Wintertime aerosol chemistry and haze evolution in an extremely 1 polluted city of North China Plain: significant contribution from coal 2 and biomass combustion 3

. The North China Plain (NCP) frequently experiences heavy haze pollution, particularly during wintertime. In 2015- 20 2016 winter, the NCP region suffered several extremely severe haze episodes with air pollution red alerts issued in many cities. 21 We have investigated the sources and aerosol evolution processes of the severe pollution episodes in Handan, a typical 22 industrialized city in the NCP region, using real-time measurements from an intensive field campaign during the winter of 23 2015-2016. The average (± 1  ) concentration of submicron aerosol (PM 1 ) during December 3, 2015 – February 5, 2016 was periods (daily PM 2.5 > 75 μg m -3 ), indicating relatively synchronous increases of all aerosol species during haze formation. The case study of a severe haze episode, which lasted 8 days starting with a steady build-up of aerosol pollution followed by a persistently high level of PM 1 (326.7 – 700.8 μg m -3 ), revealed the significant influence of stagnant meteorological conditions 38 which acerbate air pollution in the Handan region. The haze episode ended with a shift of wind which brought in cleaner air 39 masses from the northwest of Handan and gradually reduced PM 1 concentration to < 50 μg m -3 after 12 hours. Aqueous-phase 40 reactions under higher relative humidity (RH) were found to significantly promote the production of secondary inorganic 41 species (especially sulfate), but showed little influence on SOA.

events under typical fall conditions in Beijing, Guo et al. (2014) indicated that nucleation consistently preceded a polluted 68 period with high number concentrations and the development of the episode involved efficient and sustained growth from the 69 nucleation-mode particles over multiple days. In addition, organic aerosol (OA) was found to be a major component of aerosol 70 particles, accounting for more than one-third of total PM1 mass. The primary OA (POA) from traffic, cooking, biomass burning, 71 coal combustion, etc., and secondary OA (SOA) have been distinguished and quantified mainly using positive matrix 72 factorization (PMF; Paatero and Tapper, 1994). Recently, a novel PMF procedure, with the multi-linear engine (ME-2) 73 algorithm, was developed to apportion the OA sources in Beijing and Xi'an, allowing for a more objective selection of source 74 apportionment solution (Elser et al., 2016). However, our knowledge of the sources and aerosol evolution processes for the 75 whole region still remains incomplete and is especially limited for areas outside of Beijing. For other cities in the NCP region, 76 such as Hebei province, only a limited number of aerosol studies have been conducted using offline filter-based measurement 77 techniques (Zhao et al., 2013;Wei et al., 2014). Due to low time resolution varying from one day to several days, these studies 78 provided relatively limited information on aerosol emission sources and formation processes, thus it remains unclear how the 79 rapid haze evolution happens and what the driving sources are for the air pollution problems in Hebei. Therefore, it is crucial 80 to conduct research in the areas outside of Beijing, especially many provinces subjected to high anthropogenic emissions, 81 which may provide critical information to help air pollution policy making to be more direct and efficient. 82 To fill this knowledge gap, an intensive field campaign with multiple state-of-the-art research instruments was conducted 83 in Handan, a major city in southern Hebei, during the winter of 2015-2016. Handan is located in the intersectional area of four 84 provinces, Hebei, Shanxi, Henan and Shandong, all of which are heavily urbanized and industrialized (Fig. 1a). Handan itself 85 is also well known for heavy industrial production of steel, iron and cement, which result in high local emissions of air 86 pollutants. According to the routine monitoring of the China National Environmental Monitoring Center (CNEMC) from 2013 87 to 2015, Handan is always listed as one of the top 10 polluted cities in China. Hence, this location and its specific conditions 88 allow for a detailed exploration of aerosol chemistry and haze evolution processes under high anthropogenic emissions. 89 6 the US National Ambient Air Quality Standards (NAAQS, 35 μg/m 3 for the 24 h average of PM2.5) and 13 days met the Chinese 163 NAAQS Grade II (75 μg/m 3 for the 24 h average of PM2.5) for the whole study period of 65 days. In other words, the daily 164 average PM2.5 concentrations exceeded the US NAAQS and the Chinese NAAQS on 94% and 80% of the days, respectively 165 ( Fig. 2). On December 22, the daily PM2.5 concentration reached the highest value of 725.7 μg/m 3 , leading to the first "red" 166 haze alarm (http://www.cma.gov.cn/kppd/kppdsytj/201310/t20131028_229921.html) ever in Hebei Province. The 167 meteorological conditions were stagnant with calm winds throughout the study period (WS usually less than 1.5 m/s), although 168 relatively high WS (generally > 1.5 m/s) with cleaner air from northwest of Handan occasionally interrupted the haze evolution 169 process (Fig. 2b). The RH varied from 11.7% to 94.8%, generally with higher values for more polluted periods and lower 170 values during cleaner periods. No precipitation occurred throughout the entire campaign. 171 Hourly PM1 concentrations fluctuated dramatically from 4.2 μg/m 3 to 700.8 μg/m 3 (Fig. 2g). The average PM1 172 concentration was 187.6 μg/m 3 , more than twice as high as that observed in the well-known severe haze event that occurred in 173 Beijing in January 2013 (Sun et al., 2014;Zhang et al., 2014). Organics constituted a major fraction of PM1, contributing 45% 174 on average during this study, followed by sulfate (15%), nitrate (14%), and ammonium (12% biomass burning and coal combustion) (Lobert et al., 1999;McCulloch et al., 1999), and formed in the atmosphere through 179 gas-to-particle conversion (e.g., NH4Cl partitioning) (Baek et al., 2006). Considering that chloride demonstrated pronouncedly 180 enhanced peaks during night and it showed good correlations with CCOA and BBOA (r=0.72 and 0.80, respectively), a large 181 fraction of chloride during wintertime was thought to be from primary emissions at night. On average, BC accounted for 5% 182 of total PM1. Its distinct peaks at morning and evening rush hours suggested that BC was mainly associated with traffic 183 emissions. In the daytime, PM1 was dominated by secondary species because of active photochemistry, whereas the 184 contributions of primary species were significantly increased at night, probably caused by enhanced primary emissions from 185 fuel combustion coupled with shallow boundary layer height (Fig. S9). 186 Ambient CO is an indicator for the intensities of anthropogenic emissions. The hourly CO concentration was as high as 187 39% (Fig. 3). The traffic-related HOA only accounted for 7% of total OA, which was in accordance with the fact that PMF 197 analysis performed with the PMF2 algorithm had difficulty to retrieve it (see Sect of HOA under relatively low WS (<1.5m/s), suggesting that HOA was substantially influenced by local emission sources, in 213 accordance with its primary characteristics (Fig. S10). 214

Coal combustion OA 215
Although coal combustion has rarely been reported as an important source of organic aerosols in the US or Europe, it is 216 a large emitter of organics in China (Cao et al., 2006). According to Zhang et al. (2008b), organic carbon can contribute up to 217 70% of emitted PM2.5 for different types of coal combustion in China. During wintertime, coal is the primary fuel for various 218 industries (e.g. power generation, steel milling, and cement production) as well as residential heating in the NCP region. Thus 219 a considerable contribution from coal combustion to OA concentration was expected in this study. Compared to HOA and 220 BBOA, the mass spectrum of CCOA showed strong signals at higher m/z, especially a significant peak at m/z 115, and the 221 temporal trend of CCOA correlated tightly with that of m/z 115 (r=0.99, Fig. 3 has shown that the ACSM mass spectra of OA from residential coal combustion emissions tend to present a high peak at m/z 225 115. In addition, CCOA was also found to correlate relatively well with chloride (r = 0.72) during this study, consistent with 226 the fact that coal combustion is also an important emission source of chloride. 227 Figure 4 compares the OA composition in this study with those of previous winter studies in China. During wintertime, 228 CCOA was observed to contribute a significant fraction of the fine PM mass in regions to the north of the Yangtze River (e.g. 229 Beijing, Lanzhou, and Handan), due to domestic coal combustion for heating in winter. However, little to no CCOA was 230 observed in areas located to the south of the Yangtze River, e.g. Nanjing, Jiaxing, and Ziyang, mainly reflecting the lack of  representative of SOA. Although the diurnal profile of OOA was overall flat in this study, the mass fraction of OOA to total 272 OA increased significantly during daytime, reaching a maximum of 64% at 14:00 BJT (Fig. 3n). 273 As displayed in Fig. 6, the diurnal variations of meteorological parameters did not significantly change from weekdays 284 to weekends, providing a good opportunity to investigate the influence of anthropogenic activities. As expected, the diurnal 285 pattern of HOA, which is associated with traffic emissions, presented a more distinct morning peak on weekdays. This was 286 also the case for BC, CO, and NOx, which are all fossil fuel combustion tracers. However, the evening rush hour peaks of these 287 species did not show much of a difference between weekdays and weekends, indicating that human activities in the evening 288

Diurnal variations and insights into aerosol sources
were not significantly reduced on weekends. Other aerosol species showed generally similar diurnal trends for weekdays and 289 weekends, similar to the results observed in Beijing (Sun et al., 2013a). In contrast, stronger weekday vs. weekend differences 290 were observed in the US, where the mass concentrations of aerosol species are obviously lower during weekends (Young et 291 al., 2016;. Results from the present study reveal that active anthropogenic emissions tend to persist 292 throughout the entire week in polluted regions in Handan, leading to limited differences in the concentrations and compositions 293 of major air pollutants between weekdays and weekends. The exception is traffic emissions, for which the morning rush hour 294 peak is more prominent during weekdays. 295

Polluted versus non-polluted periods 296
To gain further insights into the evolution of aerosol particles throughout the day, especially during hazy conditions, we 297 explored the diurnal differences of meteorological conditions and air pollutants between polluted and non-polluted days (Fig.  298   7). According to the CNAAQS Grade II of daily PM2.5 concentrations (75 μg/m 3 ), only the 13 days (out of a total of 65 days) 299 were found to meet the requirement and are considered to be non-polluted in this study; the rest are defined as polluted periods. 300 Note that of these 13 non-polluted days, only 3 days achieved the 24 h CNAAQS Grade I level of PM2.5 (35 μg/m 3 ). 301 The temperature was relatively low throughout the period, averaging 2.1°C and 0.2°C on polluted and non-polluted days, 302 respectively. The RH during polluted periods was slightly higher during daytime, favoring the aqueous-phase processing of 303 atmospheric pollutants. The influence of RH is discussed in detail in Sect. 3.5. Stagnant weather conditions with lower wind 304 speeds were observed on polluted days, especially during nighttime, which would aggravate the accumulation of aerosol 305 pollution. Unsurprisingly, the mass concentrations of aerosol components and the mixing ratios of gaseous species were much 306 higher on polluted days. But the diurnal differences between polluted and non-polluted periods could provide some information 307 regarding their evolutionary processes. The diurnal profiles of secondary inorganic species (i.e. sulfate, nitrate, and 308 ammonium), were flatter on polluted days. For example, in the diurnal profile of nitrate during polluted periods, the maximum 309 and minimum concentrations were different by only 13 % or 4.4 μg/m 3 . This behavior is consistent with the comparison of 310 polar plots between polluted and non-polluted days, which indicated a significant effect of regional transport on polluted 311 periods for secondary species. In contrast, the diurnal trends of primary aerosol species, e.g. HOA, BBOA, and CCOA, during 312 polluted periods differed substantially from those during non-polluted periods. Compared to the relatively flat diurnal profiles 313 on non-polluted days, the mass concentrations of HOA, BBOA, and CCOA were strongly enhanced at nighttime on polluted 314 showed the highest polluted/non-polluted ratios, which were 5.3, 5.0, and 5.5, respectively (Fig. 8b). Given the higher average 325 RH on polluted days (average±1 = 56.5±18.8%) compared to non-polluted days (average±1 = 40.9±18.7%), aqueous-phase 326 processing likely has increased the production of sulfate (Wang et al., 2012;Zheng et al., 2015;Elser et al., 2016). During 327 polluted days, the average oxidation ratio of sulfur (molar ratio of sulfate to sum of sulfate and SO2) was 0.27, higher than that 328 on non-polluted days (0.16). On the other hand, the strong increases of CCOA and BBOA were possibly caused by enhanced 329 gas-to-particle partitioning associated with high PM mass loadings during polluted periods (Mader et al., 2002). Interestingly, 330 compared to aerosol species, CO showed a lower polluted/non-polluted ratio of approximately 2. A possible reason is that CO 331 has a longer atmospheric lifetime compared to aerosol particles, thus it has a more elevated regional background concentration. 332 Note that the polluted/non-polluted ratios for SO2 and NOx were also lower compared to the aerosol species. This is potentially 333 a result of enhanced aqueous phase oxidation of SO2 and NOx as well as more efficient wet deposition, since the more polluted 334 periods were generally more humid. 335  (Fig. 9c) demonstrates that higher mass fractions of organics in PM1 were associated with 351 elevated POA contributions to total OA, especially when PM1 concentrations were more than 200 μg/m 3 (Fig. 9c). Overall, the 352 results here suggest that secondary formation of sulfate (mainly from SO2 emitted by coal combustion), and primary emissions 353 of organics from coal combustion and biomass burning are important factors driving the development of winter haze pollution 354 in Handan.
build-up of air pollutants, including fine particles and CO, over a period of ~ 5 days (Dec. 17 -21, 2015) followed by 358 approximately 4 days of heavy air pollution, during which the average CO mixing ratio was 6.7 ppm and the average PM1 359 concentration was 500.1 μg/m 3 (Fig. 10). This episode ended on Dec. 25, during which winds from the northwest brought in 360 cleaner air, leading to dramatic reductions of air pollutants. This type of evolutionary process has been frequently observed in 361 Beijing during autumn and winter, and is called "sawtooth cycles" by Jia et al. (2008). In this study, the whole haze cycle was 362 divided into five stages: (1) a clean period (Stage 1), (2)  To further evaluate the influence of air mass origins on aerosol characteristics, we performed the cluster analysis of 385 HYSPLIT back trajectories for the whole study period to elucidate the relationship between aerosol concentration or 386 composition and different clusters. As shown in Fig. S12, the whole NCP region was heavily polluted, with high PM1 387 concentrations for all four clusters. Overall, the aerosol compositions were similar among different clusters. However, we 388 indeed observed an important role played by winds in altering aerosol characteristics according to the above case study. 13 classified as "Others" (Fig. 11). As expected, the PM1 concentration of "Others" was more than six times higher than that of 392 "NW_HWS". Secondary aerosol species (i.e. sulfate, nitrate, ammonium and OOA) contributed 66% of total PM1 for 393 "NW_HWS". As air masses associated with "Others" were more strongly influenced by anthropogenic sources, the main 394 primary species (i.e. HOA, BBOA, CCOA and BC), accounted for a higher fraction of 32% for "Others". These results 395 highlight the importance of high winds from the northwest of Handan in alleviating PM levels and changing aerosol 396 composition during wintertime. 397 As mentioned previously, the sulfate contribution during stage 3 was visibly enhanced under high RH, revealing the 398 effects of RH on aerosol processing. Many previous studies have observed the increased production of secondary inorganic 399 aerosol species through aqueous-phase processing. In this study, we used the oxidation ratios of sulfur and nitrogen, defined 400 as = aqueous-phase-processed SOA (aq-OOA) that tracked well with RH in Beijing during wintertime. However, the mass fraction 408 of SOA in total OA in this study remained relatively stable and showed no dependency on RH (Fig. 12c). The RH-binned bulk 409 composition of submicron aerosol also only exhibited an obvious increase of sulfate at high RH (Fig. S13) dominated the bulk composition of submicron aerosols (44.6% of PM1 mass), similar to previous observations in the NCP biomass burning, and one SOA factor. CCOA was the largest contributor to POA, on average accounting for 29%, followed 424 by BBOA (25%). The mass fraction of HOA in total OA was only 7%, indicating the minor contribution of traffic emissions 425 in Handan. Although the aerosol concentration during polluted days was more than three times higher than that during non-426 polluted days, little variation was observed in the average aerosol bulk composition, revealing the relatively synchronous 427 increase of all aerosol species during haze evolution. Stagnant weather conditions, with low wind speed and high RH, and 428 strong enhancement of primary species at nighttime, prompted haze formation during polluted days. Variation of aerosol mass 429 fractions with hourly increasing PM1 concentration further revealed that secondary formation of sulfate (mainly from SO2 430 emitted by coal combustion) and primary emissions from coal combustion and biomass burning, are important factors driving 431 haze formation. This is mainly related to large emissions of air pollutants from coal and biomass combustion during wintertime, 432 especially for simple household stoves with low combustion efficiency. Overall, sulfate, chloride, and CCOA on average 433 accounted for a total of 37% of PM1 mass (Fig. 1c) Fraction H a n d a n , 2 0 1 5 B e ij in g , 2 0 1 0 B e ij in g , 2 0 1 1 B e ij in g , 2 0 1 2 B e ij in g , 2 0 1 3 L a n z h o u , 2 0 1 4 N a n ji n g , 2 0 1 3 J ia x in g , 2 0 1 0 Z iy a n g , 2 0 1 2 North of the Yangtze River South of the Yangtze River  250  200  150  100  50  0  24  16  8  0   80   60   40   20   0  24  16  8  0   60   40   20   0  24  16  8  0   50  40  30  20  10  0  24  16  8  0   50  40  30  20  10  0  24  16  8