Impact of VOCs on the formation of ozone in a central China city during severe pollution 1 periods 2

Abstract. In this work, a five-month sampling campaign was conducted for volatile organic compounds (VOCs) for the first time in Zhengzhou City, Henan province, China, where ozone (O3) pollution has shown an increasing trend in recent years. Fifty-seven VOCs defined by the Photochemical Assessment Monitoring Stations (PAMS) were sampled using canister method. Meanwhile, other O3 precursor gases were monitored online at four different sites between the time period of May–September, 2017. The results indicated that the average mixing ratio of ΣPAMS (31.57 ± 23.35 ppbv) in Zhengzhou was lower than the other megacities in China, while alkyne was in a significantly higher proportion. The abundances, compositions and ratios of PAMS showed strong spatial and temporal variations. Alkenes were the largest contributors to the ozone formation potential (OFP). On clear days, higher O3 levels were often accompanied with high ΣPAMS/NOx ratio at each site, demonstrating that the VOCs were more sensitive during the O3 formation period in Zhengzhou. Furthermore, source apportionmemt was conducted with Positive Matrix Factorization (PMF) model, and it was found that the use of compressed natural gas (CNG), the evaporation of gasoline and the exhaust from vehicles were the important sources for ambient VOCs at all four sites. Besides, the meterological conditions and long-range transport from other surrounding provinces also had an impact on the air quality determined using the cluster analysis. It is worth mentioning that the reduction in VOCs' emissions is necessary to suppress the O3 pollution.


Non-linearity relationships between the ambient VOCs, nitrogen oxide (NOx) and O3 production indicate that the reduction in tropospheric O3 is complex (Lin et al., 1998;Hidy and Blanchard, 2015;Li et al., 2018).During the time period of 1960 -2010, the concentrations of VOCs and NOx have decreased at the rate of 7.3% yr -1 and 2.6% yr -1 , respectively, due to which, the O3 extrema declined in southern California (Pollack et al., 2013).Evaluations on the production efficiency of O3 suggested the increase in responsiveness of O3 to NOx in southeastern U.S, and therefore, the changes in anthropogenic emissions of VOCs should be taken into consideration (Blanchard and Hidy, 2017).Many modeling and field studies proved that photochemical production of O3 in several regions in China, such as Guangzhou, Shanghai and Beijing, were sensitive to VOCs (Shao et al., 2009;Gao et al., 2017;Ou et al., 2016), and the sensitivity regime was always in flux (Luecken et al., 2018).It is predicted that the formation regime of O3 will be more sensitive to VOC emissions in China by 2030 (Jin and Holloway, 2015), On the other hand, the International Agency for Research on Cancer (IARC, 2017) has identified several VOCs, such as benzene and 1,3-butadiene as the first class carcinogens (Hoque et al., 2008;Iannuzzi et al., 2010).
In recent years, investigations involving the source apportionment of VOCs, acquirement of emission profiles and interpretation of the seasonal variations in China were mainly concentrated in Yangtze River Delta (YRD), Pearl River Delta (PRD) and Beijing-Tianjin region (BJT) (An et al., 2014;Wang et al., 2014;Chen et al., 2014;Liu et al., 2016a;Guo et al., 2017).With the nationwide deterioration of air quality, studies in less developed or developing regions, such as southwestern and northwestern China, are scarce.Contributions from the burning of biomass and high abundances of toxics and reactive compounds (such as, benzene) are prominent in these areas (Li et al., 2014;Li et al., 2017a).Consequently, differences on the structure and reactivity of VOCs, which render them different abilities to influence the production of ozone, and reactivity based on the air quality control measures, have developed (Capps et al., 2012).Additionally, U.S. Environmental Protection Agency (EPA) (2005) has encouraged states to take this approach while developing the Atmos.Chem.Phys.Discuss., https://doi.org/10.5194/acp-2018-397Manuscript under review for journal Atmos.Chem.Phys.Discussion started: 10 August 2018 c Author(s) 2018.CC BY 4.0 License.
Ozone State Implementation Plans (OSIP).Aromatics and alkenes were responsible for most of the weighted reactivity of VOCs (59.4% and 25.8%, respectively) in PRD (Ou et al., 2015).With the changes in energy infrastructure, industrial construction and meteorological conditions (Wang et al., 2015;Shao et al., 2011), major emission sources of active compounds vary accordingly.For example, in less developed cities, such as Heilongjiang and Anhui, the combustion of biomass had the highest contribution to O3 formation (40% and 36%, respectively), while in developed regions, such as Shanghai, Beijing and Zhejiang, the solvent use became more important with time (>20%) (Wu and Xie, 2017).Identifying the local chemical species play vital role in O3 formation and provides useful information for policy-makers to make strategies to alleviate O3 pollution.
Zhengzhou is an important developing city in the mid-west of Huanghe-Huaihe river flood plain in China.As the capital city of Henan province, it is densely populated with more than seven million residents in 2010 (Geng et al., 2013).With the rapid growth of industrial activities, vehicle emissions and fuel combustion, air quality in Zhengzhou has notoriously deteriorated.Zhengzhou is categorized as one of the most polluted cities due to the reason that on 65% of days in a year, its air quality exceeds the allowable limits set by Air Quality Guideline in 2013.The city has the third worst air quality among 74 nationwide cities in the first half year of 2015.Particularly, O3 was the most critical pollutant in Zhengzhou in 65 -85 days of summer (Shen et al., 2017;Gong et al., 2017).Over 50% of the days in a year, the mixing ratio of O3 exceeded the Grade I standard of daily maximum average 8-hour (DMA8) (100 µg m -3 ) in Henan, while the estimated mortality of chronic obstructive pulmonary disease (COPD), caused by O3, was also higher compared to other provinces in China (Liu et al., 2018).As the major precursors of O3, the study on VOCs is of significance.However, for the city of Zhengzhou (China), the related research is non-existent in literature.In this work, a comprehensive sampling campaign aiming at VOCs has been conducted at four monitoring stations during the time period of May -September 2017.The spatial and temporal variations in VOCs in Zhengzhou were determined.The contributions of major emission sources were quantified, and the relationship between O3-VOCs-NOx was discussed in detail.The results and implications from this study can provide useful guidance for policy-makers to alleviate ozone pollution in Zhengzhou, China.

Sampling site
According to the population density and industrial facilities, and in combination with the prevailing southeastern wind, four sites have been selected for the sampling campaign.These sites are Jingkai community (JK; 113.73°E, 34.72°N), municipal environmental monitoring station (MEM; 113.61°E, 34.75°N), Yinhang school (YH; 113.68°E, 34.80°N) and Gongshui company (GS; 113.57°E, 34.81°N), which are located at the southeastern, southwestern, northeastern and northwestern of urban Zhengzhou, respectively (Fig. 1).There was a main airport highway and heavy-traffic ring roads approximately 500 m west of JK.Furthermore, the site was at a distance of 2 km from an industrial area, which involved packaging and printing plants, and material distribution factories.It is noteworthy that there were three coal power plants in the urban area of Zhengzhou city.One of the power plants with the highest production was 1.6 km northwest of MEM.Both the MEM and YH included a mix of commercial and condensed residential areas, whereas the apartments around YH were more aged.In addition, GS was surrounded by several manufacturing plants, including pharmaceuticals, materials, foods and machineries.
Ten sampling days with the rainfall record (ca.0 mm) were chosen in every month during the period of May -September, 2017.Pre-evacuated stainless-steel canisters (Entech Instrument, Inc., Simi Valley, CA, USA), each having the volume of 3.2 L, were used to instantaneously collect VOCs.Two samples, one collected at 07:00 and the other at 14:00, were obtained on every sampling day.There were a total of 400 valid samples collected in this study.The chemical analysis was accomplished within two weeks after the collection of samples.Real-time data for trace gases, including SO2, CO, NO2 and O3, and synchronous meteorological data, such as temperature (T), relative humidity (RH), wind direction (WD) and wind speed (WS), were recorded at each air monitoring station.

Chemical Analysis
In this study, the measurement of VOCs was based on Compendium Method TO-15, which was established by U.S. EPA.Air in the canister was concentrated using liquid-nitrogen at -160 °C in a cryogenic pre-concentrator (7100A, Entech Instrument, Inc.).Both the CO2 and H2O were removed from the transfer line.The air was then thermally desorbed at 120 °C and transferred for analysis in a gas chromatography (GC, 7890A, Agilent Technologies, Santa Clara, CA, USA) coupled with mass spectrometric detector (MSD)/flame ionization detector (FID) (5977E, Agilent Technology).Dual columns and detectors were applied for the simultaneous analysis of C2 -C11 hydrocarbons.A PLOT column with the length of 15 m, internal diameter of 0.32 mm and film thickness of 3.0 µm was connected to the FID for detection of C2 -C5 NMHCs, whereas C5 -C10 NMHCs, oxygenated VOCs (OVOCs) and halocarbons were separated using a DB-624 (30 m × 0.25 mm inner diameter × 3.0 µm film thickness), which was connected to the MSD.The oven temperature of GC was programed to the initial temperature of 37 °C for 5 min, which was increased at the rate of 5 °C/min to 120 °C, and maintained at this temperature for 5 min.It was then ramped at the rate of 6 °C/min to the final temperature of 180 °C, where it was maintained for 5 min.The total run time was 41.6 minutes.The MSD was operated with the electron ionization (EI) impact (70 eV), where the ions were detected in the selected ion monitoring (SIM) mode.Based on the retention time and mass spectra involving accurate standard, the targeted VOCs were identified, and then, quantified using multi-point calibration curve.Three standard gases for TO-15 and PAMS (1 ppm; Spectra Gases Inc, NJ, USA) were used.A total of 101 compounds, including 28 alkanes, 11 alkenes, 17 aromatics, 34 halocarbons, 9 OVOCs, acetylene and CS2 were quantified.This study discussed only 57 PAMS due to their critical role in O3 pollution.Detailed information on the PAMS involved in this study and their corresponding linearity of calibration (R 2 ), measurement relative standard deviation (RSD), method detection limit (MDL), maximum increment reactivity (MIR, carter, 2010) are presented in Table 1.

2.3
Quality control and quality assurance (QC/QA) QC/QA was thoroughly executed to ensure the quality of research and obtain trustworthy experimental data.In order to alleviate contamination from previous samples, canisters had to be repeatedly cleaned using humidified zero air before sampling.In order to inspect the cleanliness and vacuum in canisters, the pre-cleaned canisters, after storing for 24h, were analyzed using GC-MS according to the same analytical procedures used to analyze the field samples.The canisters without any contamination were used for sampling.No obvious emission source was present during the sampling events.Daily calibration was performed using 2 ppbv standard mixtures to ensure the consistency and sensitivity of GC-MSD/FID system.The deviation in standard values was within ±20%.Abnormal QA/QC data with extremely high or low responses was recalibrated until the deviation was within the acceptable range.

Positive matrix factorization (PMF)
U.S. EPA PMF 5.0 was used for source apportionment (Liu et al., 2008b;Lau et al., 2010;Abeleira et al., 2017;Xue et al., 2017).Paatero and Tapper (1994) and Paatero (1997) thoroughly illustrated the rationale of PMF.Briefly speaking, PMF is a diagnostic method involving multivariate analysis, and involves decomposing the integrate sample data into two matrices, namely the source profiles and the source contributions.The method takes advantage of other observation data, such as wind direction and speed.The analyst could get messages on major emission sources, which contribute to ambient VOCs.Detailed information on the application of PMF can be referred to the publications cited above and to the PMF 5.0 user manual (U.S. EPA,

2014).
Based on the chemical mass balance between the input concentrations and the chemical profiles, PMF regarded the ambient data xij, namely the concentration of j th constituent in i th sample, as the gross values contributed by p sources according to Eq. (1).
where gik stands for the contribution of k th factor in the i th sample , while fkj is the load of j th compound in the k th source.Furthermore, eij is the relevant residual.
In order to avoid negative source contributions, penalty function was adopted for constraints.
Each data point can be individually weighed in the model, while the samples with lots of missing values were excluded.

  
where uij stands for the uncertainty of the j th compound in i th sample.
In the PMF model, uncertainty is a function to evaluate the deviations in sampling and analysis procedure (Paatero, 2007).The uncertainty (U) can be calculated based on Eq. (3) (Polissar et al., 1998).
where EF signifies the error fraction that equals 100 times of the percentage uncertainty.
3 Results and discussion

Mixing ratios and meteorological variations
The total concentrations of PAMS (ΣPAMS) were different for all four sites (Table 2), and the ranks were inconsistent between each sampling month.In May 2017, the highest ΣPAMS was reported at JK (37.65±22.58ppbv), followed by GS (31.73±18.70ppbv), YH (30.05±16.43ppbv) and MEM (29.05±15.34ppbv), while the ΣPAMS values for the month of June, July, August and September were found to be in the order of: GS>JK>MEM>YH, MEM>GS>JK>YH, YH>MEM>JK>GS, and MEM> YH > GS >JK, respectively.Beside the emission sources (to be discussed in Section 3.2), the meteorological conditions also contributed to the variations.For instance, the prevailing wind in May was northwestern at MEM, GS and YH, while the southwestern wind was dominant at JK.The transportation of air pollutants from urban center and industrial plants meant that the highest level of ΣPAMS was seen at JK.In June 2017, the prevailing wind vector changed to southeastern at MEM and GS.The average wind speed at GS (0.74±0.33 m s -1 ) was significantly lower than that at MEM (1.84±0.94ms conditions at GS. Furthermore, the air pollutants emitted from MEM were more accumulated at GS, which was due to higher ΣPAMS at GS (39.29±25.37ppbv) than that at MEM (30.28±12.77ppbv).
It should also be noted that, when ΣPAMS at JK was higher than that of GS, the levels at YH were higher than those of MEM, and vice versa.Based on their geographic locations, both the GS and MEM were located in the western part of Zhengzhou city, while JK and YH were situated in the eastern side of the city.Except for the discriminations between the pollution sources at every site, the above phenomenon might be a result of topographical effect.
C2 -C5 alkanes, acetylene, ethylene, toluene and benzene were the most abundant PAMS representing the general consistency of pollution sources in the region.The mixing ratios of toluene varied within a wide range at each site.Toluene is an important aromatic and has a variety of emission sources (Wang et al., 2014;Barletta et al., 2005), among which, the vehicle exhaust and solvent usage are the major contributors.Due to variable origins and emission strengths, few compounds exhibited wide ranges in certain months.The mixing ratios of cyclohexane, isopentane, toluene, and isobutane had large deviations at MEM in June, which were attributed to the increase in liquefied petroleum gas (LPG) and gasoline evaporation due to the increase in ambient temperatures (Li et al., 2017a;Liu et al., 2008a).At YH and GS, the largest deviations were seen in the month of September.The mixing ratios of fossil fuel combustion markers, such as acetylene (5.08±8.36ppbv), toluene (4.37±5.28ppbv), and m,p-xylene (1.09±1.62 ppbv) were varied significantly at YH (Zheng et al., 2017).The variations in propane (10.77±17.73ppbv), n-butane (7.55±13.79ppbv), isobutane (5.46±10.76ppbv) and isopentane (5.25±8.76ppbv) at GS were reasonably significant due to the evaporation of gasoline and the use of LPG (Li et al., 2017a;Liu et al., 2008a).Both cases demonstrate the widespread combustion sources and fuel evaporation in Zhengzhou city.
The highest composition of alkane was observed at MEM due to the stronger contributions of ethane, iso-pentane and other branched alkanes (such as, the ones having carbon number of 7 -8).
detected at all sites (Table 3).These chemicals had the contribution of >60% for ΣPAMS,  Wang et al. (2017b) reported that these compounds could originate from the cooking process, which is consistent with the results reported here, as there was a shopping mall with a number of canteens around 200 m northeast to the sampling station.In addition, many restaurants were also scattered around.
In comparison, the aromatic compounds showed lower compositions in the afternoons (for example, around 14:00 LT) than the mornings (for example, around 07:00 LT) (Fig. 2).Increases in alkene compositions were mainly ascribed to higher contributions of isoprene, which was emitted from biogenic sources at higher temperature and solar irradiation.Isoprene's contribution to ΣPAMS value at JK at was 12% at 14:00 LT (Fig. 2), which was much higher than the other three sites.
These results are in accordance with its larger vegetation coverage in the surrounding areas.
The average ΣPAMS values in Zhengzhou city was significantly lower than those in Beijing (65.55 ppbv), Hangzhou (55.9 ppbv), Guangzhou (47.3 ppbv) and Nanjing (43.5 ppbv), though they were higher than that in Wuhan (23.3±0.5ppbv)(Guo et al., 2012;Li and Wang, 2012;An et al., 2014;Lyu et al., 2016;Raysoni et al., 2017).Furthermore, population density, industrial activity, fuel composition, local stringent regulations for environmental protection, terrain, and weather contributed to their abundances.Alkanes were generally the largest contributor in most areas (Fig. 2), while the compositions of aromatics were relatively lower in Zhengzhou city.It is well known that aromatics mainly originate from solvent usage and vehicle exhaust.It is also worth mentioning that Zhengzhou has less manufacturer involved in shoemaking and shipbuilding than Guangzhou and Nanjing.The registered number of vehicles was also lower than that in Beijing.On the other hand, higher composition of alkyne was observed in Zhengzhou.Alkyne typically originates from combustion sources.Henan is the largest agricultural province in China.As the sampling duration covered the crop harvest in June, the residents often used crop residues as the biofuel for their subsistence, thus resulting in higher alkyne compositions.

Temporal variations
The time series of mixing ratios of CO, SO2, NO, NO2, O3 and PAMS at every site are shown in Fig. 3.Both the T and RH are also included in Fig. 3.The results showed that there is a distinctive temporal characteristic, during which, lower levels of all air pollutants were observed in July and August (mid-summer) than in other months.These results were similar to those obtained for other urban areas (Li and Wang, 2012;Cheng et al., 1997;Na et al., 2001).Changes in boundary layer depth, human activities, and abundance of hydroxyl radicals (•OH) were the potential causes for the phenomenon.Pal et al.( 2012) reported that the height of atmospheric boundary layer was positively correlated with temperature.In the current study, the highest ambient temperature was observed in July, while the occurrence of precipitation was more frequent in most areas of China during this period.Furthermore, stronger dispersion efficiently offered a wash-out effect for diluting the air pollutants to lower levels in ambient air.Additionally, a series of effective local policies, such as prohibition of painting and coating in open air and limitations on fuel supply between 10:00 -17:00 LT during hot summer days assisted in suppressing the VOCs' emissions.
Meanwhile, many organizations, such as schools, institutes and scattered private workshops, were closed due to summer vacations.Some large-scale industries also stopped manufacturing processes for two weeks during this period.Due to this, the anthropogenic emissions reduced, which in turn resulted in a decrease in VOCs, SO2, and NOx emissions.The reduction of precursor levels and unfavorable photochemical conditions (such as, higher RH and WS) resulted in the O3 levels to be constantly below the standard.
As demonstrated in Fig. 3, the observed ΣPAMS values at 07:00 LT were often higher than those at 14:00 LT.LPG consumptions resulted in an increase in mixing ratio of propane, isobutane and n-butane (Huang et al., 2015b), which was found consistent with the observations at this site.
Acetylene is a stable chemical compound that mainly originates from the combustion of biomass/fossil fuel.On the morning of 5 th September (highlighted in pink in Fig. 3), alkyne was found in extremely high concentrations (14.65 -39.42 ppbv).Its mixing ratio in most of the urban areas was <10 ppbv (Louie et al., 2013;Duan et al., 2008;Guo et al., 2012).It was learnt that the 5 th September is a festival day for the people, who worship their ancestors.A large number of incenses and offerings, made up of wood and paper, were burnt during the festival, resulting in an increase in the mixing ratio of acetylene all over the Zhengzhou city.

Spatial variations
Average ΣPAMS were higher at industrially impacted sites of GS (31.66±28.73ppbv) and JK (28.63±22.04ppbv) than those at the other two sites (MEM and YH).Additionally, the air pollutants related to the combustion process, such as SO2 and CO were also in abundance in western area of Zhengzhou (GS and MEM), where stronger emission sources were related to the combustion of biomass or fossil fuel.Meanwhile, the O3 levels did not present the same trend with ΣPAMS among the four sites, indicating that the transformation between O3-NOx-VOCs was more complex than the linear reactions.
In June, the O3 concentration often exceeded the national standard level of 80 ppbv.The average mixing ratio of O3 during daytime (07:00-18:00 LT) at JK, MEM, YH, and GS were 74.87 ±39.55 ppbv, 73.50±40.59ppbv, 73.81±35.69ppbv, and 87.99±46.11ppbv, respectively.The order of these values is consistent with that of the ΣPAMS, in which the higher levels were observed at JK and GS.Combined with the higher solvent usages, highest aromatics levels and the severer O3 pollutions were found at GS as expected.Even though both the ΣPAMS and higher O3 formation potential compounds (such as, alkenes and aromatics) at MEM were a bit higher than those at YH, a reverse order of O3 concentration was observed.Such phenomenon could be attributed to other critical precursors, such as NOx, which are involved in the formation of O3.NO at MEM (8.48± 25.74 ppbv) was significantly higher than that at YH (2.57±3.08 ppbv) during daytime, indicating that the reaction between O3 and NO was more efficient at MEM.
It is well known that O3 pollution is a regional problem.Streets et al. (2007) reported that the contribution of air pollutants, derived from Hebei, was around 35 -60% to the pollution in Beijing, which experienced O3 episodes.Along with the decrease in ambient temperature and light intensity, and an increase in RH, the number of non-attainment O3 days theoretically reduced after June, 2017, while the maximum was observed on August 10 th , 2017 (highlighted in yellow in Fig. 3).VOCs are the critical contributors to the formation of O3.The reductions in ΣPAMS in the afternoons (around 14:00 LT) compared to mornings (around 07:00 LT) may have originated from the uptake by O3 formation (Fig. 4).Furthermore, the highest VOCs concentration was observed at YH, which was due to its downwind position to the other three sites between 08:00 -15:00 LT on August 10 th (Fig. 4).Additionally, the transmission of pollutants led to an abnormally high O3 level at YH.
Overall, when the synoptic conditions, such as T, RH and WS were favorable, the O3 level was accompanied by the increase in VOCs concentrations, which potentially demonstrated that the O3 formation was more sensitive to the abundance of VOCs in Zhengzhou city.

Ratios of specific compounds
Ratios of specific VOCs are useful indicators to identify the emission sources (Raysoni et al., 2017;Liu et al., 2015;Li et al., 2017c;Huang et al., 2017;Ho et al., 2009).In order to characterize the differences in the contribution of various sources at each site, two ratios of toluene/benzene and i-pentane/n-pentane were discussed.The ratio of toluene to benzene (T/B) is an efficient tool to differentiate between pollution sources, such as vehicle exhaust, coal combustion and solvent use (Sun et al., 2018).As shown in Fig. 5, the T/B ratio showed variation during days and months.In addition, the correlations between toluene and benzene were different throughout the sampling period, indicating that their emission sources were always changing.The atmospheric lifetimes for toluene and benzene are 1.9 days and 9.4 days, respectively, while the abundance of •OH was assumed to be 10 6 rad cm -3 (Monod et al., 2001).Therefore, toluene was consumed more rapidly than benzene during the photochemical reaction.Due to this reason, the T/B ratios at 14:00 LT were almost lower than those at 07:00 LT.However, in September, higher T/B ratios at 14:00 LT were observed more frequently at MEM and YH than the other station.Meanwhile, the values were increased regionally, which was in accordance with the weaker photochemical reactions in late summer.
The R 2 value for the correlation between benzene and toluene was better in May (0.73 -0.84) than during other months for all sites, and the average ratio in this month varied within the range of 1.81 -3.36 for all four sites.Both the tunnel studies and roadside researches indicated that T/B ratio varied within the range of 1 -2 when the atmosphere was heavily impacted by vehicle emissions (Gentner et al., 2013;Tang et al., 2007;Huang et al., 2015b;Wang et al., 2002).When the ratio was less than 0.6, it may be due to other sources, such as coal combustion and biomass burning (Tsai et al., 2003;Akagi et al., 2011).The industrial activity becomes more important when the value of T/B ratio is higher than 3 (Zhang et al., 2015).The results suggest that evident impacts were observed from vehicle emissions on the T/B ratio during the sampling period.Except for September, the average T/B ratio at JK lied within the range of 1.47 -2.72, which was due to the vehicle emissions.
Due to less interruption from photochemical reaction, the impact from industrial activities, which were mainly related to industrial clusters composed of automotive-related workshops and printing shops, and located approximately 2 km southwest of the observation site became important and resulted in higher T/B value (3.89) in September at JK.It is apparent that the T/B ratio at MEM was on the low side during the period of May -August (0.96 -1.68), highlighting the impact of coal-fired thermal power plant.After excluding the abnormal values, the average ratio at GS (0.44 -1.17) exhibited clear characteristics related to the combustion of coal/biomass during the period of June -August.This was related to the fact that the region was located on the edge of Zhengzhou.
The impacts from biomass burning would be heavier than in other areas during the harvesting season.On the other hand, the coal fired power plant next to SZ was located at the southeast of GS with the linear distance of 5 km between the two points.Considering that the prevalent wind direction was southeastern during the period of June -August, the air pollutants emitted from the power plant could be easily transported to this region.Overall, the emission sources related to industrial activities were more evident in September.It is likely that the traffic emissions at JK and i-Pentane has similar reactivity to n-pentane (Jobson et al., 1998).The ratio between these two VOCs is useful in differentiating the potential sources, such as natural gas drilling and vehicle emissions.In this study, i-pentane and n-pentane were highly correlated (R 2 =0.78 -0.89) throughout the whole sampling campaign (Fig. 6), indicating stable sources for these two compounds.In the area, heavily impacted by natural gas drilling, the ratios lied within the range of 0.82 -0.89 (Gilman et al., 2013;Abeleira et al., 2017).Higher values were often reported for automobiles, such as 2.2 -3.8 for vehicle emissions, 1.8 -4.6 for fuel evaporation, and 1.5 -3.0 for gasoline (Russo et al., 2010;McGaughey et al., 2004;Jobson et al., 2004;Wang et al., 2013b).Low values within the range of 0.56 -0.80 were found for coal combustion.The highest ratio was observed at JK (2.73), which was comparable to the value of 2.93 reported in a Pearl River Tunnel (Liu et al., 2008a), thus indicating strong impacts from traffic-related sources.The average ratio at MEM (2.02) was inconsistent with the characteristics of coal combustion.Yan et al. (2017) reported that the ratio for areas, which include several coal-fired thermal power plants, was approximately 0.55.The variation could be explained by factors, such as production scales, fuel compositions, sampling time and terrain.Additionally, it is remarkable that MEM was surrounded by a main road with four traffic lanes.The distance between the nearest traffic light and the sampling site was just 200 m.Frequent idling may cover up the contribution from coal combustion, thus overshadowing the effect of traffic emissions.The average ratio at GS (1.85) was close to that at YH (1.68).This was due to the similarity between the traffic related sources (such as, gasoline and fuel evaporation).The results showed that the vehicle emissions were an important source for VOCs in Zhengzhou city.

Reactive chemicals
Due to large differences in the reactivity of individual species, the contribution of each identify local active species that contribute the most to O3 pollution (Huang et al., 2017).OFP is calculated using Eq. ( 4).

OFP=Ci×MIR (4)
where Ci represents the concentration level of i th species, while MIR is a constant taken from Carter (2010) (Table 1).
Different from the results reported by (Wu and Xie, 2017), based on the emission inventory for North China Plain (NCP), YRD and PRD, where the largest contributor was the aromatics, followed by alkenes, alkenes were the most significant contributors (55.91±14.17%) to the sum of OFP in Zhengzhou city.The biggest contributors included ethylene, isoprene, m,p-xylene, toluene, propylene, acetylene, n-butane, i-pentane and propane, and their contributions lied within the range of 74.64 -79.90% of the sum of OFP (Table 5).The composition of acetylene was higher compared to other areas (Li and Wang, 2012;Jia et al., 2016), demonstrating that it is necessary to conduct emission controls on sources related to combustion in Zhengzhou city.
It is well known that the ratio between VOCs and NOx is an important index to analyze variable O3 concentrations (Toro et al., 2006;Simon et al., 2015;Geng et al., 2009;Jin and Holloway, 2015;Geddes et al., 2009;Li et al., 2017b).Zhengzhou suffered from the severest O3 pollution in June, 2017.The relationships between OFP of each organic group, ΣPAMS, and the concentrations of NOx and O3, as well as the corresponding meteorological conditions, are shown in Fig. 7.At every site at 07:00 LT, WS was generally lower than that at 14:00 LT, indicating weaker vertical distribution.With the lower RH and higher T and OFP (88.13±30.32ppbv) values, lower O3 was unexpectedly seen at YH than MEM on sunny days, which was attributed to lower ΣPAMS/NOx ratios at YH (highlighted in pink in Fig. 6).The total OFP was higher at JK than at other sites, although the actual O3 levels at JK were very close to those at MEM and even lower than those at GS.A lower WS at GS (0.65±0.26 m s -1 ) promoted the pollutant accumalation The ratio of ΣPAMS/NOx was more scattered at JK and MEM than at YH and GS along with the variable WS, thus demonstrating significant impacts from metelogical conditions.During clear days, higher O3

Source apportionment
In order to select appropriate chemical species for PMF model, following principles were used.
Species with mixing ratios usually below MDL were eliminated.Except for the source markers, species with high reactivity should be excluded (Shao et al., 2016;Guo et al., 2011).Finally, 28 VOC species and NO2 were chosen for the source analysis.Q(robust), Q(true) and Q(true)/Q(exp) are presented in Table 6.Based on the model results, seven factors were defined at MEM, YH, and GS, while eight factors were defined at JK (Fig. 8).
Source profiles showed that the sites had similar regional characteristic.There was a strong common source (factor 1) for C2 -C5 n-alkanes and certain amounts of C2 -C4 alkenes, benzene, toluene, acetylene and NO2.It is reported that i-Pentane and aromatics (such as, benzene, toluene, ethylbenzene and m/p-xylene) usually originated from gasoline evaporation, while isobutane and n-butane were emitted from LPG/CNG usage (Li et al., 2017a;Liu et al., 2008a).Considering the wide consumption of CNG for residential cooking in Zhengzhou, this factor was defined as CNG+gasoline evaporation.
The second factor was characterized based upon significant loading of toluene, ethylbenzene, m/p-xylene, and 1,2,4-trimethylbenzene.According to previous studies (Yuan et al., 2010;Wang et al., 2014), toluene and C8 -C9 aromatics were the major VOCs emitted from paint applications.
Therefore, Source 2 was assigned as the solvent use.This is consistent with the real situation that there were widespread road paving and building constructions in Zhengzhou city.In addition, car decoration, printing, and furniture manufacturing, which are associated with the use of adhesives, were also included in this source category.
The third source was associated with the significant portion of long chain alkanes (C7 -C9) and high amounts of toluene, which are tracers for diesel evaporation (Liu et al., 2008a).Therefore, Source 3 was identified as the diesel vapor.Beside on-road vehicles, many non-road mobile machineries used for building or railway station constructions were correlated with this source.Source 4 was dominated by trans-2-butene and 1-butene.These species were mainly produced in coal combustion (Liu et al., 2008a;Yan et al., 2016).Liu et al. (2016b) reported that 1-butene was a critical VOC generated during the smoldering process of coal combustion.This factor was thus categorized as the coal combustion.
The fifth factor shows a dominant loading of cyclohexane, methylcyclohexane, styrene, 2-methylhexane and 3-methylhexane.The first three VOCs were identified as the components heavily impacted by petrochemical industries (Jobson et al., 2004).Therefore, this source was defined as the petrochemical.
Source 6 was composed of ethylene, propylene, benzene, toluene, 3-methylpentane, n-heptane, acetylene and NO2.In urban areas, vehicle emissions are a major source for ethylene and propylene (Wang et al., 2017a).Fossil fuel combustion produces large amounts of acetylene (Ho et al., 2009), while NO2 is a typical tracer for vehicle exhaust (Huang et al., 2015a).The source was referred to as the vehicle exhaust.
Source 7 was distinguished by extremely high compositions of isoprene, a species mainly produced by vegetation through photosynthesis (Millet et al., 2016).Even though, it can be emitted from traffic-related sources (Yuan et al., 2009), this possibility can be ignored by its poor correlations with other source makers of vehicle exhaust (such as, i-pentane and ethylene).
Therefore, this source was identified as the biogenic emissions.
Factor 8 was only resolved for the JK site.The major compounds present in this source were ethylene, propylene, acetylene and NO2.These compounds mainly originated from on-road vehicles (Wang et al., 2017a).Since there were minor contributions of alkanes and aromatics, they were possibly derived from on-road vehicles under different running conditions.This factor was named as the traffic-related source.
According to the source apportionment results (Fig. 9), CNG usage, gasoline evaporation and vehicle exhaust had considerable portions and were the important sources for ambient VOCs at every site.The smallest contribution from CNG and gasoline evaporations was found at GS (24.99%), which was lower than the values observed at JK (35.06%) and YH (34.28%), and was uneven on the regional scale, with the largest portion observed at MEM (11.56%) and comparable values found for JK (3.44%), YH (6.21%) and GS (5.56%).These results are in accordance with the fact the distance between the petrochemical plants and the sampling site of MEM was the shortest (ca. 2 km).The weighted percentages attributed to the solvent use were similar among the four sites, with the highest value of 12.41% at MEM, which was consistent with the fact that there was less abundance of aromatics in Zhengzhou city.The emission source related to diesel fuel was more evident at JK (11.76%).The area was rife with logistics companies and frequently suffered from disturbance of heavy-duty cargo vehicles powered by diesel fuel.Coal combustion impacted the ambient air heavily at MEM (17.84%) and GS (13.50%), which were only 2 km and 5 km away from the thermal power plants, respectively.Overall, the traffic-related source was the most important emission source for VOCs in Zhengzhou, which is in accordance with the results obtained for other urban areas (Yuan et al., 2013b;Lyu et al., 2016;Lau et al., 2010).

Long-range transport
Air transported from surrounding areas had impacts on the air quality at the study site (Langford et al., 2010;Sun et al., 2018) in the concentration or composition of VOCs.As illustrated in Fig. 2, the average concentration level of ΣPAMS in July sharply decreased to 15.91±6.54ppbv.This was accompanied by the weighted percentage of aromatics, which dropped to 10.30±4.23%.It is well known that alkenes and aromatics have high photochemical reactivity.The T/B ratio also showed the lowest level (1.15±0.99)around this period, illustrating that the air plumes had longer lifetime.Both of these evidenced that Zhengzhou was impacted by air pollutants, which originated from Hubei province.
According to previous studies in Wuhan, the capital city of Hubei province, the average total ΣPAMS (23.3±0.6 ppbv) (Lyu et al., 2016) was much lower than the average level in Zhengzhou (31.57±23.35ppbv).The cleaner air mass clusters, originating from Hubei in July, can thus be designated as a dilute flow.

Conclusions
In this study, PAMS were collected at four different sites in Zhengzhou, Henan (China) for the first time.C2 -C5 alkanes, acetylene, ethylene, toluene and benzene were the most abundant VOCs in the region.On the basis of monthly average, the maximum ΣPAMS was observed at GS, which was impacted by various emission sources.In comparison to other Chinese cities, the weighted percentage of aromatics was lower, while higher alkyne level was observed in Zhengzhou city.Due to less anthropogenic emissions and more favorable dispersion conditions, most of the air pollutants had the lowest levels in the mid-summer month of July.Overall, the O3 levels were correlated with ΣPAMS, which potentially demonstrated that the O3 formation was more sensitive to the abundance of VOCs in Zhengzhou.Different from other megacities, alkenes were the biggest contributors to OFP, while acetylene was particularly critical at each site.The CNG usage, gasoline evaporation and vehicle exhaust were the important sources for ambient VOCs.Cleaner air clusters from Hubei occasionally arrived at Zhengzhou.This study provides the first-hand information on the characteristics of VOCs and assists in overcoming the O3 pollution issue in Zhengzhou city, China.toluene-14:00 benzene-14:00 toluene-14:00 benzene&toluene(ppbv) Atmos.Chem.Phys.Discuss., https://doi.org/10.5194/acp-2018-397Manuscript under review for journal Atmos.Chem.Phys.Discussion started: 10 August 2018 c Author(s) 2018.CC BY 4.0 License.
Atmos.Chem.Phys.Discuss., https://doi.org/10.5194/acp-2018-397Manuscript under review for journal Atmos.Chem.Phys.Discussion started: 10 August 2018 c Author(s) 2018.CC BY 4.0 License.combustion sources related to coal or biomass at MEM and GS would account for more contributions to VOCs, while there were no consistent results observed for YH due to the wide span of T/B ratio (1.07 -5.38).
Atmos.Chem.Phys.Discuss., https://doi.org/10.5194/acp-2018-397Manuscript under review for journal Atmos.Chem.Phys.Discussion started: 10 August 2018 c Author(s) 2018.CC BY 4.0 License.probably caused by the heavier traffic at JK and larger consumption of CNG (from the condensed canteens and active household cooking) at YH.The distribution of petrochemical source was . In this work, Hybrid Single-Particle Lagrangian Integrated Trajectory (HYSPLIT) model was used to present the long-range transport effects on the air quality of Zhengzhou city (Fig. 10).Meteorological conditions are important factors that impact both the compositions and levels of VOCs.Clusters arriving at Zhengzhou in May, 2017 demonstrated longer paths, and included six clusters in total.The largest one (27.2%)was originated from Yinchuan city, which is a central city in northwest China.The cluster passed over several non-capital cities (such as, Yanan, Yuncheng and Luoyang) in Shanxi and Sichuan provinces.Such long-range transportation of pollutants might have less impact on the air quality of Zhengzhou, which agrees well with the comparable levels and similar compositions of VOCs during the period of May -June.In the months of June, August and September, approximately half of the air currents originated from the areas of Henan province.The air current originating from Hubei province took up the largest portion (ca.88.68%) of clusters in July, and potentially led to a significant variation Atmos.Chem.Phys.Discuss., https://doi.org/10.5194/acp-2018-397Manuscript under review for journal Atmos.Chem.Phys.Discussion started: 10 August 2018 c Author(s) 2018.CC BY 4.0 License.

Fig 5 .
Fig 5. Time series of benzene and toluene and the correlation analysis between these two
The accumulation of pollutants during night-time and the temperature inversion in the morning were the most reasonable explanations for this phenomenon.Stronger photochemical reaction during noon-time led to the reduction in atmospheric VOCs.It should be noted that pronounced ΣPAMS were occasionally observed at MEM and GS, which were potentially ascribed to sharp changes in local emissions and meteorological conditions.At MEM, the distinctive increment was always accompanied by obvious uplift of alkanes or aromatics.Combined with the changes in other parameters, such as levels of NO, NO2, SO2, CO, RH and T, it was concluded that the abnormally high ΣPAMS values at MEM were mainly caused by enhanced combustion sources.Due to the disturbance from solvent use for building renovation, extremely high levels of aromatics were observed at GS in June.In addition, there were potential impacts from gas-fueled power Atmos.Chem.Phys.Discuss., https://doi.org/10.5194/acp-2018-397Manuscript under review for journal Atmos.Chem.Phys.Discussion started: 10 August 2018 c Author(s) 2018.CC BY 4.0 License.plants, which were located about 1 km northwest of the site.Compressed natural gas (CNG) or
a m-Xylene and p-Xylene are co-eluted in the chromatographic separation.