Nighttime atmospheric chemistry of iodine

. Little attention has so far been paid to the nighttime atmospheric chemistry of iodine species. Current atmospheric models predict a buildup of HOI and I 2 during the night that leads to a spike of IO at sunrise, which is not observed by measurements. In this work, electronic structure calculations are used to survey possible reactions that HOI and I 2 could undergo at night in the lower troposphere, and hence reduce their nighttime accumulation. The new reaction NO 3 + HOI → IO + HNO 3 is proposed, with a rate coefﬁcient calculated from statistical rate theory over the temperature range 260–300 K and at a pressure of 1000 hPa to be k(T ) = 2.7 × 10 − 12 (300 K/ T ) 2 . 66 cm 3 molecule − 1 s − 1 . This reaction is included in two atmospheric models, along with the known reaction between I 2 and NO 3 , to explore a new nocturnal iodine radical activation mechanism. The results show that this iodine scheme leads to a considerable reduction of nighttime HOI and I 2 , which results in the enhance-ment of more than 25 % of nighttime ocean emissions of HOI + I 2 and the removal of the anomalous spike of IO at sunrise. We suggest that active nighttime iodine can also have a considerable, so far unrecognized, impact on the reduction of the NO 3 radical levels in the marine boundary layer (MBL) and hence upon the nocturnal oxidizing capacity of the marine atmosphere. The effect of this is exempliﬁed by the indirect effect on dimethyl sulﬁde (DMS) oxidation.

Abstract. Little attention has so far been paid to the nighttime atmospheric chemistry of iodine species. Current atmospheric models predict a buildup of HOI and I 2 during the night that leads to a spike of IO at sunrise, which is not observed by measurements. In this work, electronic structure calculations are used to survey possible reactions that HOI and I 2 could undergo at night in the lower troposphere, and hence reduce their nighttime accumulation. The new reaction NO 3 + HOI → IO + HNO 3 is proposed, with a rate coefficient calculated from statistical rate theory over the temperature range 260-300 K and at a pressure of 1000 hPa to be k(T ) = 2.7 × 10 −12 (300 K/T ) 2.66 cm 3 molecule −1 s −1 . This reaction is included in two atmospheric models, along with the known reaction between I 2 and NO 3 , to explore a new nocturnal iodine radical activation mechanism. The results show that this iodine scheme leads to a considerable reduction of nighttime HOI and I 2 , which results in the enhancement of more than 25 % of nighttime ocean emissions of HOI + I 2 and the removal of the anomalous spike of IO at sunrise. We suggest that active nighttime iodine can also have a considerable, so far unrecognized, impact on the reduction of the NO 3 radical levels in the marine boundary layer (MBL) and hence upon the nocturnal oxidizing capacity of the marine atmosphere. The effect of this is exemplified by the indirect effect on dimethyl sulfide (DMS) oxidation.

Introduction
Active nighttime iodine chemistry was first evidenced a decade ago when it was shown that nocturnal I 2 emitted by macroalgae could react with NO 3 leading to the formation of IO and OIO, which were measured in the coastal marine boundary layer (MBL) at Mace Head, Ireland . The nitrate radical has also been recently suggested as a nocturnal loss of CH 2 I 2 , which helps to reconcile observed and modelled concentrations of this iodocarbon over the remote MBL . However, most of the work on reactive atmospheric iodine has focused on the use of daytime observations and models to assess its role in the catalytic destruction of ozone and the oxidizing capacity of the troposphere (e.g. Saiz-Lopez et al., 2012b, and references therein). In the MBL, iodine-catalysed along with bromine-catalysed ozone destruction contributes up to 45 % of the observed daytime depletion (Read et al., 2008;Mahajan et al., 2010a), although this contribution shows large geographical variability (Mahajan et al., 2012;Prados-Roman et al., 2015b;Volkamer et al., 2015). Iodine compounds have also been implicated in the formation of aerosols, although the mechanisms and magnitudes of these processes are not fully understood (Hoffmann et al., 2001;O'Dowd et al., 2002;McFiggans et al., 2004;Allan et al., 2015;Roscoe et al., 2015). Reactive forms of inorganic iodine may also contribute to the oxidation of elemental mercury over the tropical oceans (Wang et al., 2014). In recent years, iodine sources and chemistry have also been implemented in global models demonstrating the effect of iodine chemistry in the oxidation capacity of the global marine troposphere Saiz-Lopez et al., 2012aSherwen et al., 2016).
Iodine is emitted into the atmosphere from the ocean surface in both organic and inorganic forms. The main organic compounds emitted are methyl iodide (CH 3 I), ethyl iodide (C 2 H 5 I), propyl iodide (1-and 2-C 3 H 7 I), chloroiodomethane (CH 2 ICl), bromoiodomethane (CH 2 IBr) and diiodomethane (CH 2 I 2 ) (Carpenter, 2003;Butler et al., 2007;Jones et al., 2010;Mahajan et al., 2012). However, these organic compounds contribute only up to a fourth of the MBL iodine loading Mahajan et al., 2010a;Großmann et al., 2013;Prados-Roman et al., 2015b). Inorganic emissions of HOI and I 2 , which result from the deposition of O 3 at the ocean surface and subsequent reaction with I − ions in the surface microlayer, account for the main source of iodine in the MBL (Carpenter et al., 2013). Recent laboratory experiments have shown that HOI is the major compound emitted and provided parameterizations of the fluxes of both species depending on wind speed, temperature, and the concentrations of O 3 and I − (Carpenter et al., 2013;MacDonald et al., 2014). These parameterized fluxes of HOI and I 2 have then been used in a one-dimensional model to study the diurnal evolution of the IO and I 2 mixing ratios at the Cape Verde Atmospheric Observatory (CVAO) (Carpenter et al., 2013;Lawler et al., 2014). The model simulations replicate well the levels and general diurnal profiles of IO and I 2 , although an early morning "dawn spike" in IO is predicted by the models but has not been observed (Read et al., 2008;Mahajan et al., 2010a). The morning peak predicted by current iodine chemistry models is due to a buildup of the emitted I 2 and HOI (which is converted into IBr / ICl through heterogeneous sea-salt recycling) over the course of the night, followed by rapid photolysis at sunrise.
Traditionally it has been thought that iodine chemistry has a negligible effect on oxidizing capacity of the nocturnal marine atmosphere. As a consequence, unlike the demonstrated effect of iodine on the levels of daytime oxidants, the impact of active iodine upon the main nighttime oxidant, NO 3 , remains an open question. This is important given that in many parts of the ocean the NO 3 + dimethyl sulfide (DMS) reaction is at least as important as OH + DMS in oxidizing DMS (Allan et al., 2000), and hence a reduction of NO 3 may have an effect in the production of SO 2 and methane sulfonic acid (MSA). Here, we discuss possible mechanisms of nighttime iodine radical activation and their potential effect on nighttime iodine ocean fluxes and the currently modelled dawn spike in IO. A new reaction of HOI with NO 3 is proposed, supported by theoretical calculations. We explore the implications of this new reaction both for iodine and NO 3 chemistries.

Nocturnal iodine radical activation mechanism
We use the reaction mechanism that has recently been described in the global modelling studies by  and Ordóñez et al. (2012) (see Supplement). In addition to the reactions included in that scheme, we also include nighttime gas-phase reactions based on the theoretical calculations described below. The additional reactions are listed in Table 1, and a scheme with this new nocturnal chemistry is included in Fig. 1.
To the best of our knowledge, reactions of HOI specific to nighttime have not been studied, either theoretically or through laboratory experiments. Currently, HOI is thought to build up overnight until sunrise, with only heterogeneous uptake on sea-salt aerosol as a nighttime loss process (Saiz-Lopez et al., 2012b;Simpson et al., 2015). In addition to the well known I 2 + NO 3 → I + IONO 2 (R1) (Chambers et al., 1992), here we consider several possible HOI reactions that could occur at night, in the absence of photolysis and OH:

Theoretical calculations
In order to explore the feasibility of Reactions (R2)-(R4) taking place under the conditions of the lower troposphere, we carried out electronic structure calculations using the hybrid density functional/Hartree-Fock B3LYP method from within the Gaussian 09 suite of programs (Frisch et al., 2009), combined with a G2 level basis set for I (Glukhovtsev et al., 1995)

No. Reaction Notes
Reaction (R1) I 2 + NO 3 → I + IONO 2 1.5 × 10 −12 cm 3 molecule −1 s −1 (Chambers et al., 1992). Reaction (R2) HOI + NO 2 → I + HNO 3 Endothermic by 9 kJ mol −1 and the transition state is 73 kJ mol −1 above the reactants. Reaction (R3) HOI + HNO 3 → IONO 2 + H 2 O Exothermic by 11 kJ mol −1 . The reaction first forms a complex 21 kJ mol −1 below the reactants, but this rearranges to the products via a transition state that is 110 kJ mol −1 above the reactants. Reaction (R4) HOI + NO 3 → IO + HNO 3 Exothermic by 11 kJ mol −1 with all transition states below the reactants. k(T ) = 2.7 × 10 −12 (300 K/T ) 2.66 cm 3 molecule −1 s −1 . and the standard 6-311 + g(2d,p) triple-zeta basis set for O, N and H. Following geometry optimizations of the relevant points on the potential energy surfaces, and the determination of their corresponding vibrational frequencies and (harmonic) zero-point energies, energies relative to the reactants were obtained at the same level of theory. Spin-orbit corrections of −30.0 (Mečiarová et al., 2011), −14.4 (Khanniche et al., 2016), −5.9 (Šulková et al., 2013) and −4.8 (Kaltsoyannis and Plane, 2008) kJ mol −1 were applied to the energies of I, IO, HOI and IONO 2 , respectively. Reaction (R2) is endothermic by 2.6 kJ mol −1 , and therefore within the expected error of ±10 kJ mol −1 at this level of theory. Hence, it might be reasonably fast. However, the transition state of the reaction, which is illustrated in Fig. 2a, is 73 kJ mol −1 above the reactants, so this reaction will not occur at tropospheric temperatures. Reaction (R3) is exothermic by 19.8 kJ mol −1 . An HOI-HNO 3 complex first forms ( Fig. 2b), which is 21 kJ mol −1 below the reactants. However, this complex re-arranges to the IONO 2 + H 2 O products via the cyclic transition state shown in Fig. 2c, which is 110 kJ mol −1 above the reactants.
The stationary points on the potential energy surface (PES) for Reaction (R4) are illustrated in Fig. 3. HOI and NO 3 associate to form a complex which is 24 kJ mol −1 below the reactant entrance channel. Hydrogen atom transfer involves a submerged transition state to form an IO-HNO 3 complex, which can then dissociate to the products IO + HNO 3 . The vibrational frequencies, rotational energies and geometries (in Cartesian co-ordinates) of these intermediates are listed in Table 2. Overall, the reaction is exothermic by 14 kJ mol −1 . The energies of the HOI-NO 3 complex and the transition state are assigned the same spin-orbit correction as HOI (−5.9 kJ mol −1 ; Šulková et al., 2013), whereas the IO-HNO 3 complex is assigned the spin-orbit correction of IO (−14.4 kJ mol −1 ; Khanniche et al., 2016). This reflects the H-OI bond only increasing from 0.97 Å in HOI to 1.1 Å in the transition state, compared with 1.7 Å in the IO-HNO 3 complex. The spin-orbit correction for the transition state is therefore likely to be closer to that of HOI. Assigning the HOI spin-orbit correction therefore means that the barrier is highest with respect to the reactants so that the estimated rate coefficient (see below) may be a lower limit.
The rate coefficient for Reaction (R4) was then estimated using Rice-Ramsperger-Kassel-Markus (RRKM) theory, employing a multi-well energy-grained master equation solver based on the inverse Laplace transform method -MESMER (Master Equation Solver for Multi-well Energy Reactions) (Roberston et al., 2014). The reaction proceeds via the formation of the excited HOI-NO 3 complex from HOI + NO 3 . This complex can then dissociate back to the reactants or rearrange to the IO-HNO 3 intermediate complex over the transition state, which can in turn dissociate to the products IO + HNO 3 . Either of the intermediates can also be stabilized by collision with the third body (N 2 ). The time evolution of all these possible outcomes is modelled using the master equation. The internal energies of the intermediates on the PES were divided into a contiguous set of grains (width 10 cm 1 ), each containing a bundle of rovibrational states calculated with the molecular parameters in Table 2, using the rigid-rotor harmonic oscillator approximation for all species. It should be noted that the HOI-NO 3 and IO-HNO 3 complexes both have low-frequency vibrational modes (< 100 cm −1 ) which should more correctly be treated as hindered rotors rather than vibrations. However, in our experience this is not worth doing this until experimental rate coefficients are available to fit the rotor barrier heights. In any case, the energies of both complexes are far enough below the energy of the entrance channel (Fig. 3) that relatively small changes in their densities of states will have a minor effect on the overall rate coefficient. Each grain was then assigned a set of microcanonical rate coefficients linking it to other intermediates, calculated by RRKM theory. For dissociation to products or reactants, microcanonical rate coefficients were determined using in-verse Laplace transformation to link them directly to the capture rate coefficient, k capture . For Reaction (R4) and the reverse reaction IO + HNO 3 involving neutral species, k capture was set to a typical capture rate coefficient of 2.5 × 10 −10 (T /300 K) 1/6 cm 3 molecule −1 s −1 , where the small positive temperature dependence is characteristic of a long-range potential governed by dispersion and dipole-dipole forces (Georgievskii and Klippenstein, 2005).
The probability of collisional transfer between grains was estimated using the exponential down model, where the average energy for downward transitions was set to < E> down = 300 cm −1 for N 2 as the third body (Gilbert and Smith, 1990). MESMER determines the temperatureand pressure-dependent rate coefficient from the full microcanonical description of the system time evolution by performing an eigenvector/eigenvalue analysis (Bartis and Widom, 1974). The resulting rate coefficient over the temperature range 260-300 K at a pressure of 1000 hPa is k 4 (T ) = 2.7 × 10 −12 (300 K/T ) 2.66 cm 3 molecule −1 s −1 . Because the intermediate complexes are not strongly bound, and the transition state and products are below the entrance channel, the only products formed in Reaction (R4) under atmospheric conditions are IO + HNO 3 . The uncertainty in k 4 arises principally from the estimated capture rate coefficient (see above) and the height of the barrier below the entrance channel. As discussed above, the spin-orbit correction of the transition state is likely to be larger than the value of −5.9 kJ mol −1 corresponding to HOI, so k 4 is possibly a lower limit. For instance, if the barrier height is decreased by 3 kJ mol −1 , k 4 increases by a factor of 1.9. If the barrier is lower by 8.5 kJ mol −1 (corresponding to the transition state having the same spin-orbit correction as IO), then k 4 would increase by a factor of 5.1. Nevertheless, noting that the capture rate coefficient could be lower -perhaps by a factor of 2 -than the estimate used here, we prefer to use the value for k 4 calculated with the potential surface in Fig. 3. Of course, if k 4 is larger, then the atmospheric impacts of Reaction (R4) discussed in Sect. 4 will be even more pronounced.
Note that NO 3 also reacts with CH 2 I 2 with a rate constant ∼ 2-4 × 10 −13 cm 3 molecule −1 s −1 , which can have a significant effect on nighttime CH 2 I 2 concentration . However the products of this reaction are still uncertain (Nakano et al., 2006;Carpenter et al., 2015) and its rate is considerably slower than that of Reaction (R4).
In summary, the only likely gas-phase reactions that I 2 and HOI undergo in the nighttime troposphere are Reactions (R1) and (R4), respectively. These are included in the model reaction scheme to examine their impacts on the evolution of iodine species in the atmosphere.

Atmospheric modelling
We use two atmospheric chemical transport models to study (i) the impact of this new chemistry on the nighttime chemistry and partitioning of iodine species and (ii) the resulting geographical distribution of nocturnal iodine and impact on NO 3 within the global marine boundary layer.
The first model, Tropospheric HAlogen chemistry MOdel (THAMO) , is used for a detailed kinetics study of the impact of the different reactions shown in Table 1 as well as to assess which uptake rates best reproduce observations from a field study at the CVAO (Carpenter et al., 2011). THAMO has been used in the past to study iodine chemistry at the CVAO, and further details including the full chemical scheme can be found elsewhere Mahajan et al., 2009Mahajan et al., , 2010aLawler et al., 2014) (Read et al., 2008Lee et al., 2009Lee et al., , 2010. The model is initialized at midnight and the evolution of iodine species, O 3 , NO x and HO x is followed until the model reaches steady state. The second model is the global 3-D chemistry-climate model CAM-Chem (Community Atmospheric Model with chemistry, version 4.0), which is used to study the impact of Reaction (R1) and (R4) on a global scale. The model includes a comprehensive chemistry scheme to simulate the evolution of trace gases and aerosols in the troposphere and the stratosphere . The model runs with the iodine, bromine and chlorine chemistry schemes from previous studies Saiz-Lopez et al., 2014, including the photochemical breakdown of bromoand iodocarbons emitted from the oceans  and abiotic oceanic sources of HOI and I 2 (Prados-Roman et al., 2015a). CAM-Chem has been configured in this work with a horizontal resolution of 1.9 • latitude by 2.5 • longitude and 26 vertical levels, from the surface to ∼ 40 km altitude. All model runs in this study were performed in the specified dynamics mode  using offline meteorological fields instead of an online calculation, to allow direct comparisons between different simulations. This offline meteorology consists of a high-frequency meteorological input from a previous free running climatic simulation.
It should be noted that during nighttime the uptake on aerosols of emitted species such as I 2 and HOI, and the uptake of reservoir species such as IONO 2 , can play a major role in the cycling of iodine. Observations at CVAO show that I 2 peaked at about 1 pmol mol −1 during the night and that ICl was not detected above the 1 pmol mol −1 detection limit of the instrument (Lawler et al., 2014). In order to match these observations, we need to reduce the uptake and heterogeneous recycling of iodine species. The uptake rates of chemical species on the background sea-salt aerosols are determined by their uptake coefficients (γ ). The database of mass accommodation and/or uptake coefficients is rather sparse and essentially limited to I 2 , HI, HOI, ICI and IBr on pure water/ice and on sulfuric acid particles (Sander et al., 2006). Other iodine species which are likely to undergo uptake onto aerosol are OIO, HIO 3 , INO 2 , IONO 2 and I 2 O 2 (Saiz-Lopez et al., 2012a;Sommariva et al., 2012). Uptake of HOI is very uncertain, with γ (HOI) ranging from 2 × 10 −3 to 0.3 depending on the surface composition and state (Holmes et al., 2001). Sommariva et al. (2012) assumed γ (HOI) to be 0.6, similar to the value for HOBr measured by Wachsmuth Figure 5. THAMO modelled diurnal variation of IO, NO 3 and the IONO 2 . The right-hand panels are from Scenario 1, which do not include nighttime reactions of HOI and I 2 with NO 3 , while the lefthand panels include the reactions in Scenario 2. et al. (2002). In the case of IONO 2 , the uptake coefficient has not been measured, with most models using values of 0.1 (von Glasow et al., 2002;Saiz-Lopez et al., 2008;Mahajan et al., 2009Mahajan et al., , 2010aLeigh et al., 2010;Sommariva et al., 2012;Lawler et al., 2014). The modelled levels of I 2 and ICl change with different values of uptake coefficients. To match the CVAO I 2 and ICl observations (Lawler et al., 2014), we have used γ = 0.01 for HOI and IONO 2 , which is within the uncertainty in the literature, and assumed that 80 % is recycled as I 2 . Further measurements of these dihalogen species are needed to better constrain their heterogeneous recycling on sea-salt aerosols.

Results and discussion
Of the possible nocturnal iodine activation reactions involving the inorganic iodine source gases I 2 and HOI, only Reactions (R1) and (R4) appear to be likely candidates (see Sect. 3). We therefore designed two modelling scenarios: Scenario 1 (S1), without nighttime reactions of I 2 or HOI with NO 3 ; and Scenario 2 (S2), including Reactions (R1) and (R4) for the degradation of HOI and I 2 by NO 3 . In the onedimensional model THAMO, the I 2 and HOI are injected into the atmosphere from the ocean surface using the flux parameterizations derived from laboratory experiments (Carpenter et al., 2013;MacDonald et al., 2014). Figure 4 shows the resulting diurnal evolution of the HOI and I 2 mixing ratios in the two scenarios, after 2 days of simulation time. The I 2 mixing ratio peaks during the night in both the scenarios due to quick loss by photolysis during the daytime. By contrast, HOI is present during daytime due to its production through the reaction of IO with HO 2 and peaks just before sunset. In the first scenario, without the inclusion of Reactions (R1) and (R4), Fig. 4 (panels on right-hand side) shows that I 2 builds up during the night, reaching a concentration peak just before dawn. This is especially noticeable as the daytime concentrations are much lower than during the night. On the other hand, HOI concentrations decrease during night until dawn, when they drop to zero. For both species, inclusion of reactions with NO 3 causes a decrease in their respective nocturnal concentrations (Fig. 4, panels on left-hand side). The inclusion of Reactions (R1) and (R4) also leads to a modelled I 2 concentration which is in better agreement with the observations of the molecule made at CVAO (Lawler et al., 2014), reaching peak values of about 1 pmol mol −1 , as compared to about 3 pmol mol −1 for the scenario without nighttime reactions. An additional consequence of including Reactions (R1) and (R4) is the significant increase of the seaair fluxes of HOI and I 2 at night due to their atmospheric removal by NO 3 (Fig. 4, bottom panel). Figure 5 shows the diurnal evolution of IO, NO 3 and IONO 2 in both model scenarios after 2 days of simulation time. Although the daytime peak values of IO are well reproduced in both scenarios, reaching about 1.5 pmol mol −1  around noon similar to the ground-based observations (Read et al., 2008), the inclusion of Reactions (R1) and (R4) leads to the removal of the dawn spike in IO, which is predicted by current iodine models but was not observed at CVAO   (Read et al., 2008;Mahajan et al., 2010a). The IO dawn spike predicted by models is due to a buildup of the emitted I 2 and HOI (which is converted into IBr / ICl through heterogeneous recycling) over the night, followed by rapid photolysis after first sunlight. However, due to the considerable removal of HOI and I 2 through the night due to reaction with ambient NO 3 , this spike does not appear in the second scenario, leading to a modification of the diurnal profile of IO that better matches with observations. Reactions (R1) and (R4) also reduce the NO 3 mixing ratio (Fig. 5, middle panels). In Scenario 1, the NO 3 is modelled to peak at about 14 pmol mol −1 just before dawn. However, the inclusion of Reactions (R1) and (R4) leads to near complete depletion of NO 3 close to the surface, with the peak level at the surface reaching only 2 pmol mol −1 , since Reactions (R1) and (R4) become the main atmospheric loss processes for NO 3 in the lower MBL. These reactions lead however to the buildup of IONO 2 during the night (Fig. 5, bottom panels). In the absence of Reactions (R1) and (R4), significant levels of IONO 2 are seen only at dawn and dusk since no other reactions produce IONO 2 at night, and during the day IONO 2 is removed by photolysis. However, with continuous conversion of I 2 and HOI to IONO 2 by Reactions (R1) and (R4) in Scenario 2, IONO 2 is modelled to reach up to 3 pmol mol −1 in the nocturnal MBL.
Given the associated uncertainty in the theoretical estimate of the k 4 , we used THAMO to assess the sensitivity of surface NO 3 to k 4 . Figure 6 shows that NO 3 peak nighttime concentration is in fact highly coupled to k 4 , with the expected uncertainty in k 4 of 1 order of magnitude (see above) giving rise to a factor-of-2 change in NO 3 . A laboratory measurement of k 4 should therefore be undertaken in the future.
We now implement the nighttime reactions in the 3-D global model (CAM-Chem) to assess the resulting geographical distributions and impacts of these reactions. We have also run two different scenarios in CAM-Chem, the first without Reactions (R1) and (R4) in the chemical scheme, and the second including the new nighttime iodine chemistry. Figure 7 shows how the inclusion of Reactions (R1) and (R4) reduces globally the nighttime concentrations of I 2 and HOI. The plots correspond to the nighttime averaged (from 00:00 to 01:00 LT (local time)) differences between the model scenarios. Considerable reductions of up to 0.5 and 10 pmol mol −1 (i.e. up to 100 % removal) are observed for I 2 and HOI, respectively, particularly over coastal polluted regions where continental pollution outflow leads to higher levels of NO 3 in the nighttime MBL. Major shipping routes also show strong nocturnal iodine activity due to the characteristically high NO x , and resulting NO 3 , associated with shipping emissions. Figure 8 shows the effect of this nocturnal chemistry on the concentrations of IONO 2 and NO 3 . As in the previous figure, the plots correspond to the nighttime averaged difference between the second and the first scenarios. The maps show an increase of IONO 2 of up to 15 pmol mol −1 (∼ 600 %) over polluted coastal areas, due to efficient conversion of NO 3 into IONO 2 . The bottom panel of Figure 7 shows the expected decrease of NO 3 levels associated with the inclusion of Reactions (R1) and (R4), with decreases of up to ∼ 4 pmol mol −1 (up to 60 %) over marine polluted regions. We model global percentage reductions in the NO 3 concentrations of 7.1 % (60 • S-60 • N), with nitrate removal of up to 80 % in nonpolluted remote oceanic regions with low NO 3 levels. This in turn can affect the modelled oxidation of DMS by NO 3 . We estimate that the reduction in NO 3 , due to the inclusion of Reactions (R1) and (R4), results in a model increase in DMS levels of up to 7 pmol mol −1 (about 20 %) in marine regions affected by continental pollution outflow (Fig. 9). We therefore suggest that the inclusion of the new nighttime iodine chemistry can have a large, so far unrecognized, impact on the nocturnal oxidizing capacity of the marine atmosphere.
The hourly evolution of the main species involved in this study is shown in Figs. 10 and 11, which include the levels of HOI, I 2 , IONO 2 and NO 3 in the MBL over regions where nocturnal iodine is modelled to be particularly active. The first region is located within the Mediterranean Sea, an area that shows large differences during the summer months when high ozone levels drive large emissions of HOI and I 2 from the sea, and the high levels of NO 3 at nighttime make this chemistry especially important. The hourly average in August is shown in Fig. 10 for HOI, IONO 2 and I 2 . HOI and IONO 2 (Fig. 10) are the species whose concentration differ most between scenarios as HOI is removed and IONO 2 produced by Reaction (R4) (and, to a lesser extent, Reaction R1). Over the Pacific Ocean at the south of the Baja California Peninsula, the modelled differences between the two scenarios are even higher than over the Mediterranean Sea (Fig. 11). Large differences in MBL NO 3 , up to 28 %, are modelled during the night caused by pollution outflow from the west coasts of Mexico and USA.

Summary and conclusions
The viability of the reaction of HOI with NO 2 , HNO 3 and NO 3 has been studied by theoretical calculations. The results indicate that only the reaction of HOI with NO 3 , to yield IO + HNO 3 , is possible under tropospheric conditions. The inclusion of this reaction, along with that of I 2 + NO 3 , has a number of significant implications: -Nocturnal iodine radical chemistry is activated.
-This causes enhanced nighttime oceanic emissions of HOI and I 2 .
-Nighttime iodine species are partitioned into high levels of IONO 2 .
-The IO spike, modelled by current iodine models but not shown by observations, is removed.
-A reduction of the levels of nitrate radical in the MBL, with the associated less efficient oxidation of DMS, has important implications for our understanding of the nocturnal oxidizing capacity of the marine atmosphere.

Data availability
Data are available upon request to the corresponding author.
The Supplement related to this article is available online at doi:10.5194/acp-16-15593-2016-supplement.