Investigations on HONO formation from photolysis of adsorbed HNO3 on quartz glass surfaces

During the last few decades, nitrous acid (HONO) has attracted significant attention as a major source of the OH radical, the detergent of the atmosphere. However, the different daytime sources identified in the laboratory are still the subject of controversial discussion. In the present study, one of these postulated HONO sources, the heterogeneous photolysis of nitric acid (HNO3), was studied on quartz glass surfaces in a photo flow-reactor under atmospherically relevant conditions. In contrast to other investigations, a very low HNO3 photolysis frequency for HONO formation of J(HNO3 HONO) = 2.4 10 7 s 1 (01 SZA, 50% r.h.) was determined. If these results can be translated to atmospheric surfaces, HNO3 photolysis cannot explain the significant HONO levels in the daytime atmosphere. In addition, it is demonstrated that even the small measured yields of HONO did not result from the direct photolysis of HNO3 but rather from the consecutive heterogeneous conversion of the primary photolysis product NO2 on the humid surfaces. The secondary NO2 conversion was not photoenhanced on pure quartz glass surfaces in good agreement with former studies. A photolysis frequency for the primary reaction product NO2 of J(HNO3 NO2) = 1.1 10 6 s 1 has been calculated (01 SZA, 50% r.h.), which indicates that renoxification by photolysis of adsorbed HNO3 on non-reactive surfaces is also a minor process in the atmosphere.


Introduction
During the last two decades, unexpectedly high nitrous acid (HONO) mixing ratios have been observed during daytime in remote, 1-5 semi urban [6][7][8][9] and urban regions. [10][11][12][13][14][15][16][17] These results stimulated laboratory investigations on potential daytime HONO sources, by (i) the reduction of nitrogen dioxide (NO 2 ) in the presence of organic photosensitisers, [18][19][20][21] (ii) the photolysis of adsorbed nitric acid [22][23][24][25][26] and (iii) bacterial production of nitrite in soil [27][28][29][30] and/or desorption of adsorbed HONO from soil surfaces during daytime. [31][32][33] Another intensely discussed source, the gas-phase reaction of excited NO 2 with water, 34 was found to be of minor importance as demonstrated by laboratory studies. 35,36 Also the photolysis of nitro-phenols or similar compounds is too slow to explain significant HONO levels in the daytime atmosphere. 37 Finally, a gas-phase HONO source recently postulated by Li et al., 38 i.e. the reaction of HO 2 ÁH 2 O complexes with NO 2 , could not be confirmed by the same group in simulation chamber experiments 39 and is also in conflict with recent aircraft measurements. 40 In summary, mainly ground surface sources have been identified to date in laboratory studies to explain atmospheric HONO formation during daytime, 41 which has also been confirmed by recent flux measurements in the atmosphere. 25,42,43 Among the different proposed sources, the photolysis of adsorbed nitric acid (HNO 3 ) has been suggested as an important HONO formation pathway, especially in rural areas.
Zhou and coworkers 24 have published a photolysis frequency of 1.2 Â 10 À5 s À1 for HNO 3 adsorbed on Pyrex glass, at 50% relative humidity under tropical noontime conditions, which is at least two orders of magnitude larger compared to those of liquid-and gas-phase HNO 3 . In addition, the same group recently presented a correlation of HONO fluxes above a forest canopy with HNO 3 adsorbed on leaves from the surrounding trees. 25 These observations are in contrast to a smog chamber study, in which photolysis of adsorbed HNO 3 was excluded as a source of photochemical HONO formation on Teflon surfaces. 44 In these experiments significant photochemical HONO formation was still observed at wavelengths 4370 nm, for which no significant light absorption of HNO 3 is expected. However, for HNO 3 adsorbed on fused silica and aluminium, a red-shift of the np* HNO 3 absorption band and more than two orders of magnitude larger UV cross-sections compared to the gas phase have been recently observed by Brewster angle cavity ring-down spectroscopy. [45][46][47][48] These observations have rekindled interest in HNO 3 as a potential HONO and NO 2 source. However, the mechanism of heterogeneous HNO 3 photolysis is still under discussion, and besides reaction (R1) secondary HONO formation pathways have also been proposed. 22,24,25,48 To elucidate the importance of reaction (R1) as a HONO source in the atmosphere, photolysis of adsorbed HNO 3 was studied in a photo flow-reactor under atmospherically relevant concentrations, humidity and actinic flux conditions.

Experimental setup
Experiments were performed in a cylindrical flow-reactor (5.0 cm i.d.) made of smooth quartz glass with an illuminated length of 56.5 cm (see Fig. 1). For a few experiments a similar flow reactor made of Pyrex glass was used. The reactor was irradiated with 6 UV-Vis lamps (Philips TL/05 40 Watt, 300-500 nm, l max E 370 nm; or Wolff System Helarium 40 W, 290-450 nm, l max E 350 nm), which were placed in an air-cooled (fan), UV-reflecting aluminium box. The actinic flux inside the photoreactor (see Fig. 2) was measured using a spectroradiometer (Dl = 1 nm; Metcon Inc., USA). Calculated photolysis frequencies J(NO 2 ) and J(HCHO) using published cross-sections and quantum yields [49][50][51][52] were compared with values directly measured by chemical actinometry of NO 2 and HCHO in the flow-reactor and showed good agreement (see Table 1).
Corrections of the observed HONO formation for its gasphase photolysis were typically only ca. 5 and 8% for the two types of lamps used, respectively. Photolysis frequencies of HONO ( J(HONO), see Table 1) were calculated using published cross-sections. 53

Instruments
HONO was measured using the sensitive LOPAP technique 10,55 with a detection limit and an accuracy of 2 pptV and 10%, respectively. NO 2 and NO were detected by a recently developed NO 2 -LOPAP 56 which was modified for the additional detection of NO, details of which will be published elsewhere. In short, NO passing the NO 2 collection stripping coil was oxidized to NO 2 by an acidic KMnO 4 solution in a consecutive stripping coil and was detected downstream using the Griess-Saltzman reaction, similar to the NO 2 channel of the instrument. Detection limits and accuracies for NO 2 /NO under the experimental conditions applied were 10/20 pptV and 7/10%, respectively. For the experimental determination of the photolysis frequency of HCHO an Aero Laser AL4001 Hantzsch instrument was used. 57 Finally, adsorbed HNO 3 was quantified by ion-chromatography (IC) 58 after elution from the flow-reactor surface with ultra-pure water (MilliPore). IC detection was also used to measure gasphase levels of HNO 3 after bubbling through ultrapure water. The accuracy of the nitrate concentration determined by IC was 7% and those of HNO 3 in the gas phase and on the surface were 10 and 12%, respectively.

Experimental procedure
Prior to the experiments the reactor was cleaned with HF (5%) followed by ultra-pure water to minimize the contamination on the irradiated reactor surfaces. This was confirmed in blank experiments in which neither in the dark nor under irradiation any significant formation of HONO or NO x was observed. Different reactant mixtures were generated by mixing humidified synthetic air with a pure HNO 3 gas-phase source 58 and were flushed through the flow-reactor. For some experiments low levels of NO 2 (Messer, 2 ppmV) were also added using a calibrated 10 mL min À1 flow meter and a needle valve. During one experiment a mixture of 29 VOCs (methane, ethane, propane, n-butane, iso-butane, n-pentane, n-hexane, n-heptane, n-octane, ethene, propene, iso-, trans-, and cis-butene, isoprene, b-pinene, limonene, ethine, benzene, toluene, o-, m-, and p-xylene, methanol, ethanol, n-propanol, n-butanol, acetone, and methacrolein) with   50 The absorption spectrum s of HNO 3 adsorbed on fused silica is also shown. 45,47  individual mixing ratios in the range 48-69 ppbV was added using a 100 mL min À1 flow controller. In some experiments additional quartz glass plates were added to the flow-reactor to increase the irradiated surface area. The total gas flow rate was typically 2.7 L min À1 , but was varied between 1.1 and 3.4 L min À1 in a few experiments leading to reaction times in the range 26-80 s. When constant reactant concentrations were reached, the lamps were switched on and the system was analysed for photochemical HONO and NO x formation. Immediately after each experiment and after disconnection from the HNO 3 source, the reactor was eluted with ultra-pure water and the amount of adsorbed HNO 3 was quantified. After saturation of the reactor surfaces with the continuously running HNO 3 source (typically overnight), adsorbed HNO 3 levels were assumed to be constant during each experiment. This assumption is considered plausible as the HNO 3 source produced fairly stable gas-phase levels, as demonstrated also in a former laboratory study. 58 In addition, constant gas phase HNO 3 levels at the end of the flow tube confirmed equilibrium adsorption during the experiments. This observation also indicates that the surface concentration of HNO 3 was homogeneously distributed on all the irradiated surfaces. For the calculation of surface concentrations and flux densities the geometric surface area of the flow tube was used.

Results and discussion
In Fig. 3, a typical experiment is shown in which adsorbed HNO 3 was irradiated without (first and last step in Fig. 3) and with different levels of added NO 2 . In the absence of additional NO 2 , clear positive steps in the HONO, NO and NO 2 concentrations were observed, pointing to their photochemical production.

Surface dependency
To verify whether HONO formation was caused by the photolysis of adsorbed HNO 3 , reaction (R1), and not by its gas-phase photolysis, the surface area of the reactor was varied. For these experiments, additional quartz glass plates of different sizes were added to the flow-reactor. The HONO production during irradiation increased linearly with the surface area ( Fig. 4), which demonstrates the heterogeneous nature of the process involved.
This result is in agreement with theoretical calculations performed using known gas-phase absorption cross-sections of HNO 3 , 51 measured actinic flux, HNO 3 gas-phase concentration, residence time in the flow-reactor and assuming an upper limit quantum yield for HONO formation of one (for calculations, see eqn (3)). The calculated maximum HONO formation by HNO 3 photolysis in the gas phase of only 6 pptV is much smaller compared to the experimental results shown in Fig. 4 and thus, a heterogeneous reaction is proposed here.

HNO 3 dependency
By varying the gas-phase concentration of HNO 3 in the range 2-120 ppbV, different quantities of surface adsorbed HNO 3 in the range 0.15-3.3 Â 10 14 molecules cm À2 were investigated. Recently, a surface monolayer concentration of HNO 3 of 1.1 Â 10 14 molecules cm À2 was estimated based on its van der Waals radius, 45 which is in fair agreement with the measured value for SiO 2 particles (7 Â 10 13 molecules cm À2 ). 59 Thus, the investigated surface levels varied from sub-monolayer to multilayer coverage, which were obtained at atmospherically relevant low ppbV mixing ratios of the sticky HNO 3 .
Photochemical HONO formation was linearly correlated with the concentration of adsorbed HNO 3 , both, for the long and short wavelength lamps used (see Fig. 5). This observation is in line with a first-order surface photolysis process, reaction (R1), described by a photolysis frequency of adsorbed HNO 3 for HONO formation, J(HNO 3 -HONO), which is defined by: Here, d[HONO] surface /dt represents the HONO formation rate described as a surface flux density (molecules cm À2 s À1 ).
From weighted orthogonal regression plots according to eqn (2) and after correction for gas-phase photolysis of HONO, photolysis frequencies J(HNO 3 -HONO) of (1.24 AE 0.19) Â 10 À6 s À1 and (3.4 AE 1.8) Â 10 À7 s À1 were determined at B50% r.h. for the short-(sw) and the long-wavelength (lw) setups, respectively, (see Fig. 5). In between the experimental errors photolysis frequencies were found to be independent of the gas flow rate.
The measured values are much lower than those obtained in another study on HONO formation from the photolysis of adsorbed HNO 3 on glass surfaces, 24 for which photolysis frequencies of B10 À5 s À1 were calculated under atmospheric conditions. The difference between the two studies is actually even higher, since the small values determined in the present study are still upper limits as the lamps used in the experiments have higher actinic fluxes compared to the atmosphere (see Fig. 2, Table 1 and Section 4). Potential underestimation of the surface area of the used quartz tube, e.g. by any surface roughness, will not affect the measured J(HNO 3 -HONO) since the surface area cancels out in eqn (2) in the two terms D[HONO] surface and [HNO 3 ] surface . In addition, the potential inhomogeneous surface concentration throughout the flow tube would also not affect the measured J(HNO 3 ), since first order kinetics was observed.

Wavelength dependency
To estimate the atmospheric implication of the HNO 3 photolysis, the results obtained in laboratory experiments cannot be used directly, since the actinic flux spectra of laboratory systems are typically very different from those of the atmosphere (cf. Fig. 2).
To calculate atmospherically relevant HNO 3 photolysis frequencies for HONO formation according to: spectroscopic information on the absorption cross-section (s(HNO 3 ) l ), the quantum yield (f(HNO 3 -HONO) l ) and the actinic flux spectra (F l ) of the laboratory system and the atmosphere are necessary. Since the absorption cross-sections and quantum yields are still uncertain, extrapolation to atmospheric conditions, as done in another study, 24 is highly questionable. To better estimate the atmospheric implications of HNO 3 photolysis, a limited wavelength dependence of the HONO formation was investigated in the present study using two different sets of lamps (see Fig. 5), for which the spectra are shifted by B20 nm (see Fig. 2). The ratio of the experimentally determined photolysis frequency, J(HNO 3 -HONO) sw /J(HNO 3 -HONO) lw , of 3.6 (see Section 3.2) is much lower compared to theoretical ratios of 10.9 and 14.1 for the liquid-and gas-phase photolysis, respectively. The latter was calculated using eqn (3) and known liquid-60 and gasphase 51 absorption cross-sections, measured actinic fluxes and assuming a wavelength independent quantum yield. 49 This result is another indication of surface photolysis of HNO 3 and thus confirms recent publications on a red-shift of the np* absorption of adsorbed HNO 3 . 45,47 Using these published cross-sections, measured HONO formation (see Fig. 5) and actinic fluxes (see Fig. 2), formal quantum yields for HONO formation, f(HNO 3 -HONO), for the heterogeneous photolysis reaction (R1) of 6.9 Â 10 À5 and 8.4 Â 10 À5 were calculated using eqn (3) for the shortand long-wavelength setups, respectively. Taking into account the uncertainties of the measured HONO formation, actinic flux spectra and published absorption cross-sections, the similarity of the two values indicates no strong wavelength dependence of the quantum yield, and thus an average value of f(HNO 3 -HONO) = (7.6 AE 1.5) Â 10 À5 is preferred and applied in further calculations.

Influence of relative humidity
Similar to the study by Zhou et al., 24 a strong influence of the relative humidity on the HONO formation inside the flow-reactor was observed. HONO immediately decreased to very low levels after stopping the humidifier, indicating a direct influence of water on the HONO production. There are different potential explanations for this observation: (i) photochemical HONO formation is active only for H 2 O/HNO 3 clusters 26 and/or (ii) is caused by secondary chemistry via the additional reaction product NO 2 , for which strong humidity dependence of the heterogeneous HONO formation is known. 61

NO 2 dependency
To distinguish between these two explanations, the NO 2 dependence of the HONO formation was studied. Besides HONO, NO 2 is a main product of the HNO 3 photolysis (see Fig. 3): Thus, in addition to possible direct HONO formation via HNO 3 photolysis (R1), the reaction product NO 2 may also form HONO via secondary heterogeneous conversion: 61 or by the reaction of NO 2 with chemisorbed water on the reactor surfaces: 62 Furthermore, formation of electronically excited NO 2 was recently proposed in the photolysis of adsorbed HNO 3 on aluminium surfaces, which may react with adsorbed water: 46,48 Finally, based on theoretical calculations, photolysis of complexes of HNO 3 with nitrogen oxides has been proposed to explain HONO formation by the photolysis of adsorbed HNO 3 . 63 To elucidate the contribution of NO 2 to the observed HONO formation, NO 2 was added to the reaction mixtures (see Fig. 3). From this and other similar experiments total HONO production rates were calculated from the concentration differences between the flow-reactor bypass and the dark or light steps, respectively, and plotted against the final NO x level at the flow-reactor exit (see Fig. 6). NO x was used here instead of the proposed precursor NO 2 to take into account the significant NO 2 photolysis in the flow-reactor (see Table 1). Thus, NO measured at the exit of the reactor mainly originated from the fast gas-phase photolysis of NO 2 : and was available as a reactant for heterogeneous HONO formation, reactions (R3)-(R5). Two important conclusions can be drawn from the experimental data: (a) HONO formation depends strongly on the NO 2 concentration and within the error of the intercept (AE2 Â 10 8 molecules cm À2 s À1 ) extrapolation to 0 ppbV NO x indicates no additional HONO formation by HNO 3 photolysis alone (R1) under the experimental conditions applied. Thus, it is proposed that the observed HONO formation is caused by the heterogeneous conversion of NO 2 on the reactor surfaces, reactions (R3) or (R4).
(b) HONO formation by heterogeneous NO 2 conversion in the dark and under irradiation is similar, confirming a proposed dark formation of HONO, e.g. by reaction (R3) on clean quartz glass surfaces. This result is in excellent agreement with other studies, 26,44 in which no photoenhancement of reaction (R3) was observed. Thus, the experimental results of the present study do not confirm photochemical HONO formation via excited NO 2 , reaction (R5), as proposed in another study by Abida et al. 48 when using excimer laser photolysis. Very recently, the same group 64 also excluded the direct formation of excited NO 2 at least for the gas-phase photolysis of HNO 3 and proposed secondary HONO formation by the heterogeneous reaction (R3). This conclusion is in agreement with studies of reaction (R5) in the gas phase, 35,36 in which HONO formation via excited NO 2 was excluded at atmospherically relevant actinic fluxes. Finally, the missing light-enhancement also excludes photolysis of complexes of HNO 3 with nitrogen oxides as sources of HONO. 63 Based on these two conclusions a two-step formation mechanism of HONO is proposed here. First, NO 2 is formed in the photolysis of adsorbed HNO 3 , reaction (R2), at longer wavelengths compared to the gas and liquid phases. Second, NO 2 is consecutively converted into HONO via the heterogeneous dark reactions (R3) or (R4) on the pure quartz glass surfaces. This consecutive formation of HONO via the intermediate NO 2 was already proposed for the photolysis of nitrate in snow 22,65 and in the aqueous phase. 66 However, for snow the significant redshift of the nitrate absorption compared to the liquid nitrate spectra was not observed leading to negligible NO x formation at wavelengths 4345 nm. 67 Finally, the observed NO formation (see Fig. 3) is explained by the gas-phase photolysis of NO 2 , reaction (R6).
To determine HNO 3 photolysis frequencies for the primary reaction product NO 2 : the sum of photochemically formed NO 2 , HONO and NO was considered in calculations similar to eqn (2). Using weighted orthogonal linear regression plots, photolysis frequencies J(HNO 3 -NO 2 ) of (6.6 AE 1.3) Â 10 À6 s À1 and (1.3 AE 1.0) Â 10 À6 s À1 were determined at B50% r.h. for the short and the long wavelength setups, respectively. Using the absorption cross-sections of adsorbed HNO 3 from the studies of Zhu et al. 45 and Du and Zhu, 47 measured actinic fluxes of both setups and measured values of J(HNO 3 -NO 2 ) from the present study, similar quantum yields for NO 2 formation, f(HNO 3 -NO 2 ), of (3.6 AE 0.8) Â 10 À4 and (3.2 AE 2.5) Â 10 À4 were determined for the short and the long wavelength setups, respectively. The averaged quantum yield for both View Article Online setups of (3.4 AE 2.6) Â 10 À4 is orders of magnitude lower compared to the studies of Zhu et al. 46 and Abida et al., 48 in which quantum yields near unity were observed in excimer laser photolysis experiments at 308 and 351 nm. Reasons for the strong discrepancy are still unclear and may be explained by: (a) Overestimation of the HNO 3 absorption cross-sections in the studies of Zhu et al. 45 and Du and Zhu. 47 In these studies, the concentrations of adsorbed HNO 3 were only indirectly determined by adsorption isotherms based on measured light absorption and by using an estimated van der Waals radius for HNO 3 resulting in a monolayer coverage only at very high gasphase levels (15-20 ppmV) of the highly sticky HNO 3 . In contrast, in the present study three orders of magnitude lower gas-phase concentrations (B10 ppbV) were necessary to obtain similar levels of adsorbed HNO 3 , which were directly measured by ion chromatography. If surface levels of HNO 3 were much higher than proposed in the studies of Zhu et al. 45 and Du and Zhu, 47 absorption cross-sections may be overestimated, which was also recently proposed by Tadić; 68 (b) Overestimation of the quantum yields f(HNO 3 -NO 2 ) in the studies of Zhu et al. 46 and Abida et al. 48 caused by higher photon flux densities of the excimer laser used for the photolysis experiments compared to atmospheric conditions. High photon flux densities may have caused stronger NO 2 formation, e.g. by a potential multi-photon excitation, which was also recently observed for the gas-phase reaction of NO 2 * + H 2 O. 36 Independent of which explanation is correct, photolysis of pure HNO 3 will be of much lower importance for renoxification than previously assumed, since the presented NO 2 flux densities were determined under realistic atmospheric conditions (HNO 3 level, humidity, actinic flux). A strong indication for an overestimation of either the absorption cross-sections or the quantum yield of adsorbed HNO 3 in the former studies [45][46][47] is the resulting unrealistic short photolytic lifetime of adsorbed HNO 3 of only B5 min (see Section 4). If published absorption cross-sections are overestimated, our calculated quantum yields, which are based on these values, will be strongly underestimated.

Influence of VOCs
To explain enhanced HONO formation, photosensitized conversion of HNO 3 /nitrate by adsorbed organic compounds was recently proposed. [69][70][71][72] In addition, increased HONO formation by scavenging of OH radicals formed via reaction (R2) by organic compounds was proposed for the liquid phase photolysis of nitrate. 66 To test the influence of organic compounds on the reaction system, in one experiment a GC calibration mixture of 29 VOCs (rC10, see Section 2.3. for details; each with a final mixing ratio of 3-4 ppbV) was continuously added to the humidified HNO 3 mixture, both, during surface saturation overnight and during the photolysis. While some of the used VOCs clearly do not adsorb onto surfaces (e.g. C1-C4), others are expected to be surface active (e.g. b-pinene, limonene, xylenes, n-butanol, methacrolein) and substituted aromatic compounds are even proposed to participate in redox reactions of NO y . 18 However, within the experimental errors photochemical HONO formation was not enhanced compared to the absence of the VOCs. A similar result was obtained when the flow tube was not cleaned after flushing with laboratory air for one day before the photolysis experiment. Thus, at least for the smaller VOCs and at atmospherically relevant concentrations, HONO formation by photosensitized conversion of HNO 3 could not be confirmed. However, further experiments using more complex VOCs, known as photosensitizers (e.g. aromatic carbonyls, hydroxy-PAHs, humic acid etc.), are recommended in the future.

Atmospheric implications
Based on the experimental results HONO formation from the photolysis of adsorbed nitric acid was estimated under typical atmospheric conditions. The present study indicates no direct photochemical HONO formation via reaction (R1), but secondary formation by the heterogeneous dark conversion of NO 2 on humid surfaces, reaction (R3) or (R4). At high NO x levels, reaction (R3) can be a significant night-time ground surface source of HONO in the atmosphere, 12,61 but is too slow to explain daytime HONO levels. 41 Here, other NO 2 reactions, e.g. the photosensitized conversion on humic acid surfaces have been proposed. 19,20 Although HONO is not the primary HNO 3 photolysis product, reaction (R1), the formal quantum yield for HONO formation of f(HNO 3 -HONO) = (7.6 AE 1.5) Â 10 À5 , as determined in Section 3.3., was considered here. Using this wavelength independent quantum yield, published absorption cross-sections of adsorbed HNO 3 45,47 and the upper limit actinic flux in the atmosphere at 01 SZA 50 a formal photolysis frequency J(HNO 3 -HONO) of only 2.4 Â 10 À7 s À1 was derived using eqn (3). This value is almost two orders of magnitude lower compared to another laboratory study by Zhou and coworkers, 24 in which atmospheric HONO formation in a rural environment could be well explained by reaction (R1). In contrast, using the results of the present study and assuming a high surface coverage of HNO 3 on atmospheric surfaces of 10 14 cm À2 (Bone monolayer 45 ), an upper limit HONO flux density of only 2.4 Â 10 7 cm À2 s À1 can be calculated using eqn (1) at 01 SZA. This upper limit is 2-3 orders of magnitude lower compared to HONO flux densities typically observed over irradiated surfaces in the atmosphere. 25,42,43 Thus, HONO formation by photolysis of pure HNO 3 is too slow to explain field observations, even in rural environments. Reasons for the strong discrepancy between the present study and the one by Zhou et al. 24 are yet unclear but may possibly be explained by: (a) the indirect extrapolation of the laboratory results to the atmosphere via the chemical actinometry used by Zhou et al., 24 by which OH radicals formed in the liquid phase photolysis of nitrate were used to quantify J(HNO 3 ) for the red-shifted (compared to the liquid phase) photolysis of adsorbed HNO 3 . 45,47 In contrast, in the present study a well calibrated spectroradiometer was used to quantify actinic fluxes and to calculate J(HNO 3 ) using eqn (3). Here, the wavelength dependent absorption cross-sections of adsorbed HNO 3 45,47 were considered; (b) different surface properties of the reactors and the different NO 2 levels used. While in the present experiments, the flowreactor was chemically cleaned by HF prior to each experiment, impurities on irradiated surfaces and higher NO 2 levels in the experiments by Zhou et al. 24 may have caused increased HONO yields via reaction (R3) and (R4), or by photosensitized conversion of NO 2 on adsorbed organic impurities. [18][19][20][21] Alternatively, direct photosensitized conversion of HNO 3 /nitrate by organic impurities [69][70][71][72] may also be another explanation for the higher HONO yields observed by Zhou et al. 24 However, in the present study the latter mechanism could not be confirmed at least for a mixture of 29 smaller VOCs (rC10) or when the reactor was not cleaned after flushing with laboratory air for one day before the experiment. Thus, in the future, HONO formation from the photolysis of adsorbed HNO 3 should be studied on more realistic organic surfaces, for example, on leaves, urban grime or humic acid. However, for the photolysis of pure HNO 3 adsorbed on non-reactive surfaces, HONO formation is of minor importance.
Similar conclusions can be drawn for a possible renoxification by photolysis of pure adsorbed HNO 3 in the absence of photosensitizers. Extrapolation of the experimental results from the present study to atmospheric conditions (see above) yields an upper photolysis frequency limit of J(HNO 3 -NO 2 ) = 1.1 Â 10 À6 s À1 (01 SZA). This value translates into photolytic lifetimes of HNO 3 on surfaces of B30 days when the diurnal variability of the actinic flux in the atmosphere is considered. This contrasts starkly with an atmospheric photolytic lifetime of adsorbed HNO 3 of only B5 min (01 SZA), which is derived when published cross-sections and quantum yields are used. 48 Such a short lifetime is highly unreasonable and would result in negligible amounts of nitrates on atmospheric surfaces during daytime, which is not in agreement with direct field observations. 25

Summary
In the present study heterogeneous formation of nitrous acid (HONO) from photolysis of adsorbed HNO 3 was studied in a photo flow-reactor, from which a HNO 3 photolysis frequency for HONO formation, J(HNO 3 -HONO), of only 2.4 Â 10 À7 s À1 is extrapolated for atmospheric noontime conditions (01 SZA). The slow HONO formation is explained by secondary heterogeneous formation via the primary HNO 3 photolysis product NO 2 . On pure quartz glass surfaces, the heterogeneous conversion of NO 2 into HONO shows no photoenhancement, in good agreement with other studies in simulation chambers.
For the primary photolysis product NO 2 a low photolysis frequency of J(HNO 3 -NO 2 ) = 1.1 Â 10 À6 s À1 (01 SZA) is calculated. From the wavelength dependence of the HONO formation a red-shift of the UV absorption of adsorbed HNO 3 is inferred in agreement with other recent studies. However, in contrast to these studies, it is concluded that photolysis of pure adsorbed HNO 3 is only a minor source of HONO and NO 2 in the atmosphere, at least on non-reactive substrates. For the future, HONO formation by the photolysis of adsorbed HNO 3 should also be studied on more complex surfaces, for example, on leaves, urban grime or humic acid.