High contribution of anthropogenic combustion sources to atmospheric inorganic reactive nitrogen in South China evidenced by isotopes

Due to the intense release of reactive nitrogen (Nr) from anthropogenic activity, the source layout of atmospheric nitrogen aerosol has changed. To comprehensively clarify the level, sources, and environmental fate of NH${}_{\mathrm{4}}^{+}$ and NO${}_{\mathrm{3}}^{-}$, their concentrations and stable isotopes (δ¹⁵N) in fine particulate matter (PM_2.5) were measured in a subtropical megacity of South China. The inorganic nitrogen (NH${}_{\mathrm{4}}^{+}$ and NO${}_{\mathrm{3}}^{-})$ was an essential part of atmospheric nitrogen aerosol, and the N-NH${}_{\mathrm{4}}^{+}$ and N-NO${}_{\mathrm{3}}^{-}$ contributed 45.8 % and 23.2 % to total nitrogen (TN), respectively. The source contributions of NH${}_{\mathrm{4}}^{+}$ and NO${}_{\mathrm{3}}^{-}$ were estimated by δ¹⁵N, suggesting that the dominant sources were from anthropogenic combustion activities, including coal combustion, biomass burning, and vehicles, contributing 63.2 % and 88.3 % to NH${}_{\mathrm{4}}^{+}$ and NO${}_{\mathrm{3}}^{-}$, respectively. In particular, biomass burning was the predominant source of NH${}_{\mathrm{4}}^{+}$ (27.9 %), whereas coal combustion was the dominant source of NO${}_{\mathrm{3}}^{-}$ (40.4 %). This study emphasized the substantial impacts of human activities on inorganic Nr. With the rapid development of industry and transportation, nitrogen emissions will be even higher. The promotion of clean energy and efficient use of biomass would help to reduce nitrogen emissions and alleviate air pollution.

1 Introduction

Nitrogenous aerosols are ubiquitous in environment and play an important role as nutrients in ecosystems (Bhattarai et al., 2019). With the massive combustion of fossil fuels and the development of livestock, the proportion of total nitrogen (TN) in particulate matter (PM) ranges from 1.2 % to 17.0 % and has shown a rapid increase in the last few decades (Bhattarai et al., 2019; Galloway et al., 2004; Holland et al., 1999). Mostly nitrogenous aerosols formed from atmospheric reactive nitrogen (Nr) will be deposited into terrestrial and aquatic ecosystems (Huang et al., 2015). Excessive external nitrogen deposition accelerates nitrogen loss in soil, decreases species diversity, disturbs terrestrial ecosystems, and leads to eutrophication in aquatic ecosystems (Breemen, 2002; Wedin and Tilman, 1996; Yang et al., 2015). Furthermore, nitrogenous aerosols have adverse impacts on the climate, air quality, and human health (Bhattarai et al., 2019; Song et al., 2021).

As inorganic Nr, N-NO${}_{\mathrm{3}}^{-}$ and N-NH${}_{\mathrm{4}}^{+}$ are dominant species in the deposition of nitrogen (Zhu et al., 2015); N-NH${}_{\mathrm{4}}^{+}$ was the highest in nitrogen deposition, and NH${}_{\mathrm{4}}^{+}$ was gradually considered to be an important component of secondary inorganic aerosols (SIA) (Sun et al., 2021). Ammonia (NH₃), the precursor of NH${}_{\mathrm{4}}^{+}$, is a vital atmospheric alkaline gas, which can participate in nucleation to promote new particle generation, and can react with acid gas to produce ammonium sulfate and ammonium nitrate (Dunne et al., 2016; Fu et al., 2017). The excessive NH₃ emission from anthropogenic sources will partially offset the benefits of reducing SO₂ and NO_x and trigger urban haze in China (Sun et al., 2021; Meng et al., 2018; Pan et al., 2018a). In many urban environments, NO${}_{\mathrm{3}}^{-}$ has replaced sulfate as the component with the highest proportion in SIA. Nitrogen oxides (NO_x), precursors of NO${}_{\mathrm{3}}^{-}$, are also closely related to the formation of atmospheric oxidants and exert important effects on atmospheric oxidation. In addition, NH₄NO₃ in PM plays an increasingly important role in promoting the formation of sulfate and organic matter, and it has profound effect on the physical and chemical properties of PM (Liu et al., 2021 2020; Hodas et al., 2014). Therefore, to mitigate nitrogen deposition and air pollution, the control of NH${}_{\mathrm{4}}^{+}$ (NH₃) and NO${}_{\mathrm{3}}^{-}$ (NO_x) should not be neglected.

Considerable efforts have been made to comprehensively understand the budget of atmospheric NH${}_{\mathrm{4}}^{+}$ and NO${}_{\mathrm{3}}^{-}$; δ¹⁵N is effective to quantify source contributions of nitrogenous species (Elliott et al., 2007). The anthropogenic combustion sources (combustion of coal, biomass, and gasoline) play a key role in the emission of NO${}_{\mathrm{3}}^{-}$ (NO_x) in many regions of China, suggested by δ¹⁵N (Zong et al., 2020), which also have large effects on NH₃ (Chen et al., 2022b). Ammonia (NH₃) is released by agricultural sources (agricultural activity and livestock) and non-agricultural sources (fossil fuel combustion and vehicles) (Bhattarai et al., 2019). A previous study showed that agricultural sources were the dominant source (80 %–90 %) of NH₃ in China (Kang et al., 2016). However, NH₃ emissions from agricultural sources have been reduced due to intensive farming and efficient fertilization (Wang et al., 2022). The incomplete burning of biomass leads to massive NH₃ emissions and is gradually becoming the second largest non-agricultural source of NH₃ (Yu et al., 2020), which may be responsible for the lag of the decline in the deposition of air pollutants behind the reduction in emission of precursors (Zhao et al., 2022b). Biomass burning in the suburbs also has a potential impact on urban NH₃ (Xiao et al., 2020). As for urban NH₃, combustion sources (including coal combustion, vehicle emissions, and biomass burning) were gradually becoming dominant sources in recent years verified by δ¹⁵N–NH_x (NH₃+ NH${}_{\mathrm{4}}^{+})$ (Xiao et al., 2020; Pan et al., 2018b). In addition, the super clean emission of coal-fired power plants and strict emission standards of vehicles will change the source layout of NH${}_{\mathrm{4}}^{+}$ and NO${}_{\mathrm{3}}^{-}$. Selective catalytic reduction technology equipped with vehicles and industrial boilers reduces NO_x but increases NH₃ emissions (Meng et al., 2017; Pan et al., 2016). The occurrence of haze in North China was closely related to NH₃ emissions from combustion sources (Pan et al., 2018a, b). As the main components of SIA, NH${}_{\mathrm{4}}^{+}$ and NO${}_{\mathrm{3}}^{-}$ play a vital role in the formation of secondary aerosol (Meng et al., 2017), so it is necessary to revisit their sources.

Reactive nitrogen (Nr) emissions from densely populated subtropical areas increased rapidly with the high development of industry and transportation (Wang et al., 2013). Guangzhou is the core megacity in the South subtropical region of China, where the atmospheric environment is complex and the atmospheric oxidation level is high (Tan et al., 2019). The high emissions of inorganic nitrogen from anthropogenic combustion sources have serious and profound impacts on the environment. In this study, we aimed to comprehensively clarify the level of inorganic Nr and revisit the source layout of atmospheric inorganic Nr.

2 Experimental and theoretical methods

2.1 Sampling and chemical concentration analysis

Fine particulate matter (PM_2.5) samples (n=66) were collected from May 2017 to June 2018 in Guangzhou (23.13 ^∘N, 113.27 ^∘E). Details of the sample collection can be found in our previous study (Jiang et al., 2021a). The chemical components including water-soluble ions (i.e., NH${}_{\mathrm{4}}^{+}$, K⁺, Na⁺, Ca²⁺, Mg²⁺, Cl⁻, NO${}_{\mathrm{3}}^{-}$, and SO${}_{\mathrm{4}}^{\mathrm{2}-})$, organic carbon (OC), element carbon (EC), and organic molecular markers (e.g., levoglucosan) were analyzed in our previous studies (Sect. S1 in the Supplement) (Jiang et al., 2021a, b). Moreover, meteorological parameters (temperature, relative humidity (RH), atmospheric pressure, and wind speed) and the concentration of trace gases (CO, SO₂, NO, NO₂, and O₃) were acquired by online instruments (details shown in Sect. S1). A circular punch (r=1 cm) of the sample filter was wrapped in a tin boat and then measured in an elemental analyzer to determine the concentrations of TN.

2.2 Isotope analysis

The δ¹⁵N–NO${}_{\mathrm{3}}^{-}$ and δ¹⁸O–NO${}_{\mathrm{3}}^{-}$ values in PM_2.5 were analyzed by methods of nitrous oxide (N₂O), which was described in a previous study in detail (Zong et al., 2017). Briefly, NO${}_{\mathrm{3}}^{-}$ was reduced to NO${}_{\mathrm{2}}^{-}$ using cadmium powder and an imidazole solution, and N₂O was made by adding NaN₃ to a NO${}_{\mathrm{2}}^{-}$ solution. The production of 75 nmol N₂O gas was needed for measurement. The N₂O gas produced by above processes was measured by the MAT253 stable isotope mass spectrometer. The values of δ¹⁸O and δ¹⁵N were expressed in per mil (‰) shown in Eqs. (1) and (2), relative to the international oxygen and nitrogen isotope standard, respectively.

$$\begin{array}{}\text{(1)}& \mathit{\delta }{}^{\mathrm{15}}\mathrm{N}=\left[{\displaystyle \frac{({}^{\mathrm{15}}\mathrm{N}/{}^{\mathrm{14}}\mathrm{N}{)}_{\mathrm{sample}}}{({}^{\mathrm{15}}\mathrm{N}/{}^{\mathrm{14}}\mathrm{N}{)}_{\mathrm{standard}}}}-\mathrm{1}\right]\cdot \mathrm{1000}\end{array}$$

$$\begin{array}{}\text{(2)}& \mathit{\delta }{}^{\mathrm{18}}\mathrm{O}=\left[{\displaystyle \frac{({}^{\mathrm{18}}\mathrm{O}/{}^{\mathrm{16}}\mathrm{O}{)}_{\mathrm{sample}}}{({}^{\mathrm{18}}\mathrm{O}/{}^{\mathrm{16}}\mathrm{O}{)}_{\mathrm{standard}}}}-\mathrm{1}\right]\cdot \mathrm{1000}\end{array}$$

The δ¹⁵N–NH${}_{\mathrm{4}}^{+}$ was measured by methods of hypobromite oxidation coupled with the reduction of hydroxylamine hydrochloride (Sun et al., 2021). Briefly, NH${}_{\mathrm{4}}^{+}$ was oxidated to NO${}_{\mathrm{2}}^{-}$ using alkaline hypobromite (BrO⁻), and N₂O was made by adding sodium arsenite and hydrochloric acid to a NO${}_{\mathrm{2}}^{-}$ solution. The production of 120 nmol N₂O gas was needed for measurement. The N₂O gas produced by above processes was measured by the MAT253 stable isotope mass spectrometer. The values of δ¹⁵N were expressed in per mil (‰) in Eq. (1). To ensure the stability of the instrument, standard samples were tested for every 10 samples. The standard deviation of replicates was generally less than 0.4 ‰, 0.8 ‰, and 0.5 ‰ for δ¹⁵N–NO${}_{\mathrm{3}}^{-}$, δ¹⁸O–NO${}_{\mathrm{3}}^{-}$, and δ¹⁵N–NH${}_{\mathrm{4}}^{+}$, respectively. The instrumental values of δ¹⁵N–NO${}_{\mathrm{3}}^{-}$ and δ¹⁸O–NO${}_{\mathrm{3}}^{-}$ were corrected by multi-point correction (δ¹⁸O r²=0.99, δ¹⁵N r²=0.999) based on international standards (IAEA-NO-3, USGS32, USGS34, and USGS35). The measured values of δ¹⁵N–NH${}_{\mathrm{4}}^{+}$ were also corrected by multi-point correction (r²=0.999) based on international standards (IAEA-N1, USGS25, and USGS26). In addition, ⁷Be and ²¹⁰Pb were acquired and details were shown in Sect. S1.

2.3 IsoSource and Bayesian mixing model

2.3.1 IsoSource model

The IsoSource model, released by the Environmental Protection Agency (EPA), could calculate ranges of source contributions to a mixture based on the conservation of isotopic mass when the number of sources is too large to permit a unique solution and provide the distribution of source proportions (Phillips et al., 2005). The IsoSource model coupled with δ¹⁵N–NH₃ of atmospheric initial and potential sources (shown in Table 1) were applied to quantify the contribution of various sources to NH₃. Nitrogen fertilizer applications, livestock, human waste, biomass burning, coal combustion, and vehicles were considered as sources of NH₃ in this study (details shown in Sect. S2). Atmospheric initial δ¹⁵N–NH₃ was calculated by following Eq. (3).

$$\begin{array}{}\text{(3)}& \begin{array}{rl}{\mathit{\delta }}^{\mathrm{15}}\mathrm{N}& -{\mathrm{NH}}_{\mathrm{3}-\mathrm{initial}}={\mathit{\delta }}^{\mathrm{15}}\mathrm{N}-{\mathrm{NH}}_{\mathrm{4}}^{+}-\mathit{\epsilon }\left({\mathrm{NH}}_{\mathrm{4}}^{+}-{\mathrm{NH}}_{\mathrm{3}}\right)\\ & \times (\mathrm{1}-f),\end{array}\end{array}$$

where δ¹⁵N–NH${}_{\mathrm{4}}^{+}$ and δ¹⁵N–NH_3−initial represent the δ¹⁵N of particulate NH${}_{\mathrm{4}}^{+}$ and atmospheric initial NH₃, respectively; ε (NH${}_{\mathrm{4}}^{+}$–NH₃) represents the isotope fractionation factor in the gaseous NH₃ conversion to particulate NH${}_{\mathrm{4}}^{+}$ in the atmosphere. The f value represents the proportion of the initial NH₃ converted to NH${}_{\mathrm{4}}^{+}$, referring to NH₃ and NH${}_{\mathrm{4}}^{+}$ observed in Guangzhou (Liao et al., 2014).

The ε(NH${}_{\mathrm{4}}^{+}$–NH₃) value is temperature dependent (Huang et al., 2019), which can be deduced from Urey (1947), as shown in Eq. (4). The atmospheric average temperature was 24.5 ^∘C in our sampling period, and the corresponding ε(NH${}_{\mathrm{4}}^{+}$–NH₃) value was 34.2 ‰ calculated by Eq. (4). In addition, the ε(NH${}_{\mathrm{4}}^{+}$–NH₃) in Guangzhou was estimated to be 32.4 ‰ according to Eq. (8). Equation (8) was deduced by Eqs. (5)–(7). According to Eq. (8), a linear fitting equation was observed between fNH${}_{\mathrm{4}}^{+}$ and δ¹⁵N–NH${}_{\mathrm{4}}^{+}$ (Fig. S1), and the absolute value of the slope (32.4 ‰) was equal to ε(NH${}_{\mathrm{4}}^{+}$–NH₃). The ε(NH${}_{\mathrm{4}}^{+}$–NH₃) average of the two methods (34.2 ‰ and 32.4 ‰) was 33.3 ‰ and approximated to the experimental isotope enrichment factor (33 ‰) (Heaton et al., 1997). Therefore, 33 ‰ was used for deducing the δ¹⁵N of the initial NH₃.

$$\begin{array}{}\text{(4)}& {\mathit{\epsilon }}_{({\mathrm{NH}}_{\mathrm{4}}^{+}-{\mathrm{NH}}_{\mathrm{3}})}=\mathrm{12.4678}\cdot {\displaystyle \frac{\mathrm{1000}}{T+\mathrm{273.15}}}-\mathrm{7.6694}\end{array}$$

$$\begin{array}{}\text{(5)}& {\mathit{\delta }}^{\mathrm{15}}\mathrm{N}-{\mathrm{NH}}_{\mathrm{4}}^{+}-{\mathit{\delta }}^{\mathrm{15}}\mathrm{N}-{\mathrm{NH}}_{\mathrm{3}}={\mathit{\epsilon }}_{({\mathrm{NH}}_{\mathrm{4}}^{+}-{\mathrm{NH}}_{\mathrm{3}})}\end{array}$$

$$\begin{array}{}\text{(6)}& f{\mathrm{NH}}_{\mathrm{4}}^{+}+f{\mathrm{NH}}_{\mathrm{3}}=\mathrm{1}\end{array}$$

$$\begin{array}{}\text{(7)}& \begin{array}{rl}{\mathit{\delta }}^{\mathrm{15}}\mathrm{N}-{\mathrm{NH}}_{\mathrm{4}}^{+}& \cdot f{\mathrm{NH}}_{\mathrm{4}}^{+}+\left({\mathit{\delta }}^{\mathrm{15}}\mathrm{N}-{\mathrm{NH}}_{\mathrm{4}}^{+}-{\mathit{\epsilon }}_{({\mathrm{NH}}_{\mathrm{4}}^{+}-{\mathrm{NH}}_{\mathrm{3}})}\right)\\ & \cdot \left(\mathrm{1}-f{\mathrm{NH}}_{\mathrm{4}}^{+}\right)={\mathit{\delta }}^{\mathrm{15}}\mathrm{N}\end{array}\end{array}$$

$$\begin{array}{}\text{(8)}& \begin{array}{rl}{\mathit{\delta }}^{\mathrm{15}}\mathrm{N}-{\mathrm{NH}}_{\mathrm{4}}^{+}& =-{\mathit{\epsilon }}_{({\mathrm{NH}}_{\mathrm{4}}^{+}-{\mathrm{NH}}_{\mathrm{3}})}\cdot f{\mathrm{NH}}_{\mathrm{4}}^{+}\\ & +\left({\mathit{\delta }}^{\mathrm{15}}\mathrm{N}+{\mathit{\epsilon }}_{({\mathrm{NH}}_{\mathrm{4}}^{+}-{\mathrm{NH}}_{\mathrm{3}})}\right)\end{array}\end{array}$$

Here, T represents the atmospheric temperature (^∘C); δ¹⁵N–NH${}_{\mathrm{4}}^{+}$ and δ¹⁵N–NH₃ represent the δ¹⁵N of particulate NH${}_{\mathrm{4}}^{+}$ and atmospheric NH₃, respectively; δ¹⁵N represents the sum of δ¹⁵N–NH${}_{\mathrm{4}}^{+}$ and δ¹⁵N–NH₃; fNH₃ and fNH${}_{\mathrm{4}}^{+}$ represent the proportion of atmospheric NH₃ and particulate NH${}_{\mathrm{4}}^{+}$, respectively.

2.3.2 Bayesian mixing model

The δ¹⁵N were used for tracing sources based on the conservation of isotopic mass. The Bayesian mixing model improved upon linear mixing models by explicitly considering uncertainty in prior information and isotopic equilibrium fractionation. Recently, the Bayesian mixing model was applied to trace the sources of atmospheric pollutants (Zong et al., 2017, 2020). The model coupled with δ¹⁵N–NO${}_{\mathrm{3}}^{-}$ and δ¹⁸O–NO${}_{\mathrm{3}}^{-}$ was used to identify the formation process and quantify the source contributions of NO${}_{\mathrm{3}}^{-}$.

In the central Pearl River Delta (PRD), NO${}_{\mathrm{3}}^{-}$ formed through ⋅OH and N₂O₅ pathways contributed to 94 % simulated by the Community Multiscale Air Quality (CAMQ) model (Qu et al., 2021). In this study, only ⋅OH and N₂O₅ formation pathways were considered. Details of the NO${}_{\mathrm{3}}^{-}$ formation pathway were also shown in Sect. S2. The atmospheric δ¹⁸O–NO${}_{\mathrm{3}}^{-}$ can be expressed by Eq. (9). The [δ¹⁸O–HNO₃]_OH can be further expressed by Eq. (10), assuming no kinetic isotope fractionation (Walters and Michalski, 2016), and [δ¹⁸O–HNO₃]${}_{{\mathrm{H}}_{\mathrm{2}}\mathrm{O}}$ can be estimated by Eq. (11) (Walters and Michalski, 2016). The δ¹⁸O values in tropospheric H₂O, NO_x, O₃, and OH were within a certain range. The tropospheric δ¹⁸O–H₂O, δ¹⁸O–NO_x, δ¹⁸O–O₃, and δ¹⁸O–OH ranged from −25 ‰ to 0 ‰ (Baskaran et al., 2011; Walters and Michalski, 2016), 112 ‰ to 122 ‰ (Michalski et al., 2014; Walters and Michalski, 2016), 90 ‰ to 122 ‰, and −15 ‰ to 0 ‰, respectively (Fang et al., 2011; Johnston and Thiemens, 1997). Therefore, the γ (the contribution of the ⋅OH formation pathway) can be estimated by fNO₂ and the oxygen isotope fractionation i.e., αNO${}_{\mathrm{2}}/$ NO, αOH $/$ H₂O, and αN₂O₅ $/$ NO₂. The oxygen isotope fractionations are temperature dependent and can be estimated by Eq. (13) and Table S1. The fNO₂ varied from 0.20 to 0.95 (Zong et al., 2017; Walters et al., 2016). Based on δ¹⁸O–NO${}_{\mathrm{3}}^{-}$, δ¹⁸O–H₂O, δ¹⁸O–NO_x, δ¹⁸O–O₃, and temperature (Eqs. 9–13), γ (maximum γ and minimum γ) was estimated by Monte Carlo simulation nested in the Bayesian mixing model (Zong et al., 2017). Assuming no kinetic isotope fractionation, the nitrogen isotope fractionation value in the formation process of NO${}_{\mathrm{3}}^{-}$ (εN) was calculated by Eqs. (13)–(16) combined with γ and temperature (Zong et al., 2017; Walters and Michalski, 2016; Walters et al., 2016). The εN value in our sampling period was 5.1±2.5 ‰, which was comparable to that in Beijing(average 6.5 ‰) (Fan et al., 2020). The contributions of different sources to atmospheric NO_x were quantified by the Bayesian mixing model coupled with εN, δ¹⁵N–atmospheric-NO${}_{\mathrm{3}}^{-}$, and δ¹⁵N–NO_x endmembers shown in Table 1. We considered coal combustion, mobile traffic sources, biomass burning, and soil microbial process as dominant atmospheric NO_x sources in Guangzhou (details shown in Sect. S2). The specific details of the Bayesian mixing model were reported by our previous studies (Zong et al., 2017, 2020).

$$\begin{array}{}\text{(9)}& \begin{array}{rl}{\mathit{\delta }}^{\mathrm{18}}\mathrm{O}& -{\mathrm{NO}}_{\mathrm{3}}^{-}=\mathit{\gamma }\times {\left[{\mathit{\delta }}^{\mathrm{18}}\mathrm{O}-{\mathrm{NO}}_{\mathrm{3}}^{-}\right]}_{\mathrm{OH}}+(\mathrm{1}-\mathit{\gamma })\\ & \times {\left[{\mathit{\delta }}^{\mathrm{18}}\mathrm{O}-{\mathrm{NO}}_{\mathrm{3}}^{-}\right]}_{{\mathrm{H}}_{\mathrm{2}}\mathrm{O}}=\mathit{\gamma }\times {\left[{\mathit{\delta }}^{\mathrm{18}}\mathrm{O}-{\mathrm{HNO}}_{\mathrm{3}}\right]}_{\mathrm{OH}}\\ & +(\mathrm{1}-\mathit{\gamma })\times {\left[{\mathit{\delta }}^{\mathrm{18}}\mathrm{O}-{\mathrm{HNO}}_{\mathrm{3}}\right]}_{{\mathrm{H}}_{\mathrm{2}}\mathrm{O}}\end{array}\end{array}$$

$$\begin{array}{}\text{(10)}& \begin{array}{rl}& {\left[{\mathit{\delta }}^{\mathrm{18}}\mathrm{O}-{\mathrm{HNO}}_{\mathrm{3}}\right]}_{\mathrm{OH}}={\displaystyle \frac{\mathrm{2}}{\mathrm{3}}}{\left[\left({\mathit{\delta }}^{\mathrm{18}}\mathrm{O}-{\mathrm{NO}}_{\mathrm{2}}\right)\right]}_{\mathrm{OH}}+{\displaystyle \frac{\mathrm{1}}{\mathrm{3}}}\\ & {\left[{\mathit{\delta }}^{\mathrm{18}}\mathrm{O}-\mathrm{OH}\right]}_{\mathrm{OH}}={\displaystyle \frac{\mathrm{2}}{\mathrm{3}}}\left[{\displaystyle \frac{\mathrm{1000}\times \left({}^{\mathrm{18}}{\mathit{\alpha }}_{{\mathrm{NO}}_{\mathrm{2}}/\mathrm{NO}}-\mathrm{1}\right)\left(\mathrm{1}-f_{{\mathrm{NO}}_{\mathrm{2}}}\right)}{\left(\mathrm{1}-f_{{\mathrm{NO}}_{\mathrm{2}}}\right)+\left({}^{\mathrm{18}}{\mathit{\alpha }}_{{\mathrm{NO}}_{\mathrm{2}}/\mathrm{NO}}\times f_{{\mathrm{NO}}_{\mathrm{2}}}\right)}}\right.\\ & \left.+\left[{\mathit{\delta }}^{\mathrm{18}}\mathrm{O}-{\mathrm{NO}}_X\right]\right]+{\displaystyle \frac{\mathrm{1}}{\mathrm{3}}}[\left({\mathit{\delta }}^{\mathrm{18}}\mathrm{O}-{\mathrm{H}}_{\mathrm{2}}\mathrm{O}\right)\\ & +\mathrm{1000}\times \left({}^{\mathrm{18}}{\mathit{\alpha }}_{\mathrm{OH}/{\mathrm{H}}_{\mathrm{2}}\mathrm{O}}-\mathrm{1}\right)]\end{array}\end{array}$$

$$\begin{array}{}\text{(11)}& {\left[{\mathit{\delta }}^{\mathrm{18}}\mathrm{O}-{\mathrm{HNO}}_{\mathrm{3}}\right]}_{{\mathrm{H}}_{\mathrm{2}}\mathrm{O}}={\displaystyle \frac{\mathrm{5}}{\mathrm{6}}}\left({\mathit{\delta }}^{\mathrm{18}}\mathrm{O}-{\mathrm{N}}_{\mathrm{2}}{\mathrm{O}}_{\mathrm{5}}\right)+{\displaystyle \frac{\mathrm{1}}{\mathrm{6}}}\left({\mathit{\delta }}^{\mathrm{18}}\mathrm{O}-{\mathrm{H}}_{\mathrm{2}}\mathrm{O}\right)\end{array}$$

$$\begin{array}{}\text{(12)}& {\mathit{\delta }}^{\mathrm{18}}\mathrm{O}-{\mathrm{N}}_{\mathrm{2}}{\mathrm{O}}_{\mathrm{5}}={\mathit{\delta }}^{\mathrm{18}}\mathrm{O}-{\mathrm{NO}}_{\mathrm{2}}+\mathrm{1000}\times \left({}^{\mathrm{18}}{\mathit{\alpha }}_{{\mathrm{N}}_{\mathrm{2}}{\mathrm{O}}_{\mathrm{5}}/{\mathrm{NO}}_{\mathrm{2}}}-\mathrm{1}\right)\end{array}$$

$$\begin{array}{}\text{(13)}& \begin{array}{rl}\mathrm{1000}\left({}^m{\mathit{\alpha }}_{X/Y}-\mathrm{1}\right)& ={\displaystyle \frac{\mathrm{A}}{T^{\mathrm{4}}}}\times {\mathrm{10}}^{\mathrm{10}}+{\displaystyle \frac{\mathrm{B}}{T^{\mathrm{3}}}}\times {\mathrm{10}}^{\mathrm{8}}+{\displaystyle \frac{\mathrm{C}}{T^{\mathrm{2}}}}\\ & \times {\mathrm{10}}^{\mathrm{6}}+{\displaystyle \frac{\mathrm{D}}{T}}\times {\mathrm{10}}^{\mathrm{4}}\end{array}\end{array}$$

$$\begin{array}{}\text{(14)}& \begin{array}{rl}\mathit{\epsilon }\mathrm{N}& =\mathit{\gamma }\times \mathit{\epsilon }{\left({\mathit{\delta }}^{\mathrm{15}}\mathrm{N}-{\mathrm{NO}}_{\mathrm{3}}^{-}\right)}_{\mathrm{OH}}+\left(\mathrm{1}-\mathit{\gamma }\right)\\ & \times \mathit{\epsilon }{\left({\mathit{\delta }}^{\mathrm{15}}\mathrm{N}-{\mathrm{NO}}_{\mathrm{3}}^{-}\right)}_{{\mathrm{H}}_{\mathrm{2}}\mathrm{O}}=\mathit{\gamma }\times \mathit{\epsilon }{\left({\mathit{\delta }}^{\mathrm{15}}\mathrm{N}-{\mathrm{HNO}}_{\mathrm{3}}\right)}_{\mathrm{OH}}\\ & +\left(\mathrm{1}-\mathit{\gamma }\right)\times \mathit{\epsilon }{\left({\mathit{\delta }}^{\mathrm{15}}\mathrm{N}-{\mathrm{HNO}}_{\mathrm{3}}\right)}_{{\mathrm{H}}_{\mathrm{2}}\mathrm{O}}\end{array}\end{array}$$

$$\begin{array}{}\text{(15)}& \begin{array}{rl}\mathit{\epsilon }& {\left({\mathit{\delta }}^{\mathrm{15}}\mathrm{N}-{\mathrm{HNO}}_{\mathrm{3}}\right)}_{\mathrm{OH}}=\mathit{\epsilon }{\left({\mathit{\delta }}^{\mathrm{15}}\mathrm{N}-{\mathrm{NO}}_{\mathrm{2}}\right)}_{\mathrm{OH}}\\ & =\mathrm{1000}\times \left[{\displaystyle \frac{\left({}^{\mathrm{15}}{\mathit{\alpha }}_{{\mathrm{NO}}_{\mathrm{2}}/\mathrm{NO}}-\mathrm{1}\right)\left(\mathrm{1}-f_{{\mathrm{NO}}_{\mathrm{2}}}\right)}{\left(\mathrm{1}-f_{{\mathrm{NO}}_{\mathrm{2}}}\right)+\left({}^{\mathrm{15}}{\mathit{\alpha }}_{{\mathrm{NO}}_{\mathrm{2}}/\mathrm{NO}}\times f_{{\mathrm{NO}}_{\mathrm{2}}}\right)}}\right]\end{array}\end{array}$$

$$\begin{array}{}\text{(16)}& \begin{array}{rl}\mathit{\epsilon }& {\left({\mathit{\delta }}^{\mathrm{15}}\mathrm{N}-{\mathrm{HNO}}_{\mathrm{3}}\right)}_{{\mathrm{H}}_{\mathrm{2}}\mathrm{O}}=\mathit{\epsilon }{\left({\mathit{\delta }}^{\mathrm{15}}\mathrm{N}-{\mathrm{N}}_{\mathrm{2}}{\mathrm{O}}_{\mathrm{5}}\right)}_{{\mathrm{H}}_{\mathrm{2}}\mathrm{O}}\\ & =\mathrm{1000}\times \left({}^{\mathrm{15}}{\mathit{\alpha }}_{{\mathrm{N}}_{\mathrm{2}}{\mathrm{O}}_{\mathrm{5}}/{\mathrm{NO}}_{\mathrm{2}}}-\mathrm{1}\right)\end{array}\end{array}$$

Here, γ is the contribution of the ⋅OH formation pathway to NO${}_{\mathrm{3}}^{-}$, and εN is the nitrogen isotope fractionation value. In the total NO_x, fNO₂ is the fraction of NO₂; ¹⁸α${\mathrm{NO}}_{\mathrm{2}}/\mathrm{NO}$, ¹⁸α$\mathrm{OH}/\mathrm{H}$₂O, and ¹⁸αN₂${\mathrm{O}}_{\mathrm{5}}/{\mathrm{NO}}_{\mathrm{2}}$ are the oxygen isotope equilibrium fractionation factors between NO₂ and NO, ⋅OH and H₂O, and N₂O₅ and NO₂, respectively. The nitrogen isotope equilibrium fractionation factors between NO₂ and NO and N₂O₅ and NO₂ are ¹⁵α${\mathrm{NO}}_{\mathrm{2}}/\mathrm{NO}$ and ¹⁵αN₂${\mathrm{O}}_{\mathrm{5}}/{\mathrm{NO}}_{\mathrm{2}}$, respectively.

Table 1 The estimation of δ¹⁵N–NH₃ and δ¹⁵N–NO_x from various sources.

3 Results and discussion

3.1 Concentration and seasonal variation of NH${}_{\mathrm{4}}^{+}$ and NO${}_{\mathrm{3}}^{-}$

The concentration of NH${}_{\mathrm{4}}^{+}$ and NO${}_{\mathrm{3}}^{-}$ in PM_2.5 was 1.6±1.3 µg m⁻³ and 2.8±3.4 µg m⁻³, contributing 18.7 % and 32.6 % to SIA. The concentration of N-NH${}_{\mathrm{4}}^{+}$ and N-NO${}_{\mathrm{3}}^{-}$ was 1.2±1.0 µg m⁻³ and 0.6±0.8 µg m⁻³, contributing 45.8 % and 23.2 % to TN, respectively. Thus, NH${}_{\mathrm{4}}^{+}$ and NO${}_{\mathrm{3}}^{-}$ were an essential part of nitrogen aerosols; NH${}_{\mathrm{4}}^{+}$ and NO${}_{\mathrm{3}}^{-}$ showed similar seasonal variations with higher concentrations in winter than in summer (Fig. 1). During winter, the air mass was often dry and cold with low wind speed, which meant the decrease of the atmospheric self-purification capability. In addition, primary combustion sources related to fossil fuel and biomass burning always showed significant increase in North China during winter, which greatly increased the concentration of atmospheric pollutants in Guangzhou by long-range transportation. However, during summer, the air mass from the sea was relatively clean with high wind speed facilitating the diffusion of pollutants. Moreover, high temperature in summer was conducive to the decomposition of NH₄NO₃ (Song et al., 2008). Thus, the levels of NH${}_{\mathrm{4}}^{+}$ and NO${}_{\mathrm{3}}^{-}$ were lower in summer. In addition, concentrations of NH${}_{\mathrm{4}}^{+}$ and NO${}_{\mathrm{3}}^{-}$ in our study were lower than in North China (Beijing – Wu et al., 2019; Fan et al., 2022; Tianjin – Xiang et al., 2022; Shijiazhuang – Xiang et al., 2022; and Harbin – Sun et al., 2021), East China (Nanchang – Xiao et al., 2020), and Central China (Wuhan and Changsha – Xiao et al., 2020; Zong et al., 2020), suggesting that the level of air pollution in Guangzhou has been alleviated to a certain extent. Therefore, it is necessary to conduct a comprehensive study on the emission sources of NH${}_{\mathrm{4}}^{+}$ and NO${}_{\mathrm{3}}^{-}$ to take more effective measures to mitigate air pollution.

Figure 1 (a) The concentration and δ¹⁵N of NH${}_{\mathrm{4}}^{+}$ and (b) the concentration and δ¹⁵N and δ¹⁸O of NO${}_{\mathrm{3}}^{-}$.

3.2 Characteristic and seasonal variation in δ¹⁵N–NH${}_{\mathrm{4}}^{+}$ and source apportionment of NH${}_{\mathrm{4}}^{+}$

The δ¹⁵N–NH${}_{\mathrm{4}}^{+}$ values over Guangzhou ranged from 2.1 ‰ to 45.5 ‰, with an annual mean of 20.2 ‰±10.1 ‰. In our study, the δ¹⁵N–NH${}_{\mathrm{4}}^{+}$ values were comparable to those at suburban sites (Fig. S2) such as sites in Japan (22.1 ‰±8.3 ‰, 16.1 ‰±6.6 ‰; Kawashima and Kurahashi, 2011) and Korea (Jeju Island, 17.4 ‰±4.9 ‰; Kundu et al., 2010) but heavier than those in polluted regions, such as Guangzhou during the summer haze (average 7.17 ‰; Liu et al., 2018) and Beijing (−37.1 ‰ to 5.8 ‰; Pan et al., 2016). The δ¹⁵N–NH${}_{\mathrm{4}}^{+}$ values were lower in autumn (17.3 ‰) and winter (14.4 ‰) than in spring (22.5 ‰) and summer (25.7 ‰), which was similar to the trends in Japan (Kawashima and Kurahashi, 2011).

The seasonal differences in δ¹⁵N–NH${}_{\mathrm{4}}^{+}$ values were significant between warm (summer and spring) and cold seasons (winter and autumn) (p<0.05). The δ¹⁵N–NH${}_{\mathrm{4}}^{+}$ was affected by the ratio of NH${}_{\mathrm{4}}^{+}$ $/$ (NH₃+NH${}_{\mathrm{4}}^{+})$ (Eq. 8 and Fig. S1). A linear fitting equation was observed between NH${}_{\mathrm{4}}^{+}$ $/$ (NH₃+NH${}_{\mathrm{4}}^{+})$ and δ¹⁵N–NH${}_{\mathrm{4}}^{+}$, and the absolute value of the slope (32.4) approximated the isotope equilibrium fractionation value (33 %) between atmospheric NH₃ and NH${}_{\mathrm{4}}^{+}$ (Fig. S1). The linear fitting suggested that the lower the NH${}_{\mathrm{4}}^{+}$ proportion was, the heavier the δ¹⁵N–NH${}_{\mathrm{4}}^{+}$ value was. The lower NH${}_{\mathrm{4}}^{+}$ level was in accordance with the higher δ¹⁵N–NH${}_{\mathrm{4}}^{+}$ in summer, which was the opposite of winter. In addition, a previous study suggested that the marked variation in δ¹⁵N–NH${}_{\mathrm{4}}^{+}$ values was largely controlled by the emission sources of NH₃, the precursor gas of NH${}_{\mathrm{4}}^{+}$ (Liu et al., 2018). According to the δ¹⁵N–NH${}_{\mathrm{4}}^{+}$ results, the source of NH${}_{\mathrm{4}}^{+}$ was assigned as biomass burning (27.9 % ± 16.4 %), coal combustion (16.0 % ± 3.9 %), vehicles (19.8 % ± 5.3 %), fertilizer (10.9 % ± 6.1 %), livestock (12.7 % ± 5.8 %), and urban waste (11.9 % ± 6.1 %) as shown in Fig. 2a.

Figure 2 The source apportionment results of atmospheric NH${}_{\mathrm{4}}^{+}$ (a) and NO${}_{\mathrm{3}}^{-}$ (b) in Guangzhou, and the comparison of sources results between NH${}_{\mathrm{4}}^{+}$ and NO${}_{\mathrm{3}}^{-}$ (c).

In our study, non-agriculture sources were the dominators of NH${}_{\mathrm{4}}^{+}$ (75.6 %). Unexpectedly, the contribution of biomass burning was the highest. In particular, from late June to July, the contribution of biomass burning enhanced, which possibly resulted from sugarcane leaf burning. The δ¹⁵N in sugarcane leaves was as high as 38 ‰ (Martinellia et al., 2002). The δ¹⁵N of NH${}_{\mathrm{4}}^{+}$ formed from NH₃ released by sugarcane leaf burning was 44.1 ‰ (Sect. S3), which was consistent with the highest δ¹⁵N–NH${}_{\mathrm{4}}^{+}$ values (45.5 ‰ and 45.1 ‰) in July. In PRD, south winds prevail in July and the sampling site is located downwind of sugarcane planting area. Therefore, the air mass to the sampling site might carry the pollutants related to sugarcane leaf burning. K⁺ is a typical biomass burning tracer (Cui et al., 2018). Considering the impact of primary emission intensity, [NH${}_{\mathrm{4}}^{+}$ $/$ EC] and [K⁺ $/$ EC] were used to calculate the correlation coefficient (r = 0.435, p < 0.01), which verified that NH${}_{\mathrm{4}}^{+}$ was influenced by biomass burning. In recent years, biomass burning has been gradually identified as an important source of NH${}_{\mathrm{4}}^{+}$ (Meng et al., 2017; Xiao et al., 2020). The results based on emission inventories showed that the contribution of residential biomass combustion to NH₃ ranged from 33 % to 53 % in China (Meng et al., 2017). According to δ¹⁵N, biomass burning contributed 18 % (Harbin, Northeast China – Sun et al., 2021), 46 % (Wuhan, Central China), 40 % (Changsha, South Central China – Xiao et al., 2020), 35 % (Nanchang, East China – Xiao et al., 2020), and 23 % (Guangzhou, South China – Chen et al., 2022a) to NH${}_{\mathrm{4}}^{+}$. Particularly, in Guangzhou, the contribution of biomass burning in the ground was higher than that in the Guangzhou tower with a height of 488 m, suggesting the influence of regional biomass burning (Chen et al., 2022a). Furthermore, ⁷Be mainly originates from upper atmosphere, whereas ²¹⁰Pb is derived from terrestrial surface (Jiang et al., 2021b). High levels of ⁷Be observed in the ground suggested the sink influence of the upper atmosphere; ⁷Be and ²¹⁰Pb are chemically stable and with unique sources, which can effectively reflect the transport of continental air mass and the air exchange between stratosphere and troposphere. In our study, the correlation coefficient between NH${}_{\mathrm{4}}^{+}$ and ²¹⁰Pb (r=0.701, p<0.01) was higher than that between NH${}_{\mathrm{4}}^{+}$ and ⁷Be (r=0.432, p<0.01), suggesting that NH${}_{\mathrm{4}}^{+}$ was mainly affected by regional emission. Therefore, biomass burning exerted essential influence on NH${}_{\mathrm{4}}^{+}$ levels, which should no longer be ignored.

In addition, with the acceleration of urbanization, combustion sources related to fossil fuels have become the main sources of NH₃. In previous studies, the source of NH_x (NH₃+ NH${}_{\mathrm{4}}^{+})$ was mainly from agricultural activity due to the rough way of farming (Chang et al., 2016; Pan et al., 2020). However, with the improvement of efficient fertilization practices, agricultural NH₃ decreased significantly (Wang et al., 2022). Fossil fuels, such as coal and gasoline, are major energies for production and domestic use, and their contribution to NH₃ has become increasingly important. In North China, fossil fuel combustion contributed 92 % to NH₃ during hazes (Zhang et al., 2020; Pan et al., 2016). In a previous study of Guangzhou, the contribution of NH₃ from fossil sources in ground observations (43 %) was higher than that observed in the Guangzhou tower (18 %), indicating the importance of locally related fossil fuel combustion sources (Chen et al., 2022a). In our study, vehicle emission and coal combustion contributed 19.8 % ± 5.3 % and 16.0 % ± 3.9 % of NH${}_{\mathrm{4}}^{+}$, respectively, which was lower than that in North China but higher than agricultural sources. The share of NH₃ from vehicle exhaust emissions was estimated to be 18.8 % based on the emission factor of NH₃ from on-road vehicles in Guangzhou, which was similar to our results (Liu et al., 2014). The selective catalytic reduction process for vehicles can reduce NO_x but increase the emission of NH₃, which has been confirmed as an important source of NH₃ (Heeb et al., 2006; Meng et al., 2017). Despite the efforts of government to promote electric vehicles in recent years, their share is still relatively low (about 5 %). As car ownership increases, this has an important impact on atmospheric NH₃. Coal combustion was the second most important source of fossil combustion after vehicle emissions in our study, although the contribution was lower than in North China (Wu et al., 2019; Zhang et al., 2020; Pan et al., 2016). The absence of heating in Guangzhou may explain the lower contribution of coal combustion compared to North China. On an annual basis, the contribution of fossil-fuel-related combustion sources in our study (35.8 %) was comparable to that in North China (37 %–52 %) (Pan et al., 2018a).

The source contributions of NH${}_{\mathrm{4}}^{+}$ in our study were compared to other regions (shown in Fig. S3). The combustion-related sources (biomass burning, coal combustion, and vehicle) have gradually become the dominant source of urban atmospheric NH₃. Biomass burning and vehicle emissions could emit massive carbon monoxide (CO) (Li and Wang, 2007; Wang et al., 2005). In Guangzhou, NH${}_{\mathrm{4}}^{+}$ was positively related to CO (r=0.637, p<0.01), which confirmed that combustion sources played a key role in NH${}_{\mathrm{4}}^{+}$. From a historical perspective, NH₃ emissions from anthropogenic combustion and industry have been steadily increasing since 1960 (Meng et al., 2017). The optimization of energy structure and encouragement of the development of new energy vehicles would be hopeful to reduce NH₃. The results of this study would be conducive to reducing NH₃ scientifically and effectively and would relieve the pressure on the reduction from agricultural sources.

3.3 Characteristic and seasonal variation in δ¹⁸O–NO${}_{\mathrm{3}}^{-}$ and δ¹⁵N–NO${}_{\mathrm{3}}^{-}$ and source apportionment of NO${}_{\mathrm{3}}^{-}$

3.3.1 Seasonal variation of δ¹⁸O–NO${}_{\mathrm{3}}^{-}$

The δ¹⁸O–NO${}_{\mathrm{3}}^{-}$ in Guangzhou was 68.1 ± 9.7 ‰ (44.9 ‰ to 90.5 ‰) comparable to that in precipitation (66.3 ‰, ranging from 33.4 ‰ to 86.2 ‰) (Fang et al., 2011), but lower than those regions with weak light intensity, such as Beihuangcheng Island (ranging from 49.4 ‰ to 103.9 ‰) (Zong et al., 2017) and the Bermuda islands (cold season 76.9 ‰±6.3 ‰) (Hastings et al., 2003). In this study, δ¹⁸O–NO${}_{\mathrm{3}}^{-}$ was higher in winter and spring than in summer and autumn, which was similar to the seasonal variation in δ¹⁸O–NO${}_{\mathrm{3}}^{-}$ in previous studies (Fang et al., 2011; Gobel et al., 2013). On the one hand, the δ¹⁸O–NO${}_{\mathrm{3}}^{-}$ value was associated with the formation pathways of NO${}_{\mathrm{3}}^{-}$. The results simulated by the Bayesian mixing model suggested that the contributions of the N₂O₅ channel to NO${}_{\mathrm{3}}^{-}$ were 56.8 %, 58.9 %, 29.2 %, and 27.0 % in winter, spring, autumn, and summer, respectively. The δ¹⁸O value of NO${}_{\mathrm{3}}^{-}$ formed by the N₂O₅ channel is higher than that by the ⋅OH pathway (Sect. S2). The nights in cold seasons were longer than those in warm seasons, which favored NO${}_{\mathrm{3}}^{-}$ formation through the N₂O₅ channel. In addition, the illumination intensity was weakened in cold seasons compared with that in warm seasons, which constrained the production of ⋅OH (Zong et al., 2020; Tan et al., 2019; Wang et al., 2017). Thus, the contribution of the N₂O₅ channel in cold seasons was higher than that in warm seasons. Furthermore, concentration of NO${}_{\mathrm{3}}^{-}$ was high when contribution of the N₂O₅ channel enhanced (Fig. 3), suggesting that NO${}_{\mathrm{3}}^{-}$ pollution was related to the N₂O₅ hydrolysis pathway. The air mass to Guangzhou was derived from the South China Sea in summer and the North Continental Region in winter. The higher δ¹⁸O–NO${}_{\mathrm{3}}^{-}$ and NO${}_{\mathrm{3}}^{-}$ concentration might be affected by long-range and high-altitude transport from North China, which might carry abundant precursors. Massive NO${}_{\mathrm{3}}^{-}$ could be formed by N₂O₅ hydrolysis at high altitude and transported to the ground. The index of f (⁷Be,²¹⁰Pb) was expressed in Sect. S1 and could reflect the influence of atmospheric dynamic transport on aerosol pollutants (Jiang et al., 2021b). Generally, air masses with low values of f(⁷Be,²¹⁰Pb) suggested that pollutants were associated with continental surface emission, whereas high f(⁷Be, ²¹⁰Pb) were influenced by long-range transport from upper air masses. The contribution of the N₂O₅ channel was positively correlated with f(⁷Be,²¹⁰Pb) (r=0.319, p<0.05), indicating the long-range transport influence of upper air mass on the N₂O₅ channel. For example, on 25 January 2018, the contribution of the N₂O₅ channel (nitrate) was 81.1 % (3.6 µg m⁻³) when the upper air mass was from North China. However, on 7 July 2017, the N₂O₅ channel (nitrate) contributed only 5.7 % (0.5 µg m⁻³), corresponding to the air mass mainly from the South China Sea transported at low altitude (Fig. S4).

Figure 3 (a) The contribution of the OH radical oxidation and N₂O₅ hydrolysis pathway to NO${}_{\mathrm{3}}^{-}$. (b) The vertical position of dots corresponded to the contribution of N₂O₅ pathway and the size of the dots corresponded to the concentration of NO${}_{\mathrm{3}}^{-}$.

The δ¹⁸O–NO${}_{\mathrm{3}}^{-}$ decreased from 76.7 ‰ in 2014 to 68.1 ‰ in 2017–2018 (Zong et al., 2020), which indicated that the ⋅OH channel became more important in Guangzhou. The enhanced contribution of the ⋅OH pathway indicated the increasing atmospheric oxidation capacity. In recent years, although the concentration of PM_2.5 in Guangzhou has significantly decreased, the photochemical pollution caused by high O₃ concentrations was not optimistic (Tan et al., 2019). The O₃ concentration in the PRD showed a fluctuating upward trend from 2013 to 2020; especially in 2017–2018, O₃ concentrations were at high levels (Environmental Status Bulletin of Guangdong Province Fig. S5). In our study, the NO${}_{\mathrm{3}}^{-}$ formation pathway inferred from δ¹⁸O–NO${}_{\mathrm{3}}^{-}$ proved the enhancement of the atmospheric oxidation capacity.

3.3.2 Seasonal variation of δ¹⁵N–NO${}_{\mathrm{3}}^{-}$ and source apportionment of NO${}_{\mathrm{3}}^{-}$

Seasonal variation of δ¹⁵N–NO${}_{\mathrm{3}}^{-}$

The δ¹⁵N–NO${}_{\mathrm{3}}^{-}$ in Guangzhou was 4.9 ‰±2.2 ‰ (−0.4 ‰ to 10.8 ‰), which was similar to the wet deposition (Fang et al., 2011). The δ¹⁵N–NO${}_{\mathrm{3}}^{-}$ was comparable to that from the northeastern United States (6.8 ‰) (Elliott et al., 2009) and lower than regions in China, where NO${}_{\mathrm{3}}^{-}$ was predominantly derived from anthropogenic sources, such as Heshan in Guangdong (7.50 ‰±3.30 ‰) (Su et al., 2020), Beihuangcheng Island (8.2 ‰±6.2 ‰) (Zong et al., 2017), and Beijing (12.1 ‰±3.3 ‰) (Fan et al., 2022). Nevertheless, the δ¹⁵N–NO${}_{\mathrm{3}}^{-}$ in this study was significantly higher than those from clean background regions, where NO${}_{\mathrm{3}}^{-}$ was mainly from natural sources, such as the coast of Antarctica ($-\mathrm{12.0}\mathrm{‰}\pm \mathrm{15.6}\mathrm{‰}$) (Savarino et al., 2007) and Bermuda ($-\mathrm{2.1}\mathrm{‰}\pm \mathrm{1.5}\mathrm{‰}$ warm season, $-\mathrm{5.9}\mathrm{‰}\pm \mathrm{3.3}\mathrm{‰}$ cold season) (Hastings et al., 2003). The values of δ¹⁵N–NO${}_{\mathrm{3}}^{-}$ in winter, spring, summer, and autumn were 5.6 ‰, 5.3 ‰, 4.4 ‰, and 4.5 ‰, respectively. The δ¹⁵N–NO${}_{\mathrm{3}}^{-}$ in winter and summer showed significant difference (p<0.05). The values of δ¹⁵N–NO${}_{\mathrm{3}}^{-}$ were influenced by atmospheric processes and emission sources (Elliott et al., 2009). For the N₂O₅ channel, NO${}_{\mathrm{3}}^{-}$ is characterized by higher δ¹⁵N values (Freyer et al., 1993; Elliott et al., 2009). The N₂O₅ channel was the predominant formation pathway of NO${}_{\mathrm{3}}^{-}$ in winter, which was in accordance with the seasonal variation in δ¹⁵N–NO${}_{\mathrm{3}}^{-}$. In addition, the difference in δ¹⁵N–NO${}_{\mathrm{3}}^{-}$ reflected the variation in the emission source of NO${}_{\mathrm{3}}^{-}$. The δ¹⁵N–NO_x from coal combustion was relatively high. In winter, the higher δ¹⁵N–NO${}_{\mathrm{3}}^{-}$ was probably related to long-range transport from North China, where coal combustion enhanced during winter.

Source apportionment of NO${}_{\mathrm{3}}^{-}$

Based on the Bayesian mixing model coupled with δ¹⁵N–NO${}_{\mathrm{3}}^{-}$, NO${}_{\mathrm{3}}^{-}$ sources were assigned as coal combustion 40.4 % ± 8.7 %, biomass burning 25.6 % ± 2.1 %, mobile sources (vehicles) 22.3 % ± 3.1 %, and microbial process 11.7 % ± 3.8 %. Figures 2b and S6 showed the source contribution of NO${}_{\mathrm{3}}^{-}$ in Guangzhou and other regions in China, respectively. Compared to earlier periods (2013–2014), the concentration of NO${}_{\mathrm{3}}^{-}$ from vehicles and coal combustion decreased significantly (Zong et al., 2020), which resulted from the stricter vehicle emission standard, promotion of new energy electric vehicles, and ultraclean transformation of coal combustion (Guangdong Province, 2014; http://www.gd.gov.cn/gkmlpt/content/0/142/5 mpost_142687.html, last access: 14 February 2014; Tang et al., 2019). However, almost all production and domestic segments rely on energy generated from coal combustion, which was still the dominant source of NO${}_{\mathrm{3}}^{-}$ in 2017–2018. Coal combustion was affected not only by local emissions but also external air mass transmission. The contribution of coal combustion was higher in winter than in summer, which probably related to the long-range transportation from North China. Taking 10 January 2018 as an example, the contribution of coal combustion sources to NO${}_{\mathrm{3}}^{-}$ was 67.5 %, and the corresponding air mass was from North China and transmitted to Guangzhou through high altitude. However, the air mass on 26 July 2017 was mainly from the South China Sea, which was transmitted through low altitude to Guangzhou. The contribution of coal burning to NO${}_{\mathrm{3}}^{-}$ on 26 July 2017 was 28.5 % lower than that on 10 January 2018.

As a non-fossil combustion source, biomass burning was also an important source of NO${}_{\mathrm{3}}^{-}$ and accounted for 25.6 %. The contribution of biomass burning and vehicles was stable throughout a year. Generally, high intensity biomass burning occurred in winter in the Guangdong province (dry season, i.e., from November to March) (Xu et al., 2019); K⁺ is a typical tracer of biomass burning. The concentration of K⁺ enhanced in winter (0.4 µg m⁻³) was higher than that in summer (0.2 µg m⁻³) and autumn (0.2 µg m⁻³), respectively, indicating enhancement of biomass burning intensity. Also, NO${}_{\mathrm{3}}^{-}$ concentration of biomass burning remarkably enhanced in winter (1.2 µg m⁻³) and was higher than that in summer (0.4 µg m⁻³) and autumn (0.3 µg m⁻³), respectively. However, coal combustion also enhanced in winter due to the demand for heating in North China. Our sampling site was influenced by the air mass with a high coal combustion contribution from the north by long-range transportation, which may relatively reduce the contribution of biomass burning. Thus, the contribution of biomass burning showed stability compared with coal combustion. Another non-fossil source is related to soil microbial activity and only contributed 11.7 % to NO${}_{\mathrm{3}}^{-}$, which was unexpectedly lower than the results in earlier periods (2013–2014). Generally, the microorganisms in soil emit NO through nitrification or denitrification, which was affected by the amount of carbon and nitrogen nutrients in soil (Hall and Matson, 1996). In earlier periods, due to the higher level of aerosols, the amount of nutrients settling in soil was also higher, which was exemplified by the observation of dry and wet deposition in Guangzhou (He et al., 2022; Zheng et al., 2020). In addition, the reduction of cultivated land from 2013 to 2018 might also reduce the contribution of microbial source emissions. Therefore, emissions from natural sources were also influenced by human activities to some extent. The contribution of microbial processes was higher in summer than in winter. In summer, higher RH and temperature were favorable for the intense activity of soil microorganisms (Zong et al., 2017). The contributions of microbial processes to NO${}_{\mathrm{3}}^{-}$ also decreased in winter compared with summer at regional background sites and five Chinese megacities, including Guangzhou (Zong et al., 2017, 2020).

The source comparisons between NO${}_{\mathrm{3}}^{-}$ and NH${}_{\mathrm{4}}^{+}$ were shown in Fig. 2c. Coal combustion, biomass burning, and vehicles were three significant sources of NO${}_{\mathrm{3}}^{-}$ and NH${}_{\mathrm{4}}^{+}$. Coal combustion and biomass burning were the dominant sources of NO${}_{\mathrm{3}}^{-}$ and NH${}_{\mathrm{4}}^{+}$, respectively. The vehicles were also an important source of atmospheric inorganic Nr that contributed to 22.3 % and 19.8 % of NO${}_{\mathrm{3}}^{-}$ and NH${}_{\mathrm{4}}^{+}$, respectively. Recently, the government has actively taken many measures to reduce the pollution from vehicles, such as stricter automobile emission standards and the promotion of new energy vehicles. However, due to the large vehicle ownership base, the pollutants emitted from vehicles are not optimistic. In addition, vehicle emissions could contribute half of the fresh secondary organic aerosol in urban environments (Zhang et al., 2022; Zhao et al., 2022a).

4 Conclusions

A year-long field observation was conducted in Guangzhou to clarify the atmospheric fate of inorganic nitrogen aerosol. Inorganic nitrogen species were the most essential component of TN including NH${}_{\mathrm{4}}^{+}$ (45.8 %) and NO${}_{\mathrm{3}}^{-}$ (23.2 %), which are also dominant components of SIA and play a key role in China hazes. The δ¹⁵N is a powerful tool to quantify the source contribution of NH${}_{\mathrm{4}}^{+}$ and NO${}_{\mathrm{3}}^{-}$, which suggested that anthropogenic combustion sources (coal combustion, biomass burning, and vehicles) were the dominant sources.

Anthropogenic combustion sources contributed 63.2 % to NH${}_{\mathrm{4}}^{+}$, which is more than the 23.6 % contribution from agricultural sources. Ammonia (NH₃) largely facilitates the formation of sulfate and nitrate. Meanwhile, sulfate and nitrate promote each other with the positive feedback effect, which could trigger hazes. In megacities of China, the focus of NH₃ reduction should be on anthropogenic combustion sources, especially on biomass burning, which might be responsible for the lag of the decline in the deposition of air pollutants behind the reduction in emission (Zhao et al., 2022b). In addition, anthropogenic combustion sources accounted for 88.3 % of NO${}_{\mathrm{3}}^{-}$. Coal combustion and vehicles contributed 40.4 % and 22.3 % to NO${}_{\mathrm{3}}^{-}$, respectively. Despite a series of measures to reduce emissions of NO_x, fossil fuels, as the main energy source for production and living, will still inevitably emit a large amount of NO_x. Our results emphasized that the emission of atmospheric inorganic nitrogen is largely related to anthropogenic combustion sources. The development and promotion of clean energy and efficient use of biomass are conducive to the deep reduction of atmospheric nitrogen.