The experiments were conducted in a large outdoor photochemical smog chamber located on one rooftop of the building at the Chinese Research Academy of Environmental Sciences (CRAES). It was a 56 m³ Teflon chamber covered with a well-controlled enclosure, inside where three fans were fixed to ensure uniform mixing (Li et al., 2021b). Before each experiment, the chamber was purged with zero air at a flow rate of 100 L min⁻¹ for at least 24 h to ensure cleanliness. Herein, all the experiments were performed from October to the ensuing January, and the initial conditions and experimental results could be found in Table S1 in the Supplement. As significant organic precursors of SOA, the liquid-phase n-heptylcyclohexane, and/or α-pinene with known volumes were injected into a U-shaped tube and then blown into the chamber with zero air after heating. In all experiments, the initial concentration of n-heptylcyclohexane (TCI Development Co., Ltd, Shanghai) was in the range of 49.0–119.8 ppb, while α-pinene (Beijing Inno Chem Science & Technology Co., Ltd) was in the range of 43.7–86.7 ppb. With HONO as the precursor for OH radicals and NO, it was prepared by the dropwise addition of 0.2 mL 5 wt % NaNO₂ into 0.4 mL 30 wt % H₂SO₄, formed NO, NO₂, and HONO from which were blown into the chamber by zero air likewise. The measured initial NO_x concentration in the chamber ranged from 89.9 to 157 ppb. NH₃ was introduced from a 500 ppm standard gas cylinder, the concentration of which was calculated based on the chamber volume and added volume of NH₃. The initial NH₃ concentration was in the range of 74.85–748.5 ppb. When the target species that introduced into the chamber were thoroughly mixed, the enclosure was opened, thus the photochemical reaction began. The entire photochemical reaction process of each experiment lasted about 7 h (usually from 10:00 to 17:00 LT), with a maximum temperature (T) of 14.4–40.7° and a maximum relative humidity (RH) of less than 15 % inside the chamber. The sunlight was utilized as the natural light source at JNO₂ of 0.9–4.4×10-3 s⁻¹. Sampling commenced immediately upon opening of the enclosure. The first filter collected samples continuously for 3.5 h (i.e., the first half of the experiment, designated as “former”) and was then replaced with a fresh one for an additional 3.5 h of continuous collection (i.e., the second half of the experiment, designated as “latter”). Each set of experiments was conducted at least 2 times to verify reproducibility and ensure robust results. Under background conditions (i.e., blank measurements), the chamber air contained SO₂ ≤ 0.5 ppb, NO_x ≤ 2 ppb, and O₃ ≤ 0.5 ppb, while released or formed VOCs and particles were negligible (<0.01 µg m⁻³).

The collected precursor samples (n-heptylcyclohexane and α-pinene) in the Tenax TA tube underwent quantitative analysis using thermal desorption gas chromatography (TD-GC, GC, 8890, Agilent, USA; TD, UNITY-xr, Germany) to obtain precursor concentrations and consumption rates within the chamber. The mass concentration, number concentration, surface mean diameter, and volume concentration of formed particles were monitored with a scanning mobility particle sizer (SMPS, model 3080, model 3081, and model 3772, TSI Inc., USA). Additionally, the qualitative analysis of organic chemical compounds, which were subjected to ultrasonic extraction from quartz films using 10 mL of methanol (20 min), was performed using an electrospray ionization quadrupole time-of-flight mass spectrometer (ESI-Q-ToF-MS, Bruker, Germany). Furthermore, an FTIR spectrometer (Bruker, Germany) equipped with a RTDLaTGs detector and a UV-Vis light spectrometer (Hitachi, U-3900) were applied to the qualitative detection of potential functional groups in the organic components. The qualitative and quantitative analysis of eluted aldehyde-ketones from collected samples were carried out using an HPLC-UV/MS (Shimadzu, Japan). The T and RH inside the chamber were monitored using the temperature and humidity sensors (Beijing Star Sensor Technology Co., Ltd.), while the irradiance was detected with a JNO₂ filter radiometer (Metcon, Germany). The consumption and generation of gaseous SO₂, NO_x and O₃ inside the chamber were monitored in real-time by online SO₂ (EC 9850, Ecotech, Australia), NO_x (EC 9841, Ecotech, Australia) and O₃ analyzers (EC 9830, Ecotech, Australia).

In this work, the OH exposure during the photo-oxidation process of precursor compounds were quantified by utilizing TD-GC, as illustrated by Eq. (1) (Mao et al., 2009): 1OHexposure=ln⁡[tracer]0[tracer]tk, where k denoted the rate constant for the reaction of the tracer and the OH radical.

Here, the tracer referred to a precursor compound capable of reacting with OH radicals. Given the rapid photo-oxidation rate of α-pinene, n-heptylcyclohexane was appointed as the trace. In experiments without n-heptylcyclohexane, α-pinene served as the tracer. At the standard temperature (298 K), the reaction rate constant for OH radical with n-heptylcyclohexane was 1.91×10-11 cm³ molec.⁻¹ s⁻¹, and with α-pinene was 5.23×10-11 cm³ molec.⁻¹ s⁻¹ (Atkinson and Arey, 2003). Furthermore, the physical significance of OH exposure was the total OH concentration from the onset to the detection period; hence, the cumulative time was employed in the calculation of OH exposure.

The SOA yield (Y) was defined as the fraction of reactive organic gases (ROG, µg m⁻³) converted into aerosols, and it could be calculated by Eq. (2): 2Y=ΔMoΔROG, where ΔMo (µg m⁻³) was the maximum mass concentration of SOA, obtained by deriving the data from the Scanning Mobility Particle Sizer (SMPS); ΔROG is the amount of precursor substances involved in the reaction.

The absorption of SOA generated in the experiment across the 300 to 700 nm was measured using a UV-Vis absorption spectrophotometer. The specific absorption coefficient, denoted as Abs_λ, was calculated based on Eq. (3) (Liu et al., 2021b): 3Absλ=Aλ-A700⋅V1Va⋅L⋅ln⁡10, where Aλ was the light absorption coefficient of SOA at the λ wavelength, A700 was the light absorption intensity background value, V1 was the volume of methanol with dissolved aerosols, Va was the volume of the sampled air, and L (1 cm) was the optical path length.

Based on the UV-Vis absorption spectra, the mass absorption efficiency (MAE_λ; m² g⁻¹) was characterized the ultraviolet absorption intensity of SOA, which was calculated according to Eq. (4) (Liu et al., 2021b): 4MAEλ=AbsλM, where Abs_λ (m⁻¹) was the light absorption coefficient of SOA at the λ wavelength, and M (µg m⁻³) was the mass concentration of methanol-soluble organic matter.

Mass defect was defined as the difference between exact mass and nominal mass of a certain compound (Roach et al., 2011).

The experiments were carried out according to the scheme of n-heptylcyclohexane + HONO (AL), n-heptylcyclohexane + HONO + NH₃ (AL-N), α-pinene + HONO (AP), α-pinene + HONO + NH₃ (AP-N), n-heptylcyclohexane + α-pinene + HONO (MIX) and n-heptylcyclohexane + α-pinene + HONO + NH₃ (MIX-N), and the initial conditions and results were summarized in Table S1. Since consistent results from replicate experiments effectively ruled out interference from random uncertainties, herein selected one representative dataset for presentation. In the background experiment, the mass concentration of inorganic particles formed from the photochemical reaction of NH₃ with NO_x was below 10 µg m⁻³, and the number concentration was <2000 # cm⁻³ (Fig. S1a in the Supplement). This is approximately 1–2 orders of magnitude lower than the total particle concentration in the organic reaction system. Therefore, the contribution of inorganic particle formation in the organic precursor system is considered negligible. By the conclusion of the reaction, the maximum particle diameters at varying NH₃ concentration converged to statistically similar values (Fig. S1b). As showed in Fig. 1a, the maximum of OH exposure was in the range of (5.6–7.9) × 10¹⁰ molec. cm⁻³ s corresponding to 10.4–14.6 simulated hours, assuming a diurnal average OH concentration of 1.5×106 molec. cm⁻³ (Mao et al., 2009; Li et al., 2024). The OH exposure continuously increased over time, leading to an enhanced atmospheric oxidation capacity. The reaction commenced with a decline in the concentration of reactants, and the consumption rate of which seemed to be accelerated by the addition of NH₃ (Fig. 1b). Upon illumination, the SOA mass concentration rose rapidly while NO_x levels declined. Compared with the NH₃-free system, the NH₃-containing system produced significantly higher SOA mass concentrations; moreover, the mixed organic system yielded higher SOA mass concentrations than the single one.

In terms of particle size, the average SOA diameter in the MIX-N experiment was distinctly larger than that in the AL-N experiment but almost identical to that in the AP-N experiment, implying that the disparity stemmed primarily from precursor type rather than NH₃. Previous work likewise indicated that the average SOA diameter in mixed systems was governed by the highly reactive BVOC precursors (Cui et al., 2024). As showed in Fig. 2, the average diameter of particles from photochemical oxidation ranged from 300 to 450 nm in the MIX-N and MIX experiment, ranged from 200 to 300 nm in the AL-N and AL experiment, and ranged from 250 to 450 nm in the AP-N and AP experiment. Obviously, the absence of any appreciable shift in average diameter between NH₃-present and NH₃-free conditions further corroborated that NH₃ exerted no significant enhancement on particle size. However, Li et al. (2018) reported that NH₃ could increase SOA size-growth potential; this discrepancy was attributable to their lower reactive organic compounds-to-primary NO_x concentration ratio (VOC / NO_x) (1.6–4.9 ppb C ppb⁻¹). Under such high-NO_x conditions, the photo-oxidation of organics preferentially drove post-nucleation growth, an effect that was largely muted in our experiments with higher VOC / NO_x (3.1–16.2 ppb C ppb⁻¹) (Li et al., 2018, 2022c).

In the process of each experiment, particles were generated rapidly within the first hour of the reaction (Fig. 2). When the maximum number concentration reached approximately 10⁴ # cm⁻³, strong particle growth appeared as a clear banana shape. Compared to the NH₃-free experiments, the maximum SOA yield of the mixed organic system rose from 14.3 % to 40 % when NH₃ was present, indicating that the introduction of NH₃ led to a marked escalation in SOA production (Fig. 3a). Although temperature could significantly affect the partitioning process of organic compounds and lower temperatures could sensibly facilitated the formation of particles (Li et al., 2022d), the particle mass concentration exhibited a pronounced enhancement in the NH₃-present systems under identical organic precursor as well as comparable temperature and VOC / NO_x ratio, suggesting that NH₃ played a facilitative role in the particle formation aligned with other research (Xu et al., 2021). It was probably because NH₃ could react with certain compounds found in SOA, contributing to convert itself into nitrogen-containing organic compounds (NOC) that remained in particles (Laskin et al., 2015), such as imines and (O'Brien et al., 2013). On the other hand, both nuclei coagulation and condensation caused by acid–base reactions between NH₃ and organic acids (such as pinic acid and pinonic acid) were responsible for the SOA growth (Xu et al., 2021).

Identifying the chemical class of SOA was essential to elucidating how NH₃ influences the photo-oxidation of binary organic mixtures; to this end, we derived functional group information of the SOA by resolving the absorption spectrum. The infrared absorption spectrum featured peaks that correspond to the functional groups of organic matter in the SOA. Various characteristic functional groups of the main products could be differentiated by identifying the positions of these peaks (Fig. 4). The peak observed at 2923 and 2855 cm⁻¹ originated from the O–H stretching vibration. The characteristic peak at 1645 cm⁻¹ corresponded to the C=O stretching vibrations of the carbonyl group in either aldehyde or amide. The peak around 1054 cm⁻¹ was considered as the O–H bending vibration or C–O stretching vibration, while the distinctive peak at 1382 cm⁻¹ represented the –NO₂ stretching vibration of the nitro group (Babar et al., 2017; Huang et al., 2018). These peaks were consistently found in both MIX-N, AL-N and AP-N experiments, indicating that the predominant products from the NH₃-containing systems were compounds with carbonyl functionalities, along with certain NOC (Liu et al., 2015). Especially in systems containing NH₃, the enhanced peak at 1382 cm⁻¹ suggested that the introduction of NH₃ led to an increased light absorption of SOA, resulting from the enhancement of NOC formation (Wang et al., 2022).

In the UV absorption profiles, the peak at 300 nm was found when NH₃ participating in the reaction, which suggested the formation of carbonyl groups (Martinez et al., 1992). However, the MAE in each system were relatively small, and the light absorption within the UV-visible spectrum were weak. This could be attributed to the absence of conjugated double bonds in the SOA derived from n-heptylcyclohexane and α-pinene (Lambe et al., 2012). Notably, the latter phase of each experiment showed a more pronounced light absorption than the former phase. The MAE at 300 nm during the former half of the experiments followed the sequence MIX-N > AL-N > AP-N, whereas followed MIX-N ≈ AP-N > AL-N in the latter half. These observations implied that more light-absorbing substances generated as the reaction progressed. Furthermore, the relative magnitude of SOA mass concentrations can be inferred based on the MAE values. The SOA generated from the n-heptylcyclohexane and α-pinene mixture contained a higher concentration of light-absorbing substances per unit mass.

To gain a more comprehensive understanding of the mechanisms of SOA formation, we conducted an analysis of oxygenated VOCs (OVOCs) intermediates, key carbonyl compounds identified in the reaction products (Fig. 9). In the MIX experiment, hexanal peaked at 1.1 µg m⁻³ during the former half of the reaction, and was subsequently consumed by fragmentation or further oxidation. The highest concentration in the latter half of the reaction was heptanal (2.3 µg m⁻³), which could be attributed to the fact that the heptane branch was first broken and further oxidation was required to break into smaller branches, but sufficient oxidation environment cannot be provided by largely consumed oxidants in the later stage of the reaction. In this system, the total amount of aldehydes and ketones was relatively small, presumably because the aldehydes/ketones reacted with intermediates or products and were transferred to the particle phase. When NH₃ was present, the mass concentrations of all carbonyls rose sharply, indicating that NH₃ either initiated cross reactions and side-reactions within the mixture or reacted with specific intermediates, driving more oxidation/fragmentation of the intermediates into aldehydes and ketones.

In the AL experiment, every carbonyl except crotonaldehyde and heptanal was more abundant in the latter half of the reaction. This might be due to the unstable structure of these two compounds, which caused their rapid oxidation to other substances. Moreover, relative to the NH₃-free case, the NH₃-present system produced substantially more aldehydes and ketones with ≤4 carbon atoms, demonstrating that NH₃ enhanced carbonyl compounds formation in the system, especially that with small molecule.

In the AP experiment, compared with the NH₃-free system, every carbonyl except acetaldehyde increased when NH₃ was present. The decrease of acetaldehyde presumably because NH₃ reacted directly with it to form nitrogen-containing products that partitioned into the particle phase, lowering its gas-phase concentration.

In summary, NH₃ redirected the oxidation pathway through competitive reactions, yielding a larger fraction of lower molecular weight products and small carbonyls.

Under high-NO_x conditions, the photo-oxidation reaction of n-heptylcyclohexane and α-pinene mixture began with H-abstraction by OH radical and thus reaction with O₂ to form an alkyl peroxy radical (RO₂). When NH₃ co-existed with n-heptylcyclohexane, α-pinene and NO_x, it markedly enhanced SOA production from the photochemical oxidation of the binary mixture. Cross-reactions between n-heptylcyclohexane, α-pinene and their organic intermediates promoted nucleation and growth of high-molecular-weight oligomers; in particular, peroxy radicals derived from n-heptylcyclohexane added across the double bond of α-pinene, yielding mixed oxidation products such as alkylated terpene oxides. Concurrently, NH₃ neutralised organic acids to form low-volatility ammonium salts, strongly depressing their saturation vapour pressures. NH₃ also reacted with OH radicals or organic radicals to generate nitrogen-containing radicals that further evolved into N-containing molecules, again lowering volatility. Moreover, NH₃ participated in Maillard chemistry with carbonyls, producing N-heterocycles (e.g. imidazoles, Fig. 10) that subsequently fragmented into smaller N-heterocycles and organic acids. Existing observational evidence had indicated that the presence of NH₃ could substantially enhance the formation of imidazole compounds (Liu et al., 2023). These low-volatility products underwent further multigenerational oxidation, giving highly oxygenated and nitrated species. Through these intertwined pathways NH₃ substantially reduced the overall volatility of the photo-oxidation products from the n-heptylcyclohexane and α-pinene mixture, efficiently enhancing SOA formation. By reacting with organic peroxy radicals formed during oxidation, NH₃ produced low volatility compounds that partitioned directly into the particle phase. This partitioning interrupted self-/cross-reactions of RO2⚫ and suppressed the formation of large OOMs. Meanwhile, multi-step oxidation initiated by NH₃ reacting with carbonyls additionally generated aldehydic and ketonic intermediates, further enriching the gas-phase carbonyl compounds.