Bimolecular sinks of Criegee intermediates derived from hydro ﬂ uoroole ﬁ ns – a computational analysis †

A novel range of stabilised Criegee intermediate (sCI) species with halogenated substituent groups have been identi ﬁ ed as products to the reaction between with gaseous ozone and hydro ﬂ uoroole ﬁ ns (HFOs), a series of recently-developed and increasingly prevalent haloalkene refrigerants. The bimolecular chemistry of this group of hydro ﬂ uoroole ﬁ n-derived sCIs (HFO-sCIs) has yet to be explored in any signi ﬁ cant detail so this work evaluates the reaction chemistry of common tropospheric gaseous species with the following group of HFO-sCIs: syn - & anti -CF 3 CHOO & syn - & anti -CF 3 CFOO. Using high-level theoretical calculations (DF-HF/DF-LCCSD(T)-F12a//B3LYP/aug-cc-pVTZ), this study demonstrates that HFO-sCIs will deplete many pollutants ( e.g. HCHO, SO 2 & H 2 S) but also act as a source of other atmospheric contaminants ( e.g. SO 3 & TFA). The bimolecular reactivity of the HFO-sCIs were compared against CH 2 OO, the most frequently studied sCI, for which the general reactivity trend has been identi ﬁ ed: k THEO ( syn -CF 3 CHOO) < k THEO ( anti -CF 3 CHOO) z k THEO (CH 2 OO) (cid:1) k THEO ( anti -CF 3 CFOO) < k THEO ( syn -CF 3 CFOO). In general syn & anti -CF 3 substituents reduce overall sCI reactivity compared to similar non-halogenated sCI species, whereas both syn & anti -F substituents signi ﬁ cantly increase HFO-sCI reactivity. While HFO-sCI reactivity is largely dictated by the identity and location of the sCI substituent groups, there are co-reactants that alter these observed trends in reactivity


Introduction
2][3][4][5][6] Furthermore, as well as being used as blowing foam and as propellants, HFOs are very important working uids for waste heat recovery applications in heat pumps, as they are stable, have favourable toxicity proles and are compatible with many plastics and elastomers used in such set-ups. 7,8When haloalkenes breakdown it is the Cl & Br radicals that are primarily implicated in ozone destruction in the upper atmosphere and therefore only HFOs containing Cl & Br atoms (e.g.0][11][12] Additionally, they are relatively non-toxic, have minimal impact on global warming due to their short lifespans, and have high ignition energies compared to the rst generation of refrigerants used in the 19th century (e.g.13][14][15][16][17] Moreover, whereas CO 2 dominates the lower IR wave numbers (600-750 cm −1 ), HFCs/HCFCs are also strongly IR active in the higher wavenumbers (1000-1400 cm −1 ), increasing their heat trapping ability. 9,18][24][25][26] In areas as widespread as China, the USA and the EU, the phasing in of HFOs has been increasingly integrated into regional regulations, to try and eliminate HCFC & HFC emissions, and this has led to HFOs seeing increasing use in refrigerators, insulation and vehicle cooling units. 1,27,28For example, a UK government 2022 report on the "F gas regulation in Great Britain" documents that HFO-1234yf has dominated total refrigerant usage in new small vehicles since 2017 and HFO-HFC blends use could eliminate high GWP HFCs from transport refrigeration by 2050. 29Modest concerns about both the tropospheric implications of HFO breakdown (e.g.TFA formation) and about the carbon-intensity of HFO production has meant that both the UK and the EU are currently assessing the role of HFOs in future climate regulations. 30Many refrigerant experts and industrial leaders continue to strongly support HFO use and are consulting on such regulations with governments currently, especially as the alternatives to HFOs also have trade-offs in the areas of toxicity, ammability and/or high-pressure requirements. 31,32Examples of CFCs, HCFCs, HFCs and HFOs with their GWP 100 and ODP values and lifespan are displayed in Table 1.
52,53 The maximum ozonolysis yield for most HFOs reviewed are around 1%, but a few studies indicate that up to 10% of the removal of some haloalkenes occur via reaction with O 3 , a signicant enough pathway that HFO ozonolysis is being Table 1 A tabulation of important atmospheric attributes of examples from three different generations of refrigerants: chlorofluorocarbon (CFCs), hydrofluorocarbons (HFCs) and hydrofluoroolefins (HFOs).Attributes featured include: global warming potential over 100 years (GWP 100 ), ozone depletion potential (ODP), atmospheric lifetime, and any hydrofluoroolefin-derived stabilized Criegee intermediates (HFO-sCIs) produced from the ozonolysis of this species Fig. 1 The general schematic for the ozonolysis of a standard alkene via a short-lived primary ozonide (POZ) intermediate products, which produces in a pair of final products: a Criegee intermediate (CI) and a carbonyl product (either aldehyde or ketone).A significant portion of the CI population is formed with a high degree of energy indicated by asterisk.
investigated thoroughly at present. 47,52Furthermore, while direct CI branching fraction measurements for HFO ozonolysis are not found in the literature, indirect measurements of their aldehyde/ketone co-products show strong yields for CF 3 CHOO species and other similar CIs (>40%). 52,54This indicates that ozonolysis of haloalkenes in general are a major pathway for producing HFO-sCIs, even if these halogenated sCIs are low in abundance and ozonolysis is not the dominant pathway for removal of HFOs.This is a highly important topic and of relevance to the study of Criegee intermediates.Relative HFO ozonolysis percentages have been avoided here due to this need for a much more thorough re-investigation.CIs are produced from the ozonolysis of a large variety of both halo-alkenes and non-halogenated alkenes and the atmospheric role of these CIs have been the subject of increased scientic study in the literature in recent years. 5,6,34,48This is due to the fact that a large portion of the CIs produced by alkene ozonolysis (35-50%) are formed with such excess of internal energy that they undergo a rapid unimolecular fragmentation process, which is an important non-photolytic source of the atmospheric 'detergent' HOc species.The remaining fraction are stabilised by collision, generating stabilised Criegee intermediates (sCIs), that can participate in bimolecular reactions with other atmospheric species.These sCIs deplete pollutants to such an extent that the simplest sCI, formaldehyde oxide (CH 2 OO), can compete with the atmospheric "detergent" HOc species as a sink for polar pollutants, such as HNO 3 and SO 2 , especially in CH 2 OO-rich boreal forest environments ([CH 2 OO] ∼ 1-5 × 10 4 molec.per cm 3 ). 42,55To examine the capacity of sCIs to deplete pollutants is the purpose of this study, as well as much of the increased scientic examination of sCIs overall.
The theoretical literature (see Table 1) shows that the ozonolysis of prominent HFOs produces several particular HFO-derived stabilised Criegee intermediates (HFO-sCIs): CH 2 OO, syn-& anti-FCHOO, syn-& anti-ClCHOO, syn-& anti-CF 3 CHOO and syn-& anti-CF 3 CFOO. 4,5,45,46,53,568][59][60][61][62][63][64][65][66] In a small number of studies, the bimolecular chemistry of HFO-sCIs has been interrogated using computational chemistry techniques: CH 2 OO somewhat exhaustively; [67][68][69][70][71] 72,73,77 The literature has very little analysis of syn-or anti-CF 3 CFOO.][51] Some co-reactants (HNO 3 , HCl & H 2 S) are selected for study here because there are some environments in which CH 2 OO competes with OH radicals to facilitate the depletion of these co-reactants, as noted by Khan et al. 42 Furthermore, some coreactants studied here, such as SO 2 , are of interest because their reactions with sCIs are a competitive source for tropospherically relevant species such as SO 3 , which is considered vital in aerosol nucleation pathways. 84,85Another example is that sCI + MeOH & H 2 O reactions produce a large proportion of the global ux of the a-alkoxyalkylhydroperoxides (AAAH) and ahydroxy-hydroperoxides (HHP).3,86 One last area of consideration in co-reactant targeting is that Kumar et al. propose that sCI reactivity with H 2 X (X = O, S, Se, and Te) increases with the size of the X atom, referred to as heteroatom tuning of the co-reactants. 87In this study, the co-reactants with low mass heteroatoms, H 2 O & HF, are compared with species which have the same structures but larger heteroatoms of the same periodic group, H 2 S & HCl, to see if all the HFO-sCIs in this chapter replicate the co-reactant heteroatom tuning patterns observed in the Kumar et al. study. 87The last reactions featured are examined because the HCHO + sCIs 2 & 3 and sCI 1 + CF 3 CHO reactions are part of the same multistep potential energy surface (PES) and so the theoretical rate constants (k THEO ) and product branching ratios (G THEO ) of sCI 1 + CF 3 CHO can be determined with little additional computational cost.This is also the case for HCHO + sCIs 4 & 5 and sCI 1 + CF 3 CFO reactions which are all part of the same multistep potential energy surface.
By using computational chemistry methods to review reactions with this variety of common tropospheric gas species, this study aims to evaluate to what degree the bimolecular chemistry of HFO-sCI (syn-& anti-CF 3 CHOO and syn-& anti-CF 3 CFOO) differs from that of CH 2 OO, as well as the other nonhalogenated sCIs.

Methodologies
9][90][91][92][93][94][95][96][97][98][99] Each minimum or TS structure is veried through evaluation of the number of imaginary frequencies from subsequent harmonic frequency calculations, performed at the same level of theory as the optimisation.TSs identied in this manner are mapped to the geometric minima that they bisect using intrinsic reaction coordinate (IRC) calculations.For very shallow TSs and barrierless reactions, IRC calculations become computationally challenging, therefore in these cases relaxed potential energy scans are performed using incremental separation of a molecular bond (which forms a key component of the reaction coordinate) until the local minima is reached.The relax surface scans for non-barrierless channels each have a local energy maximum, which denote the location of the TS structure on the potential energy surface, whereas the energy diagram for barrierless channels only display asymptotic Morse-like behaviour.Optimisations, harmonic frequencies, IRCs and potential energy scans are all performed using the Gaussian09 computational chemistry programme. 100All optimisations in this study were performed on singlet surfaces.
Rate constants and PESs are constructed by using DFT optimised structures with energies rened using a higher level, ab initio approach.Single point molecular energy calculations, for all stationary points, are calculated using a localised, density tted coupled cluster method with the same Dunning basis set (DF-HF/DF-LCCSD(T)-F12a/aug-cc-pVTZ). 91,[94][95][96][101][102][103][104][105][106][107][108][109][110][111] These molecular energies are calculated in the MOLPRO computational chemistry package. 112 All ro-point energies or Gibbs free energies in this study are calculated using the zero-point and free energy corrections acquired from the DFT harmonic frequency calculations.All relative energies (DE) referenced herein in the main body of this manuscript and in all PESs are zero-point corrected. Thi overall computational approach, when applied to the variety of sizes of sCIs and co-reactants in this study, provides both accurate energies and manageable computational cost.This computational approach is used in other studies in the literature too and it effectively describes the chemistry of sCI reactions with TFA & alcohols.61,74 For further verication of the veracity of this computational approach, the theoretical rate constants (k THEO ) for the bimolecular sCI 1 reactions obtained in this study are benchmarked against experimental rate constants (k EXP ) from the literature.
The open source Master Equation Solver for Multi-Energy Well Reactions (MESMER) soware is used to calculate the microcanonical kinetic reaction rates by deriving the individual energygrained master equation (EGME) rate coefficients, which are then subsequently used to generate the overall master equation rate constant (k THEO ) and product branching ratios (G THEO ) of the system over a range of temperatures. 113Computational chemistry rate constants from other studies are also referred to as k THEO constants, even if generated using differing methods.Master equation rate constants are referred to as k ME values only in the ESI † to distinguish them from other theoretical rate constants.][116][117][118][119] These MESMER calculations include a rigid rotor harmonic oscillator approximation, with structures and vibrational frequencies again taken from the DFT calculations, as shown by the example MESMER input in ESI Section S9. † 113 An inverse Laplace transform (ILT) capture rate coefficient of 1 × 10 −10 cm 3 s −1 and an excess reactant concentration of 1 × 10 16 molecule cm −3 are assumed. 76,113The collisional energy transfer factor, hDE down i, was assigned as 300 cm −1 for all minima.This has been shown to be a good approximation for systems involving nitrogen as a gas bath. 76,113,120A 10 cm −1 grain size was routinely used for the EGME calculations, however, for the larger systems, grain sizes are adjusted to reduce computational expense (ESI Sections 1.1-1.4†).
Low mass atom motion within the reaction mechanism is sometimes facilitated by quantum tunnelling, oen accelerating the reaction.The effect of tunneling on the rate constant is determined herein using a non-ab initio tunneling correction called the asymmetrical Eckart constant (k ECKART ). 121This k ECKART is calculated and then integrated directly into the k THEO and G THEO values through the MESMER soware package, unless the reaction barriers are too low to support this methodology (these exceptions are stated in the ESI Section S1.1 †). 76,113,121In calculating the k ECKART function, the probability of transmission through the one-dimensional energy barrier, p(E), needs to be calculated using eqn (1)-( 7): Calculating the probability of transmission involves estimating the barrier width between the reactants and products using: the forward and reverse zero-point corrected energy barriers, DH ‡,0K f and DH ‡,0K r , and the imaginary frequency of the associated TS describing the reaction coordinate, v ‡ .The k ECKART value is then determined using p(E) in eqn (8): The irreversible nal product formation processes are accommodated by the MESMER soware by means of using an innite "sink" approximation.In a system where a particular reaction channel that produces multiple products via a postreaction complex (i.e.reactant / TS / post-reaction complex / product 1 + product 2) herein the formation of the postreaction complex is treated as irreversible by MESMER when determining the k ECKART & G THEO values and it is assumed that ∼100% of this post-reaction complex separates into the nal individual bimolecular products.This approximation reduces the computational intensity of MESMER calculations and introduces an error of less than 1%. 122,123This use of the postreaction complex can be seen in the MESMER example le in the ESI.† In some multi-step reactions, a small number of the intermediate product / nal products fragmentations involve progressing over an asymptotic Morse-like energy curve until the nal separated products are formed, rather than fragmenting via a transition state.When Peltola et al. examined CH 2 OO decomposing into HCO + OH and Blitz et al. studied CH 3 NH breaking down into CH 3 + NH, both used a reverse inverse Laplace transformation (reverse ILT) method in MES-MER to interpret these chemical pathways, 117,124 and both use the reverse ILT method with the experimental rate constant (k EXP ) of the equivalent forward reaction as the pre-exponential factor. 117,124However, as the reaction channels in this study where the reverse ILT method is used have no literature k EXP values instead the dipole-dipole capture limit rate constant, k d- d , (discussed in the next paragraph) are used as the preexponential factors instead.Furthermore, this reverse ILT method requires using the term reverse = true in the activation energy component and assigning both nal products as separate sink structures.
In a few previous studies, sCI reactions with other trace atmospheric species (e.g.4,125 Under such conditions, it is proposed that k THEO values can be more accurately modelled using a collision frequency model, such as the dipole-dipole capture limit, k d-d .The k d-d value is calculated using eqn ( 9) and (10): C is a constant dependent on the anisotropy of the capture potential; m D1 and m D2 are the dipole moments of reactants 1 & 2; and m is the total reduced mass of the system calculated using the mass of reactants 1 & 2 (m 1 & m 2 ). 126The C value adopted can be either isotropic (4.08), adiabatic anisotropic (2.68) and nonadiabatic anisotropic (1.953). 126The isotropic k d-d constant is used in this study, as it is much used in the literature for bimolecular sCI reactions. 61,125requently in this study, the rate determining step is barrierless or k THEO > k d-d (usually where k THEO $ ∼10 −10 ) and so the k d-d is used as the overall rate constant, as it is a more true reection of the tropospheric rate constant.As sCIs are polar in character, the isotropic k d-d value is used over the more traditional collision limit (k COLL ) too, as k d-d is better able to model the effect this polarity has on sCIco-reactant collision frequency.However, the k COLL values for each reaction, along with the adiabatic anisotropic and non-adiabatic anisotropic k d-d values, are all listed in the ESI Section S3 † for comparison.The rate constants generated in this work are analysed using the effective rst-order rate constant, k EFF , which accounts for the concentration of the co-reactant: These k EFF values in this work are calculated using previously reported co-reactant concentrations in example 'important' local environments.This data is displayed in Table 2 in the next section, alongside any known experimental rate constants for the same reactions.Further examples of other local co-reactant abundances in other environments and the effect they have on the k EFF values are listed in ESI Section S4. †

Results & discussion
The bimolecular master equation rate constant (k THEO ), the dipole-dipole capture limit (k d-d ), and the effective rate constant (k EFF ), for the HFO-sCI reactions calculated in this study are found in Table 2.The co-reactant abundances featured in Table 2 are selected from locations where either the pollutant has a high enough concentration to be a potentially signicant sCI sink and/or an area where humans may face a signicant degree exposure of either the pollutant or the potential product of this reaction.These values are then used to generate the k EFF values seen in Table 2 (to view the extended literature survey of the tropospheric abundances of the coreactants, see ESI Section S4.2 †).As most of these values are averages from single locations across a protracted period of time or a generalised upper-limit in certain urban environments from a variety of locations, most measurements occur over a range of temperatures that include 298 K. Either the k THEO value or the k d-d value are in bold to highlight which the rate constant is used to calculate each k EFF value.Any G THEO branching fractions of the two-step HFO-sCI reactions with aldehydes & SO 2 are not displayed in Table 2 and can be instead found in Sections 3.1 & 3.2.

HFO-sCI reaction with formaldehyde (HCHO)
During an ozonolysis event that generates a Criegee intermediate, a carbonyl containing species (aldehydes and ketones) are   very common co-products and as such these carbonyl moieties are co-located with sCIs during formation such that a further reaction between these two proximate species is probable.Formaldehyde (HCHO) is produced from the ozonolysis of a variety of alkenes, notably for this work that includes HFO-1234yf (see Fig. 3), and therefore examining any possible secondary reaction with HCHO is likely vital to understanding the overall tropospheric chemistry of these HFO-sCIs. 4,48,140lso, as HCHO emissions are strong in areas of heavy urban traffic (up to 170 mg m −3 ) and HFOs are used in vehicle coolant units, there is a high probability of emissions overlap between HCHO and HFO-sCIs. 36,128,141,142][145] 3.1.1The sCI 1 + HCHO reaction.All sCI + HCHO reactions in this study undergo an initial 1,3-cycloaddition (TS C ) step to a short-lived heteroozonide (HOZ) intermediate and the HOZ almost instantaneously fragments.The fragmentation is likely rapid because the HOZ is formed with signicant excess energy and the HOZ structure has a high degree of tortional strain.When the HOZ fragments, it can form acid + aldehyde products via two different mechanistic processes: a high energy, 1-step channel or a lower energy, 2-step channel via a hydroxyalkyl formate (HAE) species.The schematics and barrier heights of the 1-step "TS FAc " channel (in blue) and the 2-step "HAE channel" (in red) for sCI 1 + HCHO are seen in Fig. 4. The "TS D " & "TS HAE " labels are used to refer to the respective formation and breakup of the HAE species.
Computational assessments of the sCI 1 + HCHO reaction (Fig. 4) are characterised by a low energy TS C barrier (∼−25 to −20 kJ mol −1 ), which leads to high theoretical rate constants both here (k THEO ∼ 2.79 × 10 −11 cm 3 s −1 ) and in the literature (k THEO ∼ 4.03 × 10 −10 cm 3 s −1 ). 71,146The k THEO value determined here is also similar to the k EXP values measured for sCI 1 + HCHO by Luo et al., (4.11 ± 0.25) × 10 −12 , as well as those determined for studies of the analogous sCI 1 + CH 3 CHO reaction (3.0 × 10 −13 to 1.7 × 10 −12 cm 3 s −1 ). 64,65,127,147he sCI 1 + HCHO reaction produces only one nal product set, HCOOH + HCHO, and of the two HOZ fragmentation mechanism, the TS D barrier of the two-step HAE channel has a lower energy than the TS FAc barrier to the one-step HOZ fragmentation channel (Fig. 4).Both ndings are authenticated by similar results in prior studies. 71,146Obtaining reliable TS D & TS FAc barriers is important because whereas the absolute height of the energy barrier(s) of the rst reaction step dictates the k THEO value, the crucial factor in determining the G THEO branching fractions is the energy difference between the HOZ fragmentation barriers.This study also agrees with other computational studies in that that the second step of the twostep HOZ fragmentation, the breakdown of the HAE, involves a very low TS HAE barrier (Fig. 4). 71,146iven that the HAE is formed with considerably higher energy than the HAE / TS HAE barrier height, the MESMER results discussed here are calculated on the assumption that formation of the HAE is the nal step of that reaction channel and that all the HAE population reacts further to produce their acid + aldehyde products, with negligible populations of HAE remaining.These MESMER simulations yield a very similar result (G THEO ∼ 0.973) to those in which the HCOOH + HCHO are set as the nal products (G THEO ∼ 0.978).As this methodology has been shown to have little effect on the accuracy, this ∼100% HAE / acid + aldehyde conversion principle is applied to the HCHO reactions with sCIs 2-5, for the purpose of reducing computational cost.
3.1.2The HCHO + sCIs 2 & 3 and CF 3 CHO + sCI 1 reactions.The cycloaddition of sCI 2 + HCHO & sCI 3 + HCHO produce different HOZ conformers, HOZs 1 & 2 respectively, but these HOZ conformers can interconvert via an inversion TS HOZ mechanism.This TS HOZ barrier is much lower in energy than both the cycloreversion and HOZ fragmentation mechanisms (Fig. 5).Therefore, these HOZs generally interconvert freely and are treated as a single HOZ species when using MESMER to calculate the product branching fractions.Furthermore, during any of these reactions, (e.g.sCI 2 + HCHO), the HOZ species can fragment to produce any of the other reactant pairs (here sCI 3 + HCHO or sCI 1 + CF 3 CHO), via cycloreversion pathways.The HOZ species produced by these reactions can also fragment (Fig. 5) to produce two different "acid + carbonyl" sets of products, TFA + HCHO or HCOOH + CF 3 CHO, either through a one-step channel (via TS TFA or TS FAc2 respectively) or a twostep HAE channel (via TS D1 & TS HAE1 or TS D2 & TS HAE2 Fig. 3 The general schematic for the ozonolysis of HFO-1234yf producing a short-lived primary ozonide (POZ) followed by one of the pairs of final products, a Criegee intermediate (sCI 5 or anti-CF 3 CFOO) and a carbonyl product (formaldehyde).respectively).HOZ fragmentation can also proceed via "TS ESTER " mechanisms to produce CF 3 OCHO + HCHO.
The relatively low TS C barriers seen for the sCIs 2 & 3 reactions with HCHO in Fig. 5, lead to large k THEO values (2.69 & 24.9 × 10 −13 cm 3 s −1 ), although both reactions are less reactive than sCI 1 + HCHO (see Table 3).Studies of the bimolecular HFO-sCI chemistry are limited, but computational work on sCIs 1-3 reactions with H 2 , CO 2 & CF 3 CH]CH 2 are present in the literature and show similar trends to those identied here (Table 3). 72,73,77or instance, the barriers to the HFO-sCIs + CO 2 reaction follow the same reactivity order (E TS (sCI 2) > E TS (sCI 1) > E TS (sCI 3)) seen for these HCHO reactions. 72Furthermore, the anti-orientated sCI 3 is more reactive with CF 3 CH]CH 2 than the syn-orientated sCI 2, the same result observed for sCIs 2 & 3 + HCHO. 77hile the inclusion of a single -CF 3 group does not signicantly enhance the reactivity of sCIs 2 & 3, the presence of a single -CF 3 substituent in the aldehyde co-reactant increases its reactivity to such a degree that the cycloaddition of sCI 1 + CF 3 CHO is barrierless and the k d-d value (∼6.50 × 10 −10 cm 3 s −1 ) is used as the rate constant.This could be due to a hypothetical inductive effect caused simply by adding any alkyl substituent to the aldehyde.But, despite the increased number of alkyl groups on CH 3 CHO & (CH 3 ) 2 CO the experimental literature suggests that their reactions with sCI 1 (k EXP ∼ 10 −1 -10 −12 cm 3 s −1 ) do not show the enhanced reactivity seen for the sCI 1 + CF 3 CHO reaction. 65,71,147However, the role of uorine in -CF 3 substituents is explored in an experimental study by Taatjes and co-workers and they show that sCI 1 + (CF 3 ) 2 COO has a larger k EXP (∼3.0 ± 0.3 × 10 −11 cm 3 s −1 ) compared to the k EXP of sCI 1 + (CH 3 ) 2 COO (∼2.3 ± 0.3 × 10 −13 cm 3 s −1 ). 65This conrms the high reactivity seen for sCI 1 + CF 3 CHO must emerge from the inductive impact that the uorenes in the -CF 3 substituents have on the aldehyde reactivity.
Both HCHO + sCI 2 and HCHO + sCI 3 (Fig. 6) produce high yields of the sCI 1 + CF 3 CHO (∼0.91-0.94),due to the barrierless nature of their cycloreversion mechanisms, with small contributions from G TFA+HCHO (∼0.012-0.016)and G HCOOH+CF 3 CHO (∼0.052-0.068).In contrast, during the sCI 1 + CF 3 CHO reaction, the production of HCHO + sCIs 2 & 3 gives very low branching ratios (<0.001) because the energy barriers for producing HCHO + sCIs 2 & 3 are much larger than other HOZ fragmentation pathways.Consequently, the overall G THEO of sCI 1 + CF 3 CHO is dominated by G HCOOH+CF 3 CHO ∼ 0.823 with most of this contribution coming from the two-step HAE pathways (G TS D1 ∼ 0.746) with a smaller contribution from the one-step channel (G TS FAc ∼ 0.077).The secondary product yield, G TFA+HCHO ∼ 0.177, also has most of its contribution emerge from the two-step HAE pathways, G TS D2 &TS D3 ∼ 0.166.The TS ESTER barrier is too high for the ester formation to have anything but a negligible yield for all three reactions.
While no experimental or theoretical product ratios for sCI 1 + CF 3 CHO are available in the literature, some of the analysis of the similar sCI 1 + CH 3 CHO & PhCHO reactions does include computational calculations of HOZ fragmentation energy barriers.Examination of both the sCI 1 + CH 3 CHO & PhCHO systems show patterns consistent with this work, including: that the formation of smaller acid nal products was preferred over larger acid products, that the 2-step HAE channel was the favoured formation mechanism for these acids, and that ester formation was negligible.
3.1.3The HCHO + sCIs 4 & 5 and CF 3 CFO + sCI 1 reactions.sCI 5 + HCHO exhibits has a low energy TS C barrier (∼−28.9kJ mol −1 ) whereas the cycloaddition of sCI 4 + HCHO is barrierless.This demonstrates that sCI 4 is more reactive than sCI 5 and that both these HFO-sCIs are signicantly more reactive than sCIs 1-3.sCIs with larger substituents (e.g.-CH 3 or -CF 3 ) in the syn-position have been shown to decrease sCI reactivity, however here it is observed that sCI 4, which has a syn-orientated  ), it is clear that the inductive effect of the -CF 3 substituent in the CF 3 CFO is largely cancelled out by the larger deactivating effect of the -F substituent that is also present in CF 3 CFO.sCI 4 & sCI 5 reactions with HCHO are both integrated into the same reaction system as sCI 1 + CF 3 CFO and therefore all sCI + aldehyde reactant pairs can also be produced during HOZ fragmentation in this overall reaction system.The G THEO for both the sCIs 4 & 5 + HCHO reactions are dominated by yields of sCI 1 + CF 3 CFO (∼0.84) with the remainder being the G HCOOH+CF 3 CFO contribution (∼0.16).In contrast, the sCI 1 + CF 3 CFO reaction has only negligible yields for sCIs 4 & 5 + HCHO (<0.0001) because their cycloreversion TS C barriers are >100 kJ mol −1 higher than the lower energy HOZ fragmentation barriers.
The primary product of the sCI 1 + CF 3 CFO reaction is HCOOH + CF 3 CFO (G HCOOH+CF 3 CFO > 0.99), with the direct Htransfer TS FAc channel producing almost all the formic acid yield.The higher barriers to the HAE pathway compared to the direct H-transfer channel seen here for the sCI 1 + CF 3 CFO reaction system (see ESI Section S5 †) is uncommon both in this study and in the literature. 71,146However, in a literature analysis of anti-PhCHOO + PhCHO, the direct H-transfer mechanism also has a higher barrier (−19.2 kJ mol −1 ) to phenylacetic acid formation than the two step HAE channel (−17.6 kJ mol −1 ). 146s noted in Table 4, ester formation continues to be negligible for all sCI + aldehyde reactions.
3.1.4Overview of sCI + aldehyde reactions.The key features of the sCI + aldehyde reactions are that the chemistry can be broadly divided into two steps: the cycloaddition into an HOZ intermediate, and the fragmentation of that intermediate into various products.The HFO-sCI + aldehyde reactions have large rate constants (k THEO ∼ 10 −13 -10 −10 cm 3 s −1 ) as the cycloaddition is either barrierless or proceeds via low energy TS C barriers.The inclusion of -CF 3 substituents in the syn-position produces a reductive impact on the COO group of the sCI, as shown by the smaller rate constant of sCI 2 compared to the unsubstituted sCI 1 or the anti-substituted sCI 3.However, an -F substituent has an inductive effect on the sCI, particularly in the anti-position, as shown by the high reactivity of sCI 4. Considering the factors given above the reactivity of the HFO-sCIs studied here follows the following trend: k THEO (sCI 2) < k THEO (sCI 3) < k THEO (sCI 1) < k THEO (sCI 5) < k d-d (sCI 4).The role these -H, -F or -CF 3 substituents have on the aldehyde reactivity trend, k THEO (CF 3 CFO) < k THEO (HCHO) < k THEO (CF 3 CHO), is the opposite to that seen for the sCIs, k THEO (CF 3 CFOO) < k THEO (CH 2 OO) < k THEO (CF 3 CHOO).In the HOZ fragmentation stage of the sCI 1 + HCHO/CF 3 CHO/CF 3 CFO reactions, the yield is dominated by formation of formic acid + an aldehyde, with sCI 1 + CF 3 CHO also producing sizable product branching fractions for HCHO + CF 3 -COOH (G THEO ∼ 0.16).However, for sCIs 2-5 + HCHO reactions, the dominant product is sCI 1 + CF 3 CHO/CF 3 CFO (G THEO ∼ 0.80-0.99),due to the low barrier to formation of this product pair.
Fig. 6 The fractional populations of reactants and products for the sCI 3 + HCHO reaction over time (the product branching fractions for the barrierless cycloreversion channels to sCI 1 + CF 3 CHO are calculated using the same reverse ILT method described in the Methods section. 117,124Excess HCHO reagent concentration is ∼1.0 × 10 16 molec.per cm 3 .For other details on the conditions of the reaction see ESI Section S11 †).The product branching fractions of the barrierless cycloreversion channels that leads to either sCI 1 + CF 3 CHO or sCI 4 + HCHO are calculated using the same reverse ILT method described in Method section. 117,124.2 HFO-sCIs reactions with sulphur dioxide (SO 2 ) 3][154][155][156][157][158] This reaction is considered responsible for between ∼1% of the loss of syn-CH 3 CHOO in rural environments to ∼22% of the loss of "syn-CH 3 -anti-(trans-CH 2 ]CH)-COO" in boreal forests. 159Furthermore, the importance of this reaction is increased because the dominant nal product of this reaction is SO 3 , which is a source of organosulphates and is involved in the formation and growth of secondary organic aerosol (SOA) nuclei. 84,85,1602][163][164][165] Also, depending on the aerosol type, they can absorb and scatter solar radiation and therefore inuence climate temperatures. 161,164,165.2.1The sCI 1 + SO 2 reaction.All HFO-sCIs in this study, as well as many sCIs studied in the literature, react with SO 2 through a general two-step reaction scheme, via a short-lived 5member ozonide ring. 84,159These pathways produce cyclic intermediates, which are oen called heteroozonides too, but it can also be referred to as a secondary ozonide (SOZ) and to prevent confusion with the HCHO reaction, that SOZ label is used in these SO 2 reactions. 84The SOZs produced from sCI 1 + SO 2 are generated in either an exo-(SOZ 1) or an endo-orientation (SOZ 2), depending on the orientation of the SO 2 (Fig. 7) in the cycloaddition transition state, TS EXO and TS ENDO respectively.As also observed in the experimental and theoretical literature, the sCI 1 + SO 2 reaction is highly reactive and so when both TS ENDO and TS EXO pathways were evaluated it was found that both reactions were barrierless. 63,64,84,85,153,154,166This meant that therefore means that the k d-d limit (7.25 × 10 −10 cm 3 s −1 ) is adopted as the rate constant for the sCI 1 + SO 2 reaction.
The SOZ interconversion via the inversion of the S atom is known to require >180 kJ mol −1 excess energy, which exceeds barriers to both cycloreversion (via TS EXO & TS ENDO ) and SOZ fragmentation (Fig. 8). 84This would impede SOZ interconversion but, as the R 1 and R 2 substituents in sCI 1 are identical, SOZs 1 & 2 can freely interconvert over low energy pseudorotation TS SOZ barriers (see Fig. 8), which this means that SOZs 1 & 2 behaves as a single SOZ species.
The next stage of the sCI 1 + SO 2 reaction is the fragmentation of the SOZ to produce either SO 3 + HCHO via the TSSO 3 channel or SO 2 + HCOOH via the TS acid channel.All SOZ fragmentation channels studied here are single-step mechanisms and the acid-producing mechanisms from each HFO-sCI + SO 2 reaction system exclusively produces a single acid species.Kuwett et al. have performed a comprehensive analysis of both SO 2 reactions with sCIs 1 & (CH 3 ) 2 COO comprising a numerous range of major and minor SOZ fragmentation pathways, including a homolytic O-O bond ssion to produce a 1,5-(bis) oxy diradical intermediate product. 84While the self-reaction of this diradical intermediate produces important products, such as SOCO rings, sulphurous anhydrides or 1,3-(bis)oxy diradicals, its yield is so minor that 1,5-(bis)oxy diradical pathways are not explored for sCIs 1-5 + SO 2 here. 84he nal product branching fraction for sCI 1 + SO 2 is dominated by SO 3 + HCHO (G SO 3 ∼ 0.966) with a subordinate yield HCOOH + SO 2 (G acid ∼ 0.034) because the TSSO 3 1 & 2 barriers (Fig. 8) are signicantly lower in energy than the TS acid 1 & 2 structures.This dominance of SO 3 + CF 3 CHO in the G THEO value is in agreement with that calculated in the literature (G SO 3 ∼ 0.973-0.984). 84.   reverting to the initial reactants.This likewise applies to the mono-substituted sCI 3 + SO 2 , which also produces endo and exo conformers SOZs 3 & 4.During some sCI + SO 2 reactions, the SOZs produced can interconvert via a pseudorotation TS SOZ mechanism and circumvent the higher barrier via S atom inversion, but this can only transpire during reactions with sCIs with the same R 1 & R 2 substituents (e.g.sCI 1 or (CH 3 ) 2 COO).Instead, when SOZ 1 surmounts the pseudorotation TS SOZ 1 barrier it isomerises into the SOZ 3 conformer and the TS SOZ 2 pseudorotation makes possible the SOZ 2 4 SOZ 4 interconversion (Fig. 9).This also means that SO 2 reactions with sCIs 2 & 3 are part of the same integrated PES, through the TS SOZ 1 & 2 isomerisation channels (Fig. 9 and 10).This phenomena of 2 × non-interchangeable sets of SOZs also applies to SO 2 reactions with sCIs 4 & 5 and, according to the literature, SO 2 reactions with syn-& anti-CH 3 CHOO. 84uring the SOZ fragmentation of either sCI 2 + SO 2 & sCI 3 + SO 2 reactions, the reverse sCI + SO 2 reactant pairs can be produced through cycloreversion, as both reactions are part of the same potential energy surface.The SOZ fragmentation can also proceed through: TSSO  reactions are part of the same integrated reaction system and the phenomenon of 2 sets of non-interchangeable SOZs also applies to this system (Fig. 11) but this has little impact on SOZ fragmentation yields (Table 5).The only other SOZ fragmentation calculated are TSSO 3 & TS ESTER channels, as an SOZ / acid channel cannot transpire from SO 2 reactions with sCIs 4 & 5 (Fig. 11).The cycloreversion channel has a negligible yield in the sCI 5 + SO 2 reaction and a minor role in sCI 4 + SO 2 reaction (Table 5).While TS ESTER 1 & 2 are not extremely high in energy, the ester yield is still small (G SO 2 + CF 3 OCHO ∼ 0.01).The low TSSO 3 barriers and the minor yields of other products means that the predominant products from SOZ fragmentation are SO 3 + CF 3 CHO (∼0.986 & 0.996).

Overview of sCI + SO 2 reactions.
There are several key results from this analysis of the HFO-sCI + SO 2 reactions, including that all these reactions are initiated via a cycloaddition generating a short-lived SOZ ring, that then fragmented.sCI 2 + SO 2 was the only reaction without a barrierless cycloaddition step, which conrms that the syn-orientated sCI 2 is less reactive than the anti-orientated sCI 3.This contributes to producing an HFO-sCI + SO 2 general reactivity trend of: Lastly, the most dominant products of these reactions were SO 3 + aldehydes (G SO 3 +aldehyde ∼ 0.96-0.99)with only small yields of either acids, esters or other sCIs.This implies that emissions of the HFO-sCIs will lead to consequentially higher local concentrations of SO 3 and subsequently higher concentrations of atmospheric aerosols.

HFO-sCIs reactions with organic and inorganic acids
Both organic acids such as TFA & carboxylic acids, and inorganic acids including HNO 3 & HCl, are prevalent in various tropospheric environments (see ESI Section S4.2 † for more details) and sCIs is their capacity to deplete these acids. 49,59,77,130,138,139,167.3.1 HFO-sCI reactions with HNO 3 .The sCI + HNO 3 reaction almost exclusively forms a nitrooxyalkyl hydroperoxide (NAHP) species by reacting via a 1,4-insertion mechanism, such as that seen in Fig. 12.In previous studies of sCI 1 + HNO 3 the chemistry of NAHP fragmentation was also calculated, but as this process occurs over a long timescale, NAHP breakdown is beyond the scope of this study.In contrast, the HOZ/SOZ fragmentation in the sCI + HCHO or SO 2 reactions occurs almost instantaneously.168,169 NAHP can also be produced via a 1,2insertion mechanism, but it has been excluded from this study as this barrier is ∼30 kJ mol −1 greater in energy than the 1,4insertion barrier.169 None of the 1,4-insertion processes for HFO-sCI + HNO 3 reactions studied here are barrierless, but the high reactivity of sCI 1 + HNO 3 means that the k THEO value (8.21 × 10 −9 cm 3 s −1 ) is so great that the k d-d limit (8.45 × 10 −10 cm 3 s −1 ) is used as the rate constant instead.This k d-d limit is also similar to the k EXP [295 K] value (5.4 × 10 −10 cm 3 s −1 ) from the experimental Foreman et al. study.59 The trend in reactivity indicated by these HFO-sCI + HNO 3.3.2HFO-sCI reactions with TFA.In general, sCIs 1-5 react with TFA to produce a hydroperoxy ester (HPE) species through a barrierless minimum energy pathway (e.g.Fig. 13).The product branching fractions of the barrierless cycloreversion channels that lead to SO 2 + sCIs 3, 4 or 5 are calculated using the same reverse ILT method described in Method section.117,124 barrierless minimum energy pathway calculated here for sCI 1 + TFA is similar to literature schemes and its k d-d capture limit (7.54 × 10 −10 cm 3 s −1 ) is found to be in close proximity to the k EXP rate constant (3.4 × 10 −10 cm 3 s −1 ).61 Altering the sCI substituents does not alter the barrierless nature of the HFO-sCI + TFA reaction proles studied here (see ESI Section S8.5 †).For this reason, all the sCIs 1-5 + TFA reactions use the k d-d values as rate constants (3.98-7.54× 10 −10 cm 3 s −1 ).

HFO-sCI reactions with water monomer and dimer
Due to the high tropospheric concentration of water vapour, water acts as the most prevalent bimolecular sink for sCIs, and therefore examining the chemistry of sCIs 1-5 reactions with H 2 O & (H 2 O) 2 is essential.sCIs 1-5 reactions with H 2 O proceed through only two reaction channels (TSH 2 O 1 and TSH 2 O 2), whereas sCIs 1-5 reactions with (H 2 O) 2 proceed via four pathways: TS(H 2 O) 2 1, TS(H 2 O) 2 2, TS(H 2 O) 2 3 and TS(H 2 O) 2 4 (see ESI Section S8.9 †).One commonality between these reactions is that the primary product is an a-hydroxy-hydroperoxide (HHP), which is implicated in both forest damage and generating crucial secondary products (inc.8][179] Under standard conditions, photolysis, deposition, and reaction with hydroxyl radicals are usually the main sinks for HHPs, but the excess internal energy within the HHP at the point of formation increases the signicance of thermal unimolecular HHP decay. 131,177The primary unimolecular decay mechanism for hydroxymethyl hydroperoxide (HMHP), the product of CH 2 OO + H 2 O, is the elimination of H 2 O 2 to produce formaldehyde, with some yields from minor channels that produce either H 2 O + HCOOH or OH + OCH 2 OH. that while ∼55% of the HMHP remained stable, this excess energy led to ∼40% of the HMHP fragmenting to produce H 2 O 2 and formaldehyde, both of which are implicated in biogenic degradation. 131Halogenated HHPs produced from sCIs 2-5 are also likely to be produced with excess energy, and therefore exploring the breakdown of these products may be useful in the future.
The k THEO value produced for sCI 1 + H 2 O (1.18 × 10 −16 cm 3 s −1 ) is within the range of literature k EXP values (0.25-13 × 10 −16 cm 3 s −1 ), and is particularly close to the rate constant found by Sheps et al. (k EXP [293 K, 50 Torr] ∼ 2.4 × 10 −16 cm 3 s −1 ). 64,78,131,182,183Furthermore, both the k THEO value in this study and the k EXP value found in the literature agree that sCI 1 + H 2 O reactions slower than sCI 1 reactions with HNO 3 , SO 2 or TFA. 59,61,63,64,85,153,154,166he computational rate constants of the sCIs 4 & 5 + H 2 O reactions were found to have a negative temperature dependence, where 200 K < T < 400 K, whereas sCIs 1, 2 & 3 + H 2 O all have a positive temperature dependence within the same range (see ESI Section 1.2 †).The general reactivity trend for these sCI + H 2 O reactions continues the trend observed throughout most of this study k THEO (sCI 2) < k THEO (sCI 3) # k THEO (sCI 1) < k THEO (sCI 5) < k THEO (sCI 4).The theoretical trend of sCI 4 being more reactive than sCI 5 is partially obscured by rate constants being near to or exceeding the k d-d limit, however as neither the k THEO values for sCI 4 + H 2 O (5.13 × 10 −12 cm 3 s −1 ) or sCI 5 + H 2 O (2.27 × 10 −13 cm 3 s −1 ) reach the k d-d limit, sCI + H 2 O would provide an experimental case study to verify if k THEO (sCI 4) does exceed k THEO (sCI 5) in bimolecular reactions.
In contrast, HFO-sCI + (H 2 O) 2 reactions are faster as a result of sCI 5 + (H 2 O) 2 having a lower reaction barrier than sCI 5 + H 2 O (Fig. 14).58]184 In line with previous HFO-sCI reactions, sCI 2 + (H 2 O) 2 has the lowest k THEO value (2.71 × 10 −12 cm 3 s −1 ) but unusually the   12 The potential energy surfaces for the sCI 1 + HNO 3 (black) and sCI 4 + HNO 3 (red) reactions.Energies are relative to raw reactants.Fig. 13 The barrierless minimum energy pathway for the sCI 4 + TFA reaction.Energies displayed here are derived from using the B3LYP/ aug-cc-pVTZ approach.Energies displayed here are relative to raw reactants.
the rate constants instead.Overall the trend for HFO-sCI + (H 2 O) 2 reactions: k THEO (sCI 2) < k THEO (sCI 1) z k THEO (sCI Criegee intermediates exhibit a mixed biradical and zwitterionic character and it has been shown that the more reactive zwitterionic dominated COO moieties have the larger ratio of OO and CO bond lengths, R OO /R CO or q. 68,74 Given these observations, sCIs 2 & 3 (q = 1.064 & 1.071) appear to have a more biradical COO moiety than sCI 1 (q = 1.078), whereas sCIs 5 & 4 (q = 1.101 & 1.114) are more zwitterionic.Anglada et al. noted in a computational study of sCI + H 2 O reactions that the degree of this zwitterionic nature of the carbonyl oxide moiety, and consequentially it's reactivity, can be tuned by modication of the substituents. 68Furthermore, in a computational study of sCI + alcohol reactions it is noted that the electron-withdrawing nature of F substituents in syn-& anti-FCHOO is likely to increase the unsaturated nature of the zwitterionic carbonyl oxide.This then generates a more electropositive central C atom, which is more vulnerable to nucleophilic attack by the alcohol's electronegative O atom.This study shows the same positive relationship between electron-withdrawing groups, zwitterionic COO moiety and high reactivity seen in the literature, as shown by the fact that increases in q values in the HFO-sCI series mostly correlates with the growth in rate constant: k THEO (sCI 2) < k THEO (sCI 3) z k THEO (sCI 1) < k THEO (sCI 5) < k THEO (sCI 4).

Impact of heteroatom tuning on HFO-sCI reactions
It has been suggested that for sCI + H 2 X (X = O, Si, Se & Te) reactions, as the central X atom in the H 2 X co-reactant is substituted with larger group XVI elements, the rate constant increases. 87In this section, the computational analysis of the sCIs 1-5 + H 2 S reactions is compared with sCIs 1-5 reactions with H 2 O (& MeOH) to determine if these heteroatom tuning trends apply to HFO-sCIs too.To see if the effects observed from heteroatom tuning group XVI-centred co-reactants also apply to group XVII-centred co-reactants, analysis of HFO-sCI + HF & HCl reactions is also included in this section.
3.5.1 HFO-sCI reactions with hydrogen sulphide (H 2 S).The HFO-sCI + H 2 S reactions proceed through two hydrogen abstraction transition states to produce hydrosulphide alkyl hydroperoxide, following a reaction process similar to that observed for most sCI + H 2 S reactions studied in the computational and experimental literature. 87,136,185The rate constant for sCI 1 + H 2 S determined here (k THEO ∼ 7.06 × 10 −15 cm 3 s −1 ) is larger than that for sCI 1 + H 2 O in this study (k THEO ∼ 1.18 × 10 −16 cm 3 s −1 ).This difference in reactivity is substantiated by the fact that a literature k EXP value for sCI 1 + H 2 S observed by Smith et al. (1.7 × 10 −13 cm 3 s −1 ) is much larger than the k EXP range observed in the literature for sCI 1 + H 2 O (0.25-13 × 10 −16 cm 3 s −1 ). 64,78,131,136,182,183When analysing sCIs 2 & 3 reactions with H 2 S each also show much larger k THEO values than their reactions with H 2 O, as observed in Fig. 15.Therefore, the theoretical results of H 2 S reactions with sCIs 1-3 help to substantiate the general trend observed by Kumar et al. and the experimental literature that, if the central X atomic element of H 2 X is substituted with an element further down group XVI on the periodic table, then the k THEO value for the sCI + H 2 X reaction increases. 87ne other point that may be noteworthy is that sCI 3 & sCI 1 exhibit similar rate constants throughout all three reactions featured in Fig. 15, as well as throughout much of this study.In much of the literature the non-uorinated equivalent to sCI 3, anti-CH 3 CHOO, exhibits a higher bimolecular rate constant to sCI 1 in reactions with H 2 O, MeOH & SO 2 . 61,74,79,148,151This may well be due to the presence of hyperconjugative a-H atoms in the anti-CH 3 substituent and that a-F atoms on the anti-CF 3 group rather than hyperconjugative a-H atoms reduces the inductive impact the substituent has on sCI 3 reactivity. 56n contrast to the patterns seen for the sCIs 1-3 + H 2 S reactions, the rate constants observed for the H 2 S reactions with sCIs 4 & 5 are both smaller than those of the equivalent H 2 O reactions (Fig. 15).To determine whether this stagnation of the sCI + H 2 S rate constant was due to increased overall steric bulk,  k THEO values were for the HFO-sCI reactions with the bulkier MeOH (which also proceeds via a hydrogen abstraction mechanism) were determined.However, no such stagnation in the rate constants of sCI 4 & 5 + MeOH reactions emerges (Fig. 15).
3.5.2HFO-sCIs reactions with HF & HCl.The Gr17-centred HX compounds, HF & HCl, react with sCIs via similar hydrogen abstraction mechanisms as the sCI + H 2 X reactions, which means the heteroatom tuning trends seen for the HFO-sCI + H 2 X reactions may be replicated. 59,186For instance, sCI 1 + HCl has a considerably lower barrier (Fig. 16) to reaction than sCI 1 + HF, following the trend that the co-reactant with larger heteroatoms (HCl) will have greater reactivity in reactions with sCIs 1-3.The resulting k THEO value for sCI 1 + HF (3.32 × 10 −13 cm 3 s −1 ) is smaller than both the k THEO value (4.70 × 10 −10 cm 3 s −1 ) and its literature k EXP rate constant (1.7 × 10 −13 cm 3 s −1 ) for sCI 1 + HCl. 59All sCIs 1-3 reactions with HCl yield a rate constant that is ∼10 3 -10 4 cm 3 s −1 larger than the equivalent HFO-sCI + HF reaction, much larger than the equivalent reactivity difference of sCIs 1-3 + H 2 S over that of H 2 O (∼10 1 -10 2 cm 3 s −1 ).
Fig. 16 displays that the uorinated sCI 4 has much lower reaction barrier with HF than the unsubstituted sCI 1 has, demonstrating the continued activating effect that the -F substituent has on the sCI reactivity.The inductive impact of the -F substituent results in sCI 4 and sCI 5 having substantially larger rate constants (1.19 × 10 −11 & 3.66 × 10 −13 cm 3 s −1 ) compared to the sCI 1 + HF reaction.In contrast, the energy barrier of the sCI 4 + HCl reaction does not decrease compared to that of sCI 1 + HCl (Fig. 16), replicating the observation that heteroatom tuning does not have an inductive effect on reactions involving sCIs 4 & 5.In fact, -F substituents in sCIs 4 & 5 in combination with the heteroatom tuning of the co-reactant leads not just to the same stagnation effect seen for sCIs 4 & 5 + H 2 S reactions, but instead to a reductive effect on reactivity for the sCIs 4 & 5 + HCl reactions, compared to sCIs 4 & 5 + HF.This reductive effect is so clear that the rate constants for HCl reactions with sCI 4 (1.64 × 10 −10 cm 3 s −1 ) and sCI 5 (1.13 × 10 −10 cm 3 s −1 ) are both smaller than that of the equivalent sCI 1 reaction for the only time in this study.The cause of these stagnation/reductive effects can potentially be explained in terms of the repulsion between the sCI substituent groups and increased steric interactions, although a deeper exploration into such factors would probably require experimental analysis and verication. 56 Atmospheric implications of HFO-sCIs bimolecular sinks HFO emissions, and subsequently HFO-sCIs, are projected to become much more abundant as the use of HFO become increasingly widespread in refrigerators, insulation and vehicle cooling units, as is being seen in China, the USA and the EU.1,27,28 The chemistry of sCIs and other atmospheric species are typically included in atmospheric models in sets of like chemistries or taxonomic groups, where their reactivity can be estimated on the basis of their structure, and this work suggest sCIs 2-5 can be grouped into any such wider sCI classication system.40,70,187,188 Similar taxonomic groupings are already used in atmospheric models to estimate the chemistry of large groups of molecules and therefore reduce the computational processing power required to run simulations.189 However, integrating HFO-sCIs into a wider taxonomic sCI framework faces particular difficulties, principally because of the differing effects non-halogenated groups (-CH 3 and -H) and halogenated groups (-CF 3 and -F) have on the conjugation, hyperconjugation and steric interactions the sCI.56,[72][73][74][75] One example of this is the difficulty of including the stagnation/reduction in reactivity of sCIs 4 & 5 seen during reactions with H 2 S & HCl, a phenomenon not exhibited by other sCIs studied here or in the prior literature.59,64,78,87,131,136,182,183 This shows that systemby-system computational analysis, or indeed ultimately experimental analysis is still vital for understanding the chemistry of many bimolecular sCI reactions.
The effective rate constants, k EFF , a product of the rate constant and the atmospheric abundance of the co-reactant, is used here as a primitive system of measuring the potential tropospheric impact of each bimolecular HFO-sCI reaction studied (Table 1).
The sCI reactions with water are the dominant bimolecular sink for the HFO-sCIs due to the abundance water Fig. 16 The potential energy surfaces for selected sCI reactions with HF and HCl.On the potential energy surfaces on the left, feature sCI 1 + HF, in black, and sCI 4 + HF, in red.On the potential energy surfaces on the right, feature sCI 1 + HCl, in black, and sCI 4 + HCl, in red.Energies are relative to raw reactants.
,170 HFO-sCI reactions with aldehydes are quite uncompetitive compared to the reactions with H 2 O & (H 2 O) 2 but the products of their reactions include organouoride aldehydes and TFA, both known to contribute to respiratory problems. 192Furthermore the sCIs 2-5 reactions with HCHO lead to high yields of sCI 1 sometimes with a large excess of energy, that easily leads to CI fragmentation and therefore contributes to the tropospheric population of the 'atmospheric detergent' OH radical. 70,82he sCI + SO 2 reaction is known to act as a sink for ∼20% of some sCIs in boreal forest, however the high uptake of the reactions with H 2 O & (H 2 O) 2 means that they do not contribute hugely to the overall depletion of sCIs 2-5. 159][195][196] TFA may have a disproportionate impact on HFO-sCI bimolecular chemistry because both HFO-sCIs and TFA emerge from the breakdown of HFOs and so they may see a large population overlap with HFO-sCIs.The large rate constants from sCI reactions with TFA, HNO 3 and HCl found here indicate a high reactivity between sCIs and other organic and inorganic acids.However, even in seaside areas, where HCl concentrations are elevated, the contribution that sCI 1 could make the minor and the role of sCIs 4 & 5 would be even more reduced, due to the high k EFF values of the sCI + H 2 O & (H 2 O) 2 reactions.The non-water reactions (e.g.][199] Aer collisional stabilisation, sCIs are less energetic so fewer unimolecular fragmentation mechanisms have energy barriers low enough to compete with bimolecular reactions as sinks for sCIs under tropospheric temperature and pressure conditions. 42ccording to Guidry et al., the central mechanism for the fragmentation of both syn-& anti-CF 3 CHOO is via a 1,3-ring closure to produce a dioxirane ring, both of which produce energy barriers (∼80 kJ mol −1 ) higher than anti-CH 3 CHOO (∼70 kJ mol −1 ) and more similar to those of CH 2 OO (70-100 kJ mol −1 ). 200The unimolecular rate constant (k UNI ) of Table 6 A tabulation of the effective rate constants (k EFF ) for the HFO-sCI + co-reactant reactions, determined in this study using the product of the theoretical rate constant and the co-reactant abundance.This table shows a primitive calculation of the atmospheric impact of each reaction.If the k EFF value is written in italics, the dipole-dipole capture limit (k d-d ) was used to calculate the k EFF value because the master equation rate constant (k THEO ) exceeded the k d-d limit or because the reaction was barrierless  66,200 Therefore, the k UNI for CH 2 OO (∼0.3 s −1 ) is used here instead as it may be the closest analogue to that of syn-& anti-CF 3 CHOO. 42,70On the basis of these premises, while the k UNI value used here is greater than many of the k EFF value for syn-& anti-CF 3 CHOO found in Table 6, the k EFF value for other reactions, particularly syn-& anti-CF 3 CHOO + (H 2 O) 2 , are greater still.This indicates that bimolecular reactions can compete with unimolecular decomposition.

Co
No equivalent k UNI analogue could be found for syn-& anti-CF 3 CFOO especially considering the role of the -F substituent in CI fragmentation in isolation is also challenging.One study by Ljubić and Sabljić calculated that the anti-F substituent induced a lower anti-FCHOO fragmentation barrier (TS UNI ∼ 47.3 kJ mol −1 ), whereas syn-FCHOO saw no such effect (TS UNI ∼ 83.5 kJ mol −1 ). 201It is quite possible that the anti-F substituent could therefore theoretically induce a higher k UNI for syn-CF 3 -CFOO, but this might be offset by the lack of any tunnelling effect and the steric hinderance of the syn-CF 3 substituent.Nevertheless, no computational or experimental literature exists of the syn-& anti-CF 3 CFOO unimolecular fragmentation reactions, and an assessment of these reactions has been not been deliberated here, due to the need for a more in-depth analysis of these reactions than the kind of provisional evaluation that would be covered in this study.

Conclusion
It has been shown that the individual steric and electronic features of HFO-sCIs impact upon their bimolecular chemistry in a way that transcends the choice of tropospheric co-reactant species.The chemistry of sCI 2 (syn-CF 3 CHOO) & sCI 3 (anti-CF 3 CHOO) have consistency with existing reactivity trends in the literature, particularly that sCIs with a syn-orientated sterically bulky group are signicantly less reactive than anti-orientated sCIs.However, sCI 3 is distinctive because it exhibits similar rate constants to the unsubstituted sCI 1 (CH 2 OO), whereas sCIs with similar orientations like anti-CH 3 CHOO yields larger rate constants.The literature indicates that the presence of hyperconjugative a-H atoms in the anti-CH 3 group has an inductive impact on the sCI, but here the presence of a-F atoms in the anti-CF 3 group appears to reverse this activating effect.
The especially large bimolecular rate constants for sCI 4 (syn-CF 3 CFOO) & sCI 5 (anti-CF 3 CFOO) emerge from the inductive -F substituents they possess.Moreover, the inductive impact of the -F substituent is greater in the case of sCI 4 because of the antiposition of the -F substituent, and so the k THEO (sCI 4) exceeds the k THEO (sCI 5).However, with some notable exceptions, the HFO-sCIs studied here follow the reactivity series: k THEO (sCI 2) < k THEO (sCI 3) ∼ k THEO (sCI 1) < k THEO (sCI 5) < k THEO (sCI 4).
One exceptionally distinct observation in this study is that the overall sCI reactivity series is inuenced by the choice of coreactant, as shown when contrasting the reactivity of sCIs 1-5 reactions with different group-XVI and group-XVII centred co-reactants.When sCI + H 2 O & HF are compared with sCI + H 2 S & HCl respectively, sCIs 1-3 reactions with H 2 S & HCl produced comparatively larger rate constants, in line with previous literature observations. 87However, this does not always apply as the -F substituent appear to obstruct the large S and Cl atoms from reaching the reaction site during sCIs 4 & 5 reactions with H 2 S and HCl and this leads to a smaller rate constant.
This work demonstrates that for all sCIs considered, reaction with H 2 O or (H 2 O) 2 is the biggest tropospheric sink of HFO-sCIs, due primarily to the natural abundance of water vapour.4,170 While bimolecular reaction with other trace pollutants are less likely to be sinks for HFO-sCIs, sCIs 2-5 all have a signicant capacity to react with tropospheric species such as MeOH, TFA & HNO 3 and yield other functionalised hydroperoxides.
The uncommon nature of these HFO-sCIs is also demonstrated in the multi-step reactions with HCHO, where sCIs 2-5 react (via a short-lived HOZ) to produce another sCI, a highly energised sCI 1, a product not commonly found in other studies of sCI + aldehyde reactions. 65,71,147A similar full mechanistic analysis of the reactions using a two-step HFO-sCI + SO 2 systems show that these reactions have very high rate constants and all produce high yields of SO 3 + R 1 R 2 CO (0.90-0.99).This production of SO 3 would also lead to H 2 SO 4 formation and aerosol nucleation.Furthermore, HFO-sCI reactions with both HCHO & SO 2 systems are found to produce CF 3 CHO or CF 3 CFO, which are known to contribute to respiratory problems.
In summary, HFO-sCIs are shown to have very unique chemistry, because the -F substituent has a large activation affect on the COO group, whereas the a-F atoms in the CF 3 group are mildly deactivating.If any sCI taxonomic classication is established, sCIs 2 & 3 could be categorised as part of syn-sCI and anti-sCI groups respectively, whereas sCIs 4 & 5 could be categorised in a separate "electronegative substituent" taxonomic group.

Paper
Environmental Science: Atmospheres

Fig. 4
Fig. 4 The potential energy surface for the sCI 1 + HCHO reaction.The lowest energy transition state for each HOZ fragmentation pathway is displayed.The 1-step direct transfer channel via the TS FAc barrier is in blue and a 2-step hydroxyalkyl ester (HAE) channel, via the TS D & TS HAE barriers, is red.Energies are relative to the raw reactants.

Fig. 5
Fig.5The collective potential energy surface for sCIs 2 & 3 + HCHO and sCI 1 + CF 3 CHO reactions with minima energy structures of each HOZ fragmentation displayed.Energies are relative to the raw sCI 3 + HCHO reactants.

2 . 2
The sCIs 2 & 3 + SO 2 reactions.The sCI 2 + SO 2 reaction is unique amongst the HFO-sCI + SO 2 reactions studied here in that the cycloaddition process is not barrierless and structures for the TS ENDO & TS EXO barriers can be readily computationally determined.The TS ENDO barrier (−18.3 kJ mol −1 ) is lower in energy than TS EXO (−17.6 kJ mol −1 ) and so the endo-orientated SOZ 2 yield is greater than exoorientated SOZ 1 at a ratio of 0.69 : 0.31.Even though all other SO 2 reactions with sCIs 1-5 are barrierless, sCI 2 + SO 2 still has a substantially large theoretical rate constant (k THEO = 1.90 × 10 −12 cm 3 s −1 ).As both the TS EXO & TS ENDO channels of sCI 3 + SO 2 are barrierless, the k d-d value (4.08 × 10 −10 cm 3 s −1 ) is assigned as the rate constant and SOZs 3 & 4 are produced in equal fractions.Due to the high energy barrier for interconversion via inversion of the S atom (>180 kJ mol −1 ), the SOZ 1 & 2 conformers, produced from the SO 2 reaction with the monosubstituted sCI 2, cannot interconvert with each other without

Fig. 8
Fig.8The potential energy surface of sCI 1 + SO 2 with minima energy structures of each SOZ fragmentation displayed.Energies are relative to the raw reactants.
3 into SO 3 + CF 3 CHO, TS acid into SO 2 + TFA, or TS Ester into SO 2 + CF 3 OCHO.It was noted above that the cycloaddition step of the SO 2 + sCI 2 reaction produced a large yield of the SOZ 2 & 4 conformer set and a small yield of the SOZ 1 & 3 conformer set, whereas the SO 2 + sCI 3 produced fairly even yields of each SOZ set.Furthermore, these stereochemically distinct SOZ sets each have their own fragmentation pathways, as SOZ 1 & 3 breakdown via TSSO 3 2, TS acid 1 & TS Ester 1 and SOZ 2 & 4 fragments via TSSO 3 1, TS acid 2 & TS Ester 2. The overall product branching ratio for sCI 2 + SO 2 is very similar to that of sCI 3 + SO 2 , so it appears that the differing SOZ yields seen in these two reactions has little overall effect on the product branching fraction.sCI 2 + SO 2 does produce a small yield of sCI 3 + SO 2 from the cycloreversion channel, whereas the sCI 2 + SO 2 yield from sCI 3 + SO 2 reaction is negligible.This is because cycloreversion to sCI 2 + SO 2 has to overcome TS barriers, whereas cycloreversion to sCI 3 + SO 2 only has to overcome the energy of required to reform the raw reactants.The product branching fraction for ester formation is slight.The dominant nal products for both the sCIs 2 & 3 + SO 2 reactions are CF 3 CHO + SO 3 (G THEO ∼ 0.995) with only a very minor yield of TFA + SO 2 (G THEO ∼ 0.003).3.2.3sCIs 4 & 5 + SO 2 reactions.Both SO 2 reactions with sCIs 4 & 5 are barrierless, so their k d-d values (4.51 & 4.42 × 10 −10 cm 3 s −1 ) are used for the rate constants for these reactions and their exo-& endo-SOZ conformers are produced in almost equal yields.The sCIs 4 & 5 + SO

Fig. 10
Fig.10The collective potential energy surface for sCIs 2 & 3 + SO 2 reactions with minima energy structures of each SOZ fragmentation displayed.Energies are relative to the raw sCI 2 + SO 2 reactants.

Fig. 11
Fig. 11 The collective potential energy surface for sCIs 4 & 5 + SO 2 reactions with minima energy structures of each SOZ fragmentation displayed.Energies are relative to the raw sCI 4 + SO 2 reactants.

Fig. 14
Fig. 14 The potential energy surfaces for the sCI 5 + H 2 O (in black) and sCI 5 + (H 2 O) 2 (in red) reactions.Energies are relative to raw reactants.

Fig. 15 A
Fig. 15 A comparison of the impact that heteroatom tuning of the group XVI-centred co-reactant (identity in the legend) has on the bimolecular rate constant (k THEO ) of different HFO-sCI reaction.Of note in this figure is the large increase in reactivity for sCIs 4 & 5 + H 2 O and MeOH that is not observed for H 2 S reactions.

Table 4
A tabulated breakdown of the results from the HOZ fragmentation of all HFO-sCI + aldehyde & ketone studied here, giving the final product branching for each product set 2

Table 5
A tabulated breakdown of the outcomes from SOZ fragmentation of the sCIs 1-5 + SO 2 reactions studied here, giving the final product branching for each product set 180,181Sheps et al. showed in a study of CH 2 OO + (H 2 O) 2

Table 6 .
3owever, whether the sCI with H 2 O or (H 2 O) 2 is more prevalent depends not only on the relative abundance of the H 2 O or (H 2 O) 2 , but also on the identity of the sCI.For instance, the k EFF ((H 2 O) 2 ) for sCI 2 is around 10 3 s −1 larger than the equivalent k EFF (H 2 O), because the natural abundance of H 2 O (6.2 × 10 17 molec.percm 3over that of (H 2 O) 2 (8.7 × 10 14 molec.percm3) is overcome by the fact that the sCI 2 has a 10 6 cm 3 s −1 larger k THEO ((H 2 O) 2 ) than k THEO (H 2 O).sCIs 1 & 3 also exhibit larger k EFF ((H 2 O) 2 ) than k EFF (H 2 O) by between 10 1 -10 3 s −1 , meaning that tropospheric removal of the sCI occurs more readily via reaction with the water dimer than the monomer.But, whereas the rate constants for sCIs 4 & 5 + H 2 O reactions are much enhanced, the sCIs 4 & 5 + (H 2 O) 2 systems are so reactive that the k d-d capture limits are used as the rate constants, and the gap between these k d-d ((H 2 O) 2 ) and k THEO (H 2 O) values is much reduced.Consequently, the k EFF (H 2 O) for sCI 5 is similar to the equivalent k EFF ((H 2 O) 2 ) value and the k EFF (H 2 O) for sCI 4 is larger than k EFF ((H 2 O) 2 ), as seen in This great preference for HFO-sCI + H 2 O & (H 2 O) 2 reactions leads to the signicant production of uorinated a-hydroxyhydroperoxides, which have been implicated in forest damage and associated with generating important secondary products like H 2 O 2 , organic acids and aldehydes, albeit that the primary emission locations for HFO are likely to be in urban environments.86,170-176 HFO-sCI + MeOH reactions are not as competitive as the equivalent H 2 O & (H 2 O) 2 reactions, but because, MeOH is emitted from vehicles that run off biofuel and HFOs are used in vehicle coolant units, the HFO-sCI & MeOH species have greater concentration overlap in many urban environments.
Environmental Science: Atmospheres Paper reaction path and k EFF are both equivalent and effective tools to show the efficiency of the respective reaction path as an sCI sink but, while Wang et al. have identied syn-& anti-CF 3 CHOO experimentally, no k UNI values were identied in either the Wang et al. or Guidry et al. studies.