Formation of reactive nitrogen oxides from urban grime photochemistry

Impervious surfaces are ubiquitous in urban environments and constitute a substrate onto which atmospheric constituents can deposit and undergo photochemical and oxidative processing, giving rise to “urban grime” films. HNO3 and N2O5 are important sinks for NOx in the lower atmosphere and may be deposited onto these films, forming nitrate through surface hydrolysis. Although such deposition has been considered as a net loss of NOx from the atmosphere, there is increasing evidence that surface-associated nitrate undergoes further reaction. Here, we examine the gas phase products of the photochemistry of real, field-collected urban grime using incoherent broadband cavity-enhanced absorption spectroscopy (IBBCEAS). Gas phase nitrogen oxides are emitted upon illumination of grime samples and their production increases with ambient relative humidity (RH) up to 35 % after which the production becomes independent of RH. These results are discussed in the context of water uptake onto and evaporation from grime films.


Introduction
Atmospheric NO x (= NO + NO 2 ) is an important reactant in the formation of urban pollutants such as ground-level O 3 , while HONO (g) is an important photochemical source of OH (Finlayson-Pitts and Pitts Jr., 1999).Therefore, in order to quantify the local atmospheric oxidative capacity, it is important to understand the processes mediating the concentrations of these species in the urban atmosphere.A major sink for nitrogen oxides in the troposphere is the formation of gas phase HNO 3 or N 2 O 5 , followed by the deposition of these species to surfaces and their subsequent hydrolysis to form nitrate.This anion is considered to be a sink for the gas phase NO x because its aqueous phase photochemistry is very slow.However, there is an increasing body of literature which suggests that surface bound nitrate and HNO 3 are not terminal sinks, but rather can undergo further recycling back to the gas phase.For example, HNO 3 has been shown to react on surfaces with gas phase NO and HONO to form NO 2 (Mochida and Finlayson-Pitts, 2000;Rivera-Figueroa et al., 2003;Saliba et al., 2001), and photochemical mechanisms for the conversion of HNO 3 and nitrate anion to gaseous nitrogen oxide species have been proposed on a variety of surfaces including glass (Zhou et al., 2003), snow (Grannas et al., 2007;Mochida and Finlayson-Pitts, 2000;Rivera-Figueroa et al., 2003;Saliba et al., 2001), organic films (Handley et al., 2007), leaves (Zhou et al., 2011), plants (Ye et al., 2016), building materials (Ye et al., 2016) and mineral oxide surfaces such as aluminum oxide and zeolite (Gankanda and Grassian, 2013;Nanayakkara et al., 2014;Rubasinghege and Grassian, 2009;Schuttlefield et al., 2008).There is particular interest surrounding whether such processes could explain an as of yet unconfirmed source of daytime HONO in urban centers.Field studies have indicated that this missing source is photochemical in nature and acts at or near ground level (Lee et al., 2016;Wong et al., 2013Wong et al., , 2012;;Young et al., 2012).Other processes, such as reactions of NO y (total reactive nitrogen) on aerosols (Ma et al., 2013) and soil-mediated processes (Oswald et al., 2013;Scharko et al., 2015), have also been proposed but have not been confirmed at this time.
When studying atmospheric surface reactions, an oftenoverlooked surface is that of human-made structures (e.g., buildings, roadways).These surfaces, when exposed to the Published by Copernicus Publications on behalf of the European Geosciences Union.
atmosphere, become coated in a complex surface film over time due to the deposition and subsequent processing of atmospheric constituents (Chabas et al., 2008(Chabas et al., , 2012;;Diamond et al., 2000;Duigu et al., 2009;Favez et al., 2006;Ionescu et al., 2006;Lam et al., 2005;Law and Diamond, 1998;Liu et al., 2003;Lombardo et al., 2010Lombardo et al., , 2005;;Simpson et al., 2006;Wu et al., 2008a, b).Referred to as "urban grime", these films have generally been thought of as merely a surface for deposition as a terminal sink for species.However, there is increasing understanding that these films could also play a role in mediating environmental cycling.Most attention has been brought to the idea that the films can sequester gas phase compounds and enhance pollutant concentrations in rainfall runoff (Diamond et al., 2010(Diamond et al., , 2001;;Priemer and Diamond, 2002), but there is evidence suggesting that they can also impact the reactivity of species contained within the film, such as PAHs (Ammar et al., 2010;Kwamena et al., 2007), and nitrate / HNO 3 (Baergen and Donaldson, 2013;Baergen et al., 2015).
Additionally, it has been predicted that there is enough water present on all environmental surfaces, even those hydrophobic in nature, to influence heterogeneous reactions (Sumner et al., 2004).Rubasinghege and Grassian (2013) have discussed the role of water on environmental surfaces outlining a wide range of mechanisms through which water can impact reactivity.These include altering reaction pathways, promoting hydrolysis reactions, ionic dissociation and solvation of ions, inhibiting reactivity through blocking reactive sites, enhancing ion mobility on the surface and altering the stability of surface species.For example, both the extent of reaction and the distribution of products change as a function of relative humidity (RH) for nitrate photolysis on aluminum oxide and Pyrex substrates, but the response is substrate dependant (Rubasinghege and Grassian, 2009;Zhou et al., 2003).Such studies have generally investigated the impact of water on atmospheric surface chemistry by varying the ambient RH.However, because different surfaces have different water affinities, they may be expected to display different responses to changes in relative humidity.For example, a study by Nguyen et al. (2015) shows that estimating water content in the aerosol, rather than just using RH data, is important for predicting the formation of biogenic secondary organic aerosol.Related to grime surfaces, Sumner et al. ( 2004) have shown how different surfaces, representing building surfaces, vary in their water uptake behavior.There are only minimal studies performed looking at water interactions with grime, but they show that grime films impact water uptake on surfaces (Baergen and Donaldson, 2013;Chabas et al., 2014).Thus it is important to characterize the change in surface water content as a function of RH as well as grime photochemistry.In the following we present results of experiments which monitor the photochemical release of gas phase nitrogen oxides from urban grime as a function of RH, in conjunction with water uptake measurements on grime.

Sample collection
Grime samples were collected by placing the substrate, either 3 mm diameter glass beads (Fisher Scientific) or quartz crystal microbalance (QCM) crystals, outside in downtown Toronto, Canada, for up to 1 year.The beads were placed on metal mesh shelves underneath a building overhang, sheltering the sampler from precipitation.Sunlight was blocked with a black cloth covering the front of the sample and a building blocking the sunlight from the other direction.The QCM crystals were placed in holders where the face of the crystal was facing the ground while the back was within the holder, preventing the collection of grime on this backside, which would impact the QCM response.In this way both types of samples were shielded from light and precipitation while still being open to the atmosphere.

Incoherent broadband cavity-enhanced spectroscopy (IBBCEAS)
Gas phase product formation was determined using IB-BCEAS.The system is described in full elsewhere (Reeser et al., 2013).Briefly, a 10 W LED with a maximum intensity at 372 nm was focused into a 100 cm cell.The cell was sealed with two highly reflective mirrors (> 99.95 % between 367 and 380 nm).The light escaping through the back mirror was collected by a lens and focused onto a fiber optic bundle, which was directed into a spectrograph with a CCD detector.The mirrors were continually purged using a flow rate of 25 mL min −1 of N 2 directed onto the mirror surfaces.Transmission spectra were collected for 30 s (averages of 30 scans with an integration time of 1 s each) over a wavelength range of 362 to 385 nm.The concentrations of HONO and NO 2 were calculated using the method described by Fiedler and Gherman (Fiedler et al., 2003;Gherman et al., 2008) and previously used by us (Reeser et al., 2013).This uses measured mirror reflectivity (Washenfelder et al., 2008), Rayleigh cross sections of the carrier gas (Bodhaine et al., 2010) and the absorption cross sections of NO 2 (Vandaele et al., 1998) and HONO (Stutz et al., 2000) to fit the experimental spectra with DOASIS software (Lehmann, 2009).DOASIS uses a linear least-squares method to fit the absorption bands to reference spectra and a polynomial to fit broad features such as those from Rayleigh scattering, Mie scattering and temperature drifts.The fit is optimized by including terms that allow for small shifts in absorption wavelengths and spread of peaks.A sample fit is displayed as the solid line in Fig. 1.Calculated detection limits (signal/noise = 3) are 1.50×10 11 molecules cm −3 (∼ 6 ppb) for NO 2 and 6.5×10 10 molecules cm −3 (∼ 3 ppb) for HONO.Experimental absorption spectrum fit with a reference HONO spectrum (Stutz et al., 2000) using DOASIS (Lehmann, 2009).This spectrum was measured at RH = 37 % and represents a concentration of 2.17 × 10 11 molecules cm −3 .

Photochemistry
Samples of 10.0 g of exposed glass beads were weighed into a glass petri dish for illumination.These were placed within a stainless steel chamber (3.2 × 2.2 × 1.5 ), and nitrogen was flowed through the chamber into the IBBCEAS cell at a rate of 0.3 L min −1 .RH and temperature in the chamber were monitored using a Traceable ® Memory Hygrometer/Thermometer.The reported accuracy is ±2 % at midrange and ±4 % elsewhere in the range of 10 to 95 % RH.
The calibration was checked by measuring the RH above a series of saturated salt solutions in comparison to the known deliquescent RH and was the same as the reported values within the stated uncertainties.
Nitrogen was flowed through the system for 1 h prior to illumination to establish a stable background in the spectrum, and equilibrate the water vapor in the chamber for the RH used in the experiment.The samples, initially at a RH of 35 %, were illuminated through a quartz window at the top of the sample chamber with a xenon arc lamp (λ > 295 nm) for 60 min.The light was then blocked, the signal allowed to return to baseline, and the RH adjusted for the next illumination period.After 60 min the sample was again illuminated for 60 min before blocking the light and repeating the cycle for a third time.Average concentrations were calculated for the second 30 min of illumination, where the signal appeared to reach steady state, and then normalized to the initial steady-state value at the RH of 35 % to adjust for experiment variability such as variations in sample, light intensity and temperature.These experiments were carried out without temperature control, with the chamber operating at a temperature between 28 and 34 • C during illumination.
Control experiments were carried out in which 10.0 g of clean beads was illuminated for 1 h at a relative humidity of 35 %.In addition, 10.0 g of the grime coated beads was subjected to heating up to 36 • C to study the impact of increased temperature on product formation.Both of these tests were completed in triplicate with a different sample being used each time.Neither experiment showed detectable levels of HONO or NO 2 .
The setup was further tested by flowing a known concentration of NO 2 through the empty chamber and IBBCEAS cell.A flow containing 6.0 ppm NO 2 in N 2 was diluted in a stream of N 2 down to (4.76 ± 2.4) × 10 12 molecules cm −3 at varying RHs.NO 2 and HONO steady-state concentrations measured at each RH were used to characterize the IB-BCEAS response to changes in humidity and the efficiency of NO 2 hydrolysis to HONO on the walls of the reaction chamber and IBBCEAS cell.

Ion analysis
Two different ion extraction techniques were used.For ion analysis of the illuminated beads, 4.0 g of grime-coated beads was shaken for 5 min with 3.00 mL of deionized water with a resistivity of greater than 18 M cm (Baergen et al., 2015).For ion analysis to accompany water uptake measurements, samples were collected on 5 cm × 7.6 cm pieces of window glass over the same time period as the quartz crystals and extracted with 45 mL deionized water.Glass samples were first placed in 25 mL of water and sonicated for 1 min.Each side of the glass was then washed twice with 5 mL of water.Solutions were filtered through a 0.2 µm IC Millex ® -LG syringe filter before being analyzed by ion chromatography on a Dionex ICS-2000; 1.33 mL samples were injected onto a concentrator/analytical column system: Ionpac ® TAC-ULP1/AS19 with KOH eluent for anion detection and Ionpac ® TCC-ULP1/CS17 with methanesulfonic acid eluent for cation detection.A second extraction resulted in concentrations of less than 10 % of the first extraction for all ions.The inorganic ion content of the grime used for photochemistry experiments is given in Table S1 in the Supplement.

Water uptake
The mass of water taken up onto an urban grime film as a function of RH was measured using a quartz crystal microbalance (QCM), as described in Demou et al. (2003).Grime was collected directly onto a quartz crystal and then placed in the QCM.The QCM was housed in a Plexiglas chamber whose humidity was increased by flowing air through a water bubbler at room temperature at a variable flow rate to maintain a rate of change of RH of 1 % min −1 and decreased by flowing dry air through the chamber at variable flow rates to maintain a rate of change of −1 % min −1 .The frequency change of the microbalance from the change in water content of the film was converted to a mass using the Sauerbrey equation ( m = C f ), where m is the mass change, C is a proportionality constant and f is the frequency change from the deposited mass.In a previous study, the Sauerbrey relationship was confirmed to hold for this apparatus only up to 1 % of the fundamental frequency of the crystal (Demou et al., 2003).Due to this mass restriction, samples for QCM analysis were collected for only 4 weeks instead of the 1 year for the photochemistry samples.The value of the constant C is reported to be 8.147 × 10 7 Hz cm 2 g −1 for the 0.550-inch, 6 MHz crystals used in this study (Sauerbrey, 1959).The RH was measured using a Traceable ® Memory Hygrometer/Thermometer.

Photochemical production of nitrogen oxides
Figure 1 shows a typical absorption spectrum collected upon illumination of a grime sample.One can see two features, typical of HONO absorption: a stronger signal at 368nm, and a second peak appearing at 384 nm, at the longer edge of our wavelength range.The IBBCEAS is also sensitive to NO 2 , which absorbs in this wavelength region.However, this molecule was not detected in any of the photochemical experiments performed here.We argue in the Supplement that NO 2 hydrolysis on the walls of the chamber and/or IBBCEAS cell (Finlayson-Pitts et al., 2002) would prevent NO 2 from being detected even if it was originally formed in the chamber.Because of this hydrolysis there is an uncertainty as to whether the HONO we observe in the illumination experiments was originally NO 2 , which was hydrolyzed prior to detection, or if it is HONO being produced directly from the sample.Therefore, we cannot attribute the observed HONO product exclusively to direct photochemistry of the grime sample; rather we use the HONO signal to indicate the combined total emission of NO 2 and HONO.We further note that the total product detected decreases when NO 2 is flowed through the apparatus in the light as compared to in the dark by approximately 60 %, indicating gas phase photolysis of products (see Fig. S1 in the Supplement).This highlights the importance of such considerations to be made whenever HONO and NO 2 are being measured.Each system needs to be classified individually over a range of RH conditions.
Figure 2 depicts the results of a typical experiment where a grime sample is placed within the chamber and exposed to three separate 60 min illumination periods at different relative humidities.The yellow highlighted regions indicate illumination.It is clear that nitrogen oxides are released to the gas phase during illumination and that the amount of products formed is dependent on the relative humidity.A repeat illumination of a sample at an RH of 35 % showed an average ratio of 0.88 ± 0.06 compared to the original illumination.This provides evidence of some precursor depletion due to illumination; however, the smaller signal level at 20 % RH, apparent in Fig. 2, indicates that there is an RH dependence to product formation, in addition to any sample depletion.The nitrate to sulfate ratio of the grime before and after illumination was used to examine nitrate behavior.Sulfate is not expected to have any photoreactivity on the film and thus was used to account for sample variability as was done in our previous work (Baergen et al., 2015).No nitrate loss was detected between water extracts of beads before and after illumination at 35 % for three 1 h periods.The average change in the nitrate to sulfate ratio from before to after illumination was 3.6 ± 6.6 %.There was also no nitrate loss detected for the samples that were heated for 3 h; these show an average change in the nitrate to sulfate ratio from before to after heating of 1.0 ± 3.6 %.The amount of nitrate loss expected during illumination, based on the integrated amount of gas phase nitrogen oxides produced, is in agreement with the above results.
In order to further investigate the RH dependence on product formation, the initial illumination period at 35 % RH was used to normalize the concentrations detected for the next illumination periods.These data are shown for a range of RH values in Fig. 3. Up to an RH of approximately 35 % the amount of products formed increases, after which product formation becomes independent of RH.At a RH of 0 %, no products were detected.However, from the NO 2 control experiments, there was evidence that gas phase NO 2 / HONO is lost to the chamber walls for these dry conditions, and thus this value was not plotted in Fig. 3.

Water uptake by grime samples
This interesting RH dependence of the amount of nitrogen oxides emitted photochemically from urban grime motivates the study of water uptake onto grime.Grime-water interaction has been reported before using ATR-FTIR with 1-weekold grime, showing equilibrium with ambient water vapor (Baergen and Donaldson, 2013).Chabas et al. (2014) also reported that mass measurements on 100-month-old films showed enhanced water uptake on grime-coated substrates compared to clean ones.Here we use 4-week-old samples collected throughout the year-long collection onto the glass beads, and look at both water uptake and evaporation, to better probe water-grime interactions.The uptake and loss curves displayed in Fig. 4   lected at different time points through the year normalized to the mass of major ions in the film (Cl − , NO − 3 , SO 2− 4 , Na + , K + , Mg 2+ and Ca 2+ ), extracted from a glass slide exposed to the atmosphere for the same length of time as the quartz crystal and scaled to the same surface area as the crystal.The shaded region indicates the 95 % confidence interval.The water uptake onto a clean crystal was subtracted from each sample's uptake curve before averaging, so the figure displays the mass of water taken up mediated by the grime itself.The degree of uncertainty captures some of the seasonality of grime water uptake, which will be discussed in an upcoming paper along with the seasonality of grime ion content.The water uptake onto and evaporation from grime are both smooth curves, with no indication of phase changes over the RH values spanned here.The lack of hysteresis also gives confidence that the illumination experiments reflect the true state of the "real" urban grime, as the film remains equilibrated with the ambient RH as this changes.

Discussion
The illumination of urban grime results in the release of gas phase nitrogen oxides in the form of NO 2 and/or HONO.While previously predicted (Baergen and Donaldson, 2013;Baergen et al., 2015), this is one of the first observations of such gas phase products.Field-collected grime samples were illuminated, without any alteration, clearly showing that urban grime is a source of nitrogen oxides back into the atmosphere.Our previous work, as well as that of others, has shown that nitrate is present within urban grime films (Favez et al., 2006;Lam et al., 2005) and that this nitrate is photolabile (Baergen and Donaldson, 2013;Baergen et al., 2015;Ye et al., 2016).Nitrate photolysis is known to form NO 2 within  2016) also looked at surfaces exposed to the atmosphere for much shorter exposure times detecting HONO and NO 2 at varying ratios depending on the surface (Ye et al., 2016).If HONO is a product of this chemistry one possible source is via the protonation of nitrite, another known product of aqueous nitrate photochemistry.However, our previous study was consistent with an alkaline film due to the loss of ammonium; therefore, we do not expect the film to be acidic enough for this mechanism to be important (Baergen et al., 2015).The organic fraction of the film could also play a role in the conversion of NO 2 to HONO, such as has been seen on organic surfaces such as humic acid (Stemmler et al., 2006) and PAH films (Ammar et al., 2010).NO 2 to HONO conversion could also occur through NO 2 hydrolysis within the film.Although it seems likely that nitrate is responsible for the observed chemistry due to its high concentration and known photoactivity on other surfaces, it is also possible that photochemically active organo-nitrogen compounds may be present, though they have yet to be detected within grime films.If present, these compounds may react as indicated by Han et al. (2013), who have reported the formation of R-NO, R-NO 2 and R-ONO 2 species on NO 2 exposed soot, which can photolyze to form NO and HONO.In contrast to our previous studies showing the photolability of nitrate in grime (Baergen and Donaldson, 2013;Baergen et al., 2015), the current study does not show nitrate depletion upon illumination.This apparent discrepancy can be explained by the difference in experimental methods between studies.In both previous studies, the films were "younger", with between 1 and 6 weeks of collection time, in comparison to the year-long collection here.In addition, for the Leipzig samples described in Baergen et al. (2015) the "light" sample was continually exposed to ambient sunlight, whereas in the present experiment, like the previous Toronto study (Baergen and Donaldson, 2013), the samples were shielded from the light for the entire collection and then illuminated in a controlled laboratory setting.Both of the previous studies suggested that only a portion of the film is photolabile and the current result indicates that this non-photoactive proportion of the film forms a greater proportion of the film over time.Continued growth of the film may block photoactive sites or bury photoactive components of the film, making a smaller portion available for reaction.Ye et al. (2016) found a logarithmic relationship between surface density of nitrate / HNO 3 and reaction rate which could also indicate only nitrate / HNO 3 on the surface remains reactive in comparison to the nitrate / HNO 3 within the film.Whether a film grown under continual exposure to ambient light would show the same trend is an open question.Exposure to precipitation could also impact the photoactive fraction, both in potential compositional changes as different fractions are removed from the film during precipitation and in preventing such long-term film growth.This large nonphotoactive fraction may also explain the disparity between the depletion of gas phase products over time and the lack of a corresponding nitrate drop; the photolabile fraction is small enough that the approximately 12 % loss of reactive precursor implied by the gas phase result is too small of a proportion of the total nitrate to be detected within the extracts of the whole film.
The photochemical release of gas phase NO 2 and/or HONO is clearly dependant on relative humidity and therefore, as seen through the water uptake experiments, on the water content of the film.In particular, the product formation increases as the amount of water on the film increases, up to a relative humidity of 35 % after which case, the chemistry is not impacted by further addition of water up to 60 %.This behavior is different from what has been seen from nitrate photolysis experiments on other surfaces.In a study performed on HNO 3 deposited on Pyrex glass, the combined NO x and HONO formation rate was highest at 0 % and decreased for 20 and 50 % while the reported HONO production rate was lowest at 0 % and then increased up to 80 % (Zhou et al., 2003).However, the authors assumed a constant NO 2 to HONO wall conversion independent of relative humidity taken from a measurement in a different system, and thus the determined ratios may not reflect the real distribution of products emitted as a result of the photochemistry (Zhou et al., 2003).Humidity dependence has also been seen for nitrate photochemistry on mineral dust surfaces.In this case, a minimum was seen for nitrate loss and NO 2 production at 0 % and a maximum at 20 % which subsequently decreased between 20 and 80 %, while NO production continually decreased from 0 to 80 % (Rubasinghege and Grassian, 2009).HONO production was not reported in this study.
The difference in nitrate photolysis behavior between grime and other surfaces as a function of RH is indicative of the grime providing a unique environment for the photochemistry.Many different mechanisms for the role of water in surface reactions have been discussed, such as enhanc-ing the mobility of reagents, allowing them to move to more photolabile positions within in the film or enhanced hydrolysis and dissociation of species such as HNO 3 , NH 4 NO 3 or N 2 O 5 producing more of the photolabile precursors (Rubasinghege and Grassian, 2013).The increased reactivity could also be the result of a viscosity change within the film.It is known that the viscosity of particles changes based on relative humidity (Renbaum-Wolff et al., 2013), and therefore it is expected that the same would be true for grime, with particles being a source to the film.The film's water uptake/evaporation curve is consistent with continuous viscosity change rather than phase transitions over the RH region studied.In a highly viscous film, the photochemical products are more likely to be trapped and thus recombine.However, a less viscous film would allow for faster diffusion and thus the release of products could become competitive and then dominate in comparison to recombination.Such an impact has recently been suggested to explain a smaller mass loss from illuminated secondary organic aerosol under low RH conditions in comparison to high (Wong et al., 2014), and faster PAH ozonation within a secondary organic aerosol coated particles at high RH as compared to lower RH (Zhou et al., 2013).This sort of behavior would not be anticipated for a clean glass or metal oxide surface.The leveling off of product formation at relative humidities greater than 35 % could indicate that a critical amount of water has been reached.The case of a viscosity effect would suggest that the process is no longer diffusion limited.Another process that could be playing a role is the re-adsorption of the products to the film, as discussed by Rubasinghege and Grassian (2009), which would compete with further growth causing a net leveling off of product formation.
While specific atmospheric implications require a better speciation of products, the production of such species can be discussed in the context of multiple recent field studies.In SHARP 2009, field measurements indicate that there was a photolytic source of HONO within 20 m of the ground (Wong et al., 2012(Wong et al., , 2013)).Studies done in other urban centers such as London (Lee et al., 2016) and Los Angeles (Young et al., 2012) also suggest there is an unknown photochemical HONO source.Many suggest that this source is correlated to NO 2 ; however, in a study carried out in Bakersfield and Pasadena, the HONO source does not correlate with NO 2 (Pusede et al., 2015).As discussed by those authors, the formation of HNO 3 and its subsequent incorporation into aerosol as ammonium nitrate can extend the lifetime of airborne nitrate, causing the nitrate which is deposited to not correlate temporally with NO 2(g) .Grime would likely have a similar delayed response; in addition, the RH dependence of the grime photochemistry could serve as a further mechanism for an offset in NO 2 values and HONO production, due to the cycling of RH conditions in the atmosphere and, therefore, the cycling of this source strength.However, more quantification and speciation is required to evaluate the importance of such a grime source.

Conclusions
Urban grime was collected onto glass substrates without modification and illuminated.Grime photochemistry produced nitrogen oxides in the form of NO 2 and/or HONO.Such chemistry is not currently included in urban air models, but could impact NO x and/or HONO levels in these centers.The production of these species is dependant on RH, again highlighting the need to consider water content when studying environmental surfaces.

Figure 3 :
Figure 3: HONO production as a function of relative humidity.Values are normalized to the steady state concentration of HONO formed during an initial illumination period at a relative humidity of 35%.The average of at least 3 measurements on different samples is shown; error bars represent 1 standard deviation.

Figure 4 :Figure 3 .
Figure 4: Average ratio of water mass to total ion mass within grime as a function of relative humidity.Water uptake onto clean crystals was subtracted from the grime uptake curves and thus only grime-mediated uptake is shown here.The shaded region indicates a 95% confidence interval base on 16 measurements of different samples.

16 533Figure 3 :Figure 4 :Figure 4 .
Figure 3: HONO production as a function of relative humidity.Values are normalized to the steady state 534