Important role of ammonia on haze formation in Shanghai

A haze episode occurred on 12 September 2009 in Shanghai, when the mass ratio of PM1.0/PM2.5 (PM: particulate matter) reached 0.8. A similar variation of hygroscopic growth factor distribution was observed for Aitken mode particles and accumulation mode particles, implying that the enhancement of fine particles was caused by local atmospheric processing. The hygroscopicity measurements in combination with chemical analysis provided strong support for the significant contribution of (NH4)2SO4 and NH4NO3 to the haze episode. The molar ratio of [NH4 + ]/([NO3 − ] + 2[SO42 − ]) rose up to 0.96, coincident with the large increase in NH3 concentration, suggesting that the available NH3 played a vital role in the enhancement of particulate sulfate and nitrate during the haze episode.


Introduction
In the past 20 years, China has undergone rapid industrialization and urbanization. Due to the economic development dominantly powered by coal-based energy, China has been the biggest emitter of SO 2 in the world. Also, in contrast to substantial reductions in some areas of Europe and the USA, NO 2 emission in East Central China has accelerated since 2000 [1]. Geng et al [2] found that both SO 2 and NO x emissions over the Yangtze river delta (YRD) region mostly originated from coal combustion. In 2005, the industrial source accounted for 73% of the 5.13 × 10 5 tons of the total SO 2 emissions in Shanghai [3]. The emission intensity of SO 2 in Shanghai reached 81 tons km −2 , at least six times higher than that of other provinces in the YRD. On the other hand, the emission of NO x from power plants and industry reached 3 × 10 5 tons in Shanghai in 2005, accounting for about 60% of the total emission [4]. The large emissions of SO 2 and NO x caused severe air pollution problems. For example, haze has occurred much more often in the past several years. 3 Authors to whom any correspondence should be addressed. However, the mechanism of haze formation is still not clear. A better understanding of the relation between secondary aerosol formation and haze pollution in Shanghai is key to the control of urban air quality.
Several studies regarding secondary inorganic species in Shanghai have been reported. Wang et al [5] suggested that heterogeneous reactions played a main role in the formation of SO 2− 4 in the entire year whereas NO − 3 was largely from gasphase photochemical reactions in the cold season. The high equivalent ratio of NH + 4 to SO 2− 4 from their results indicated that the aerosol was from neutral to slightly acidic. Fu et al [6] suggested that the transformation of SO 2 and NO 2 contributed much to the high concentrations of secondary aerosols during the heaviest pollution episode in Shanghai. They attributed the air pollution mass to long-range transport from the northern part of the YRD. However, Pathak et al [7] found that the molar ratio of NH + 4 to SO 2− 4 was often below 1.5 in PM 2.5 samples collected in Taicang, which is 44 km to the northwest of Shanghai. They supposed that the high concentration of nitrate in particulate phase was formed by the hydrolysis of N 2 O 5 on the surface of the pre-existing moist and acidic aerosols. In a laboratory study, Tursic et al [8] found that the rate of SO 2 conversion in the presence of NH 3 increased by a factor of about 20 in comparison with that in the absence of NH 3 . It was reported that China had consumed over one-third of the world's chemical nitrogen fertilizer [9,10]. Streets et al [11] estimated that the agriculture sector contributed 88% to the total emissions of NH 3 in China. However, the role of atmospheric NH 3 concentration on the haze formation was barely noticed.
The current work was performed in the framework of Mirage (Megacities Impact on Regional and Global Environment), Shanghai 2009-2012, as organized by the National Center for Atmospheric Research and the Shanghai Meteorological Bureau. Focusing on the formation of ozone and aerosols, the first field campaign was deployed in September 2009. This was the period with the best air quality in a year, because the prevailing wind direction in Shanghai was mainly from southeast to southwest in the summer while northerly advection brought polluted air from Jiangsu province in the winter. As shown in figure 1, the supersite measurement station was located in the Pudong Meteorological Bureau (PD, 121.55 • E, 31.22 • N), while NH 3 concentration was measured at Dongtan beach (DT, 121.92 • E, 31.52 • N), the background site located in the east of Shanghai. The purpose of our work was to investigate the chemical mechanism of haze formation in Shanghai. The hygroscopicity and ion chemistry of fine particles are reported.

Experimental section
The compositions of bulk particles were monitored online with a Monitor for Aerosols and Gases in Ambient Air (MARGA, Model ADI 2080, Metrohm Applikon BV) described previously [12].
Briefly, atmospheric aerosol (with a cutoff of 2.5 µm using a PM 2.5 cyclone impactor) passed through a steam jet aerosol collector where the soluble components were captured and dissolved into the supersaturated steam. Subsequently, the mass concentrations of NH + 4 , Na + , K + , Ca 2+ , Mg 2+ , SO 2− 4 , NO − 3 , and Cl − were analyzed with an ion chromatograph. The aerosol hygroscopicity was measured at 26 • C with a custom-built hygroscopic tandem differential mobility analyzer (HTDMA) described previously [13]. Briefly, the system was operated alternately in DMA and TDMA modes through which the particle size distribution and sizeresolved hygroscopicity of submicrometer ambient aerosols were measured, respectively. The sample stream was dried to approximately 10% relative humidity (RH) and neutralized to charge equilibrium before entering the first DMA (Model 3081L, TSI Inc.). The hygroscopic growth was measured at RH = 20-91% (only results at 68% RH are reported here) and quantitatively expressed as hygroscopic growth factor (GF), defined as relative increase in particle size due to water uptake at a certain RH. The particle size distributions were determined by online DMA inversion, while hygroscopic growth factor distributions were obtained through TDMAinv inversion [14]. The number fraction of any particle mode was obtained by integrating the related segment of the normalized growth factor distribution.
Both the local meteorological data and air quality parameters including the daily average concentrations of ambient particulate matter, NH 3 , O 3 , and NO 2 , were provided by the Shanghai Meteorological Bureau.

Results and discussion
There was no precipitation during our observation. The average temperature and dew point over the observation were 26.1± 2.3 and 20.8± 2.6 • C, respectively. The prevailing wind was from the northeast, so air parcels arriving in Shanghai were mainly from the East China Sea. Figure   Health Organization Air Quality Guidelines. The mass ratio of PM 1.0 /PM 2.5 was 0.80 on 12 September, indicating that PM 1.0 made a significant contribution to the fine particle pollution.
Investigation on the hygroscopicity of ambient aerosols below deliquescence RH of ammonium sulfate may provide some insight into the particle mixing state and particle sources. Figure 3 displays the temporal evolution of hygroscopic growth factor distribution at 68% RH and hourly concentrations of sulfate and nitrate during the haze episode. Aitken mode particles in urban areas mostly originated from traffic emissions and condensational growth of nucleation mode particles. Soot particles from traffic sources were in general nearly hydrophobic while secondary inorganic aerosols from ammonia transformation were hygroscopic in nature. Some particles of diameter 50 nm significantly deliquesced at 68% RH, indicating that gas-to-particle phase conversion and subsequent condensation contributed to the Aitken mode particles significantly. A similar variation of hygroscopic growth mode was observed for particles in the whole size range, indicating that both Aitken mode and accumulation mode particles during the haze episode had undergone the same atmospheric processing. Because Aitken mode particles in urban areas were mostly influenced by local sources [16], it is evident that the fine particles from long-range transport made a minor contribution to the haze formation. The hygroscopic growth factor distribution split into the lessdeliquesced group (GF < 1.1) and more-deliquesced group (GF > 1.1) during the period from 21:00 on 11 September to 19:00 on 12 September (local time). The presence of a more-deliquesced group at 68% RH suggested that NH 4 NO 3 or NH 4 HSO 4 was a significant aerosol constituent, because they were typical components in PM 2.5 with deliquescence RH below 68%. Compared to theoretical calculation according to the hygroscopic growth calculator described previously [17], the average GFs of the more-deliquesced group were about 92% of the theoretical GF of NH 4 NO 3 regardless of the particle size. The number fraction of the more-deliquesced groups displayed a feature of size dependence during the haze episode, with a larger fraction for bigger particles. This feature indicated that the atmospheric process was dominated by the deposition of hygroscopic materials.
As shown in figure 3, the hourly concentrations of sulfate and nitrate in PM 2.5 reached 0.285 and 0.345 µmol m −3 , equivalent to 27.4 and 21.4 µg m −3 , respectively. These results suggested that the fine particle enhancement on 12 September should be attributed to secondary inorganic aerosol formation. The nitrate concentration varied in accordance with the number fraction of the more-deliquesced group, providing strong support for the significant contribution of NH 4 NO 3 indicated by hygroscopicity. The more-deliquesced group disappeared when the molar ratio of sulfate to nitrate (S/N) was above 1.5. These results indicated a negligible contribution of NH 4 HSO 4 to the haze formation. Figure 4 illustrates the correlation between the daily average concentrations of ammonium, sulfate, and nitrate during the whole period. The linear fit to the equivalent concentration of the sum of sulfate and nitrate versus that of ammonium had a square correlation coefficient of 0.998 and a slope of 0.94. The measured slope was approximately 1.0, suggesting the important role of gaseous NH 3 in the formation of particulate sulfate and nitrate.   Figure 5 displays the daily evolution of NH 3 , NO 2 and O 3 during the whole period. The daily average concentrations of NH 3 exhibited the highest concentration on 12 September, in agreement with the increase in PM 2.5 , whereas the concentration of O 3 decreased abruptly. The remarkable increase in NH 3 concentration was attributed to the change in wind direction. In contrast to the prevailing winds from the eastern direction during the observation, air parcels arriving in Shanghai on 12 September were from the northwest and had passed through Jiangsu Province. Although Shanghai is normally an ammonia-deficient city, Jiangsu Province to the north of Shanghai is one of the biggest NH 3 emitters in China [9]. The annual average concentration of NH 3 in Huian (Jiangsu Province) in 2008 reached 14.1 ± 8.6 ppb, corresponding to 41.5 ± 25.3 µg m −3 (NH 4 ) 2 SO 4 or 50.3 ± 30.7 µg m −3 NH 4 NO 3 . Particulate NH 4 NO 3 was formed from the reaction between gaseous NH 3 and HNO 3 through a reversible process: NH 3 (g) + HNO 3 (g) = NH 4 NO 3 (s or aq).
The formation of NH 4 NO 3 depended on the reaction dissociation constant and the product of the partial pressure of NH 3 and HNO 3 . Under urban atmosphere, NH 3 was liable to react with acidic sulfate before HNO 3 could condense onto aerosols as ammonium nitrate. Thus, the formation of NH 4 NO 3 was often suppressed in NH 3 -deficient atmosphere. This was the typical case for our observation. The increase in NH 3 was consistent with the enhancement of particulate sulfate and nitrate, indicating that the ammonia-rich atmosphere further promoted SO 2 conversion and provided a more favorable environment for the formation of NH 4 NO 3 in the particulate phase. The large enhancement of NH 4 NO 3 would result in the rapid depletion of gaseous HNO 3. A number of studies revealed that continental O 3 concentration was correlated with [NO z ] (NO z is mainly composed of HNO 3 , HONO, and PAN) [18,19]. Neuman et al [20] suggested a positive correlation between O 3 and HNO 3 under the lower stratosphere. Therefore, the daily average concentration of O 3 could be regarded as an indicator of HNO 3 variation. The concentration of O 3 significantly decreased on 12 September, further supporting that the formation of NH 4 NO 3 in this haze episode was driven by the high concentration of NH 3 . Our conclusion was supported by Sharma et al [21], who reported that NH 3 played a vital role in the formation of secondary aerosols in Kanpur city.
Due to the wide application of flue-gas desulfurization technology, the annual average concentration of atmospheric SO 2 in Shanghai in 2010 was 52.5% lower than that in 2005. However, the annual average PM 10 loading only decreased 10.2% [15]. A reduction in SO 2 emission would undoubtedly result in the decrease in particle sulfate. However, the sulfate loss would increase the amount of available free NH 3 , which may result in the formation of more NH 4 NO 3 . This replacement behavior would compromise the effect of SO 2 control. This work suggests that the effect of NH 3 from longrange transport sources on haze formation in Shanghai should be emphasized both in field work and modeling calculation in the future.

Conclusions
Previous observations in Shanghai have found that the ion chemistry during a haze episode was characterized by high concentrations of SO 2− 4 and NO − 3 . This case study revealed that the enhancement of (NH 4 ) 2 SO 4 and NH 4 NO 3 in the fine particles were locally influenced, and additional ammonia played a vital role on the haze formation. The high occurrence of slightly acidic aerosols was attributed to the ammoniadeficient atmosphere over the largest megacity in China. The formation of ammonium nitrate was often limited by the availability of NH 3 . Under adverse meteorological conditions, advection of air parcels from a northerly direction resulted in the enhancement in NH 3 concentration, which favored the formation of sulfate and nitrate.