Comprehensive multiphase chlorine chemistry in the box model CAABA/MECCA: Implications to atmospheric oxidative capacity

. Tropospheric chlorine chemistry can strongly impact the atmospheric oxidation capacity and composition, especially in urban environments. To account for these reactions, the gas-and aqueous-phase Cl chemistry of the community atmospheric chemistry box model CAABA/MECCA has been extended. In particular, an explicit mechanism for ClNO 2 formation following N 2 O 5 uptake to aerosols has been developed. The updated model has been applied to two urban environments with different concentrations of NO x ( NO + NO 2 ): New Delhi (India) and Leicester (United Kingdom). The model shows a sharp 5 build-up of Cl at sunrise through Cl 2 photolysis in both the urban environments. Besides Cl 2 photolysis, ClO + NO reaction, and photolysis of ClNO 2 and ClONO are also prominent sources of Cl in Leicester. High-NO x conditions in Delhi tend to suppress the night-time build-up of N 2 O 5 due to titration of O 3 and thus lead to lower ClNO 2 , in contrast to Leicester. Major loss of ClNO 2 is through its uptake on chloride, producing Cl 2 , which consequently leads to the formation of Cl through photolysis. The reactivities of Cl and OH are much higher in Delhi, however, the Cl / OH reactivity ratio is up to ≈ 9 times greater 10 in Leicester. The contribution of Cl to the atmospheric oxidation capacity is signiﬁcant and even exceeds (by ≈ 2.9 times) that of OH during the morning hours in Leicester. Sensitivity simulations suggest that the additional consumption of VOCs due to active gas-and aqueous-phase chlorine chemistry enhances OH , HO 2 , and RO 2 near the sunrise. The simulation results of the updated model have important implications for future studies on atmospheric chemistry and urban air quality


Introduction
Chlorine (Cl) radicals are one of the most important players in the tropospheric chemistry (Seinfeld and Pandis, 2016;Ravishankara, 2009).Cl impacts the oxidative capacity of the atmosphere and radical cycling, and, therefore, can significantly alter the atmospheric composition (Seinfeld and Pandis, 2016;Faxon and Allen, 2013).In comparison with hydroxyl (OH) radicals, Nevertheless, the heterogeneous chemistry of Cl species remains poorly represented in models, and often neglected in large scale numerical simulations.For example, in several models, the heterogeneous uptake of N 2 O 5 on aqueous aerosols yielded nitric acid (HNO 3 ) via reaction HET1: N 2 O 5 (g) + H 2 O(aq) → 2 HNO 3 (aq) (HET1) However, N 2 O 5 uptake on chloride-containing particles can produce ClNO 2 (Behnke et al., 1997;Thornton et al., 2010) especially in urban environments with strong NOx emissions (Osthoff et al., 2008;Young et al., 2012).Incorporating heterogeneous mechanism of ClNO 2 into the regional models led to 3-12 % increase in O 3 over Northern China (Sarwar et al., 2014;Zhang et al., 2017;Liu et al., 2017).In addition, heterogeneous reactions of Cl-containing species including particulate chloride (pCl − ), Cl 2 , ClNO 2 , chlorine nitrate (ClNO 3 ), and hypochlorous acid (HOCl) are suggested to result in the formation of Cl radicals as well as in recycling of NOx, and HOx (OH, and HO 2 ) (Ravishankara, 2009;Qiu et al., 2019a;Hossaini et al., 2016;Faxon and Allen, 2013).Very recent measurements suggest a reduction in ClNO 2 formation due to the competition of N 2 O 5 uptake among chloride, sulphate and acetate aerosols (Staudt et al., 2019).These heterogeneous reactions can be of paramount significance in the Cl budget, however, to the best of our knowledge, these are not yet considered in model simulations.
The main goal of the present study is to investigate the role of chlorine chemistry in chemically contrasting urban environments.In this regard, we incorporate comprehensive gas-phase and heterogeneous Cl chemistry into a state of the art box model.Section 2 provides a detailed description of the Cl chemistry mechanism with gas-phase and heterogeneous reactions.
Section 3 describes the model setup and Section 4 shows the simulation results which include a detailed investigation on (i) the sensitivity of air composition to chlorine chemistry, (ii) the production and loss of Cl and ClNO 2 , (iii) the role of Cl in the Atmospheric Oxidative Capacity (AOC), and (iv) the sensitivity to ClNO 2 + Cl − reaction.

Mechanism Development
The community box model "Chemistry As A Boxmodel Application/Module Efficiently Calculating the Chemistry of the Atmosphere" (CAABA/MECCA, Sander et al., 2019), has been used in this work.A comprehensive gas-and aqueous-phase mechanism of chlorine chemistry has been added to MECCA, here used within the box model CAABA.The gas-phase and heterogeneous chemistry implemented in MECCA is described in the following subsections and the full mechanism is shown in the supplementary section.
These reactions could suppress the formation of ClNO 2 and also the corresponding rate constants for reactions A11-A14 are similar to the NO + 2 + Cl − reaction yielding ClNO 2 i.e. 7.5×10 9 mol −1 L s −1 (Staudt et al., 2019;Ryder et al., 2015;Heal et al., 2007).Methanol reacts with NO + 2 (A15) and forms aqueous methyl nitrate (CH 3 NO 3 ) (Iraci et al., 2007).Phase exchange for CH 3 NO 3 and nitrophenol (HOC 6 H 4 NO 2 ) is shown by reactions H4 and H5, respectively.The heterogeneous chemistry just discussed is implemented in MECCA and is summarized in Fig. 1.The rate constant for NO + 2 reaction with methoxyphenol is about ≈10000 times smaller than NO + 2 + phenol reaction (Kroflič et al., 2015), so it is not considered in this study.In addition, nitration reactions of other alcohols (e.g.catechol and polyphenols) could be potentially important, however due to unavailability of corresponding rate constants, these reactions are not considered in this study, nonetheless future studies calculating the kinetics of these reactions are recommended.In this study, Cl reacts with hydrocarbons and acetone via H-abstraction, and hence does not lead to the formation of any Cl-containing molecules, such as chloroacetone.This means that there are no such reactions in MECCA in which the Cl atom becomes part of the organic molecule.The reaction of Cl atoms with isoprene proceeds mainly via addition, and it produces chlorine-containing organics (Ragains and Finlayson-Pitts, 1997;Fan and Zhang, 2004).However, here we have simplified the mechanism by not considering the fate of organohalogens.Therefore, for future research, it would be valuable to investigate the chemical kinetics of such reactions kinetics and their importance in the formation of organohalogen compounds.

Box model setup
The chemistry described in section 2 has been added into community box model CAABA/MECCA v4.4.2 (Sander et al., 2019).
A comprehensive gas and aqueous phase tropospheric chemistry involving total 3330 reactions was utilized for the simulations, and the full set of reactions are presented in the electronic supplement.The gas-phase chemistry of organics like terpenes and aromatics is treated by the Mainz Organic Mechanism (MOM) (Taraborrelli et al., 2012;Nölscher et al., 2014;Hens et al., 2014;Taraborrelli et al., 2021).The aqueous-phase chemistry of oxygenated VOCs is treated by the Jülich Atmospheric Mechanism of Organic Chemistry (JAMOC) (Rosanka et al., 2021).The numerical integration of the chemical mechanism is performed by the kinetic preprocessor v2.1 (KPP) (Sandu and Sander, 2006).The photolysis rate constants (J values) are calculated by the submodel JVAL, based on the method by Landgraf and Crutzen (1998).The Cl chemistry is expected to be more prominent during winter conditions due to higher concentration of Cl-containing species in the boundary layer (Thornton et al., 2010;Gunthe et al., 2021;Sommariva et al., 2021), and therefore, simulations are performed for the winter season.Hence, the model is set-up for typical winter conditions of two different urban environments: Delhi (28.6 • N, 77.2 • E), India and Leicester (52.4 • N, 01.1 • W), United Kingdom.Simulations are performed for a 5-day period (17-21 February 2018) and output of 5 th day has been considered for the analysis; by then, radicals had achieved almost a steady state.The typical environmental conditions used in the simulations for Delhi (Tripathi et al., 2022) and Leicester (Sommariva et al., 2021) are summarized in Tab. 3 and Tab.S1.

Results and Discussion
The model captures the patterns in O 3 variability at both locations (Sommariva et al., 2018;Nelson et al., 2021;Chen et al., 2021;Sommariva et al., 2021;Nelson et al., 2023) to an extent, as shown in Fig. 2. O 3 is underestimated after ≈16:00 h LT in Leicester mainly due to titration by high NO and lack of adequate dynamics/transport of O 3 in the model.Entrainment seems to improve O 3 after mid-night, towards the observed values (Fig. 2i).Simulated isoprene is in agreement with diurnal observations in Delhi (Tripathi et al., 2022) and in accordance with observed mean level in Leicester (Sommariva et al., 2021).
The nitrate radical (NO 3 ), which is a nighttime oxidant, is formed through reaction between NO 2 and O 3 (G37).NO 3 can react with NO 2 forming N 2 O 5 , which can again produce NO 3 and NO 2 through thermal dissociation (G38). NO As seen in Fig. 2e, NO 3 remains negligible during the night-time (≈18:00-07:30 h LT) in Delhi due to unavailability of O 3 under high-NO conditions (up to 200 nmol/mol).Interestingly, despite its very short lifetime (≈5 s), about ≈0.1 pmol/mol of NO 3 sustains during daytime.This is primarily due to prevailing levels of NO 2 (≈30 nmol/mol) and O 3 (≈40 nmol/mol).Such unusual daytime enhanced NO 3 have been reported in recent studies, for example, 5-31 pmol/mol of NO 3 in Texas, USA (Geyer et al., 2003).Aircraft measurements during the New England Air Quality Study showed ≈0.5 pmol/mol of NO 3 within boundary layer (≤1 km) during noon time (Brown et al., 2005).The calculated NO 3 levels using steady state approximation showed 0.01-0.06pmol/mol of NO 3 for the 1997-2012 period at urban sites in the UK (Marylebone Road London, London Eltham, and Harwell) (Khan et al., 2015a).Horowitz et al. (2007) suggested that NO 3 in tenths of pmol/mol during daytime over the eastern United States results in formation of ≈50 % isoprene nitrates through oxidation of isoprene, which could further affect the formation of O 3 and SOA significantly (Horowitz et al., 2007).Following to higher NO 3 , up to 8 pmol/mol of N 2 O 5 is simulated during daytime in Delhi (Fig. 2f).Similar unusual daytime high levels of N 2 O 5 (≈21.9±29.3pptv) during wintertime were recently measured at Delhi using a high-resolution iodide adduct chemical ionization mass spectrometer (Haslett et al., 2023).
Enhanced NO 3 ≈ 2.6 pmol/mol and N 2 O 5 ≈ 330 pmol/mol are simulated after mid-night in Leicester (Fig. 2k, 2l).In contrast to Delhi, the daytime simulated levels of NO 3 are negligible as it gets removed rapidly during the daytime by photolysis and through its reactions with NO, HO 2 , RO 2 , and VOCs (Khan et al., 2015b).In conjunction with high NO from ≈16:00 h LT to near midnight that titrates O 3 , the corresponding NO 3 and N 2 O 5 is negligible (following reactions G37 and G38).
Night-time high and negligible day-time levels of NO 3 and N 2 O 5 are their typical features which are generally reported in the literature (Brown et al., 2001;Seinfeld and Pandis, 2016).

Sensitivity of air composition to chlorine chemistry
To investigate the effects of Cl chemistry on air composition, other than comprehensive chemistry simulation discussed in previous section (simulation: NEW i.e. chemistry already present in the model + newly added gas and aqueous phase chlorine chemistry), two additional simulations have been performed, which are: (1) OLD -this includes default chemistry already present in the model, and (2) NOCL -OLD minus chlorine chemistry (i.e.without Cl chemistry).OLD simulation also encompassed some basic chlorine chemistry that was part of the model prior to its update (full mechanism is also shown in supplement).Figure 3 shows the comparison of Cl, ClNO 2 , ClONO, OH, HO 2 , and RO 2 variations among the three simulations in Delhi and Leicester.Figure S5 shows the differences in diurnal variations of Cl, ClONO + ClNO 2 , OH, HO 2 , and RO 2 in NEW simulation with: NOCL and OLD simulations.
The Delhi environment is mainly characterized by two peaks in Cl, a predominant sharp peak just after sunrise followed by a broad shallow peak during noontime, corresponding to different mechanisms as discussed in the next section.With newly added chemistry (NEW simulation), a sharp peak in Cl is seen near sunrise, with the maximum values attained is ≈3.5 fmol/mol (8.75x10 4 molec cm −3 ) in Delhi (Fig. 3a).A broad smaller peak with magnitude of ≈0.8 fmol/mol maximizing around noontime is seen, which is ≈4 times smaller than the first morning peak.OLD simulation also show a sharp peak in Cl near sunrise in Delhi, with a maximum of ≈11 fmol/mol (2.75 x 10 5 molec cm −3 ).Cl get suppressed by up to ≈ 0.01 pmol/mol of maximum value in the OLD simulation, in the presence of added chlorine chemistry (NEW) as shown in The model-predicted Cl peaks at ≈2 fmol/mol (5.2×10 4 molec cm −3 ) during sunrise in Leicester (Fig. 3g).In contrast to Delhi, suppressed Cl (up to ≈ 3.2 times) with a narrow peak is simulated by OLD simulation in comparison with NEW simulation containing newly added Cl chemistry, at Leicester.In contrast to negligible night-time ClONO + ClNO 2 in Delhi, it shows a strong build-up over Leicester during 0-4 hours with a maximum of ≈40 pmol/mol, with higher levels (up to 50 pmol/mol) prevailing until about sunrise.ClONO + ClNO 2 is negligible during mid-day until mid-night, in accordance with N 2 O 5 in Leicester as shown in Fig. 2l.Previous studies have demonstrated that the formation of ClNO 2 occurs within the nocturnal residual layer, which contains lower levels of NO compared to the surface layer.Subsequently, ClNO 2 mixes downward during the morning when the convective mixed layer develops (Bannan et al., 2015;Tham et al., 2016).However, the present study does not account the the effect of transport processes due to the limitations of the box model.The effects of added Cl chemistry on OH, HO 2 , and RO 2 are more prominent in Leicester compared to Delhi.NEW simulation show strong enhancements in OH (up to ≈ 2 times), HO 2 (up to ≈ 5 times), and RO 2 (up to ≈ 8 times) after sunrise which is gradually progressive, resulting in higher levels during noon-time as well (Fig. 3, Fig. S5).Remarkably elevated levels of RO 2 (by a factor of ≈ 2) are prominent during the noon hours.Such elevated levels of RO 2 could favour enhanced levels of secondary organic aerosols in Leicester.The impact of Cl chemistry on aerosols (NO + 2 , NO − 3 , and oxalic acid) is discussed in Supplementary section 2.2 (Fig. S6).Though significant differences in NO + 2 , NO − 3 , and oxalic acid are seen due to Cl chemistry, further measurements are required for validation.In the next sections, we have analysed the observed behaviour of Cl and ClNO 2 in the NEW simulation over both the locations in more detail.Cl 2 with a maximum rate of 1.2 x 10 7 molec cm −3 s −1 .The shallow secondary peak is due to the reaction HCl + OH with a noon time rate of ≈0.4 x 10 7 molec cm −3 s −1 .However, there is a smaller contribution from other reactions (photolysis of ClNO 2 , ClONO and reaction of ClO with NO) to the morning peak, which have negligible contributions during the daytime.
Interestingly, there is a strong consumption of Cl to oxidize VOCs (peak rate ≈2.4 x 10 7 molec cm −3 s −1 ) during sunrise, and a lesser consumption during the rest of the day.Cl + NO 2 is also a Cl sink during the morning time in Delhi.The Cl-initiated oxidation of VOCs in the morning hours in Delhi may lead to formation of secondary organic aerosols and new particle formation, which opens up pathways of future research in this direction.In addition to Cl 2 photolysis (≈1.0 x 10 6 molec cm −3 s −1 ), photolysis of ClNO 2 and ClONO, and ClO + NO reaction (total rate ≈0.8 x 10 6 molec cm −3 s −1 ) are other prominent sources of Cl in Leicester.VOCs are the major sink for Cl (rate ≈1.3 x 10 6 molec cm −3 s −1 ), followed by NO 2 (rate ≈0.6 x 10 6 molec cm −3 s −1 ).We further analyzed the production and loss pathways of ClNO 2 , as shown in Fig. 4c,d.While the major source of ClNO 2 is through the Cl + NO 2 reaction with a reaction rate ≈3 × 10 5 molec cm −3 s −1 in Delhi, the aqueous phase reaction Cl − + NO + 2 (≈3.4 × 10 5 molec cm −3 s −1 ) is the prominent source in Leicester corresponding to the peak ClNO 2 (Fig. 2h,p).Though gasphase reaction Cl + NO 2 is discussed in the literature (Burkholder et al., 2015;Qiu et al., 2019a), however, to the best of our knowledge, such an unusually higher contribution of this reaction (seen in Delhi) as compared to the aqueous-phase reaction of Cl − + NO + 2 has not been reported in any study.The reaction of Cl with NO 2 (≈1.1 × 10 5 molec cm −3 s −1 ) is the major ClNO 2 source during sunrise in Leicester.In contrast, there is lesser contribution of Cl − + NO + 2 reaction (rate ≈ 1 × 10 3 molec cm −3 s −1 ) in ClNO 2 production in Delhi.The prominent sink for ClNO 2 is through its heterogeneous reaction with Cl − (≈1.8 × 10 5 molec cm −3 s −1 or 7.2 × 10 −15 mol mol −1 s −1 ) in Delhi almost throughout the day, while it's loss through the photolysis (≈0.5 × 10 5 molec cm −3 s −1 or 2 × 10 −15 mol mol −1 s −1 ) is also an important sink during the daytime.We are using ClNO 2 uptake coefficient, γ = 9E-3 from Fickert et al. (1998) in the simulation.Sensitivity simulation with γ = 1E-5 (Haskins et al., 2019) results in considerably slower (by a factor of ≈270 and ≈17, near sunrise and during mid-day, respectively) loss rate of ClNO 2 with Cl − than in the NEW simulation over Delhi.ClNO 2 loss through the reaction ClNO 2 + Cl − (≈2.7 × 10 5 molec cm −3 s −1 or 1.0 × 10 −14 mol mol −1 s −1 ) is its major sink in Leicester from mid-night to mid-day, while photolysis (≈0.3 × 10 5 molec cm −3 s −1 or 1.1 × 10 −15 mol mol −1 s −1 ) is smaller sink from sunrise to mid-day here.The diurnal variation in Cl 2 , and its production and loss mechanisms over Delhi and Leicester are shown by Fig. S1 and Fig. S2.
In conjunction with major loss of ClNO 2 , ClNO 2 + Cl − reaction is the major contributor to Cl 2 formation over Delhi and Leicester.
We also calculated ClNO 2 yield from NO + 2 (Fig. S3), which is the ratio of P ClNO2 /L total , where P ClNO2 is the rate of ClNO 2 production through Cl − + NO + 2 reaction and L total denotes the loss rate of NO + 2 through it's reactions with Cl − , H 2 O, SO 2− 4 , HCOO − , CH 3 COO − , phenol, CH 3 OH, and cresol (A4, A10-A16).ClNO 2 yield is ≈0.9 over Delhi, indicating the strongest loss of NO + 2 is through it's reaction with Cl − , which is also mimicked in Fig. S4a showing the same concentrations of ClNO 2 as in NEW simulation and when other NO + 2 reactions (A10-A16) are turned off (simulation: without other NO + 2 reactions).
ClNO 2 yield over Leicester is between ≈0.40-0.55,which is about half the yield in Delhi.Stronger ClNO 2 yield in Delhi could be attributed to ≈2 times higher Cl − than Leicester.Lesser ClNO 2 yield in Leicester portrays the importance of NO + 2 loss reactions (A10-A15) other than with Cl − , which could be seen through Fig. S4b where ClNO 2 is increased by more than twice during early morning hours when A10-A15 reactions are kept inactive in the model.The determination of ClNO 2 yield using cavity ring-down spectroscopy and chemical ionization mass spectrometry, shows yield ranging between 0.2 to 0.8 for Cl − concentrations of 0.02 to 0.5 mol/L (Roberts et al., 2009).The measurements of ClNO 2 yield for coastal and open ocean waters were found to be between 0.16-0.30which is suppressed by up to 5 times than equivalent salt containing solutions, due to the addition of aromatic organic compounds (e.g., phenol and humic acid) to synthetic seawater matrices (Ryder et al., 2015).

Role of Cl in Atmospheric Oxidative Capacity (AOC)
In order to understand the role of Cl as oxidising agent with respect to the OH radical, we calculated the reactivity of Cl and , where radical is Cl or OH, and [X i ] is the concentration of specie X i (here X i includes CO, CH 4 , primary VOCs and NMHCs which are initialized in the model) (Fig. 5).The corresponding rate constants for Cl + X and OH + X reactions are taken from the MECCA.The reactivity of both Cl and OH decreases rapidly nearly from sunrise to noon time and afterwards increases gradually at both locations.In comparison to Leicester, the magnitudes of Cl and OH reactivity in Delhi are higher by up to ≈1.4 and ≈12 times, respectively.However, the Cl/OH reactivity ratio in Leicester is up to ≈9 times higher than that in Delhi.Cl reactivity is lower (Delhi: ≈685 s −1 , Leicester: ≈553 s −1 ) during noontime and higher (Delhi: ≈750 s −1 , Leicester: ≈554 s −1 ) during nighttime and early morning hours at both locations.The OH reactivity follows a similar pattern as that of Cl in Delhi and Leicester.The ratio of Cl to OH reactivity starts increasing after sunrise, reaching a maximum value of ≈42 at nearly 16:00 h LT and then decreases further in Delhi.As mentioned above, Cl/OH reactivity ratio in Leicester shows a double peak pattern, with one peak (≈270) during early morning ≈04:00 h LT and other peak (≈276) at about 16:00 h LT.Leicester.
We quantified the relative contribution of Cl in atmospheric oxidative capacity (AOC) using the model.AOC represents the sum of oxidation rates of specie X i by oxidants Y (OH, Cl, and other radicals: NO 3 and O 3 ) (Elshorbany et al., 2009): where, k Xi is the corresponding rate constant for X i + Y reaction.Accordingly, the magnitude of AOC depends upon the concentration and reactivity of Cl. Figure 6 shows the contribution of individual oxidants in AOC at both locations.Besides OH, Cl is the second most important oxidant in Delhi, with a significant contribution of 23.4 % during morning (averaged over 07:00-09:00 h LT), and 8.2 % throughout the day (06:00-16:00 h LT).In Leicester, Cl is the highest contributor (74.0 %) towards AOC during morning.In fact, with 34.1 % contribution, Cl is major oxidant after OH, during the daytime.Besides the abundance of Cl, higher reactivity enhances the contribution of Cl in AOC, which is further substantiated by the ratio of Cl reactivity to OH reactivity (Fig. 5b).This ratio indicates that Cl reactivity exceeds OH reactivity by a significant margin,  ranging from 265 to 276 times greater throughout the day in Leicester.Such a substantial contribution of Cl in AOC leads to enhancements of RO 2 as seen in Fig. 3(f,l).Especially, a prominent peak in RO 2 during early morning (07:00-09:00 h LT) 305 is imparted to strong participation of Cl in atmospheric oxidation during this time.Notably strongest contribution of Cl in AOC during early morning in Leicester, strengthens RO 2 peak by up to a factor of 8 (Fig. 3l).The role of Cl is predominant in Leicester as well as in Delhi during early morning, compared to a polluted environment of Hong Kong, China where Cl contribution was estimated to be 21.5 % (Xue et al., 2015).NO 3 and O 3 were found to play a relatively minor role in AOC at both urban environments.

Sensitivity to ClNO
In a study conducted by Haskins et al. (2019), using the reacto-diffusive length-scale framework, it was demonstrated that field and laboratory observations could be reconciled by considering an aqueous-phase reaction rate constant for the ClNO 2 + Cl − reaction on the order of ≈ 10 4 s −1 .This reaction rate constant is considerably lower (by ≈179 times) than reported in 10 4 mol −1 L s −1 (Haskins et al., 2019) for the ClNO 2 + Cl − reaction, for both Delhi and Leicester.As depicted in Figure S7a, the concentration of Cl remains nearly the same in the NEWrate simulation compared to the NEW simulation over Delhi.
However, there are significant changes in the concentration of ClNO 2 , as shown in Fig. S7b.The simulated ClNO 2 exhibits a broader peak and is approximately 30 pmol/mol higher near sunrise in the NEWrate simulation when compared to the NEW simulation.During the nighttime, approximately 20 pmol/mol of ClNO 2 is simulated in the NEWrate simulation, whereas it is negligible in the NEW simulation (see Fig. 3b).Since the Cl concentration is almost similar in both the NEW and NEWrate simulations, the differences in the simulated concentrations of OH, HO 2 , and RO 2 remain consistent between the NEWrate or NEW simulations and the OLD and NOCL simulations (refer to Fig. S7d, e, f, and Fig. 3d, e, f).The production and loss mechanisms of Cl are similar in both the NEW and NEWrate simulations (see Fig. S8a and Fig. 4a).The contributions from ClNO 2 formation reactions are also similar.However, in contrast to the NEW simulation, the loss of ClNO 2 through photolysis becomes dominant and is ≈6 times greater than its loss through ClNO 2 + Cl − reaction, in NEWrate simulation.
The contribution of radicals to AOC is also similar between the NEW and NEWrate simulation, as depicted in Fig. 6a,c and Fig. S9a,c respectively, over Delhi.
In contrast to Delhi, significant differences are seen in atmospheric composition in Leicester when the rate coefficient of the ClNO 2 + Cl − reaction is altered (as shown in Fig. S7).The peak concentration of Cl becomes ≈0.6 fmol/mol during the morning hours of NEWrate simulation (Fig. S7g), which is about 4 times lower than the concentration of Cl in NEW simulation (Fig. 3g).However, due to slower rate of ClNO 2 consumption with Cl − , the simulated ClNO 2 using the NEWrate is significantly enhanced (by ≈ 5 times) compared to NEW simulation, reaching a maximum of about 210 pmol/mol around sunrise (see Fig. S7h).Due to lower Cl concentrations, the levels of ClONO also decrease by 3.5 times in NEWrate simulation (as shown in Fig. S7i) compared to NEW simulation (Fig. 3i).The dominant peak seen at sunrise in the NEW simulation for OH, HO 2 , and RO 2 is significantly reduced with the lower rate of the ClNO 2 + Cl − reaction, as illustrated in Fig S7j ,k,l.
Significant changes in the production and loss mechanisms of Cl and ClNO 2 are seen in Leicester when the reaction rate of A6 is changed, as shown in Fig. S8 and Fig. 4b.For example, in the NEWrate simulation, other reactions, including the photolysis of ClNO 2 and ClONO, and ClO + NO reaction, become prominent sources of Cl (with a rate of approximately 6.0 x 10 5 molec cm −3 s −1 ), whereas in the NEW simulation, the major source for Cl is photolysis of Cl 2 .The primary source for ClNO 2 production remains the Cl − + NO + 2 reaction in both the NEW and NEWrate simulations.However, in the NEWrate simulation, ClNO 2 loss from photolysis becomes the major sink, whereas in the NEW simulation, loss from the ClNO 2 + Cl − reaction is prominent.In addition, remarkable changes in AOC are seen between the NEWrate (Fig. S9b, d) and the NEW simulation (Fig. 6b,d).In the NEWrate simulation, even though Cl remains the major oxidant its contribution is notably reduced from 74% (in NEW simulation) to 58.1% during the early morning hours.
Extended gas-and aqueous-phase chemistry of chlorine compounds has been added to the MECCA mechanism.It consists of 36 gas-phase reactions (inorganic, organic, and photolysis reactions).A total of 24 aqueous-phase and heterogeneous reactions have been added, containing detailed chemistry of N 2 O 5 uptake on aerosols to yield ClNO 2 and various other competing reactions.The updated model is applied to two different urban environments: Delhi (India) and Leicester (United Kingdom) during winter time.The major conclusions are: 1.The model predicts up to 0.1 pmol/mol of NO 3 and up to 8 pmol/mol of N 2 O 5 during daytime in Delhi.However, night-time production of NO 3 and N 2 O 5 is seen to be negligible primarily due of the unavailability of O 3 .In contrast to Delhi, NO 3 and N 2 O 5 after mid-night in Leicester is ≈2.6 pmol/mol and ≈330 pmol/mol, respectively.N 2 O 5 uptake on aerosols yields ClNO 2 , which produces Cl via photolysis.
2. A sharp build-up of Cl with sunrise is mainly through Cl 2 photolysis in Delhi.Besides Cl 2 , photolysis of ClNO 2 and ClONO and the reaction of ClO with NO are prominent Cl sources in Leicester.VOCs are the main sink for Cl at both locations, whereas NO 2 is also an important sink for Cl in Leicester.The latter results in the formation of ClNO 2 with a major contribution in Delhi, while Cl − + NO + 2 is a stronger source in Leicester.Photolysis is the major sink for ClNO 2 in Delhi, however, its uptake on chloride aerosols is a prominent sink in Leicester.
3. The magnitude of Cl (≈750 s −1 ) and OH (≈25 s −1 ) reactivities are significantly greater in Delhi, particularly during the morning hours, when compared to Leicester.However, Cl to OH reactivity ratio (≈270) is pronounced in Leicester coinciding with higher contribution of Cl in AOC.
4. Sensitivity simulations reveal substantial post-sunrise enhancements of in OH, HO 2 , and RO 2 radicals, with a prominent secondary peak due to Cl chemistry.Up to 8 times higher RO 2 is simulated in Leicester primarily because of leading role of Cl in AOC potential.
It is important to note that box models, despite their general limitation of neglecting transport phenomena and assuming species to be well mixed, do include highly detailed chemical mechanisms.Furthermore, because the model is initialized with measurements of chemical species at both locations and the modeled levels align with observed data, significant discrepancies in model estimates would be unexpected.Future studies focussing on modeling vertical gradients, in particular for radical reservoir species such as HONO, and ClNO 2 (Young et al., 2012) are recommended.
This study highlights the vital role of Cl chemistry in governing the oxidation capacity of the atmosphere and air quality, and therefore it is important to account for it in detailed photochemical as well as in 3-D chemical transport models.This will lead to better quantification of the importance of radicals in atmospheric oxidation and hence, the formation of ozone as well as secondary aerosols, over regional to global scale.Future studies focusing on secondary aerosol formation and new particle formation from heterogeneous reactions are needed to deepen the understanding of transformation of trace gases to aerosols.

Figure 1 .
Figure 1.Aqueous-phase and heterogeneous chemistry added to MECCA.

Figure 2 .
Figure 2. Diurnal variations of NO, NO2, O3, C5H8, NO3, and N2O5 mixing ratios in Delhi (left) and Leicester (right).The unusal and negligible nighttime NO3 in Delhi is attributed to the nearly non-existent O3, due to titration by higher concentrations of NO.This leads to the negligible nighttime N2O5 in this region.Although mixing ratios of NO3 and N2O5 peak during the daytime, their levels remain quite low.Mean value of observed C5H8 in Leicester is shown by red colored long dashed line.

4. 2
Production and loss of Cl and ClNO 2 The sources and sinks of Cl in Leicester and Delhi are presented in Fig. 4. The left-upper panel (a) delineates the sources and sinks of Cl radical on diurnal scale in Delhi.The morning sharp peak in Cl radical is caused mainly by the photolysis of

Figure 4 .
Figure 4. Production and loss rates of (a, b) Cl and (c, d) ClNO2 in Delhi (left panel) and Leicester (right panel).

Figure 5 .
Figure 5. Reactivity of Cl and OH with CO, CH4, and VOCs, and Cl/OH reactivity ratio during the simulation period in (a) Delhi and (b)

Table 1 :
Gas-phase chlorine reactions and corresponding rate constants added to MECCA.The rate constants are expressed in units of cm 3 molecule −1 s −1 unless otherwise specified.Model-simulated maximum noontime J-values for Delhi are provided.

Table 2 :
Aqueous-phase and heterogeneous chlorine reactions added to MECCA

Table 3 .
Environmental conditions of Delhi and Leicester in the model simulations.