Main

We first studied this photochemically driven conversion of nitrogen dioxide (NO2) into nitrous acid (HONO) on films of humic acid (HA), representing the complex unsaturated organic materials ubiquitously present in the environment. We exposed these HA films to various NO2 mixtures in an irradiated flow-tube and detected the NO2 and HONO gas phase concentrations at its exit. Typically, 1 mg (8 µg cm-2) of a given HA was coated onto the inner flow-tube wall. Figure 1a shows the results from a typical experiment, where we irradiated such an organic layer in the 400–700-nm wavelength range with an irradiance of 162 W m-2. A flow of synthetic air containing 20 p.p.b. NO2 at a relative humidity of 20% passed through the reactor (with a residence time of 0.6 s). During the irradiation we observed a substantial loss of NO2 and a corresponding formation of HONO of similar magnitude, which was a factor of 30 greater than production in the dark. The high conversion yield of about 80% indicates that NO2 is reduced by photochemically activated electron donors being present in the organic film and not by a catalysed disproportionation reaction18. No comparable reactivity was observed on clean glass surfaces, in contrast to surfaces containing HAs originating from peat, soil or lignite coal (Supplementary Fig. 1). As a consequence, this photochemically driven conversion is probably common to many surfaces ‘rich’ in partly oxidized aromatic structures, which facilitate the appearance of photochemically activated electron donors. Because such materials are commonly found on ground but potentially also on airborne surfaces19 (due to soil abrasion, biomass burning or oxidation of volatile organic compounds), the recognition of their photoreactivity may change our understanding of various processes occurring in the planetary boundary layer.

Figure 1: Conversion of NO 2 → HONO on 1 mg layers of humic acids initiated by visible light (400–700 nm).
figure 1

a, The filled circles represent the concentrations of HONO; the empty circles represent the amount of NO2 removed during an experiment. b, The HONO formation as a function of different initial NO2 concentrations. In the inset the data of the saturation curve is linearized according to a simple photochemical mechanism (reactions (1)–(3), see Methods section). c, The dependence of the HONO formation ([NO2]0 = 20 p.p.b.) on the light intensity. The black line is a simple model describing the dependence (see Methods section). The error bars (± 2σ) represent the estimated overall accuracy of the chemical analyses.

Full size image

An experiment under the same conditions as that described above was performed over an extended irradiation time of 13 h. Over the total extent of the irradiation 2 × 1015 molecules of NO2 per cm2 have been reduced on the HA surface (one molecule per 2,500 daltons of HA). This high value indicates that on average each molecule of HA (average molecular weight ∼15 kDa) present in the reactor reacted several times with NO2. However, the reaction is not catalytic because HA is oxidized by reaction with NO2. This is consistent with the observation that the reactivity of the HA surface decreased by 55% during the 13-h irradiation. HAs are steadily formed by the degradation of biota (about 1 g m-2 of soil organic matter is formed per day over the average global continental area20), so the deactivation by NO2 is small compared to the reformation of HA.

Figure 1b shows the effectiveness of the NO2 → HONO conversion at typical tropospheric NO2 concentrations. The experiments were performed on a single 1 mg HA sample and are corrected for sample deactivation using measured deactivation rates. The HONO yields saturate at high NO2 concentrations (higher than in the ambient atmosphere) under the reaction conditions described before. An elementary photochemical mechanism—activation of reductive centres (Ared) within the organic film by light (reaction (1)), the corresponding deactivation process (reaction (2)), and the reaction of Ared with adsorbed NO2 (reaction (3))—predicts such a saturation (see the Methods section for the mathematical treatment of this reaction system). Compounds ‘X’ introduced in reactions (1) and (2) indicate that the reactions may involve a photo-induced intra- or intermolecular electron transfer and its back-reaction. Therefore, ‘X’ should be viewed as oxidants. Clearly, alternative rate-limiting factors, such as a saturation of the adsorption sites by NO2, might also explain the observed saturation effect.

The irradiance at the reactor surface in this experiment was about 40% of that corresponding to clear-sky irradiance on a horizontal surface for a solar zenith of 48° in the 300–700-nm range (Supplementary Fig. 2). Thus in the atmosphere the HONO formation can be even more effective.

Figure 1c shows the increase of HONO production with light intensity, which demonstrates the photochemical nature of the reaction. The nonlinearity of the dependency can only be described by reactions (1)–(3), when oxidants ‘X’ responsible for the deactivation of Ared in reaction (2) are photochemically produced transient oxidants (that is, they are formed in reaction (1)). A simple model (see equations (1)–(3) and the Methods section) assuming that the concentrations of oxidants ‘X’ are proportional to the irradiance is shown in Fig. 1c to match the observations.

The photochemical NO2 → HONO conversion occurs over a broad spectral region with approximate overall quantum yields of about 3.9 × 10-6 (300–420 nm), 1.4 × 10-6 (400–700 nm), and 1.5 × 10-6 (mainly 500–700 nm) at light intensities of 144–162 W m-2, 20 p.p.b. NO2 and 20% relative humidity (see Supplementary Fig. 2 for lamp spectra). The given values should be taken as a relative measure, because they describe the number of HONO molecules formed per light quantum absorbed by the HA films in the reactor and depend on the experimental parameters (such as thickness of the HA coating or NO2 concentration). The error in the absolute values is about a factor of two21. From the solar spectral distribution, we conclude that the formation of HONO occurs not only in the ultraviolet-A spectral region, where the irradiation can also initiate photodissociation of nitrogen oxides (NO2, HONO or HNO3/NO3-), but is also very effective in the visible region under atmospheric conditions. This is in contrast to the photolysis of nitrate, which has previously been proposed as a daytime source of HONO15,22.

In view of the importance of superoxide as an electron carrier in aquatic photochemistry of dissolved organic matter, it is of interest whether superoxide could be the reductant responsible for the NO2 → HONO conversion observed here on solid HA surfaces. Therefore, we performed identical HA irradiations in the absence of NO2 and measured the yields of H2O2, which is the product of superoxide disproportionation (2O2- + 2H+ → H2O2 + O2). No H2O2 was observed under visible light, and only negligible amounts of H2O2 in the ultraviolet-A range (0.05 p.p.b., see Supplementary Fig. 3). Additionally, nitric oxide (NO), which exhibits a reactivity23 towards superoxide in aqueous solution similar to that of NO2, did not react on the HA surfaces under identical conditions (Fig. 1) when using 20 p.p.b. NO instead of NO2. The absence of significant amounts of H2O2 and the lack of reactivity of NO on irradiated HA surfaces indicate that superoxide is probably not involved in the observed reduction of NO2.

HAs are a complex mixture of macromolecular organics. They contain aromatic moieties as visible light absorbers and high contents of phenolic functionalities which can act as electron donors and even show (dark) reactivity towards NO2 (ref. 24). We have previously shown that films of mixtures of synthetic phenols and an aromatic light absorber (a benzophenone) showed a similar NO2 → HONO conversion during ultraviolet-A irradiation, as is observed here for HAs21. Such simple mixtures could therefore to some degree be considered as model systems for the photoreactivity of HAs. But we cannot yet assign the exact chemical nature of the reducing intermediates formed on irradiated HAs nor those formed on the synthetic films.

Figure 2a shows the evolution of HONO during irradiation of a natural standard soil (agricultural loamy sand) containing 2.3 wt% organic matter. A 1.5-mm-thick layer of this soil was irradiated (300–420 nm) in a flat-bed flow reactor through a 45 cm × 1 cm glass window by relatively parallel light (70 W m-2). While 45 cm2 of soil surface were irradiated, a much larger, non-illuminated, internal surface area of at least 5,000 cm2 (based on the particle sizes) was actually exposed to NO2. This meant there was already significant HONO production in the dark owing to NO2 reduction by HAs (by phenolic moieties24) or NO2 disproportionation on humid surfaces18. We suggest that the dark reaction observed under our flow-reactor conditions should not be viewed as representative of soil under environmental conditions, where the transport of NO2 into the bulk soil is probably much more limited. Upon irradiation the HONO formation was markedly enhanced. The slow response of the HONO concentration to switching the light on and off can be related to its high retention on the large soil surface along the reactor. Under these conditions, NO is also observed as a secondary product of HONO at rates consistent with a known dark reaction of HONO with organic soil constituents25,26. Figure 2b shows HONO formation on a glass surface containing small amounts of soil dust (0.16 mg cm-2), demonstrating that surfaces contaminated with traces of soil dust (like roads, buildings, rocks or plants) can also be expected to be photoreactive. Owing to the absence of a large bulk volume of soil in the flow tube, the dark reactions of NO2 and the retention of HONO on the soil surface are drastically reduced in contrast to the photochemical production.

Figure 2: Conversion of NO 2 → HONO on irradiated soil in presence of 17 p.p.b. NO 2 and 30% relative humidity.
figure 2

a, The formation of HONO (circles) and NO (line) for an ultraviolet-A irradiation (300–420 nm, 70 W m-2) of 6 g of soil. The vertical lines indicate experimental steps 1 to 6. 1: 00 h 06 min, NO2 added to air flow; 2: 00 h 30 min, soil exposed to NO2; 3: 01 h 00 min, irradiation started; 4: 02 h 40 min, irradiation stopped; 5: 05 h 50 min, NO2 addition stopped; 6: 07 h 07 min, reactor bypassed. b, HONO formation (circles) on 20 ± 5 mg artificially acidified soil dust (pH 4.6, H2SO4) dispersed on the glass surface of the tubular photo-reactor under ultraviolet-A irradiation (150 W m-2). Thick line, removal of NO2; thin line, formation of NO. The error bars (± 2σ) represent the estimated overall accuracy of the chemical analyses.

Full size image

The light-induced HONO production is 2.5 × 1010 molecules cm-2 s-1 on soil under ultraviolet-A irradiation (70 W m-2) in the presence of 17 p.p.b. NO2. In the 400–700-nm spectral range we observe HONO photoproduction of 1 × 1010 molecules cm-2 s-1 for a 75 W m-2 irradiance, which is low intensity compared to the solar visible irradiance (for example, 400 W m-2 for 48° solar zenith irradiance). From the experimental results, we estimate the total photochemical HONO production to be 5 × 1010 molecules cm-2 s-1 for a moderately polluted atmosphere (∼20 p.p.b. NO2) and solar irradiances (300–700 nm) of ∼400 W m-2. For comparison, Staffelbach et al.16 had to introduce an artificial HONO emission of 3.6 × 1010 molecules cm-2s-1 to explain the summer daytime HONO concentrations in southern Switzerland with their model. From their measurements they estimate that HONO contributed by more than 30% to the local radical production in air near the ground during the afternoon27. Reports of HONO measurements over a forest12 and over a rural site7 inferred unknown daytime HONO sources of 500 p.p.t. h-1and 170 p.p.t. h-1, respectively. From the evaluation of the main radical sources—the photolysis of ozone, formaldehyde and HONO or the ozonolysis of alkenes—the authors concluded that HONO photolysis accounted for 33% of the noon-time radical production12 and for 24% of the 24-h-average radical production7. A photochemical HONO formation at the ground surface of 5 × 1010 molecules cm-2 s-1 is sufficient to establish these HONO source strengths in air columns (assumed to be homogeneously mixed) with heights of 150 m and 430 m, respectively.

Therefore, we conclude that the photochemical HONO formation described here is consistent with recent observed daytime HONO concentrations and can be predicted to have a large contribution (for example, 20–30%) to the OH-radical production of the lowest hundred to a few hundred metres of the atmosphere. A similar impact can also be estimated by a simple OH-radical budget for the lowermost 100 m of the atmosphere: Summer day primary OH production rates of around 107 radicals cm-3s-1 are often reported for semi-polluted environments7,8,12. Integrating this OH production over a height of 100 m results in a layer production of 1011 radicals cm-2 s-1. The estimated HONO production of 5 × 1010 molecules cm-2s-1 derived in this study could explain half of the observed OH production in this lowermost layer. As the photochemical HONO formation occurs on ground surfaces and HONO photolyses rapidly, this radical source is correspondingly less important for the formation of OH radicals at higher altitudes in the atmosphere. But this lowest part of the atmosphere is important for the oxidation of biogenic volatile organic compounds, which have a similar short atmospheric lifetime28 as HONO in the daytime atmosphere, and for the formation of secondary air pollutants and aerosols due to the fast radical reactions occurring in this generally most polluted part of the atmosphere.

Methods

Photoreactors and description of the experiments

The irradiations of HA substrates were performed in 50 cm × 0.8 cm (i.d.) Duran glass tubes installed in an air-cooled lamp-housing holding seven fluorescence lamps (44 cm × 2.6 cm o.d.), in a circular arrangement surrounding the reactor tube. Three types of lamps were used to examine the HONO production under irradiation at different wavelengths (300–420 nm, 400–700 nm and 500–700 nm). The spectral irradiance of the three light sources at the reactor cell surface were measured with a LI-COR 1800 hemispherical, cosine-corrected spectro-radiometer and are shown in Supplementary Fig. 2 and compared to the solar spectral distribution at the Earth's surface and to the absorption spectra of Aldrich-HA. The inner surface of the tubular glass flow reactors (surface = 125 cm2, surface to volume ratio = 5 cm-1) was coated with a thin layer of HA. The HA coatings on the reactor wall were produced by drying aliquots of aqueous solutions of the HA (1 mg ml-1, pH = 4.4) dispersed on the reactor walls in a nitrogen stream at room temperature. The experiments with soil were performed in a 45 cm × 1 cm × 1 cm Teflon flow reactor with a glass window (45 cm × 1 cm) at the top. This reactor was illuminated by a single fluorescence lamp (70–75 W m-2) mounted in an air-cooled aluminium housing. The flat-bottomed area of the reactor (45 cm2) was completely covered with soil.

Analytical instrumentation

For the measurement of HONO we used a Long Path Absorption Photometer (LOPAP) instrument29 (total accuracy ± 10%; ref. 29). NO2 and NO were detected by means of NO/NOx-chemiluminescence detectors (CLD, Eco Physics, model CLD 77AM, or Eco Physics model CLD AL 770ppt connected to a photolytic converter PLC 760). The total accuracy of the NOx measurement is estimated to be ± 10%. The NOx detectors were used in combination with a sodium carbonate denuder tube (50 cm × 0.8 cm) at the inlet of the analysers to remove HONO from the gas stream and therefore eliminate the known interference of HONO in the NO2 → NO conversion. H2O2 was measured with an Aerolaser AL2001CL gas phase monitor.

Materials and reagents

The HA coatings were prepared from the commercially available HA sodium salts from Aldrich. As a possibly photochemically interfering impurity30, the specific lot had an iron content of 0.56%. However, spiking of the HA sample solution by additional 1.2% and 3.6% mass portions of iron(III) did not alter the reactivity of the HA coatings significantly. Standard soil was obtained from the Landwirtschaftliche Untersuchungs- und Forschungsanstalt (LUFA), Speyer, Germany. It is a loamy sand (standard soil type Lufa 2.2) collected 15 days before the experiments at 20 cm depth from an agricultural meadow. After drying to 5% residual water content, the soil organic carbon content was 2.29 ± 0.14% and the soil pH was 5.7 ± 0.3.

Model calculations used in Fig. 1

The saturation curve in Fig. 1b can be described according to reactions (1)–(3), assuming a steady-state concentration for [Ared]ss:

with

The combination of these two equations results in a HONO formation rate which is first order in NO2 at low concentrations (that is, d[HONO]/dt = keff[NO2]), but becomes independent of NO2 at high NO2 concentrations (that is, d[HONO]/dt = kmax). A linear regression (equation (6) of the data in a plot of r(HONO)-1 versus [NO2]-1 yields the limiting HONO production kmax = k1[HA] at high NO2 concentrations from the intercept. kmax equals the rate of photochemical production of Ared. The first-order rate coefficient keff = k3k1[HA]/(k2[X]) for HONO formation at low NO2 concentrations is derived from the slope. The linearized plot is presented as an inset in Fig. 1b.

The available data allow retrieving the maximum HONO production (kmax = (1.1 ± 0.2) × 1011 molecules s-1 per cm2 reactor surface) and the first-order rate coefficient keff for the HONO formation in the reactor (keff = 0.0048 ± 0.0007 s-1 cm-2; the given value is normalized to 1 cm2 of reactor surface area). keff corresponds to a gas kinetic uptake coefficient of γ = 2 × 10-5 for the reaction of NO2 with the HA surface.

In Fig. 1c the parameters kmax and keff derived above are used to model the dependence of HONO formation on the light intensity. Again we use equation (6), but we assume both that the oxidants [X] in reaction (2) are photochemically produced transient oxidants and approximate the concentration of [X] as proportional to the light intensity and that the rate of formation of Ared (reaction (1)) is proportional to the light intensity. The model result is shown in Fig. 1c.