethene, 200 ppbv for cis- and trans-but-2-ene and 400 ppbv for TME). The chamber was monitored for an hour subsequent to the addition of ethene and forty five minutes for cis- and trans-but-2-ene and TME. The rate of alkene–ozone consumption is dependent on k₁. Roughly 25% of the ethene was consumed after an hour, while for cis- and trans-but-2-ene and TME 90% of the alkene was consumed within roughly 25 minutes, 20 minutes and 6 minutes respectively. Each experiment was performed at a constant humidity, which was increased in a step-wise manner for consecutive runs to cover the range 1.5–21% RH. Measured increases in [SO₂] agreed with measured volumetric addition across the SO₂ and humidity range used in the experiments.

2.3 Analysis approach

The following sections describe (i) a common analysis applied to all systems, but which features several approximations, and (ii) more detailed consideration of each chemical system in turn, to address potential contributions from the water dimer, and multiple criegee species within each system (e.g. contrasting reactivity of different SCI conformers).

From the chemistry presented in reactions (R1)–(R4) it is assumed that SCI will be produced in the chamber from the reaction of the alkene with ozone at a given yield, ϕ. The SCI produced can then react with SO₂, with H₂O, with other species or decompose under the experimental conditions applied. The rate at which SO₂ is lost, compared with the total production of SCI, is determined by the fraction, f, of the total SCI produced which reacts with SO₂, compared to the sum of the total loss processes of the SCI (eqn (E1)):

Here, L is the sum of any other pseudo-first order chemical loss processes for SCI in the chamber, after correction for dilution. (E2) neglects other (non-alkene) chemical sinks for O₃, such as reaction with HO₂ – also produced directly during alkene ozonolysis,¹¹ but indicated through model calculations to account for <2% of ozone loss under all the experimental conditions of this work. Eqn (E1) and (E2) treat the SCIs formed as a single species – this is the case for (e.g.) ethene and TME, but is an approximation for the 2-butenes; considered further below.

2.3.1 SCI yield calculation. Values for ϕ_SCI were determined for each ozonolysis reaction from experiments performed under dry conditions (RH < 1%) in the presence of excess SO₂ (ca. 1000 ppbv), such that SO₂ scavenges the overwhelming majority of the SCI. From eqn (E2), regressing dSO₂ against dO₃ (corrected for chamber dilution), assuming f to be unity (i.e. all the SCI produced reacts with SO₂) determines the value of ϕ_min, a lower limit to the SCI yield. Fig. 2 shows the experimental data, from which ϕ_min was derived, for all four alkene ozonolysis systems studied.

The lower limit criterion applies as in reality f will be less than one, at experimentally accessible SO₂ levels, as a small fraction of the SCI will still react with any H₂O present, or undergo decomposition. The actual yield, ϕ, was determined by combining the results from the high-SO₂ experiments with those from the series of experiments performed at lower SO₂, as a function of [H₂O], to determine k₃/k₂ and k_d/k₂ (see Section 2.3.2), through an iterative process to determine the single unique value of ϕ_SCI which fits both datasets. It is important to note that the SCI yield is to an extent an operationally defined quantity – for example, OH formation from alkene ozonolysis is known to proceed over at least hundreds of milliseconds⁹ following the alkene–ozone reaction, and so the corresponding CI population must also be evolving with time. In this work, SCI yields reflect the amount of SCI available to oxidise SO₂ on timescales of seconds to minutes.

2.3.2 k(SCI + H₂O)/k(SCI + SO₂) and k_d/k(SCI + SO₂). To determine k₃/k₂ and k_d/k₂, a series of experiments were performed for each alkene, in which the SO₂ loss was monitored as a function of [H₂O]. From eqn (E2), regression of the loss of ozone (dO₃) against the loss of SO₂ (dSO₂) (both corrected for dilution, measured through the removal SF₆, added at the start of each experiment and monitored via FTIR) for an experiment determines the factor f·ϕ at a given point in time. This quantity will vary through the experiment as SO₂ is consumed, and other potential SCI co-reactants are produced, as predicted by eqn (E1). A smoothed fit was applied to the experimental data for the cumulative consumption of SO₂ and O₃, ΔSO₂ and ΔO₃, (Fig. 1) to determine dSO₂/dO₃ (and hence f·ϕ) at the start of each experiment, for use in eqn (E3). The start of the experiment (i.e. when [SO₂] ∼ 50 ppbv) was used as this corresponds to the greatest rate of production of the SCI, and hence largest experimental signals (O₃ and SO₂ rate of change) and is the point at which the SCI + SO₂ reaction has the greatest magnitude compared with any other potential chemical loss processes for either species (see discussion below).

Fig. 1 Cumulative consumption of SO₂ and O₃, ΔSO₂versus ΔO₃, for the ozonolysis of four alkenes in the presence of SO₂ at a range of relative humidities from 1.5–21%. Open symbols are experimental data, corrected for chamber dilution. Solid lines are smoothed fits to the experimental data.

Fig. 2 ΔSO₂vs. ΔO₃ during the excess SO₂ experiments, to determine the minimum SCI yield for the four alkenes.

The value [SO₂]((1/f) − 1) can then be regressed against [H₂O] for each experiment to give a plot with a gradient of k₃/k₂ and an intercept of (k_d + L)/k₂ (eqn (E3)). Our data cannot determine absolute rate constants (i.e. values of k₂, k₃, k_d) in isolation, but is limited to assessing their relative values, which may be placed on an absolute basis through use of an (external) reference value.

Eqn (E1)–(E3) as presented above assume that only a single SCI species is present in each ozonolysis system. While this is the case for the ethene and TME systems, for the but-2-ene systems this is an approximation as two conformers of the CH₃CHOO SCI (syn and anti) are produced. Further analysis is performed in Section 3.3.2 in which the SO₂ loss in the but-2-ene systems is treated as having two components, related to the different SCI.

3. Results and discussion

3.1 Introduction

Table 1 shows the resulting SCI yields obtained for the ozonolysis of ethene, cis-but-2-ene, trans-but-2-ene and tetramethylethylene (TME); uncertainties are ±2σ, and represent the combined systematic (estimated measurement uncertainty) and precision components. Also shown are past literature values, obtained under various conditions using a range of different SCI scavengers.

Table 1 SCI yields derived in this work and reported in the literature. Uncertainty ranges (±2σ, parentheses) indicate combined precision and systematic measurement error components for this work, and are given as stated for literature studies. All referenced studies were conducted between 700 and 760 Torr. C2B – derived from cis-but-2-ene; T2B – derived from trans-but-2-ene

SCI	φ SCI	Ref.
a http://mcm.leeds.ac.uk/MCM/. b At 700–760 Torr (low pressure limit of 0.15 at 0 Torr).
CH2OO	0.37 (±0.04)	This work
	0.37	MCMv3.2^a (IUPAC)^33
	0.35 (±0.05)	Niki et al.^34
	0.39 (±0.053)	Hatakeyama et al.^35
	0.47 (±0.05)	Horie and Moortgat^36
	0.50 (±0.04)	Neeb et al.^37
	0.39 (±0.11)	Hasson et al.^38
	0.54 (±0.15)	Alam et al.^31
CH3CHOO	0.38 (±0.05)	This work
(C2B)	0.19	Rickard et al.^39
	0.18	Niki et al.^40
	0.43	Cox and Penkett^41
CH3CHOO	0.28 (±0.03)	This work
(T2B)	0.49 (±0.22)	Berndt et al.^32
	0.53 (±0.24)	Berndt et al.^19
	0.45	Cox and Penkett^41
	0.19 (±0.03)	Hatakeyama et al.^35
	0.42	Horie and Moortgat^36
	0.24 (±0.07)	Hasson et al.^38
	0.13	Rickard et al.^39
(CH3)2COO	0.32 (±0.02)	This work
	0.45 (±0.20)	Berndt et al.^32
	0.62 (±0.28)	Berndt et al.^19
	ca. 0.65^b	Drozd et al.^10
	0.10 (±0.03)	Hasson et al.^38
	0.30	Niki et al.^40
	0.11	Rickard et al.^39

---

The yield of CH₂OO from ethene ozonolysis obtained in this work is 0.37 (±0.04). This yield has been investigated in many previous studies, with values ranging from 0.34–0.50 determined – a more detailed review is available elsewhere.³¹ The yield obtained in this work is at the lower end but within the envelope of these estimates, and is in excellent agreement with the current IUPAC recommendation of 0.37.

The yield of CH₃CHOO from cis-but-2-ene ozonolysis obtained in this work is 0.38 (±0.05), with that from trans-but-2-ene being 0.28 (±0.03). These values fall within the range of reported literature values of 0.18–0.43 and 0.13–0.53 for cis-but-2-ene and trans-but-2-ene respectively (Table 1). cis and trans-but-2-ene both yield (differing) mixtures of syn and anti conformers of CH₃CHOO, the relative amounts of which are not well known, and which are treated here initially as a single SCI species (this approximation is discussed further in Section 3.3). Berndt et al.³² recently reported a yield of 0.49 (±0.22) for the CH₃CHOO produced from trans-but-2-ene ozonolysis (also treating both syn and anti conformers as a single SCI species).

The yield of (CH₃)₂COO from TME ozonolysis obtained in this work is 0.32 (±0.02). This again falls within the (wide) range in the literature of 0.10–0.65, with Berndt et al.³² most recently reporting a yield of 0.45 (±0.20).

Fig. 1 shows the cumulative consumption of SO₂ relative to that of O₃, ΔSO₂versus ΔO₃ (after correction for dilution), as a function of [H₂O] for each experiment for the four alkenes studied. A fit to each experiment, extrapolating the experimental data to evaluate dSO₂/dO₃ at t = 0 (start of each experimental run) for use in eqn (E1)–(E3), is also shown. The overall change in SO₂, ΔSO₂, is seen to decrease substantially with increasing humidity (over a relatively narrow range of RH (1.5–20%)) for all four alkenes. This trend would be expected from the understood chemistry ((R1)–(R4)), as there is competition between SO₂, H₂O, and decomposition for reaction with the SCI.

Fig. 3 shows a fit of eqn (E3) to the data for each alkene, giving a slope of k₃/k₂, and an intercept of (k_d + L)/k₂. The results appear to show a generally linear relationship; however, for cis- and trans-but-2-ene and TME, the data point at the highest relative humidity accessible in this work ([H₂O] = 1.5–2.0 × 10¹⁷ cm⁻³) appears to deviate from this relationship. These data points lie outside the 95% confidence intervals defined by all the other (lower relative humidity) data for each alkene. For the analysis to determine k₃/k₂ and (k_d + L)/k₂ presented in Table 2, the points at the highest RH are excluded and the kinetic parameters are derived from a linear fit to the measurements from all other experiments. Extended analyses to account for the non-linearity observed for CH₃CHOO and (CH₃)₂COO are presented in the following sections.

Fig. 3 Application of eqn (E3) to derive rate constants for reaction of CH₂OO, CH₃CHOO (derived from cis-but-2-ene; C2B), CH₃CHOO (derived from trans-but-2-ene; T2B), and (CH₃)₂COO with H₂O (k₃) and decomposition (k_d), relative to that for reaction with SO₂, k₃/k₂ and (k_d + L)/k₂ – see text.

Table 2 SCI relative rate constants derived in this work using the single-SCI water monomer approach (i.e.eqn (E3); see text). Uncertainty ranges (±2σ, parentheses) indicate combined precision and systematic measurement error components

SCI	10^5k3/k2	10^−11 cm^−3kd/k2
CH2OO	3.3 (±1.1)	−2.3 (±3.5)
CH3CHOO (C2B)	26 (±10)	13 (±43)
CH3CHOO (T2B)	33 (±10)	−14 (±31)
(CH3)2COO	8.7 (±2.5)	63 (±14)

---

One potential explanation for the observed curvature in the CH₃CHOO and (CH₃)₂COO data is measurement error. ΔSO₂ is relatively small at high [H₂O] compared to the precision of the measurements; however, even allowing for associated uncertainties, the points at high RH do not fit the linear relationship successfully applied to the remaining data points. Moreover, any systematic error in the measurement of O₃, SO₂ or H₂O would also be expected to affect the results for the ethene system (and to a greater extent, given the slow ethene–ozone reaction rate and consequent lower overall chemical SO₂ loss observed), suggesting that the cause lies in contrasting chemical behaviour. In terms of experimental factors, H₂O was measured using multiple approaches (two dew-point hygrometers in addition to a solid state probe) with no evidence for any divergence with RH. SO₂ monitors can exhibit humidity-dependent interferences (quenching of the SO₂ signal), commonly of the order of a few percent, observed at very high RH, and corrected through incorporation of a nafion dryer (fitted in this case); in addition the monitor-derived SO₂ concentration increments were in agreement with those calculated from the measured SO₂ addition and chamber volume, across the relative humidity range studied.

It should be noted that the k_d values reported here represent upper limits, as a consequence of possible further chemical losses for the SCI within our experimental system (as represented by L in eqn (E3), notwithstanding the approach of extrapolating to the start of each experiment to minimise these). Other potential fates for SCIs include reaction with ozone,^42,43 other SCI,⁴³ carbonyl products,⁴⁴ acids,⁴⁵ or with the parent alkene⁴³ itself. Sensitivity analyses indicate that the reaction with ozone could be significant, as predicted by theory^42,43 with a possible contribution of up to 10% of SCI loss for (CH₃)₂COO at 2% RH, while total losses from reaction with SCI (self-reaction), carbonyls and alkenes are calculated to account for <1% of the total SCI loss under the experimental conditions applied.

3.2 CH₂OO

3.2.1 Linear Fit. For CH₂OO a linear fit (i.e. using eqn (E3)vs. H₂O) describes the observations well (Fig. 3) over the entire [H₂O] range studied (1–20 × 10¹⁶ molecules per cm⁻³). This fit gives relative rates k₃/k₂ of 3.3 (±1.1) × 10⁻⁵ and (k_d + L)/k₂ of −2.3 (±3.5) × 10¹¹ cm⁻³ (Table 2).

These relative rates can be placed on an absolute basis using absolute measurements of k₂(SCI + SO₂). In Table 3 we apply the absolute k₂ values reported by Welz et al.,¹⁵ obtained using direct methods at reduced pressure (4 Torr), to the relative rates shown in Table 2. Using this method, the value obtained for k₃(CH₂OO + H₂O) is 1.3 (±0.4) × 10⁻¹⁵ cm³ s⁻¹. This is consistent with the recent determination by Welz et al.,¹⁵ that k₃ < 4 × 10⁻¹⁵ cm³ s⁻¹, but is (ca. 14–50 times) greater than the recent estimates of Ouyang et al.¹⁸ (k₃ = 2.5 (±1) × 10⁻¹⁷ cm³ s⁻¹) and Stone et al.¹⁷ (k₃ < 9 × 10⁻¹⁷ cm³ s⁻¹).

Table 3 Comparison of SCI relative rate constants derived in this work to relative and absolute values from the literature. Uncertainty ranges (±2σ) indicate combined precision and systematic measurement error components

SCI	10^5k3/k2	10^15k3 (cm^3 s^−1)	10^−11kd/k2 (cm^−3)	k d (s^−1)	Ref.	Method	Conditions^a
a Experiments were conducted at atmospheric pressure unless stated otherwise. b Assuming k2 = 3.9 × 10^−11 cm^3 s^−1 (Welz et al.^15). c Assuming k2 = 4.55 × 10^−11 cm^3 s^−1. Average of k2 for syn and anti CH3CHOO conformers from Taatjes et al.^16 d Assuming k2 = 2.4 × 10^−11 cm^3 s^−1. k2 for syn-CH3CHOO from Taatjes et al.^16
CH2OO	3.3 (±1.1)	1.3 (±0.4)^b	−2.3 (±3.5)	−8.8 (±13)^b	This work	Ethene ozonolysis	298–303 K
		<4			Welz et al.^15	Alkyl iodide photolysis	4 Torr; 298 K
		<0.09			Stone et al.^17	Alkyl iodide photolysis	200 Torr; 295 K
CH3CHOO	26 (±10)	12 (±4.5)^c	13 (±43)	59 (±196)^c	This work	C2B ozonolysis	296–302 K
	33 (±10)	15 (±4.5)^c	−14 (±31)	−64 (±141)^c	This work	T2B ozonolysis	297–302 K
		anti 10 (±4); syn <4			Taatjes et al.^16	Alkyl iodide photolysis	4 Torr; 298 K
	8.8 (±0.4)		12 (±0.1)		Berndt et al.^32	T2B ozonolysis	293 K
				76 (25–228)	Fenske et al.^26	T2B ozonolysis	296 K
(CH3)2COO	8.7 (±2.5)	2.1 (±0.6)^d	63 (±14)	151 (±35)^d	This work	TME ozonolysis	298–299 K
	<0.4		42 (±3)		Berndt et al.^32	TME ozonolysis	293 K

---

The derived (k_d + L) value for CH₂OO using this method is −8.8 (±13) s⁻¹, i.e. zero within uncertainty. Theoretical work²² has predicted k_d(CH₂OO) to be small (∼0.3 s⁻¹), in agreement with the experimentally derived value reported here.

3.2.2 CH₂OO + (H₂O)₂. Recent experimental work⁴⁶ has reported the reaction of CH₂OO with the water dimer, (H₂O)₂, (reaction (R5)) to be very fast (1.1 × 10⁻¹¹ cm³ s⁻¹ – assuming k₂ = 3.9 × 10⁻¹¹ cm³ s⁻¹ (ref. 15)), in broad agreement with theoretical predictions,⁴⁷ but in contrast to other experimental work.⁴⁵

The CH₂OO data from this study appear to be well described by a linear fit under the experimental conditions applied (a fast reaction of CH₂OO with (H₂O)₂ would be manifested as a significant upward curvature in Fig. 3). However, this does not mean the results are inconsistent with reaction of CH₂OO with (H₂O)₂.

In Fig. 4, eqn (E4) (an expanded version of eqn (E3), including the SCI + (H₂O)₂ reaction (R5)) is applied to the data, now expressed in terms of (H₂O)₂, calculated for each RH via the equilibrium constant.⁴⁸

Fig. 4 Application of eqn (E4) to derive rate constants for reaction of CH₂OO with H₂O (k₃/k₂) and (H₂O)₂ (k₅/k₂) relative to that of CH₂OO with SO₂. Inset: eqn (E4) as shown in the main figure (red line), (E4) applied using the dimer reaction rate (k₅) reported by Berndt et al.⁴⁵ (1.1 × 10⁻¹¹ cm³ s⁻¹) (dashed line) and a fit of (E4) to the upper limits of the uncertainties in the ethene data (solid green line).

The value for k₃/k₂ (water monomer) derived from the fit shown in Fig. 4 is 2.5 (±0.7) × 10⁻⁵. It is seen that this value is rather insensitive to the inclusion of the (H₂O)₂ term in eqn (E4) as the value is within the uncertainties of the linear fit to the data presented in Fig. 3 – see also Table 2. Converting this value to an absolute value using the k₂ from Welz et al.¹⁵ gives k₃ = 9.9 (±2.9) × 10⁻¹⁶ cm³ s⁻¹. The derived value of k_d/k₂ is −6.4 (±66) × 10¹⁰ cm⁻³, which, using the Welz et al.¹⁵k₂ gives an absolute value for k_d of −1.5 (±16) s⁻¹. This is again indistinguishable from zero, within uncertainty, as is the k_d determined from eqn (E3) (Fig. 3). Note the large uncertainties in k_d, resulting from allowing three parameters to vary in the optimisation; consequently k_d was fixed to zero in eqn (E4) to determine the k₃/k₂ and k₅/k₂ values.

The resulting value of k₅/k₂ (water dimer) is 1.4 (±1.8) × 10⁻². Converting this to an absolute value using the k₂ from Welz et al.¹⁵ gives k₅ = 5.6 (±7.0) × 10⁻¹³ cm³ s⁻¹. This is roughly a factor of twenty smaller than the value derived by Berndt et al.,⁴⁶ but within a factor of two of the upper limit for k₅ deduced by Welz et al. (<3 × 10⁻¹³ cm³ s⁻¹) (ref. 45) from the data presented by Stone et al.¹⁷ The inset plot in Fig. 4 also shows two additional fits generated using eqn (E4) with k₃/k₂ fixed to 9.9 × 10⁻¹⁶ and k_d fixed to zero. One fit line uses the k₅ value reported by Berndt et al.⁴⁶ (blue dashed line). This is seen to overestimate the presented data. The green dotted line shows a fit to the upper limits of the uncertainties of the measured data. This yields a k₅/k₂ value of 0.10 (±0.01), giving an upper limit k₅ value of 3.9 (±0.39) × 10⁻¹² cm³ s⁻¹.

The contribution of (H₂O)₂ to the removal of CH₂OO increases in relation to that of H₂O as [H₂O] increases. Hence at typical atmospheric [H₂O] (∼2.5–7.5 × 10¹⁷ molecules per cm⁻³), greater than was accessible in this study, reaction with (H₂O)₂ could become the dominant sink for CH₂OO. In this case using just the H₂O monomer kinetics in models would considerably underestimate the total effect of water on removal of CH₂OO in the atmosphere.

3.3 CH₃CHOO

3.3.1 Single SCI Approach. The CH₃CHOO data shown in Fig. 3 for both cis and trans-but-2-ene appear to be well described (within the uncertainties) by a linear fit to eqn (E3), with the exception of the experiment at the highest RH ([H₂O] = 1.8–1.9 × 10¹⁷ cm⁻³) in both cases. The kinetic parameters derived from a linear fit to the data (Fig. 3), using eqn (E3) (which treats the system as producing a single SCI), excluding the highest RH experiments, are shown in Table 2. Very similar results are obtained for k₃/k₂ for CH₃CHOO derived from both cis- and trans-but-2-ene ozonolysis, with values of 26 (±10) × 10⁻⁵ and 33 (±10) × 10⁻⁵ respectively. The (k_d + L)/k₂ obtained for CH₃CHOO from cis-but-2-ene ozonolysis is 13 (±43) × 10¹¹ molecule cm⁻³ and from trans-but-2-ene ozonolysis −14 (±31) × 10¹¹ molecule cm⁻³. Berndt et al.³² reported the k₃/k₂ ratio from trans-but-2-ene ozonolysis to be 8.8 (±0.4) × 10⁻⁵ (also assuming a single SCI system), a factor of 3.75 smaller than that reported here.

The relative rate constants (Table 2) can be placed on an absolute basis using the measurements of k₂(SCI + SO₂) reported by Taatjes et al.¹⁶ (derived using the same methodology as for CH₂OO) (Table 3). As eqn (E3) treats the SCI produced as a single SCI, we use an average of the syn and anti conformer rates presented in Taatjes et al.,¹⁶ 4.55 × 10⁻¹¹ cm³ s⁻¹. Using this method, the value obtained for k₃(CH₃CHOO + H₂O) from cis-but-2-ene ozonolysis is 12 (±4.5) × 10⁻¹⁵ cm³ s⁻¹ and from trans-but-2-ene ozonolysis is 15 (±4.5) × 10⁻¹⁵ cm³ s⁻¹. Taking a mean of the k₃ values reported for the two CH₃CHOO conformers by Taatjes et al.¹⁶ gives a value of 7.0 × 10⁻¹⁵ cm³ s⁻¹, while Sheps et al.⁴⁹ give a mean value of 12 × 10⁻¹⁵ cm³ s⁻¹. The values obtained for (k_d + L) are 59 (±196) s⁻¹ from cis-but-2-ene and −64 (±141) s⁻¹ from trans-but-2-ene. Clearly there is a large uncertainty associated with the k_d determined from this analysis. Fenske et al.²⁶ have reported k_d(CH₃CHOO) from trans-but-2-ene ozonolysis to be 76 s⁻¹ (accurate to within a factor of three).

3.2.2 Two conformer system. In Fig. 1 it is evident that dSO₂/dO₃ falls rapidly with increasing [H₂O] for all but-2-ene experiments as RH is initially increased, but that the experiments at higher RH all appear to display a similar dSO₂/dO₃, i.e. the trend in decreasing SO₂ removal with increasing H₂O levels off. From this observation it appears that there may be competing H₂O dependencies to the SO₂ loss present. This is manifested in Fig. 3 as a curving over of the data at high RH. We propose two possible explanations for this observation: firstly that it arises from differing kinetics of the two CH₃CHOO conformers formed in but-2-ene ozonolysis; secondly that the behaviour may reflect the presence of an additional oxidant being formed in the ozonolysis system that reacts with SO₂ but is less sensitive to H₂O. The first of these possibilities is discussed below, the second is discussed in relation to (CH₃)₂COO in the following section.

One explanation for the observed non-linearity at high RH apparent in Fig. 3 is the differing reactivities of the syn- and anti-conformers of CH₃CHOO produced in the ozonolysis of cis- and trans-but-2-ene. It has been predicted⁵⁰ that the anti-conformer reacts with H₂O several orders of magnitude faster than the syn-conformer, while the rate constant for the SCI reaction with SO₂ has been determined experimentally¹⁶ to be about a factor of three greater for the anti-conformer than the syn-conformer. The fraction of each conformer that is lost to reaction with SO₂ can be considered in the same way as illustrated in eqn (E2), leading to eqn (E5) and (E6) below, plus simplifications outlined in the following text. The total loss of SO₂ to CH₃CHOO is then the sum of the fractional loss to each conformer, multiplied by the relative SCI yield (γ) (i.e. φ^syn/φ) of that conformer (eqn (E7)).

f = γ^synf^syn + γ^antif^anti (E7)

Eqn (E8) can then be fitted to the data presented in Fig. 3 for cis- and trans-but-2-ene (Fig. 5).

Here we make two assumptions to reduce the degrees of freedom and hence make the problem tractable with the dataset available. First it is assumed that k^syn₃[H₂O] may be neglected, in keeping with theoretical predictions²⁴ predicting k^syn_d to be over three orders of magnitude greater than k^syn₃[H₂O]. Further theoretical work⁵⁰ predicts a rate constant for syn-CH₃CHOO + H₂O of 2.39 × 10⁻¹⁸ cm³ s⁻¹, and recent experimental work⁴⁹ yields an upper limit for k^syn₃ of <2 × 10⁻¹⁶ cm³ s⁻¹. Hence with k^syn_d expected to be relatively fast, given the decomposition rate presented here for (CH₃)₂COO (a syn-conformer) and the facile decomposition route available via the hydroperoxide mechanism for the syn-conformer, it can be assumed that k^syn_d ≫ k^syn₃[H₂O] at the experimental conditions reported here. Second it is assumed that k^anti_d ≪ k^anti₃[H₂O] under the experimental conditions used. If the kinetics derived from treating the but-2-ene data in Fig. 3 as representing a single SCI are dominated by the anti-conformer then the k_d derived from these kinetics, which is indistinguishable from zero within the uncertainties, suggests that k^anti_d is small. Taatjes et al.¹⁶ report k^anti₃ to be 1.0 (±0.4) × 10⁻¹⁴ cm³ s⁻¹, while Sheps et al.⁴⁹ report a value of 2.4 (±0.4) × 10⁻¹⁴ cm³ s⁻¹. Thus, even at the lowest [H₂O] considered here (∼1 × 10¹⁶ cm⁻³), loss of anti-CH₃CHOO to H₂O would be >100 s⁻¹, and decomposition negligible in comparison (including a k^anti_d of 50 s⁻¹ changes the derived k^anti₃ and k^syn_d values by <5%).

Fig. 5 Fits of eqn (E8) to the cis-but-2-ene and trans-but-2-ene data shown in Fig. 3. γ^syn and γ^anti are 0.45 and 0.55 for cis-but-2-ene and 0.25 and 0.75 for trans-but-2-ene (see Fig. 6).

Fitting eqn (E8) to the data shown in Fig. 5 derives a range of values for k^anti₃ and k^syn_d dependent on the values of γ^anti and γ^syn used. As the CIs, once formed and thermalised, are expected to show the same kinetic behaviour in these experiments irrespective of their precursor alkene, the measurements from the cis-but-2-ene and trans-but-2-ene experiments can be used in combination to constrain k^anti₃, k^syn_d and also γ^syn and γ^anti from each alkene. Fig. 6 plots the k^anti₃vs. k^syn_d determined at different γ^syn and γ^anti from cis-but-2-ene and trans-but-2-ene. Where these two lines intercept represents the unique solution for both k^anti₃ and k^syn_d and for γ^syn and γ^anti (Table 4).

Fig. 6 The ranges of k^anti₃/k^anti₂ and k^syn_d/k^syn₂ determined from fitting eqn (E8) to the but-2-ene data (Fig. 5). The colour legend shows the fraction of the total CH₃CHOO formed that is syn-CH₃CHOO (γ^syn).

Table 4 Kinetic parameters derived for syn-CH₃CHOO and anti-CH₃CHOO from two-component fits to the trans-but-2-ene (T2B) and cis-but-2-ene (C2B) experiments (see Fig. 5). Also relative (γ) and absolute (φ) yields of the two conformers

CH3CHOO	10^4k3/k2	10^14k3 (cm^3 s^−1)	10^−13kd/k2 (cm^−3)	k d (s^−1)	γ		φ		Ref.	Conditions^b
					T2B	C2B	T2B	C2B
a γ ^syn = φ^syn/φ. b Experiments were conducted at atmospheric pressure unless stated otherwise and temperatures from 293 to 303 K.
syn	—	—	1.2 (±1.1)	288 (±275)	0.25	0.45	0.07	0.17	This work	—
		<0.4							Taatjes et al.^16	4 Torr
		<0.02							Sheps et al.^49	200 Torr
				20 (3–30)					Novelli et al.^23	735 Torr
anti	3.5 (±3.1)	2.3 (±2.1)	—	—	0.75	0.55	0.21	0.21	This work	—
		1.0 (±0.4)							Taatjes et al.^16	4 Torr
		2.4 (±0.4)							Sheps et al.^49	200 Torr

---

Fig. 6 determines k^anti₃/k^anti₂ to be 3.5 (±3.1) × 10⁻⁴ and k^syn_d/k^syn₂ to be 1.2 (±1.1) × 10¹³ cm⁻³. The 2σ uncertainties presented are, unsurprisingly, large as there are two free parameters. Using the relevant values of k₂ for the syn and anti-CH₃CHOO conformers from Taatjes et al.¹⁶ to place the relative rate constants on an absolute basis gives a value for k^anti₃ of 2.3 (±2.1) × 10⁻¹⁴ cm³ s⁻¹ and for k^syn_d of 288 (±275) s⁻¹. This k^anti₃ is comparable to (a factor of two greater than) that reported by Taatjes et al.¹⁶ (1.0 (±0.4) × 10⁻¹⁴). Novelli et al.²³ have recently reported k^syn_d to be an order of magnitude smaller (3–30 s⁻¹) based on direct observation of OH formation during trans-but-2-ene ozonolysis at atmospheric pressure.

The point of interception in Fig. 6 also determines the relative yields of the two conformers, γ^syn and γ^anti (which in turn has been used to derive the optimised fits shown in Fig. 5). For cis-but-2-ene these are determined as 0.45 and 0.55 for γ^syn and γ^anti respectively. For trans-but-2-ene they are determined as 0.25 for γ^syn and 0.75 for γ^anti. The analysis performed in this section has implications for the determination of the SCI yield. Using the relative rate constant k₃/k₂(anti-CH₃CHOO) obtained, as shown in Table 4, it is calculated that ∼90% of the anti-CH₃CHOO produced in the SCI yield experiments would react with SO₂. From the determined k_d/k₂(syn-CH₃CHOO) it is calculated that ∼67% of the syn-CH₃CHOO produced in the SCI yield experiments would react with SO₂. Applying these (with the corresponding syn and anti yields shown in Table 4) corrections to φ_min determines total SCI yields of 0.29 for trans-but-2-ene and 0.42 for cis-but-2-ene. These values both lie within the uncertainties in the SCI yields presented in Table 1 for the two but-2-ene systems.

It is not practicable to assess the possible contribution of the water dimer to the SCI loss for CH₃CHOO because of the number of free parameters that would result for a small dataset. However, theoretical predictions⁴⁷ suggest that this may be less important for CH₃CHOO than for CH₂OO, indicating k(H₂O)₂/k(H₂O) to be two orders of magnitude smaller for anti-CH₃CHOO than for CH₂OO.

3.4 (CH₃)₂COO

3.4.1 Linear fit. For (CH₃)₂COO a value of 8.7 (±2.5) × 10⁻⁵ (Table 2) was obtained from a linear fit to the data in Fig. 3, excluding the experiment performed at the highest RH ([H₂O] = 1.6 × 10¹⁷ cm⁻³) from the regression. (k_d + L)/k₂ for (CH₃)₂COO was determined as 63 (±14) × 10¹¹ molecule per cm⁻³. These values can be placed on an absolute basis using the measurements of k₂(syn-CH₃CHOO + SO₂) reported by Taatjes et al.¹⁶ as there are no reported k₂ values for (CH₃)₂COO from direct experiments and (CH₃)₂COO is a syn-conformer (i.e. there is always a methyl group in a syn orientation to the terminal oxygen). This gives values for k₃ ((CH₃)₂COO + H₂O) of 2.1 (±0.6) × 10⁻¹⁵ cm³ s⁻¹ and for k_d of 151 (±35) s⁻¹ (Table 3).

Berndt et al.³² have recently reported the k₃/k₂ ratio for (CH₃)₂COO to be <0.4 × 10⁻⁵ (i.e. approximately a factor of 22 lower than the relative rate reported in this study). Theoretical predictions⁵⁰ also suggest k₃ to be very slow, 3.9 × 10⁻¹⁷ cm³ s⁻¹. No measured values have been reported for k_d ((CH₃)₂COO), but a more facile overall decomposition than for CH₂OO or the mean of the CH₃CHOO isomers might be anticipated as the vinyl-hydroperoxide isomerisation channel²⁶ is always available.

3.4.2 Additional oxidant. In the case of the SCI formed from TME ozonolysis, (CH₃)₂COO, there is always a methyl group in a syn position to the carbonyl oxide moiety, thus the analysis presented in Section 3.3.2 for the CH₃CHOO isomers does not apply. A possible alternative explanation for the observed behaviour (and possible contributor to the behaviour observed in the but-2-ene systems) is that there is a further oxidant (X) of SO₂, in addition to the SCI, being formed during the ozonolysis reaction. If this oxidant reacts relatively slowly with H₂O, it could give rise to the apparent ‘two component’ nature of the observations seen in Fig. 3. It may also provide an alternative explanation of the observed nature of the SO₂ loss from the but-2-ene experiments or could be occurring in addition to the effects of differing conformer reactivities.

Eqn (E9) (below) is an expanded version of (E2), in which we consider the contribution from a second SO₂ oxidant, making the approximation that this species does not react appreciably with water vapour. In eqn (E9), f is the sum of f^SCI (the fraction of SCI reacting with SO₂) and f^x, each multiplied by the relative yield of the total oxidant (i.e. SCI + X) γ^SCI and γ^x. Following the assumption of negligible H₂O reactivity, (dSO₂/dO₃)_x in eqn (E9) can be derived from the SO₂ loss at the highest RH experiments (i.e. when all the SO₂ loss is attributed to X + SO₂) of ∼10 ppbv. Therefore, loss of SO₂ from reaction with X, relative to the loss of O₃ (dSO₂/dO₃)_x is approximately 0.025. φ represents the total oxidant yield (i.e. φ_SCI + φ_x). Assuming that γ^x is not dominant (<0.5), then φ, as calculated from correcting φ_min as in Section 3.1, changes little (0.31–0.34) from the value presented in Table 1. Eqn (E10) is then an expanded version of eqn (E3) that includes the additional oxidant term.

Fig. 7 shows eqn (E10) fitted to the TME data from Fig. 3. It is not possible to determine unique values for the parameters included in eqn (E10) due to the degrees of freedom vs. the limited data set. The fit shown in Fig. 7 uses values of φ = 0.34, γ^SCI = 0.88, k₃/k₂ = 6.7 × 10⁻⁴ and k_d/k₂ = 1.2 × 10¹² cm⁻³. However, Fig. 7 does demonstrate that a two-oxidant system, as represented by eqn (E10), is able to describe the data within uncertainty.

Fig. 7 Data for TME ozonolysis (as shown in Fig. 3), with fit to eqn (E10), and linear fit to the full dataset.

Fig. 7 also includes a linear fit (i.e.eqn (E3)) to the full (CH₃)₂COO dataset (including the highest RH experiment). While it seems unlikely that the curvature observed in the data is a result of measurement error (as described in Section 3.1), this must be considered as a possibility for (CH₃)₂COO in light of the two conformer explanation not being applicable. The linear fit in Fig. 7 gives a k₃/k₂ value of 3.8 (±3.2) × 10⁻⁵ cm³ s⁻¹, a factor of two smaller than the k₃/k₂ value presented in Table 2.

The most obvious candidate for an additional oxidant present to consume SO₂ is OH. OH radicals are produced in the chamber, primarily (in the absence of sunlight and NO_x) through the alkene + ozone reaction.¹¹ However cyclohexane was added in excess at the beginning of each experiment to act as an OH scavenger, such that SO₂ reaction with OH was calculated to be ≤1% of the total chemical SO₂ removal in all experiments. Other potential candidates for this oxidant species include the (stabilised) vinyl hydroperoxide (VHP) intermediate, a secondary ozonide (formed through an SO₂–SCI cyclic adduct), and dioxirane (Scheme 1).

Drozd et al.¹⁰ have presented evidence for substantial VHP stabilisation (derived from the (CH₃)₂COO CI) at pressures of a few hundred Torr, with a lifetime of the order of a few hundred milliseconds with respect to decomposition to form OH, providing scope for bimolecular reactions of this species to occur. This would be consistent with the kinetic observations presented here: a small yield in the systems with syn-SCIs, while for ethene, no VHP intermediate is available, and the standard chemistry ((R1)–(R4)) can be used to satisfactorily reproduce the observations. However, no significant SO₂ reactivity is known for the peroxide or alkene functionalities present in the closed shell VHP in isolation, hence it may be surprising if this species reacted rapidly with SO₂, although theoretical studies have suggested that the VHP may react with H₂O.²³

Secondary ozonide species, formed as an adduct from the SCI + SO₂ reaction, have been suggested to account for the observed isotopic exchange in alkene–ozone–SO₂ systems,²⁵ and are suggested to have lifetimes of seconds or longer.¹⁹ Such a secondary ozonide could react with a further SO₂ molecule (i.e. a two-component secondary ozonide catalysed oxidation route for SO₂ to SO₃ conversion), however a substantial humidity dependence to the overall process may still be anticipated (on the basis of SCI removal through the SCI + H₂O reaction), which is not observed here.

The ‘hot acid/ester channel’ (rearrangement and decomposition via a dioxirane intermediate) is the dominant isomerisation route available for anti-SCIs. Although the hydroperoxide channel is the principal isomerisation route for syn-SCIs, the ester mechanism is also available,²⁵ and it is likely that a small proportion of the syn-CI will isomerise through this channel to form a dioxirane. Dioxiranes are known to be highly reactive and selective oxidising agents,^51,52 and have a particular affinity for sulphur compounds.⁵³ Additional oxidation of SO₂ by the dioxirane, over and above that arising directly from reaction with the SCI, could then explain the observed behaviour of SO₂ in the TME experiments (and contribute to the behaviour observed for the but-2-ene systems). For the small CI CH₂OO, formed in the ethene system, it has been predicted that the dioxirane formed is considerably less stable than the methyl substituted dioxiranes formed from but-2-ene and TME ozonolysis, and furthermore will decompose promptly due to chemical activation.

The possibility of non-CI products of the ozonolysis system being responsible for some of the observed SO₂ oxidation has been suggested previously as an alternative (or additional) explanation for the observed behaviour of SO₂ in the atmosphere.^1,54,55 In their laboratory study Berndt et al.¹⁹ note that their data were not perfectly described by a model in which a single (SCI related) SO₂ oxidation process was assumed, and commented that the SO₂ oxidation in ozonolysis systems may in fact be more complex. Taatjes et al.⁵⁶ have recently suggested that these data¹⁹ are consistent with a two-component oxidation system, either two processes removing SO₂ in parallel (as described above), or through a sequential two-step SO₂ removal mechanism, such as the secondary ozonide route outlined above. However, this latter mechanism cannot (in isolation) account for the observations presented here.

The presence of an additional oxidant would have implications for the role of alkene ozonolysis in oxidation of trace gases in the atmosphere. If an oxidant is being formed that reacts slowly with H₂O then this, perhaps in addition to SCI, may contribute to the additional (non-OH) SO₂ oxidation observed in recent field experiments.¹ Further investigation of this possibility is needed.

As for CH₃CHOO, the analysis performed in this section has implications for the determination of the SCI ((CH₃)₂COO) yield. Using the value of k_d derived from the linear fit to all the data shown in Fig. 7 would indicate an SCI yield of 0.33. Using the value of k_d derived from the ‘additional oxidant fit in Fig. 7, and taking into account that ∼10 ppb of the SO₂ loss in the high SO₂ experiment would be attributed to reaction with the additional oxidant rather than the SCI, determines a slightly lower φ_min of 0.23, and a corrected yield of 0.32, very similar to that shown in Table 1. Both of these possible alternative SCI yields lie within the uncertainties in the SCI yield from TME ozonolysis presented in Table 1.

As for CH₃CHOO, it is not possible to quantitatively consider the contribution of the water dimer to the SCI loss given the limited data, but theory⁵³ predicts k(H₂O)₂/k(H₂O) to be three orders of magnitude smaller than that for CH₂OO suggesting that reaction with the water dimer would be unimportant for (CH₃)₂COO at typical atmospheric boundary layer [H₂O].

4. Atmospheric implications

The derived values for k₃ reported in Table 3 correspond to loss rates for reaction of SCI with H₂O in the atmosphere of 650 s⁻¹ for CH₂OO, ∼3500 s⁻¹ for CH₃CHOO and ∼1050 s⁻¹ for (CH₃)₂COO (assuming [H₂O] = 5 × 10¹⁷ molecules per cm⁻³, equivalent to an RH of 65% at 298 K). Comparing this to the derived k_d values it is seen that reaction with H₂O is predicted to be the main sink for SCI in the atmosphere, but also that loss through decomposition cannot be neglected for some SCIs – contributing on the order of 0–1% for CH₃CHOO and 13% for (CH₃)₂COO.

An estimate of a mean steady state SCI concentration in the background atmospheric boundary layer can then be calculated using eqn (E11).

Using the data given below, a steady state SCI concentration of 1.7 × 10³ molecules per cm⁻³ is estimated for an ozonolysis source (noting that other potential atmospheric sources of SCI exist, e.g. photolysis of alkyl-iodides in the marine boundary layer, and sinks, e.g. reaction with NO and NO₂). This assumes an ozone mixing ratio of 40 ppbv, an alkene mixing ratio of 2 ppbv, φ of 0.35, and mean reaction rate constants k₁ (alkene–ozone) of 1 × 10⁻¹⁶ cm³ s⁻¹; k₂ (SCI + SO₂) of 3.5 × 10⁻¹¹ cm³ s⁻¹, k₃ (SCI + H₂O) of 2 × 10⁻¹⁵ cm³ s⁻¹, k_d of 30 s⁻¹ with [H₂O] of 5 × 10¹⁷ cm⁻³ (RH ∼ 65%).

However, in the case of CH₃CHOO the data shown in Fig. 3, and the discussion above, indicate contributions from multiple species – syn- and anti-conformers with contrasting behaviour. It is clear that the burden of CH₃CHOO in the atmosphere would be better described by considering these two fractions of SO₂ loss separately. Eqn (E12) expands eqn (E11) to treat the two conformers separately, where [SCI] = [anti-SCI] + [syn-SCI], making the same assumptions as for the analysis of CH₃CHOO in eqn (E5) and (E6). k^anti₃ is estimated to be 1 × 10⁻¹⁴ cm³ s⁻¹ (taking into account that CH₂OO is considered as an anti-SCI in this analysis and that the derived k₃(CH₂OO) is more than an order of magnitude smaller than the derived k₃(CH₃CHOO) of 2.3 × 10⁻¹⁴ cm³ s⁻¹), k^syn_d is assumed to be 200 s⁻¹ and φ^anti = φ^syn = 0.175. Additionally the anti-SCI + water dimer reaction is also considered, using a value of 5.6 × 10⁻¹³ cm³ s⁻¹ as derived for CH₂OO in this work.

Using these values in eqn (E12) determines [anti-SCI] = 164 molecules per cm⁻³ and [syn-SCI] = 4.4 × 10³ molecules per cm⁻³. The formation of an additional oxidant during alkene ozonolysis would be expected to have a similar effect to the two component contribution presented in eqn (E12) based on the apparent yields from the experiments presented here. From this analysis the atmospheric SCI burden is seen to likely be dominated by syn-SCI since this term is at least an order of magnitude greater than the anti-SCI term.

A typical diurnal loss rate of SO₂ to OH (k_OH × [OH]) is 9 × 10⁻⁷ s⁻¹,¹⁵ while the SO₂ loss rate due to reaction with SCI, using the values derived from eqn (E12), would be 1.2 × 10⁻⁷ s⁻¹. This suggests, for the conditions and assumptions given above, the loss of SO₂ to SCI to be about 13% of loss to OH. This analysis neglects additional chemical sinks for SCI, which would reduce SCI abundance but are unlikely to be competitive with the two main SCI loss processes identified herein. SCI concentrations are expected to vary greatly depending on the local environment, e.g. alkene abundance may be considerably higher (and with a different reactive mix of alkenes) in a forested environment, compared to a rural background environment. The majority of the SCI burden, particularly in forested regions, is likely to be dominated by larger SCI derived from (C₅) isoprene and (C₁₀) monoterpenes. The chemistry of these species could differ greatly from the small SCI reported here (which we have found to be structure specific, even for small alkene systems), especially for tethered SCI derived from ozonolysis of internal double bonds within (for example) some monoterpenes. It is clear that the total SCI loss rate is dependent upon SCI identity and configuration, and that further work is required to quantify speciated SCI in the atmosphere, and to accurately calculate SCI concentrations for use in atmospheric modelling.

5. Conclusions

It has been shown that at relatively low [H₂O] (<1 × 10¹⁷ cm⁻³) the loss of SO₂ in the presence of four ozone–alkene systems: ethene, cis-but-2-ene, trans-but-2-ene and 2,3-dimethyl-but-2-ene significantly decreases with increasing water vapour. This is consistent with production of a stabilised Criegee intermediate from the ozonolysis reaction and subsequent reaction of this species with SO₂ and H₂O. Competition between H₂O and SO₂ for reaction with the SCI leads to the observed relationship which is sensitive to water vapour abundance over a relatively narrow range of RH. Derived kinetic data for these ozonolysis systems shows that the reaction rates are dependent on the structure of the SCI. At [H₂O] > 1 × 10¹⁷ cm⁻³ the SO₂ loss in the presence of cis- and trans-but-2-ene, and 2,3-dimethyl-but-2-ene appears to show a reduced dependence upon H₂O. The results suggest that there is an H₂O dependent and an H₂O independent fraction to the observed SO₂ loss in these systems. These two fractions may be attributable to differing kinetics of the two conformers produced in but-2-ene ozonolysis or to other oxidant products of the alkene ozonolysis reaction. This observation means that SCI structure must be considered in atmospheric modelling of SCI production from alkene ozonolysis, and suggests that the atmospheric SCI burden (and hence the oxidation of trace gases) will be dominated by syn-SCI.

This work provides constraints on the behaviour of SCI formed through alkene ozonolysis under conditions relevant to the atmospheric boundary layer, but also highlights the complex nature and incomplete current understanding of the ozonolysis system. Further research is needed to definitively quantify the impact of this chemistry upon atmospheric oxidation.

Acknowledgements

The assistance of the EUPHORE staff is gratefully acknowledged. Marie Camredon, Stephanie La and Mat Evans are thanked for helpful discussions. This work was funded by EU FP7 EUROCHAMP 2 Transnational Access activity (E2-2012-05-28-0077), the UK NERC (NE/K005448/1) and Fundacion CEAM. Fundación CEAM is partly supported by Generalitat Valenciana, and the projects GRACCIE (Consolider-Ingenio 2010) and FEEDBACKS (Prometeo – Generalitat Valenciana). EUPHORE instrumentation is partly funded by the Spanish Ministry of Science and Innovation, through INNPLANTA project: PCT-440000-2010-003. LV is supported by the Max Planck Graduate Center with the Johannes Gutenberg-Universität Mainz (MPGC).

Notes and references

1. R. L. Mauldin III, T. Berndt, M. Sipilä, P. Paasonen, T. Petäjä, S. Kim, T. Kurtén, F. Stratmann, V.-M. Kerminen and M. Kulmala, Nature, 2012, 488, 193.
2. G. Sarwar, K. Fahey, R. Kwok, R. C. Gilliam, S. J. Roselle, R. Mathur, J. Xue, J. Yu and W. P. L. Carter, Atmos. Environ., 2013, 68, 186.
3. P. Forster and V. Ramaswamy, et al., IPCC 4, 2007, ch. 2, vol. 160.
4. R. A. Cox and S. A. Penkett, Nature, 1971, 230, 321.
5. R. Criegee and G. Wenner, Liebigs Ann. Chem., 1949, 564, 9.
6. D. Johnson and G. Marston, Chem. Soc. Rev., 2008, 37, 699.
7. H. Niki, P. D. Maker, C. M. Savage, L. P. Breitenbach and M. D. Hurley, J. Phys. Chem., 1987, 91, 941.
8. R. I. Martinez and J. T. Herron, J. Phys. Chem., 1987, 91, 946.
9. G. T. Drozd, J. Kroll and N. M Donahue, J. Phys. Chem. A, 2011, 115, 161.
10. G. T. Drozd and N. M Donahue, J. Phys. Chem. A, 2011, 115, 4381.
11. M. S. Alam, A. R. Rickard, M. Camredon, K. P. Wyche, T. Carr, K. E. Hornsby, P. S. Monks and W. J. Bloss, J. Phys. Chem. A, 2013, 117, 12468.
12. K. Izumi, M. Mizuochi, K. Murano and T. Fukuyama, Atmos. Environ., 1987, 21, 1541.
13. J. G. Calvert, R. Atkinson, J. A. Kerr, S. Madronich, G. K. Moortgat, T. J. Wallington and G. Yarwood, The Mechanisms of Atmospheric Oxidation of the Alkenes, Oxford University Press, New York, 2000.
14. C. A. Taatjes, G. Meloni, T. M. Selby, A. J. Trevitt, D. L. Osborn, C. J. Percival and D. E. Shallcross, J. Am. Chem. Soc., 2008, 130, 11883.
15. O. Welz, J. D. Savee, D. L. Osborn, S. S. Vasu, C. J. Percival, D. E. Shallcross and C. A. Taatjes, Science, 2012, 335, 204.
16. C. A. Taatjes, O. Welz, A. J. Eskola, J. D. Savee, A. M. Scheer, D. E. Shallcross, B. Rotavera, E. P. F. Lee, J. M. Dyke, D. W. K. Mok, D. L. Osborn and C. J. Percival, Science, 2013, 340, 177.
17. D. Stone, M. Blitz, L. Daubney, N. U. M. Howes and P. Seakins, Phys. Chem. Chem. Phys., 2014, 16, 1139.
18. B. Ouyang, M. W. McLeod, R. L. Jones and W. J. Bloss, Phys. Chem. Chem. Phys., 2013, 15, 17070.
19. T. Berndt, T. Jokinen, R. L. Mauldin III, T. Petaja, H. Herrmann, H. Junninen, P. Paasonen, D. R. Worsnop and M. Sipila, J. Phys. Chem. Lett., 2012, 3, 2892.
20. L. Vereecken, H. Harder and A. Novelli, Phys. Chem. Chem. Phys., 2012, 14, 14682.
21. J. R. Pierce, M. J. Evans, C. E. Scott, S. D. D’Andrea, D. K. Farmer, E. Swietlicki and D. V. Spracklen, Atmos. Chem. Phys., 2013, 13, 3163.
22. M. Olzmann, E. Kraka, D. Cremer, R. Gutbrod and S. Andersson, J. Phys. Chem. A, 1997, 101, 9421.
23. A. Novelli, L. Vereecken, J. Lelieveld and H. Harder, Phys. Chem. Chem. Phys., 2014, 16, 19941.
24. K. T. Kuwata, M. R. Hermes, M. J. Carlson and C. K. Zogg, J. Phys. Chem. A, 2010, 114, 9192.
25. J. T. Herron, R. I. Martinez and R. E. Huie, Int. J. Chem. Kinet., 1982, 14, 201.
26. J. D. Fenske, A. S. Hasson, A. W. Ho and S. E. Paulson, J. Phys. Chem. A, 2000, 104, 9921.
27. J. H. Kroll, S. R. Sahay, J. G. Anderson, K. L. Demerjian and N. M. Donahue, J. Phys. Chem. A, 2001, 105, 4446.
28. J. M. Beames, F. Liu, L. Lu and M. I. Lester, J. Am. Chem. Soc., 2012, 134, 19104.
29. J. M. Beames, F. Liu, L. Lu and M. I. Lester, J. Chem. Phys., 2013, 138, 244307.
30. K. H. Becker, EUPHORE: Final Report to the European Commission, Contract EV5V-CT92-0059, Bergische Universität, Wuppertal, Germany, 1996.
31. M. S. Alam, M. Camredon, A. R. Rickard, T. Carr, K. P. Wyche, K. E. Hornsby, P. S. Monks and W. J. Bloss, Phys. Chem. Chem. Phys., 2011, 13, 11002.
32. T. Berndt, T. Jokinen, M. Sipilä, R. L. Mauldin, H. Herrmann, F. Stratmann, H. Junninen and M. Kulmala, Atoms. Environ., 2014, 89, 603.
33. S. M. Saunders, M. E. Jenkin, R. G. Derwent and M. J. Pilling, Atmos. Chem. Phys., 2003, 3, 161.
34. H. Niki, P. D. Maker, C. M. Savage and L. P. Breitenbach, J. Phys. Chem., 1981, 85, 1024.
35. S. Hatakeyama, H. Kobayashi and H. Akimoto, J. Phys. Chem., 1984, 88, 4736.
36. O. Horie and G. K. Moortgat, Atmos. Environ., Part A, 1991, 24, 1881.
37. P. Neeb, O. Horie and G. K. Moortgat, Int. J. Chem. Kinet., 1996, 28, 721.
38. A. S. Hasson, G. Orzechowska and S. E. Paulson, J. Geophys. Res., 2001, 106, 34131.
39. A. R. Rickard, D. Johnson, C. D. McGill and G. Marston, J. Phys. Chem. A, 1999, 103, 7656.
40. H. Niki, P. D. Maker, C. M. Savage and L. P. Breitenbach, Chem. Phys. Lett., 1977, 46, 327.
41. R. A. Cox and S. A. Penkett, J. Chem. Soc., Faraday. Trans., 1972, 68, 1735.
42. H. G. Kjaergaard, T. Kurtén, L. B. Nielsen, S. Jørgensen and P. O. Wennberg, J. Phys. Chem. Lett., 2013, 4, 2525.
43. L. Vereecken, H. Harder and A. Novelli, Phys. Chem. Chem. Phys., 2014, 16, 4039.
44. C. A. Taatjes, O. Welz, A. J. Eskola, J. D. Savee, D. L. Osborn, E. P. F. Lee, J. M. Dyke, D. W. K. Mok, D. E. Shallcross and C. J. Percival, Phys. Chem. Chem. Phys., 2012, 14, 10391.
45. O. Welz, A. J. Eskola, L. Sheps, B. Rotavera, J. D. Savee, A. M. Scheer, D. L. Osborn, D. Lowe, A. Murray Booth, P. Xiao, M. Anwar H. Kahn, C. J. Percival, D. E. Shallcross and C. A. Taatjes, Angew. Chem., Int. Ed., 2014, 18, 4547.
46. T. Berndt, J. Voigtländer, F. Stratmann, H. Junninen, R. L. Mauldin III, M. Sipilä, M. Kulmala and H. Herrmann, Phys. Chem. Chem. Phys., 2014, 16, 19130.
47. A. B. Ryzhkov and P. A. Ariya, Phys. Chem. Chem. Phys., 2004, 6, 5042.
48. Y. Scribano, N. Goldman, R. J. Saykally and C. Leforestier, J. Phys. Chem. A, 2006, 110, 5411.
49. L. Sheps, A. M. Scully and K. Au, Phys. Chem. Chem. Phys., 2014, 16, 26701.
50. J. M. Anglada, J. Gonzalez and M. Torrent-Sucarrat, Phys. Chem. Chem. Phys., 2011, 13, 13034.
51. W. Adam, R. Curci and J. O. Edwards, Acc. Chem. Res., 1989, 22, 205.
52. R. W. Murray, Chem. Rev., 1989, 89, 1187.
53. R. Curci, A. Dinoi and M. F. Rubino, Pure Appl. Chem., 1995, 67, 811.
54. D. Heard, Nature, 2012, 488, 164.
55. J. Prousek, Chem. Listy, 2009, 103, 271.
56. C. A. Taatjes, D. E. Shallcross and C. J. Percival, Phys. Chem. Chem. Phys., 2013, 16, 1704.

---