Simultaneous measurements of urban and rural particles in Beijing – Part 1: Chemical composition and mixing state

Two single-particle aerosol mass spectrometers (SPAMSs) were deployed simultaneously at an urban and a rural site in Beijing during an intensive field campaign from 1 to 29 November 2016 to investigate the source and process of airborne particles in Beijing. In the first part of this research, we report the single-particle chemical composition, mixing state, and evolution at both sites. A total of 96 % and 98 % of collected particles were carbonaceous at the urban and rural sites, respectively. Five particle categories, including elemental carbon (EC), organic carbon (OC), internal-mixed EC and OC (ECOC), potassium-rich (K-rich), and metals, were observed at both sites. The categories were partitioned into particle types depending on different atmospheric processing stages. A total of 17 particle types were shared at both sites. In the urban area, nitrate-containing particle types, such as EC-Nit (Nit: nitrate) and ECOC-Nit, were enriched especially at night, sulfate-containing particles were transported when wind speed was high, and ECOC-Nit-Sul (Sul: sulfate) were mostly aged locally. In sum, these processed particles added up to 85.3 % in the urban areas. In the rural area, regional particles were abundant, but freshly emitted ECOC and OC had distinct patterns that were pronounced at cooking and heating times. Biomass burning, traffic, and coal burning were major sources of particulate matter (PM2.5) in both rural and urban areas. Moreover, particles from the steel industry located in the south were also identified. In summary, the chemical composition of urban and rural particle types was similar in Beijing; the urban particles were influenced significantly by rural processing and transport. The work is useful to understand the evolution of urban and rural particles in Beijing during winter.


Introduction
China has experienced severe haze events caused by extremely high concentrations of fine particulate matter (PM 2.5 ) since January 2013. In the worst cases, an area of 2.0 million km 2 and a population of 800 million were affected . In the Beijing-Tianjin-Hebei (BTH) area, extreme haze events frequently occur during winter, with PM 2.5 mass rapidly reaching up to 200 µg m −3 and such levels being sustained for hours (Guo et al., 2014).
Over the last 2 decades, comprehensive studies have been conducted on urban PM in Beijing. He et al. (2001) reported the first characterization of PM 2.5 . Since then, numerous studies have been published on the characterization (Huang et al., 2010), sources (Guo et al., 2012;Sun et al., 2014a), and processing of PM . The mechanism of rapid-boosting PM 2.5 in Beijing, including new particle for-mation and growth (Guo et al., 2014), regional transport (Li et al., 2015a), and both Sun et al., 2014a), have been proposed. However, discrepancies remain among these studies. For example, the mass loading of PM 2.5 can rapidly increase to hundreds of micrograms per cubic meter (µg m −3 ). Both Wang et al. (2016b) and Cheng et al. (2016) suggested the secondary formation of sulfate from the oxidation of NO 2 , while Guo et al. (2014) have proposed a mechanism of particle formation and growth. Different from local secondary formation and accumulation, Li et al. (2015b) proposed that particles via long-range transport cause the elevation of PM 2.5 . According to Sun et al. (2014b) and Zhai et al. (2018), regional transport played important roles during heavy haze episodes. However, most of the studies have focused on the urban areas of Beijing with limited attention to the rural areas. To illustrate the sources, evolution, and transport of particles, the investigation of rural areas around Beijing is necessary.
Single-particle mass spectrometers (SPMSs) have been used to investigate the size-resolved chemical composition and mixing state of atmospheric particles (Gard et al., 1997;Pratt and Prather, 2012). More recently, singleparticle aerosol mass spectrometers (SPAMSs) have been used in Chinese megacities such as Beijing (Li et al., 2014), Shanghai , Guangzhou , Xi'an , Nanjing , and Chongqing . SPAMS has proven to be a useful tool for characterizing the single-particle chemical composition, mixing state, and processing of atmospheric particles. Due to the nature of laser desorption ionization (LDI), the instrument is very sensitive to dust and other types of particles containing sodium and potassium. This may cause bias in the particle matrix ).
In Beijing, particle types, such as carbonaceous, metallic, dust, potassium-rich, and others, were reported during spring and fall (L. Li et al., 2014). Moreover, lead-containing particles have also been investigated in recent studies (Ma et al., 2016;Cai et al., 2017). Organics, sulfate, nitrate, ammonium, and other species have been found internally mixed with the atmospheric particles, and these particle types are mostly from the combustion of fuel or biomass. The abundance of secondary species can indicate the degree of aging during atmospheric processing. Particles are mostly secondary species with deeper processing. However, these studies lack the use of these data to provide a view of dynamic particulate processing. These studies have focused on the urban areas of Beijing, limiting their information to the characterization of the particles in the Beijing region. Therefore, a simultaneous study to investigate the particle chemical composition and mixing state would fill the gap.
This study is part of the APHH-Beijing (Air Pollution and Human Health in a Chinese Megacity: Beijing) intensive field campaign during winter 2016 . Two SPAMSs were deployed simultaneously at Peking University (PKU) and Pinggu (PG) in order to observe both urban and rural particles in the Beijing region. The aims of the study are (1) to characterize the single-particle chemical composition and mixing state and (2) to investigate particulate evolution at both sites during haze events. These two objectives are presented in two parts. In Part 1, particle types and their atmospheric processing (e.g., origination, source, and diurnal profiles) at both sites are reported; in Part 2 (Chen et al., 2020), the detailed analysis of haze events, effects of heating activities, and evidence of regional transport are addressed.  (Huang et al., 2006). Briefly, the site is located on a rooftop (15 m above the ground) of the PKU campus which is surrounded by residential and commercial blocks. Trace gases including NO, NO 2 , SO 2 , O 3 , and CO (Thermo Inc. series models 42i, 43i, 49i, and 48i, respectively), meteorological parameters (Vaisala Inc.), and PM 2.5 (TEOM 1430) were recorded during the observation.
The PG site (40.173 • N, 117.053 • E) is 3 km from the PG center. The site is located to the northeast of the PKU site at a distance of 70 km. The PG site also acts as a host of the AIR-LESS (effects of air pollution on cardiopulmonary disease in urban and peri-urban residents in Beijing) Project. The meteorological data are acquired from the local meteorological office. The PG village is surrounded by orchards and farmland with no main road nearby on a scale of 3 km. Coal and biomass are used for domestic heating and cooking in the nearby villages.

Instrumentation and data analysis
Two SPAMSs (model 0515, Hexin Inc., Guangzhou, China) were deployed at both PKU and PG. A technical description of SPAMS is available in Li et al. (2011). Briefly, a SPAMS has three functional parts: sampling, sizing, and mass spectrometry. In the sampling part, particles within a 0.1-2.0 µm size range pass efficiently through an aerodynamic lens. In the sizing unit, the aerodynamic diameter (D va ) is calculated using the time of flight of particles. The particles are then decomposed and individually ionized into ions using a 266 nm laser. A bipolar time-of-flight mass spectrometer measures the ions and generates the positive and negative mass spectra of each particle. The two instruments were maintained and calibrated following the standard procedures before sampling .
A neural network algorithm based on adaptive resonance theory (ART-2a) was used to resolve particle types from both datasets (Song et al., 1999). The parameters used were as fol- lows: a vigilance factor of 0.70, a learning rate of 0.05, and 20 iterations. This procedure generated 771 and 792 particle groups. Then, the groups were combined into particle types based on similar mass spectra, temporal trends, and size distributions (Dallosto and Harrison, 2006). During combining, relative areas of nitrate and sulfate were used to distinguish between the stages of processing, assuming that more sulfate and nitrate can be measured if a particle is more processed during its lifetime. Thus, particles with relative peak areas of sulfate and nitrate larger than 0.1 were marked with sulfate (-Sul) or nitrate (-Nit), respectively, or both. Indeed, the matrix effect can affect ionic intensities between different particles during single-particle mass spectrometer analysis. However, the effect can be reduced using average mass spectra of particles within the similar size distribution and chemical composition. Finally, the strategy resulted in 20 and 19 particle types at PKU and PG, respectively. Among them, 17 types appeared at both sites, and the same type at both locations had the same mass spectra (R 2 > 0.80).

Results
A total of 4 499 606 and 4 063 522 particles were collected at the PKU and PG sites, respectively. The size distributions peaked at 0.48 and 0.52 µm (Fig. 1). The smaller size distribution was due to a more substantial fraction of freshly emitted particles at PG, as described in Table 1. A total of 17 particle types (R 2 > 0.80, mass spectra) were observed both at PKU and PG (Table 1). These particle types were labeled with the suffixes "_PKU" or "_PG" to indicate their locations. The term "particle category" stands for a group of particle types with variable stages of processing.

Meteorological conditions and overview
Temperature, relative humidity (RH), and wind speed at both sites during the sampling period are summarized in Table 2. Their temporal trends are available in Part 2. The average temperature at PKU (urban, 5.7 ± 2.3 • C) was higher than at PG (rural, 3.1 ± 2.2 • C). Correspondingly, relative humidity was higher at PG (67 ± 32 %) than at PKU (49 ± 30 %). The wind was stronger at PKU (2.5 ± 1.8 m s −1 ) than at PG (1.7 ± 0.9 m s −1 ). As shown in Fig. 2, wind speed at PKU peaked at 12:00 LT (local time, UTC+8; hereafter all times are local time), while wind speed at PG reached its maximum at 15:00. Various wind speeds determined the different dispersion patterns of pollutants near the surface. It should be noted that wind speed up to 2 m s −1 represents a scale of 172 km in diurnal transport. Therefore, at PKU, the wind could bring the pollutants from Hebei province under stagnant air conditions.

Elemental carbon category
As shown in Fig. 3a, the elemental carbon (EC) particle category was represented by ions peaking at m/z 12, 24, 36, 48, and 60 in positive mass spectra (Sodeman et al., 2005;Toner et al., 2008). EC is emitted from solid fuel combustion, traffic (Sodeman et al., 2005;Toner et al., 2008Toner et al., , 2006, and industrial activities (Healy et al., 2012). Due to the various ionic intensities of nitrate (m/z −46 and −62) and sulfate (m/z −80 and −97), the EC category has four types including EC-Nitrate (EC-Nit), EC-Sulfate (EC-Sul), and EC-Nit-Sul. Moreover, the EC category was more abundant after the heating began than before (Part 2), indicating that coal burning was one of the primary sources.
EC-Nit_PKU and EC-Nit_PG accounted for 7.0 % and 2.0 % in PKU and PG datasets, respectively. In the diurnal profiles of EC-Nit_PKU, there was an apparent early morning peak at 05:00 along with an evening peak (22:00). There was also an early morning NO x peak in the urban area of Beijing, providing sufficient precursors for secondary nitrate . Wang et al. (2018) validated the role of N 2 O 5 uptake on the nitrate formation in PM. Therefore, the early morning peak of EC-Nit_PKU occurred due to the uptake of nitrate on the freshly emitted EC in the early morning (Sun et al., 2014a). The evening peak could be due to the low temperature after the heating supply started (D. . Diurnally, EC-Nit_PG exhibited an early morning peak (05:00) but no evening peak and mainly came from the southeast.
EC-Sul was a minor type at both sites, accounting for 0.7 % at PKU and 0.1 % at PG. EC-Sul was pronounced in the afternoon when the wind was strong at both sites. It was unlikely for either EC-Sul_PKU or EC-Sul_PG to be local because their concentrations were associated with high wind speed, as shown in Fig. 3c. More specifically, EC-Sul_PKU came from the southeast and northeast of Hebei Province when the wind speed exceeded 6 m s −1 . EC-Sul_PG could have come from the west when the wind speed exceeded 2 m s −1 and the east when the wind speed exceeded 3 m s −1 as coal-using industries are located in both directions. Also, at both sites, the concentrations of SO 2 were elevated in the afternoon due to transport, providing sufficient precursors for the formation of sulfate .
OC-Sul_PG had morning (08:00) and afternoon (16:00) peaks, while the diurnal profile of OC-Sul_PKU showed a trend with early morning (03:00), morning (10:00), and afternoon (16:00) peaks. The diurnal trends OC-Sul at both PKU and PG were consistent with the heating patterns depending on the variation of local temperature. Moreover, OC-Sul_PG increased after the heating supply began. Polar plots suggest that OC-Sul_PKU came from the surrounding southwestern areas via transport, while OC-Sul_PG came from villages to the east and west (Fig. 4). These results suggest that OC-Sul_PG was emitted from coal burning for residential heating in nearby areas.
OC-Nit-Sul accounted for 7.4 % and 21.3 % of all detected particles at PKU and PG, respectively. OC-Nit-Sul_PKU had a diurnal peak at 07:00 during rush hour, suggesting that OC-Nit-Sul could be formed due to the uptake of nitrate on OC-Sul, while OC-Nit-Sul_PG had a diurnal peak at 08:00 due to traffic in nearby towns. As an aged particle type, OC-Nit-Sul_PKU and OC-Nit-Sul_PG, also acting similar to background types with hourly counts, remained low but had elevated to high levels at night. Polar plots suggest that OC-Nit-Sul_PKU mainly came from the surrounding areas, while OC-Nit-Sul_PG mainly came from the south and east, where populous villages are located (Fig. 4).

Potassium-rich category
All K-rich category particles showed different atmospheric evolution processes at both PKU and PG. K-rich_PKU illustrated a typical pattern that was at low levels in the daytime but high levels at night (22:00). As shown in Fig. 6c, at an average wind speed of 3 m s −1 , it took 5 h for particles at a distance of 50 km to arrive at PKU. This is also the reason why BB-related particles were abundant in urban Beijing where household BB is prohibited. The origination of K-rich_PKU was from nearby and the southwest. K-rich_PG, however, showed a pattern with cooking and heating activities, peaking at 07:00 and 17:00. The peak at 07:00 was due to the local emissions; the 17:00 emissions could be transported from a distance of 50 km at a wind speed of 3 m s −1 from the east and west.
The secondary process contributed to the early morning peak (05:00) of K-Nit_PKU due to the nighttime formation of nitrate via hydrolysis of N 2 O 5 in the NO x -rich urban areas . In the daytime, after rush hour, the number concentration of K-Nit_PKU increased again via the uptake of nitrate due to daytime photoactivity. K-Nit_PKU mainly originated from the local and southern areas (Fig. 6c). Besides the early morning peak, K-Nit_PG showed cooking and heating patterns that were abundant when the temperature was low in the early morning and afternoon. K-Nit_PG largely originated from both local and regional sources via long-range transport.

Metal category
Two metal-rich particle types were identified, namely Fe-rich and Ca-rich. The Fe-rich type contained iron (m/z 56 and 54), K + , Na + , NH + 4 , Cl − (m/z −35 and −37), sulfate, and nitrate. The Ca-rich type was composed of Ca + (m/z 40), CaO (m/z 56), K − , Na + , Cl − , sulfate, and nitrate. As shown in Fig. 6b, Ca-rich_PKU (0.4 %) and Ca-rich_PG (0.1 %) were likely of regional origin with no distinct diurnal variations. Since SiO − 2 and SiO − 3 (m/z −60 and −76) were not abundant in the Ca-rich particles, they are not likely to come from dust (Silva et al., 2000). According to its weak peaks during rush hour at PKU, a possible source of the Ca-rich particles was from road dust resuspension. Such rush hour peaks were not observed at PG.
Fe-rich_PKU (3.1 %) and Fe-rich_PG (1.8 %) had similar diurnal profiles that arose in the early morning when heavyduty vehicles were allowed to enter the 5th Ring expressway. The peak occurred earlier at PG (04:00 rather than 05:00) because these vehicles got closer to PG earlier than to PKU. The daytime peak occurred in the afternoon at both PKU and PG when wind speed was high. Therefore, there were also multiple sources for Fe-rich particles, including resuspended dust particles from traffic and fly ash from the steel industry. In Beijing, daytime Fe-rich particles were reported and assigned to long-range transport and industrial sources from Hebei Province (Fig. 7c) (Li et al., 2014). The steel industry moved out of Beijing more than a decade ago (L. . Currently, most of these steel industries are located in Hebei Province.

NaK category
As shown in Fig. 8, mass spectra of the NaK category contained Na + , K + , C + n , C − n , nitrate, and PO − 3 (m/z −79). The aged NaK particles contained strong signals of nitrate (NaK-Nit), sulfate (NaK-Sul), or both (NaK-Nit-Sul). In general, the NaK category contained stronger signals of Na + than the EC and K-rich categories. The NaK category may also come from incomplete solid fuel combustion processes such as coal, peat, or wood Healy et al., 2010;Xu et al., 2017). The NaK category was more abundant at PKU (9.5 %) than PG (5.8 %), suggesting a stronger contribution of emissions from coal boilers (Xu et al., , 2018. Additionally, after heating began, the fraction of NaK-Nit_PG and NaK-Sul-Nit_PG increased by 1.2 times (see Part 2).
NaK_PKU showed no distinct diurnal variations, suggesting that it was a regional particle type arriving at the PKU site via transport, while NaK_PG showed an apparent diurnal variation consistent with cooking and heating patterns. Polar plots also suggest that they are from the east and the west. NaK-Nit, with a considerable uptake of nitrate, was more abundant at PKU (6.4 %) than PG (1.7 %). Both NaK-Nit_PKU and NaK-Nit _PG increased in the afternoon when photochemical activities were most active (Fig. 8c). Both of them may be from regional transport (Fig 8b and c).
NaK-Sul was a minor particle type at both PG and PKU, accounting for 0.2 % and 0.4 %, respectively. The diurnal profile of NaK-Sul_PG was also following the local cooking and heating patterns, while NaK-Sul_PKU showed a typical transport pattern that became abundant in the afternoon as the southwesterly wind speed increased. As a heavily aged particle type, NaK-Nit-Sul was transported to both PKU and PG from the southwest. In short, NaK-related particle types mainly came from the solid fuel burning process, e.g., coal. Due to its different origins, it showed different levels of processing at PKU and PG.

Unique particle types at the PKU site
OC-Nit_PKU (0.9 %) and ECOC-Nit_PKU (3.1 %) with strong ion intensities of nitrate were observed at the PKU site. OC-Nit_PKU and ECOC-Nit_PKU showed a peak at nighttime rather than at daytime, similar to the diurnal profiles of OC-Nit-Sul_PKU and ECOC-Nit-Sul_PKU. Such nitrate-rich particle types could have come from the uptake of nitrate in OC and ECOC (Qin et al., 2012;Chen et al., 2016). Polar plots suggest that both types were formed locally when the wind speed was lower than 4 m s −1 . The NO x -rich environment in urban Beijing provides a favorable condition Figure 6. (a) Average mass spectra of BB, K-Nit, K-Sul, and K-Nit-Sul observed at both sites; (b) diurnal patterns of the hourly count of K-rich, K-Nit, K-Sul, and K-Nit-Sul at both sites; (c) polar plots of BB, K-Nit, K-Sul, and K-Nit-Sul; the gray circles indicate wind speed (m s 1 ).
A minor amount (0.10 %) of amine-containing particles were observed at the PKU site, and trimethylamine ion fragments (m/z 58 and 59) were influential in the mass spectrum of K-amine-Nit-Sul_PKU (Fig. 9a). The diurnal profile of K-amine-Nit-Sul_PKU showed an afternoon peak, indicating a regional source (Fig. 9c). K-amine-Nit-Sul_PKU was transported to the site from nearby locations. The amines may come from animal husbandry, BB, traffic, or vegetation . Amines were ubiquitous in the atmospheric environment, playing essential roles in new particle formation and growth, as well as fog and cloud processing (Ge et al., 2011;Chen et al., 2019).

Unique particle types at the PG site
OC_PG (5.9 %) and ECOC_PG (3.3 %) were only observed at the rural PG site (Fig. 10). The major components of these two types were consistent with the OC and ECOC categories but with limited uptake of sulfate and nitrate, respectively, suggesting that they were possibly freshly emitted particles (Peng et al., 2020b). Their diurnal profiles are consistent with cooking and heating patterns which peaked at 07:00 in the morning and 17:00. Polar plots suggest that OC_PG mainly came from nearby and from other remote areas in all directions except the north. ECOC mainly came from the east of the PG site. These results support the assumption that the two types were mainly from local emission sources. Also, the emission of OC_PG and ECOC_PG is prevalent in the region.

Discussion
Multiple models for source apportionment have been used in Beijing to quantify the sources of particles (Sun et al., 2014a;Xu et al., 2015;Zhai et al., 2018). Biomass burning, coal combustion, traffic, and dust are the key sources of PM (Sun et al., 2014a;Liu et al., 2018;Huang et al., 2014). Multiple studies have confirmed that biomass burning is an essential source of PM in urban Beijing Sun et al., 2014a;. In this study, biomass burning and other solid fuel burning were identified as crucial sources of PM in not only urban but also rural areas of Beijing. We observed that BB-related particles (K-rich category) were more abundant at PG than at PKU. In particular, we found freshly emitted K-containing particles at the Pinggu site, confirming the importance of local emissions to PM. Furthermore, K-containing particles in the urban area were more aged, suggesting that they are aged mostly from the surrounding areas. This result is consistent with that of Y.  based on combined receptor and footprint models. Nevertheless, household emissions in the BTH region caused 32 % and 15 % of primary PM 2.5 and SO 2 . These studies have proven the importance of household emissions from BB in the BTH area (J. . Especially at the PG site, the ambient PM was mainly controlled by longrange transport and household emissions from cooking and heating. Due to the nature of SPAMS, the chemical composition of PM cannot be precisely quantified. However, single-particle aerosol mass spectrometers have advantages in studying the chemical composition, mixing state, source, and process of particles . Mass-based technologies can not differentiate between the origins of the bulk of nitrate or whether it is transported or formed locally. Indeed, single-particle types in urban Beijing have been reported in previous studies (Li et al., 2014;, and the major types are consistent with this study. However, in this study, we adopted a cluster strategy considering the relative ion peak area of sulfate and nitrate as indicators of particle processing. Therefore, more detail could be extracted from both datasets. We confirmed that the source, origin, and processes were different for these particles in the urban and rural areas. For example, the seriously processed particles, such as K-Nit-Sul, OC-Nit-Sul, and NaK-Nit-Sul, acted with no distinct diurnal patterns as background or regional sources Figure 8. (a) Average mass spectra of NaK, NaK-Nit, NaK-Nit-Sul, and NaK-Sul observed at both sites; (b) diurnal patterns of the hourly count of NaK, NaK-Nit, NaK-Nit-Sul, and NaK-Sul at both sites; (c) polar plots of NaK, NaK-Nit, NaK-Nit-Sul, and NaK-Sul; the gray circles indicate wind speed (m s 1 ). . The processed particles, such as OC-Nit, ECOC-Nit, and NaK-Nit, were affected by emissions and secondary formations.
The emission and transport patterns were different in the urban and rural areas, resulting in different characteristics of PM. For example, EC particles were a key component at PKU (18.2 % in total) but a minor particle type at PG (5.6 %). Meanwhile, in the urban area of Beijing, direct emissions of K-rich particles should be small due to strict con-trol measures; thus, the K-Nit-Sul particles are mainly from long-range transport. Transported particles were aged and commonly coated in a thick layer of nitrate and sulfate, but the local particles were affected by both emission and the near-surface aging process. For example, at PKU, the primary emission sources were traffic and central heating supply, causing an NO x -rich region in which freshly emitted particle types could undergo processing due to the uptake of nitrate (Wang et al., 2016a). In the nearby villages of PG, do- mestic heating and cooking were the major contributors of primary particles when the temperature was low in the morning and afternoon, resulting in the emission of multiple primary particle types such as OC_PG and ECOC_PG. In short, the characteristics of PM in urban and rural areas of Beijing were affected by local emissions, and they interacted with each other due to regional transport.
SO 2 was controlled strictly in Beijing. However, the emission of SO 2 is still significant in the nearby Hebei and Shandong provinces . The different control measures produced a low concentration area of SO 2 around Beijing. The sulfate-rich particle types, such as EC-Sul, OC-Sul, K-Sul, and NaK-Sul, arrived at the PKU site usually when wind speed was high (> 3 m s −1 ). The wind directions, along with the transport of sulfate-rich particles, were easterly, southwesterly, and southerly. In these directions, sulfate was primarily emitted from coal burning from residential heating, power generation, and industry, or there was secondary uptake on the preexisting particles . Likewise, a part of the sulfate-rich particles arrived at the PG site when wind speed was high. However, the locally formed particles were also pronounced, especially for ECOC-Sul, K-Sul, and NaK-Sul. As discussed in Sect. 3, ECOC-Sul and NaK-Sul were mainly from the coal burning for residential heating. The K-Sul was formed due to the uptake of secondary sulfate. Conclusively, the particulate characterization in rural areas around Beijing is influenced significantly by residential coal burning.
Secondary nitrate formation is still a critical issue in Beijing. The daytime rise in nitrate has been reported in studies , and we also found a similar predominance of nitrogen-containing particles in this study. Recent studies have reported the early morning peaks of nitrate using a soot particle aerosol mass spectrometer (SP-AMS) , which is consistent with our results. Interestingly, the early morning peak was only observed for several particle types at both sites, including EC-Nit_PKU, K-Nit_PKU, EC-Nit-Sul-PG, and EC-Nit_PG. This result is not surprising Figure 10. (a) Average mass spectra of OC_PG and ECOC_PG, (b) diurnal plots of OC_PG and ECOC_PG, and (c) polar plots of OC_PG and ECOC_PG. All these particle types appeared at the PG site.
because PG is also an NO 2 -rich region . The increasing contribution of nitrate-containing particles suggests the role of nighttime chemistry in nitrate uptake on particles. Wang et al. (2017) revealed the importance of nighttime N 2 O 5 chemistry on nocturnal nitrate formation in the urban area of Beijing. The heterogeneous hydrolysis of N 2 O 5 was most favorable when NO was at a low level. Moreover, the polar plots suggested a small role of long-range transport to the nitrate in individual particles. The contribution of local traffic was insignificant at the PG site as it was far from highways and major roads, the nighttime formation of nitrate appeared to be important in PG as well.

Conclusions
Two SPAMSs were simultaneously deployed at urban and rural sites in Beijing in order to characterize PM during wintertime. The results at both sites indicate that they shared 17 types of common clusters, most of which belonged to particle categories such as EC, OC, ECOC, BB, and NaK. The origins and sources of these particle types at both sampling sites were also comprehensively analyzed. Most of the processed PM, including EC-Nit-Sul_PKU, ECOC-Nit-Sul_PKU, and NaK-Nit-Sul_PKU, were aged locally in an NO x -rich environment, while EC-Nit-Sul_PG, ECOC-Nit-Sul_PG, NaK-Nit-Sul_PG, and OC-Nit-Sul_PG were regional. Domestic heating in the rural area was found to be an important source of PM, and such heating activities typically caused three diurnal peaks in the early morning, morning, and afternoon (after sunset). Moreover, the early morning peak of nitrate was observed at both sites, suggesting the contribution of the heterogeneous hydrolysis of N 2 O 5 in the dark during the winter. The insights gained in this study can provide useful references for understanding the relationship between regional transport and local aging in both urban and rural areas in Beijing. In Part 2, we focus on haze events observed at both sites and attempt to determine the effects of heating activities and possible regional transport between PKU and PG.
Data availability. All data described in this study are available upon request from the corresponding authors.
Author contributions. FY, MZ, TZ, QZ, and KH designed the experiments. YC, JC, ZW, MT, CP, and HY carried them out. XiY, XiaY, YL, GS, and ZS analyzed the experimental data. YC prepared the article with contributions from all coauthors.
Competing interests. The authors declare that they have no conflict of interest.
Special issue statement. This article is part of the special issue "In-depth study of air pollution sources and processes within Beijing and its surrounding region (APHH-Beijing) (ACP/AMT interjournal SI)". It is not associated with a conference.