Atmospheric oxidation capacity in Chinese megacities during photochemical polluted season: radical budget and secondary pollutants formation

Atmospheric oxidation capacity is the core of converting freshly-emitted substances to secondary pollutants. In this study, we present in-situ measurements at four Chinese megacities (Beijing, Shanghai, Guangzhou, and Chongqing) during photochemical polluted seasons. The atmospheric oxidation capacity is evaluated using an observational-based model with the 25 input of radical chemistry precursor measurements. The radical budget analysis illustrates the importance of HONO and HCHO photolysis, which contribute nearly half of the total primary radical sources. The radical propagation is efficient due to abundant NO in the urban environments. Hence, the production rate of secondary pollutants, i.e. ozone and fine particle precursors (H2SO4, HNO3, and extreme low volatile organic compounds (ELVOCs)) is fast, resulting in secondary air pollution. The ozone budget demonstrates that strong ozone production occurs in the urban areas, which results in fast ozone concentration increases locally 30 and through transport in downwind areas. The O3-NOx-VOC sensitivity tests show that ozone production is VOC-limited and that alkenes and aromatics should be first mitigated for ozone pollution control in the four megacities investigated. In contrast, NOx emission control will lead to more severe ozone pollution due to the decrease of NOx. For fine particle pollution, the role of HNO3NO3partitioning system is investigated with a thermal dynamic model (ISORROPIA2) due to the importance of particulate nitrate during photochemical polluted seasons. The strong nitric acid production converts efficiently to nitrate particles due to high RH 35 and ammonia-rich conditions during photochemical polluted seasons. This study highlights the efficient radical chemistry that maintains the atmospheric oxidation capacity in Chinese megacities, which results in secondary pollution characterized by ozone and fine particles.


Introduction
Air pollution is the one of the major threat to human health in cities (Kan et al., 2012).In China, the rapid economic development 40 accompanied by degradation of air quality for the last decades in the eastern areas (Chan and Yao, 2008).More than 300 million Atmos.Chem.Phys.Discuss., https://doi.org/10.5194/acp-2018-959Manuscript under review for journal Atmos.Chem.Phys.Discussion started: 2 October 2018 c Author(s) 2018.CC BY 4.0 License.people live in the North China Plain (NCP), Yangtze River Delta (YRD) and Pearl River Delta (PRD) regions in eastern China.Among all, Beijing, Shanghai, and Guangzhou are the metropolitan cities in these regions and suffering from severe air pollution.
To improve the air quality, emission mitigations have been conducted since the 2000s.As a result, the primary pollutant concentrations decline since then.However, secondary pollution characterized by ozone and fine particle has become the major contributor to air pollution.The major composition of fine particles are secondary components (Tie et al., 2013;Sun et al., 5 2006;Huang et al., 2014;Guo et al., 2014;Cheng et al., 2016;Zheng et al., 2005;He et al., 2001;Sun et al., 2004), e.g.sulfate, nitrate, and oxidized organic aerosol.This indicates high oxidation capacity in the Chinese pollution environments.The national-wide measurements showed ozone is the only one out of six air quality index substance that increases in the last five years.Therefore, it's difficult to control secondary pollution given the non-linear relation between primary and secondary pollutants.After all, atmospheric oxidation capacity is the key factor that converts primary pollutants to secondary ones.Therefore, it requires the 10 knowledge of the role oxidation processes on the secondary pollution formation.So far, only limited studies have been performed in China to elucidate the oxidation processes (Lu et al., 2018).The studies in 2006 found an OH source is missing in the current chemical mechanism for low NOx conditions (Lu et al., 2012;Hofzumahaus et al., 2009;Lu et al., 2013).The study in Wangdu (summer) and Beijing (winter) found an evidence of missing RO2 sources, which could lead to strong underestimation of ozone production (Tan et al., 2017;Tan et al., 2018c).The radical observation and model 15 comparisons highlight the uncertainty of the radical chemistry in China.The nitrous acid (HONO) were measured and constraint to the model in these studies, which were the major source of the OH-HO2-RO2 radical system.However, the majority of HONO sources is unclear so far (Su et al., 2011;Ye et al., 2016;Li et al., 2014b).On the other hand, the large aerosol content offers a large surface to conduct heterogeneous reactions.Radical loss on the aerosol surface could also play a role, which is not well understood due to the limited information based on the Chinese aerosol composition.The Chlorine chemistry is gaining increasing attention 20 recently, which acts as a radical source and source of particulate nitrate formation (Tham et al., 2016;Wang et al., 2016;Wang et al., 2017b;Wang et al., 2017c;Wang et al., 2017a).However, the large variability of uptake coefficient and ClNO2 yield adds large uncertainty to heterogeneous reactions (Xue et al., 2014;Tham et al., 2018).
With respect to secondary formation, the radical chemistry is insightful to reveal the key processes.The ozone production rate can be determined by the oxidation rate of NO by HO2 and RO2 directly.The gas phase oxidation produced semi-and/or low volatile 25 compounds, which are important precursors for particle formation.Sulfate, nitrate, and SOA are dominant contributors to the particle during heavy polluted episodes (haze events), all of which could be produced in the OH-initiated oxidation processes.
The Chengdu-Chongqing city group (population 90 million) locates in Sichuan Basin (SCB), southwest of China, representing the developing city clusters.Although the new city clusters also suffer from air pollution, only sparse researches have been conducted in these regions, especially for the secondary pollution formation.For the SCB region, only limited studies have been performed 30 regarding the oxidation capacity.Chengdu is evaluated using an observational-based model which found similar radical concentration and ozone production rate (Tan et al., 2018b).The VOC and ozone formation is evaluated in Chongqing (Su et al., 2018;Li et al., 2018b).
In this study, we present the measurement at four megacities, Beijing, Shanghai, Guangzhou, and Chongqing.The focus of this study is to illustrate the atmospheric oxidation processes in the urban areas.The observation periods were selected in typical 35 photochemical polluted seasons to explore the photochemistry and secondary pollution formation.An observational-based model is used to explore the oxidation capacity from the aspect of radical chemistry.This study aims to provide insight into the secondary pollution formation in Chinese megacities.With the intention to illustrate the common feature of the megacities oxidation capacity, we compared the results from different studies.Two questions need to be addressed in this study.First, what is the oxidation capacity in these megacities and which source(s) sustain it?Especially the comparison of city centers to suburban and rural locations, to foreign countries.Second, what is the secondary formation rate, e.g.ozone and nitrate?What is the limiting factor in secondary pollution formation?Finally, the diagnosis of atmospheric oxidation capacity and secondary pollution formation could provide fundamental knowledge for further air pollution control in China.

Measurement sites 5
This study presents the measurements at four Chinese megacities, namely Beijing, Shanghai, Guangzhou, and Chongqing.These cities were located in the highly polluted region (Fig. 1).Beijing, Shanghai, and Guangzhou are the main cities in North China Plain (NCP), Yangtze River Delta (YRD), and Pearl River Delta (PRD), which are representative of the most developed region in China.Chongqing is one of the biggest cities in southwestern China (population of 30 million), representing the Chinese developing city clusters.10 The detail of the field campaigns is summarized in Table 1.All measurements sites locate within the city downtown areas to represent the conditions in the city center.Beijing is located on the north edge of the North China Plain and is the most north station (39.9 °N) than others.The campaigns were mainly conducted in summer to represent the most active photochemical season.The Beijing campaign took place in mid-summer (June) and thus the strongest solar radiation input.The latitude of Shanghai (31.1 °N) and Chongqing (29.6 °N) are similar as well as the observation period (August).Therefore, the solar input level should be 15 comparable.Given the synoptic flow pattern interaction, the atmospheric pollution is expected to be most serious in PRD in autumn (Zhang et al., 2007;Li et al., 2014a).The measurement in Guangzhou was performed in late October to represent the photochemical polluted period.However, the latitude is lowest (23.1 °N), which can partly compensate the seasonal effect.The maximum of 1-h averaged O3 concentrations were more than 100 ppbv in Shanghai, Beijing, Guangzhou, but only 79 ppbv in Chongqing, which demonstrated a separation between the two classes of cities. 20

instrumentation
In all sites, similar instrumentations were deployed as a standard super site.All Thermo instruments were carefully maintained and calibrated during the campaigns.Only a brief description of the measurement techniques is presented.Ozone was measured by UV absorption method using the Thermo O3 analyzer (Model 49i).The NO2 measurement was performed by the chemiluminescence after chemical conversion to NO. CO was measured by the infrared absorption technique using Thermo instrument (Model 20). 25 Speciated VOCs measurement was performed by the GC-MS/FID.The photolysis frequencies were measured by spectrum radiometer.Meteorological parameters were measured simultaneously, e.g.ambient temperature, pressure, and relative humidity.

The model
A box model based on the Regional Atmospheric Chemical Mechanism version 2 (Goliff et al., 2013) to simulate the concentrations of the short-lived OH, HO2 and RO2 radicals and other unmeasured secondary species.The newly proposed isoprene mechanisms 30 are also incorporated (Peeters et al., 2014;Fuchs et al., 2013).The model was constrained to the observation of photolysis frequencies, long-lived trace gases (NO, NO2, O3, CO, C2−C12 VOCs), and meteorological parameters.Since nitrous acid (HONO) was not measured in these campaigns, it was fixed to 2% of the observed NO2 concentrations because good correlation was found between HONO and NO2 in different field studies with a constant ratio being 0.02 (Elshorbany et al., 2012).The uncertainty of such parameterization is discussed in section 3. and the reaction rate constants.Taking into account the uncertainties of both measurements and kinetic rate constants, the model calculations is approximately 40% (Tan et al., 2017).

Overview of measurements
The mean diurnal profiles of measured ambient temperature, j(O 1 D), and CO, O3 (Ox = O3+NO2), NOx(=NO+NO2), and AHC 5 concentrations are shown in Fig. 2 (the time series are shown in Fig. S1-S4).The ambient temperature is relatively similar in Beijing, Shanghai, and Chongqing, but lower in Guangzhou because the campaign was conducted in a later time of a year.Similarly, the photolysis frequencies are smaller in Guangzhou.However, j(O 1 D) is highest in Beijing and are comparable in Shanghai and Chongqing.However, diurnal maximum O3 concentrations are highest in Shanghai (80 ppbv) followed by Beijing (72 ppbv), Guangzhou (65 ppbv), and Chongqing (56 ppbv).The diurnal peak of O3 appears at 15:00~16:00 LT in Beijing, Guangzhou, and 10 Chongqing.In Shanghai, the peak of O3 shows up at 13:00 LT due to the fast increase in the morning.During the measurement period, the observed ozone concentrations exceed the Chinese National Air Quality Standard Grade II (99.3 ppbv) in Beijing, Shanghai, and Guangzhou (Table 1).When a measurement site is close to NOx emission sources, part of the O3 is titrated to NO2 by fresh emitted NO.Although O3 is regenerated in a few minutes to half hour after the photolysis of NO2, O3 is stored temporally in the form of NO2.Therefore, Ox, the sum of O3 and NO2, is a better metric to describe ozone pollution in the urban area.Ox 15 concentrations are also shown with broken lines in Fig. 2. In this case, the Ox mean diurnal profiles in Beijing, Shanghai and Guangzhou show maximum values of about 90 ppbv (1-hour resolution), indicating the ozone pollution are comparable in these cities during the measurement period.In Chongqing, the maximum of the diurnal average is 66 ppbv.The ozone pollution is serious in autumn in Guangzhou due to the unique synoptic system, including the surface high-pressure system, hurricane movement and the sea-land breeze (Fan et al., 2008).In Shanghai, the synoptic weather is crucial to pollution accumulation, and the ozone 20 concentrations are reduced in August and September due to the cleaning effect by the summer Monsoon (Dufour et al., 2010;Geng et al., 2015).
The ozone precursors, NOx, and AVOCs are shown in Fig. 2. In all cases, the NOx concentrations show a typical diurnal profile with a minimum in the afternoon.In the morning, the peak is caused by the transportation emission during the rush hours.In Shanghai, the nighttime NOx concentrations also decrease after sunset but increase after midnight, which anti-correlates with the 25 O3.The AVOCs concentrations show similar diurnal profiles as those of NOx, which suggests both AVOCs and NOx are originated from the same sources, e.g.traffic emission and/or manipulated by the same factor, e.g.boundary layer development.CO is also a precursor of ozone.The diurnal profiles of CO are almost flat due to their long lifetime versus OH.A small peak appears in the morning rush hours due to poor dilution condition and enhanced emission from transportation.

OH reactivity and composition 30
OH reactivity (kOH) represents the pseudo first-order reaction rate constant of OH radical.It is a measure of the sum of sink terms due to OH radical reactants Xi, which depends on their ambient concentration [Xi] and their rate coefficient with OH radical.
Mathematically, kOH equals to the inverse of ambient OH radical lifetime.The use of OH reactivity is of importance to understand the OH consumption potential.
The mean diurnal profiles of OH reactivity calculated from a box model are presented in Fig. 3, including the contribution from 35 CO, NOx, VOCs, and model generated secondary species.In general, the OH reactivity is lowest in the afternoon and highest in the morning rush hours due to the change of dilution condition over a day.The total OH reactivity can be measured directly with three techniques (Fuchs et al., 2017).In China, OH reactivity measurements were performed in NCP and PRD, showing large seasonal and spatial variation.In Beijing, kOH measurements were conducted in an urban and suburban site during summer (Lu et al., 2013;Williams et al., 2016;Yang et al., 2017).The OH reactivity was on average in the range of 10 to 30 s -1 and showed large daily variation due to the meteorology changes.In PRD, OH reactivity was measured in a rural site (Backgarden) in summer and suburban site (Heshan) in autumn, both of which were close to Guangzhou 5 (<100km).In Backgarden, the average OH reactivity was in the range of 20 to 50 s -1 with a large contribution from isoprene (Lou et al., 2010).In Heshan, downwind of Guangzhou, OH reactivity was dominated by anthropogenic species.However, the averaged kOH was comparable in two sites (Tan et al., 2018a).
In this study, the modeled OH reactivity in Guangzhou is highest (20-30 s -1 ) among all the cities (Fig. 3), indicating a strong influence of anthropogenic emission in Guangzhou.In Beijing and Chongqing, OH reactivity is comparable, in the range of 15 to 10 25 s -1 .In Chengdu, another big city in southwestern China, the modeled OH reactivity was also found to be in the range of 15 to 30 s -1 at three urban sites, with an exception that large OH reactivity was contributed by alkenes due to petrochemical complex industry (Tan et al., 2018b).In Shanghai, the OH reactivity was lowest (<15 s -1 ) due to the small contribution from CO and NOx.
The speciation is clearly shown in Fig. 4. The contribution of OH reactivity was relatively similar among all cities as a typical fingerprint of anthropogenic emission.NOx is the most important OH reactants that contribute 28%~35% to the total reactivity.In 15 total, the inorganic species (CO and NOx) contribute more than half of the reactivity.The measured AVOCs contribute to the total OH reactivity from 14% to 26%.In Beijing, the AVOC contribution is lowest as well as the absolute reactivity (2.3 s -1 ).In Guangzhou, the contribution and absolute reactivity of AVOC (6 s -1 compared to 2~3 s -1 in other cities) are highest.However, isoprene is relatively small in Guangzhou (0.4 s -1 ) due to the observation season.In comparison, isoprene reactivity is up to 1 s -1 in Beijing and Chongqing but becomes negligible in Shanghai.The small isoprene concentration was also reported in another study 20 (Geng et al., 2011).
The relative contributions of grouped VOCs are shown in Fig. 5. Anthropogenic VOCs are usually more important than the biogenic ones.The isoprene accounted for 15% of the measured VOC reactivity in Beijing and Chongqing.The shares of different VOCs groups are comparable with a slightly different portion of alkanes and alkenes.Aromatics become the most important VOC group in Guangzhou, accounted for about half of the measured VOC reactivity, which is related to vehicle emissions and industry 25 that produce VOC-related products (Zheng et al., 2009).
The OH reactivity concept is useful to estimate the ozone production for VOCs because it describes the VOC degradation rate initiated by OH oxidation that leads to net ozone production in the presence of NO and sunlight (see section 4.1).On the other hand, the ozone formation potentials (OFPs) are used to describe the theoretical ozone production maximum.This metric shows the temporal Lagrangian evolution of O3 production potential within a mixture of air that undergoes fully oxidation.The OFP of 30 individual VOCs is calculated by the product of measured VOC concentrations and its MIR value (Carter, 2009), which are sum up later according to the VOC classification.Aromatics become dominant species in the OFPs due to their large carbon numbers and thus high MIR values.In Guangzhou, aromatics contributed up to 70% of the share of OFPs, followed by Shanghai (55%), Beijing (43%), and Chongqing (42%).
In comparison, the mixing ratios of different VOC are shown.It is worth noting that alkanes (including ethyne) accounted for a 35 large fraction in the mixing ratio, their contribution to OFPs and OH reactivity was small.Therefore, they are not important in ozone formation and radical chemistry.It highlights the importance of the concept of OH reactivity and OFPs to describe photochemical processes accurately.

Concentrations
The OH, HO2, and RO2 concentrations are derived from box model calculations (Fig. S1-S4).The mean diurnal profiles are shown in Fig. 6.Modeled OH concentrations show distinct diurnal variation in cases.The maximum of OH means the diurnal profile is largest in Beijing and Shanghai (7×10 6 cm -3 ).The maximum is 4×10 6 cm -3 in Chongqing and further decrease to 2×10 6 cm -3 in 5 Guangzhou.The relative change of OH maximum is consistent with the observed photolysis frequencies (Fig. 2).In fact, modeled OH concentrations show good correlation with the observed j(O 1 D) (R=0.83,0.93, 0.87, and 0.87 for Beijing, Shanghai, Guangzhou, and Chongqing) because OH chemistry relies on solar radiation input tightly.The OH maximum ranking is consistent with that of j(O 1 D).The j(O 1 D)-OH correlation slope was largest in Shanghai (3.0×10 11 cm -3 s -1 compared to 2.0×10 11 cm -3 s -1 in the other cities) because the OH reactivity is lower and OH lifetime becomes longer.The good correlation between OH and j(O 1 D) 10 was found in other field campaigns conducted in China with a correlation coefficient more than 0.8 (Tan et al., 2017;Rohrer et al., 2014;Lu et al., 2013;Lu et al., 2012).However, the correlation slopes were about (4.5±0.5)×10 11cm -3 s -1 , which are on average two times larger than the modeled results in this study.The larger correlation slope suggests that solar radiation converts more efficiently to radical concentrations.The different conversion factor could be due to the different chemical regimes in each location.
The larger factor was found in suburban and rural sites, where the air masses were more oxidized.The slope is comparable to the 15 results obtained in urban environments where the slope was in the range of (2-4)×10 11 cm -3 s -1 (Holland et al., 2003;Michoud et al., 2012;Griffith et al., 2016;Whalley et al., 2018).
The peroxy radical concentrations are variable in different campaigns depending on the chemical conditions.A general feature is found in all cases that peroxy radical concentrations are suppressed in the morning.Since then, peroxy radical concentration increases gradually and reach a peak at about 14:00 (two hours later than the maximum of solar radiation), because of the 20 suppression by high NO.The peroxy radicals concentrations are highest in Chongqing with a maximum of mean diurnal profiles being 5×10 8 cm -3 for HO2 and 7×10 8 cm -3 for RO2.Meanwhile, the HO2 and RO2 concentrations are comparable in the other cities despite the difference in the solar radiation and chemical conditions.In Chongqing, the relative large VOC/NOx ratio leads to the highest peroxy radical concentration in the model, which reflects in the efficient radical recycling and ozone production.

Radical budget analysis 25
All radical reactions are classified into four groups (initiation, termination, propagation, and thermos-equilibrium with reservoir species).The reaction turnover rate illustrates the important processes in the ROx radical reactions framework.The initiation and termination rate are shown in Fig. 7.The following radical budget analysis will focus on the daytime conditions (06:00-18:00) if no additional clarification.
In contrast, alkenes ozonolysis is the dominant radical source during nighttime and the yields of OH, HO2, and RO2 radicals depend on individual alkenes.The maximum of P(ROx) mean diurnal profile is largest in Beijing (5 ppbv/h), followed by Shanghai 35 (4.6 ppbv/h) and Chongqing (4.3 ppbv/h).The daytime averaged P(ROx) is smaller in Shanghai due to the narrower peak of photolysis frequencies (Fig. 2) and shorter photolysis reaction time.The primary radical source is smallest in Guangzhou (3.2 ppbv/h) due to the later observation period in a year.However, the alkene ozonolysis contributed significantly to the radical sources in Guangzhou (43% of the total primary source for daytime conditions), which could attribute to higher abundance of alkenes due to special emission inventory (see section 3.2).
Radical termination can be divided into two groups, the nitrogen-containing compounds, including HONO, HNO3, RONO2, and PAN-type species (LN).The other pathway leads to peroxides formation result from the combination of two peroxy radicals (LH).
The ratio between LN and LH depends on the NOx concentrations.In urban environments, the limiting factor for radical propagation 5 is the abundance of VOCs.In our case, the radical termination is dominated by LN (>70%).Among all, the nitric acid formation was the major contributor to the radical termination in all cities (>50%).In Chongqing, the peroxide formation path contributes 26% to the radical termination, especially the ratio increase to 32% during the afternoon due to the higher VOC/NOx ratio.Net PAN-type species formation as a radical loss becomes relatively important in Guangzhou (about 20%) due to the lower temperature (Fig. 2).It is reported that on average 25% of the radical can be lost via forming PAN-type species in Beijing during winter (Tan 10 et al., 2018c).Besides, PAN-type species formation becomes important in the urban area, e.g. it contributes 30% to the total radical loss in London downtown area (Whalley et al., 2018).
The comparison of the four cities is clearly shown in Fig. 8.The HONO photolysis is the dominant OH source in all cities except in Shanghai.The O3 photolysis is more important than HONO photolysis in Shanghai, contributing 55% to the total OH primary sources and 23% of the total radical sources.In all cities, the primary production of HO2 is comparable to that of OH, which is 15 mainly contributed by the HCHO photolysis and alkene ozonolysis.These results are consistent with the model calculation performed in Beijing (Yang et al., 2018) and Hong Kong (Xue et al., 2016).The importance of HONO and HCHO photolysis to radical primary production is also found in suburban and rural environments (Lu et al., 2012;Lu et al., 2013;Tan et al., 2017).In the base model scenario, HONO is scaled to NO2 measurements and the uncertainty of this assumption is further discussed following.In the base model, the HONO concentrations are scaled to the observed NOx concentration using a scaling factor 0.02 20 (Elshorbany et al., 2012).In this study, we use this scaling factor between HONO and NOx to simplify the discussion of unknown HONO sources.In the original RACM2 model, only homogenous source is included, i.e.OH+NOHONO, which is not sufficient to support the high daytime HONO concentrations and, as a result, leads to a strong underestimation of OH concentrations (Su et al., 2011;Yang et al., 2014;Tong et al., 2016;Ye et al., 2016;Li et al., 2012;Li et al., 2014b).To evaluate the impact of missing HONO source on the radical chemistry, we switched off the scaling between HONO and NOx in a sensitivity test.Therefore, the 25 results show that OH concentrations reduce by about 20% if the only homogenous source is considered.The modeled HO2 and RO2 concentrations are also reduced correspondingly (15-20%).
The RO2 source strength is in the range of 0.2 to 0.3 ppbv/h, which is balanced by the RO2 loss via organic nitrate formation (RONO2).Therefore, the imbalance between primary source and termination of HO2 (P>D) results in a net flow from HO2 to OH.
Since the sum of ROx initiation and termination rate should be balanced, the excess HO2 radical production leads to the large 30 formation of nitric acid through the net flow from HO2 to OH.The equilibrium between RCO3 (acetyl peroxy radicals) and PANtypes species tends to result in a radical loss (0.1~0.3 ppbv/h).The equilibrium between HO2 and HNO4 is fast with null effect on the radical budget.
The OH reacts with VOCs or CO and produces peroxy radicals.The peroxy radicals react with NO producing NO2, which lead the net production of O3 in the presence of sunlight (see section 4.1).Nitric acid can be formed from the reaction between NO2 and 35 OH, which is an important precursor of fine particles (see 4.2).Therefore, the efficient radical propagation facilitates the secondary pollution formation.
In addition, the radical propagation between OH, HO2, RO2, RCO3 are also shown in the inner part of Fig. 8.In RACM2, RCO3 is mainly produced in the reaction between OH radical and aldehydes.RCO3 is separated from the sum of RO2 family and explicitly shown in the radical propagation because RCO3 reacts with NO and converts to RO2.In the sum of RO2 concentrations, RCO3 is considered as a subgroup of RO2 radicals (e.g.Fig. 6).
The OH reactions with CO and VOC produce to HO2 and RO2, respectively.The conversion from OH to RO2 is slightly faster than OH to HO2.In addition, the conversion from OH to RCO3 is about one-third of the rate from OH to RO2.In the presence of NO, OH is regenerated from peroxy radicals and ozone is produced in the same process (see section 3.5).The flow rate from RO2 to 5 HO2 is less than half of the flow rate from HO2 to OH.Even considering the contribution from RCO3+NO, RO2 (+RCO3) contributes less than half of the P(O3).Surprisingly, the HO2 contribution to the total ozone production is constant at 63% in all cases.In this study, we find that the ratio between P(O3) and kvoc× [OH] is rather constant in the range of 1.5 to 1.6.Therefore, the robust relation between OH oxidation and ozone production indicates that it would be justified to estimate the ozone production rate using the reaction rate of OH+VOC (CO) in the future.However, the nature of such a robust relation need to be investigated 10 before real application.

Local ozone production
Ozone is generated from the NO2 photolysis, which produces NO simultaneously.The removal of NO without consuming O3 leads 15 to net ozone production.In the photochemical system, peroxy radicals (HO2 and RO2) are the major NO consumers in all photochemical reactions.
In this study, the ozone formation rate F(O3) is calculated from the NO oxidation rate by HO2 and RO2 radicals as denoted in E2.
The chemical loss of O3 includes O3 photolysis (producing O 1 D followed by H2O reaction) and its reactions with alkenes, OH, 20 HO2 (E3).Since NO2 can be regarded as a reservoir species of O3, the reaction between NO2 and OH is also considered as ozone chemical loss.In fact, the use of Ox is helpful to avoid the interruption of NO titration due to fresh emission, which is conservative to describe the ozone concentration change due to photochemical reactions (Liu, 1977).
Therefore, the net ozone production rate P(O3) is determined by the difference between E2 and E3.The mean diurnal profiles of 25 P(O3) are shown in Fig. 9.The ozone production rate is highest in Beijing and Shanghai with a diurnal averaged maximum reaching 19 ppbv/h.Although the peroxy radical concentrations are highest in Chongqing, the ozone production rate only shows a board peak at 13 ppbv/h.The duration of ozone production is different with the longest duration in Beijing (13 hours).Therefore, the daily integrated ozone production rate is largest in Beijing (136 ppbv).For comparison, the integrated ozone production rate were 92, 40, and 105 ppbv in Shanghai, Guangzhou, and Chongqing, respectively.30 The Ox concentration change depends on both the local production and physical processes (E4).R(O3) represents the combined effect of all physical processes, including horizontal transportation, vertical mixing, deposition and so on.
As shown in Fig. 9, the Ox concentration changes are derived from the derivative of observed Ox concentrations.In Beijing, Guangzhou, and Chongqing, the d(Ox)/dt show similar increasing trends in the morning starting at 06:00 (Fig. 9).In Shanghai, 35 positive derivative shows up at 05:00, one hour earlier than the other three cities.Also, the increase rate is fastest in Shanghai In the bottom panel of Fig. 9, the difference between d(Ox)/dt and P(O3) are shown, which denotes the local transportation of Ox (positive: inflow; negative: outflow).In Beijing and Chongqing, the local production rate is larger than the Ox concentration increase.Hence, it suggests that the photochemical produced Ox at the measurement site is transported to downwind regions.In 5 Shanghai and Guangzhou, both positive and negative values appear during daytime, indicating the local ozone budget changes from importation to exportation.In the morning, the increases in Ox concentrations are larger than the local ozone production rate, which is supported by additional Ox import from the volume outside.The Ox is most likely entrained from the air aloft because the raising up boundary layer mixes in the air mass aloft from the residual layer, which maintains the high load of Ox that produced from the last day and then isolated from the surface layer.Especially in Shanghai, the fast Ox concentration increase before 08:00 10 is mainly caused by transportation given the relatively small local production rate.The Ox import stops at about 10:00 and the surface layer becomes a net Ox source region.The Ox importation also stops in the later time of a day (at about 14:00) in Guangzhou.
The Ox transportation to downwind areas (negative R(O3)) are observed in Beijing, Shanghai, and Chongqing during the afternoon, and in Guangzhou after sunset, which suggest the city centers are important for Ox formation in a regional scale.

O3-NOx-VOC sensitivity 15
The OH-HO2-RO2 radical budget is useful to diagnose the O3-NOx-VOC sensitivity as discussed in section 3.3.2.In this section, we use the ratio of nitrate formation rate (LN) to total radical production/termination rate (Q), known as LN/Q ratio to evaluate the ozone production sensitivity as suggested by Kleinman et al. (1997).The threshold is 0.5 of the LN/Q ratio.When LN/Q is greater than 0.5, the radical termination is outweighed by the nitrate formation which indicates the ozone production is in limited to VOCs abundance; On the other hand, when LN/Q is less than 0.5, peroxy radicals self-combination dominates radical termination 20 indicating ozone production is under NOx-limited control.In the radical budget analysis, we found the LN contributed more than 70% to the radical termination in all cities.With a larger than 0.5 LN/Q ratio, ozone production was in VOC-limited regime, which persist in all the cities in this study.
The Relative Incremental Reactivity (RIR) method is also used to evaluate the O3-NOx-VOC sensitivity.RIR is a useful metric for ozone sensitivity to individual precursors.The model input parameters are changed by a certain amount and the corresponding 25 ozone concentration change is compared and summarized to reveal the O3-NOx-VOC sensitivity.The calculation of RIR is expressed in E5.

RIR(
In equation 1, X represents a set of primary pollutants and O3 represents the modeled O3 concentrations in the base case.∆C(X)/C(X) represents the relative change in the primary pollutants in one of the sensitivity tests.As a result, the relative change in modeled 30 ozone concentrations is given by ∆O3 (X)/O3.
As shown in Fig. 10, the RIR values are calculated for NOx, AHC, NHC, and CO, respectively.The AHC has the highest RIR (> 1 %/%) in all cities because ozone production is limited to the abundance of VOC in urban areas.In comparison, the RIR of NHC and CO are small (< 0.2 %/%), which demonstrates isoprene and CO are not important ozone precursors in all cities.For NOx, the RIR values are negative indicating the ozone production is in NOx-titration regime and thus reducing NOx could lead to 35 increase in Ox concentrations.It can be explained from the respect of radical budget that the OH+NO2 reaction rate is a dominant part of the radical termination in all cities (Fig. 7).If the radical termination is reduced, the OH-HO2-RO2 radical propagation will become more efficient and thus the modeled radical concentrations will increase.One should keep in mind that HONO is scaled to NO2 in our base model.The reduction in NO2 also leads to a reduction in HONO, which means less primary radical sources.
Therefore, less ozone will be produced from the radical recycling, which compensates for the titration effect partly.We performed further sensitivity study that HONO is free running.The results show that the negative effect becomes more significant compared to the base model scenario (Fig. S5).In comparison, a small negative effect showed up in the RIR analysis in the base model run.
It is because the HONO photolysis only contributed about 40% to the primary source, smaller than the portion of OH+NO2 to the 5 termination.In another word, if HONO is As discussed above, VOCs emission control is critical to ozone pollution reduction.To perform accurate VOC mitigation for O3 pollution control, the AHC is further split into alkanes, alkenes, and aromatics.As shown in section 3.2, alkenes and aromatics are the dominant VOC groups with respect to OH reactivity.The RIR analysis also showed that reduction in alkenes and aromatics are important for ozone pollution control in these megacities (Fig. 10).Xue et al. (2014) compared the observation in four cities 10 (Beijing, Shanghai, Guangzhou, and Lanzhou) but located in downwind suburban areas.The results showed that ozone increase could be attributed to local ozone production in Shanghai, Guangzhou, and Lanzhou (Xue et al., 2014).Besides, ozone production was in VOC-limited regime in Shanghai and Guangzhou, among which aromatics were the most important contributor.A comparison of two megacities in China (Shanghai and Tianjin) showed that the ozone production is highly variable depending on the VOC speciation at certain NOx concentrations (Ran et al., 2012).Alkenes were important ozone precursors in Tianjin, while 15 in Shanghai, aromatics dominated the ozone production.Based on a one-year measurement in Nanjing (YRD), the ozone production was in the VOC-limited regime and the Nanjing-Shanghai axis with its city clusters in between was subjected to a regional photochemical pollution (Ding et al., 2013).The regional model (WRF-Chem) also showed ozone production was under strong VOC-limited not only in urban but a larger regional area (Tie et al., 2013).The PRD region has been well studied from field campaigns, emission inventories, and regional modeling (Zhang et al., 2008a;Zheng et al., 2009;Ding et al., 2004).The results 20 showed both urban (Guangzhou) and downwind rural site (Xinken) were under the VOC-limited regime and negative RIR values for NOx were found (Zhang et al., 2008b).Meanwhile, the ozone-NOx-VOC sensitivity was found to be VOC-limited in Chongqing and the traffic emission contributed 44% to the VOCs in an urban area (Su et al., 2018;Li et al., 2018b).

Nitrate production potential
Nitric acid is one of the major products generated by the radical system for high NOx conditions, which is an important precursor 25 of particulate nitrate (NO3 -).Nitrate has become a significant portion in particles in Beijing, Shanghai, and Nanjing (YRD) during summertime (Li et al., 2018a).The gas phase nitric acid HNO3 together with ammonium NH3 form a gas-particle partitioning equilibrium with NH4NO3 (R3), which depends on the relative humidity, temperature, and the aerosol contents (Seinfeld and Pandis, 2016).
The gas-particle partitioning is calculated by the aerosol thermodynamic model (ISORROPIA).The partition is assumed to reach equilibrium because the time scale of reaching equilibrium is in range of minutes, 1-2 orders smaller than those of deposition and chemical production (Morino et al., 2006;Neuman et al., 2003).Therefore, the photochemical produced HNO3 will deposit on to the aerosol if the ambient NH3 is sufficient.The nitric acid could also lose via deposition.The decomposition rate is set to be 7 cm s -1 , which results in a deposition timescale being 8 hours by assuming the boundary layer height to be 2 km (typical values 35 for summertime).The total ammonia (NH4TOT) is calculated from an iterative method for each case to reproduce the gas-phase NH3 concentrations reported by Pan et al. (2018).From the field measurements, the averaged NH3 concentrations were Beijing: 16.3 μg/m 3 , Shanghai: 14.6 μg/m 3 (the number is adapted from a close-by city Nanjing), Guangzhou: 5.9 μg/m 3 , and Chongqing:  The modeled nitrate concentration and partitioning in Beijing are shown to illustrate the typical pattern of particulate nitrate formation (Fig. 11a).The total nitrate maximizes in the late afternoon while the particulate nitrate shows a board peak at night, which is mainly driven by the stronger gas-to-particle partitioning due to higher RH.Since deliquesce relative humidity (DRH) of NH4NO3 is about 60% in all cases, the partitioning changed dramatically with the relative humidity above DRH (nighttime) and below DRH (daytime).5 To investigate the nitrate concentration dependence on the nitrate production rate and ambient ammonia concentrations, the averaged nitrate concentrations are plotted as a function of daily integrated nitric acid production rate and total ammonium (NH4 + (a)+NH3(g)) concentrations.As shown in Fig. 11b, the isopleth diagrams are split into two parts by the dashed line to represent the nitrate-(upper left) and ammonium-sensitive (lower right) regimes.However, the threshold for nitrate-and the ammoniumsensitive regime is not distinct in the small chemical range.Actually, the nitrate concentrations are sensitive to both precursors.10 The daily integrated nitrate production rate and averaged total ammonium concentrations for each city are denoted by the circles (Fig. 11b).The circles are located above the ridgeline, which means nitrate concentrations are more sensitive to the change of nitrate production rate.In comparison, the averaged nitrate concentrations derived from the time-dependent calculations (as shown in Fig 11a) are 64±7, 27±8, 16±4, and 20±4 μg/m 3 per day in Beijing, Shanghai, Guangzhou, and Chongqing, respectively.These numbers are consistent with the EKMA plot (Fig. 11b), which indicates the results are reasonable.15 In the base scenario, the other chemical compositions are set to zero (total SO4 2-, Na + , and Cl -).The simplification uncertainty is evaluated by the following sensitivity tests.As shown in Table 2, the NH4NO3 concentration and the particulate nitrate to total nitrate ratio ε(NO3 -) is calculated for different model scenarios.First, if NH3 concentrations are changed by a factor of 2, the change to partitioning is relatively small in Beijing because the NH3 concentrations are high.It's worth noting that NH4NO3 concentration is still high even if the NH3 concentration is reduced by two-fold in Beijing, which highlights the difficulty in particulate nitrate 20 reduction.In contrast, the change almost linearly correlates with the change of NH3 in Guangzhou during the daytime due to the limited amount of NH3.The role of other anions and cations is investigated by setting the SO4 2-and Na + to be 10 μg/m 3 , respectively.Since SO4 2-reacts with NH4 + and thus competes with the formation of NH4NO3.The NH4NO3 formation will be limited if NH3 is not sufficient given the NH3 to tend to react with SO4 2-before NO3 -.The nitrate concentrations decrease by 1-3 μg/m 3 when SO4 2- is fixed in all cities to be 10 μg/m 3 during day and night (Table 2).In contrast, additional cations can neutralize more nitrates and 25 enhance the gas-to-particle partitioning.For example, if 10 μg/m 3 of Na + is added, stronger nitrate production is found (Table 2).
It is caused by two kinds of effect.First, the stronger partitioning enhances the particulate nitrate directly.Second, the total nitrate loss is reduced by less HNO3(g) deposition because more nitrate remains in the particle phase.

Atmospheric oxidation capacity and secondary pollution formation
Atmospheric oxidation capacity is mainly contributed by OH radical which dominates the chemical removal of trace gases (e.g. 30 CO, NO2, VOCs).The OH reactions convert primary pollutants to oxidized products (e.g.CO2, HNO3, OVOCs).As shown in Fig. 12, the daytime averaged OH oxidation rate is up to 10 ppbv/h in Beijing, indicating strong oxidation capacity (daily integrated oxidation rate > 100 ppbv).The OH oxidation rate is comparable in Shanghai and Chongqing (~5 ppbv/h) and reduces to 4 ppbv/h in Guangzhou.In this study, the OH oxidation rate correlates with the strength of primary radical sources P(ROx).Actually, the ratio between the radical recycling rate and the primary production rate indicates the efficiency of radical propagation (E6), also 35 known as the radical chain length.The radical chain length is on average 2.9±0.3 for all cases, which is consistent with the results (3-5) derived from radical observation in urban areas (Kanaya et al., 2008;Ren et al., 2006;Emmerson et al., 2005).
As shown in section 3.3.2,P(ROx) is highly correlated with j(O 1 D) because photolysis reactions dominate the ROx primary sources.
An exception happens in Guangzhou that P(ROx) decrease by a factor of 2 compared to Beijing although j(O 1 D) is reduced by a factor of 3. It is because alkenes ozonolysis reactions increase to 0.7 ppbv/h in Guangzhou (0.2-0.3 ppbv/h for other cities) and 5 contribute nearly half of the primary source.The reduction of P(ROx) in Guangzhou is partly compensated.
More than half of the OH oxidation rate was contributed by the reaction with VOC that produce less volatile species (OVOCs) (Fig. 12).Therefore, these oxidized compounds have the potential to contribute to the particle formation due to their low volatility (Odum et al., 1997).Besides, the oxidation of NO2 produces HNO3, which can contribute to the particle formation given the relatively high ambient NH3 concentrations (see in section 4.2). 10 O3 is another important secondary pollutants generated from the OH-HO2-RO2 radical system.The averaged P(O3) is consistent with the OH oxidation rate in four cities (Fig. 12).The ratio between ozone production and radical primary production rate is used to evaluate the ozone production efficiency (OPE).In this study, the OPE is highest in Chongqing (3.6 on average) due to the relative high VOC/NOx ratio.In contrast, OPE is only 2.2 in Guangzhou and increases to 3.4 and 3.1 in Beijing and Shanghai.In comparison, the OPE was found to be in the range of 3 to 7 in other cities (Kleinman et al., 2002;Lei et al., 2008).The OPE 15 determined in this study is in the low range compared to other cities in the US, which is due to the suppression of high NOx conditions.Since OPE generally increases with time when a plume is transported and diluted (Kleinman et al., 2002), the ozone production becomes more efficient in the suburban areas of the megacities.
Finally, the radical inner recycling rate is on average 4-5 times larger than that of radical initiation (termination), which demonstrates that the secondary pollutant formation rate (e.g.ozone, nitric acid) is enhanced by the efficient radical recycling.20

Summary and conclusion
Secondary pollution has been increasing in the last decade in Chinese cities despite the reduction of primary pollution.Ozone and fine particle precursors (e.g.H2SO4, HNO3, and ELVOCs) are generated from the radical reactions.Therefore, atmospheric oxidation capacity is the core of secondary pollution formation.In this study, we present the observation of the radical precursors at four Chinese megacities during photochemical polluted seasons, namely Beijing (July), Shang (August), Guangzhou (October), 25 and Chongqing (August).A box model is used to simulate OH, HO2 and RO2 concentrations.The key processes are elucidated with explicit radical budget analysis.The formation mechanism of secondary pollutants (ozone, particle precursors) is investigated with the chemical model.The major findings for this study are shown in the following: 1) The metric of OH reactivity is used to demonstrate the air pollutants burden.The modeled OH reactivity show typical diurnal profiles with a maximum in the morning and minimum in the afternoon.The OH reactivity is highest in Guangzhou (20-30 s - 30 1 ), followed by Beijing and Chongqing (15-25 s -1 ), and smallest in Shanghai (<15 s -1 ).More than half of the OH reactivities are contributed by inorganic species, i.e.CO and NOx.The measured AVOCs contribute to the total OH reactivity from 14% to 26%.In Guangzhou, the contribution and absolute reactivity (6 s -1 compared to 2~3 s -1 in other cities) are highest due to the large contribution from aromatics.The shares of different VOCs groups are comparable with slightly different contribution between alkanes and alkenes (in total make up 40% of the VOC reactivity except in Guangzhou).35 2) Modeled OH concentrations show distinct diurnal variation.The OH mean diurnal profile maximum is largest in Beijing and Shanghai (7×10 6 cm -3 ) and decreases to 4×10 6 cm -3 in Chongqing and 2×10 6 cm -3 in Guangzhou.The modeled OH concentrations are correlated with the photolysis frequencies (R 2 > 0.7) with a correlation slope to 2.0×10 11 cm -3 s -1 (3.0×10 11 cm -3 s -1 in Shanghai).The peroxy radicals concentrations are highest in Chongqing with a diurnal maximum of 5×10 8 cm -3 for HO2 and 7×10 8 cm -3 for RO2 due to the relatively high VOCs/NOx ratio.
3) The dominant radical sources are photolysis reactions, including HONO, O3, HCHO and other carbonyl compounds, among which photolysis of HONO and HCHO make up nearly half of the primary sources.The mean diurnal maximum is largest in Beijing (5 ppbv/h), followed by Shanghai (4.6 ppbv/h) and Chongqing (4.3 ppbv/h), and is smallest in Guangzhou (3.2 ppbv/h) 5 due to the later observation period in a year.However, the alkene ozonolysis contributes significantly to the radical sources in Guangzhou (43% of the total primary source for daytime conditions).
4) The daily integrated local ozone production rate is largest in Beijing (136 ppbv) and reduced to 92, 40, 105 ppbv in Shanghai, Guangzhou, and Chongqing, respectively.The measurement site represents city center conditions, where ozone precursors are fresh emitted.With the advection of fresh-emitted air mass, NOx and VOCs undergo efficient photochemical processes 10 producing a large amount of ozone and transmitted to downwind regions.The outflow of Ox is identified in Beijing, Shanghai, and Chongqing during the afternoon and in Guangzhou after sunset.5) Ozone production is in VOC limited in all the cities presented because the LN/Q ratio is greater than 0.5.Besides, we find that the RIR values of AHC are highest in all cities compared to CO, NOx, and isoprene.The speciation shows that alkanes and alkenes are a major contributor to total OH reactivity except in Guangzhou.In the sense of ozone formation potential, aromatics 15 become dominant species due to their large carbon numbers and thus high MIR value.In Guangzhou, aromatics contribute up to 70% of the share of OFPs, followed by Shanghai (55%), Beijing (43%), and Chongqing (42%).In comparison, alkanes are the major contributor to the mixing ratios but had limited impact on the ozone formation and radical chemistry.To avoid the bias in the understanding of photochemistry process, it's more proper to use the concept of OH reactivity.
6) The coexistence of high OH and NO2 concentrations resulted in fast nitric acid production rate.The partitioning between 20 HNO3 and NO3 -is analyzed using a thermal dynamic model (ISORROPIA2).In the presence of abundant ammonium, the photochemical produced HNO3 can partition to aerosol phase efficiently under high RH conditions.The daily integrated nitrate production rates are 64±7, 27±8, 16±4, and 20±4 μg/m 3 per day in Beijing, Shanghai, Guangzhou, and Chongqing, respectively.This study demonstrates that active radical chemistry could contribute to the particle pollution.
Atmos.Chem.Phys.Discuss., https://doi.org/10.5194/acp-2018-959Manuscript under review for journal Atmos.Chem.Phys.Discussion started: 2 October 2018 c Author(s) 2018.CC BY 4.0 License.during the morning hours.However, d(Ox)/dt turn sharply from positive to negative at 12:00 in Shanghai.While positive d(Ox)/dt last until 16:00 in the other three cities.The difference in the d(Ox)/dt results in an early Ox peak (around noon time) in Shanghai.

Figure 1 .
Figure 1.The location of the four measurement sites in Chinese megacities.10

Figure 2 .
Figure 2. Mean diurnal variation of measured temperature, CO, O3, j(O 1 D), NOx and Anthropogenic Volatile Organic Compounds (AHC) in four field studies.Ox is denoted in the same panel as O3 with dashed lines.

Figure 3 .
Figure 3. Mean diurnal profile of contributions from all measured species for OH reactivity in Beijing, Shanghai, Guangzhou, and 5

Figure 4 .
Figure 4. Contributions of different atmospheric constituents to OH reactivity in Beijing, Shanghai, Guangzhou, and Chongqing.

Figure 5 . 5
Figure 5.The group compositions (mixing ratios) in percentages for VOCs as well as their shares in OFPs and OH reactivity for four cities.5

Figure 6 .
Figure 6.Mean diurnal profiles of modeled OH, HO2, RO2 concentrations in four measurement sites.The vertical bars denote the variability of model calculations.

Figure 7 .
Figure 7. Hourly averaged primary sources and sinks of ROx radicals derived from model calculations in four measurement sites.5

Figure 8 . 5 Figure 9 .
Figure 8.Comparison of OH-HO2-RO2 radical budget in four cities for daytime conditions (06:00-18:00).The numbers are sorted in the order of Beijing, Shanghai, Guangzhou, and Chongqing from left to right.Blue boxes denote radical primary sources, black boxes denote radical termination, red boxes denote radical propagation, and yellow boxes denote equilibrium between radicals and reservoir species.

Figure 10 .Figure 11 .
Figure 10.The RIR analysis for NOx, AHC, CO and NHC at four sites.