Introduction

Formation of gaseous nitric acid and particulate nitrate (hereafter both referred to as HNO3), followed by its deposition on ambient surfaces, has been traditionally considered to be the permanent removal of nitrogen oxides (NOx = NO + NO2) from the troposphere1. However, increasing number of studies have demonstrated that photolysis rate constant of HNO3 on ambient surfaces and in aerosol particles, referred to as HNO3(s), is enhanced by 1–4 orders of magnitude2,3,4,5,6,7,8,9,10,11, compared to that of nitric acid in the gas phase, with HONO and NO2 as the main products2,3,4,5,8,9,11. As such, cycling of HNO3(s) to reproduce HONO and NO2 is competitive to its deposition removal, which results in a longer apparent lifetime and farer transport distance of NOx in the atmosphere than originally expected. HONO and NO2 are known as precursors of hydroxyl radical (OH) and ozone (O3), respectively. Production of HONO and NO2 by photolysis of HNO3(s) have a great impact on the atmospheric oxidative capacity, both for the polluted high-NOx environments5,10 and the remote low-NOx environments2,3,4,12.

The reported photolysis rate constant of HNO3(s) varies over a range of three orders of magnitude2,3,4,5,6,7,8,9,10,11,12,13. For example, several laboratory studies have reported photolysis rate constants of HNO3(s) of 2.2 × 10−5 s−1, 2.0 × 10−5 s−1 and 7.7 × 10−4 s−1 on silicon, glass, and sapphire surfaces6,7,8,9. The photolysis rate constant of HNO3(s) in urban grime was found to be as high as 1.2 × 10−3 s−1 under dry condition5. We have studied various natural and artificial surfaces in a previous paper and measured the photolysis rate constant of HNO3(s) in the range from 9 × 10−6 s−1 to 3.7 × 10−4 s−1, depending on the types of surfaces3.

The surface-enhanced and highly-varied photolysis rate constant of HNO3(s) could be, at least in part, rationalized by the underlying micro-mechanisms14,15,16,17,18. The interaction of HNO3(s) with surface reactive sites or associated molecules in the surface matrix, such as H2O, organic compounds, and HNO3(s) itself, has the potential to distort its molecular structure14,15,16,17,18. The distorted molecular structure of HNO3(s) is believed to be responsible for the “red shift” and the cross section enhancement of its absorption spectra from that of isolated nitric acid in the gas phase6,7,13,15,16,17,19. Although structure distortion of HNO3 also occurs to some extent in the solution14, the quantum yield in aqueous nitrate photolysis is lower than 0.1 due to the recombination of the primarily-produced photo-fragments, e.g. NO2 and OH, before they leave the surrounding cage of HNO3(s) - a so-called “cage” effect20,21,22,23. The quantum yield of HNO3(s) photolysis was measured near unity in the laboratory19,24. The red-shift absorption spectra into the actinic region with increased photon density, absorption cross section enhancement and high quantum yield together lead to the enhanced photolysis rate constant of HNO3(s).

Despite of the general information collected, detailed understanding on the kinetics and mechanisms of HNO3(s) photolysis, e.g., how exactly the surface matrix affects HNO3(s) photolysis, is still lacking. Organic compounds and HNO3(s), are ubiquitous on the environmental surfaces. In this study, photolysis of HNO3 on Pyrex glass is investigated in a photochemical flow reactor over a wide range of HNO3 surface density (DHNO3), with or without model organic compounds in the surface matrix. The photolysis rate constant of HNO3(s) are measured and described as a function of DHNO3 and the presence of the model organic compounds. The photolysis rate of HNO3(s) is inhomogeneous as suggested by surface catalysis mechanism, and the influence of organic matters on the photolysis production of HONO and NO2 is discussed in the context of an integrated mechanism.

Results

Surface-catalyzed photolysis of HNO3(s)

Upon light exposure, production rates of HONO and NO2 (PHONO and \({P}_{N{O}_{2}}\)) increased immediately, followed by a rapid drop (non-exponential) in the initial few minutes and then a pseudo-exponential decay afterwards (Fig. S1). The signals nearly returned to the baseline level after the light was turned off. It was evidential that the immediate increase in PHONO and \({P}_{N{O}_{2}}\) was in response to photolysis of HNO3(s). The apparent or average photolysis rate constant of HNO3(s) (JHNO3(s)) during the light exposure period is then determined by equation (Eqs 7–9) in the Method section and listed in Table 1. The rapid drop in the initial PHONO and \({P}_{N{O}_{2}}\) suggests the inhomogeneity of photolysis reactivity of HNO3(s), that is, some HNO3(s) is more reactive than the rest2,3. It is proposed here that some HNO3(s) molecules directly associated with surface reactive sites, thus acquire superior photochemical reactivity and tends to be photolyzed at much higher rates in the initial stage, relative to the rest, when exposed to the UV light.

Table 1 Summary of photolysis rate constant of HNO3 on Pyrex glass surface without and with model organic (~16 µmol m−2).
Full size table

The inhomogeneous photolysis reactivity of HNO3(s) could be attributed to the surface catalysis effect on HNO3(s) photolysis and be quantitatively described (Eqs 1–3)3. It is known that HNO3 distributes on the surface irregularly even under the “sub-monolayer” conditions24, with high affinity to the surface reactive sites25. The association of HNO3(s) with surface reactive sites could lead to structure distortion of HNO3(s) molecule, and thus change its absorption spectra and thus its photolysis reactivity6,7,26. In the proposed surface catalysis mechanism, the catalysis power depends on the nature of the surface reactive site and the type of the reaction, and it dissipates as more molecules are deposited on the upper layer and further away from the neighboring surface reactive site27,28. That is, for each infinitesimal amount of HNO3(s), the corresponding photolysis rate constant (j) is an inverse function of surface coverage or surface density of HNO3(s) (\({D}_{HN{O}_{3}}\)):

$${j}=\frac{a}{1\,+\,b\,{D}_{HN{O}_{3}}}+c$$
(1)

where a, b and c are the fitting constants. The constant a represents the maximum photolysis rate constant (s−1) for infinitesimally small amount of HNO3(s) that is directly associated with the surface reactive sites. The constant b reflects the dissipation rate (m2 mol−1) of the catalysis power towards increased \({D}_{HN{O}_{3}}\), and c is the photolysis rate constant (s−1) of the upper-layer HNO3(s) that is not closely associated with the surface reactive site. Then the total production of HONO and NO2 (R) from photolysis of all the HNO3 molecules on unit surface area is:

$$\begin{array}{rcl}R & = & {\int }_{0}^{{D}_{HN{O}_{3}}}jd({D}_{HN{O}_{3}})\\ & = & \frac{a}{b}\,\mathrm{ln}(1+b\,{D}_{HN{O}_{3}})+c\,{D}_{HN{O}_{3}}\end{array}$$
(2)

And the apparent or average photolysis rate constant of HNO3(s) is the ratio of the production (R) to the amount of HNO3(s) exposed to light on unit surface area (\({D}_{HN{O}_{3}}\)):

$$\begin{array}{rcl}{J}_{HNO3(s)} & = & \frac{R}{{D}_{HN{O}_{3}}}\\ & = & \frac{a}{b\,{D}_{HN{O}_{3}}}\,\mathrm{ln}(1+b\,{D}_{HN{O}_{3}})+c\end{array}$$
(3)

The equation (Eq. 2) fits the experimental data well, with an r2 of 0.96 for “HNO3” experiments, an r2 of 0.91 for “HNO3 + HA” experiments, and an r2 of 0.94 for “HNO3 + SA” experiments (Fig. 1a). The equation (Eq. 3) again fits all the experimental data well with strong correlations (r2 values ≥ 0.94) (Fig. 1b). The higher a (5.0 × 10−3) and lower b (5.8 × 107) values fitted for “HNO3 + SA” than those for “HNO3” suggest extra catalytic power of salicylic acid, in addition to the surface reactive site on Pyrex glass. This point is in good agreement with the substantial influence of salicylic acid on HNO3(s) photolysis (see the next section). A single fitted c value of ~3 × 10−6 s−1 for all the three experiments represents a common photolysis rate constant of HNO3(s) in the upper layer, where HNO3 complexed with water and itself, but is nearly unaffected by the catalysis power of surface reactive sites and the organic matters in the matrix3. A common c for all the experiments also seems reasonable, as all the experiments fitted in Fig. 1 were conducted under 50% RH, and HNO3-H2O complex in the upper layer would be the same.

Figure 1
figure 1

Relationships of total production rate of HONO and NO2 from HNO3(s) photolysis (a) and the photolysis rate constant (b) with HNO3 surface density. The lines are the best fits to the data for equation (Eq. 2): the experiments of “HNO3” (, a = 6.5 × 10−4, b = 1.5 × 108, c = 3.0 × 10−6, r2 = 0.96), “HNO3 + HA” (, a = 9.5 × 10−4, b = 1.5 × 108, c = 3.0 × 10−6, r2 = 0.91), and “HNO3 + SA” (□, a = 5.0 × 10−3, b = 5.8 × 107, c = 3.0 × 10−6, r2 = 0.93) and for equation (Eq. 8): the experiments of “HNO3 only” (, r2 = 0.95), “HNO3 + HA” (, r2 = 0.94), and “HNO3 + SA” (□, r2 = 0.98). The fitting constants a, b, and c are the same as in the two panels. HA and SA represent humic acid and salicylic acid, respectively.

Full size image

In our previous paper on photolysis of ambient particulate nitrate and HNO3 deposited on natural and artificial surfaces, equation (Eq. 3) fits well those reported photolysis rate constants2,3. The good fitting of our data by Eq. 3 here and in our previous papers is quantitative evidence for the surface catalysis mechanism, and provides a robust parameterization method for the varied photolysis rate constant of HNO3(s).

Matrix effect of water and organic matters

Water and organic matter are ubiquitous on the environmental surfaces and may have significant effects on both JHNO3(s) and HONO/NO2 production ratio. As the surface -adsorbed water decreased slowly with the drying time of the sample, the JHNO3(s) value was found to increase and the production ratio of HONO/NO2 to decrease (Fig. 2). This is consistent with a previous result, which shows higher JHNO3(s) but lower HONO/NO2 production ratio at 0% RH than 50% RH9. Surface-adsorbed water could affect the photolysis reactivity of HNO3(s) and HONO/NO2 production ratio via complex mechanisms14,16,22,26. At high RH, competitive adsorption of surface water with HNO3(s) on surface reactive sites may reduce the apparent photolysis reactivity of HNO3(s). Water cage potentially formed at high RH may further reduce the apparent photochemical reactivity by promoting the recombination of photo-fragments20,21,22,23, and favor the formation of HONO from reaction between primary product of NO2 and adjacent water as a H-donor. While at low RH, HNO3 and H2O cluster formed on the surface reactive site might enhance the photolysis reactivity of HNO3(s)28. The complex interaction of HNO3(s) with co-absorbed H2O is consistent with varied absorption state and chemical environments of HNO3(s), which determines the photolysis reactivity of HNO3(s).

Figure 2
figure 2

Plots of JHNO3(s) and production ratio of HONO to NO2 as a function of dry time after coating the Pyrex glass surface with same amount of HNO3.

Full size image

To investigate the matrix effect of organic matters on HNO3(s) photolysis, 13 model organic compounds were chosen as proxies for the naturally occurring organics in the atmosphere (Table 1). Aromatic compounds were of special interest because they have absorption bands in the UVB (280–315 nm) actinic region as shown in the absorbance measurement of their water solution in Fig. S2 and may serve as photosensitizers. The isomers of hydroxybenzoic acids and benzenediols were examined and compared. Some non-aromatic organic acids and polyols were chosen to study matrix effects other than photosensitization because they generally do not absorb light in the UV actinic regions. Humic acid was selected due to its ubiquitous presence and its photosensitization effect in the photo-enhanced conversion of NO2 to HONO29. All these model organic compounds do not contain nitrogen and thus are not direct precursors to the target products of HONO and NO2.

Table 1 summarized the photolysis rate constant of HNO3(s) at the presence of different model organic compounds. While multiple measurements were made for each model organic compound at different \({D}_{HN{O}_{3}}\), the table lists only the results at \({D}_{HN{O}_{3}}\) of ~1.1 × 10−6 mol m−2 and ~25 × 10−6 mol m−2 to represent the sub-monolayer and multilayer conditions, respectively. One or two of the model organic compounds was co-adsorbed with HNO3(s) onto the Pyrex glass surface at a surface density of about 16 × 10−6 mol m−2 to form ~2 layers3, except for humic acid, which was coated at a surface density of 1.6 mg m−2.

In the absence of model organic compounds, a mean (±SD) JHNO3(s) value of ~2.1 (±0.4) × 10−5 s−1 was measured for the sub-monolayer conditions, which was similar to a previous measurement value of 2.2 (±0.2) × 10−5 s−1 on an unpolished Pyrex glass surface at 50% RH9. Photolysis of HNO3(s) is enhanced by the presence of all the model organic compounds relative to “HNO3” conditions; the magnitude of enhancement depends on both functional groups and substitution patterns of organic compounds.

The presence of non-light-absorbing organic compounds significantly enhances the photolysis rate constant of HNO3(s). Citric acid and oxalic acid enhance the \({J}_{HN{O}_{3}(s)}\) by a factor of 5–6, and succinic acid, ascorbic acid and glucose by a factor of ~2. The observed enhancement on HNO3(s) photolysis by non-light-absorbing organic compounds is due to mechanisms other than photosensitization, such as direct participation in the reaction as H-donors29,30. All these model organic compounds, especially organic acids and polyols, are strong H-donors. They distribute with water molecules in the surrounding of HNO3(s), weaken the water molecular “cage” and modify the chemical environment of HNO3(s)19,20,21,22. The reactions between these organic compounds and the initial photo-fragments of HNO3(s) photolysis, i.e., OH and NO2, suppress the recombination of photo-fragments and thus enhance JHNO3(s). The reactions between organic compounds and primary NO2 also produces secondary HONO and shifts the HONO/NO2 production ratio. \({J}_{HN{O}_{3}(s)}\) was then plotted against HONO/NO2 production ratio for both monolayer HNO3 and multiple-layer HNO3 experiments to test the hypothesis of H-donation reaction in HNO3(s) photolysis (Fig. 3). Consistently strong correlation was found both in the monolayer HNO3 (Fig. 3a) and the multiple-layer HNO3 (Fig. 3b) experiments for organic acids and polyols, i.e., citric acid, oxalic acid, succinic acid, ascorbic acid and glucose (see Fig. S3 analysis to include more organic species). One exception was found in ascorbic acid in monolayer HNO3 experiment (Fig. 3a). Ascorbic acid is a strong reducing agent31; it may have reduced NO2 not just to HNO2 but further to NO, which was not detected by our NO2 measurement method, resulting in a lower apparent photolysis rate constant. To further test the H-donation hypothesis, different amount of catechol was co-absorbed with HNO3, and \({J}_{HN{O}_{3}(s)}\) was indeed found increased with catechol amount (Fig. 4). These two corroborative pieces of evidence in Figs 3 and 4 suggest the participation of H-donors in HNO3(s) photolysis, affecting both JHNO3(s) and the production ratio of HONO/NO2.

Figure 3
figure 3

Correlation analysis between the JHNO3(s) value and the production ratio of HONO/NO2 for non-aromatic compounds at a HNO3 surface density of ~1.1 × 10−6 mol m−2 (a) and 25 × 10−6 mol m−2 (b). OA, CA, GL, SA, and AA represent oxalic acid, citric acid, glucose, succinic acid and ascorbic acid, respectively. The solid circle represents an outlier datapoint of AA, which was not included in the fitting.

Full size image
Figure 4
figure 4

Log-log plot of the photolysis rate constant against catechol surface density. The photolysis rate constant increases with catechol surface density from data point for experiment without catechol (solid black cycle) to data points for increasing catechol surface density (black open cycle) Data points for ~ monolayer HNO3 (red open cycle) are above the data points for multilayer HNO3 (black open cycle) as catechol surface density is the same.

Full size image

The presence of light-absorbing organic compounds has an even stronger influence on the photolysis rate constant of HNO3(s). Salicylic acid and hydroquinone exhibit highest enhancement effect on JHNO3(s), by approximate one order of magnitude. Their isomers, i.e., 3-hydrobenzoic acid, 4-hydrobenzoic acid, catechol, and resorcinol only show modest enhancement effect, with enhancement factors of 3–4. Benzoic acid and humic acid also show small enhancement effect by a factor of ~2. To evaluate the photosensitization effect, the relative light absorption of organic compound solution (A) was calculated as the product of their absorption cross section from 290 nm to 360 nm and intensity spectra of the experimental light source at corresponding wavelength (Eq. 4).

$${\rm{A}}={\int }_{290\,nm}^{360\,nm}\,{\sigma }_{i}{\rho }_{i}d{\lambda }_{i}$$
(4)

where σi and ρi represent the absorption cross section of organic compounds and photon density of the light source as a function of wavelength. Within the same group of isomers (hydroxybenzoic acids or benzene-diols) photolysis rate is significantly enhanced by the light absorption (Fig. 5), suggesting the enhancement effect of photosensitization. However, no correlation between relative light absorption and the enhancement factor was found if all the tested organic compounds were considered (Fig. 5). Possible reason might lie in the fact that photosensitization reaction consists multiple primary steps, e.g., absorbing light, photo-electron generation, photo-electron transition and induction of final photolysis reaction. The yields of these steps are expected to vary with the structure of organic compounds and thus consistent relationship between light absorbance and photolysis rate would not be always expected. For example, despite of small difference in molecular structure of hydroxybenzoic acids and benzene-diols, they have demonstrated substantially different quantum yield as shown in the blue line and red line in Fig. 5. As such, the light absorption appears only a rather rough indicator of the photosensitization of organic compounds on HNO3(s) photolysis (Fig. 5).

Figure 5
figure 5

Correlation analysis between the enhanced JHNO3(s) value and relative light absorption. The light with wavelengths below 300 nm was filtered by a Pyrex glass filter. HE, SA, 4-HBA, CL, RL, 3-HBA and 3-HB represent hydroquinone, salicylic acid, 4-hydroxybenzoic acid, catechol, resorcinol, 3-hydroxybenzoic acid and 3-hydrobenaldehyde, respectively.

Full size image

Mechanisms

Photolysis of HNO3 on Pyrex glass surface is enhanced by 1–4 orders of magnitude compared to those in the aqueous solution and gas phases, depending on its surface density and the type of organic compounds coexisting in the surface (Table 1, Fig. 1b). The surface catalysis mechanism explains the observed enhancement on HNO3(s) photolysis, and the derived Eq. 3 can fits the photolysis rate constants at different HNO3 surface density (Fig. 1), presenting Eq. 3 as a quite good parameterization method for HNO3(s) photolysis.

In the catalysis mechanisms discussed above, HNO3 absorbs preferably on the surface reactive sites25,26,32. The association of HNO3(s) with the surface reactive site distorts its molecular structure and results in enhanced absorption cross section and “red shift” of the absorption spectra into the actinic region6,7,14,16,18,33. Both changes in its absorption spectra contribute to the significant enhancement in the light absorption and the production of excited HNO3(s), HNO3*(s):

$${{\rm{HNO}}}_{3({\rm{s}})}+h\nu \to {{\rm{HNO}}}_{3({\rm{s}})}^{\ast }$$
(R1)

Organic compounds on the surface may associate with HNO3(s), causing structure distortion and enhancement of the light absorption of HNO3(s), analogous to the catalysis power of the surface reactive site, and could thus further increase the apparent yield of HNO3*(s). This will result in a higher a value and a lower b value in the equations (Eq. 2-Eq. 3), for example, in “HNO3 + SA” experiments than in “HNO3” experiments (Fig. 1). In addition, organic chromophores are photosensitizers, which absorb light and transfer the energy to the adjacent HNO3(s) and thus indirectly increase the apparent yield of HNO3*(s).

The bond breakage of HNO3*(s) produces OH(s) and NO2(s) as the dominant products (R2) and HONO as a minor product (R-2)9,18,19:

$${{\rm{HNO}}}_{3({\rm{s}})}^{\ast }\leftrightarrow {{\rm{OH}}}_{({\rm{s}})}+{{\rm{NO}}}_{2({\rm{s}})}$$
(R2)
$${{\rm{HNO}}}_{3({\rm{s}})}^{\ast }\to {{\rm{HONO}}}_{({\rm{s}})}+{\rm{O}}{({}^{3}{\rm{P}})}_{({\rm{s}})}$$
(R-2)

Certain fraction of these two photo-fragments, OH(s) and NO2(s), may recombine to form HNO3*(s) (R-2)20,21,22,23, or react with organic and water molecules (R3–R5) to form new products including HONO.

$${{\rm{OH}}}_{({\rm{s}})}+{{\rm{Org}}}_{({\rm{s}})}\to {\rm{Oxidized}} \mbox{-} {{\rm{Org}}}_{({\rm{s}})}+{{\rm{H}}}_{2}{{\rm{O}}}_{({\rm{s}})}$$
(R3)
$${{\rm{NO}}}_{2({\rm{s}})}+{{\rm{H}}}_{2}{{\rm{O}}}_{({\rm{s}})}\to {{\rm{OH}}}_{({\rm{s}})}+{{\rm{HONO}}}_{({\rm{s}})}$$
(R4)
$${{\rm{NO}}}_{2({\rm{s}})}+{{\rm{Org}}}_{({\rm{s}})}\to {\rm{Oxidized}} \mbox{-} {{\rm{Org}}}_{({\rm{s}})}+{{\rm{HONO}}}_{({\rm{s}})}$$
(R5)

The water molecular cage effect could sufficiently decrease the quantum yield and thus photolysis rate constant9. The NO2(s) produced from (R2) may possess excessive energy and may be more reactive than ground-state NO2 towards H donors (R4 and R5), leading to HONO production31,34,35. The H-donation reactions consume the photolysis fragments, prevent them from recombination (R-2), and thus enhance the quantum yield and JHNO3(s)9,16,17. Increase of JHNO3(s) with HONO/NO2 production ratio (Figs 3 and S3) is consistent with the H-donation reaction mechanism (R4 and R5). The dependence of JHNO3(s) on the abundance of a typical H donor, catechol, supports the argument that H donor directly participates in HNO3(s) photolysis (R3–R5).

Finally, the produced NO2 from reaction (R2) and HONO from reactions (R-2, R4 and R5) release from surface to gas phase (R6 and R7):

$${{\rm{HONO}}}_{({\rm{s}})}\to {{\rm{HONO}}}_{({\rm{g}})}$$
(R6)
$${{\rm{NO}}}_{2({\rm{s}})}\to {{\rm{NO}}}_{2({\rm{g}})}$$
(R7)

Our measurement systems measured the released HONO and NO2 in the gas phase.

Conclusion

The environmental surfaces, the deposited molecules of HNO3, organic matter and water form a complex system and play varied roles in HNO3(s) photolysis, such as surface catalysis, photosensitization, and H-donation. An integrated photolysis mechanism is proposed here with detailed fundamental reactions and Eq. 3 to qualitatively and quantitatively describe the variation of JHNO3(s), such as the dependence of JHNO3(s) on HNO3 surface density, H donors (both water and organic compounds) and photosensitizer. To note, this mechanism investigation on the surface-catalyzed HNO3 photolysis in laboratory need to be supplemented with quantitative analysis on the photolysis rate constant and its dependence on ambient conditions, such as temperature. For that, further study should evaluate the kinetics and mechanisms of HNO3 photolysis in ambient conditions.

Method

Chemicals

All the chemicals were at least reagent grade or better, and were used without further purification, including: HNO3 (Ultrapure, J.T. Baker), Salicylic acid (ACS grade, Sigma-Aldrich), 3-hydrobenzoic acid (99%, Aldrich), 4-hydrobenzoic acid (ACS grade, Aldrich), catechol (≥99%, Sigma-Aldrich), resorcinol (ACS grade, Sigma-Aldrich), hydroquinone (≥99%, Sigma), citric acid (ACS grade, Sigma-Aldrich), oxalic acid (≥99%, Sigma-Aldrich), succinic acid (≥99%, Sigma-Aldrich), ascorbic acid (reagent grade, Sigma-Aldrich), Benzoic acid (99%, Aldrich), glucose (≥99.5%, Sigma), humic acid sodium salt (Technical grade, Aldrich), NaNO2 (99.7%, J.T. Baker), NaNO3 (99.9%, J.T. Baker), sulfanilamide (SA) (≥99%, Aldrich), N-(1-Naphthyl) ethylene-diamine (NED) (ACS grade, Aldrich), NaOH (99.99%, Aldrich), NH4Cl (99.99%, Aldrich), HCl (Aldrich), sodium benzoate (99%, Aldrich), and sodium salicylate (≥99%, Sigma-Aldrich). Water was purified with a Barnstead Nanopure Diamond system or Millipore Milli-Q water system, with resistivity ≥18.2 MΩ.

Experimental Setup

The light exposure experiment was conducted in a cylindrical Teflon flow reactor with a quartz window on the top. The diameter is 10 cm and depth is 2.5 cm, with a cell volume of ~200 ml. A sandblasted borosilicate glass (Pyrex) surface (~9 cm diameter) was used as the “substrate” for coating HNO3 and organic compounds, and was placed in the flow reactor for light exposure. Ultra-high-purity nitrogen (Airgas, UHP200) was used as the carrier gas. The carrier gas flowed through a thermostatic water bubbler at 9.2 ± 0.1 °C at a flow rate of 450 ml min−1 and then went through a long thermal-equilibration coil at the room temperature of 21 ± 1 °C. The RH of the carrier gas entering the flow reactor was then at 50 (±2) %. The gaseous products were sampled by two coil samplers connected in series at a gas flow rate of 400 mL min−1. Purified water was used to scrub HONO in the first 10-turn coil sampler at 100% efficiency. A reagent solution containing 60 mM sulfanilamide (SA) and 0.8 mM N-(1-Naphthyl) ethylene-diamine (NED) in 2.5 M acetic acid was used to scrub NO2 in the second 40-turn coil sampler at 60% efficiency. The remaining outflow of ~50 ml min−1 from the reactor was vented through a 10-cm 1/16” ID Teflon tubing to maintain slightly positive pressure in the reactor. All tubing used in the cabinet was wrapped by aluminum foil to shield from the UV light. The collected HONO and NO2 were derivatized by SA and NED and analyzed as azo dye by two long-path absorption photometric (LPAP) systems36. Each LPAP system consists of a miniature fiber optic spectrometer (USB2000, Ocean Optics), a 1-m liquid-waveguide capillary flow cell (LWCC-3100, WPI) and a tungsten light source (FO-6000, WPI). The detection limits for HONO and NO2 are 6 pptv and 15 pptv, respectively.

A 450-watt medium pressure mercury arc lamp (ACE Glass, model 7825) was placed 20 cm above the reactor as the light source. The light was filtered by a Pyrex sleeve to remove low wavelength UV light (<290 nm) and by a quartz well filled with circulated water to remove heat-generating infrared light. Temperature in the photochemical reactor increased slightly during light exposure, by 1–2 °C. Effective UV intensity was monitored using a nitrate actinometer34.

Surface preparation

The Pyrex glass surface was first cleaned thoroughly with Micro-90 cleaning solution (Cole-Parmer), and then repeatedly rinsed by ethanol and DI water. The cleaned Pyrex glass surface was dried in a vacuum desiccator before applying surface coating.

Thirteen model organic compounds with and without chromophores are chosen as proxies for the naturally occurring organic matters in the atmospheric environment. HNO3 with or without model organic compounds were coated onto Pyrex glass surface by applying 0.1 ml coating solution of known concentrations, and spreading the solution out uniformly with a hydrophobic Teflon blade. Organic solutions were made from their corresponding sodium salts for better water solubility and were acidified to pH ~4 with 1 M H2SO4 before being mixed with HNO3 solution. HNO3 mainly partitions in the NO3− form at pH value of ~4. The coated surface with or without model organic compounds was allowed to dry overnight in the vacuum desiccators before use in the light exposure experiments.

To quantify the amount of HNO3 exposed to light, the coated HNO3 was carefully rinsed off from the Pyrex glass surface with 10 ml 1% NH4Cl buffer solution (pH = 8.5). The wash solution was analyzed immediately in a LPAP system with a copperized Cd-column to convert nitrate into nitrite37. HNO3 surface density was then calculated from the determined HNO3 amount and the geometric area of the Pyrex glass surface (62 cm2). A range of HNO3 surface density of (0.1–80) × 10−6 mol m−2 was used to simulate sub-monolayer and multilayer conditions38. To account for the light absorption of organic compounds, absorbance spectra in the wavelength of 200–600 nm of the coating solution of HNO3 and organic compounds with a pH value around 4 were scanned using a UV-visible spectrometer (JENWAY, model 6405) with 1-cm path length (Fig. S2).

Background level and corrections

Several corrections are made when calculating HONO and NO2 production rates and photolysis rate constants. Background signals from dark experiment were used as experiment baseline and were subtracted from the light exposure signals. Blank signals contributed from photolysis of deposited HNO3 on reactor and window surface were corrected by subtracting one half of the blank signals from the light exposure signals, since the bottom of the flow reactor was covered by the Pyrex glass surface in light exposure experiment. Photolytic losses of HONO and NO2, the products from HNO3(s) photolysis, in the flow reactor during light exposure experiment were also considered. With a residence time of about 30 seconds in the flow reactor, ~5% HONO loss was calculated, and about 25% NO2 loss was observed when a gaseous NO2 standard (Matheson Tri-Gas Inc., CP) was introduced into the flow reactor.

The production rate (nmol s−1) of HONO, PHONO, and production rate of NO2, \({P}_{N{O}_{2}}\), were calculated by:

$${P}_{HONO}=\frac{{C}_{si}^{HONO}-{C}_{ri}^{HONO}/2}{60\times 1000}\times {F}_{l}^{HONO}\times \frac{450}{400}\times \frac{1}{0.95}\,$$
(5)
$${P}_{N{O}_{2}}=\,\frac{{C}_{si}^{NO2}-{C}_{ri}^{NO2}/2}{60\times 1000}\,\times {F}_{l}^{NO2}\times \frac{450}{400}\,\times \frac{1}{0.6\times 0.75}\,$$
(6)

where Csi and Cri are the concentrations (nM) of HONO or NO2 measured in the scrubbing solutions during the light exposure and the blank control experiments, respectively; Fl is the scrubbing solution flow rates at 0.24 ml min−1 and 0.4 ml min−1 for HONO and NO2, respectively; the ratio of \(\frac{450}{400}\) is the correction for minor overflow loss from the reactor; the coefficient of 0.6 is the collection efficiency in the NO2 channel; the coefficients of 0.95 and 0.75 are the corrections for photolysis losses of HONO and NO2, respectively.

The photolysis rate constants (s−1) of HNO3 leading to productions of HONO (\({j}_{HN{O}_{3}\to HONO}\)) and NO2 (\({j}_{HN{O}_{3}\to N{O}_{2}}\)), and the overall photolysis rate constant of HNO3 (\({J}_{HN{O}_{3}(s)}\)) on the surface were determined by equations (Eqs 6–8):

$${j}_{{{\rm{HNO}}}_{3}\to {\rm{HONO}}}\,=\frac{{{\rm{P}}}_{{\rm{HONO}}}\,}{{{\rm{N}}}_{{{\rm{HNO}}}_{3}}}\times \frac{3.0\times {10}^{-7}}{{J}_{nitrate}}$$
(7)
$${j}_{{{\rm{HNO}}}_{3}\to {{\rm{NO}}}_{2}}=\frac{{{\rm{P}}}_{{{\rm{NO}}}_{2}}}{{{\rm{N}}}_{{{\rm{HNO}}}_{3}}}\times \frac{3.0\times {10}^{-7}}{{J}_{nitrate}}$$
(8)
$${J}_{HN{O}_{3}(s)}={j}_{HN{O}_{3}\to HONO}+{j}_{HN{O}_{3}\to N{O}_{2}}$$
(9)

where \({{\rm{N}}}_{{{\rm{HNO}}}_{3}}\) is the amount of HNO3 exposed to light; \({J}_{nitrate}\) is the photolysis rate constant of nitrate in the actinometer solution exposed to the light source34. The calculated photolysis rate constants from Equations (Eqs 3–8) have been normalized to tropical noontime conditions on the ground (Solar elevation angle ϴ = 0°) where photolysis rate constant is ~3 × 10−7 s−1 for aqueous nitrate and ~7 × 10−7 s−1 for gaseous HNO334. It should be pointed out that the calculated photolysis rate constant \({J}_{HN{O}_{3}(s)}\) in equation (Eq. 8) is based on the productions of HONO and NO2, the two dominant products. It may underestimate the true photolysis rate constant of HNO3 on the surface if other products are also produced and not corrected for. Production of NO from HONO and NO2 photolysis has been corrected in equations (Eq. 4) and (Eq. 8). The reproducibility of results from repeated experiments was better than 30%. The overall uncertainty of the measurement was determined to be within 50%, considering the uncertainty contributed by measurements of HONO, NO2, HNO3(s) amount, gas phase flow rate, liquid phase flow rate and effective light intensity.