Exploring the atmospheric chemistry of nitrous acid ( HONO ) at a rural site in Southern China

We performed measurements of nitrous acid (HONO) during the PRIDE-PRD2006 campaign in the Pearl River Delta region 60 km north of Guangzhou, China, for 4 weeks in June 2006. HONO was measured by a LOPAP in-situ instrument which was setup in one of the campaign supersites along with a variety of instruments measuring hydroxyl radicals, trace gases, aerosols, and meteorological parameters. Maximum diurnal HONO mixing ratios of 1–5 ppb were observed during the nights. We found that the nighttime build-up of HONO can be attributed to the heterogeneous NO 2 to HONO conversion on ground surfaces and the OH + NO reaction. In addition to elevated nighttime mixing ratios, measured noontime values of ≈200 ppt indicate the existence of a daytime source higher than the OH + NO→HONO reaction. Using the simultaneously recorded OH, NO, and HONO photolysis frequency, a daytime additional source strength of HONO ( PM ) was calculated to be 0.77 ppb h −1 on average. This value compares well to previous measurements in other environments. Our analysis ofPM provides evidence that the photolysis of HNO3 adsorbed on ground surfaces contributes to the HONO formation.


Introduction
Nitrous acid (HONO) is an important trace gas in the atmosphere due to its contribution to the cycles of nitrogen oxides (NO x ) and hydrogen oxides (HO x ) (Finlayson-Pitts and Pitts, 2000).The photolysis of HONO in the near ultraviolet spectral region (320 nm<λ<400 nm): provides a daytime source of hydroxyl radicals (OH) and nitrogen oxides (NO) independently from ozone photolysis and NO 2 concentrations.This feature is even more important after HONO accumulates under dark or low light conditions.Observations of HONO in the troposphere were performed at many locations over the last 30 yr.These measurements revealed concentrations from several ppts in remote and clean areas (e.g.Zhou et al., 2002Zhou et al., , 2007;;Beine et al., 2006;Ren et al., 2010) up to several ppbs in more polluted environments (e.g.Febo et al., 1996;Stutz et al., 2002;Kanaya et al., 2007;Su et al., 2008a;Yu et al., 2009).Due to photolysis, HONO exhibits a diurnal variation with minima during the day.The highest reported values occurred during nighttime in heavily polluted megacities like Santiago, Chile (Elshorbany et al., 2009), and Guangzhou, China (Qin et al., 2009).The level Fig. 1.Diagram of HONO budget in the troposphere.Boxes in light blue refers to gas phase reactions, while grey boxes refer to processes with various surfaces involved.Reactions in the yellow frame need solar radiation and thus only take place in daytime.When significant amount of OH radicals and NO exist, reactions in the light orange frame will also happen in night.
of HONO is related to the NO x levels during all of these field experiments, however this relation is highly variable as indicated by the HONO to NO x ratios determined in the above experiments, ranging from way below 1 % to well above 10 %.
Despite the importance of HONO in atmospheric chemistry, the HONO budget in the troposphere is not well understood since both emissions, homogeneous and heterogeneous processes influence the gas phase concentration of HONO. Figure 1 depicts the different pathways of HONO formation and loss.Direct emissions of HONO have been accessed through tunnel experiments and car exhaust analysis; the emission ratios of HONO/NO x are found to be 0.3-0.8% (Kurtenbach et al., 2001, and references therein).These values are generally much lower than that from the field observations, indicating that the atmospheric HONO is mostly secondarily formed.
The reverse reaction of Reaction (R1) is the most important homogeneous reaction forming HONO.
This reaction plays a role mainly during daytime when OH and NO are high (e.g.Alicke et al., 2003).However, in regions with significant amount of nighttime OH, e.g., Pearl River Delta in China (Lu et al., 2011), the contribution of Reaction (R2) to the nocturnal HONO formation might not be neglected.Li et al. (2008) found that the homogeneous reaction of photolytically excited NO 2 with H 2 O can form OH and HONO.
NO 2 +hν(λ > 420 nm) But its reaction rate and yield are still under discussion (Carr et al., 2009) and this mechanism does not significantly contributes to the atmospheric HONO formation (Sörgel et al., 2011a;Wong et al., 2011;Amedro et al., 2011).Laboratory studies found that the photolysis of ortho-nitrophenols also leads to homogeneous HONO formation (Bejan et al., 2006).
Although the estimated HONO formation rate of 100 ppt h −1 for 1 ppb nitrophenols at overhead sun could be a significant fraction of the total HONO formation rate during daytime, the relevance of this mechanism needs to be validated through field measurements.Based on theoretical calculations, Zhang and Tao (2010) proposed a gas phase HONO formation pathway through the homogeneous nucleation of NO 2 , H 2 O and NH 3 .
Compared to the homogeneous HONO formation, heterogeneous formation of HONO on various surfaces is thought to be the most likely explanation of the observed high HONO concentrations.Laboratory studies revealed that NO 2 can be converted to HONO on wet surfaces.
However, the source strength of Reaction (R5) is not exactly known and depends on various parameters including the gas phase NO 2 concentration, the surface water content, and the surface area density (Finlayson-Pitts et al., 2003).Laboratory studies also found a heterogeneous reduction of NO 2 forming HONO on soot particles or surfaces containing organic substrates (Ammann et al., 1998(Ammann et al., , 2005;;Kleffmann et al., 1999;Bröske et al., 2003;Aubin and Abbatt, 2007).
The proposed mechanism of this redox HONO formation is Reaction (R6) can be relevant for HONO formation under conditions of high NO 2 and soot loadings, i.e. in plumes (Kleffmann et al., 1999).Following Akimoto et al. (1987) who reported that Reaction (R6) was photo-enhanced, a photosensitized NO 2 reduction on organic surfaces has been observed recently in many studies as an effective pathway producing HONO (George et al., 2005;Stemmler et al., 2006).The production rate of HONO is a complex function of the actinic flux, the surface area density and gas phase NO 2 concentration (Stemmler et al., 2007).At low NO 2 , the HONO production is linearly increasing with NO 2 ; whilst the increase levels off when the NO 2 concentration reaches a certain point.The same phenomenon is also found for the actinic flux.The rate of the photosensitized reaction was 1-2 orders of magnitude more efficient than the dark reaction (George et al., 2005).
Both laboratory studies and field observations found that HONO can also be heterogeneously formed through the photolysis of adsorbed nitric acid (HNO 3 ) and nitrate (NO − 3 ) at UV wavelengths (Zhou et al., 2002(Zhou et al., , 2003)): It is still an open question why the HNO 3 -to-HONO conversion frequency necessary to explain the observed HONO production is orders of magnitude higher than the photolysis of HNO 3 in the gas and liquid phase (Zhou et al., 2003;Rohrer et al., 2005).However, a new field study by Zhou et al. (2011) provided evidence for a daytime source of nitrous acid caused by the photolysis of nitric acid on forest canopies.
Recently, Su et al. (2011) illustrated that the soil nitrite, which is formed through the biological nitrification and denitrification processes, can be a prominent HONO source.The HONO formation is via the following reversible acid-base reaction and partitioning between air and the aqueous phase of humid soil: Its HONO production rate depends on the nitrite concentration, the pH value, the water content and the temperature of the soil.Values from 2 to 5 ppb h −1 were estimated for the field measurements in Xinken in Southern China, which can explain most HONO productions besides Reaction (R2) during daytime.
Compared to HONO formation, the loss processes of HONO are well understood.Besides Reaction (R1), HONO can be removed by reaction with OH.
Reactions (R1) and (R9) are well established and they determine the HONO lifetime during daytime.Under typical daytime conditions the HONO lifetime is in the order of 20 min.Other gas phase reactions of HONO, e.g.self reaction or reaction with HNO 3 , are not important under tropospheric conditions.Heterogeneous loss of HONO on various surfaces is too slow to be relevant in most cases (Stutz et al., 2002, and references therein).During night, the ma-jor HONO removal pathway is the dry deposition on ground surfaces (Trick, 2004), i.e.
where ν ground HONO is the deposition velocity of HONO and H the height of the well-mixed layer.The reported value of ν ground HONO ranges from 0.077 to 3 cm s −1 (Harrison and Kitto, 1994;Harrison et al., 1996;Spindler et al., 1999;Stutz et al., 2002).
The question of HONO budget is two-fold, a day and a night problem.Due to the absence of solar radiation and low concentration of OH radicals during night, it is well accepted that the nocturnal HONO is mainly formed by the heterogeneous conversion of NO 2 .Regardless the detailed chemical mechanisms, the strength of the HONO formation from NO 2 can be expressed as the HONO conversion frequency C HONO (Alicke et al., 2002(Alicke et al., , 2003)), where [NO 2 ] is the average NO 2 mixing ratio in the time interval of t 2 − t 1 .In order to correct the effects of source emission and diffusion, complex scaling methods has been tested by Su et al. (2008a) during the calculation of C HONO .
They found that values of C HONO derived from using different scaling species or without applying scaling method do not show a big difference.During daytime, recent field observations find that the HONO concentration is too high to be explained by the gas phase Reactions (R1), (R2) and (R9).Therefore, the daytime HONO budget was expanded by an additional production rate term P M (Su et al., 2008b).
Here we omitted the influence of direct emission, deposition, horizontal and vertical transport on the HONO concentration since their influence is relatively small compared to the gas phase reactions and can be neglected in Eq. (3).Under quasi stationary conditions, i.e. d[HONO]/dt ≈ 0, the additional production rate P M can be expressed by which also corresponds to an additional HONO concentration [HONO] M calculated from from the difference of the measured HONO and the budget without P M : which is equivalent to For the first time Kleffmann et al. (2005) had a full set of OH, NO, and photolysis frequency measurements simultaneously with HONO so they could directly derive P M for their measurements in a mixed deciduous forest in Germany.Their value of 0.55 ppb h −1 exceeds the possible source by Reaction (R5) therefore indicating an additional source of HONO.Similar findings, using direct measurements of HONO, OH, NO, and J HONO , were obtained by Acker et al. (2006b) and Wong et al. (2012).
With the fast economic growth, the Pearl River Delta (PRD) region in Southern China has been identified as one of the hot-spots of NO 2 by satellite measurements (Richter et al., 2005).The first continuous HONO measurements in the PRD region was conducted by Su et al. (2008a,b) at a non-urban site in XinKen (22.61 • N, 113.61 • E).Besides the high nighttime HONO concentration (0.4-3.5 ppb) a surprisingly high daytime average of 1.2 ppb was observed.However, due to the lack of measured key parameters (i.e.photolysis frequencies and OH) and the low time resolution of their HONO measurement, the investigation of the daytime HONO chemistry by Su et al. (2008b) left some open questions.Here we present the HONO data recorded during the PRIDE-PRD2006 campaign.We performed 4 weeks of continuous HONO measurements at a rural site in PRD region by the LOPAP technique side-by-side to measurements of trace gases, HO x radicals, aerosol parameters and meteorological conditions.The contribution of HONO to the HO x production has been investigated and discussed by Lu et al. (2011).The relative contribution was found to be 8-28 % during the campaign.In this study, we are focusing on the analysis of the HONO budget.The simultaneous measurements that are used for the identification and quantification of HONO sources and sinks are summarized in Table 1.

LOPAP instrument
HONO was measured by LOPAP (long path absorption photometer) technique which was pioneered by Kleffmann and co-workers (Heland et al., 2001;Kleffmann et al., 2006).The instrument used in this study was a commercial version of the original instrument (QUMA GmbH, Wuppertal).Since the LOPAP instrument has been described in detail previously we give only a short outline.Ambient air is sampled into a sampling unit consisting of two stripping coils in series.Stripping solution of 100 g sulfanilamide and 1 l HCl (37 % volume fraction) in 9 l MilliQ water is pumped through the coils.HONO and any interfering species in the sampled air will dissolve in the liquid phase forming NO − 2 followed by a rapid reaction of NO − 2 forming a stable diazonium salt.Almost all HONO and a small fraction of interfering species are taken out by the first coil, while the second coil takes out the same amount of interfering species but has no HONO in it.After debubbling the remaining air, the formed diazonium salt solution of each coil is pumped from the sampling unit to the detection unit through a 2.5 m insolated Teflon tubing.Each solution is then reacted with a dye solution made by dissolving 1.6 g n-(1-naphthyl)-ethylenediaminedihydrochloride in 8 l MilliQ water forming an azo dye.The concentration of the azo dye is monitored by a 2 channel optical absorption spectrometer which absorption cell consists of a 1.2 m long Teflon tubing (called Liquid Core Waveguide, LCW).Visible light from a white LED lamp (Lumileds, Typ: LUXEON V star white, LXHLLW6C) is focused into the tubing via fiber optics.At the opposite end of the LCW, the light is collected by a fiber and detected by the spectrometer.The absorption spectra are recorded on a computer for later data evaluation.All the processes above are continuously operated.
The sampled air flow and the liquid flow of the stripping solution was set to 1 l min −1 and 0.28 ml min −1 , respectively, and both were calibrated every third day during the campaign.Zero measurements by sampling zero air were done manually every 12 h.With above settings, the sampling efficiency of HONO in the sampling unit was determined to be 99.93 %.The time resolution, detection limit and accuracy of the measurement was 5 min, 7 ppt and 10 %, respectively.

Measurement site and campaign setup
The PRIDE-PRD2006 intensive campaign took place from 3 to 31 July 2006 in the Pearl River Delta (PRD) area in Southern China.Our measurements were conducted at the Back Garden (BG) supersite (23.50 • N, 113.03 • E) which is about 60 km northwest of Guangzhou.The BG site is located near a water reservoir and is surrounded by farmland (peanuts, lychees, trees, small forests).It experienced little local emissions from traffic, but road and building construction as well as biomass and cable burning were occasionally observed within the surrounding area.The campaign was within the framework of the Program of Regional Integrated Experiments of Air Quality over the Pearl River Delta (PRIDE-PRD2006, Zhang et al., 2012).This program targets on understanding the formation mechanism of O 3 and fine particulate matter within the densely populated PRD area.
The BG supersite was equipped with a comprehensive suite of atmospheric measurements (e.g.Garland et al., 2008;Hofzumahaus et al., 2009;Li et al., 2010;Lou et al., 2010;Lu et al., 2011).On-site measurements used in this study are reported in Table 1 with their time resolutions and positions relative to the LOPAP.The sampling unit of the LOPAP was fixed on the top of the FZJ lab container housing the HO x (Lu et al., 2011) and OH reactivity (Lou et al., 2010) instruments.The container was stacked on top of a second container hence the sampling height was 6 m above the ground.The detection unit (i.e.pumps, spectrometers etc.) was placed inside the lab container which was held at constant temperature.OH concentrations were measured by laser induced fluorescence (LIF) spectroscopy (Hofzumahaus et al., 2009;Lu et al., 2011).Ambient air was sampled by gas expansion into a low-pressure (3.5 hPa) fluorescence cell, where the OH radicals were electronically excited by tunable, pulsed UV laser radiation at a wavelength of 308 nm.The resulting OH resonance fluorescence was detected by time-delayed gated photon counting.The limit of detection was (0.5-1) × 10 6 cm −3 at 5 min integration time and the accuracy was 20 % (1σ ).The laser of the LIF was also used for the OH reactivity instrument (Lou et al., 2010) which was setup in the same container.This instrument provided OH reactivities in the range between 1 s −1 and 100 s −1 at 1 min time resolution.
NO and NO 2 were measured by commercial instruments (Takegawa et al., 2006).NO was detected by NO-O 3 chemiluminescence (Thermo Electron, Model 42CTL), while NO 2 was first converted to NO in a photolytical reactor (Droplet Measurement Technologies, Model BLC).The 1 min detection limits for NO and NO 2 were 50 ppt and 170 ppt, respectively, and the corresponding accuracies were 7 % and 13 %.However, during the campaign the instruments were stopped occasionally for repair and maintenance.The gaps (less than 10 % of the data set) were filled with data of a second instrument operated by PKU at the BG site.
The aerosol surface density S a was calculated from the particle number size distributions between 3 nm and 10 µm by assuming the particles are in spherical shape (Yue et al., 2010).The particle number size distributions were measured by a system consisting of a Twin Differential Mobility Particle Sizer (TDMPS) and an APS (TSI model 3321).The relative humidity within the whole system was kept below 30 % by silica gel tubes within the inlet line and both sheath air cycles.The accuracy of derived S a was estimated to be around 20 %.A hygroscopic factor f (RH) was applied to correct S a to the aerosol surface density in the real atmosphere (S aw ).
The empirical factors a and b used to estimate f (RH) were set to 2.06 and 3.6, respectively, as described by Liu et al. (2008).Total aerosol particle scattering coefficients (σ s,λ ) and hemispheric backscattering coefficients (σ bs,λ ) at three different wavelengths (λ = 450 nm, 550 nm, and 700 nm) were measured by an integrating nephelometer (Model 3563, TSI).The aerosol particle absorption coefficient at 532 nm (σ a,532 ) was determined with a photoacoustic spectrometer (PAS; Desert Research Institute), which provides highly sensitive absorption measurements without interference by scattering signals.These two instruments shared the same PM 10 inlet with the TDMPS-APS system, and thus provided data at ≈30 % relative humidity.The accuracy of the nephelometer and the PAS data was 10 %.The aerosol single scattering albedo at 532 nm (ω 532 ) was calculated as the ratio of the interpolated σ s,532 and measured σ a,532 (Garland et al., 2008).
Gaseous nitric acid, HNO 3 , was measured by a GAC-IC system (Su et al., 2008b).The system consists of a gas and aerosol collector (GAC) and an ion chromatograph detector (IC).The GAC uses an annular wet denuder to collect HF, HCl, SO 2 , HONO, HNO 3 in gas phase and F − , Cl − , SO 2− 4 , NO − 2 , NO − 3 in PM 2.5 .The ambient air was sampled into the denuder through a 6 m Teflon tube with an air flow rate of 16.7 l min −1 .The stripping solution used in the denuder was 25 µmol l −1 Na 2 CO 3 .The time resolution for all species was 30 min, while the detection limits for HNO 3 and aerosol nitrate were 65 ppt and 0.4 µg m −3 , respectively.
Photolysis frequencies were calculated from solar actinicflux spectra recorded by a spectroradiometer (Meteorologie

X. Li et al.: HONO atmospheric chemistry in Southern China
Consult) as described by Bohn et al. (2008).The accuracy of the HONO photolysis frequency, J HONO , was estimated to be 10 % at solar zenith angles smaller than 80 • .
Wind speed and temperature were measured by an ultrasonic anemometer (USA-1, Metek, Germany) on the top of the lab container.A weather station (WXT510, Vaisala, Finland) was located on a small elevation behind the container (20 m distance).The accuracies for temperature, relative humidity, and pressure were 0.3 K, 3 %, and 0.5 hPa, respectively.

Results
We performed nearly continuous HONO measurements from 3 to 26 July 2006 (Fig. 2).HONO exhibits a clear diurnal cycle with peaks in the early morning and minima around noontime.The nighttime maxima ranged from 1 ppb to 5 ppb.Especially during two days (24 and 25 July) when intense electric cable combustion events happened, the nocturnal HONO concentration exceeded 4 ppb in the early morning hours.During an extended rainfall period (15 to 18 July, influence of typhoon Bilis) the concentration of HONO as well as other trace species were at their lowest level of the campaign.When we average the entire data set the mean maximum HONO concentration was 1.5 ppb, occurring around sunrise (Fig. 3b).The average HONO mixing ratio at noon time was about 200 ppt, however with large variations between 10 ppt and 490 ppt (Fig. 3b).
The entire campaign was characterized by tropical conditions with high temperatures (28-36 • C) and high humidities between 60 % and 95 % relative humidity (Fig. 2).The rainfall period (Typhoon Bilis) is clearly seen in the relative humidity record with constant values above 90 % throughout the day.It is also represented in the solar radiation which is strongly reduced as shown by the photolysis frequency of HONO (J HONO ) and NO 2 (J NO 2 ).At the field site, local wind speeds were generally slow and had values mostly less than 2 m s −1 measured at 8 m height directly above the HONO inlet.These conditions correspond to the typical summer weather condition in the PRD region.The general weather pattern consists of high temperatures, high humidities, and dense cloud cover as reported by Chan and Yao (2008).The prevailing geostrophic winds are from the southeast, south and southwest due to the monsoon circulation.The typical ground wind speed at inland locations is around 1.5 m s −1 , a situation which favors the accumulation of air pollutants.
Figure 2 also presents the time series of simultaneously measured trace gases and aerosol parameters.Noontime OH concentrations reach (1-2) × 10 7 cm −3 on every day when OH data were reported.However, the LIF instrument could not be operated during the rainfall period due to technical problems.The measured OH reactivity (k OH ) record indicates the high level of reactants present.Combining OH concentrations and reactivities show the enormous turnover under these conditions (Hofzumahaus et al., 2009).At the same time the levels of NO and NO 2 show a large variability.Maximum mixing ratios of NO, NO 2 , and NO x were 20 ppb, 30 ppb, and 50 ppb, occurring in the early morning when all ozone has been titrated by NO emissions.The gas phase HNO 3 concentrations are in the range of 100 ppt to 3 ppb with an average of 700 ppt, which are comparable with those observed at other rural sites (e.g.Zhou et al., 2007).Similar variations can be found between HNO 3 and HONO.
In contrast to the gas phase parameters shown here, the aerosols exhibit a less pronounced diurnal variation, as demonstrated by the aerosol surface density (S a and S aw ) and aerosol light absorption (σ a,532 ) in Fig. 2. The major variation comes from the different pollution levels, namely the burning events at the end of the campaign.The elevated values strongly reduced the visibility to values in the order of 1-2 km.The aerosol surface density and aerosol light absorption show a very similar pattern.Therefore, the aerosol light absorption can fill gaps in the aerosol surface density record, i.e. in the rainfall period.

HONO concentrations and the influence of direct HONO emissions
The observed HONO concentrations with maximum values above 5 ppb at night and of 200 ppt at overhead sun (Fig. 2) are among the highest ever recorded at a site outside of big cities (cf.Alicke et al., 2003;Kleffmann et al., 2005;Cheng et al., 2007;Shon et al., 2007;Zhou et al., 2007).However, when we refer to campaign averages and relate them to measurements which were done in or in the vicinity of big cities (Table 2) we see that the BG data are at the low end for nighttime and almost average for daytime observations.Our nighttime values compare to Beijing and Tokyo values, while daytime mixing ratios are lower than other observations in Asia.In order to relate the HONO concentration to the level of local pollution and the possible HONO sources, the HONO to NO 2 and/or the HONO to NO x ratio has been used in many publications to create a more stable value for charactering the level of HONO.
The ratio HONO/NO 2 has the advantage of being less influenced by convection or transport processes than individual concentrations and it can be used to indicate the heterogeneous conversion efficiency of NO 2 to HONO.However, when a large fraction of HONO comes from direct emissions the ratio might be influenced.Emission ratios of HONO to NO 2 were measured to be less than 1 % as reported by Kurtenbach et al. (2001, and references therein).
Figure 3c displays the mean diurnal variation of HONO/NO 2 at the BG site.In general, the highest ratios with large variability were found from 11:00 to 17:00 LT, even higher than those during the nights.This can be attributed to the efficient daytime formation of HONO which is not correlated to NO 2 (see Fig. 9 and the discussion about the additional HONO source below).On average, HONO/NO 2 started to increase shortly after sunset and kept stable at ≈6 % in the rest of the night.The increase of HONO/NO 2 was accompanied by the simultaneous increase of HONO and NO 2 , suggesting the heterogeneous conversion from NO 2 to HONO.Once the conversion reached a steady state, the HONO/NO 2 became stable.The covariation between HONO and NO 2 during nighttime (19:00-06:00 LT) can further be seen from the correlation plot (Fig. 4b), indicating NO 2 as the precursor of HONO in this time interval.Figure 4b also shows the dependence of the HONO/NO 2 ratio on the aerosol surface density (S aw ) during night.Higher values of HONO/NO 2 are usually found at higher levels of S aw , suggesting the possible influence of aerosol surfaces on the heterogeneous NO 2 to HONO conversion.However, as will be discussed later, compared to the ground surfaces, the influence of aerosol surfaces on the HONO formation is small.The nighttime stable HONO/NO 2 values were slightly higher than that observed by Su et al. (2008a) (4.3 ± 1.5 %) in the same region.The decrease of HONO/NO 2 can be caused by emissions or transport.The increase of HONO/NO 2 and its high value during daytime suggests a high HONO formation efficiency, which however might not be solely explained by the NO 2 conversion given the weak correlation between HONO and NO 2 (Fig. 4a).Moreover, as higher HONO/NO 2 ratios are usually found at lower RH level (Fig. 4a), the daytime HONO formation is unlikely caused by Reaction (R5).As will be discussed later, additional HONO production mechanisms are needed to sustain the high daytime level of HONO as well as HONO/NO 2 .
As the BG site was characterized by occasional local emissions from biomass and electric cable burning processes, it is necessary to separate HONO formed by chemical reactions from directly emitted.Emission factors of HONO are only available for vehicles obtained through either tun- nel experiments or laboratory studies in the US and Europe (Kurtenbach et al., 2001, and references therein).Considering the differences in combustion processes, fuel compositions, etc., the reported emission of HONO relative to NO x (i.e.HONO/NO x ) might not be representative for PRD region.In order to evaluate the influence of direct emissions, we tried to estimate the local HONO emission factor from the ambient measurements.
When the air mass is getting aged, due to the conversion from NO 2 to HONO, the HONO/NO x ratio will increase; the freshly emitted air mass should have the lowest HONO/NO x .Thus, we used the observed lowest HONO/NO x as the upper limit for the emission factor.Besides, two other filters as proposed by Su et al. (2008a) were applied: (1) only nighttime data were included in order to avoid the fast photolysis of HONO; (2) NO ≥ 20 ppb was used to catch the fresh emissions.The derived HONO/NO x is 1.8 % much higher than 0.8 % reported by Kurtenbach et al. (2001).During the experiments conducted by Kurtenbach et al. (2001), air masses were characterized by NO/NO x ≥95 %.The typical NO/NO x ratio in our measurements is ≈50 %, suggesting the air detected is a mixture of freshly emitted and aged.Therefore, a substantial part of HONO is secondary.Since the observed NO/NO x and HONO/NO x during the entire campaign period has an average of 13 % and 6 %, respectively, the influence of direct emissions on HONO should be small and hence be ignored in the following analysis.

HONO during night
The nocturnal pattern observed here is similar to ever reported: HONO increases after sunset to a maximum at sunrise.Since most of the dynamic range of the HONO concentration is during the night, the correlation between HONO and NO 2 is mostly created during nighttime hours.In this section, we will focus on the formation pathways of the nocturnal HONO and explore the relevant factors.

Gas phase formation through OH + NO
During the PRIDE-PRD2006 campaign, a substantial measured OH concentration of ≈1×10 6 cm −3 together with NO mixing ratios larger than 2 ppb were found in the night (Fig. 5b, c), which brings up the importance of OH + NO on the nighttime HONO formation.Taking Reactions (R2) and (R9) into account, the net HONO production is The mean nocturnal variation of P net OH + NO is illustrated in Fig. 5a.Since k OH + NO and k OH + HONO are similar (7.2 and 5.0 in units of 10 −12 cm 3 s −1 at 298 K), the P net OH+NO depends on the relative abundance of NO and HONO.Before midnight, due to the low NO level, small negative value of P net OH+NO was found.The formation of HONO in this time period might mostly come from the NO 2 conversion as will be discussed later.After midnight, with the sharp increase of NO mixing ratios, the P net OH+NO increased from 0 to 0.28 ppb h −1 .Higher P net OH+NO of 0.4-0.9ppb h −1 were found by Wong et al. (2011) for the nocturnal urban atmosphere of Houston.By integrating P net OH+NO from 00:00 to 06:00 LT, an accumulated HONO formation of 0.66 ppb can be derived.However, the averaged measured HONO accumulation in this time period was only 0.36 ppb.This suggested that, (1) the reaction of OH + NO is enough to explain the increase of HONO in the second half of the night, even though sources like NO 2 conversion might still existed; (2) the strength of HONO sinks other than OH + HONO should be at least in the order of 0.05 ppb h −1 (i.e.(0.66 ppb−0.36ppb)/6 h).

Heterogeneous conversion of NO 2
Figure 6 illustrates the correlation between HONO, HONO/NO 2 , RH corrected aerosol surface density from 3 nm to 10 µm (S aw ) and aerosol light absorption at 532 nm (σ a,532 ).As the HONO production by OH + NO was prominent in the second half of the night, we restrict the correlation analysis to the time frame 18:00-0:00 LT.Both HONO and HONO/NO 2 increase with S a and σ a,532 .Compared with HONO/NO 2 , HONO has better correlation with S aw and σ a,532 , which was probably due to the decrease of boundary layer height leading to the simultaneous concentration change of trace gases and aerosols.Whilst this effect can largely be canceled out in the HONO/NO 2 .The positive correlation between HONO/NO 2 and S aw indicates the HONO formation on aerosol surfaces through the heterogeneous conversion of NO 2 , which has already been found in Fig. 4b.However, considering the correlation is weak (r 2 = 0.184), the relative amount of HONO formed on aerosol surfaces might be small.The σ a,532 provides the information of black carbon or soot content in the atmosphere.The even weaker correlation between HONO/NO 2 and σ a,532 indicates a small Fig. 6.Correlations between HONO/NO 2 , HONO, aerosol surface density (RH corrected), aerosol light absorption during the time interval of 18:00-00:00 LT.Fig. 7. Scatter plot of HONO/NO 2 against RH in the time interval of 18:00-00:00 LT (time interval avoids influence of OH+NO, see Fig. 5).Triangles are the averaged top-5 HONO/NO 2 in each 5 % RH interval.role of soot in the heterogeneous HONO formation.The relationship between HONO/NO 2 and relative humidity RH is illustrated in Fig. 7. Following the method introduced by Stutz et al. (2004), we analyzed the top-5 HONO/NO 2 values in each 5 % RH interval.These are less influenced by factors like the time of the night, the surface density, advection, etc.The dependence of HONO/NO 2 on RH can be divided into two groups.An increase of HONO/NO 2 along with RH was found when RH was less than 70 %, which is consistent with the kinetics of Reaction (R5).Once RH was higher than 70 %, the NO 2 to HONO conversion efficiency depended negatively on RH.The same phenomenon was also observed by Yu et al. (2009) in Kathmandu.This finding can be associated with the number of water layers formed on various surfaces increases rapidly with relative humidity, leading to an efficient uptake of HONO.Thus HONO is lost in the gas phase when on the surface loss becomes dominant (Lammel, 1999;Stutz et al., 2004).
Using the data from the first half of the night (18:00-00:00 LT), we calculated the HONO conversion frequency C HONO by Eq. ( 2).The average value of C HONO during the campaign was 0.024 ± 0.015 h −1 , corresponding to a HONO production rate by NO 2 (P NO 2 ) of 0.38 ± 0.25 ppb h −1 (i.e.C HONO × [NO 2 ]).It is necessary to note that, (1) the C HONO assumes all the measured HONO is coming from the conversion of NO 2 ; and (2) the derived P NO 2 is the net HONO production, which means sinks of HONO (e.g.deposition, HONO + OH, etc.) have already been taken into account in P NO 2 .The HONO conversion frequency obtained in this work is comparable to observations of Su et al. (2008b) in the same region (0.016 ± 0.014 h −1 ) and by Alicke et al. (2003) at a rural site in Pabstthum (0.018 ± 0.009 h −1 ), but much higher than observation at other places (e.g.Lammel, 1999;Alicke et al., 2002;Acker et al., 2005;Sörgel et al., 2011a,b).
The formation of HONO through heterogeneous NO 2 conversion on aerosol surfaces (S aw ) depends on many factors and can be approximated by where γ NO 2 →HONO is the uptake coefficient of NO 2 to HONO, and ν NO 2 the mean molecular velocity of NO 2 (Ammann et al., 1998).If the entire HONO formation were taking place on aerosol surfaces of with density S aw , γ NO 2 →HONO calculated from Eqs. ( 2) and ( 10) would have an average of (2.8 ± 1.5) × 10 −5 .Using the measured single scattering albedo at 532 nm ω 532 , and considering the soot particles are usually hydrophobic, the soot surface density can be roughly estimated as S soot = (1 − ω 532 ) × S a .Further assuming all HONO was formed on soot surfaces, the average value of γ soot NO 2 →HONO was found to be (3.0±1.8)×10−4 .As discussed by Kleffmann et al. (1998), this value seems too high for ambient aerosols.Especially for soot particles, along with the air mass is getting aged, the γ soot NO 2 →HONO will decrease dramatically to a value which is in the order of 10 −5 (Aubin and Abbatt, 2007).Moreover, given the weak correlation of NO 2 to HONO conversion efficiency against S aw and σ a,532 , we conclude that the HONO production was mostly taking place at ground surfaces rather than aerosols during nighttime.These ground surfaces could include the surfaces of plants, buildings, soils and so on in surrounding areas which can typically provide orders of magnitude higher surface density than aerosols (Sörgel et al., 2011b).A similar conclusion was drawn from recent studies in both forest and urban areas (Sörgel et al., 2011b;Wong et al., 2011).

Removal of HONO through deposition
In Sect.4.2.1, we mentioned that the production of HONO estimated from gas phase reactions (i.e.Reactions (R2) and (R9)) was larger than that from the measurement.Moreover, as the heterogeneous NO 2 conversion can also take place in the second half of the night, this discrepancy will become even larger.Assuming a constant C HONO over night and the extra HONO are all removed through deposition on the ground, the HONO deposition velocity ν ground HONO can be roughly estimated by, The ν ground HONO calculated from the mean nocturnal variation (00:00-06:00 LT in Fig. 5) is ≈0.8 cm s −1 taken the night time H of 100 m.This value is substantially higher than observations by Stutz et al. (2002) (0.077 cm s −1 ) but within the range of other published results (0.5-3 cm s −1 ) (Harrison and Kitto, 1994;Harrison et al., 1996;Spindler et al., 1999).The estimated high value of ν ground HONO seems to be in line with the high RH during night which increases the water adsorbed at ground surfaces and favor the adsorption of HONO.

Formation of HONO on ground surfaces
As illustrated above, the nighttime HONO production was mainly from the heterogeneous NO 2 conversion on ground surfaces, which can be described as, where ν ground NO 2 and φ NO 2 →HONO is the dry deposition velocity of NO 2 and the yield of HONO from deposited NO 2 , respectively (Su et al., 2008a).By definition, φ NO 2 →HONO varies between 0 and 1, accounting for the fact that the deposited NO 2 molecule is not necessarily converted to HONO.As the controlling resistances were the aerodynamic resistance and the quasi-laminar layer resistance during night, similar deposition velocities were expected for NO 2 and HONO.Thus, using the measured average value of C HONO , NO 2 , HONO and the estimated ν ground HONO , and assuming a 100 m height of the well-mixed layer, a φ NO 2 →HONO value of 0.15 was found which means that one out of seven deposited NO 2 forms HONO being released into the atmosphere.Our derived φ NO 2 →HONO is lower than that by Su et al. (2008a) (0.34), which is mainly caused by using the different value of ν ground NO 2 during the calculation.

Daytime HONO budget
The HONO data presented here were accompanied by simultaneous measurements of OH, NO, and J HONO .Therefore, we can examine the HONO budget and HONO lifetime directly from measured quantities.The HONO chemical lifetime (τ HONO ) due to photolysis (Reaction R1) and reaction with OH (Reaction R9) was ≈50 min after sunrise and before sunset and decreased to ≈13 min under overhead sun conditions.Due to the short lifetime, especially around noon, HONO is in quasi steady state and we can calculate the budget of HONO from gas phase production and loss in the same way as Kleffmann et al. (2005) and Acker et al. (2006b) did for two rural sites in Germany.In order to justify the quasi steady state assumptions associated with Eqs. ( 4) and ( 5), we restricted the data set to J HONO > 8 × 10 −4 s −1 .Under these conditions, the HONO lifetime was less than 20 min, and d[HONO]/dt calculated from the measurements was small and evenly distributed around zero.
The P M and HONO M (see Eqs. 4, 5, and 6) for the "high J subset" are shown in Fig. 8 (blue dots).The value of P M was significantly higher than zero in most of the time, which corresponds to a substantial amount of HONO M .Also shown in Fig. 8 are the mean diurnal variations of P M , HONO M and measured HONO (open circles).The average P M during the entire campaign period was 0.77±0.55ppb h −1 , which is comparable to observations by Kleffmann et al. (2005) in the forest in Jülich (≈0.55 ppb h −1 ) and by Wong et al. (2012) in downtown Houston (≈0.6 ppb h −1 ), but smaller than the estimation by Su et al. (2008b) in XinKen (≈2 ppb h −1 ).Since Fig. 8. Upper frame: Calculated additional HONO production rate P M (Eq.4).The blue dots correspond all 5-min averages where measurements of J HONO , OH, NO and HONO were present.The red circles refer to the mean over 1 h with 1-σ error bars.Lower: additional HONO concentration HONO M (Eq.5).Red circles are the mean values over 1 h, black triangles are the mean value of the measured HONO.
the daytime average minima of HONO in XinKen was as high as 1.2 ppb, higher value of P M can be expected.P M in the order of 0.1-2 ppb h −1 are also found in other rural and urban areas (Elshorbany et al., 2009;Ren et al., 2010;Sörgel et al., 2011a).The HONO M was the dominant part of measured HONO, accounting for >95 % since 11:00 LT.This means the daytime HONO formation was controlled by P M while only a minor part (<5 %) came from gas phase OH + NO reaction.
Similar temporal variation pattern can be found between P M and HONO/NO 2 (Fig. 3c).The simultaneous increase of P M and HONO/NO 2 after 09:00 LT indicates either the elevated conversion efficiency from NO 2 to HONO or the existence of unknown HONO sources during daytime.Recent studies showed the heterogeneous NO 2 reaction forming HONO on various surfaces being photosensitive or photoenhanced (George et al., 2005;Stemmler et al., 2006).Therefore, higher HONO production rate can be expected when surfaces are exposed to sunlight.The production of HONO through the proposed mechanisms should be related to parameters including the gas phase NO 2 concentration, RH, aerosol/ground surface area density, soot/organic composition on the surfaces, etc.The correlations of P M against NO 2 , RH, J NO 2 , S aw and σ a,532 do not provide a clear picture (Fig. 9).Relationship is neither found neither between P M and J NO 2 nor between P M and RH.According to Stemmler et al. (2006), the photosensitized HONO production will become saturated when the irradiance is above 200 W m −2 .Since our data are restricted to high J HONO we did not explore the sensitive region in the laboratory study.A similar phenomenon was observed by Stemmler et al. (2007) with regard to RH, thus the range of humidities explored here might not be sensitive again.The correlation between P M and NO 2 is two-fold.For NO 2 less than 4.5 ppb, a weak increase can be observed (Fig. 9a), while at higher NO 2 the values decrease (Fig. 9b).The higher NO 2 were only observed during the early morning (08:00-10:00 LT) and might caused by the NO to NO 2 conversion, the lower light levels and the lower boundary height.
The weak dependence of P M on S aw and σ a,532 suggests the aerosol surfaces might not be the places where HONO M is formed.Even weaker relationship between P M and parameters in Fig. 9 was found for individual measurement days except 24 July.The prevailing cable combustion events in surrounding areas happened during the night of 23 to 24 July.The high aerosol surface density and the soot concentration caused by the strong primary emissions could facilitate the HONO formation on aerosols.However, since the dependence of P M on parameters in Fig. 9 is generally weak, we thought the contribution of heterogeneous NO 2 reaction to P M should be quite small.To further prove our assumption, we calculated the γ NO 2 →HONO , γ soot NO 2 →HONO , and φ NO 2 →HONO around noon (11:00-13:00 LT) by assuming P M is only taking place on aerosol, soot particle and ground surfaces, respectively.The mean value of γ NO 2 →HONO and γ soot NO 2 →HONO derived from Eq. ( 10) is 0.03 and 0.39, respectively.These values are three to four orders of magnitude higher than those obtained in laboratory studies.Using a mixing height of 250 m as suggested by Stemmler et al. (2006) and assuming the same dry deposition velocity between day and night, the mean value of φ NO 2 →HONO calculated by Eq. ( 12) is 135.This value is far away from the theoretical range (i.e.0 to 1) and could not be explained by the uncertainty of ν ground HONO (ν ground NO 2 ).Therefore, we concluded that the contribution of heterogenous NO 2 reaction to P M was quite small.Other HONO formation mechanisms are needed during daytime.
The photolysis of adsorbed HNO 3 on ground surfaces was suggested to be the major HONO source at Whiteface Mountain in the US (Zhou et al., 2003(Zhou et al., , 2007)).As the adsorption of HNO 3 on ground surfaces occurs day and night, and the amount of adsorbed HNO 3 is proportional to the gas phase HNO 3 concentration, we correlated the average noontime P M with the GAC-IC measured HNO 3 concentrations averaged over the previous 24 h period (Fig. 10a).A possitive correlation was found for the limited days when continuous HNO 3 measurements were available (r 2 = 0.81).The firstorder approximation of production rate of HONO through this mechanism can be calculated as (Zhou et al., 2007), where f is the fraction of noontime full sun to which the surfaces were exposed, ν HNO 3 is the dry deposition velocity of HNO 3 over t d period, and j HNO 3 →HONO the photolysis rate constant of HNO 3 via HONO pathway.Here, by assuming the surfaces containing HNO 3 within a mix-ing height of 250 m were exposed to f = 1/4 noontime full sun, taking ν HNO 3 = 5 cm s −1 over t d = 24 h, and using the average value of P M = 0.97 ppb h −1 and [HNO 3 ] = 1.1 ppb in Fig. 10a, the photolysis frequency j HNO 3 →HONO derived from Eq. ( 13) was 5.7 × 10 −5 s −1 .This value is comparable to measurments by Zhou et al. (2003) in the laboratory ((1.2 ± 0.2) × 10 −5 s −1 ).In addition, we also correlated the noontime P M with the sum of the HNO 3 concentrations averaged over the previous 24 h period and the aerosol nitrate, both measured by GAC-IC (Fig. 10b).This scatter plot has

X. Li et al.: HONO atmospheric chemistry in Southern China
an even higher correlation (r 2 = 0.88) and the photolysis frequency j HNO 3 →HONO of 2.5 × 10 −5 s −1 is clsoer to the value obtained by Zhou et al. (2003).Given the good correlation of P M with HNO 3 and the sum of HNO 3 and aerosol nitrate, we concluded that the photolysis of adsorbed HNO 3 on ground surfaces in surrounding areas could be one likely source for the missing daytime additional HONO formation.This is also supported the value of the calculated j HNO 3 →HONO .
Recently, Su et al. (2011) demonstrated that the soil nitrite, which is formed through the biological nitrification and denitrification processes, can be the source of HONO.The soil HONO emission (P soil ) was estimated to be 2-5 ppb h −1 for Xinken site.Since the BG site is surrounded by farmland, we can not exclude the possibility that the soil nitrite contributes to P M .However, even if Xinken and BG are located in the same region, they represent different environments: Xinken is located at the southeast end of PRD around 200 km south of the Backgarden site.Xinken is at the seaside (at the Zhujiang river estuary) and located in a land reclamation area.However, the magnitude of the estimated P soil covers our observed P M .Since we did not measure the relevant information with regard to the nitrite concentration, the temperature, the pH value and the water content of the soil in the surrounding area during the campaign, it is difficult to quantify the P soil and thus its contribution to P M .

Conclusions
We performed continuous measurements of atmospheric nitrous acid (HONO) at Back Garden supersite (BG) in Pearl River Delta (PRD) in China, for 4 weeks in July 2006.The observed HONO concentrations are among the highest ever reported for rural areas, consistent with the fact that the PRD is one of the "hot-spots" of the HONO precursor NO 2 in the world.
Different from former studies in the same region (Su et al., 2008a,b;Qin et al., 2009), we for the first time have a full set of online measurements which are relevant for exploring the HONO formation.It is found that the direct emissions from combustion processes only have minor influence on the observed HONO concentration.During night, the reaction of OH + NO and the heterogeneous conversion of NO 2 on ground surfaces are the major sources of HONO.The observed net HONO production from OH + NO reaction can be up to 0.28 ppb h −1 .The average conversion frequency of NO 2 to HONO was found to be 0.024 h −1 which is higher than other places in the world.During daytime, to sustain the HONO concentration at ≈200 ppt, an additional HONO production (P M ) rate of 0.77 ppb h −1 is needed in addition to the productions from the well-known gas phase reactions.Our analysis demonstrated that, P M did not relate to the heterogeneous NO 2 conversion, but relate to the photolysis of adsorbed HNO 3 on ground surfaces.There is the possibility that an unknown portion of P M is correlated with the soil nitrite.However, the validation of this assumption requires on-site measurements of the soil chemical and physical properties.

Fig. 2 .
Fig. 2. Time series of HONO, J HONO , J NO 2 , RH, OH, k OH , NO, NO 2 , NO x , gas phase HNO 3 , aerosol light absorption (σ a,532 ), aerosol surface density (S a ) and RH corrected aerosol surface density (S aw ) at the Back Garden supersite in July 2006.

Fig. 3 .
Fig.3.Mean diurnal cycle of J HONO , HONO, NO 2 and HONO/NO 2 at the Back Garden supersite in July 2006.In (c), the + refers to the median, boxes represent 50 % (25 % to 75 %), the whiskers 90 % of the data.The blue diamond is the average.

Fig. 4 .
Fig. 4. Correlation between HONO and NO 2 .The dashed lines correspond to the HONO/NO 2 ratio of 10 %, 5 % and 1 %.The nighttime data are in the time period of 19:00-06:00 LT and color coded by the aerosol surface density; while the daytime data are color coded by the relative humidity.

Fig. 5 .
Fig. 5. Mean nocturnal variation of P net OH+NO , OH, NO and HONO.The color bar illustrates the local day (orange) and night (grey).These box-whisker diagram are based on 1 h averages for time intervals where both measurements were present.The + refers to the median, boxes represent 50 % (25 % to 75 %), the whiskers 90 % of the data.The blue × is the average.

Table 1 .
Lu et al. (2011)).(2009)angzhoubackgardensupersite.The instruments listed here do only refer to data used in this study.For a comprehensive overview of all instruments participating seeHofzumahaus et al. (2009);Lou et al. (2010);Lu et al. (2011); the measured parameters are explained in the text.: horizontal distance from inlet to HONO sampling.The sampling height on top of the LIF container was 6 m less than on the hotel roof.
a b : Laser Induced Florescence.c : Laser-flash Photolysis Laser Induced Florescence.d : Chemiluminescence.e : Spectro radiometer.f : Gas and aerosol collector followed by Ion Chromatograph detection.g : Twin Differential Mobility Particle Sizer and Aerodynamic Particle Sizer.h : Nephelometer and Photoacoustic spectrometer.i : Meteorology station.

Table 2 .
Overview on HONO measurements performed in or in the vicinity of big cities.The daytime values correspond to ± 3 h around local noon, the nighttime values reperesent averages between sunset and sunrise.
a : near Berlin.b : near Beijing.c : near Guangzhou.