A controlling role for the air−sea interface in the chemical processing of reactive nitrogen in the coastal marine boundary layer

Significance

Reactions occurring at the air−sea interface have the potential to alter the chemical composition of the atmosphere. However, our knowledge of the extent to which these reactions impact the concentration of oxidants and their precursors is derived from laboratory measurements using systems that mimic the chemical, biological, and physical complexity of the surface ocean. Here, we present direct measurements of the vertical fluxes of a reactant−product pair using eddy covariance coupled with chemical ionization time-of-flight mass spectrometry to directly assess the role of the ocean surface in the exchange of reactive nitrogen and halogens. Our observations suggest that the ocean surface plays a critical role in controlling the lifetime of N₂O₅, a primary nocturnal reservoir for tropospheric reactive nitrogen.

Abstract

The lifetime of reactive nitrogen and the production rate of reactive halogens in the marine boundary layer are strongly impacted by reactions occurring at aqueous interfaces. Despite the potential importance of the air−sea interface in serving as a reactive surface, few direct field observations are available to assess its impact on reactive nitrogen deposition and halogen activation. Here, we present direct measurements of the vertical fluxes of the reactant−product pair N₂O₅ and ClNO₂ to assess the role of the ocean surface in the exchange of reactive nitrogen and halogens. We measure nocturnal N₂O₅ exchange velocities (V_ex = −1.66 ± 0.60 cm s⁻¹) that are limited by atmospheric transport of N₂O₅ to the air−sea interface. Surprisingly, vertical fluxes of ClNO₂, the product of N₂O₅ reactive uptake to concentrated chloride containing surfaces, display net deposition, suggesting that elevated ClNO₂ mixing ratios found in the marine boundary layer are sustained primarily by N₂O₅ reactions with aerosol particles. Comparison of measured deposition rates and in situ observations of N₂O₅ reactive uptake to aerosol particles indicates that N₂O₅ deposition to the ocean surface accounts for between 26% and 42% of the total loss rate. The combination of large V_{ex, N2O5} and net deposition of ClNO₂ acts to limit NO_x recycling rates and the production of Cl atoms by shortening the nocturnal lifetime of N₂O₅. These results indicate that air−sea exchange processes account for as much as 15% of nocturnal NO_x removal in polluted coastal regions and can serve to reduce ClNO₂ concentrations at sunrise by over 20%.

The production rate of tropospheric ozone (O₃), a criteria air pollutant, depends critically on the concentrations of nitrogen oxides (NO_x ≡ NO + NO₂), volatile organic compounds (VOCs), trace oxidants (e.g., OH, NO₃, and Cl), and the wavelength-dependent actinic flux. Accurate model representation of O₃ mixing ratios and the sensitivity of O₃ to changes in NO_x and VOC emissions rely heavily on a complete description of the factors that control NO_x lifetimes and, in turn, the concentrations of atmospheric oxidants. Modeling studies, constrained by laboratory and field observations, suggest that nocturnal processes involving the nitrate radical (NO₃) and N₂O₅, both products of NO_x oxidation, can account for as much as 50% of the NO_x removal (1). Incorporation of the heterogeneous reaction of N₂O₅ on chloride containing aerosol particles (2, 3) serves as both an efficient NO_x recycling and halogen activation mechanism via the production of photolabile nitryl chloride (ClNO₂) in both coastal (4) and continental air masses (5).

To date, study of the impact of nocturnal processes on the lifetime of NO_x and the production of reactive halogen species in the marine boundary layer has concentrated on gas-phase reactions and heterogeneous and multiphase processes occurring on/within aerosol particles, with little attention paid to reactions occurring at the air−sea interface (6, 7). With nearly half of Earth’s population living within 200 km of a saltwater coastline, a significant fraction of NO_x emissions are found near coastal waters (4, 8). As such, the chemical evolution of polluted air masses stemming from coastal megacities occurs to a large extent over the ocean (e.g., Beijing plume). If air−sea exchange of reactive nitrogen compounds is rapid and the reaction kinetics at the air−ocean and air−particle surface are comparable, we expect air−sea exchange processes to play an important role in setting the lifetime of compounds such as N₂O₅ in coastal environments. Specifically, dry deposition of N₂O₅ to the ocean surface could serve to help close the existing gap between models and measurements of N₂O₅ mixing ratios in the polluted marine boundary layer (9). In what follows, we describe direct measurements of the vertical flux of N₂O₅ and ClNO₂ obtained via eddy covariance at a polluted coastal site to provide observation-based constraints on the role of the air−sea interface in setting the lifetime of reactive nitrogen and the production rate of reactive halogens in the marine boundary layer.

The vertical flux of trace gases across the air−sea interface is a complex function of both atmospheric and oceanic processes, where gas exchange is controlled by molecular diffusion in the interfacial regions surrounding the air−water interface (10) and the solubility and chemical reactivity of the gas in the molecular sublayer. The flux (F) of trace gas across the interface is described by Eq. E1, as a function of both the gas-phase (C_g) and liquid phase (C_l) concentrations and the dimensionless gas over liquid Henry’s law constant (K_H),

where K_t, the total transfer velocity for the gas (cm s⁻¹), encompasses all of the chemical and physical processes that govern air−sea gas exchange (11). As such, accurate, molecule-specific parameterization of K_t is critical for assessing the role of the ocean as a net source or sink for both greenhouse gases and criteria air pollutants, and the trace gases that control their abundances in the atmosphere.

With respect to N₂O₅ air−sea exchange, we expect the reaction mechanism to closely follow that described for reactions occurring at the air−particle interface, particularly that of sea spray aerosol. Generally, the reactive uptake of N₂O₅ to aqueous interfaces in the troposphere has been proposed to follow the concerted reaction mechanism (12, 13):

This mechanism is consistent with laboratory evidence that the reactive uptake of N₂O₅ to aqueous interfaces is dependent on: (i) liquid water content (13), (ii) nitrate (NO₃⁻) and chloride (Cl⁻) concentrations (13–15), and (iii) the presence of organic surfactants and/or films (16–19). The ClNO₂ product yield, Φ (ClNO₂), following N₂O₅ hydrolysis has been shown to be a strong function of chloride concentration, where Φ (ClNO₂) is 0.8 for [Cl⁻] = 0.5 M, increasing to 1.0 for [Cl⁻] > 1.0 M (2, 13, 20).

Extension of laboratory determined reaction rates (21) and equilibrium constants (2, 22) to the air−sea interface would suggest that N₂O₅ deposition to the ocean should be rapid (e.g., K_Η = 1.9 × 10⁻², k₂ = 5 × 10⁶ s⁻¹) and that Φ (ClNO₂)_ocean would be >0.8, based on an oceanic [Cl⁻] of 0.55 M. However, the a priori estimate for the magnitude and direction of the air−sea flux for ClNO₂ (K_H = 1.66) is less clear (2). Although laboratory results suggest that ClNO₂ should be made at high yield at the ocean surface, it is not clear if water-side transport and subsequent chemical reactions may suppress ClNO₂ release back to the atmosphere. Alternatively, reaction of the nitronium ion (NO₂⁺) in the organic-rich sea surface microlayer (23) may also serve to reduce Φ (ClNO₂)_ocean, a reaction that may also proceed in organic-rich aqueous aerosol.

Results and Discussion

Eddy Covariance Measurements of N₂O₅ and ClNO₂ Air−Sea Exchange.

Concentration and vertical flux measurements of N₂O₅ and ClNO₂ were made at 13 m above mean lower low water (13 m above mean low water) from the end of the 330-m Scripps Institution of Oceanography (SIO) pier during January and February 2013. Briefly, N₂O₅ and ClNO₂ mixing ratios were measured using chemical ionization time-of-flight mass spectrometry (24), using I⁻ reagent ion chemistry (25). Spectra were saved at 10 Hz, coincident in time with measurements of 3D winds acquired with a colocated ultrasonic anemometer sampling at 20 Hz. Details on instrument calibration, inlet performance, and flux measurements can be found in Materials and Methods and SI Text.

Here, we discuss a subset of these measurements obtained on February 20, 2013 where the true wind direction ranged between 205° and 295°, resulting in a purely ocean fetch with pollution from Los Angeles entrained into the sampled air mass. Ten-meter wind speeds (u₁₀) ranged between 6.5 and 11.1 m s⁻¹ with a mean and SD of 9.07 ± 1.29 m s⁻¹. The diel profile in N₂O₅ and ClNO₂ mixing ratios is shown in Fig. 1 A and B, where N₂O₅ and ClNO₂ mixing ratios track one another for much of the night, peaking at midnight. As expected, N₂O₅ mixing ratios drop sharply to zero at sunrise due to rapid photolysis of the nitrate radical (NO₃), which is in thermal equilibrium with N₂O₅, whereas ClNO₂ decays to zero with a time constant (τ = 2.71 h) consistent with its photolysis lifetime (26). The magnitude of the N₂O₅ and ClNO₂ mixing ratios are comparable with those found in previous studies in coastal California (27).

Fig. 1.

(A) N₂O₅ and (B) ClNO₂ mixing ratios as measured from the SIO pier in La Jolla, CA, on February 20, 2013 at 0.1, 1, and 10 Hz time resolution.

N₂O₅ flux measurements are shown in Fig. 2A. As expected, N₂O₅ displays a net downward flux, into the ocean. The magnitude of the flux tracks ambient N₂O₅ mixing ratios, yielding a nocturnally averaged exchange velocity (V_ex, or flux divided by concentration) of −1.66 ± 0.60 (1σ) cm s⁻¹. We note that a negative V_ex indicates a downward flux from the atmosphere to the ocean. V_ex,N2O5 can be interpreted within the resistance framework developed for O₃ dry deposition, where V_ex depends on the aerodynamic resistance, quasi-laminar boundary layer resistance, and the surface resistance that includes chemical reactions at the interface (28). To this end, we calculate the total transfer velocity (K_t) for N₂O₅ (Eq. E2) for comparison with the observed exchange rate,

where k_a is the air-side transfer velocity, k_w is the water-side transfer velocity, and K_H is the dimensionless gas over liquid Henry’s law constant. Over the past two decades, a series of parameterizations (ref. 11 and references therein) have been developed that permit calculation of both k_a and k_w as a function of both the molecular properties of the gas (e.g., diffusivity, reactivity, solubility) and physical forcing data (e.g., wind speed). In the case of N₂O₅, the hydrolysis rate (k₂) is sufficiently fast (>1 × 10⁶ s⁻¹) that we expect N₂O₅ deposition to be limited only by the air-side transfer rate (k_a), despite its moderate solubility (K_H = 1.9 × 10⁻²). As such, we calculate k_a (Eq. E3) following the numerical approach of Johnson (29) where the still air diffusive flux of Mackay and Yeun (30) has been added to the representation of k_a found in the NOAA COARE model (31), with a numerical representation of the wind speed dependent drag coefficient (C_D).

Here, u* is the friction velocity, S_c is the Schmidt number for N₂O₅, and κ is the von Karman constant (taken as 0.4). For comparison, if we neglect fast hydrolysis, N₂O₅ deposition is controlled by the surface resistance due to its moderate solubility and slow diffusion rate in water (D_l = 1.9 × 10⁻⁵ cm² s⁻¹) (32). Calculations of the total transfer velocity of N₂O₅ as a function of wind speed are shown in Fig. 3, with (solid black line) and without (dashed black line) surface hydrolysis. In the latter case, we calculate k_w, again following the numerical approach of Johnson (29). Also included in Fig. 3 are the wind speed-dependent parameterizations for k_a(N₂O₅) of Liss (33) and Mackay and Yeun (30) both derived from wind tunnel studies and that of Duce et al. (34), which is calculated from micrometeorological theory. The nocturnally averaged measured N₂O₅ exchange velocity is shown with a blue square in Fig. 3. As shown, the observed exchange rate is on average a factor of 2.17 larger than that calculated using Eq. E3; however, it is in agreement with the parameterizations of Liss (10) and Mackay and Yeun (30), to within the uncertainty and variability of the measurements. Longer-term observations of F_N2O5, designed to capture wide variability in wind speed, will provide unique observation-based constraints on k_a, while permitting the opportunity to assess the impact of entrainment of N₂O₅ and ClNO₂ from the free troposphere and vertical gradients in temperature and aerosol surface area on the retrieved fluxes.

Fig. 2.

Measured vertical fluxes of N₂O₅ and ClNO₂. Errors are determined for each 30-min flux segment as the covariance between vertical wind speed and concentration at lag times significantly longer than the delay (or lag) time. Calculated ClNO₂ vertical fluxes (lines in B), as determined from the coupled time-dependent ocean−atmosphere model, constrained by the measured N₂O₅ vertical fluxes (A). Three different model scenarios are shown and described in detail in SI Text: C1 (a priori), model inputs taken as suggested in the literature [e.g., K_H (ClNO₂) = 1.66, Φ (ClNO₂)_ocean = 0.8, k_r = 5 × 10⁶ s⁻¹, and δ = 1.5 × 10⁻⁶ cm]; C2, model inputs taken as 90% confidence limits of those suggested in the literature [e.g., K_H (ClNO₂) = 0.32, Φ (ClNO₂)_ocean = 0.5, k_r = 2 × 10⁵ s⁻¹, and δ = 7.1 × 10⁻⁶ cm]; and C3, same as C2, with Φ (ClNO₂)_ocean = 0.

Fig. 3.

(Upper) Average N₂O₅ exchange velocity (V_ex, cm s⁻¹) as a function of 10-m wind speed (m s⁻¹). Modeled exchange velocities, determined via Eq. E3 [Johnson (29)] are shown using the dimensionless gas over liquid Henry’s Law Constant, K_H (N₂O₅) = 1.9 × 10⁻² with and without incorporation of N₂O₅ hydrolysis (k_r = 5 × 10⁶ s⁻¹). Parameterizations of Liss (33), Mackay and Yeun (30), and Duce (34) are also shown for comparison. (Lower) The 2012 annual median (black line), interquartile range (dark gray shaded region), and full range (light gray shaded region) for 10-m wind speed measured at the SIO pier.

Our a priori supposition for the direction of the ClNO₂ flux, based on laboratory-determined reaction rates and equilibrium constants, is assumed to be out of the ocean, where hydrolyzed N₂O₅ produces NO₂⁺ ions that rapidly react with Cl⁻, forming ClNO₂ in high yield at the ocean surface. Given the N₂O₅ deposition rate measured above, combined with the moderate solubility of ClNO₂, one would expect the magnitude of the flux to be nearly equal, yet opposite in direction, to N₂O₅. Surprisingly, vertical fluxes of ClNO₂ show net deposition, with the exception of the time period between 0300 and 0600, where the ClNO₂ flux indicates net emission from the sea surface (Fig. 2B).

Modeling Sea Surface Chemistry of Deposited N₂O₅.

To further our understanding of N₂O₅ reactions at the ocean interface, we construct a coupled atmosphere−surface ocean time-dependent box model following the framework outlined in Carpenter et al. (35). The model is constrained by the measured N₂O₅ vertical flux reported above and N₂O₅ and ClNO₂ reaction rates, product branching ratios, and diffusion constants as previously determined in the literature for reactions occurring at the air−particle interface. As in Carpenter et al. (35), we define a surface aqueous layer (δ) with a depth equal to the reacto-diffusive length (Eq. E4) for the N₂O₅ reaction mechanism (Reactions R1−R4)

where D_l is the N₂O₅ liquid-phase diffusion constant (D_l = 1.9 × 10⁻⁵ cm² s⁻¹) (32), and k_r is the total reactivity of N₂O₅ in seawater (k_r), taken as the N₂O₅ hydrolysis rate (k₂) (21) as the lifetime of the nitronium ion (NO₂⁺) product is estimated to be less than 10⁻⁹ s, making the hydrolysis the rate limiting step in the mechanism (2). We expect the reaction to occur in a thin film (δ < 100 nm), well within the molecular sublayer (ca. 10⁻³ m) where transport is driven by molecular diffusion (28). As a result, rapid volatilization of reaction products of relatively high solubility may occur following the supersaturation of dissolved gas in the thin film. For simplicity, gas transfer velocities (K_t) and the associated vertical flux for ClNO₂ are calculated in the model using k_a (Eq. E1) and k_w (Eq. E2) parameterizations as described in Johnson (29). As in Carpenter et al. (35), mixing from the bulk, where [ClNO₂] is taken to be zero, to the interfacial region is determined by the wind speed dependent expression for the transfer velocity.

To explore the apparent downward measured flux of ClNO₂, we drive the coupled atmosphere−ocean model using three plausible sets of input values in the treatment of K_t. In model case 1 (C1), our a priori conditions were based on existing laboratory-based measurements, where model input parameters were taken as suggested in the literature: K_H (ClNO₂) = 1.66, Φ (ClNO₂)_ocean = 0.8, k_r = 5 × 10⁶ s⁻¹, and δ = 1.5 × 10⁻⁶ cm. As shown in Fig. 2B, this results in a strong upward flux (sea to air) of ClNO₂ that, at its maximum, is approximately half the magnitude of the N₂O₅ peak flux. This difference is due to the prescribed ClNO₂ product yield coupled to exchange with the bulk ocean. In model case 2 (C2), we set the model input parameters at the 90% confidence limits of those suggested in the literature, in the direction of reducing the upward flux of ClNO₂ [e.g., K_H (ClNO₂) = 0.32, Φ (ClNO₂)_ocean = 0.5, k_r = 2 × 10⁵ s⁻¹, δ = 7.1 × 10⁻⁶ cm]. This results in a factor of two reduction in peak ClNO₂ flux, but the direction of the flux is still positive and outside the uncertainty of the measurements for much of the night. In model case 3 (C3), we take Φ (ClNO₂)_ocean = 0, suggesting that perhaps NO₂⁺ does not react with Cl⁻ as expected but proceeds via nitration reactions (36) with enriched organic material found in the sea surface microlayer (23). It is only by setting the ClNO₂ product yield to zero or invoking rapid ClNO₂ aqueous-phase reaction kinetics or hydrolysis that we can force the model near the uncertainty limits of the measurements.

For the conditions sampled here, ClNO₂ production rates were less than 3.0 × 10⁻³ ppt s⁻¹ [calculated for the median observed k_het, Φ (ClNO₂)_ocean = 1, and [N₂O₅] = 50 pptv] and the ClNO₂ steady-state lifetime with respect to loss to gas-phase and aerosol reactions is estimated at greater than 30 h (4). As a result, it is unlikely that significant gradients in ClNO₂ exist in the nocturnal marine boundary layer due to vertical gradients in the ClNO₂ atmospheric production rate. As such, our measurements of the vertical flux of ClNO₂, presented here, are most consistent with a model where the product yield of ClNO₂ is near zero for reactions of NO₂⁺ in the sea surface microlayer, and/or ClNO₂ aqueous-phase reactions are significantly faster than the volatilization rate.

Relative Roles of the Particle and Ocean Surface in Regulating τ(N₂O₅).

To assess the relative roles of the ocean and aerosol surface in the net removal of N₂O₅, both reactant transport and surface reactivity must be considered. At a marine boundary layer inversion height (z_i) of 810 m, taken as the nocturnally averaged z_i observed during DYCOMS-II off the coast of San Diego (37), the total aerosol surface area ranges between 5% and 50% of the surface area of a slab ocean, based on aerosol surface area measurements made at the SIO pier. In gas−aerosol reactions, transport of the reactant to the particle surface is set by the gas−aerosol collision rate, which is a function of the mean molecular speed of the reactant and the total suspended aerosol surface area (Eq. E6). In contrast, reactions occurring at the ocean surface require turbulent transport of the reactants to the diffusive sublayer, where turbulence is suppressed and molecular diffusion controls the collision rate of the reactant with the ocean surface. We compare N₂O₅ deposition rates (k_dep, Eq. E5) with N₂O₅ heterogeneous aerosol reaction rates (k_aerosol, Eq. E6), both directly measured at the SIO pier.

Here, V_ex is the measured N₂O₅ exchange velocity, z_i is the marine boundary layer inversion height, γ(N₂O₅) is the N₂O₅ reactive uptake coefficient, ω is the molecular velocity for N₂O₅, and S_a is the aerosol surface area concentration. Measurements of k_aerosol at this location have been described previously (38), where the median k_aerosol was observed to be 6.04 × 10⁻⁵ s⁻¹ with an interquartile range of 4.5–7.4 × 10⁻⁵ s⁻¹, where k_aerosol was a strong function of particle nitrate and organic mass fractions. As described above, the mean V_ex was −1.66 ± 0.60 cm s⁻¹ measured at an average wind speed of 9 m s⁻¹, corresponding to a range in k_dep of 0.17–1.7 × 10⁻⁴ s⁻¹ for z_i between 1000 and 100 m, respectively.

The relative strengths of the ocean and the aerosol surface in the net removal of N₂O₅ from the atmosphere are shown in Fig. 4, where the fraction of N₂O₅ lost to the ocean surface is shown as a function of k_dep and k_aerosol. As indicated by the dashed box, the fraction of N₂O₅ removed by the ocean surface ranges between 19% and 79% for the conditions sampled at the SIO pier, where shallow boundary layers (1000 < z_i < 100 m) and high wind speeds combined with suppressed N₂O₅ heterogeneous reactivity, result in large fractions of N₂O₅ being lost at the air−sea interface. For z_i = 810 m (37) and k_aerosol = 6.04 × 10⁻⁵ s⁻¹, the fraction of N₂O₅ lost at the air−sea interface is 32%, ranging between 26% and 42% for the interquartile range in k_aerosol. It is important to note that the measurements described here were made at wind speeds significantly larger than the annual median wind speed measured at the SIO pier (2.1 m s⁻¹). Based on the parameterized dependence of K_t on wind speed, it is expected that the range in k_dep reported in Fig. 4 may be a factor of 3 smaller at lower wind conditions, resulting in the fraction of N₂O₅ removed by the ocean surface to between 7% and 55% (1000 > z_i > 100 m). In addition, increased aerosol surface area concentrations under low wind speeds, due to reduced dilution of aerosol from an urban source, may also serve to reduce the fraction of N₂O₅ removed by the ocean surface at low wind speed.

Fig. 4.

Fraction of the total N₂O₅ removal attributed to deposition to the ocean shown as a function of the N₂O₅ deposition rate (k_dep) and the heterogeneous loss rate to aerosols (k_aerosol). The boxed region represents the range in k_dep and k_aerosol measured at the SIO pier in La Jolla, CA. The median and interquartile range in the observed k_aerosol is shown on the ordinate, while the range in the observed k_dep [based on a marine boundary layer inversion height (z_i) of 100–1000 m, at an average wind speed of 9.1 ± 1.3 m s⁻¹] is shown on the abscissa.

NO_x Removal Rates and ClNO₂ Production.

At present, the majority of steady-state box model analyses as well as regional-scale chemical transport models designed to assess nocturnal NO_x chemistry do not include either N₂O₅ or ClNO₂ deposition to the ocean surface or the possibility for reaction at the air−sea interface (39, 40). Here, we use a 0D time-dependent box model to assess the impact of air−sea exchange on NO_x removal rates and Cl atom production rates in the polluted marine boundary layer. The box model was run under four different N₂O₅ deposition rates at two different values of Φ (ClNO₂)_ocean, 0 and 0.8. It is important to note that the impact of N₂O₅ loss mechanisms on NO_x removal rates and Cl atom production is coupled to nitrate radical (NO₃) chemistry, where reaction of NO₃ with dimethyl sulphide (DMS) can act as the primary loss process for nocturnal nitrogen oxides. In the model described here, DMS concentrations are set to 100 pptv, corresponding to a loss rate of 2.6 × 10⁻³ s⁻¹ for NO₃. Model details can be found in SI Text as well as in Materials and Methods.

To assess the importance of deposition processes on reactive nitrogen and halogen budgets in coastal, polluted regions, we first examine the fraction of NO_x present at sunset, [NO_x]_sunset, that is lost to terminal sinks at night as a function of the N₂O₅ exchange velocity and ClNO₂ product yield.

As shown in Fig. 5A, constraining the model with the measured V_ex (−1.66 cm s⁻¹), z_i = 500 m and our best estimate of the yield of ClNO₂ produced at the ocean surface [Φ (ClNO₂)_ocean = 0], results in an approximate 14% increase in the fraction of NO_x that is lost to terminal sinks at night, compared with the model that neglects deposition (0.58–0.66). This trend shown in Fig. 5A describes the increasing fraction of NO_x that is deposited to the ocean surface, where at the limit of Φ (ClNO₂)_ocean → 1, approximately half of deposited N₂O₅ is returned to the atmosphere as ClNO₂ and in the limit of Φ (ClNO₂)_ocean → 0, all of the deposited N₂O₅ is terminally lost from the atmosphere. Model scenarios that neglect deposition or prescribe a high ClNO₂ product yield at the air−sea interface are efficient in recycling NO_x, where as much as 50% of reacted N₂O₅ is returned as NO₂ in the early morning following the photolysis of ClNO₂ (Fig. 5B). Further, these model scenarios will result in higher concentrations of ClNO₂ at sunrise, compared with models that include deposition and low Φ (ClNO₂)_ocean.

Fig. 5.

(A) Model calculations of the fraction of available NO_x ([NO_x]_sunset) lost during a 12-h night and (B) the mixing ratio of ClNO₂ at sunrise as a function of the prescribed N₂O₅ exchange velocity and ClNO₂ product yield [Φ (ClNO₂)_ocean]. Calculations were conducted using a 0D time-dependent box model constrained by the mean N₂O₅ reactive uptake coefficients and particle surface area (SA) concentrations measured at this site [γ(N₂O₅) = 0.005, SA = 500 μm² cm³]. The model was initialized with the following conditions: T_air = 283 K, T_water = 287 K, [O₃]_i = 60 ppb, [NO_x]_i = 1.0 ppb, γ(NO₃) = 0, and NO₃ reactivity = 0.16 min⁻¹.

Our observations suggest that N₂O₅ deposited to the ocean surface is terminally lost, thus limiting NO_x recycling rates and Cl atom production that would otherwise be sustained by heterogeneous mechanisms at the air−particle interface. As shown in Fig. 5B, including deposition to the ocean surface and neglecting ClNO₂ surface production results in nearly a 20% reduction in the concentration of ClNO₂ at sunrise, compared with a model that does not include N₂O₅ deposition. For comparison, including deposition to the ocean surface at the rates measured here, with a high ClNO₂ surface yield (0.8), results in nearly a 10% increase in the concentration of ClNO₂ at sunrise. This analysis highlights the sensitivity of the nocturnal nitrogen and halogen budget to N₂O₅ deposition and the resulting chemistry in the sea surface microlayer.

Atmospheric Implications.

We present an analysis of direct measurements of N₂O₅ and ClNO₂ air−sea exchange using eddy covariance. Our results indicate that N₂O₅ deposition to the ocean surface is rapid (V_ex = −1.66 ± 0.60 cm s⁻¹). We find no evidence for net ClNO₂ production at the air−sea interface in this study, suggesting either that rates of aqueous-phase reactions of ClNO₂ in the sea surface microlayer (SSML) are competitive with volatilization, or that the product yield for ClNO₂ is small in the organic-rich SSML. Comparison with direct measurements of the N₂O₅ loss rate to aerosol particles at the same sampling location indicates that the ocean surface serves on average to remove 32% of N₂O₅ in the marine boundary layer under the conditions sampled here (assuming z_i = 810 m). Our results suggest that future measurements of the exchange velocities of N₂O₅ under a wide range of wind conditions will provide needed constraints on parameterizations of the air-side transfer rate (k_a). We hypothesize that measurements of the flux of ClNO₂ will display large spatiotemporal variability and be responsive to the chemical composition and concentration of dissolved organic material (DOM) in the surface ocean due to the competition reactions of the nitronium ion with DOM and halogen ions. Our results suggest that regional modeling efforts designed to assess the impact of nocturnal nitrogen chemistry on oxidant loadings in coastal polluted environments need to properly represent air−sea interactions for adequate representation of the lifetime of both N₂O₅ and ClNO₂.

More broadly, we show that direct measurements of trace gas vertical fluxes when combined with in situ determinations of the reactive uptake to aerosol particles provides an experimental constraint on the lifetime and reactivity of trace gases to the wide array of available surfaces in the marine boundary layer. These results indicate that under conditions of shallow boundary layer heights, uptake to the ocean surface can outpace uptake to aerosol surfaces, highlighting the vast difference in the chemical composition, morphology, phase, and pH of the aerosol and ocean surface. The results presented here highlight the future utility of combining high sensitivity and precision time-of-flight mass spectrometric measurements of reactant and product pairs with micrometeorological techniques for direct, in situ study of chemical reactions occurring at the air−sea interface. This approach will permit study of complex interfacial processes under ambient conditions, where coupled biological, chemical, and physical mechanisms often prohibit the extension of laboratory results to environmental conditions.

Materials and Methods

Sampling Location.

Concentration and vertical flux measurements of N₂O₅ and ClNO₂ were made at 10 m from the northwest boom of the 330-m SIO Pier during February 2013. Air sampled at this location is impacted by local emissions in the La Jolla cove region as well as regional pollution attributed to both San Diego and Los Angeles. The observations presented here are for time periods where winds were sustained from the west (true wind direction between 205° and 295°) so as to ensure an ocean fetch. Backward air trajectories indicate that air sampled during these time periods was influenced by the Los Angeles plume, thus sustaining concentrations of N₂O₅ well above that observed in clean, marine air (41).

N₂O₅ and ClNO₂ Concentration Measurements.

N₂O₅ and ClNO₂ mixing ratios were measured using chemical ionization time-of-flight mass spectrometry (24), using I⁻ reagent ion chemistry (25). N₂O₅ sensitivities were determined using the output of a portable N₂O₅ generation system, described previously (42), where N₂O₅ is made in situ from the dark reaction of NO₂ and O₃, and subsequent reaction of the NO₃ product with NO₂. ClNO₂ sensitivities were determined by passing the output of the N₂O₅ source over concentrated NaCl slurry for unit conversion of N₂O₅ to ClNO₂ (4, 25, 43). We sample N₂O₅ and ClNO₂ through a 17-m, 3/8” o.d. Teflon perfluoroalkoxy tube. The inlet manifold is constructed of fluoropel-coated glass, closely resembling that of Ellis et al. (44), where air is drawn through a critical orifice, reducing the sample line pressure to 200 mbar. The resulting mass flow rate of 10 standard liters per minute results in a laminar flow profile, with a measured gas exchange time for the inlet of 0.7 s.

N₂O₅ and ClNO₂ Flux Measurements.

Mass spectra, acquired at 80 kHz, were saved at 10 Hz, coincident with measurements of 3D winds acquired with a colocated ultrasonic anemometer sampling at 20 Hz (HS-50; Gill Instruments). Fluxes were determined by the eddy covariance technique (45). In this method, the vertical turbulent flux is the covariance of vertical wind speed (w) and mixing ratio from the mean (c), F = <w′c′>. Details on the application of time of flight mass spectrometry to eddy covariance flux measurement can be found elsewhere (46) and described in more detail in SI Text.

Time-Dependent Air−Sea Model.

The chemical evolution of the nocturnal boundary layer was tracked using a one-dimensional time-dependent dimensional box model, where coupled differential equations were solved using custom code written in MATLAB, analyzed with the built-in ordinary differential equation solvers. As time propagates in the model, we calculate the production and loss of NO, NO₂, O₃, N₂O₅, NO₃, HNO₃, and ClNO₂ to gas-phase and heterogeneous reactions occurring on/within aerosol particles, as well as air−sea exchange with the ocean surface. The model is initialized with concentrations representative of those measured at the SIO pier, and reaction rates from the NASA Jet Propulsion Laboratory Chemical Kinetics and Photochemical Data for Use in Atmospheric Studies, Evaluation Number 14. Detailed descriptions of the modeling efforts described here can be found in SI Text.