Variability in Nocturnal Nitrogen Oxide Processing and Its Role in Regional Air Quality

See allHide authors and affiliations

Science  06 Jan 2006:
Vol. 311, Issue 5757, pp. 67-70
DOI: 10.1126/science.1120120


Nitrogen oxides in the lower troposphere catalyze the photochemical production of ozone (O3) pollution during the day but react to form nitric acid, oxidize hydrocarbons, and remove O3 at night. A key nocturnal reaction is the heterogeneous hydrolysis of dinitrogen pentoxide, N2O5. We report aircraft measurements of NO3 and N2O5, which show that the N2O5 uptake coefficient, g(N2O5), on aerosol particles is highly variable and depends strongly on aerosol composition, particularly sulfate content. The results have implications for the quantification of regional-scale O3 production and suggest a stronger interaction between anthropogenic sulfur and nitrogen oxide emissions than previously recognized.

Photochemical ozone production in the lower troposphere depends critically on the concentration of NOx (which is equal to the sum of NO and NO2)(1). In sunlight, NOx acts as a catalyst in cycles that produce O3 while oxidizing volatile organic compounds (VOC). At night, when the photochemical cycles are inactive, NOx enters a different pathway via the oxidation of NO2 by O3 to give the nitrate radical, NO3 (2, 3). The nocturnal nitrogen oxide reactions tend to counteract the daytime photochemical production of O3; they remove NOx and initiate the nocturnal oxidation of reactive VOC (e.g., higher alkenes, particularly biogenic compounds) (4). The nighttime pathway depletes the two main ingredients required for photochemical O3 production (NOx and VOC) while at the same time consuming, rather than producing, O3.

A key step in nocturnal NOx removal is the heterogeneous hydrolysis of dinitrogen pentoxide, N2O5. Math(1) Math(2a) Math(2b) Math(3) Math(4) Although reaction 3 is very slow in the gas phase, it can be rapid on aerosol particles as a heterogeneous process. Therefore, it has a substantial impact on the lifetime and abundance of NOx (3, 5).

In spite of its importance, relatively little is known about the actual efficiency of N2O5 hydrolysis in the nocturnal atmosphere. In this study, we provide direct experimental determinations of N2O5 uptake coefficients on the basis of in situ measurements of NO3, N2O5, and related compounds from an aircraft platform (the NOAA P-3) as part of an air quality study in the northeast United States (6). The analysis constrains the rate of N2O5 hydrolysis in anthropogenic NOx plumes from both urban sources and power plants. It shows that there can be considerable chemical processing of NOx at night via N2O5 hydrolysis and, importantly, that this processing is highly variable and depends strongly on aerosol composition. Because the variability exceeds that suggested by parameterizations of γ(N2O5) used in atmospheric models to date (710), these observations are pertinent to the understanding of NOx processing and regional air quality and to global budgets of NOx, O3, and other oxidants.

A map of the northeast United States (Fig. 1) indicates the location of major NOx sources. Overlaid is the flight track for the NOAA P-3 on 9 and 10 August 2004. This flight occurred mainly at night, with local sunset about 1 hour into the 8-hour flight. The map indicates three regions with distinctly different NO3 and N2O5 chemistry. In region I (over Ohio and western Pennsylvania), there was relatively little NO3 and N2O5 even though the ozone mixing ratios were sufficient to rapidly oxidize NO2 via Eq. 1. There was no measurable NO in these air masses, so the reaction of NO3 with NO also could not account for the small concentrations of NO3 and N2O5. Over eastern Pennsylvania, New Jersey, and downwind of New York (regions II and III), mixing ratios of NO3 and N2O5 were consistently large [peak values of 0.4 and 3.1 parts per billion by volume (ppbv), respectively] even though NO2 and O3 mixing ratios in these regions were similar to those in region I (6).

Fig. 1.

P-3 flight track for 9 and 10 August 2004, color-coded according to the N2O5 lifetime, τ(N2O5). [This lifetime varied from <1 up to 250 min after sunset, but the color bar saturates at 100 min for clarity of contrast. Values of τ(N2O5) for N2O5 mixing ratios less than twice the 1-pptv detection limit have been arbitrarily settozero.]NOx point sources (sized according to emission rate) and urban areas are shown. Wind barbs give local wind speed and direction along the flight track.

That the NO3 and N2O5 concentrations were small relative to their source (i.e., reaction 1) in region I indicates that their sinks were rapid or that their steady-state lifetimes (11) were short. Equations 5 and 6 show the relationship of the steady-state lifetimes (τSS) to the actual first-order sink rate coefficients for NO3 and N2O5, kNO3 and kN2O5 (12). Math(5) Math Math(6) Math Here, k1 is the rate coefficient for reaction 1 (NO2 + O3) and Keq is the temperature-dependent equilibrium constant for reaction 2. The unitless quantity Keq[NO2], equal to the ratio of N2O5 to NO3 at equilibrium, serves as a weighting factor that shifts in favor of N2O5 at large NO2 concentration or low temperature. The validity of a steady-state analysis of NO3 and N2O5 sinks for these data was verified with a chemical box model (6, 12).

The lifetimes of NO3 and N2O5 varied considerably between regions I and III (color code in Fig. 1). Equations 5 and 6 provide a means to determine whether this difference was due primarily to reactions of NO3 or N2O5; a plot of τss(NO3)–1 against Keq[NO2] gives kNO3 as the intercept and kN2O5 as the slope (Fig. 2 left). The slope of the fit to the data from region I shows clear evidence for a rapid sink for N2O5, with kN2O5–1 = 10 min. Analysis of several different NOx plumes within this region gave similar results. By contrast, the slope of a fit to the data from region III is indistinguishable from zero and gives a lower limit (2σ uncertainty in the fit) of kN2O5–1 > 23 hours. Again, this behavior is typical of several different plumes analyzed in this region. The intercepts differ by less than a factor of two, with values for kNO3–1 of a few tens of minutes in both air masses. This lifetime is consistent with the presence of unsaturated VOC (e.g., isoprene and anthropogenic alkenes measured from canister samples) in these air masses, although an accurate budget for gas phase sinks of NO3 would have required measurements for a larger array of VOC. A similar analysis of the same two flight segments using Eq. 6 (Fig. 2 right) gives kNO3 and kN2O5 as the slope and intercept, respectively, from a plot of τss(N2O5) against (Keq[NO2])–1 and corroborates the result from Fig. 2 (left).

Fig. 2.

Plots of τSS(NO3)–1 versus (Keq[NO2]) (left) and τSS(N2O5)–1 versus (Keq[NO2])–1 (right) for selected plumes from regions I and III of the flight in Fig. 1. The two plumes shown here were likely urban in origin (6), although air masses containing power plant plumes in both regions gave similar results. The solid lines are linear fits to the data, and the inverse of the slopes from each fit are given on the plot. The intercepts of each plot are equal to the slopes of the other plot to within the stated uncertainty on the figure, except for kN2O5–1 in the upper figure, which is derived from the slope uncertainty as described in the text.

The measured aerosol surface area, in combination with kN2O5 from plots such as those in Fig. 2, gives a direct determination of the N2O5 uptake coefficient, γ(N2O5) (13). Math(7) Here, cmean is the mean molecular speed of N2O5 and A is the aerosol surface area density (μm2 cm–3). Table 1 shows the average values of kN2O3–1, kN2O5–1, aerosol surface area density, and γ(N2O5) for the three regions marked in Fig. 1. The determined γ(N2O5) in region I had an average value near 0.02, consistent with laboratory measurements on ammonium bisulfate particles (1417), the largest component of the aerosol mass on this flight segment. However, in regions II and III, the γ(N2O5) were at least an order of magnitude smaller; the values in Table 1 in these regions are conservative upper limits determined from the intercepts of the τ(N2O5)–1 versus (Keq [NO2])–1 plots (Fig. 2 right). Values determined from the slopes of the τ(NO3)–1 versus Keq[NO2] plots (Fig. 2 left) are several times smaller.

Table 1.

Averaged actual (not steady-state) lifetimes of NO3 and N2O5, aerosol surface area densities, and N2O5 uptake coefficients for different regions of the 9 and 10 August 2004 flight. The aerosol surface area densities have been corrected by factors from 1.1 to 1.65 to account for the hygroscopic growth of the particles between the sampled RH and the ambient RH. Uncertainties in kNO3-1, kN2O5-1, and aerosol surface are standard deviations from several determinations within each region.

Region kNO3-1 (min) kN2O5-1 (min) Aerosol surface (μm2 cm-3) γ(N2O5)
I 13 ± 2 13 ± 2 1000 ± 140 0.017 ± 0.004
II 20 ± 2 >260 820 ± 80 <0.0010
III 31 ± 4 >400 340 ± 50 <0.0016

The N2O5 loss rate coefficients in regions II and III were slow enough to be inconsistent with recent parameterizations from smog chamber measurements of the proposed, slow gas-phase homogeneous reaction of N2O5 with water vapor (18), which would give an N2O5 lifetime of about 2.5 hours (rather than the determined >6.5 hours) for the New York City plumes. If there is a small, homogeneous component to the N2O5 hydrolysis, then the upper limits to γ(N2O5) in Table 1 for regions II and III are smaller.

The sharp change in γ(N2O5) between the western and eastern flight legs was accompanied by a similarly sharp change in the aerosol fine particle composition (DP < 1 μm, accounting for the majority of the particle mass and surface area) (Fig. 3). The largest observed difference between regions I and III was in sulfate aerosol mass loading. Because the organic content of the particles remained relatively constant, the sulfate/organic ratio also decreased considerably. The sulfate aerosol in region I was diffusely distributed rather than associated with the spatially narrow NOx plumes sampled in this region. This observation, combined with meteorological analysis of the sampled air masses by the FLEXPART model (19) in region I, implies that the sulfate aerosol came from prior photochemical conversion of SO2 to H2SO4 over the course of several days in the Midwest United States.

Fig. 3.

(A) Aerosol sulfate, nitrate, and water-soluble organic carbon (WSOC, labeled “organic”) mass loadings from the particle into liquid sampling (PILS) instrument for the 9 and 10 August flight for altitudes < 1.5 km. WSOC is in units of μg C m–3 and provides only a lower limit to the total organic mass. (B) Molar ratio of ammonium to sulfate from the same instrument. (C and D) Mixing ratios of N2O5 and NO3, respectively. Local sunset occurred just before 20:00 local time.

The change in aerosol composition between regions I to III may have influenced γ(N2O5) in several ways. First, the fraction of the total mass that was organic increased on the eastern leg. Laboratory measurements have shown small values for γ(N2O5) on some pure organic substrates (20) and reductions in γ(N2O5) upon addition of organic coatings to particles (21, 22). The majority of particles sampled during the New England Air Quality Study–Intercontinental Transport and Transformation (NEAQS-ITCT) were internal mixtures of sulfate and organic compounds (23). Second, the apparent aerosol acidity also decreased on the eastern leg, as shown by the time series of the average NH4+/SO4–2 molar ratio for each region (Fig. 3B). Acidic particles (NH4+/SO4–2 < 2) tend to exclude NO3, and laboratory studies have shown that NO3 inhibits N2O5 uptake (14, 24). This effect may explain why γ(N2O5) is small in region II, where the only measurable particulate NO3 was observed (Fig. 3). Lastly, the change from ammonium bisulfate to ammonium sulfate aerosol between regions I to III may have changed the relative humidity (RH) range over which the particles were present in a hydrated state. Although recent studies (25) suggest that real atmospheric aerosol particles likely remain hydrated even to low RH, the RH values in region III (44 to 52%) were near the efflorescence point of ammonium sulfate, below which laboratory data show a drop in γ(N2O5) for pure (NH4)2SO4 (14, 16, 17). Regardless of the precise mechanism, the important conclusion is that γ(N2O5) is highly variable and appears to be a function of aerosol composition, particularly sulfate mass or sulfate-to-organic ratio.

This result has implications for the regional-scale NOx burden and VOC oxidation, both of which affect the photochemical production of ozone. For example, a plume of NO2 emitted at dusk into an air mass with loss rate coefficients for NO3 and N2O5 characteristic of region I will undergo >90% irreversible loss of NOx in a 10-hour night, whereas the same plume emitted into an air mass with NO3 and N2O5 loss rate coefficients similar to those in region III will lose only 50% of its NOx in the same time period. Nocturnal VOC processing will also be affected by the N2O5 loss rate because it influences the amount of NO3, a potent oxidant. Mixing ratios of NO3 in region I (Fig. 3D) were on the order of tens of parts per trillion by volume (pptv), whereas those in region III varied up to nearly 400 pptv, almost exclusively because of the difference in the loss rate for N2O5. The corresponding lifetime for many anthropogenic alkenes (other than ethene) with respect to NO3 oxidation would change from about 50 to 5 hours (26). Such larger alkenes are among the most efficient for photochemical ozone production during the day (27). Therefore, the efficiency of N2O5 hydrolysis at night regulates both NOx and reactive hydrocarbon abundances at sunrise, when photochemical ozone production begins.

That the variability in γ(N2O5) appears to be most strongly correlated with the aerosol sulfate loading implies that there is a potential interaction between the emissions of SO2, the NOx budget, and regional O3 formation. It is well understood that sulfate aerosol arising from anthropogenic SO2 emissions provides a surface on which heterogeneous NOx processing may occur. Less well understood, however, has been the magnitude of and variability in γ(N2O5) and its dependence on aerosol sulfate. The only previous modeling study (10) to explicitly consider the interaction between SO2 emission, sulfate aerosol, NOx loss, and O3 production assumed a constant, large value of γ(N2O5) = 0.1 on the basis of laboratory data available at that time (28). The study concluded that changes in SO2 emissions would have a negligible effect on NOx lifetimes and regional O3 because the heterogeneous hydrolysis of N2O5 would be saturated (i.e., it would go fully to completion) for any reasonable aerosol surface area. The range of γ(N2O5) determined here and the apparent variation of γ(N2O5) with aerosol sulfate show that emission of SO2, followed by its conversion to particulate sulfate, can indeed decrease the lifetime of NOx and therefore influence photochemical ozone production. Further regional-scale model studies with realistic parameterizations of γ(N2O5) are required to quantify such an effect; if it were shown to be important, however, it could have implications for ozone mitigation strategies.

The variability of γ(N2O5) with aerosol composition has potential impacts on other issues. They include the export of NOx from the boundary layer to the free troposphere, the global burden of oxidants such as O3 and OH (10), and the seasonal variations in NOx and related chemistry. Our results point toward such influences, but additional in situ measurements of NOx, NO3, N2O5, O3, VOC, and aerosol from aircraft and other platforms in different locations and seasons will be required to address these questions.

Supporting Online Material

Materials and Methods

SOM Text

Figs. S1 to S3


References and Notes

View Abstract

Stay Connected to Science

Navigate This Article