Investigating atmospheric nitrate sources and formation pathways between heating and non-heating seasons in urban North China

In urban North China, nitrate ( NO3− ) is a primary contributor to haze formation. So far, the production processes and source apportionments of atmospheric NO3− during the heating season (i.e. the wintertime) have not yet been well understood. This study determined δ15N– NO3− , δ18O– NO3− , and Δ17O– NO3− of aerosol samples to compare the potential sources and formation pathways of atmospheric NO3− during heating (November to March) and non-heating (April to May) seasons. Combining stable isotope composition with the MixSIAR model based on Δ17O– NO3− showed that NO3 + DMS/HC (dimethyl sulfate/hydrocarbon) pathway was the dominant process of atmospheric nitrate formation during the heating season (mean = 52.88 ± 16.11%). During the non-heating season, the contributions of NO3 + DMS/HC (mean = 37.89 ± 13.57%) and N2O5 + H2O (mean = 35.24 ± 3.75%) pathways were comparable. We found that Δ17O– NO3− was negatively correlated with wind speed and positively correlated with relative humidity during the heating season, possibly associated with the sources and production of atmospheric NO3− . In specific, in a dust storm event, the very low Δ17O– NO3− is likely associated with particles from land surface. Under the premise of considering 15N fractionation, the constraint-based on δ15N– NO3− illustrated that coal combustion was the major source of NO x emission during the heating season, and the relative contribution of coal combustion decreased rapidly from the heating season (mean = 42.56 ± 15.50%) to the non-heating season (mean = 21.86 ± 4.91%). Conversely, the proportion of NOx emitted by soil microbes rose significantly from the heating (mean = 9.67 ± 5.99%) to non-heating season (mean = 24.02 ± 11.65%). This study revealed differences in the sources and formation processes of atmospheric NO3− during the heating and non-heating seasons, which are of significance to atmospheric nitrogen oxide/nitrate pollution mitigation.


Introduction
Atmospheric nitrate (NO − 3 ) is a major chemical constituent of aerosols, and it can affect terrestrial system nitrogen cycling by wet and dry deposition (Galloway et al 2008, Guo et al 2010, Fowler et al 2013, Ghaly and Ramakrishnan 2015. In the atmosphere, NO − 3 mainly from the oxidation of NO x (NO x = NO + NO 2 ). Figure 1 simply summarizes the photochemical cycle of NO x and formation pathways of NO − 3 in the atmosphere. Upon emitted, NO is immediately oxidized by ozone (O 3 ) or peroxy radical (HO 2 /RO 2 ) to NO 2 (R1-R3). The formation pathways for atmospheric NO − 3 may vary under different environmental conditions (text S1). For example, the NO 2 + OH pathway (P1) is the dominant pathway for NO − 3 formation during the daytime. However, during the nighttime, the NO 2 + OH pathway is unimportant because of the absence of photochemically produced OH radicals. Under this condition, NO 2 is further oxidized by O 3 , which forms nitrate radical (NO 3 ), and NO 3 will react with dimethyl sulfate (DMS) or hydrocarbons (HC) (P2) to directly form HNO 3 or react with NO 2 to form pentoxide (N 2 O 5 ) which will undergo hydrolysis to form HNO 3 via heterogeneous processes on the surface of aerosol particles (P3). In the atmosphere, a variety of sources can contribute to NO − 3 . Lighting, wildfires, soil microbial nitrogen cycle, and stratospheric transmission are the primary natural sources of NO x (Felix et al 2012), while anthropogenic sources of NO x are mainly fossil fuel combustion, biomass combustion, vehicle exhaust and human waste (Fibiger and Hastings 2016). The nitrogen and oxygen stable isotope composition of NO − 3 provides a useful tool for a better understanding of the sources and production pathways of atmospheric NO − 3 . Nitrogen and oxygen isotopic ratio of NO − 3 (δ) was defined as,  (Hoering 1957, Alexander et al 2009, Hastings et al 2013. However, isotopic fractionation of nitrogen may occur during the conversion of NO x to NO − 3 (Freyer 1978, Freyer et al 1993, which leads to δ 15 N-NO − 3 maybe not identical to the primary δ 15 N-NO x (Freyer 1978, Freyer et al 1993, Luo et al 2019, potentially hampering the precise source identification of NO x . In this case, corresponding δ 15 N-NO x values that has been transformed to NO − 3 should be calculated by taking the nitrogen isotopic fractionation (ε) in relation to the observed δ 15 N-NO − 3 values as follows (Song et al 2019).
The ε values are mainly determined by the fractional contributions and isotopic effects of the three main pathways of NO x oxidation to NO − 3 (P1-P3; figure 1 NO 3 , and N 2 O 5 ) via oxidation processes . Therefore, ∆ 17 O-NO − 3 is a more robust indicator of the oxidation pathways involving the main atmospheric oxidants than δ 18 O (Savarino et al 2013).
Heavy air pollution (e.g. haze pollution) frequently occurs during the heating season in North China (Pang et al 2020). Coal combustion for heating emits a large amount of NO x and SO 2 , resulting in increasing concentrations of secondary inorganic aerosols (SO 2− 4 and NO − 3 ) during the heating season, and consequently affecting the atmospheric chemistry. Previous reports suggested that NO − 3 has replaced SO 2− 4 as the most important component in particulate pollution in North China (Fu et al 2020, Yan et al 2021. To date, a number of studies have been performed to investigate the sources and formation mechanisms of atmospheric NO − 3 with the aid of nitrogen and oxygen isotopes in urban areas (Zong et al 2017, Luo et al 2019, Fan et al 2020, Feng et al 2020. However, uncertainties remained regarding the source identification and production of NO − 3 as a result of isotopic effects associated with NO x /NO − 3 chemistry and the overlap of δ 15 N among varied sources. In this study, we aim to quantitatively understand the source and formation pathways of atmospheric NO − 3 from heating season to non-heating season in North China with the MixSIAR model, driven by the stable isotopes of NO − 3 . Here, we quantified the contribution of the major oxidizing pathways to atmospheric NO − 3 by ∆ 17 O-NO − 3 , and then estimated the isotopic fractionation of δ 15 N (ε), that corrected the N isotope effect of the conversion of NO x to NO − 3 . The δ 15 N-NO − 3 analysis in the MixSIAR model was applied to quantify the contribution of the main potential sources of NO x . In addition, we discussed the potential effects of meteorological factors on NO − 3 concentration and its isotopic composition.

Field sampling
The sampling site (figure 2) was on the roof of a building (the height of ∼30 m) in urban Tianjin, a metropolitan city in north China (39 • 3 ′ 36.42 ′′ N, 117 • 7 ′ 23.06 ′′ E). In this city, the heating period is from the beginning of November to the end of March in the next year. For a comparison to the heating season, atmospheric NO − 3 was also collected from April to the end of May, 2021, referred to as the nonheating season. The samples were collected using a low-volume air sampler (ARA N-FRM Sampler, ARA Instruments, USA) with a flow rate of 16.7 L·min −1 . Quartz filters (Whatman 1851-047), which were calcined for 5 h in a muffle furnace at ∼500 • C, were used to collect atmospheric NO − 3 . The sampling interval was about 24 h, and the sampling was conducted once a week, generally at the beginning of each week. In total, 29 samples were collected. After sampling, the filters were folded in half and wrapped in aluminum foil before being placed in polyerhylene bags and stored at −20 • C for further processing and analysis. The details on chemical and isotopic analysis are in text S2. In addition, two blank filters were prepared by installing the filters in the ARAN-FRM Sampler without pumping. Gaseous HNO 3 is widely assumed to be absorbed by aerosols on the glass and quartz fiber filters (Frey et al 2009, Savarino et al 2016, He et al 2020. Thus, the atmospheric NO − 3 collected here was taken as the sum of gaseous HNO 3 and particulate NO − 3 . It is acknowledged that the samples collected during April and May cannot represent the full non-heating season in North China, and here we only make a comparison with the results of heating season.

Meteorological data and the trace gases
The meteorological data and the trace gases concentrations during the sampling period were also collected. The daily meteorological data, including temperature, ultraviolet radiation, wind speed, relative humidity was from National Climate Data Center (http://cdc.cma.gov.cn). The daily O 3 and NO 2 concentrations in the study area were obtained from the China National Urban air quality real-time publishing platform (https://air.cnemc.cn:18014/). In this study, the Bayesian stable isotope mixing model (MixSIAR) based on R (https://github.com/ brianstock/MixSIAR) was used to estimate the contribution fractions of different sources and production pathways of atmospheric NO − 3 (text S3).

Calculation of N isotope fractionation (ε)
Assuming no kinetic isotope fractionation is associated with the conversion of NO x to NO − 3 , the isotope fractionation ε incorporated in the model can be estimated as follows (Song et al 2021b).
where f P1 , f P2 and f P3 are the relative contribution of NO 2 + OH, NO 3 + HC/DMS and f NO2 and f NO were respectively assumed to be 0.6 ± 0.04 and 0.4 ± 0.04 (Song et al 2021b). F NO is the ratio of the emitted NO to emitted NO x , averaging 0.60 ± 0.22 (Burling et al 2010, Stockwell et al 2014). More details on the calculation refer to text S4.

Results
Concentrations and isotopic composition of NO − 3 , as well as environmental parameters during the study period, are shown in figure 3. Concentration of NO − 3 was relatively stable over time (figure 3(d)), with the means of 18.14 ± 17.38 and 11.83 ± 9.54 µg·m −3 during the heating and non-heating seasons, respectively. The independent sample t-test showed that there was no significant difference in NO − 3 concentration between the two seasons (p = 0.23; figure 4(a)), suggesting that the concentration of NO − 3 did not change significantly in a short period of time after the heating season. Different from concentration, there was a downward trend in δ 15 N-NO − 3 from the heating to non-heating season. A significant difference in δ 15 N-NO − 3 (p < 0.01; figure 4(b)) between the heating season (−3.62-17.10‰, mean = 9.37 ± 5.44‰) and the non-heating season (−3.55-7.61‰, mean = 2.15 ± 4.18‰) may suggest the shifts in NO x sources and/or NO − 3 chemistry between the two seasons.
The δ  3 observed in this study may be associated with the particulate matter originated from the land surface, which likely carries ∆ 17 O-NO − 3 of ∼0‰. It is interesting that a negative linear relationship between ∆ 17 O-NO − 3 and wind speed during the heating season (R 2 = 0.56, p < 0.001; figure 5(a)) was observed, possibly suggesting that the strong wind carrying more particulate matter from the ground contributed to atmospheric NO − 3 and consequently lower ∆ 17 O-NO − 3 . In addition to wind speed, relative humidity may also play an important role in the formation of NO − 3 , especially for the heterogeneous hydrolysis reaction of N 2 O 5 (Wahner et al 1998, Kane et al 2001. It is found that the three samples with extremely low ∆ 17 O-NO − 3 were all collected during the days with very low relative humidity (∼20% versus ∼30% for other time period in heating season) ( figure 5(b)). Previous investigations suggested that low relative humidity is not conducive to heterogeneous reaction of nocturnal NO x (Lin et al 2022). Then, the N 2 O 5 pathway, which may play an important role in atmospheric NO − 3 formation in urban sites (Brown and Stutz 2012), would be restricted by the low humidity (figure S2), and consequently the ∆ 17 O-NO − 3 may be lowered. In addition, there was a positive linear relationship between the concentration of NO − 3 and relative humidity during the heating season (R 2 = 0.53, p < 0.001; figure 5(d)), likely indicating the impacts of humidity on atmospheric NO − 3 formation (especially the N 2 O 5 channel). Following the MixSIAR model, the contribution of N 2 O 5 + H 2 O pathway to atmospheric NO − 3 in the three samples was estimated to be 11.90 ± 3.16%, which is lower than the other samples (discussed below).
Contributions of the three main production pathways of atmospheric NO − 3 (NO 2 + OH, NO 3 + DMS/HC, and N 2 O 5 + H 2 O) are shown in figure 6. The contribution of the NO 2 + OH pathway (∼18% and ∼27% in heating and non-heating periods, respectively) was lower than the other two pathways. This may be due to the lack of solar radiation in the cold season. Indeed, the solar UV radiation was relatively weak during the heating season (13.29 ± 4.68 W·m −2 ; figure 3(a)), compared to the non-heating season (25.58 ± 5.81 W·m −2 ; figure 3(a)). In addition, the contribution fraction of NO 2 + OH pathway is generally lower than previous reports in summer (Fan et al 2022, Li et al 2022, when the solar radiation is much stronger and OH concentrations are higher than in winter (Zeroual et al 1995).
Contribution fractions of the NO 3 + DMS/HC pathway were significantly higher during the heating season (∼53%) than during the non-heating season (∼38%; p < 0.05; figure 6(b)). During the heating season, natural gas and coal combustion for heating can release plenty of volatile organic compounds (VOCs) (Wang et al 2013. In urban North China, the mean concentration of VOCs during heating season has been reported to be ∼1.8 times that of clean period (Gu et al 2020), and the contribution of coal combustion to VOCs emission during heating season is about twice than that of nonheating season (Niu et al 2022). The contribution of N 2 O 5 + H 2 O pathway during the non-heating season was significantly higher (p < 0.01; figure 6(c)) than that during the heating season. Considering that the meteorological condition (i.e. relative humidity; figure 3(a)) favorable to N 2 O 5 + H 2 O pathway has not changed significantly during the two seasons, the lower fraction of this channel during heating season may be associated more NO x participating in NO − 3 production via the NO 3 + DMS/HC pathway. The contribution fraction of N 2 O 5 + H 2 O pathway in this study is generally similar to the previous reports in urban North China (Beijing of ∼30%; Fan et al 2022).

3
The ε values estimated by proportion of NO − 3 formation pathways were 4.12 ± 6.72‰ and 13.80 ± 3.64‰ during the heating and non-heating seasons, respectively (figure S3), suggesting that most of the fractionation processes in the formation of NO − 3 from NO x were dominated by the enrichment The contribution of coal combustion (mean = 42.56 ± 15.50%) was higher than the other three sources during the heating season, suggesting that coal combustion was the primary source of NO x . The contribution of coal combustion during the non-heating season (mean = 21.86 ± 4.91%) was significantly lower (p < 0.01; figure 7(a)) than that of the heating season. This may be associated with a significant decrease in coal combustion during the non-heating season. In addition, there was a good linear correlation between the concentrations of SO 2− 4 and NO − 3 (R 2 = 0.79, p < 0.001; figure 8) during the heating season, possibly indicating that SO 2− 4 and NO − 3 had the common source. And SO 2 , as a precursor of SO 2− 4 , has been proved to be predominantly generated from coal combustion (Xu et al 2000, Pyshyev et al 2017. Thus, it further supports that NO − 3 in the atmosphere were mostly generated by NO x from coal combustion in the heating season. However, this correlation during the nonheating season is rather weak (R 2 = 0.47, p = 0.059; figure 8).
The contributions of soil microbial emission during the non-heating season were significantly (p < 0.01; figure 7(b)) higher than that during the heating season. It may be associated with that the low temperature (figure 3(c)) and consequently the weak microbial activities lead to lower emissions of NO x from soil in heating season (Stark 1996). With the ambient temperature gradually rising, enhanced soil microbial activities may lead to increased NO x emission. In addition, March and April are spring ploughing in North China, and N fertilizer input can lead to enhanced NO x emissions from soil (Miller et al 2018, Song et al 2018. The contributions of biomass burning during the heating and non-heating seasons were not significantly different (p > 0.05; figure 7(c)). It is noted that during the heating season, both the contributions of biomass burning and vehicle exhaust were significantly lower than those of coal combustion, but these three contributions were very similar during the non-heating season (about 22-27%). A previous study suggested that non-fossil fuel NO x emissions (including soil microbial emission and biomass burning) may be just as significant as fossil fuel NO x emissions globally (Song et al 2021a). In this study, we found that non-fossil fuel emissions (34.55%) are lower than fossil fuel emissions (65.45%) during the heating season, while these two types of sources are comparable in non-heating seasons. It suggests that the nonfossil fuel emissions are important in atmospheric Figure 8. Relationship between atmospheric SO 2− 4 and NO − 3 during heating and non-heating seasons. The grey and red bands represent 95% confidence intervals during the heating and non-heating seasons, respectively. NO − 3 budget, but its contribution fraction varies seasonally.

Conclusion
Concentrations and isotopic composition of atmospheric NO − 3 were investigated during the heating and non-heating season in urban North China. Concentrations of NO − 3 were comparable in the two seasons, while δ 15 N-NO − 3 and δ 18 O-NO − 3 were much higher in heating season than in non-heating season. It is found that ∆ 17 O-NO − 3 was negatively correlated with wind speed and positively correlated with relative humidity during the heating season, likely suggesting the meteorological factors influencing the sources and production of atmospheric NO − 3 . The extremely low ∆ 17 O-NO − 3 was found in several heating season samples, likely associated with particles from land surface in a dust storm event. It is estimated that the NO 3 + DMS/HC was the dominant pathway for atmospheric NO − 3 production during the heating season, possibly associated with larger VOCs from natural gas and coal combustion for heating. Based on the δ 15 N of NO x and NO − 3 , it is estimated that NO x was primarily from coal combustion during the heating season, but the relative contribution of coal combustion sharply decreased during the nonheating season. The contribution of non-fossil fuel NO x emissions was over 50% during the non-heating season, suggesting an important contribution of this source to the atmospheric NO − 3 budget.

Data availability statement
The data that support the findings of this study are available upon reasonable request from the authors.