Characterization of Organic Aerosol at a Rural Site in the North China Plain Region: Sources, Volatility and Organonitrates

Qiao ZHU, Li-Ming CAO, Meng-Xue TANG, Xiao-Feng HUANG,Eri SAIKAWA, and Ling-Yan HE*

1Key Laboratory for Urban Habitat Environmental Science and Technology,Peking University Shenzhen Graduate School, Shenzhen 518055, China

2Department of Environmental Sciences, Emory University, Atlanta, GA 30322, USA 3Rollins School of Public Health, Emory University, Atlanta, GA 30322, USA

ABSTRACT The North China Plain (NCP) is a region that experiences serious aerosol pollution. A number of studies have focused on aerosol pollution in urban areas in the NCP region; however, research on characterizing aerosols in rural NCP areas is comparatively limited. In this study, we deployed a TD-HR-AMS (thermodenuder high-resolution aerosol mass spectrometer) system at a rural site in the NCP region in summer 2013 to characterize the chemical compositions and volatility of submicron aerosols (PM1). The average PM1 mass concentration was 51.2 ± 48.0 μg m−3 and organic aerosol(OA) contributed most (35.4%) to PM1. Positive matrix factorization (PMF) analysis of OA measurements identified four OA factors, including hydrocarbon-like OA (HOA, accounting for 18.4%), biomass burning OA (BBOA, 29.4%), lessoxidized oxygenated OA (LO-OOA, 30.8%) and more-oxidized oxygenated OA (MO-OOA, 21.4%). The volatility sequence of the OA factors was HOA ＞ BBOA ＞ LO-OOA ＞ MO-OOA, consistent with their oxygen-to-carbon (O:C)ratios. Additionally, the mean concentration of organonitrates (ON) was 1.48−3.39 μg m−3, contributing 8.1%–19% of OA based on cross validation of two estimation methods with the high-resolution time-of-flight aerosol mass spectrometer (HRToF-AMS) measurement. Correlation analysis shows that ON were more correlated with BBOA and black carbon emitted from biomass burning but poorly correlated with LO-OOA. Also, volatility analysis for ON further confirmed that particulate ON formation might be closely associated with primary emissions in rural NCP areas.

Key words: organic aerosols, volatility, organonitrates, biomass burning, North China Plain

1. Introduction

Atmospheric fine aerosols have attracted much attention from the public owing to their adverse effects on visibility, global climate and human health (Dockery et al., 1993;Charlson and Heintzenberg, 1995; Haywood and Boucher,2000; Ramanathan et al., 2001; IPCC, 2013). The North China Plain (NCP), which is one of the most important city clusters in China, is a global hotspot of aerosol pollution originated from both natural (e.g., dust storms) and anthropogenic (e.g., industrial emissions, vehicle emissions and biomass burning) sources. As a result, numerous studies on the sources, formation and evolution processes of aerosols have been conducted in urban regions, such as Beijing, Tianjin,and some cities in Hebei Province that have experienced severe air pollution (Song et al., 2006; Gu et al., 2010;Huang et al., 2010; Liu et al., 2012; Sun et al., 2012; Wang et al., 2014; Li et al., 2017). Compared to studies on the urban atmosphere, studies focusing on aerosol characterization in rural NCP areas are much more limited. For example, as a key property for aerosols, volatility can greatly influence the mass concentration and size distribution of aerosols via gas–particle partitioning (Lazaridis,1999; Pankow and Barsanti, 2009; Bilde et al., 2015), and previous studies suggest that the volatility of different aerosol compounds may vary from site to site (Huffman et al.,2009a, b; Bi et al., 2015; Cao et al., 2018, 2019; Xu et al.,2019). However, to the best of our knowledge, aerosol volatility measurements in rural NCP areas do not exist. Research on the volatility of different aerosol compounds in rural NCP areas is needed to obtain a more comprehensive understanding of aerosol fate on the regional scale. Three methods are frequently applied to measure the volatility of aerosols: VTDMA (volatility tandem differential mobility analyzer) (Dassios and Pandis, 1999; Kuhn et al., 2005; Saleh et al., 2008), dilution samplers (Shrivastava et al., 2006),and the combination of a thermodenuder and aerosol mass spectrometer (TD-AMS system) (Huffman et al., 2008).Among them, the TD-AMS system can measure the volatility of different chemical compositions with a high temporal resolution and has been widely used in field campaigns (Huffman et al., 2009a, b; Bi et al., 2015; Cao et al., 2018, 2019;Xu et al., 2019) and laboratory studies (Kolesar et al., 2015;Saha and Grieshop, 2016).

Organonitrates (ON) are generating increasing interest owing to their crucial role in the chemistry of atmospheric oxidation and potentially significant influence on regional air quality, climate change and global biogeochemical cycles (Bertman et al., 1995; Jenkin and Clemitshaw, 2000;Russo et al., 2010). The broad definition of ON includes peroxy nitrates (RONO) and multifunction alkyl nitrates(RONO). Lacking the ideal analytical instruments to detect the bulk ON was the biggest obstacle for fully understanding the chemistry of ON in the past. In recent years, several advanced techniques have emerged that are capable of measuring various kinds of ON, especially for particulate ON. Aerosol mass spectrometry (AMS), which has been widely used to characterize organic compounds of aerosols with high temporal resolution, can quantify the number of ON species based on indirect estimation methods (Farmer et al., 2010;Hao et al., 2014; Xu et al., 2015a, b). Rollins et al. (2012)reported the first direct ambient measurement of particulate ON by the Berkeley thermal dissociation-laser induced fluorescence method. These measurements provided significant evidence that ON contribute a substantial portion of the secondary organic aerosols (Rollins et al., 2012; Fry et al.,2013; Ayres et al., 2015; Boyd et al., 2015; Xu et al.,2015b; Lee et al., 2016; Yu et al., 2019). There are two main secondary formation pathways for ON: OH-initiated oxidation of hydrocarbons in the presence of NOin the daytime, and NO-induced oxidation of alkenes at night. For the secondary formation pathway, several studies have shown that particulate ON yields were high via nighttime NOreacting with biogenic unsaturated alkenes (i.e., isoprene and monoterpene) (Fry et al., 2009; Ayres et al.,2015; Boyd et al., 2015, 2017). Very recently, Joo et al.(2019) reported that particulate ON can be formed from the reaction of NOoxidation of 3-Methylfuran, which is an important tracer species in biomass burning plumes.However, the properties, sources and formation mechanisms of particulate ON are still not well understood (Perring et al., 2013). To the best of our knowledge, most research on particulate ON has been conducted in the U.S.and Europe (Perring et al., 2013; Ng et al., 2017), and very few studies have reported detailed particulate ON results for urban areas in China (Xu et al., 2017; Yu et al., 2019),which are characterized by an atmosphere with a large variety of anthropogenic pollutants.

In this study, we conducted aerosol measurements using a TD coupled with an Aerodyne high-resolution timeof-flight aerosol mass spectrometer (TD-HR-AMS) and other collocated instruments at a rural site in the NCP region in summer 2013. Besides investigating the sources and volatility of organic aerosol (OA), we also focused on characterizing the particulate ON with mass concentration,volatility and their possible sources in the NCP rural region.

2. Materials and methods

2.1. Site description and instrumentation

The sampling site (39.80°N, 116.96°E; 15 m MSL) was at Xianghe Atmospheric Observatory of the Institute of Atmospheric Physics, Chinese Academy of Sciences. This site is a rural site surrounded by agricultural fields and located between Beijing and Tianjin, which are the two most important megacities in the NCP region (Fig. 1). The site’s location means that it experiences plumes derived from local agricultural activities or regional transport from urban areas.Measurements were performed from 9 June to 9 July 2013.

The TD-HR-AMS system was deployed at the site to characterize the chemical compositions, mass concentrations and volatility of non-refractory particulate matter (NRPM) (DeCarlo et al., 2006; Canagaratna et al., 2007). A detailed description of the setup and operation of the TDHR-AMS during measurements can be found in our previous studies (Cao et al., 2018, 2019). Briefly, a PMcyclone was used to remove coarse particles and introduce the ambient air into the TD-HR-AMS and a scanning mobility particle sizer system (3775 CPC and 3080 DMA, TSI Inc.)through a copper tube with a flow of 10 L min. The TD was positioned upstream of the HR-ToF-AMS, and the flow rate passing through the TD was 0.45 L minwith a residence time of ～27.9 s in the heating section. The TD included a heating section and denuder section. The temperatures in the heating section were set at 48°C, 95°C, 143°C and 192°C, corresponding to the measured temperatures of 50°C, 100°C, 150°C and 200°C, respectively. Ambient aerosols can pass through directly (bypass flow) or through the TD section (TD flow), and are then dried by a nafion to eliminate the potential influence of relative humidity on the particle collection efficiency (CE) (Matthew et al., 2008).The ionization efficiency (IE) and size calibrations were performed with size-selected pure ammonium-nitrate particles every two weeks. A composition-dependent CE was applied to the data based on the method of Middlebrook et al.(2012). The NR-PMspecies were quantified by the Vmode with unit mass resolution, while the high-resolution mass spectral data were obtained by the W-mode. We operated the HR-ToF-AMS in four menus: bypass flow in Vmode, TD flow in V-mode, TD flow in W-mode, and bypass flow in W-mode, each with a sampling time of 2 min.Besides the TD-HR-AMS system, an aethalometer (AE-31,Magee Inc.) was simultaneously used to measure refractory black carbon (BC) with a temporal resolution of 5 min. The wavelength of 880 nm was used to calculate the mass concentration of BC. The total measured BC can be divided into BC derived from traffic emissions (BC_tr) and biomass burning (BC_bb) based on the aerosol absorption as described in Sandradewi et al., (2008), shown in Text S1 in the Electronic Supplementary Material (ESM).

2.2. HR-ToF-AMS data routine processing and positive matrix factorization analysis

The HR-ToF-AMS data routine analysis was performed using the software SQUIRREL (version 1.57) and PIKA (version 1.16) downloaded from the ToF-AMS software downloads webpage (http://cires1.colorado.edu/jimenezgroup/ToFAMSResources/ToFSoftware/index.html)written in Igor Pro 6.37 (Wave Metrics Inc.). The relative IEs for organics, nitrates and chlorides were assumed to be 1.4, 1.1 and 1.3, respectively. A composition-dependent CE was applied to the data based on the method of Middlebrook et al. (2012). Organic elemental analyses, such as the oxygen-to-carbon ratio (O:C), and hydrogen-to-carbon ratio, are determined by the latest procedures proposed by Canagaratna et al. (2015).

2.3. Estimation of ON mass concentration

3. Results and discussion

3.1. PM1 composition and OA source apportionment

The daily average ambient PMmass concentration was 51.2 ± 48.0 μg m(mean ± standard deviation), ranging from 1.5 to 466 μg m, shown in Fig. 2a. The biggest contributor of PMmass loading during summer 2013 was organic (35.4%), followed by sulfate (31.3%), ammonium(12.8%) and nitrate (11.4%) (Fig. 2b). Figure 2c indicates the variation in relative contributions of different species as a function of the total PMmass loading. We can see that organics show a continuously increasing fraction when PMwas accumulating, implying OA played a more important role in extremely polluted periods. Therefore, we discuss the source apportionment of OA further below.

Fig. 1. Location of the sampling site (Xianghe).

Fig. 2. (a) Time series of PM1 species. (b) Pie chart showing the average chemical compositions. (c)Evolutions of PM1 compositions (left-hand axis) as a function of PM1 mass concentration, and the probability distributions of PM1 mass concentration (white line to the right-hand axis).

3.2. Estimating results of ON and correlation analysis

Figure 5a shows scatterplots of NOversus OA factors resolved by PMF analysis, BC from biomass burning (BC_bb), and BC from traffic emissions (BC_tr). We find that NOhad good correlations with BBOA (r =0.71) and BC_bb (r = 0.67), but a poor correlation with LOOOA (r = 0.20), which is quite different from the results in other regions that show the highest ON correlation with LOOOA (Xu et al., 2015a, b; Yu et al., 2019). Particulate ON formation is found to be through photooxidation of biogenic VOCs in the presence of NOin the daytime (Teng et al., 2015, 2017) and NOradicals oxidation of biogenic VOCs at night (Fry et al., 2013; Ayres et al., 2015; Boyd et al., 2015; Xu et al., 2015b; Lee et al., 2016; Yu et al., 2019).However, recent studies show that NOradicals reacting with typical VOCs in biomass burning plumes could also produce a substantial fraction of particulate ON (Ahern et al.,2019; Joo et al., 2019). The good correlation between ON and biomass burning aerosols in this study indicate the possible existence of different formation mechanisms of ON relevant to biomass burning plumes in the real atmosphere.Here, we further compared the diurnal variation of NO, inorganic nitrates (NO), BBOA and BC_bb in Fig. 4c. First, the result shows that NOhad a quite different diurnal variation from NO, implying that ON has been well separated from inorganic nitrates in this study. Furthermore, NOincreased by nearly two times from 1700 LST to 2200 LST and maintained a relative high mass loading level during the nighttime. We note that there were two similar peaks at 2100–2200 LST and 0300–0400 LST in the NO, BBOA and BC_bb variation trends. A number of studies have proposed that nighttime biomass burning contributes to OA compositions in field campaigns, in particular with some specific ON species formation (Allan et al., 2010; Iinuma et al., 2010, 2016;Mohr et al., 2013). However, these studies did not describe the influence of biomass burning on overall ON. In order to better understand ON in this case, we will attempt to characterize it from the property of volatility in the following section.

Fig. 3. (a) MS profiles of the OA factors. (b) Pie chart showing the average OA components. (c) Diurnal patterns of the OA components.

Table 1. Summary of ON estimations using the NO+/NO2+ ratio method and the PMF method.

3.3. Volatility characteristics of OA factors and ON

The mass fraction remaining (MFR) of different OA factors resolved by PMF analysis, ON and inorganic nitrates are shown in Fig. 6. MFRs varied differently among different OA factors. The MFR of HOA was 0.58 at 50°C and decreased by 1.70 % °C; then the evaporation rate slowed down from 50°C to 200°C with a nearly constant rate of 0.35% °C, and only 4.9% was left at 200°C. BBOA had a similar MFR variation to HOA, with a fast decrease from ambient temperature to 50°C (evaporation rate was 2% °C)and a slower decrease from 50°C to 200°C (0.5% °Cfor 50°C–150°C and 0.006% °Cfor 150°C–200°C), but much wider standard deviation (SD) areas at temperature stages,suggesting BBOA contained more compounds that have different evaporation, which agrees with the BBOA factor in this case consisting of fresh and aged ones. The MFR of LO-OOA at 50°C was 0.72, lower than that of MO-OOA(0.80) and with increased temperature, and the MFR of MO-OOA decreased much slower than LO-OOA and other OA factors, both implying that MO-OOA was less volatile compared to other OA factors. The volatility sequence of OA factors in this study was HOA ＞ BBOA ＞ LO-OOA ＞MO-OOA, determined by the MFR at 50°C (Cao et al.,2018, 2019; Xu et al., 2019).

Fig. 4. (a) Time series of NO3, org concentration estimated by the NO+/NO2+ ratio method and PMF method for the study period. (b) Correlations between NO3_org_ratio_1 and NO3_org_PMF. (c) Diurnal trends of NO3_org_ratio_1, BC_bb(left-hand axis), inorganic nitrates (NO3_inorg) and BBOA (right-hand axis).

Fig. 5. (a) Correlations of NO3, org1_ratio with OA factors resolved by PMF, BC from biomass burning (BC_bb), and BC from traffic emissions (BC_tr). (b) Time series of NO3, org concentration estimated by the NO+/N ratio method(NO3_org), BC_bb (left axis) and BBOA (right-hand axis).

4. Conclusions

The ON were quantified by two methods, i.e.,ratio method and PMF method. Both methods provided reasonable and comparable results in abstracting ON functionality from total measured nitrates. The nitrate functionality of ON (i.e., NO) accounted for about 7.8%–12% of total measured nitrates (NO). Furthermore, the mass concentration of ON was estimated assuming the molecular weight of bulk ON, and they contributed a substantial fraction(8.1%–19%) to total OA. ON show good correlations with BBOA (r = 0.71) and BC from biomass burning (r = 0.67)but poor correlation with LO-OOA (r = 0.20). High levels of biomass burning emissions at night and several similar nighttime peaks of ON, BBOA and BC from biomass burning in diurnal trends suggest ON were influenced by biomass burning. The volatility analysis shows that the overall ON were more volatile than other OA factors at 50°C.Based on these results, we can conclude that particulate ON in rural NCP areas are more relevant to primary emissions,with biomass burning as the most likely source responsible.

Fig. 6. Variation of the average MFR of OA factors (a–d) resolved by PMF, ON (NO3_org) (e), and inorganic nitrates(f) with the TD temperature. The shaded regions indicate the average ± SD.

Acknowledgements

. This work was supported by the Ministry of Science and Technology of China (Grant No.2017YFC0210004), the National Natural Science Foundation of China (Grant No. 91744202), and the China Postdoctoral Science Foundation and Guangdong Province Outstanding Young Talents for the International Education & Development Plan: Post-Doctoral Program.

Electronic supplementary material:

Supplementary material is available in the online version of this article at https://doi.org/10.1007/s00376-020-0127-2.