Nocturnal atmospheric synergistic oxidation reduces the formation of low-volatility organic compounds from biogenic emissions

Volatile organic compounds (VOCs) are often subject to synergistic oxidation by different oxidants in the atmosphere. However, the exact synergistic-oxidation mechanism of atmospheric VOCs and its role in particle formation remain poorly understood. In particular, the reaction kinetics of the key reactive intermediates, organic peroxy radicals (RO₂), during synergistic oxidation is rarely studied. Here, we conducted a combined experimental and kinetic modeling study of the nocturnal synergistic oxidation of α-pinene (the most abundant monoterpene) by O₃ and NO₃ radicals as well as its influences on the formation of highly oxygenated organic molecules (HOMs) and particles. We find that in the synergistic O₃ + NO₃ regime, where OH radicals are abundantly formed via decomposition of ozonolysis-derived Criegee intermediates, the production of C_xH_yO_z HOMs is substantially suppressed compared to that in the O₃-only regime, mainly because of the depletion of α-pinene RO₂ derived from ozonolysis and OH oxidation by those arising from NO₃ oxidation via cross reactions. Measurement–model comparisons further reveal that the cross-reaction rate constants of NO₃-derived RO₂ with O₃-derived RO₂ are on average 10–100 times larger than those of NO₃-derived RO₂ with OH-derived RO₂. Despite a strong production of organic nitrates in the synergistic-oxidation regime, the substantial decrease in C_xH_yO_z HOM formation leads to a significant reduction in ultralow- and extremely low-volatility organic compounds, which significantly inhibits the formation of new particles. This work provides valuable mechanistic and quantitative insights into the nocturnal synergistic-oxidation chemistry of biogenic emissions and will help to better understand the formation of low-volatility organic compounds and particles in the atmosphere.

1 Introduction

Earth's atmosphere is a complex oxidizing environment in which multiple oxidants coexist. During nighttime, NO₃ radicals (generated by the reaction of NO₂ and O₃) and O₃ contribute significantly to the oxidation of volatile organic compounds (VOCs) (Huang et al., 2019), while during daytime, the fast photolysis of NO₃ radicals and rapid photochemical formation of OH radicals and O₃ make the latter two the major oxidants for VOCs (Zhang et al., 2018). Therefore, the degradation of ambient VOCs is subject to concurrent oxidation by different oxidants. Gas-phase oxidation of VOCs from biogenic emissions (BVOCs) by these major atmospheric oxidants produces a key type of reactive intermediate, organic peroxy radicals (RO₂), a portion of which can undergo fast autoxidation, forming a class of highly oxygenated organic molecules (HOMs) with low volatilities (Jokinen et al., 2014; Mentel et al., 2015; Berndt et al., 2016; Zhao et al., 2018; Iyer et al., 2021; Shen et al., 2022; Ehn et al., 2014). HOMs typically contain six or more oxygen atoms and play a key role in the formation of atmospheric new particles and secondary organic aerosol (SOA) (Kirkby et al., 2016; Berndt et al., 2018; Zhao et al., 2018; Ehn et al., 2014; Bianchi et al., 2019), which have important influences on air quality (Huang et al., 2014), public health (Pye et al., 2021), and Earth's radiative forcing (Shrivastava et al., 2017).

Due to the complexity of oxidation mechanisms of BVOCs, previous laboratory studies typically featured only one oxidant and a single SOA precursor (Berndt et al., 2016; Berndt, 2021; Claflin et al., 2018; Iyer et al., 2021; Boyd et al., 2015). However, the synergistic oxidation by different oxidants may significantly alter the fate of RO₂ intermediates, therefore influencing the formation of HOMs and SOA (Bates et al., 2022). Recently, a field study at a boreal forest site in Finland observed a series of nitrate-containing HOM dimers from the coupled O₃ + NO₃ oxidation of monoterpenes (Zhang et al., 2020). At the same site, Lee et al. (2020) found that the synergistic oxidation of BVOCs by OH radicals and O₃ contributed to the largest fraction of SOA. These studies suggest that the synergistic oxidation of BVOCs by different oxidants plays an important role in the formation of HOMs and SOA in the atmosphere and highlight the need to investigate the synergistic-oxidation mechanisms of BVOCs for a better representation of atmospheric particle formation.

Several laboratory studies have attempted to address the role of synergistic oxidation of BVOCs in the formation of new particles and SOA (Kenseth et al., 2018; Inomata, 2021; Liu et al., 2022; Li et al., 2024). Kenseth et al. (2018) identified a suite of dimer esters in flow tube experiments that can be only formed from the OH+O₃ synergistic oxidation of β-pinene. These dimers exhibit extremely low volatility and contributed 5.9 %–25.4 % to the total β-pinene SOA. Similarly, Inomata (2021) found that the presence of OH radicals during α-pinene ozonolysis is a key factor in the production of low-volatility organic species and significantly promotes new particle formation (NPF). On the other hand, the addition of O₃ in the monoterpene photooxidation system also significantly increases the SOA mass yield (Liu et al., 2022). In addition, a recent chamber study by Bates et al. (2022) showed that the synergistic oxidation of α-pinene by NO₃ radicals and O₃ can significantly enhance the SOA yield compared to the NO₃ + α-pinene regime, which has nearly 0 % SOA yield (Fry et al., 2014; Hallquist et al., 1999; Mutzel et al., 2021), and they revealed that the SOA yield in the NO₃ + O₃ oxidation system largely depends on the RO₂ fates. Most recently, Li et al. (2024) found that during α-pinene ozonolysis, the presence of nitrooxy RO₂ radicals formed from NO₃ oxidation can significantly suppress the production of ultralow-volatility organic compounds (ULVOCs) and thereby NPF. These laboratory studies together provide growing evidence that synergistic oxidation of BVOCs by different oxidants has profound impacts on atmospheric particle formation. However, the specific synergistic mechanisms of different oxidants and oxidation pathways remain obscure. Although a few studies underscored the importance of the RO₂ fates (Bates et al., 2022; Li et al., 2024), the exact interactions between RO₂ species derived from different oxidants are still unclear, and quantitative constraints on the reaction rate of different RO₂ species are quite limited.

Here we conducted an investigation of the synergistic O₃ + NO₃ oxidation of α-pinene, one of the most abundant monoterpenes in the atmosphere, using a combination of laboratory experiments and detailed kinetic modeling and focusing on the fate of RO₂ intermediates arising from different oxidation pathways. The α-pinene oxidation experiments were conducted in a custom-built flow reactor. The molecular composition of RO₂ species and HOMs in different oxidation regimes was characterized using a chemical-ionization atmospheric-pressure interface time-of-flight mass spectrometer (CI-APi-ToF MS) employing a nitrate ion source. The measured distributions of specific RO₂ and HOMs across different oxidation regimes were fitted with a kinetic model using Master Chemical Mechanisms (MCM v3.3.1) updated with recent advances of α-pinene RO₂ chemistry (Wang et al., 2021; Iyer et al., 2021; Shen et al., 2022; Zang et al., 2023), which allows for quantitative constraints on RO₂ kinetics and synergistic-oxidation mechanisms. Atmospheric relevance of the experimental results was evaluated by modeling the investigated oxidation chemistry under typical nocturnal atmospheric conditions.

2 Materials and methods

2.1 Flow tube experiments

Experiments of α-pinene oxidation in different regimes (i.e., synergistic O₃ + NO₃ oxidation vs. the O₃-only regime) were carried out under room temperature (298 K) and dry (relative humidity <5 %) conditions in a custom-built flow tube reactor (FTR; Fig. S1 in the Supplement). O₃ and NO₂ were added to a glass tube (Fig. S1) to form the NO₃ radical and its precursor, N₂O₅:

$$\begin{array}{}\text{(R1)}& {\displaystyle }{\mathrm{NO}}_{\mathrm{2}}+{\mathrm{O}}_{\mathrm{3}}\to {\mathrm{NO}}_{\mathrm{3}}+{\mathrm{O}}_{\mathrm{2}},\text{(R2)}& {\displaystyle }{\mathrm{NO}}_{\mathrm{3}}+{\mathrm{NO}}_{\mathrm{2}}\leftrightarrow {\mathrm{N}}_{\mathrm{2}}{\mathrm{O}}_{\mathrm{5}}.\end{array}$$

O₃ was generated by passing a flow of ultrahigh-purity (UHP) O₂ (Shanghai Maytor Special Gas Co., Ltd.) through a quartz tube housing a pen-ray mercury lamp (UV-S2, UVP Inc.), and its concentration was measured by an ozone analyzer (T400, API). NO₂ was obtained from a gas cylinder (15.6 ppm, Shanghai Weichuang Standard Gas Co., Ltd.). The initial NO₂ concentration in the flow tube was 4.5–6.4 ppb. To prevent the titration of NO₃ radicals by NO, all the experiments were performed without the addition of NO. The total airflow in the NO₃ generation glass tube was 0.6 and 0.4 L min⁻¹ for the gas-phase HOM and SOA formation experiments, respectively. The produced N₂O₅ and NO₃ radicals, as well as the excessive O₃, were added into the FTR to initiate α-pinene oxidation. For the O₃-only experiments, only O₃ was added into FTR.

The α-pinene gas was generated by evaporating a defined volume of its liquid (99 %, Sigma-Aldrich) into a cleaned and evacuated canister (SilcoCan, Restek) and then added into the FTR through a movable injector at a flow rate of 22–108 mL min⁻¹. The initial concentration of α-pinene in the flow reactor ranged from 100–500 ppb. In some experiments, the gas of cyclohexane (∼100 ppm), which was generated by bubbling a gentle flow of UHP N₂ through its liquid (liquid chromatography–mass spectrometry grade, CNW), was added into the flow reactor as a scavenger of OH radicals formed from α-pinene ozonolysis.

For experiments characterizing the formation of HOMs, the total airflow in the FTR was 10.8 L min⁻¹ and the residence time was 25 s. The short reaction time and the small amount of reacted α-pinene (see Table S1 in the Supplement) in these experiments prevented the formation of particles. For the experiments characterizing the formation of SOA particles, a larger FTR was used, with a total airflow of 5 L min⁻¹ and a residence time of 180 s. A summary of the conditions including the simulated concentrations of NO₂, N₂O₅, and NO₃ radicals, as well as the concentration of α-pinene oxidized by each oxidant in different experiments, is given in Table S1.

The gas-phase RO₂ radicals and closed-shell products were measured using a nitrate-based CI-APi-ToF MS (abbreviated as nitrate CIMS, chemical-ionization mass spectrometer; Aerodyne Research, Inc.), which has been described in detail previously (Zang et al., 2023). A long-ToF MS with a mass resolution of ∼10 000 was used here. The mass spectra within the $m/z$ range of 50–700 were analyzed using the tofTools package developed by Junninen et al. (2010) based on MATLAB. The total ion counts (TICs) with values of (5.9–6.2) ×10⁴ cps (counts per second) are similar under different reaction conditions. In this study, we assume that the C_xH_yO_z HOMs derived from ozonolysis and OH oxidation of α-pinene exhibit the same sensitivity in nitrate CIMS. However, the highly oxygenated organic nitrates (ONs) may have a significantly lower sensitivity compared to the C_xH_yO_z HOM counterparts, given that the substitution of the –OOH or –OH groups by the –ONO₂ group in the molecule would reduce the number of H-bond donors, which is a key factor in determining the sensitivity of nitrate CIMS (Shen et al., 2022; Hyttinen et al., 2015). Recently, Li et al. (2024) used CI-Orbitrap with ammonium or nitrate reagent ions to detect oxygenated organic molecules in the synergistic O₃ + NO₃ regime and found that both the ion intensity of ONs and their signal contribution to the total dimers were much lower when using nitrate as reagent ions.

A scanning mobility particle sizer (SMPS, TSI), consisting of an electrostatic classifier (model 3082); a condensation particle counter (model 3756); and a long or nano differential mobility analyzer (model 3081 and 3085) with a measurable size range of 4.6–156.8 nm or 14.6–661.2 nm, respectively, was employed to monitor the formation of particles in the flow tube. During the HOM formation experiments, even under conditions with the highest initial α-pinene concentration (500 ppb), only a tiny number of particles was formed, with mass concentrations of (6.4 ± 1.7) × 10⁻³ and (1.0 ± 0.4) × 10⁻² µg m⁻³ and number concentrations of 574 ± 148 and 256 ± 68 cm⁻³ in the O₃-only (experiment 5) and O₃ + NO₃ (experiment 11) regimes, respectively. These results suggest that the formation of SOA particles in the HOM formation experiments is negligible and would have no significant influence on the fate of RO₂ and closed-shell products.

2.2 Estimation of HOM volatility

A modified composition-activity method was used to estimate the saturation mass concentration (C^∗) of HOMs in this study according to the approach developed by Li et al. (2016):

$$\begin{array}{}\text{(1)}& {\displaystyle }\begin{array}{rl}{\displaystyle }{\log }_{\mathrm{10}}{C}^{\ast }& {\displaystyle }=(n_{\mathrm{C}}^{\mathrm{0}}-n_{\mathrm{C}})b_{\mathrm{C}}-n_{\mathrm{O}}{b}_{\mathrm{O}}-\mathrm{2}{\displaystyle \frac{n_{\mathrm{C}}{n}_{\mathrm{O}}}{n_{\mathrm{C}}+n_{\mathrm{O}}}}{b}_{\text{co}}\\ {\displaystyle }& {\displaystyle }-n_{\mathrm{N}}{b}_{\mathrm{N}}-n_{\mathrm{S}}{b}_{\mathrm{S}},\end{array}\end{array}$$

where $n_{\mathrm{C}}^{\mathrm{0}}$ is the reference carbon number; n_C, n_O, n_N, and n_S are the atomic numbers of carbon, oxygen, nitrogen, and sulfur, respectively; b_C, b_O, b_N, and b_S are the contribution of each atom to log ₁₀C^∗, respectively; and b_co is the carbon–oxygen nonideality (Donahue et al., 2011). These b values were provided by Li et al. (2016).

It should be noted that the CHON compounds used in the data set by Li et al. (2016) are mostly amines, amides, and amino acids and only contain a limited number of ONs (0.07 %). Since different types of CHON compounds have very different vapor pressures (Isaacman-VanWertz and Aumont, 2021), this formula-based approach can be biased to estimate the C^∗ of ONs. Considering that the –ONO₂ and –OH groups have similar impacts on vapor pressure and that the CHON species are predominantly ONs in our study, all –ONO₂ groups are treated as –OH groups during the estimation of vapor pressure (Daumit et al., 2013; Isaacman-VanWertz and Aumont, 2021).

Gas-phase HOMs are grouped into five classes based on their log ₁₀C^∗ (Donahue et al., 2012; Bianchi et al., 2019; Schervish and Donahue, 2020), that is, ULVOCs (${\log }_{\mathrm{10}}{C}^{\ast }<-\mathrm{8.5}$), extremely low-volatility organic compounds (ELVOCs; $-\mathrm{8.5}<{\log }_{\mathrm{10}}{C}^{\ast }<-\mathrm{4.5}$), low-volatility organic compounds (LVOCs; $-\mathrm{4.5}<{\log }_{\mathrm{10}}{C}^{\ast }<-\mathrm{0.5}$), semi-volatile organic compounds (SVOCs; $-\mathrm{0.5}<{\log }_{\mathrm{10}}{C}^{\ast }<\mathrm{2.5}$), and intermediate-volatility organic compounds (IVOCs; $\mathrm{2.5}<{\log }_{\mathrm{10}}{C}^{\ast }<\mathrm{6.5}$).

2.3 Kinetic model simulations

Model simulations of specific RO₂ radicals and closed-shell HOMs formed in different oxidation regimes were performed to constrain the reaction kinetics and mechanisms using the Framework for 0-D Atmospheric Modeling (F0AM v4.1) (Wolfe et al., 2016), which employs MCM v3.3.1 (Jenkin et al., 2015). The α-pinene oxidation mechanism was updated with state-of-the-art knowledge regarding the chemistry of RO₂ autoxidation and cross reactions forming HOM monomers and dimers, respectively (Zhao et al., 2018; Wang et al., 2021; Iyer et al., 2021; Shen et al., 2022). The detailed updates have been described in our previous study (Zang et al., 2023). In particular, the formation and subsequent reactions of the ring-opened primary C₁₀H₁₅O₄ RO₂, the highly oxygenated acyl RO₂, and the C₁₀H₁₅O₂ RO₂ arising from H abstraction by OH radicals during α-pinene ozonolysis are included in the model according to recent studies (Iyer et al., 2021; Zhao et al., 2022; Zang et al., 2023; Shen et al., 2022).

To investigate the synergistic reactions of RO₂ derived from the oxidation of α-pinene by different oxidants, we added the cross reactions of the primary nitrooxy RO₂ derived from NO₃ oxidation (${}^{{\mathrm{NO}}_{\mathrm{3}}}{\mathrm{RO}}_{\mathrm{2}}$), i.e., C₁₀H₁₆NO₅ RO₂, with RO₂ derived from ozonolysis (^CIRO₂) and OH oxidation (^OHRO₂). Recently, Zhao et al. (2018) revealed the bulk rate constant for ^CIRO₂ and ^OHRO₂ self-reactions/cross reactions to be $\mathrm{2}\times {\mathrm{10}}^{-\mathrm{12}}$ ${\mathrm{cm}}^{\mathrm{3}}\mathrm{molec}{.}^{-\mathrm{1}}{\mathrm{s}}^{-\mathrm{1}}$, and Bates et al. (2022) constrained the rate constant for ${}^{{\mathrm{NO}}_{\mathrm{3}}}{\mathrm{RO}}_{\mathrm{2}}$ self-reactions/cross reactions to be $\mathrm{1}\times {\mathrm{10}}^{-\mathrm{13}}$–$\mathrm{1}\times {\mathrm{10}}^{-\mathrm{12}}$ ${\mathrm{cm}}^{\mathrm{3}}\mathrm{molec}{.}^{-\mathrm{1}}{\mathrm{s}}^{-\mathrm{1}}$. In the present study, the default rate constant for ${}^{{\mathrm{NO}}_{\mathrm{3}}}{\mathrm{RO}}_{\mathrm{2}}$ + ^CIRO₂ was set to $\mathrm{2}\times {\mathrm{10}}^{-\mathrm{12}}$ ${\mathrm{cm}}^{\mathrm{3}}\mathrm{molec}{.}^{-\mathrm{1}}{\mathrm{s}}^{-\mathrm{1}}$, the same as for self-reactions/cross reactions of ^CIRO₂ and ^OHRO₂. The default rate constant for ${}^{{\mathrm{NO}}_{\mathrm{3}}}{\mathrm{RO}}_{\mathrm{2}}$ + ${}^{{\mathrm{NO}}_{\mathrm{3}}}{\mathrm{RO}}_{\mathrm{2}}$ was set to $\mathrm{1}\times {\mathrm{10}}^{-\mathrm{12}}$ ${\mathrm{cm}}^{\mathrm{3}}\mathrm{molec}{.}^{-\mathrm{1}}{\mathrm{s}}^{-\mathrm{1}}$. The ratio of the cross-reaction rate constant of ${}^{{\mathrm{NO}}_{\mathrm{3}}}{\mathrm{RO}}_{\mathrm{2}}$ + ^CIRO₂ to that of ${}^{{\mathrm{NO}}_{\mathrm{3}}}{\mathrm{RO}}_{\mathrm{2}}$ + ^OHRO₂ was tuned to achieve a good measurement–model agreement for the distribution of specific RO₂ and HOMs across different oxidation regimes. Recent studies suggested that the ROOR^′ dimer formation rates from the highly oxygenated RO₂ are fast (Berndt et al., 2018; Molteni et al., 2019). As a result, a relatively high dimer formation branching ratio of 50 % was used for different RO₂ types (e.g., ^CIRO₂, ^OHRO₂, ${}^{{\mathrm{NO}}_{\mathrm{3}}}{\mathrm{RO}}_{\mathrm{2}}$) in the model, except for the reaction of ${}^{{\mathrm{NO}}_{\mathrm{3}}}{\mathrm{RO}}_{\mathrm{2}}$ + ${}^{{\mathrm{NO}}_{\mathrm{3}}}{\mathrm{RO}}_{\mathrm{2}}$, for which ROOR^′ dimer formation was not considered, given the extremely low signals of CHON₂ dimers observed in the synergistic-oxidation regime (see Sect. 3.1). With these default kinetic parameters, the RO₂ bimolecular lifetimes were predicted to be 10.9–25.9 s in the O₃-only regime and 8.4–11.8 s in the O₃ + NO₃ regime in the HOM formation experiments. Considering that the RO₂ cross-reaction kinetics remain highly uncertain, sensitivity analyses were performed to evaluate their influences on the results in this study (see Sect. 3.2). Previous studies indicated that the primary ${}^{{\mathrm{NO}}_{\mathrm{3}}}{\mathrm{RO}}_{\mathrm{2}}$ radicals arising from α-pinene are prone to lose the nitrate group and form pinonaldehyde with high volatility (Kurtén et al., 2017; Fry et al., 2014). Therefore, we did not consider the autoxidation of primary ${}^{{\mathrm{NO}}_{\mathrm{3}}}{\mathrm{RO}}_{\mathrm{2}}$ in the model. Considering the presence of NO₂ in the experiments, the reactions of ${\mathrm{RO}}_{\mathrm{2}}+{\mathrm{NO}}_{\mathrm{2}}\leftrightharpoons {\mathrm{ROONO}}_{\mathrm{2}}$ were also included in the model (Zang et al., 2023).

3 Results and discussion

3.1 Molecular distribution of RO₂ and HOMs in the synergistic-oxidation regime

The abundance of gas-phase RO₂ species and HOMs in different oxidation regimes is shown in Fig. 1a. The species signals are normalized by the total reacted α-pinene in each regime. Compared to the O₃-only regime, the normalized signals of total RO₂ and HOMs decrease by 63 %–68 % in the synergistic O₃ + NO₃ regime. Although NO₃ oxidation accounts for a considerable fraction of reacted α-pinene in the synergetic oxidation regime, the signal contributions of HOM ONs are not significant. This might be due to the low sensitivity of nitrate CIMS to the ONs formed involving NO₃ oxidation (Sect. 2.1). Although there remain considerable uncertainties in instrument sensitivities for different compounds, sensitivity analyses suggest that varying the CIMS sensitivities to RO₂ and HOMs by a factor of 10 would not significantly influence their relative distribution across different oxidation regimes (see Sect. S1 in the Supplement for details).

Figure 1 Distributions of RO₂ and HOMs in the O₃-only and O₃ + NO₃ regimes. (a) Signals of total RO₂, as well as HOM monomers and dimers normalized by the total reacted α-pinene in each oxidation regime (experiments 1–5, 7–11). (b) Relative changes in the normalized signals of C_xH_yO_z HOMs in the O₃ + NO₃ regime vs. the O₃-only regime. Ion signals are normalized to $\mathrm{\Delta }[\mathit{\alpha }\text{-pinene}{]}_{{\mathrm{O}}_{\mathrm{3}}}$ in each oxidation regime to highlight the suppression effect of the synergistic chemistry between ${}^{{\mathrm{NO}}_{\mathrm{3}}}{\mathrm{RO}}_{\mathrm{2}}$ and ^CIRO₂ or ^OHRO₂ on C_xH_yO_z HOM formation. (c) Difference mass spectrum between the two oxidation regimes. The positive and negative peaks indicate the species with enhanced and decreased formation in the O₃ + NO₃ regime compared to the O₃-only regime, respectively.

Note that the initial concentrations of α-pinene and O₃ in the two oxidation regimes were the same. In addition, model simulations show that in the synergistic O₃ + NO₃ regime, over 97 % of OH radicals react with α-pinene and the depletion of OH by NO₂ is minor (0.2 %–1.3 %). Also, NO₃ radicals almost entirely (over 98.5 %) react with α-pinene, and their reaction with RO₂ has negligible influence on the fate of RO₂ (Fig. S2). Meanwhile, the depletion of acyl RO₂ by NO₂ only leads to a small reduction (4 %–5 % and 7 %–12 %, respectively) in total C_xH_yO_z HOM monomers and dimers in the synergistic regime compared to the O₃-only regime. As a result, the strong reduction in C_xH_yO_z HOM formation due to the presence of NO₃ oxidation is likely mainly due to (i) the fast competitive consumption of α-pinene by NO₃ radicals, which leads to a reduction in the reacted α-pinene by O₃ ($\mathrm{\Delta }[\mathit{\alpha }\text{-pinene}{]}_{{\mathrm{O}}_{\mathrm{3}}}$, Fig. S3) and thereby C_xH_yO_z HOM signals, and (ii) the cross reactions of ^CIRO₂ or ^OHRO₂ with ${}^{{\mathrm{NO}}_{\mathrm{3}}}{\mathrm{RO}}_{\mathrm{2}}$, which suppress the autoxidation and self-reactions/cross reactions of ^CIRO₂ and ^OHRO₂ to form C_xH_yO_z HOMs.

To quantify the contribution of cross reactions of ${}^{{\mathrm{NO}}_{\mathrm{3}}}{\mathrm{RO}}_{\mathrm{2}}$ with ^CIRO₂$/$^OHRO₂ to the suppressed formation of C_xH_yO_z HOMs in the synergistic-oxidation regime, C_xH_yO_z HOM signals shown in Fig. 1a are first normalized to $\mathrm{\Delta }[\mathit{\alpha }\text{-pinene}{]}_{{\mathrm{O}}_{\mathrm{3}}}$ in each oxidation regime and then compared between different oxidation regimes (see Fig. 1b). Notably, after excluding the influence of reduced $\mathrm{\Delta }[\mathit{\alpha }\text{-pinene}{]}_{{\mathrm{O}}_{\mathrm{3}}}$, the C_xH_yO_z HOM signals still drop by 32 %–33 % in the O₃ + NO₃ regime compared to those in the O₃-only regime, indicating a significant contribution of the coupled reactions between ${}^{{\mathrm{NO}}_{\mathrm{3}}}{\mathrm{RO}}_{\mathrm{2}}$ and ^CIRO₂ or ^OHRO₂ to suppressed C_xH_yO_z HOM formation.

Figure 1c shows a difference mass spectrum, highlighting the changes in species distribution between the two oxidation regimes. Almost all C_xH_yO_z HOM species decrease significantly in the O₃ + NO₃ regime compared to the O₃-only regime. Furthermore, a large set of HOM ON species are formed, despite their relatively low signals. It should be noted that no obvious signals of highly oxygenated ${}^{{\mathrm{NO}}_{\mathrm{3}}}{\mathrm{RO}}_{\mathrm{2}}$ (C₁₀H₁₆NO_x, x≥6) were observed by nitrate CIMS in the O₃ + NO₃ oxidation system. One possible reason is that nitrate CIMS exhibits relatively low sensitivity to the ONs. Secondly, the instrument's mass resolution is not high enough to differentiate the mass closure between some of ${}^{{\mathrm{NO}}_{\mathrm{3}}}{\mathrm{RO}}_{\mathrm{2}}$ and C_xH_yO_z HOMs with strong peaks (Table S2), limiting the detection of ${}^{{\mathrm{NO}}_{\mathrm{3}}}{\mathrm{RO}}_{\mathrm{2}}$ species. Furthermore, previous studies revealed that the primary ${}^{{\mathrm{NO}}_{\mathrm{3}}}{\mathrm{RO}}_{\mathrm{2}}$ radicals (i.e., C₁₀H₁₆NO₅ RO₂) in the α-pinene + NO₃ system mainly react to form pinonaldehyde (Kurtén et al., 2017; Perraud et al., 2010). It is likely that only a very small amount of ${}^{{\mathrm{NO}}_{\mathrm{3}}}{\mathrm{RO}}_{\mathrm{2}}$ can undergo intramolecular H shift and O₂ addition to form highly oxygenated ${}^{{\mathrm{NO}}_{\mathrm{3}}}{\mathrm{RO}}_{\mathrm{2}}$. It should be pointed out that although the primary C₁₀H₁₆NO₅ RO₂ species arising from NO₃ oxidation may not undergo fast autoxidation, they tend to efficiently terminate ^CIRO₂ and/or ^OHRO₂ and suppress the formation of C_xH_yO_z HOMs.

As shown in Fig. 1c, although several closed-shell monomeric HOM ONs have been observed in the synergistic-oxidation regime, only a few of them exhibit relatively high signals. Among them, C₁₀H₁₇NO₈ may be formed by the autoxidation of C₁₀H₁₆NO₆ RO₂ derived from the intramolecular H shift in primary ${}^{{\mathrm{NO}}_{\mathrm{3}}}\mathrm{RO}$ radicals (C₁₀H₁₆NO₄). In addition, although CI is a soft ionization method, the fragmentation of chemically labile species still occurs during the ionization in nitrate CIMS. It is possible that some of the dimeric HOM ONs are fragmented to C₁₀H₁₇NO₈ during measurements with nitrate CIMS. In a recent study by Li et al. (2024), C₁₀H₁₇NO₈ was also identified during the synergistic oxidation of α-pinene by O₃ and NO₃. However, the exact origin of this species remains to be determined.

The C₂₀ dimers with only one nitrogen atom are very likely to be formed from the cross reactions of ^CIRO₂ or ^OHRO₂ with ${}^{{\mathrm{NO}}_{\mathrm{3}}}{\mathrm{RO}}_{\mathrm{2}}$, which provides direct evidence for the synergistic RO₂ chemistry in the O₃ + NO₃ regime. The CHON₂ dimers were also observed in the O₃ + NO₃ regime, despite their signals that were much lower than those of CHON dimers, which is different from the recent studies by Bates et al. (2022) and Li et al. (2024), which found CHON₂ dimers account for an important fraction of the total dimer signals in the synergistic-oxidation regime. A potential explanation for this discrepancy is the difference in the instrument sensitivity in these studies (Sect. 2.1). In general, the nitrate CIMS has lower sensitivities to ONs than to the C_xH_yO_z HOM counterparts (Shen et al., 2022; Hyttinen et al., 2015). Bates et al. (2022) used CF₃O⁻ as the reagent ion of CIMS. Its sensitivity to ONs might be significantly higher than the nitrate ion. In addition, Li et al. (2024) observed a significantly lower signal contribution of CHON₂ dimers using CI-Orbitrap with nitrate reagent ions than with ammonium ions. Despite both using nitrate reagent ions, the nitrate CI-Orbitrap in Li et al. (2024) possibly exhibits higher sensitivities to ONs than the nitrate CIMS in our study.

3.2 Synergistic reaction efficiencies of different RO₂ species

In the O₃ + NO₃ regime, synergistic reactions are likely to occur between ^CIRO₂, ^OHRO₂, and ${}^{{\mathrm{NO}}_{\mathrm{3}}}{\mathrm{RO}}_{\mathrm{2}}$. Figure 2 shows the $\mathrm{\Delta }[\mathit{\alpha }\text{-pinene}{]}_{{\mathrm{O}}_{\mathrm{3}}}$-normalized signal ratios of specific C₁₀ RO₂ as well as their related C_xH_yO_z HOM monomers and dimers in the synergistic O₃ + NO₃ regime vs. the O₃-only regime. It should be noted that the second-generation oxidation processes are strongly inhibited by the excess of α-pinene in this study; thus the predominant type of RO₂ observed here is primary RO₂. Model simulations show that the H abstraction of α-pinene by OH radicals contributes less than 2 % to the formation of C₁₀H₁₅O_x RO₂ and related HOMs under different experimental conditions (Fig. S5). Therefore, C₁₀H₁₅O_x RO₂ observed in this study are primarily ^CIRO₂. Notably, the ^CIRO₂ (C₁₀H₁₅O_x) and related C₁₀H₁₄O_x HOMs decrease by ∼20 %–80 % in the O₃ + NO₃ regime (Fig. 2a and b), while the decreasing extent of ^OHRO₂ (C₁₀H₁₇O_x) and related C₁₀H₁₈O_x HOMs are significantly smaller (0 %–30 %). In particular, some of the most oxygenated C₁₀H₁₇O_x RO₂ and C₁₀H₁₈O_x HOMs (x≥9) even increase unexpectedly in the synergistic-oxidation regime. For the C₁₀H₁₆O_x HOMs that can be derived from the self-reactions/cross reactions of both ^CIRO₂ and ^OHRO₂, their reductions are at a medium level. Because of the very small contribution of acyl RO₂ to the total C₁₀ RO₂ (0.4 %) (Zang et al., 2023), their consumption by NO₂ leads to a reduction of less than 2 % in the C₁₀ ^CIRO₂ signals. Therefore, the more significant decrease in signals of ^CIRO₂ and related HOMs as compared to the OH-derived ones in the synergistic O₃ + NO₃ regime is primarily due to the more efficient cross reactions of ${}^{{\mathrm{NO}}_{\mathrm{3}}}{\mathrm{RO}}_{\mathrm{2}}$ with ^CIRO₂ than with ^OHRO₂. Because a large amount of ^CIRO₂ is terminated by ${}^{{\mathrm{NO}}_{\mathrm{3}}}{\mathrm{RO}}_{\mathrm{2}}$, less ^CIRO₂ is available to terminate ^OHRO₂. As a result, more ^OHRO₂ can undergo autoxidation to form highly oxygenated C₁₀H₁₇O_x RO₂ and C₁₀H₁₈O_x HOMs (x≥9), leading to an increase in signals of these species. Consistently, the signals of C₂₀ HOM dimers decrease by ∼20 %–40 % in the O₃ + NO₃ regime compared to those in the O₃-only regime, and the signal reduction in dimers (C₂₀H₃₀O_x) formed by ^CIRO₂ is slightly larger than that of the dimers (C₂₀H₃₄O_x) arising from ^OHRO₂ (Fig. 2c). Note that the highly oxygenated C₂₀H₃₄O_x dimers (x≥13) that can be formed from self-reactions/cross reactions of C₁₀H₁₇O_x RO₂ (x≥9) are not observed in this study, likely due to their low abundance and the limitation of instrument sensitivity.

Figure 2 Normalized signal ratios of (a) specific and total C₁₀H_15,17O_x RO₂ radicals, as well as (b–c) their related (b) C₁₀ HOM monomers and (c) C₂₀ HOM dimers in the O₃ + NO₃ regime vs. the O₃-only regime (experiments 1–5 and 7–11). Ion signals observed in each oxidation regime are normalized to $\mathrm{\Delta }[\mathit{\alpha }\text{-pinene}{]}_{{\mathrm{O}}_{\mathrm{3}}}$.

The above results are somewhat different from the most recent study by Li et al. (2024), which found that the measured C₁₀H₁₅O_x RO₂ increased slightly with NO₃ radicals, while C₁₀H₁₇O_5,7 RO₂ from OH chemistry decreased by a factor of 9. Li et al. (2024) indicated that additional C₁₀H₁₅O_x could be produced from the H-abstraction pathway of NO₃ oxidation of α-pinene. However, in the monoterpene oxidation system, the rate constant for H abstraction by NO₃ radicals is (4–10) × 10⁻¹⁷ ${\mathrm{cm}}^{\mathrm{3}}\mathrm{molec}{.}^{-\mathrm{1}}{\mathrm{s}}^{-\mathrm{1}}$, which is 10³–10⁴ times lower than that for the NO₃ addition channel (Martinez et al., 1998). Besides this, the subsequent reactions of RO₂ species formed from H abstraction by NO₃ radicals should be very similar to those derived from H abstraction by OH radicals, which was found to be not important for C_xH_yO_z HOM formation in the absence of NO (Zang et al., 2023). Therefore, the H abstraction of α-pinene by NO₃ radicals would have a negligible influence on C₁₀H₁₅O_x formation. As Li et al. (2024) used a low α-pinene concentration and relatively high O₃ and NO₃ concentrations in their experiments, the secondary oxidation of aldehydes, such as the substantially formed pinonaldehyde, by NO₃ radicals might be important, which could contribute to the additional formation of C₁₀H₁₅O_x RO₂. However, as noted above, the second-generation oxidation processes are strongly inhibited due to the excess of α-pinene in this study; therefore the formation of secondary C₁₀H₁₅O_x RO₂ is not important.

In addition, Li et al. (2024) reported that the fraction of α-pinene oxidized by OH radicals decreased from 44 % in the O₃ oxidation system to 6 % in the O₃ + NO₃ system, mainly due to the depletion of OH radicals by NO₂ and the competitive consumption of α-pinene by NO₃ radicals, which resulted in a significant decrease in C₁₀H₁₇O_5,7 radicals from OH chemistry as observed in their experiments. However, in the present study, because of the excess of α-pinene, over 97 % of OH radicals react with α-pinene and the depletion of OH by NO₂ is minor (0.2 %–1.3 %) in the O₃ + NO₃ regime. The reduction in the reacted α-pinene by OH radicals is less than 10 % compared to the O₃-only regime. As a result, a smaller decrease in C₁₀H₁₇O_5,7 radicals was observed in our study.

To gain quantitative constraints on the relative reaction efficiency of ${}^{{\mathrm{NO}}_{\mathrm{3}}}{\mathrm{RO}}_{\mathrm{2}}$ + ^CIRO₂ vs. ${}^{{\mathrm{NO}}_{\mathrm{3}}}{\mathrm{RO}}_{\mathrm{2}}$ + ^OHRO₂ (i.e., $k_{{\mathrm{NO}}_{\mathrm{3}}+\mathrm{CI}}/k_{{\mathrm{NO}}_{\mathrm{3}}+\mathrm{OH}}$), the signal ratios of C₁₀ ^CIRO₂ and ^OHRO₂ as well as their related C₁₀ HOMs in the synergistic-oxidation regime vs. the O₃-only regime were predicted using a kinetic model (see Sect. 2.3) with different $k_{{\mathrm{NO}}_{\mathrm{3}}+\mathrm{CI}}/k_{{\mathrm{NO}}_{\mathrm{3}}+\mathrm{OH}}$ ratios. Figure 3 shows a measurement–model comparison of those signal ratios. When the ratio of $k_{{\mathrm{NO}}_{\mathrm{3}}+\mathrm{CI}}/k_{{\mathrm{NO}}_{\mathrm{3}}+\mathrm{OH}}$ is smaller than or equal to 1, the simulated signal ratios of a large amount of RO₂ and HOMs differ significantly from the measured ratios, especially for some C₁₀H₁₇O_x RO₂ and C₁₀H₁₈O_x HOMs. When the ratio of $k_{{\mathrm{NO}}_{\mathrm{3}}+\mathrm{CI}}/k_{{\mathrm{NO}}_{\mathrm{3}}+\mathrm{OH}}$ is 10–100, there is a good measurement–model agreement for most of the RO₂ and HOMs. Therefore, we conclude that the cross-reaction rate constants of ${}^{{\mathrm{NO}}_{\mathrm{3}}}{\mathrm{RO}}_{\mathrm{2}}$ + ^CIRO₂ are on average 10–100 times larger than those for ${}^{{\mathrm{NO}}_{\mathrm{3}}}{\mathrm{RO}}_{\mathrm{2}}$ + ^OHRO₂. This different RO₂ cross-reaction efficiency is the main reason for the significantly larger decrease in the abundance of ^CIRO₂ and related HOMs as compared to the OH-derived ones (see Fig. 2).

Figure 3 Measurement–model comparisons of the signal ratios of different C₁₀ RO₂ and HOMs in the synergistic O₃ + NO₃ regime vs. the O₃-only regime. The cross-reaction rate constant of ${}^{{\mathrm{NO}}_{\mathrm{3}}}{\mathrm{RO}}_{\mathrm{2}}$ + ^CIRO₂ was set to $\mathrm{2}\times {\mathrm{10}}^{-\mathrm{12}}$ ${\mathrm{cm}}^{\mathrm{3}}\mathrm{molec}{.}^{-\mathrm{1}}{\mathrm{s}}^{-\mathrm{1}}$, and the rate of ${}^{{\mathrm{NO}}_{\mathrm{3}}}{\mathrm{RO}}_{\mathrm{2}}$ + ^OHRO₂ was varied from $\mathrm{2}\times {\mathrm{10}}^{-\mathrm{11}}$ to $\mathrm{2}\times {\mathrm{10}}^{-\mathrm{14}}$ ${\mathrm{cm}}^{\mathrm{3}}\mathrm{molec}{.}^{-\mathrm{1}}{\mathrm{s}}^{-\mathrm{1}}$ in the model.

As a competitive reaction pathway, the autoxidation rates of RO₂ can affect the extent to which RO₂ cross reactions influence the RO₂ fate and HOM formation. Therefore, sensitivity analyses of the autoxidation rate of RO₂ were conducted to evaluate its influence on the changes in RO₂ and related HOM concentrations in the synergistic O₃ + NO₃ regime vs. the O₃-only regime (Fig. S6). In these analyses, a $k_{{\mathrm{NO}}_{\mathrm{3}}+\mathrm{CI}}/k_{{\mathrm{NO}}_{\mathrm{3}}+\mathrm{OH}}$ ratio of 10 was used according to the above discussions. As the autoxidation rate of ^OHRO₂ increases from 0.28 to 10 s⁻¹, corresponding to the rate range reported in previous studies (Berndt et al., 2016; Zhao et al., 2018; Xu et al., 2019), the simulated reduction in highly oxygenated ^OHRO₂ and related C₁₀H₁₈O_x HOMs in the synergistic O₃ + NO₃ regime exhibits a slight decrease (<10 %) but still agrees reasonably well with the measured value (Fig. S6a–d). Considering that the autoxidation rates of ^CIRO₂ used in the model approach their upper limits reported in the literature, i.e., ∼1 s⁻¹ for the butyl ring-opened C₁₀H₁₅O₄ RO₂ (Iyer et al., 2021) and relatively smaller rates for ring-retained C₁₀H₁₅O₄ RO₂ (0.02–0.29 s⁻¹; see Scheme S1 in the Supplement) (Zhao et al., 2021), we also lowered the autoxidation rate constants of ^CIRO₂ by a factor of 10 to see its influence on RO₂ and HOM distribution in the O₃ + NO₃ regime. The simulated reduction in ^CIRO₂ and C₁₀H₁₄O_x HOMs in this case decreases by 7 %–16 % (Fig. S6e–h), while that of C₁₀H₁₆O_x HOMs increases by up to 31 % (Fig. S6i and j). However, the simulated results are still close to the measured values. These sensitivity analyses suggest that the uncertainty in the autoxidation rates of ^OHRO₂ and ^CIRO₂ could slightly affect the simulated distribution of RO₂ and HOMs across different oxidation regimes but not significantly change the $k_{{\mathrm{NO}}_{\mathrm{3}}+\mathrm{CI}}/k_{{\mathrm{NO}}_{\mathrm{3}}+\mathrm{OH}}$ ratio obtained in this study. Further sensitivity analyses on the rate constant and dimer formation branching ratio of RO₂ cross reactions indicate that the uncertainties in these reaction kinetics do not alter the conclusion regarding the $k_{{\mathrm{NO}}_{\mathrm{3}}+\mathrm{CI}}/k_{{\mathrm{NO}}_{\mathrm{3}}+\mathrm{OH}}$ ratio either (see details in Sects. S2 and S3).

Cyclohexane was added in some experiments as an OH scavenger to elucidate the role of ^OHRO₂ chemistry in HOM formation in the O₃ + NO₃ regime. In the presence of cyclohexane, most of the ^OHRO₂ (C₁₀H₁₇O_x) and related HOM monomers (C₁₀H₁₈O_x) and dimers (C₂₀H₃₂O_x and C₂₀H₃₄O_x) decrease by more than 70 % (Fig. 4), while ^CIRO₂ (C₁₀H₁₅O_x) and related HOM monomers (C₁₀H₁₄O_x) only decrease slightly. Accordingly, the reduction in C₂₀H₃₂O_x and C₂₀H₃₄O_x dimers is significantly larger than that of C₂₀H₃₀O_x. These results are in a good agreement with previous measurements (Zhao et al., 2018; Zang et al., 2023). The C₁₀H₁₆O_x species, which can arise from both ^CIRO₂ and ^OHRO₂, exhibit a medium reduction (Fig. 4b). It is interesting to note that with the addition of cyclohexane, there is a significant increase in C₂₀H₃₁NO_x, which is formed from the cross reactions of ^CIRO₂ with ${}^{{\mathrm{NO}}_{\mathrm{3}}}{\mathrm{RO}}_{\mathrm{2}}$. Such an enhanced production of C₂₀H₃₁NO_x as compared to the slightly deceased formation of C₂₀H₃₀O_x indicates that the ^CIRO₂ + ${}^{{\mathrm{NO}}_{\mathrm{3}}}{\mathrm{RO}}_{\mathrm{2}}$ reactions are competitive compared to the ^CIRO₂ + ^CIRO₂ and ^CIRO₂ + ^OHRO₂ reactions. As a result, when the ^OHRO₂ is depleted, the ^CIRO₂ that is supposed to react with ^OHRO₂ efficiently reacts with ${}^{{\mathrm{NO}}_{\mathrm{3}}}{\mathrm{RO}}_{\mathrm{2}}$ to form C₂₀H₃₁NO_x, leading to the increase in C₂₀H₃₁NO_x signals. Consistent with the experimental measurements, the model simulations show that the concentrations of C₂₀H₃₁NO_x in the O₃ + NO₃ regime increase with the addition of cyclohexane as an OH scavenger (Fig. S9). However, the simulated enhancement is slightly lower than the measurements, which might be due to the uncertainties in the RO₂ cross-reaction kinetics in the model.

Figure 4 Relative changes in signals of (a) C₁₀ RO₂, (b) C₁₀ HOMs, and (c) C₂₀ dimers due to the addition of 100 ppm cyclohexane as an OH scavenger derived in the synergistic O₃ + NO₃ regime (experiments 6 and 12).

3.3 Influence of synergistic oxidation on low-volatility organics and particle formation

Compared to the O₃-only regime, there are a remarkable reduction in C_xH_yO_z HOMs and a strong formation of HOM ONs due to the efficient cross reactions between ${}^{{\mathrm{NO}}_{\mathrm{3}}}{\mathrm{RO}}_{\mathrm{2}}$ and ^CIRO₂ in the synergistic-oxidation regime. This significant change in HOM composition and abundance would alter the volatility distribution of HOMs and influence the formation of particles. The volatilities of HOMs formed in the two oxidation regimes are estimated using a modified composition-activity method (see Sect. 2.2) and shown in Fig. 5. The abundance of C_xH_yO_z HOMs characterized as ULVOCs and ELVOCs decreases considerably in the synergistic O₃ + NO₃ regime compared to the O₃-only regime (Fig. 5), in agreement with the very recent observations by Li et al. (2024), who found that the presence of NO₃ radicals during α-pinene ozonolysis significantly reduced the abundance of ULVOCs. Although a substantial number of HOM ONs are formed in the O₃ + NO₃ regime, they generally have higher volatilities (i.e., characterized as ELVOCs to IVOCs) (Fig. 5). Therefore, the synergistic O₃ + NO₃ oxidation of α-pinene significantly reduces the formation of ULVOCs and increases the overall volatility of total HOMs.

Figure 5 Volatility distribution of C_xH_yO_z HOMs and HOM ONs formed in the O₃ + NO₃ regime and O₃-only regime (experiments 1 and 7). Ion signals in each oxidation regime are normalized to the corresponding total reacted α-pinene.

Figure 6a shows the particle number and mass concentrations formed in the two oxidation regimes in SOA formation experiments (experiments 13 and 14). The particle number concentration decreases by ∼50 %, whereas the particle mass concentration increases by a factor of 2 in the synergistic O₃ + NO₃ regime, compared to that in the O₃-only regime. The presence of NO₃ radicals during α-pinene ozonolysis reduces the abundance of ULVOCs, which are the key species driving particle nucleation, thereby leading to a reduction in the particle number concentration in the O₃ + NO₃ regime. On the other hand, substantial formation of HOM ONs is expected from the cross reactions of ${}^{{\mathrm{NO}}_{\mathrm{3}}}{\mathrm{RO}}_{\mathrm{2}}$ with ^CIRO₂ and ^OHRO₂ in the synergistic-oxidation regime (Li et al., 2024; Bates et al., 2022), although their signals are relatively low due to the low sensitivity of nitrate CIMS to ONs in this study. The newly formed HOM ONs have relatively higher volatilities and are inefficient in initiating particle nucleation, but they are able to partition into the formed particles and contribute to the particle mass growth. Meanwhile, as the particle number concentration decreases drastically in the synergistic-oxidation regime, more condensable vapors are available for each particle to grow to larger sizes (Fig. 6b), which would in turn favor the condensation of more volatile organic species including ONs due to the reduced curvature effect of the larger particles, ultimately resulting in an increase in SOA mass concentrations.

Figure 6 (a) Number and mass concentrations, as well as (b) the size distribution of particles formed from the ozonolysis and synergistic O₃ + NO₃ oxidation of α-pinene (experiments 13 and 14).

Recently, Bates et al. (2022) also found that in chamber experiments with seed particles, the SOA mass yields were significantly higher during α-pinene oxidation by O₃ + NO₃ than during ozonolysis, mainly due to the substantial formation and condensation of dimeric ONs. However, in the absence of seed particles, synergistic O₃ + NO₃ oxidation of α-pinene does not nucleate in their study. This phenomenon might be due to the high concentrations of NO₂ (72 ppb) and O₃ (102 ppb) as well as the relatively low concentration of α-pinene (27 ppb) in their experiments. As indicated by Bates et al. (2022), under these conditions NO₃ radicals were substantially formed and contributed to a dominant fraction (75 %) of α-pinene oxidation, which strongly inhibited the production of low-volatility species and particle nucleation.

3.4 Atmospheric relevance of experimental results

In the present study, the flow tube experiments were conducted under dry conditions. Although water vapor may affect the fate of Criegee intermediates and RO₂ radicals and thereby HOM formation during the oxidation of organics under humid conditions, there is growing evidence that such effects in the α-pinene oxidation system are small. Kinetic studies have found that the stabilized Criegee intermediates (SCIs) arising from α-pinene ozonolysis can undergo fast unimolecular decay at a rate constant of 60–250 s⁻¹ (Vereecken et al., 2017; Newland et al., 2018), which is rapid compared to their reaction with water vapor, in particular for syn-SCIs, under atmospheric conditions (Vereecken et al., 2017; Newland et al., 2018). In addition, the yield of OH radicals from Criegee decomposition is independent of RH (Atkinson et al., 1992; Aschmann et al., 2002). Consistent with the fast unimolecular reaction kinetics revealed by these studies, recent laboratory measurements have shown that the contribution of SCIs to the formation of gas-phase and particle-phase dimers is small (<20 %) during α-pinene ozonolysis (Zhao et al., 2018, 2022). Furthermore, the molecular composition and abundance of HOM monomers and dimers (Li et al., 2019) and the formation of particle-phase dimers (Zhang et al., 2015; Kenseth et al., 2018) do not change significantly with RH ranging from 3 % to 92 %. These studies suggest that the humidity condition does not strongly affect the HOM formation chemistry in the α-pinene ozonolysis system.

To evaluate the relevance of our experimental findings to the real atmosphere, we performed chemical model simulations of HOM formation from nocturnal synergistic O₃ + NO₃ oxidation of α-pinene under typical atmospheric conditions. In these simulations, constant concentrations of α-pinene (1 ppb), O₃ (30 ppb), NO (5 ppt), NO₂ (1.8 ppb), NO₃ radicals (0.2 or 1 ppt), OH radicals (5–50×10⁴ molecules cm⁻³), and HO₂ radicals (4 ppt), as well as a constant RH of 50 % and temperature of 298 K were used as typical nocturnal conditions in the boreal forest according to the field studies (Stone et al., 2012; Lee et al., 2016; Brown and Stutz, 2012; Geyer et al., 2003; Kristensen et al., 2016; Hakola et al., 2012; Liebmann et al., 2018). Considering the rapid deposition of oxidized biogenic compounds (Nguyen et al., 2015), a typical dilution lifetime of 5 h (i.e., $k_{\text{dil}}=\mathrm{1}/\mathrm{5}$ h⁻¹) was assumed in the model. According to the above analysis, the cross-reaction rate constants for ${}^{{\mathrm{NO}}_{\mathrm{3}}}{\mathrm{RO}}_{\mathrm{2}}$ + ^CIRO₂ and ${}^{{\mathrm{NO}}_{\mathrm{3}}}{\mathrm{RO}}_{\mathrm{2}}$ + ^OHRO₂ were set to $\mathrm{2}\times {\mathrm{10}}^{-\mathrm{12}}$ and $\mathrm{2}\times {\mathrm{10}}^{-\mathrm{13}}$ ${\mathrm{cm}}^{\mathrm{3}}\mathrm{molec}{.}^{-\mathrm{1}}{\mathrm{s}}^{-\mathrm{1}}$ in the model, respectively. The formation of RO₂ with oxygen numbers higher than 11 was not considered in the model, due to the large uncertainty in the autoxidation rate constants of the highly oxygenated RO₂. In fact, the autoxidation rate of the highly oxygenated RO₂ is expected to be small given the significant decrease in the number of active sites for intramolecular H abstraction in the molecule. As a result, the contribution of the most oxygenated HOMs to the total HOM monomers could be relatively small (Zhao et al., 2018; Claflin et al., 2018).

In the absence of NO₃ radicals (with NO₃ concentrations and formation rates set to zero), the amount of α-pinene consumed during 4 h of simulation is 1.04 ppb. When a relatively low NO₃ concentration (0.2 ppt) is considered (Fig. 7a), the amount of α-pinene consumed is 1.48 ppb, and the ozonolysis is the primary loss pathway of α-pinene (68 %), followed by NO₃ (30 %) and OH oxidation (2 %). The reactions of RO₂ + HO₂, RO₂ + NO, and RO₂ + RO₂ account for ∼49 %, ∼27 %, and ∼24 % of the total RO₂ fate, respectively (Fig. S10a). Compared to the ozonolysis of α-pinene, the synergistic O₃ + NO₃ oxidation leads to a reduction of 3 % and 13 % in the formation of C_xH_yO_z HOM monomers and dimers, respectively (Fig. 7b). Given that the concentrations of α-pinene and oxidants were held constant during the simulation, the consumptions of α-pinene by O₃ and OH radicals are the same across different oxidation regimes. Therefore, the decreases in the concentrations of C_xH_yO_z HOM monomers and dimers in the presence of NO₃ oxidation are mainly due to the cross reactions of ${}^{{\mathrm{NO}}_{\mathrm{3}}}{\mathrm{RO}}_{\mathrm{2}}$ with other RO₂. When the NO₃ concentration is as high as 1 ppt as reported in field studies (Liebmann et al., 2018), the consumption of α-pinene reaches 3.24 ppb, of which 68 % is contributed by NO₃ oxidation (Fig. 7c). Under this condition, the RO₂ + RO₂ reactions account for ∼34 % of the total RO₂ fate (Fig. S10b). As a result, the cross reactions of ${}^{{\mathrm{NO}}_{\mathrm{3}}}{\mathrm{RO}}_{\mathrm{2}}$ with other RO₂ play a more important role in the HOM formation. The production of C_xH_yO_z HOM monomers and dimers decreases by 12 % and 43 %, respectively, due to the presence of NO₃ oxidation (Fig. 7d). We note that the variation in RH from 0 %–90 % in the model has negligible influence on the relative changes in C_xH_yO_z HOMs under these nocturnal atmospheric conditions (Fig. S11). Considering that there are uncertainties in the dilution rate constant, a sensitivity analysis was performed by varying the k_dil in the range of 0.04–0.2 h⁻¹. It is found that the variation within these rate values does not significantly influence the response of C_xH_yO_z HOM dimer formation to concurrent NO₃ oxidation (Fig. S12).

Figure 7 Model simulations of α-pinene (AP) oxidation and HOM formation under typical nighttime conditions in the boreal forest. (a, c) Contributions of different loss pathways of α-pinene by different oxidants at NO₃ concentrations of 0.2 and 1 ppt, respectively. (b, d) Concentrations of C_xH_yO_z HOMs and HOM ONs formed by synergistic O₃ + NO₃ oxidation and ozonolysis of α-pinene under conditions corresponding to (a) and (c). The simulations were run for 4 h after an 8 h spin-up for intermediates and secondary species.

Field observations have shown that NO₃ radicals, O₃, and OH radicals all had important contributions to monoterpene oxidation during the early morning after sunrise and late afternoon before sunset in the southeastern United States (Zhang et al., 2018). In addition, relatively high nighttime OH concentrations of (2–10) × 10⁵ molecules cm⁻³ were measured in some areas such as Germany and New York City (Faloona et al., 2001; Geyer et al., 2003). As a result, a model simulation was conducted using OH concentration that was 10 times higher (5×10⁵ molecules cm⁻³). The concentration of NO₃ radicals is 1 ppt, and the concentrations of other species are the same as the values mentioned above. With a higher OH concentration, O₃, NO₃, and OH radicals account for 28 %, 61 %, and 11 % to the total α-pinene consumption, respectively (Fig. S13a). Compared to the results under low OH concentration, the formation of C_xH_yO_z HOM monomers and dimers are all enhanced under high OH concentration (Fig. S13b). This is mainly due to the promoted self-reactions/cross reactions of ^OHRO₂, as well as the promoted formation of C₁₀H₁₅O_x RO₂ derived from the H-abstraction pathway by OH radicals. Nevertheless, the presence of NO₃ oxidation still reduces the formation of C_xH_yO_z HOM dimers by 26 % (Fig. S13b).

Furthermore, model simulations under typical conditions in the southeastern United States (see details in Sect. S4) suggest that the coexistence of isoprene appears to exacerbate the suppression effect of synergistic oxidation on HOM formation from monoterpenes. As shown in Fig. S14, in the absence of isoprene, the synergistic O₃ + NO₃ oxidation of α-pinene leads to a reduction of 13 % and 24 % in the formation of C_xH_yO_z HOM monomers and dimers, respectively. When isoprene is present, as the isoprene + NO₃ oxidation produces a significant amount of nitrooxy RO₂ that can also scavenge α-pinene-derived ^CIRO₂ and ^OHRO₂ via cross reactions, the synergistic oxidation leads to a slightly larger reduction in C_xH_yO_z HOM monomers and dimers (15 % and 31 %, respectively).

The above model simulations suggest that under nocturnal atmospheric conditions with a very low NO₃ concentration, the RO₂ radical pool is dominated by ^CIRO₂ and the self-reactions/cross reactions of which are a major contributor to ULVOCs such as the highly oxygenated C₂₀ dimers as observed in boreal forest (Bianchi et al., 2017). When the NO₃ concentration is high, the production of ${}^{{\mathrm{NO}}_{\mathrm{3}}}{\mathrm{RO}}_{\mathrm{2}}$ becomes significant and the cross reactions of which with ^CIRO₂ would suppress the formation of ULVOCs. Although HOM ON dimers are readily produced by cross reactions between ${}^{{\mathrm{NO}}_{\mathrm{3}}}{\mathrm{RO}}_{\mathrm{2}}$ and ^CIRO₂, they generally have higher volatilities than C_xH_yO_z HOM dimers and therefore are less efficient in initiating particle formation. However, these HOM ONs can be an important contributor to the particle mass growth. As suggested by the model simulations in Bates et al. (2022), the NO₃ oxidation of α-pinene led to a particulate nitrate yield of 7 % under nocturnal atmospheric conditions in rural Alabama during the Southern Oxidant and Aerosol Study (SOAS) campaign. Our results offer mechanistic and quantitative insights into how the synergistic oxidation of α-pinene by O₃ and NO₃ radicals can influence the formation of low-volatility organic compounds and hence particle formation and growth. They also provide a potential explanation for field observations regarding NPF events frequently occurring in monoterpene-rich regions during daytime but not at nighttime (Mohr et al., 2017; Kulmala et al., 2001; Junninen et al., 2017).

4 Conclusions

This study provides a comprehensive characterization of the nocturnal synergistic oxidation of α-pinene by O₃ and NO₃ radicals and its influence on the formation of HOMs and low-volatility organic compounds using a combination of flow reactor experiments and detailed kinetic model simulations. It is found that the formation of C_xH_yO_z HOMs in the O₃ + NO₃ regime is significantly suppressed compared to that in the O₃-only regime, mainly due to the depletion of ozonolysis-derived RO₂ (i.e., ^CIRO₂ and ^OHRO₂) by ${}^{{\mathrm{NO}}_{\mathrm{3}}}{\mathrm{RO}}_{\mathrm{2}}$ via cross reactions. In addition, the decreases in the abundance of ^CIRO₂ and related HOMs are significantly larger than those of OH-derived ones, indicating that the ${}^{{\mathrm{NO}}_{\mathrm{3}}}{\mathrm{RO}}_{\mathrm{2}}$ species react more efficiently with ^CIRO₂ than with ^OHRO₂. Detailed measurement–model comparisons for the distribution of a suite of ^CIRO₂, ^OHRO₂, and associated HOMs across different oxidation regimes further reveal that the cross reactions between ^CIRO₂ and ${}^{{\mathrm{NO}}_{\mathrm{3}}}{\mathrm{RO}}_{\mathrm{2}}$ are on average 10–100 times more efficient than those of ^OHRO₂ and ${}^{{\mathrm{NO}}_{\mathrm{3}}}{\mathrm{RO}}_{\mathrm{2}}$.

The suppressed formation of C_xH_yO_z HOMs in the synergistic O₃ + NO₃ regime results in a significant reduction in ULVOCs. Although a substantial number of HOM ONs are formed from the cross reactions between ${}^{{\mathrm{NO}}_{\mathrm{3}}}{\mathrm{RO}}_{\mathrm{2}}$ and ^CIRO₂ or ^OHRO₂ in the synergistic-oxidation regime, they have higher volatilities and are less likely to participate in the formation and initial growth of new particles. As a result, in our experiment the formation of new particles in the synergistic-oxidation regime is substantially inhibited compared to the O₃-only regime. Chemical model simulations further confirm that the synergistic oxidation of α-pinene by O₃ and NO₃ radicals can significantly inhibit the formation of C_xH_yO_z HOMs, especially the ultralow-volatility C_xH_yO_z HOM dimers under typical nighttime atmospheric conditions. Our study sheds lights on the synergistic-oxidation mechanism of biogenic emissions and underscores the importance of considering this chemistry for a better depiction of the formation of low-volatility organics and particles in the atmosphere.