Chlorine activation indoors and outdoors via surface-mediated reactions of nitrogen oxides with hydrogen chloride

Abstract

Gaseous HCl generated from a variety of sources is ubiquitous in both outdoor and indoor air. Oxides of nitrogen (NO_y) are also globally distributed, because NO formed in combustion processes is oxidized to NO₂, HNO₃, N₂O₅ and a variety of other nitrogen oxides during transport. Deposition of HCl and NO_y onto surfaces is commonly regarded as providing permanent removal mechanisms. However, we show here a new surface-mediated coupling of nitrogen oxide and halogen activation cycles in which uptake of gaseous NO₂ or N₂O₅ on solid substrates generates adsorbed intermediates that react with HCl to generate gaseous nitrosyl chloride (ClNO) and nitryl chloride (ClNO₂), respectively. These are potentially harmful gases that photolyze to form highly reactive chlorine atoms. The reactions are shown both experimentally and theoretically to be enhanced by water, a surprising result given the availability of competing hydrolysis reaction pathways. Airshed modeling incorporating HCl generated from sea salt shows that in coastal urban regions, this heterogeneous chemistry increases surface-level ozone, a criteria air pollutant, greenhouse gas and source of atmospheric oxidants. In addition, it may contribute to recently measured high levels of ClNO₂ in the polluted coastal marine boundary layer. This work also suggests the potential for chlorine atom chemistry to occur indoors where significant concentrations of oxides of nitrogen and HCl coexist.

Gaseous HCl levels reaching concentrations of a few parts-per-billion (ppb) (vol:vol) have been measured in polluted air and in some indoor settings (1–6). Direct emissions include garbage burning (7), incineration of municipal and medical wastes, burning of biomass, agricultural products and coal, and industrial processes, e.g., semiconductor and petroleum manufacturing (5). Natural sources in air include volcanic eruptions and reactions of sea salt and organochlorine compounds (5). Removal indoors and in the boundary layer is largely by deposition. Because many of these HCl sources involve combustion or occur in polluted urban areas, oxides of nitrogen are typically present simultaneously.

Although heterogeneous nitrogen oxide reactions on airborne particles and boundary layer surfaces are known to be important in the atmosphere, their kinetics and mechanisms remain elusive (8). For example, the hydrolysis of NO₂ on surfaces,

generates gas phase nitrous acid (HONO), a major OH source in continental regions (8, 9), and HNO₃. Nitric acid and other, as yet unidentified, oxides of nitrogen, NO_y (NO_y = NO + NO₂ + HNO₃ + N₂O₅ + …), are strongly adsorbed on surfaces and when they are irradiated, generate HONO, NO and NO₂ (10–13), even in supposedly “clean” systems (14). Based on a variety of experimental studies, the surface complex NO⁺NO₃⁻ has been proposed to be a key intermediate (8) in the hydrolysis of NO₂ and the precursor of HONO in this system:

Formation of NO⁺NO₃⁻ is thought to be from autoionization of asymmetric N₂O₄ (ONONO₂), possibly formed by sequential uptake and reaction of NO₂ on the surface (15). Recent theoretical studies (16) show that once ONONO₂ is formed, it is converted within femtoseconds to NO⁺NO₃⁻; based on experimental studies of NO₂ on ice films, conversion of NO⁺NO₃⁻ to HONO via reaction [3] is also fast (17, 18). To the best of our knowledge, there are no reports of other reactions of the ion pair that compete with the reaction with water.

In the case of N₂O₅, autoionization to NO₂⁺NO₃⁻ has also been proposed as a key intermediate step (19, 20), and rapid reaction with water also occurs, forming nitric acid:

This article presents a combination of experiments and theory that demonstrate new chemistry in which surface-bound oxides of nitrogen from the uptake of NO₂ or N₂O₅ react with gaseous HCl to form ClNO and ClNO₂, respectively, potentially harmful gases (21) that photolyze to form chlorine atoms. We show that uptake of oxides of nitrogen and HCl on surfaces is not necessarily a permanent removal mechanism, but rather can be an intermediate step on the way to generating more reactive gases. Fundamental molecular insights gained from theoretical calculations confirm the experimental observations that water has a remarkable effect on this chemistry and imply that this chemistry is likely under atmospheric conditions, both outdoors and indoors. Indeed, our airshed model calculations demonstrate the potential importance of this novel chemistry in polluted marine regions where HCl from sea salt reactions and NO_y from urban emissions coexist. It is clear from this work that uptake of oxides of nitrogen and HCl on surfaces may contribute to chemistry and photochemistry in previously unrecognized ways both indoors and outdoors, including in coastal regions and downwind of waste, coal and biomass combustion, and industrial processes that generate HCl.

Results and Discussion

NO₂ Surface Reaction.

The experiments use a flow system designed to deliver controlled amounts of NO₂, HCl and water vapor in a stream of nitrogen to one of a set of parallel borosilicate glass reaction chambers before passing through an optical cell in a Fourier transform infrared (FTIR) spectrometer. One chamber contains silica (SiO₂) powder pressed into pellets, which provide large surface areas for the heterogeneous chemistry, whereas the second, empty chamber acts as a bypass. Silica powder is selected to mimic silicate-rich urban surfaces and some of the oxide surfaces present in mineral dust (8, 22, 23), which are important substrates for heterogeneous tropospheric chemistry. Fig. 1A shows that ClNO is formed efficiently when a stream of reactants passes through the chamber containing SiO₂, but not the bypass. The likely mechanism involves reaction of NO⁺NO₃⁻ with HCl:

The average ratio of ClNO formed to NO₂ reacted, Δ[ClNO]/Δ[NO₂], is 0.39 ± 0.01 (2 s) at a relative humidity of 0.5%. The reactor containing SiO₂ has ≈10⁴ times more surface area than the bypass, and its effect on ClNO formation confirms the heterogeneous nature of the reaction, which is extremely slow in the gas phase (24).

Fig. 1.

Reaction of HCl with NO₂ on SiO₂ studied by infrared spectroscopy. (A) Nitrosyl chloride (ClNO) is formed efficiently when a stream of gas containing 3.7–5.5 ppm (vol:vol) NO₂, 5–10 ppm (vol:vol) HCl and H₂O (≈0.5% relative humidity) flows over a bed of high-surface area (330 m²) SiO₂ (shaded regions); the reaction does not occur when the stream is diverted through an empty reaction cell (clear regions). (B) The concentration of ClNO formed per NO₂ reacted increases from 25% in the absence of added water vapor to 50% at a relative humidity (RH) ≥5%; the triangle marks the average ClNO yield derived from the experiment described in panel A. The maximum yield expected based on reaction [5] is 0.5. (C) The addition of water vapor has a dramatic effect on gaseous NO₂ initially present in a reaction cell containing SiO₂ pellets. Sequential addition of water vapor (1.6, 3.1, and 3.2 Torr, indicated by green regions) leads to enhanced uptake of NO₂ on the SiO₂ surfaces and formation of nitrous acid (HONO). Introducing HCl to the chamber (blue region) leads to rapid ClNO formation with a yield of 47%, relative to the amount of NO₂ initially present. Units of concentration are parts per million by volume (ppmv).

To study the relative humidity dependence, a cell used in static mode (i.e., where reactants are added to the cell and allowed to react in situ) is positioned in the FTIR such that the IR beam interrogates the gas over the top of a bed of SiO₂ pellets. Nitrogen dioxide is introduced from a bulb on the attached vacuum line and anhydrous HCl or a mixture of HCl and water vapor is then added. As shown in Fig. 1B, Δ[ClNO]/Δ[NO₂] is 0.25 under anhydrous conditions, but doubles to the theoretical yield implied by reactions [2] and [5] when the relative humidity is >5% (3 × 10¹⁶ molecules cm⁻³). Not only does the conversion of NO₂ to ClNO occur at relative humidities (RH) found in the atmosphere, but surprisingly, water actually enhances the reaction despite the possibility of competition from reaction [3]. A catalytic role of water in thermal and photochemical reactions has been reported for other systems (25–27). However, to the best of our knowledge it has not been reported for a case involving thin water films on a solid substrate where water can also participate in a competing reaction. It is also notable that separate experiments (see SI Text) show that water vapor does not enhance the heterogeneous hydrolysis of ClNO in this system. This suggests that under some conditions, loss of ClNO to aerosol particles and surfaces at the terrestrial-air interface may be much slower than what is expected to occur in bulk water (28, 29).

Fig. 1C shows the effect of adding water alone to a reaction chamber containing NO₂ and SiO₂ pellets. The gas-phase concentration of NO₂ decreases with each addition of water vapor as it is taken upon the SiO₂ surface, facilitating NO₂ uptake, and forming gas phase HONO via reaction [3]. When HCl is added, ClNO is formed rapidly, indicating that under these conditions reaction [5] competes with reaction [3]. Although these experiments are performed using higher mixing ratios than found in the atmosphere, the fact that similar ClNO yields are obtained with NO₂ concentrations varying over 2 orders of magnitude (Fig. 1 A–C) is a good indication that this chemistry will occur at lower concentrations typically found in air both indoors and outdoors.

Because HONO is known to react with HCl heterogeneously to form ClNO (30, 31), the possibility that HONO is initially formed in reaction [3] and then reacts further with HCl to form ClNO must be considered. This can be ruled out because the amount of ClNO that is possible from the direct reaction of HCl with gaseous HONO at 80 min (Fig. 1C) is <3% of the reacted NO₂, much less than observed. This shows that the ClNO precursor must be adsorbed to the SiO₂ surface before the introduction of HCl. Theoretical studies predict that HONO adsorbs to SiO₂ surfaces (32), so that the gaseous HONO observed in Fig. 1C could be in equilibrium with far greater amounts of surface-adsorbed HONO or nitrite (although given that HNO₃ is formed simultaneously, HONO will be the major species). However, control experiments in which SiO₂ is exposed to gaseous HONO followed by HCl do not produce ClNO (see SI Text and Fig. S1), ruling this pathway out as a major source.

The proposed mechanism and the effect of water in enhancing ClNO formation are supported by theory. A series of high level ab initio calculations [using the General Atomic and Molecular Electronic Structure System (GAMESS) (33)] are performed to gain insight into the role of water in ClNO formation. The starting reactive species in the calculations is the asymmetric dimer, ONONO₂ in gas-phase clusters comprised of HCl and a varying number of water molecules. Previous studies have shown that gas-phase clusters are suitable models for surface reactions where the substrate is not involved in the chemical mechanism (34). Molecular dynamics “on-the-fly” calculations [using second-order perturbation theory (MP2) with the cc-pVDZ basis set, denoted MP2/cc-pVDZ], are performed to elucidate the mechanism (Fig. 2). Water acts as a conduit to transfer a proton from HCl to NO₃⁻, facilitating the formation of the products ClNO and HNO₃. Energetics are attained at the coupled cluster [CCSD(T)/cc-pVTZ] level of theory by conducting single point energy calculations on the stationary structures obtained on the MP2/cc-pVDZ surface. In the absence of water, the activation energy for ClNO formation is 11.5 kcal·mol⁻¹. Addition of 1 water molecule lowers the activation energy to 5.3 kcal·mol⁻¹. The formation of ClNO is exothermic by 10.0 kcal·mol⁻¹ in the absence of water and 12.9 kcal·mol⁻¹ in the presence of 1 water molecule. A barrierless channel is found for ClNO formation in the presence of 2 water molecules at the MP2/cc-pVDZ level of theory. Thus, the theoretical results support the experiments described above and demonstrate the critical role of water in enhancing the formation of ClNO.

Fig. 2.

Snapshots along an ab initio molecular dynamics “on-the-fly” trajectory of ClNO formation show how the reaction of HCl and ONONO₂ is catalyzed by 1 water molecule. (A) Weakly bound reactive complex formed upon geometry optimization of ONONO₂, HCl and H₂O. (B) First proton transfer from HCl to H₂O to form Cl^- and H₃O⁺. (C) Second proton transfer from H₃O⁺ to NO₃⁻ to form H₂O and HNO₃. (D) Product ClNO is formed. Calculations are carried out at the MP2/cc-pVDZ level of theory. Depicted atomic distances are given in Ångströms. The amount of time elapsed along the trajectory is provided below each structure.

N₂O₅ Reaction.

Gaseous HCl may also react with other oxides of nitrogen such as N₂O₅ on surfaces to form ClNO₂ (35). The reaction likely occurs through an analogous mechanism to that for NO₂, whereupon absorption to the surface, N₂O₅ autoionizes to NO₂⁺NO₃⁻ and then reacts with HCl,

in competition with water, reaction [4]. The gas-phase reaction is otherwise slow in the absence of sufficient surface area (24). Spectral evidence for the role of NO₂⁺ as the key intermediate in reaction of HCl with N₂O₅ was obtained using attenuated total reflection FTIR (ATR-FTIR) spectroscopy. As shown in Fig. 3, sharp bands at 2,370 cm⁻¹ and 1,667 cm⁻¹, and a weak absorbance between 3,500 and 2,500 cm⁻¹ appear in the spectrum measured immediately after the internal reflection element (IRE) is exposed to 10 Torr of N₂O₅ (solid line). The band at 2,370 cm⁻¹ is assigned to the ν₃(NO₂-antisymmetric) stretching mode of linear NO₂⁺, consistent with previous observations of this ion in the gas phase and at low temperatures in frozen matrices and on metallic substrates (35–39). The band at 1,667 cm⁻¹ is due to the ν₂(NO₂-antisymmetric) stretch of HNO₃. Broad bands between 3,500 and 2,500 cm⁻¹ are assigned to ν(O–H) stretches of nitric acid complexed to surface-adsorbed water (40). The weak signals typically observed for surface-adsorbed species using ATR-FTIR required the use of higher concentrations than found in the atmosphere. However, ionization of N₂O₅ on surfaces is expected to be operational at lower concentrations as well (19, 20).

Fig. 3.

Nitronium nitrate (NO₂⁺NO₃⁻) is the key intermediate in reaction of N₂O₅ with HCl on surfaces. ATR-FTIR was used to obtain the spectra of species adsorbed to a ZnSe crystal exposed to 10 Torr of N₂O₅ (solid line) and subsequently 20 Torr of HCl (dashed line) in the absence of added water vapor. (Inset) An expanded spectral region where the NO₂⁺ infrared absorption band occurs.

Similar to the NO₂ case where water promotes the formation of NO⁺NO₃⁻ from ONONO₂ (16), water is expected to play an essential role in autoionization of N₂O₅. Indeed, calculations at the MP2/cc-pVDZ level of theory show that ionization of N₂O₅ into a NO₂⁺NO₃⁻ ion pair takes place in the presence of 1 or more water molecules (Fig. 4). The largest effect is observed upon addition of the first water molecule, which triggers an increase in the separation of the NO₂^δ+ and NO₃^δ− moieties by 0.2 Å compared with free N₂O₅. Furthermore, the calculated partial charges on these 2 moieties increase dramatically from δ(NO₂^δ+) = +0.127 and δ(NO₃^δ−) = −0.127 in the free N₂O₅ molecule, to δ(NO₂^δ+) = +0.274 and δ(NO₃^δ−) = −0.284 in the (N₂O₅)·(H₂O) complex. This trend of increasing separation and partial charges on NO₂^δ+ and NO₃^δ− moieties continues upon addition of the second and third water molecules. Hence, water enhances the reactivity of N₂O₅ by promoting the formation of the (NO₂⁺)·(NO₃⁻) ion pair.

Fig. 4.

Geometry optimizations of N₂O₅ in the absence and the presence of 1, 2 and 3 water molecules conducted at MP2/cc-pVDZ level of theory show how ionization and dissociation of N₂O₅ occurs on water clusters. Depicted atomic distances are given in Ångströms; Mulliken charges on the NO₂^δ+ and NO₃^δ− moieties are provided in the table.

The IR absorption band stemming from NO₂⁺ is completely removed and nitric acid bands increase when the IRE is exposed to excess HCl in the ATR-FTIR experiment shown in Fig. 3 (dashed line), supporting the mechanism where surface-adsorbed NO₂⁺ is the key intermediate in the conversion of N₂O₅ to ClNO₂ (19, 20). In separate experiments (Fig. 5A) using transmission FTIR spectroscopy, ClNO₂ is clearly formed in the gas-phase when HCl is added to a cell containing N₂O₅. Again, the reaction is more efficient in the presence of water, as demonstrated in Fig. 5B.

Fig. 5.

Reaction of N₂O₅ with HCl in absence and then in the presence of added water vapor. (A) Nitryl chloride (ClNO₂) is formed essentially quantitatively (96% yield) as 300 ppm of anhydrous HCl is added at t = 0 to a reaction chamber (volume 64 cm³) containing 126 ppm N₂O₅. Rapid uptake of N₂O₅ onto walls leads to the drop in N₂O₅ levels initially. (B) The rate of reaction of N₂O₅ (300 ppm) with HCl vapor (400 ppm) is increased in the presence of water vapor (≈25% relative humidity). Note the different time scales in A and B. Units of concentration are parts per million by volume (ppmv). (C) Ab initio calculations show that the reaction between HCl and NO₂⁺ is catalyzed by water. Minima are obtained by optimizations carried out in the absence and presence of 1 and 2 water molecules at MP2/cc-pVDZ level of theory. The number of water molecules was increased by placing an additional water molecule 4 Å apart from previously obtained structure. Depicted atomic distances are given in Ångströms.

Theory again supports the experimental observations and proposed mechanism, showing that water aids the reaction of NO₂⁺ with HCl. The reaction was investigated theoretically by conducting a series of optimizations of the (NO₂⁺)·(HCl) complex in the absence and presence of water at the MP2/cc-pVDZ level of theory. As shown in Fig. 5C, a barrierless channel for the formation of ClNO₂ from NO₂⁺ and HCl is found in the presence of 2 water molecules. Upon addition of the second water molecule the product ClNO₂ is formed having a Cl–N bond length of 2.22 Å and a ∢O–N–O of 144°, comparable to the 1.96 Å and 135° found for the calculated free molecule. Water clearly facilitates the formation of ClNO₂, similar to the trend observed in the ClNO case. There is some similarity to the reactions of ClONO₂ or N₂O₅ with HCl on bulk ice at low temperatures (54); in those cases, Cl⁻ from the dissociation of solvated HCl is the reactive species (35, 41, 42). However, acids at the top of the air-water interface are largely undissociated (34, 40, 43–45). Given this and the fact that water adsorbed on solid substrates at room temperature does not behave like water in the bulk (46–48), it is likely that the reactant in the surface-mediated reaction reported here is molecular HCl.

Atmospheric Implications.

Extrapolation to atmospheric conditions both outdoors and indoors relies on similar surface-bound nitrogen oxides being present on real surfaces. Evidence for the universality of reaction [1] is that it has been shown to occur on many different surfaces, including Teflon, borosilicate glass and silica (8), boundary layer surfaces outdoors [e.g., vegetation, building materials etc. (49, 50)], on typical indoor surfaces (51, 52) and on airborne dust particles (50, 53). It is also well known that N₂O₅ is taken up on surfaces and hydrolyzed (54). Due to such uptake, there is a reservoir of strongly adsorbed reactive oxides of nitrogen on surfaces in both environmental chambers (14) and in the field (13).

Experiments were performed at RH up to 25%, typical of drier environments. However, the role of water suggests that this chemistry will continue to higher RH. When gaseous HCl is present as well as oxides of nitrogen, e.g., in coastal areas and downwind of certain industrial settings, incineration facilities (5), biomass burning (55), in buildings such as medieval churches and in volcanic plumes (2, 5), it will be converted to ClNO and ClNO₂. ClNO absorbs light well into the visible region (Fig. 6), forming Cl + NO with unit quantum yield (54). As seen in Fig. 6, its absorption cross section in the near UV overlaps strongly not only with solar radiation but also with that from typical fluorescent lights used indoors. Formation of ClNO indoors is certainly possible. Indoor surfaces are frequently exposed to NO_y generated from indoor combustion sources or from an influx of polluted outside air (56). It has been proposed that corrosive chlorine containing gases such as HCl [e.g., from cigarette smoke, decomposition of chlorine-containing polymers, and cleaning agents] are responsible for observed elevated levels of Cl⁻ found in indoor relative to outdoor air (57). Thus, if formed indoors, ClNO could serve as a source of highly reactive chlorine atoms that would participate in the complex chemistry known to occur there (52, 58). In addition, the presence of ClNO in the indoor environment could have important ramifications for the reliability and lifetime of electronics susceptible to corrosion (59). To the best of our knowledge, chlorine atom chemistry has not been considered in typical indoor air environments.

Fig. 6.

The overlap between the UV-vis absorption spectra of ClNO (104, 105) and ClNO₂ (104) and emission spectra from the Sun (106) and fluorescent lamps typical of indoor settings.

Outdoors, ClNO and ClNO₂ are important chlorine atom precursors on short time scales because of their rapid photolysis, having calculated lifetimes of only 5 and 30 min, respectively, at a solar zenith angle of 0° (54). Hydrochloric acid also forms Cl atoms through its reaction with OH, but this is slow, with a lifetime for HCl with respect to this reaction of ≈14 days at an OH concentration of 10⁶ radicals cm⁻³. In the presence of sufficient NO, chlorine atoms enhance the formation of ozone through well-known organic-NO_y cycles (54) so heterogeneous HCl-to-ClNO conversion could speed up the formation of ozone and other photochemically generated species in polluted coastal urban areas, changing their peak concentration and geographical distribution. Ozone is an important trace gas with documented health effects, a greenhouse gas, and a significant atmospheric oxidant and precursor to the OH radical (54).

The potential role of heterogeneous HCl-to-ClNO conversion in a coastal airshed (see Fig. S2) is probed using a model (60) that employs state-of-the-art modules for both gas-phase and heterogeneous/multiphase reaction mechanisms, and dynamic aerosol predictions (60, 61). Although this model represents the Southern California air basin, it is typical of polluted coastal regions elsewhere. Two simulations are examined: A base case that provides benchmark predictions of the ambient concentrations of O₃, ClNO and HCl without the surface mediated conversion, and a test case that assumes that every HCl molecule deposited generates 1 gaseous ClNO molecule. This in effect assumes that HCl is the limiting reagent. Although this might appear to be the extreme case, it is not necessarily the upper limit for conversion of HCl to ClNO because the reaction probability in the model is expressed as a function not only of the deposition velocity of HCl but also of the total surface area of the domain. Surface areas in the model do not include the additional geometric area due to structures such as buildings, nor the molecular scale porosity of surfaces; these can contribute significantly to chemistry in the boundary layer (50). It also assumes that HCl will continue to compete with water vapor for the NO⁺NO₃⁻ intermediate under typical atmospheric conditions. The ratio of HCl/H₂O that was experimentally accessible here was typically of the order of 10⁻³ whereas in air, it is ≈10⁻⁶ to 10⁻⁷. Whether this assumption is justified awaits measurement of the relative rate constants for reactions [3] and [5].

Fig. 7A shows the geographical distribution and concentrations of O₃, ClNO and HCl at the times at which they each peak within the modeling domain. Fig. 7B shows the increase in concentrations, ΔO₃, ΔClNO, and ΔHCl, above those predicted for the base case due to the inclusion of surface-mediated HCl-to-ClNO conversion. The locations with the greatest impact from the new ClNO source are mainly in the downwind regions of the domain and show up to 40 ppb more O₃ (≈20% increase) in a 1-h averaging time relative to the base case, and up to 25 ppb increase over an 8-h averaging time. For comparison, the current U.S. Environmental Protection Agency (EPA) 8-h average standard is 75 ppb. Additional test cases performed with varying conversion probabilities suggest that increases in ozone levels scale linearly with the HCl-to-ClNO conversion probability so that a smaller ClNO yield generates proportionally smaller amounts of O₃ above the base case.

Fig. 7.

Impact of HCl-NO₂ chemistry on ambient air quality in the South Coast Air Basin (SoCAB) of California. Shown are the domain-wide time maxima mixing ratios in parts-per-billion by volume (ppbv) for the base case (A). The peaks occur at 14:00, 6:00, and 12:00 for O₃, ClNO, and HCl, respectively. The increases over the values in A for O₃, ClNO, and HCl at the same hours for O₃, ClNO, and HCl are shown in B, i.e., these are the additional concentrations on top of those predicted for the base case due to inclusion of the heterogeneous HCl-to-ClNO conversion.

It is noteworthy that this heterogeneous chemistry of oxides of nitrogen with gaseous HCl is the inverse of mechanisms of chlorine activation from sea salt particles in coastal areas involving chloride, which can generate a variety of photochemically active chlorine atom precursors (5, 62–64). For example, reactions of gaseous NO₂ and N₂O₅ with chloride ions in sea salt particles are known to generate ClNO and ClNO₂ (20, 63–66). Although the former reaction is likely too slow to generate significant amounts of ClNO in coastal environments, the latter is believed to be responsible for the formation of ClNO₂ measured recently in air (67), where measured mixing ratios of ClNO₂ were greater than expected based on the measured concentrations of N₂O₅ and chloride in sea salt particles. It is possible that the chemistry reported here may have contributed to the measured ClNO₂. This heterogeneous chemistry is also expected to be important around salt lakes such as the Dead Sea (68) and the Great Salt Lake (69), and during dust storms where enhanced HCl uptake has been observed (70). Plumes from biomass (55) and garbage burning (7) are other cases where this chemistry may be significant in outdoor environments.

Unusual chlorine activation near the midlatitude tropopause has been reported that appears to be associated with high particle surface areas, relatively high water vapor concentrations and mixing of tropospheric and stratospheric air (71). Although reaction of HCl with ClONO₂ on polar stratospheric clouds (PSCs) in polar regions is known to generate Cl₂, this unusual midlatitude chemistry does not appear to require PSCs. The heterogeneous reaction of HCl with surface-bound oxides of nitrogen to form ClNO would be consistent with the need for high surface areas and water vapor for midlatitude tropopause chlorine activation.

In short, the combination of experiments and theory reported here suggests a new and unique coupling of surface-mediated nitrogen oxide and halogen activation cycles that will generate ClNO and ClNO₂ in a wide variety of outdoor air environments, and indoors where it has not been considered. Although a technique to measure ClNO₂ at ppt levels has been developed (67), this is not the case for ClNO, but is clearly needed. Similar chemistry is expected for HBr, leading to photolyzable bromine species such as BrNO and BrNO₂ that cause O₃ destruction through well-known cycles (54).

Materials and Methods

Experimental Details.

Infrared spectra were collected using a Thermo Nicolet Avatar 370 Fourier transform infrared (FTIR) spectrometer equipped with a liquid nitrogen-cooled Hg-Cd-Te (MCT) detector. Background single beam spectra were obtained from the average of 200–2,000 interferograms whereas single beam spectra collected during the reaction were obtained from the average of 6–2,000 interferograms; in all cases, spectra were recorded at 1-cm⁻¹ resolution. The IR transmission cells were made of borosilicate glass with an optical path length of either 12 or 10 cm and an inner diameter of 2.5 cm. The ends were closed with germanium windows and sealed with Viton “O”-rings. Concentrations of NO₂, ClNO and H₂O were determined from calibrations measured in our laboratory. Absolute cross sections of nitrous acid from Barney et al. (72) were applied. Hydrogen chloride absorption cross sections were obtained from a reference spectrum available in the Pacific Northwest National Laboratory vapor phase infrared spectral library (73). All experiments were conducted at 22 ± 1 °C under dark conditions.

Gases were handled with an all-glass vacuum line with Teflon stopcocks and Kalrez “O”-rings. Nitrogen dioxide was synthesized from the reaction of NO (Matheson, 99%) with excess oxygen (Oxygen Service, 99.993%), followed by trap-to-trap purification. Nitric oxide was purified by passing it through an acetone/dry ice bath trap at 195 K before use. Nitrosyl chloride (ClNO) was synthesized from the reaction of 1 eq of Cl₂ (Matheson, 99.5%) and slightly more than 2 eq of nitric oxide, followed by repeated freeze-thaw cycles at liquid nitrogen temperatures to drive the reaction to completion in the liquid phase. The vessel containing ClNO was held at 195 K and the excess nitric oxide and chlorine were removed under vacuum until a constant vapor pressure above the ClNO was attained. Nitrosyl chloride purified this way was determined to be ≥99.9% pure by FTIR (42.5 m path length). Dinitrogen pentoxide (N₂O₅) was synthesized by reacting 2 eq of NO₂ with slightly more than 1 eq of ozone (from a 3% mixture in O₂) in a previously conditioned glass bulb equipped with a cold finger. The cold finger was cooled to 195 K to condense N₂O₅; oxygen and unreacted ozone were removed under vacuum. The N₂O₅ was stored at 195 K in a storage tube under vacuum and in the dark. The product was ≈95% pure (determined by FTIR, 10 cm path length) with the main impurities being HNO₃ and NO₂. Nitrous acid was generated by the method of Wingen et al. (31). Silica pellets were formed by hand pressing SiO₂ powder (Cab-O-Sil, Cabot) in a precleaned stainless steel press. The SiO₂ powder was baked overnight at 475 °C before pressing to remove adsorbed organic impurities and a new batch of pellets were used for each experiment.

Studies of the NO₂ + HCl reaction.

The flow system used to study the reaction of HCl with surface-adsorbed NO₂ was constructed of Teflon PFA and designed to deliver a controlled amount of NO₂ (5% in Ultrapure grade N₂, Scott-Marrin) and HCl (from the vapor over a 5 M aqueous solution at 22 °C) in a stream of nitrogen to 1 of 2 parallel borosilicate glass reaction chambers (each with V = 64 cm³) before finally passing through the IR cell (12 cm path length) at a flow rate of 75 cm³·min⁻¹. One chamber was filled with 1.0 g of fumed SiO₂ (Cab-O-Sil from Cabot) which had been pressed into pellets and broken into pieces that were ≈5 mm in each dimension. The total fumed SiO₂ surface area was measured previously using the Brunauer-Emmett-Teller (BET) method and found to be 329 m²·g⁻¹ (74), resulting in a total surface area of 329 m² inside the reactor. The second chamber (geometric surface area of 0.01 m²) was empty and acted as a bypass to allow measurement of the initial concentrations of NO₂, HCl, and H₂O in the reactant mixture. Single beam spectra were continuously collected as the stream of NO₂ (5 ppm), HCl (10 ppm) and water vapor (≈0.5% relative humidity measured using the infrared absorbance of selected water lines) was alternated between flowing through the empty or SiO₂-filled chamber. Uptake of NO₂ on the SiO₂ surface decreases somewhat over time, likely due to the dehydration of the surface with repeated exposures. In addition, water may be tied up in complexes of (HNO₃)·(H₂O)_n as nitric acid (40) accumulates on the surface according to reaction [1].

Measurements of the relative humidity dependence of the ClNO yield from the HCl + NO₂ reaction were carried out in a static system using a 10-cm path length IR cell (V = 43 cm³) containing 1.0 g of fumed SiO₂ pellets. The IR cell was evacuated overnight at ≈10⁻⁴ Torr and heated at 125 °C to drive off most of the surface-adsorbed water before each experiment. After cooling to room temperature, NO₂ was introduced from a bulb on the attached vacuum line. After ≈10 min, the IR cell was opened for 10 s to an attached 493-cm³ bulb containing anhydrous HCl (99.995%, Matheson) or a mixture of HCl and water vapor. The concentration range of NO₂ used in these experiments was 35–200 ppm; the amount of HCl added to the IR cell was between 150–800 ppm. These experiments were limited to ≤20% RH because of the strong absorption of water vapor in the infrared, and condensation of water in the mixing bulb used to prepare the HCl/H₂O mixture that would be needed to give higher RH once it was expanded into the reaction cell.

Studies of the N₂O₅ + HCl reaction.

Before experiments, the surface inside an empty 10-cm path length IR cell is conditioned by repeated exposure to ≈1 Torr of N₂O₅ from a preconditioned 4-L glass bulb, followed by evacuation for 1 h at ≈10⁻⁴ Torr to eliminate adsorbed water and reduce the loss rate of N₂O₅ to the walls. For the reaction, N₂O₅ is added to the IR cell. After ≈10 min, anhydrous HCl or a mixture of HCl and water vapor is added from a 493-cm³ glass bulb attached to the IR cell. Addition of HCl occurs at t = 0 in Fig. 5 A and B. Spectra of surface-adsorbed NO₂⁺ before and and after the reaction with HCl (Fig. 3) are obtained with an ATR probe (Axiom Analytical) mounted in the sample compartment of the spectrometer and inserted into a vacuum-tight borosilicate glass reaction chamber (80 cm³). The internal reflection element (IRE) installed at the end of the probe is a ZnSe rod, 0.6 cm dia. × 4 cm long with 45° conical ends. Interior surfaces of the reaction chamber are first conditioned by repeated exposure to ≈10 Torr of N₂O₅, followed by evacuation for 15 min at ≈10⁻⁴ Torr. The IRE is then exposed to 10 Torr of anhydrous N₂O₅ for 30 min before introducing 20 Torr of anhydrous HCl from a bulb on the attached vacuum line.

Computational Details.

Ab initio calculations (33, 75) are performed with second-order Møller-Plesset (76–78) (MP2) perturbation theory using the cc-pVDZ basis set (79, 80). Geometry and saddle point optimizations (81–83) are carried out with the largest component of the analytic gradient (76, 84) being smaller than 10⁻⁵ Hartree/Bohr. Double differencing of analytic gradients is used in the Hessian calculations (85). Minima are confirmed by an all-positive Hessian, whereas transition states have only 1 negative Hessian eigenvalue. Intrinsic reaction coordinate calculations (86–90) are conducted to connect the transition state with the corresponding minima. Molecular dynamics calculations (91–95) “on-the-fly” that have been successfully applied in related studies (34) are performed starting from the located minimum structure that is obtained upon the geometry optimization of ONONO₂, H₂O and HCl.

To attain high accuracy of reported activation energies and reaction enthalpies, single point energy coupled cluster calculations [CCSD(T) (96, 97)] with the cc-pVTZ basis set (79, 80) are performed on the stationary structures already located on the MP2/cc-pVDZ potential energy surface. Zero point energy (ZPE) contributions to the activation energy are calculated by scaling the MP2/cc-pVDZ harmonic ZPE by 0.95 (98). All of the calculations are carried out using the GAMESS (33, 75) package and MacMolPlot (99) is used for molecular visualization.

Airshed Model Details.

The emission inventory selected for the University of California Irvine-California Institute of Technology airshed model simulations of this study is from August 6, 1997, which was used in the 2003 Air Quality Management Plan designed by the South Coast Air Quality Management District of California. A map of the airshed and surrounding areas is shown in Fig. S2. The emission profiles are coupled with meteorological measurements from August 28, 1987 that have been used widely to test air quality models (100, 101). The airshed source function for sea-salt particles is a function of the wind speed (102), and the RH ranged from 30–64%. In this study, the strength of the sea-salt particle source function is that of Cohan et al. (61), which predicts peak HCl concentrations for the base case (Fig. 7A) that are similar to those reported by Keene et al. in New England coastal air [0.005–5.7 ppb (103)] and Osthoff et al. [0.2–1.8 ppb (67)] near the coastline of Houston, TX. The HCl levels for the test case are in some regions a factor of 2 greater than those in the base case due to a positive feedback from ClNO formation. Chlorine atoms from ClNO photolysis react with volatile organic compounds through H-abstraction reactions to form HCl, which then drives further HCl-to-ClNO conversion (54). Chlorine formation initiated by the heterogeneous reaction of the hydroxyl radical with chloride ions at the gas-liquid interface of deliquesced salt particles (60, 62) was not included in the mechanism but would increase the HCl concentration and hence the ClNO and O₃. Although additional ClNO production pathways may exist, the current approach aims to provide insight into the importance of the proposed HCl-to-ClNO conversion chemistry and to understand how ambient ozone levels respond to this source.

Summary.

This work shows that when both HCl and oxides of nitrogen coexist indoors or outdoors, their interaction with surfaces does not result solely in their removal by deposition, but rather can also lead to the formation of new photochemically active gases. Particularly surprising is that this chemistry is enhanced by the presence of water, which has important implications not only for the chemistry occurring in the atmosphere but also for the fundamental chemical insights it provides. Clearly, the role of this chemistry needs to be further explored, particularly in indoor air environments where chlorine atoms have not been recognized as potential oxidants.