The chamber experiments were conducted in Fluorinated Ethylene Propylene (FEP, 200A, DuPont) reactors under dark conditions, with background air supplied by purified zero air. Styrene was injected into the reactor with zero air using a glass microsyringe, O3 was produced by an ozone generator with pure O2, and NH3 was directly injected into the reactor. The reactants and their concentration ranges used in the experiment are styrene (0.3–3 ppm), O3 (1–10 ppm), and NH3 (0–10 ppm), respectively. Because ozonolysis of styrene can form OH radicals, n-Hexane was used as an OH radical scavenger (>100 ppm with a removal efficiency >90%). Detailed experimental conditions are provided in Table S1 in the Supplement.

To collect particles and determine the SOA yields, experiments 1–5 were conducted in a 1.2 m3 chamber. During these experiments, styrene was measured online using a proton transfer reaction-mass spectrometer (PTR-MS P1000-L-AI, Anhui Province Key Laboratory of Medical Physics and Technology) with a time resolution of 20 s in the gas phase. O3 was measured every 0.5 h lasting for 5 min with an O3 analyzer (Model 49C, Thermo Scientific) with a time resolution of 10 s in the gas phase. The particle concentrations and size distributions were determined by a scanning mobility particle sizer (SMPS, Model 3936, DMA-3080, CPC-3776, TSI) with a time resolution of 5 min. The online measurements covered the entire experimental process (4–5 h). Particles were collected on a 25 mm polytetrafluoroethylene (PTFE) membrane with a pore size of 0.45 µm at the 4th hour, and the sample flow rate was 6 Lmin-1 and lasted for 40 min. The collected particles were extracted with methanol for composition analysis in the particle phase, which were injected by a high-performance liquid chromatography (HPLC, Thermo Scientific), ionized by a heated electrospray ionization source (ESI), and then the molecular composition was measured by a high-resolution Orbitrap mass spectrometer (Orbitrap MS, Q-Exactive, Thermo Scientific) with a resolution R=70000 at m/z 200. To determine the kinetics and mechanism of the reaction between C7-SCI and NH3, experiments 6–10 were performed with higher concentrations in a 150 L chamber. During these experiments, the products were online ionized by a gas aerosol in-situ ionization source (GAIS), and then measured by Orbitrap MS in the gas phase. The time resolution of GAIS-Orbitrap MS measurement is about 0.5 s, and all the experiments lasted about 1 h.

To detect peroxides in the sample, experiment 11 was conducted in the 1.2 m3 chamber. The collected sample was immediately extracted by 400 µL acetonitrile (ACN) before being injected into HPLC-HRMS. Using ACN as extraction solvent to minimize other unwanted decomposition processes such as hydrolysis. Half of the liquid (180 µL) from the combined extract mixed with 10 µL acetic acid (600 mM in ACN) in a vial, followed by the addition of 10 µL KI (99.5 %, Sigma-Aldrich) (400 mM in H2O) to trigger the iodometry reaction; another 180 µL aliquot was treated in a same way by adding 10 µL acetic acid (600 mM in ACN) and 10 µL H2O, instead of KI. These two SOA samples are designated as KI-treated and non-treated respectively, which were injected into HPLC-HRMS (Li et al., 2025a).

Raw spectra were processed using Xcalibur (v4.1.31.9, Thermo Scientific). Tandem MS (MS²) was used to determine molecular structures, and Mass Frontier (v7.0.5.9, Thermo Scientific) can simulate potential product ions for molecule with known structure, which were then compared to the MS² spectra of molecular ion species to confirm the final structures of the molecules. Gas-phase reactions were simulated using the Master Chemical Mechanism (MCM v3.3.1, website: https://mcm.york.ac.uk/MCM, last access: 29 April 2024). To evaluate the influence of NH3-SCI reactions, we added four reactions to the MCM mechanism, including those between NH3 and C1-/C7-SCIs (CH2OO/PHCHOO) and the subsequent decomposition of C7H9O2N into C7H7N and C7H7ON. (Bloss et al., 2005; Jenkin et al., 2003; Jia et al., 2023; Jia and Xu, 2021). The OECD QSAR Toolbox (Version 4.7, http://qsartoolbox.org, last access: 27 March 2025) is used for the calculation of molecular toxicity, and additional details are presented in the Supplement.

As NH3 concentrations increased, SOA mass yields decreased significantly from (4.9±0.3) % (0 ppm NH3) to (1.0±0.1) % (0.8 ppm NH3), showing an obvious inhibitory effect (Fig. 1a). The observed yields with 0 ppm NH3 are within the range of those previously reported for styrene ozonolysis under no NH3 conditions (2.7 %–6.5 %) (Bracco et al., 2019; Díaz-de-Mera et al., 2017; Yu et al., 2022, 2024a), which demonstrates the rationality of our experiments. A strong negative correlation was observed between NH3 levels and SOA yields (R2=0.98), confirming significant suppression of SOA by the presence of NH3. In the styrene-O3 reaction system, SOA is primarily derived from SCI-related products. As the concentration of NH3 increases, the SOA concentration decreases linearly. This indicates that the observed reduction of SOA is attributed to the competitive consumption of SCI by NH3.

MS analysis reveals that NH3 suppresses SOA formation by affecting oligomerization pathways of styrene-derived SCIs. As shown in Fig. 1b, the most significant peaks C6H14O9Na+ (m/z=253.052), C5H12O7Na+ (m/z=207.047) and C13H20O11Na+ (m/z=375.089) exhibit regular mass differences corresponding to C1-SCI (CH2OO, Δm/z=46.005) and C7-SCI (C7H6OO, Δm/z=122.036), consistent with our previous works that the oligomerization of SCIs is the main mechanism for SOA formation from styrene (Yu et al., 2022). These oligomers are significantly reduced by 51 % with increasing NH3 concentrations, which strongly supports the result that NH3 can inhibit the formation of oligomers from SCIs.

Meanwhile, benzoic acid (C7H5O2-, m/z=121.028), as the dominant compound in styrene-ozonolysis system, is significantly suppressed by 51 % with increasing NH3 concentration (Fig. 1c), which is consistent with the trend of SOA yield inhibition (50 %). Since benzoic acid is mainly formed from the reaction of C7-SCI with H2O (Na et al., 2006; Banu et al., 2018), the presence of NH3 apparently competes with H2O for SCIs and inhibits the formation of benzoic acid (Fig. 1d). In addition, to maximize the potential of NH3 and H2O to compete for C7-SCI under extreme conditions, we further conducted two experiments with low NH3/normal humidity vs. high NH3/extremely low humidity. The strong suppression on the formation of benzoic acid can be most clearly demonstrated when the concentration of NH3 is much higher than that of H2O. Experimental observation show that benzoic acid was suppressed by over 90 % under high NH3 concentration (10 ppm) and low relative humidity (2 %) conditions (Fig. 1d). The simulation results from MCM show that high concentration NH3 (10 ppm) can suppress benzoic acid formation over 70 % at 2 %RH, and 50 % inhibition at 17 %RH. This inhibition intensifies to 80 % when comparing high-NH3/2% RH to low-NH3/17% RH conditions. The consistency between the results from ESI, GAIS and MCM simulation confirms the role of NH3 in competitively reacting with C7-SCI.

Since C7-SCI-derived products (oligomers and benzoic acid) were greatly suppressed with the presence of NH3, where did C7-SCIs go? Our previous study on the isoprene-ozonolysis system found that NH3 can react with SCIs to produce amines, thereby inhibiting the original oligomerization pathway of SCIs and reducing SOA yields (Li et al., 2024). This is consistent with the phenomenon observed in this study and may be due to the same mechanisms, indicating that the reaction between NH3 and SCIs may be common in alkenes.

Referring to the reaction mechanism between C1-SCI and NH3 (Jørgensen and Gross, 2009; Misiewicz et al., 2018; Liu et al., 2018b; Chao et al., 2019a, b; Chhantyal-Pun et al., 2019), and the reaction mechanism between C4-SCI and NH3 from isoprene (Li et al., 2024), C7-SCI should react with NH3 to produce a molecule C7H9O2N. Online GAIS-Orbitrap MS measurements identified a nitrogen-containing product at m/z 140.071 with the molecular formula C7H10O2N+ (Fig. 2a), which is in good agreement with the predicted product C7H9O2N. However, it should be noted that the ammonium adduct ion of benzoic acid is also 140, and its molecular formula is the same as C7H10O2N+ (C7H6O2+NH4+, m/z 140.071). We worried that this might affect the determination of C7H10O2N+. Therefore, to rule out the potential interference introduced by benzoic acid-ammonium adducts, we first compared the MS² spectra of m/z 140.071 (C7H10O2N+) and benzoic acid. Since ammonium ions are easily separated, the MS² of the ammonium adduct ion of benzoic acid may be mainly from m/z 123.044 (C7H7O2+). Results show that the MS² C7H10O2N+ (m/z 140.071) is different to the MS² C7H7O2+ (m/z 123.044) (Fig. 2b). Different MS² spectra confirm that the molecule C7H10O2N+ is a unique new amine species, rather than an ammonium adduct derived from benzoic acid. We also conducted online observations by introducing NH3 into pure benzoic acid vapor and found that no signal at m/z 140.071 was detected, thus excluding the possibility of adducts. These prove that (C7H10O2N+) is not an adduct ion of benzoic acid and NH4+, but a newly generated species.

Based on our previous study on the reaction mechanism between NH3 and SCI from isoprene, the molecule C7H10O2N+ (m/z 140.071) should contain a peroxide bond. To determine the presence of peroxide bond in the molecule of C7H10O2N+, we further conducted iodometry kinetic experiments based on the selective reaction of I- ions with peroxide bonds. The chromatographic results of iodometry kinetic experiments showed that the peak of C7H10O2N+ appeared at 21.07 min. While in the control sample with added KI, its peak intensity at 21.07 min was suppressed by almost 100 % (Fig. 2a). This verifies the presence of a peroxide bond in C7H10O2N+, and meanwhile also confirms the molecular structure of the product from the reaction between C7-SCI and NH3.

Due to the high reactivity of peroxide bonds, the peroxide amine C7H9O2N is expected to be highly unstable and easily decomposed by removing one H2O2 or H2O (Smith and March, 2020), and may further decompose into imines and amides based on theoretical calculation (Banu et al., 2018; Ma et al., 2018). Online MS measurements detected an imine C7H8N+ (m/z=106.066) and an amide C7H8ON+ (m/z=122.060) as the dominant products (Fig. 2c and e). We compared the MS² spectra of C7H8N+ and C7H8ON+ with the simulated fragments of C7H7N with imine structure and of C7H7ON with amide structure from Mass Frontier, respectively. Results show that the MS² spectra of C7H8N+ and C7H8ON+ matched well with the simulation results by Mass Frontier (Fig. 2d and f). These results demonstrate that the unstable C7H9O2N further decomposes into C7H7N and C7H7ON. Previous theoretical study calculated that the reaction between NH3 and C7-SCI may produce the products C7H7N and C7H7ON (Banu et al., 2018; Ma et al., 2018), which further supports our findings.

Accurate quantification of C7H9O2N and its degradation products typically requires the use of standard gases to establish a calibration coefficient between mass spectrometry signal abundance and actual concentration. However, due to the current unavailability of standard materials for C7H9O2N and its products, direct quantification is challenging. Nevertheless, a previous study (Ma et al., 2018) estimated the rate constant for the reaction of C7-SCI with NH3 forming C7H9O2N (1.65×10-15cm3molecule-1s-1) via quantum chemical calculations. Based on this rate constant, we added the corresponding reaction into the MCM mechanism. Under Exp.10 experimental conditions, the simulated maximum concentration of C7H9O2N after 50 min of reaction was 28 ppb. Since the decomposition of C7H9O2N was not considered in the simulation, this concentration actually represents the total concentration of C7H9O2N and its two decomposition products. To further distinguish the specific concentrations of C7H9O2N and its two decomposition products, it needs to determine their decomposition rate constants. Fortunately, using online GAIS-Orbitrap MS monitoring data on abundance-time evolution, we can obtain the relative proportions among the three species: C7H9O2N(m/z140):C7H7N(m/z106):C7H7ON(m/z122). Based on this ratio, we introduced two decomposition reactions into the MCM mechanism and adjusted their rate constants so that the simulated concentration ratios matched the experimentally observed values. The corresponding concentrations of C7H9O2N (m/z 140), C7H7N (m/z 106) and C7H7ON (m/z 122) at the 50th minute were determined to be 23.8, 1.6 and 2.7 ppb in Exp. 10, with a deviation of ±17%. This allowed us to derive the two decomposition rate constants as (3.0±0.4)×10-5s-1 and (5.1±0.6)×10-5s-1. To date, only Banu et al. (2018) have reported theoretical values for the two decomposition rate constants of C7H9O2N, which are 7.02×10-16s-1 and 1.22×10-13s-1, respectively. It shows that the experimentally derived decomposition rate constants are approximately eight orders of magnitude higher than the theoretical values, indicating that C7H9O2N is a highly unstable compound. Then, the maximum yields of C7H9O2N, C7H7N and C7H7ON can be determined to be 8.1 %, 3.0 %, and 5.1 % in styrene-O3 system under conditions of Exp.10, respectively.

To quantify the expected atmospheric lifetime of C7H9O2N, we have considered 3 primary removal pathways: (1) Reaction with OH radicals, the reaction rate constant between C7H9O2N and OH was estimated to be 4.77×10-11cm3molecule-1s-1 using a tool of AOPWIN (Atmospheric Oxidation Program for Microsoft Windows) in EPI (Estimation Program Interface). Using an average OH radical concentration of 1.0×106moleculescm-3, the atmospheric lifetime of τOH=5.8h; (2) Photolysis: Based on the general photolysis rates of peroxides 1.3×10-6s-1 (Roehl et al., 2007), the photolytic lifetime τhν=214h; (3) Thermal decomposition: Based on our results, the decomposition rate of C7H9O2N is 8.1×10-5s-1, and its self-decomposition lifetime τdecomp=3.4 h. The total atmospheric lifetime was calculated to be 2.1 h based on 1/τ=1/τOH+1/τhν+1/τdecom. This suggests that C7H9O2N predominantly exists in the atmosphere as its more stable transformation products, namely the imine C7H7N and the amide C7H7ON.

Combining multiple experimental evidence, we propose the following reaction mechanism between NH3 and styrene-derived products and their role in SOA formation (Fig. 2g). Styrene reacts with O3 to form C7-SCI, which then generates benzoic acid and forms SOA through oligomerization (Yu et al., 2022, 2025). However, the addition of NH3 leads to a competitive reaction between both NH3 and H2O with C7-SCI, forming an unstable peroxide amine C7H9O2N, which rapidly further produces more stable imine C7H7N and amide C7H7ON. Furthermore, due to the presence of peroxide bonds and nitrogen, toxicity calculations show that the toxicity of C7H9O2N, C7H7N, and C7H7ON (High, class III) is significantly higher than that of benzoic acid (Low, class I) based on Cramer classification (Cramer et al., 1976).

The reaction pathway between NH3 and SCI identified in both isoprene and styrene systems indicates a general mechanism by which NH3 affects SOA molecular composition across in different olefin VOCs, highlighting the widespread impact of NH3 on aerosol chemistry, independent of the type of olefins. NH3 entering aerosols through reaction results in the generation of NOCs (e.g., amines, imines), which changes aerosol composition and potentially enhances light absorption and toxicity (Updyke et al., 2012) NH3 reduces SOA yields but increases NOC diversity. In recent years, NH3 emissions have increased globally, driven by agricultural and industrial activities (Fu et al., 2017; Kuttippurath et al., 2020; Liu et al., 2018a; Meng et al., 2020). Our study suggests that increasing NH3 levels may suppress SOA from isoprene and styrene, and affect regional aerosol budgets. Further research is needed to determine whether it has an impact on other olefins. Current models ignore the role of NH3 in SOA chemistry, and may overestimate the formation of SOA in NH3-rich environments. Integrating the novel NOC formation pathway from NH3 and SCI into the current model framework is crucial for improving climate and health predictions of aerosols.