Measurement report: High contribution of N₂O₅ uptake to particulate nitrate formation in NO₂-limited urban areas

Particulate nitrate (p${\mathrm{NO}}_{\mathrm{3}}^{-}$) is a major component of fine particles in Chinese urban areas. However, the relative contributions of p${\mathrm{NO}}_{\mathrm{3}}^{-}$ formation pathways in urban areas remain poorly quantified, particularly under the NO₂-limited regime that governs its formation (as defined by the ${\mathrm{NO}}_{\mathrm{2}}/{\mathrm{O}}_{\mathrm{3}}$ ratio), which hinders effective particulate pollution control. In this study, comprehensive winter field observations were conducted in urban Xiamen, Southeast China. We observed significantly elevated nighttime p${\mathrm{NO}}_{\mathrm{3}}^{-}$ levels concurrent with increased N₂O₅ concentrations. Quantification using an observation-constrained model revealed that N₂O₅ uptake contributed 51.2 % to total p${\mathrm{NO}}_{\mathrm{3}}^{-}$ formation, which was comparable to that of the OH+NO₂ reaction. The N₂O₅ uptake was found to be mainly driven by nocturnal NO₃ oxidation capacity (modulated by NO₂ and O₃ levels) rather than by heterogeneous reaction conditions. Sensitivity simulations further demonstrated that p${\mathrm{NO}}_{\mathrm{3}}^{-}$ formation rate was more sensitive to NO_x variations than to VOCs variations. Implementing NO_x control measures at nighttime was shown to effectively reduce p${\mathrm{NO}}_{\mathrm{3}}^{-}$ by abating N₂O₅ uptake while simultaneously preventing daytime O₃ increase. Our findings enhance the understanding of p${\mathrm{NO}}_{\mathrm{3}}^{-}$ formation in NO₂-limited urban areas and provide valuable insights for developing joint PM_2.5 and O₃ mitigation strategies.

1 Introduction

Fine particulate matter (PM_2.5) contributes to various atmospheric environmental issues, including visibility deterioration, radiative forcing change, and adverse impacts on human health (Seinfeld, 1989; Lelieveld et al., 2015). Among its chemical components, particulate nitrate (p${\mathrm{NO}}_{\mathrm{3}}^{-}$) has attracted increasing attention due to its rising mass fraction in PM_2.5 and its nonlinear responses to emission mitigation strategies (Xie et al., 2022; Zhai et al., 2021; Li et al., 2021; Zhang et al., 2021; Zhou et al., 2022; Zong et al., 2022; Wang et al., 2020). The primary formation pathways of p${\mathrm{NO}}_{\mathrm{3}}^{-}$ include gas-phase oxidation through the reaction of hydroxyl radicals (OH) and nitrogen dioxides (NO₂) (R1 and R2), and heterogeneous uptake of dinitrogen pentoxide (N₂O₅) which is produced via NO₂ oxidation by nitrate radicals (NO₃) (R3–R5) (Brown and Stutz, 2012). It is well recognized that the OH+NO₂ reaction dominates in daytime, while N₂O₅ uptake dominates in nighttime. During nocturnal p${\mathrm{NO}}_{\mathrm{3}}^{-}$ formation, particulate chlorides can induce N₂O₅ heterogeneous uptake to produce ClNO₂, thereby competing with p${\mathrm{NO}}_{\mathrm{3}}^{-}$ formation.

$$\begin{array}{}\text{(R1)}& \mathrm{OH}\left(\mathrm{g}\right)+{\mathrm{NO}}_{\mathrm{2}}\left(\mathrm{g}\right)+\mathrm{M}\to {\mathrm{HNO}}_{\mathrm{3}}\left(\mathrm{g}\right)+\mathrm{M}\end{array}$$

$$\begin{array}{}\text{(R2)}& {\displaystyle }& {\displaystyle }{\mathrm{HNO}}_{\mathrm{3}}\left(\mathrm{g}\right)+{\mathrm{NH}}_{\mathrm{3}}\left(\mathrm{g}\right)\rightleftharpoons {\mathrm{NH}}_{\mathrm{4}}{\mathrm{NO}}_{\mathrm{3}}\left(\mathrm{p}\right)\text{(R3)}& {\displaystyle }& {\displaystyle }{\mathrm{NO}}_{\mathrm{2}}\left(\mathrm{g}\right)+{\mathrm{O}}_{\mathrm{3}}\left(\mathrm{g}\right)\to {\mathrm{NO}}_{\mathrm{3}}\left(\mathrm{g}\right)\text{(R4)}& {\displaystyle }& {\displaystyle }{\mathrm{NO}}_{\mathrm{2}}\left(\mathrm{g}\right)+{\mathrm{NO}}_{\mathrm{3}}\left(\mathrm{g}\right)\rightleftharpoons {\mathrm{N}}_{\mathrm{2}}{\mathrm{O}}_{\mathrm{5}}\left(\mathrm{g}\right)\text{(R5)}& {\displaystyle }& {\displaystyle }\begin{array}{rl}{\displaystyle }& {\displaystyle }{\mathrm{N}}_{\mathrm{2}}{\mathrm{O}}_{\mathrm{5}}\left(\mathrm{g}\right)+{\mathrm{H}}_{\mathrm{2}}\mathrm{O}/{\mathrm{Cl}}^{-}\left(\mathrm{p}\right)\to \\ {\displaystyle }& {\displaystyle }(\mathrm{2}-\mathit{\phi }){\mathrm{NO}}_{\mathrm{3}}^{-}\left(\mathrm{p}\right)+\mathit{\phi }{\mathrm{ClNO}}_{\mathrm{2}}\left(\mathrm{g}\right)\end{array}\end{array}$$

Many studies have focused on quantifying the potential formation pathways of p${\mathrm{NO}}_{\mathrm{3}}^{-}$ in urban areas of China. In major urban agglomerations such as the Beijing-Tianjin-Hebei (BTH) region (Chen et al., 2020; Ma et al., 2023; Zhao et al., 2023), Yangtze River Delta (YRD) (Sun et al., 2022; Zhai et al., 2023; Zhang et al., 2023b), and Pearl River Delta (PRD) (Yang et al., 2022; Niu et al., 2022; Cheng et al., 2024), p${\mathrm{NO}}_{\mathrm{3}}^{-}$ formation was typically dominated by the gas-phase oxidation of OH+NO₂. In contrast, under special conditions such as the COVID-19 pandemic and PM_2.5 pollution events (Yan et al., 2023; Zhai et al., 2023), N₂O₅ uptake became the main pathway. Previous research has demonstrated that the formation rate of p${\mathrm{NO}}_{\mathrm{3}}^{-}$ via N₂O₅ uptake is closely related to its precursor NO₂ and O₃, and the N₂O₅ formation can be classified into NO₂-limited and O₃-limited regimes based on the ${\mathrm{NO}}_{\mathrm{2}}/{\mathrm{O}}_{\mathrm{3}}$ ratio (Ma et al., 2023). The winter ${\mathrm{NO}}_{\mathrm{2}}/{\mathrm{O}}_{\mathrm{3}}$ ratios in the BTH, YRD, and PRD regions were generally above 1, placing N₂O₅ formation in the O₃-limited or transition regime (Ma et al., 2023; Wen et al., 2018; Li et al., 2021; Zhang et al., 2023b). However, the N₂O₅ uptake served as the dominant pathway for p${\mathrm{NO}}_{\mathrm{3}}^{-}$ formation, typically under NO₂-limited conditions (e.g., reduced emissions during the pandemic) or under large aerosol surface areas (e.g., severe particulate pollution episodes). Collectively, these findings indicate that spatial variations in NO₂ and O₃ levels are likely a key driver of regional differences in the dominant formation pathways of p${\mathrm{NO}}_{\mathrm{3}}^{-}$. The formation of p${\mathrm{NO}}_{\mathrm{3}}^{-}$ primarily depends on precursors OH, NO₂, and O₃, with OH and O₃ concentrations being influenced by VOCs and NO_x emissions. Thus, the different formation pathways of p${\mathrm{NO}}_{\mathrm{3}}^{-}$ result in complex responses to NO_x/VOCs emissions. The response of p${\mathrm{NO}}_{\mathrm{3}}^{-}$ formation via OH+NO₂ to precursors variation is relatively well-understood, as most Chinese urban areas are located in VOC-limited regimes for O₃ (Wang et al., 2023a, 2022c; Zhang et al., 2023a; Mao et al., 2022), and ammonia-rich regimes for p${\mathrm{NO}}_{\mathrm{3}}^{-}$ (Xing et al., 2018; Sun et al., 2022; Fu et al., 2024; Liu et al., 2019). Under these conditions, VOCs reduction suppresses p${\mathrm{NO}}_{\mathrm{3}}^{-}$ formation by decreasing OH concentrations, whereas NO_x reduction enhances p${\mathrm{NO}}_{\mathrm{3}}^{-}$ formation by weakening the NO_x titration effect. Given the regional variations in the ${\mathrm{NO}}_{\mathrm{2}}/{\mathrm{O}}_{\mathrm{3}}$ ratio across urban areas of China (Ma et al., 2023), the response of p${\mathrm{NO}}_{\mathrm{3}}^{-}$ formation via N₂O₅ uptake to precursor changes (VOCs, O₃) likely exhibits spatial heterogeneity. A recent study has revealed that under O₃-limited conditions for N₂O₅ formation (Zhang et al., 2023b), reducing NO_x emissions had negligible effects, while reducing VOCs decreased the consumption of NO₃ by VOCs, thereby enhancing p${\mathrm{NO}}_{\mathrm{3}}^{-}$ formation from N₂O₅ uptake. However, the response of p${\mathrm{NO}}_{\mathrm{3}}^{-}$ formation to precursors under NO₂-limited conditions remains unclear. Aside from precursor availability, N₂O₅ uptake is also greatly influenced by heterogeneous reaction conditions like aerosol composition and aerosol surface area (Mcduffie et al., 2018b, a; Tham et al., 2018; Yu et al., 2020), which introduces additional uncertainty in determining the contribution of p${\mathrm{NO}}_{\mathrm{3}}^{-}$ formation pathways and the effectiveness of precursor control strategies.

The ${\mathrm{NO}}_{\mathrm{2}}/{\mathrm{O}}_{\mathrm{3}}$ ratios in southeastern China predominantly fell within the NO₂-limited regime for N₂O₅ formation (Ma et al., 2023). Xiamen, as one of the most developed cities in southeastern China, exhibits relatively better air quality with low levels of VOCs and NO_x compared to China's megacities (Table S1 in the Supplement). This pattern well represents the future urban atmospheric conditions following the implementation of air pollution control measures in China. From December 2022–February 2023, we conducted comprehensive multi-parameter observations in urban Xiamen, including N₂O₅ and related chemical constituents. An observation-constrained box model incorporating the heterogeneous reaction parameters was utilized to quantify the rates of different p${\mathrm{NO}}_{\mathrm{3}}^{-}$ formation pathways. An explainable machine learning (ML) method was applied to identify the driving factors for high p${\mathrm{NO}}_{\mathrm{3}}^{-}$ formation rate via N₂O₅ uptake. Additionally, multi-scenario simulations were performed to examine the joint responses of p${\mathrm{NO}}_{\mathrm{3}}^{-}$ and O₃ formation to various NO_x and VOCs emissions. These findings enhance our understanding of p${\mathrm{NO}}_{\mathrm{3}}^{-}$ formation pathways and their environmental implications in NO₂-limited regions, providing valuable insights for developing joint PM_2.5 and O₃ mitigation strategies.

2 Methods

2.1 Field observation

Field observations were conducted during the winter period from 1 December 2022–3 February 2023, at an urban site (marked by the red star in Fig. S1 in the Supplement) in Xiamen, which is located in the southeastern coastal region of China. Detailed site information has been described in our previous studies (Yang et al., 2023; Liu et al., 2022). Trace gases (including PAN, HCHO, HONO, VOCs, O₃, NO_x, CO, and SO₂), chemical components in PM_2.5 (including organic carbon and elemental carbon, ${\mathrm{SO}}_{\mathrm{4}}^{\mathrm{2}-}$, ${\mathrm{NO}}_{\mathrm{3}}^{-}$, ${\mathrm{NH}}_{\mathrm{4}}^{+}$, Cl⁻), PM_2.5 mass concentration, and meteorological parameters (including ambient temperature (T), relative humidity (RH), atmospheric pressure (P), wind speed (WS), wind direction (WD), and photolysis rates) were continuously measured during the campaign. Detailed information about measurement methods and instruments is summarized in Sect. S1 in the Supplement. In addition, boundary layer height (BLH) data were obtained from the ERA5 dataset (Hersbach et al., 2020).

A chemical ionization time-of-flight mass spectrometer equipped with an iodide source (iodide-TOF-CIMS, Aerodyne Research Inc., USA) was deployed to measure N₂O₅ and ClNO₂. A nearly 2 m long perfluoroalkoxy (PFA) tube with a 1$/$4-inch inner diameter was used for sampling. The total sampling flow rate was set as 10 standard liters per minute (SLPM), of which only 2SLPM was diverted to the CIMS. A nitrogen (N₂) flow (99.999 %, 2.7 SLPM), carrying methyl iodide (CH₃I) vapor released from a heated permeation tube, passed through a soft X-ray source (Tofwerk AG, P-type) to generate reagent ions I⁻. The I⁻ was combined with the target gas in an ion molecule reaction (IMR) chamber and then detected by the ToF-CIMS. Ambient N₂O₅ and ClNO₂ were detected as the I(N₂O₅)⁻ and I(ClNO₂)⁻ clusters at 235 and 208 $m/z$. The detailed calibration procedures of N₂O₅ and ClNO₂ are described in Sect. S2, following established methods (Wang et al., 2022b, a; Thaler et al., 2011). Briefly, N₂O₅ was generated from the reaction between O₃ and excessive NO₂, while ClNO₂ was synthesized via the reaction of Cl₂ (6 ppm in N₂) with a moist mixture of NaNO₂ and NaCl. The calibration curves for N₂O₅ and ClNO₂ at different RH are shown in Fig. S2, with mean sensitivities of 0.110±0.063 and $\mathrm{0.055}\pm \mathrm{0.018}\mathrm{ncps}{\mathrm{ppb}}^{-\mathrm{1}}$, respectively. The instrument background was determined by introducing dry N₂ into the inlet for 20 min. Based on three times the standard deviation (3σ) of the background signal, the typical 1 min detection limits for N₂O₅ and ClNO₂ were estimated to be 0.61 and 1.3 ppt, respectively.

2.2 Determination of p${\mathrm{NO}}_{\mathrm{3}}^{-}$ formation rate

The iterative box model developed by Wagner et al. with a simplified mechanism was employed to obtain key parameters of the N₂O₅ uptake process (Wagner et al., 2013), including the loss rate of N₂O₅ (kN₂O₅) and the production yield of ClNO₂ (φClNO₂, see in Sect. S3). To validate the iterative box model results, these parameters were calculated concurrently based on the classical steady-state approximation method (Sect. S4) (Brown et al., 2003; Chen et al., 2022). The derived parameters of N₂O₅ uptake were adopted for subsequent multiphase box model.

A Framework for 0-D Atmospheric Modeling (F0AM), incorporating the Master Chemical Mechanism (MCM v3.3.1) and heterogeneous mechanisms (Table S2), was employed to simulate nitrate formation rates for each day during the study period (Wolfe et al., 2016; Atkinson and Arey, 2003; Jenkin et al., 2015). The heterogeneous parameters derived from the iterative box model were implemented in F0AM. In addition, hourly interval data of trace gases, photochemically active species, meteorological variables, and reanalysis data were also applied to constrain the multiphase chemical box model. Detailed model configurations are provided in Sect. S5. As shown in Fig. S3, the model performed well in simulating the trends of N₂O₅ and ClNO₂ with R² of 0.88 and 0.49, respectively. However, a systematic underestimation existed in the simulated N₂O₅ and ClNO₂ concentrations, which likely resulted from the model configuration including overestimated physical removal rates, elevated concentration of intermediate VOC species, or uncertainties in transport processes. Consequently, the simulated p${\mathrm{NO}}_{\mathrm{3}}^{-}$ formation from N₂O₅ uptake in this study could be regarded as a lower limit. The simulated OH concentrations agreed well with parameterized method suggested by Ehhalt and Rohrer (Fig. S4, R²=0.86) (Ehhalt and Rohrer, 2000). Based on model simulation and precursor observations, we quantified p${\mathrm{NO}}_{\mathrm{3}}^{-}$ formation rates through both OH+NO₂ and N₂O₅ uptake pathways by model integral.

2.3 Identification of influencing factors for p${\mathrm{NO}}_{\mathrm{3}}^{-}$ formation via N₂O₅ uptake

Extreme gradient boosting (XGBoost), a machine learning technique, has been widely applied in atmospheric chemistry research (Gui et al., 2020; Wang et al., 2023b; Requia et al., 2020). Here, we built a XGBoost model to reproduce the p${\mathrm{NO}}_{\mathrm{3}}^{-}$ formation rate via N₂O₅ uptake with selected variables. The model was built using the “xgboost” library (https://github.com/dmlc/xgboost/tree/master, last access: 8 November 2025) in a python environment. Explanatory variables included meteorological parameters (BLH, T, and RH), nocturnal atmospheric oxidation capacity P(NO₃) calculated by $k_{{\mathrm{NO}}_{\mathrm{2}}+{\mathrm{O}}_{\mathrm{3}}}\left[{\mathrm{NO}}_{\mathrm{2}}\right]\left[{\mathrm{O}}_{\mathrm{3}}\right]$, TVOCs, the logarithm of the ratio of NO₂ to O₃ ($\log \left(\right[{\mathrm{NO}}_{\mathrm{2}}]/[{\mathrm{O}}_{\mathrm{3}}]$), NO, and heterogeneous uptake parameters (φClNO₂ and kN₂O₅). Only nighttime (18:00–06:00 LT the next day) data were considered to identify key drivers of p${\mathrm{NO}}_{\mathrm{3}}^{-}$ formation via N₂O₅ uptake. The hyperparameters of the XGBoost model were tuned by grid searching method and the established model was evaluated using R², Mean Absolute Error (MAE) and Root Mean Square Error (RMSE). By incorporating SHAP interpretation, the XGBoost-SHAP method could quantify factor contributions through SHAP values, where absolute SHAP values denote the relative importance. Detailed description and setup of the XGBoost-SHAP method can be found in Sect. S6 and our previous study (Lin et al., 2024).

2.4 Emission scenario modelling

Using the aforementioned multiphase chemical box model, we investigated changes in formation rates of p${\mathrm{NO}}_{\mathrm{3}}^{-}$ (${\mathrm{PNO}}_{\mathrm{3}}^{-}$) and O₃ (PO₃) under different VOCs and NO_x emission scenarios. The base model simulation was performed using mean diurnal values from the winter 2022 observations. A series of emission scenarios were tested by scaling normalized VOCs and NO_x concentrations from 0–2 times baseline levels to examine their impacts on ${\mathrm{PNO}}_{\mathrm{3}}^{-}$ and PO₃. Prior to each scenario simulation, 3d spin-up was set to stabilize intermediate species concentrations. Isopleth diagrams of simulated ${\mathrm{PNO}}_{\mathrm{3}}^{-}$ and PO₃ were obtained from the base scenario and 120 emission change scenarios. In addition, response strength (RS) was calculated using Eq. (2) as an indicator of emission sensitivity.

$$\begin{array}{}\text{(1)}& {\mathrm{PO}}_{\mathrm{3}}=k_{\mathrm{1}}\left[{\mathrm{HO}}_{\mathrm{2}}\right]\left[\mathrm{NO}\right]+\sum k_{\mathrm{2}i}\left[{\mathrm{RO}}_{\mathrm{2}}\right]\left[\mathrm{NO}\right]\end{array}$$

where, k_i is the corresponding chemical reaction rate constants.

$$\begin{array}{}\text{(2)}& \text{RS}={\displaystyle \frac{X_i-X_{\text{base}}}{V_i-V_{\text{base}}}}\end{array}$$

where, X_i and X_base are the mean formation rates of dependent variables e.g. ${\mathrm{PNO}}_{\mathrm{3}}^{-}$, PO₃ in scenario i and base simulations, respectively. V_i and V_base are the emission rates for the scenario i and base simulations, respectively. Notably, the emission rates ranged from 0–2 times baseline levels, with the base simulation emission rate normalized to 1.

3 Results and discussion

3.1 Overview of observations

The mean diurnal patterns of p${\mathrm{NO}}_{\mathrm{3}}^{-}$, gaseous pollutants and relevant meteorological parameters are shown in Fig. 1. During the entire observation period, mean concentrations of NO₂, O₃, total VOCs, and PM_2.5 were 10.9 ppb, 27.3 ppb, 18.2 ppb, and 14.3 µg m⁻³, respectively, lower than those observed in most of China's key cities (refer to Table S1). Despite the low NO_x levels, p${\mathrm{NO}}_{\mathrm{3}}^{-}$ contributed 29.5 % to PM_2.5 mass concentration, which was higher than proportions reported in Beijing urban area (24.7 %) (Ma et al., 2023), Guangdong (24.0 %) (Yun et al., 2018), and Nanjing (24 %–27 %) (Huang et al., 2020). This discrepancy suggests efficient conversion from NO₂ to p${\mathrm{NO}}_{\mathrm{3}}^{-}$ in the study area. In addition, the proportion of p${\mathrm{NO}}_{\mathrm{3}}^{-}$ increased with rising PM_2.5 concentration (Fig. S6), indicating its importance to particulate pollution. This is consistent with the phenomenon widespread in urban areas of China where p${\mathrm{NO}}_{\mathrm{3}}^{-}$ became dominant in inorganic aerosols despite NO_x reduction, underscoring the need for efficient p${\mathrm{NO}}_{\mathrm{3}}^{-}$ control strategies (Zhai et al., 2021; Zhao et al., 2020; Zhang et al., 2022).

Figure 1 Diurnal variations of key parameters during the winter of 2022. The concentrations of p${\mathrm{NO}}_{\mathrm{3}}^{-}$, NO_x, O₃, N₂O₅, PM_2.5 and ClNO₂. The levels of the photolysis frequencies of NO₂ (JNO₂), ambient temperature (T), relative humidity (RH), the lifetime of N₂O₅ (τN₂O₅), wind speed (WS) and the boundary layer height (BLH). Shaded areas of p${\mathrm{NO}}_{\mathrm{3}}^{-}$, O₃, N₂O₅, PM_2.5, ClNO₂, T, RH and BLH represent 95 % confidence intervals.

The diurnal pattern of p${\mathrm{NO}}_{\mathrm{3}}^{-}$ exhibited a bimodal characteristic, with peaks occurring at 04:00 and 15:00 LT, respectively. The daytime peak (07:00–17:00 LT was accompanied by low concentrations of NO_x and high levels of O₃ and JNO₂, indicating that active photochemical conditions promoted daytime p${\mathrm{NO}}_{\mathrm{3}}^{-}$ formation. During the nighttime (18:00–06:00 LT the next day), p${\mathrm{NO}}_{\mathrm{3}}^{-}$ concentrations increased together with NO₂, N₂O₅ and ClNO₂ from 18:00 LT onward and remained elevated until early morning. This nighttime accumulation can be attributed to two factors. First, lower temperature, shallower boundary layer height, and reduced wind speed at night favored the accumulation of p${\mathrm{NO}}_{\mathrm{3}}^{-}$ and related nitrogen-containing species. Second, higher RH and PM_2.5 concentrations at night enhanced aerosol water content and surface area, providing favorable conditions for heterogeneous hydrolysis of N₂O₅ to form p${\mathrm{NO}}_{\mathrm{3}}^{-}$. The mean concentration of N₂O₅ was 0.19±0.26 ppb (peaking at 2.52 ppb), which is relatively higher than values reported for China's megacities (Chen et al., 2020; Wang et al., 2017; Tham et al., 2018; Wang et al., 2022a; Liu et al., 2025; Li et al., 2023). Moreover, the observed elevation in nighttime ClNO₂, primarily produce via the reaction of N₂O₅ with Cl-containing particles, strongly supports the presence of active heterogeneous processes of N₂O₅. Collectively, these findings imply a likely significant contribution of N₂O₅ uptake to p${\mathrm{NO}}_{\mathrm{3}}^{-}$ formation during the nighttime.

Figure 2 Simulated rates of key p${\mathrm{NO}}_{\mathrm{3}}^{-}$ formation pathways obtained from the chemical box model incorporating heterogeneous parameters. Diurnal formation rates of p${\mathrm{NO}}_{\mathrm{3}}^{-}$ via the OH+NO₂ and N₂O₅ uptake pathways (a) and comparison of the two pathways during the nighttime (b), daytime (c), and the whole day (d). Note that the results in panel (a) represent the mean simulated formation rates over the entire observation period. The box shows the 25th–75th percentiles with whiskers representing the 5th–95th percentiles. The black line and dot inside the box represent the mean and median values, respectively. Statistical significance was determined using pair-sample t-tests with *** indicating p<0.001.

3.2 High contribution of N₂O₅ uptake to p${\mathrm{NO}}_{\mathrm{3}}^{-}$ formation in NO₂-limited conditions

In view of the observed importance of daytime and nighttime p${\mathrm{NO}}_{\mathrm{3}}^{-}$ formation, we further employed an observation-constrained model to quantify the potential formation pathways, including the gas-phase reaction of OH+NO₂ and heterogeneous N₂O₅ uptake. This model incorporated heterogeneous chemical mechanisms, with key heterogeneous parameters (e.g. kN₂O₅ and φClNO₂) obtained through simulation (See Methods for details). As shown in Fig. S7, the simulated kN₂O₅ and φClNO₂ exhibited good agreement with the classical steady-state method (R²=0.76 and 0.73, respectively), demonstrating the model's capability to characterize heterogeneous uptake processes and thereby effectively evaluate p${\mathrm{NO}}_{\mathrm{3}}^{-}$ formation processes.

As illustrated in Fig. 2a, the diurnal pattern of p${\mathrm{NO}}_{\mathrm{3}}^{-}$ formation rates exhibited a classical characteristic, with daytime dominated by gas-phase oxidation and nighttime dominated by N₂O₅ uptake. Specifically, the daytime OH+NO₂ reaction had a mean p${\mathrm{NO}}_{\mathrm{3}}^{-}$ formation rate of 1.62 $\mathrm{\mu }\mathrm{g}{\mathrm{m}}^{-\mathrm{3}}{\mathrm{h}}^{-\mathrm{1}}$, while the nighttime N₂O₅ uptake pathway showed a formation rate of 1.18 $\mathrm{\mu }\mathrm{g}{\mathrm{m}}^{-\mathrm{3}}{\mathrm{h}}^{-\mathrm{1}}$ (Fig. 2b and c). For the whole day, N₂O₅ uptake contributed an average of 51.2 % to p${\mathrm{NO}}_{\mathrm{3}}^{-}$ formation, which was comparable to the contribution of the OH+NO₂ pathway (Fig. 2d). To exclude year-specific effects, we further analyzed p${\mathrm{NO}}_{\mathrm{3}}^{-}$ formation during the winters from 2019–2023. The results revealed that the p${\mathrm{NO}}_{\mathrm{3}}^{-}$ formation rates via N₂O₅ uptake (0.75–1.40 $\mathrm{\mu }\mathrm{g}{\mathrm{m}}^{-\mathrm{3}}{\mathrm{h}}^{-\mathrm{1}}$) were comparable to those from the OH+NO₂ reaction (0.88–1.66 $\mathrm{\mu }\mathrm{g}{\mathrm{m}}^{-\mathrm{3}}{\mathrm{h}}^{-\mathrm{1}}$; Fig. 3a), with the N₂O₅ uptake pathway consistently accounting for approximately half of the total p${\mathrm{NO}}_{\mathrm{3}}^{-}$ formation in the study area (Fig. 3b). Such a high contribution of N₂O₅ uptake to p${\mathrm{NO}}_{\mathrm{3}}^{-}$ is generally uncommon in urban areas. A study in urban Beijing showed that during non-polluted periods, N₂O₅ uptake contributed only 18.9 % to nitrate formation rates (Ma et al., 2023). Similarly, the contributions of N₂O₅ uptake were 10 %–38 % and 4 % in urban areas of the YRD (Sun et al., 2022; Zhai et al., 2023; Zhang et al., 2023b) and PRD regions(Yang et al., 2022), respectively.

Figure 3 Inter-annual patterns of key p${\mathrm{NO}}_{\mathrm{3}}^{-}$ formation pathways in urban Xiamen. The average p${\mathrm{NO}}_{\mathrm{3}}^{-}$ formation rate from OH+NO₂ and N₂O₅ uptake (a), and the average ratio of the sum of nocturnal p${\mathrm{NO}}_{\mathrm{3}}^{-}$ concentrations to the sum of all-day p${\mathrm{NO}}_{\mathrm{3}}^{-}$ concentration (b) in different winters from 2019–2023 based on the measured p${\mathrm{NO}}_{\mathrm{3}}^{-}$ in PM_2.5. The sensitivity of nocturnal p${\mathrm{NO}}_{\mathrm{3}}^{-}$ formation via N₂O₅ uptake to NO₂ and O₃ from 2019–2023 (c). And probability distribution of observed ${\mathrm{NO}}_{\mathrm{2}}/{\mathrm{O}}_{\mathrm{3}}$ at nighttime in winter 2022 (d). The observed periods of different winters from 2019–2023 are summarized in Table S3. In panel (b), the black line and the solid circle in the boxplot represent the mean and median value, respectively. In panel (c), the black triangle indicates the base case of winter 2022, solid circles in different colors represent the average NO₂ to O₃ ratios in different years, and the predicted average formation rate of N₂O₅ uptake as the normalized emissions (average concentrations of O₃ and NO₂) varied between 0–2.

Previous studies have found that nocturnal p${\mathrm{NO}}_{\mathrm{3}}^{-}$ formation via N₂O₅ uptake strongly depends on the ratio of NO₂ to O₃ (Ma et al., 2023). This process is suppressed in the O₃-limited regime (${\mathrm{NO}}_{\mathrm{2}}/{\mathrm{O}}_{\mathrm{3}}>\mathrm{2}$) but enhanced in the NO₂-limited regime (${\mathrm{NO}}_{\mathrm{2}}/{\mathrm{O}}_{\mathrm{3}}\le \mathrm{1}$). The COVID-19 lockdown period was a typical example of this ratio dependence(Yan et al., 2023). In regions like Beijing, substantial reductions in NO_x emissions caused a shift in nocturnal p${\mathrm{NO}}_{\mathrm{3}}^{-}$ formation from the O₃-limited to the NO₂-limited regime. This shift resulted in elevated nighttime O₃ levels and a weakened NO titration effect, collectively promoting N₂O₅ formation and subsequent p${\mathrm{NO}}_{\mathrm{3}}^{-}$ formation. The sensitivity of p${\mathrm{NO}}_{\mathrm{3}}^{-}$ formation via N₂O₅ uptake to NO₂ and O₃ during the campaign is presented in Fig. 3c and d. The observed mean values of ${\mathrm{NO}}_{\mathrm{2}}/{\mathrm{O}}_{\mathrm{3}}$ (0.40) and the probability distributions of ${\mathrm{NO}}_{\mathrm{2}}/{\mathrm{O}}_{\mathrm{3}}$ ratios both indicate that N₂O₅ uptake was in the NO₂-limited regime. Based on NO₂ and O₃ observational data during 2015–2021 from the China National Environmental Monitoring Centre(Ma et al., 2023), most key urban regions in China (e.g., the NCP, YRD, and Beijing) were found to lie in the O₃-limited or transition regimes ($\mathrm{1}<{\mathrm{NO}}_{\mathrm{2}}/{\mathrm{O}}_{\mathrm{3}}\le \mathrm{2}$), whereas nocturnal p${\mathrm{NO}}_{\mathrm{3}}^{-}$ formation in southeastern China was distinctly in NO₂-limited regime. These results confirm that the dominant p${\mathrm{NO}}_{\mathrm{3}}^{-}$ formation mechanisms in our study area significantly differs from those in most urban areas of China, which might be attributed to the dependence of N₂O₅ uptake on precursor NO₂ and O₃. In addition, the dominance of N₂O₅ uptake in p${\mathrm{NO}}_{\mathrm{3}}^{-}$ formation also occurred during haze pollution periods (Zhai et al., 2023; Wang et al., 2017), where increased aerosol surface area under high particulate loadings created favorable conditions for N₂O₅ heterogeneous reactions. Therefore, to evaluate the role of precursors, we conducted a comprehensive analysis of the factors driving p${\mathrm{NO}}_{\mathrm{3}}^{-}$ formation via N₂O₅ uptake.

3.3 Driving factors of p${\mathrm{NO}}_{\mathrm{3}}^{-}$ formation via nocturnal N₂O₅ uptake

The N₂O₅ uptake rate is influenced by multiple factors including precursor levels, meteorological parameters, and heterogeneous reaction conditions (Ma et al., 2023; Chen et al., 2020, 2024). A machine learning method integrating these factors was employed to identify the key drivers of p${\mathrm{NO}}_{\mathrm{3}}^{-}$ formation via N₂O₅ uptake. The relative importance of each factor was evaluated by absolute SHAP values (Fig. 4a), and their impacts were elucidated by examining the relationships between individual factors and their corresponding SHAP values (Fig. 4b–e and S8). Results showed that the nocturnal NO₃ formation rate (P(NO₃)), an integrated indicator of nocturnal atmospheric oxidation capacity (Wang et al., 2021), was the most important factor. The steep slope of the positive correlation between P(NO₃) and SHAP values indicated that P(NO₃) strongly enhances p${\mathrm{NO}}_{\mathrm{3}}^{-}$ formation via N₂O₅ uptake. P(NO₃) is primarily formed through the reaction between NO₂ and O₃ ($\mathrm{P}\left({\mathrm{NO}}_{\mathrm{3}}\right)=k_{{\mathrm{NO}}_{\mathrm{2}}+{\mathrm{O}}_{\mathrm{3}}}\left[{\mathrm{NO}}_{\mathrm{2}}\right]\left[{\mathrm{O}}_{\mathrm{3}}\right]$), suggesting that NO₂ and O₃ mainly influenced p${\mathrm{NO}}_{\mathrm{3}}^{-}$ formation via N₂O₅ by modulating NO₃ radical formation. Notably, the factor $\log {\mathrm{NO}}_{\mathrm{2}}/{\mathrm{O}}_{\mathrm{3}}$ had relatively low importance, indicating concentrations of precursors were more important than ${\mathrm{NO}}_{\mathrm{2}}/{\mathrm{O}}_{\mathrm{3}}$ ratio in determining p${\mathrm{NO}}_{\mathrm{3}}^{-}$ formation via N₂O₅ uptake under extremely NO₂-limited condition (mean ${\mathrm{NO}}_{\mathrm{2}}/{\mathrm{O}}_{\mathrm{3}}$ was 0.40). Furthermore, as shown in Fig. S8b, $\log {\mathrm{NO}}_{\mathrm{2}}/{\mathrm{O}}_{\mathrm{3}}$ and its SHAP value show a positive correlation when $\log {\mathrm{NO}}_{\mathrm{2}}/{\mathrm{O}}_{\mathrm{3}}$ is less than 0. This indicates that under NO₂-limited conditions ($\log {\mathrm{NO}}_{\mathrm{2}}/{\mathrm{O}}_{\mathrm{3}}<\mathrm{0}$, i.e., ${\mathrm{NO}}_{\mathrm{2}}/{\mathrm{O}}_{\mathrm{3}}<\mathrm{1}$), p${\mathrm{NO}}_{\mathrm{3}}^{-}$ formation via N₂O₅ uptake was driven by the elevated NO₂.

Compared with P(NO₃), other factors exhibited weaker effects on p${\mathrm{NO}}_{\mathrm{3}}^{-}$ formation rate via N₂O₅ uptake. φClNO₂ emerged as the second most important factor and showed a negative correlation with SHAP values (Fig. 4c), illustrating that ClNO₂ formation inhibited p${\mathrm{NO}}_{\mathrm{3}}^{-}$ formation. This inhibitory effect could be attributed to high concentrations of Cl-containing particles ($\mathrm{0.94}\pm \mathrm{1.11}\mathrm{\mu }\mathrm{g}{\mathrm{m}}^{-\mathrm{3}}$) in the study area. Chloride-containing aerosols promote N₂O₅ uptake to produce more ClNO₂ (as evidenced by the positive correlation between φClNO₂ and chloride ions, Fig. S9), while simultaneously reducing p${\mathrm{NO}}_{\mathrm{3}}^{-}$ formation (R5). Additionally, the nighttime produced ClNO₂ can undergo photolysis in following day to release Cl radicals, which further promote O₃ formation. This indirect effect must be considered when formulating control measures for particulate matter pollution. Interestingly, as shown in Table S4 (Tham et al., 2016; Wang et al., 2018; Yun et al., 2018; Morgan et al., 2015), although the simulated kN₂O₅ ($\mathrm{7.64}\times {\mathrm{10}}^{-\mathrm{3}}\pm \mathrm{6.12}\times {\mathrm{10}}^{-\mathrm{3}}{\mathrm{s}}^{-\mathrm{1}}$) was higher than values reported in Beijing ($\mathrm{8.1}\times {\mathrm{10}}^{-\mathrm{4}}-\mathrm{1.42}\times {\mathrm{10}}^{-\mathrm{3}}{\mathrm{s}}^{-\mathrm{1}}$), Guangdong ($\mathrm{3.78}\times {\mathrm{10}}^{-\mathrm{3}}-\mathrm{9}\times {\mathrm{10}}^{-\mathrm{3}}{\mathrm{s}}^{-\mathrm{1}}$), and UK ($\mathrm{9.3}\times {\mathrm{10}}^{-\mathrm{5}}-{\mathrm{10}}^{-\mathrm{3}}{\mathrm{s}}^{-\mathrm{1}}$), kN₂O₅ exerted only a weak positive effect on N₂O₅ uptake (Fig. 4d). The large difference existing in the importance of P(NO₃) and kN₂O₅ indicated that the p${\mathrm{NO}}_{\mathrm{3}}^{-}$ formation rate via N₂O₅ uptake process was more limited by precursor levels rather than heterogeneous uptake conditions. Similar phenomenon was also found in winter in urban Beijing and Northern Utah mountain basins (Mcduffie et al., 2019; Chen et al., 2020). The total concentrations of the observed VOCs (TVOCs) showed a weak negative correlation with N₂O₅ uptake (Fig. 4e). Similar to existing research (Hu et al., 2023), specific VOC species, such as styrene, 2-butene, and isoprene, can readily consume NO₃ radicals (Fig. S10), thereby inhibiting N₂O₅ formation. However, the loss of N₂O₅ through the reaction between VOCs and NO₃ was relatively limited compared to its direct uptake, as determined by our calculations (Text S4), which supported the SHAP analysis.

Figure 4 Feature importance (a) and the effects of key factors on p${\mathrm{NO}}_{\mathrm{3}}^{-}$ formation via N₂O₅ uptake (b–e) obtained by the XGBoost-SHAP method. The relationships between SHAP values and major features: P(NO₃) (b), φClNO₂(c), kN₂O₅ (d), and TVOCs (e). Feature importance ranking (a) is determined by mean absolute SHAP values (descending order, top to bottom). Relationships between SHAP values and other factors are shown in Fig. S8.

Moreover, we found that the effects of φClNO₂, kN₂O₅, and TVOCs on p${\mathrm{NO}}_{\mathrm{3}}^{-}$ formation via N₂O₅ uptake were subject to P(NO₃) levels (Fig. 5a–c). Specifically, the negative effect of φClNO₂ and the positive effect of kN₂O₅ on p${\mathrm{NO}}_{\mathrm{3}}^{-}$ formation via N₂O₅ uptake became statistically significant when P(NO₃) exceeded approximately 1.0 and 0.5 ppb h⁻¹, respectively. The negative correlation slope of TVOCs versus p${\mathrm{NO}}_{\mathrm{3}}^{-}$ formation via N₂O₅ uptake intensified with increasing P(NO₃) levels, indicating that the N₂O₅ removal effect was enhanced through VOC-induced NO₃ depletion. These findings highlight the critical role of precursor NO₂ and O₃ in nocturnal p${\mathrm{NO}}_{\mathrm{3}}^{-}$ formation, demonstrating that these precursors mainly affect this pathway by modulating NO₃ radical formation.

Figure 5 Relationships between p${\mathrm{NO}}_{\mathrm{3}}^{-}$ formation via N₂O₅ uptake and φClNO₂ (a), kN₂O₅ (b), and TVOCs (c) colored by P(NO₃). Linear fit curves in purple, blue and orange represent the fitting results for P(NO₃) in the ranges of 0–0.5, 0.5–1.0 and $>\mathrm{1.0}\mathrm{ppb}{\mathrm{h}}^{-\mathrm{1}}$, respectively.

Figure 6 Results of multi-scenario simulations obtained from an observation-constrained box model. Isopleths of simulated ${\mathrm{PNO}}_{\mathrm{3}}^{-}$ (a) and PO₃ (b) with normalized VOCs and NO_x. Simulated mean formation rates of p${\mathrm{NO}}_{\mathrm{3}}^{-}$ and O₃ (c, d), as well as p${\mathrm{NO}}_{\mathrm{3}}^{-}$ formation rates via N₂O₅ uptake and OH+NO₂ (e, f) with normalized VOCs and NO_x. The ${\mathrm{PNO}}_{\mathrm{3}}^{-}$ and PO₃ denote the formation rates of p${\mathrm{NO}}_{\mathrm{3}}^{-}$ and O₃, respectively. The simulated results are daily mean values, and the black triangle indicates the base case for winter 2022. In addition, the results in panel (c–f) were obtained by maintaining either NO_x or VOCs at the base emission rate while varying the other.

3.4 Optimal mitigation strategies of p${\mathrm{NO}}_{\mathrm{3}}^{-}$ under NO₂-limited conditions

The above results revealed that p${\mathrm{NO}}_{\mathrm{3}}^{-}$ formation through both the daytime OH+NO₂ reaction and nocturnal heterogeneous N₂O₅ uptake was closely linked to VOCs–NO_x–O₃ chemistry (Yang et al., 2022). Using a multiphase box model, we systematically examined the responses of both p${\mathrm{NO}}_{\mathrm{3}}^{-}$ and O₃ to varying NO_x and VOC emission scenarios. Figure 6a shows p${\mathrm{NO}}_{\mathrm{3}}^{-}$ formation located in the transition regime of VOCs and NO_x. The formation rate of p${\mathrm{NO}}_{\mathrm{3}}^{-}$ (${\mathrm{PNO}}_{\mathrm{3}}^{-}$) decreased with the reductions of VOCs and NO_x, and this trend became more pronounced under aggressive NO_x reduction scenarios (Fig. 6c and d). Figure S11a and b reveal that the mean response strength (RS, as defined in Methods) of ${\mathrm{PNO}}_{\mathrm{3}}^{-}$ to NO_x was 0.75, higher than that for VOCs (RS=0.29), suggesting that NO_x reduction had a greater potential for p${\mathrm{NO}}_{\mathrm{3}}^{-}$ mitigation compared to VOCs control. However, NO_x and VOCs reductions exerted different impacts on O₃ formation rate (PO₃). In our study area, PO₃ located in the VOC-limited regime (Fig. 6b). We found that PO₃ declined with VOCs reduction but increased with NO_x reduction until NO_x dropped below 20 % of the base (Fig. 5c and d). Moreover, detailed results distinguishing daytime and nighttime major formation pathways of p${\mathrm{NO}}_{\mathrm{3}}^{-}$ are presented in Fig. 6e and f and Fig. S11c and d. For VOC reduction scenarios, both the OH+NO₂ reaction and N₂O₅ uptake pathways showed declining nitrate formation rates, with comparable RS of 0.11 and 0.18, respectively. This occurs because reduced VOCs concentrations decrease OH radical and O₃ concentrations, thereby suppressing p${\mathrm{NO}}_{\mathrm{3}}^{-}$ formation via both pathways. In contrast, NO_x reduction yielded more complex behavior. The OH+NO₂ reaction rates remained nearly constant until NO_x dropped to 60 % of the base. This stability arises because NO_x reduction diminishes the NO titration effect on O₃, thereby increasing OH radicals through O₃ photolysis. The competing effects of NO_x reduction and OH enhancement led to an initial plateau in the OH+NO₂ reaction rate before its eventual decline. Differently, the N₂O₅ uptake rate decreased consistently and significantly with NO_x abatement, exhibiting a high mean RS value of 0.61. This phenomenon was closely associated with the NO₂-limited regime of N₂O₅ uptake in the study area. As shown in Fig. S12, the variation trends of ${\mathrm{PNO}}_{\mathrm{3}}^{-}$, P(O₃), OH+NO₂, and N₂O₅ uptake were consistent across all VOCs/NO_x combinations, indicating that the results robustly reflect the response mechanisms to precursor emission changes.

As mentioned above, while VOCs reduction proved effective in mitigating both p${\mathrm{NO}}_{\mathrm{3}}^{-}$ and O₃, its effectiveness in reducing p${\mathrm{NO}}_{\mathrm{3}}^{-}$ remained limited compared to NO_x reduction. However, the effectiveness of NO_x reduction exhibited significant regional and temporal variations. In China's megacities, including PRD, YRD, and BTH regions, p${\mathrm{NO}}_{\mathrm{3}}^{-}$ initially increased and then decreased in response to the reduction of NO_x emissions (Li et al., 2021; Zhang et al., 2023b; Yang et al., 2022). Under high-NO_x conditions, mild NO_x reduction would raise daytime OH and O₃ concentrations (Zhang et al., 2023b), rendering OH (rather than NO_x) the limiting factor for the OH+NO₂ reaction, which consequently enhanced daytime p${\mathrm{NO}}_{\mathrm{3}}^{-}$ formation. Additionally, as the season most susceptible to PM pollution, wintertime N₂O₅ formation in these regions was in an O₃-limited or transition regime (Ma et al., 2023), wherein the elevated daytime O₃ significantly enhanced NO₃ radical generation, thereby promoting nocturnal N₂O₅ uptake and subsequent p${\mathrm{NO}}_{\mathrm{3}}^{-}$ formation. Conversely, in NO₂-limited regions (e.g., southeastern China), NO_x reduction showed limited impact on daytime p${\mathrm{NO}}_{\mathrm{3}}^{-}$ formation via the OH+NO₂ pathway but effectively suppressed nighttime p${\mathrm{NO}}_{\mathrm{3}}^{-}$ formation via N₂O₅ uptake. This approach concurrently reduced ClNO₂ formation from N₂O₅ heterogeneous processes, consequently diminishing next-day Cl radical generation and its positive feedback on O₃ formation. Considering NO_x reduction during the daytime would cause O₃ formation and only a slight reduction in p${\mathrm{NO}}_{\mathrm{3}}^{-}$, it is preferable to regulate NO_x at night (18:00–06:00 LT the next day). Our findings demonstrate that in regions with a NO₂-limited for p${\mathrm{NO}}_{\mathrm{3}}^{-}$ formation, targeted NO_x reduction can synergistically decrease both p${\mathrm{NO}}_{\mathrm{3}}^{-}$ and O₃ concentrations, highlighting the critical need to tailor mitigation strategies for different regions.

4 Conclusions and implications

Our observations revealed a bimodal diurnal pattern of p${\mathrm{NO}}_{\mathrm{3}}^{-}$ in winter in urban Xiamen. The co-occurrence of elevated nighttime p${\mathrm{NO}}_{\mathrm{3}}^{-}$ levels with increased N₂O₅ implied a significant contribution of N₂O₅ uptake to p${\mathrm{NO}}_{\mathrm{3}}^{-}$ formation. Quantitative model analysis showed that N₂O₅ uptake contributed 51.2 % of the total p${\mathrm{NO}}_{\mathrm{3}}^{-}$, which was comparable to the OH+NO₂ reaction. This high contribution of N₂O₅ uptake to p${\mathrm{NO}}_{\mathrm{3}}^{-}$ is not commonly observed across Chinese cities. Comparative analysis among different cities suggests that this phenomenon is likely associated with NO₂-limited conditions for N₂O₅ uptake in our study area. Machine learning results further demonstrated that p${\mathrm{NO}}_{\mathrm{3}}^{-}$ formation via N₂O₅ uptake was driven by nocturnal atmospheric oxidation capacity (PNO₃) rather than heterogeneous uptake conditions. The underlying mechanism is that the weakened NO_x titration effects lead to nighttime O₃ accumulation, which promotes NO₃ radical generation and consequently enhances N₂O₅ and p${\mathrm{NO}}_{\mathrm{3}}^{-}$ formation. The joint response of p${\mathrm{NO}}_{\mathrm{3}}^{-}$ and O₃ to various NO_x and VOCs emission scenarios indicated that p${\mathrm{NO}}_{\mathrm{3}}^{-}$ was more sensitive to NO_x reduction than to VOCs reduction. However, mild NO_x reduction showed limited effectiveness in reducing daytime p${\mathrm{NO}}_{\mathrm{3}}^{-}$ while simultaneously increasing O₃ concentrations. Our findings suggest that NO_x reduction is more effective when implemented during nighttime, particularly in regions where N₂O₅ formation is NO₂-limited. This approach can effectively control p${\mathrm{NO}}_{\mathrm{3}}^{-}$ formation by suppressing nocturnal NO₃ radical generation and consequently inhibiting N₂O₅ uptake, while simultaneously alleviate O₃ pollution by reducing ClNO₂ formation. With continuous NO_x and VOCs emission reductions and renewable energy adoption in China, urban areas are transitioning from NO_x-saturated to NO_x-limited conditions, potentially increasing the importance of the N₂O₅ uptake pathway. In this context, comprehensive assessment of NO_x reduction impacts on urban p${\mathrm{NO}}_{\mathrm{3}}^{-}$ and O₃ pollution, along with the development of region-specific mitigation strategies, becomes critically important.