Ethylamine-driven amination of organic particles: mechanistic insights via key intermediates identification

Atmospheric amines critically contribute to secondary aerosols formation via heterogeneous reactions, yet the molecular mechanisms governing heterogeneous amination chemistry of aerosols remain unclear. Here, we utilize an integrated tandem flow-tube system coupled with online ultrahigh-resolution mass spectrometry to elucidate the amination chemistry of ethylamine (EA) with representative organic aerosol components, including C₂₀–C₅₄ secondary ozonides (SOZs), C₁₇–C₂₇ carboxylic acids, and aldehydes. Our experiments provide evidence for the formation of four key intermediates: hydroxyl peroxyamines, amino hydroperoxides, peroxyamines, and amino ethers, which mediate SOZs conversion to hydroxyimines, amides, and imines. Furthermore, dihydroxylamines and hydroxylamines are identified as characteristic intermediates in carboxylic acids and aldehydes amination. Quantitative heterogeneous reactivity measurements (Δγ_eff) reveal that SOZs exhibit a pronounced inverse dependence on carbon chain length, e.g., C₂₁ SOZ (Δγ_eff = 1.0 × 10⁻⁴) > C₄₉ SOZ (Δγ_eff = 5.7 × 10⁻⁶), with consistently lower reactivity than acids and aldehydes, e.g., C₁₇ acid (Δγ_eff = 2.3 × 10⁻⁴). The amination mechanism of SOZs is initiated by EA addition, followed by either hydroxyl peroxyamines-mediated dehydration yielding hydroxyimines and amides, or amino hydroperoxides-driven H₂O₂ elimination forming imines. For carboxylic acids and aldehydes, EA addition leads to dihydroxylamines and hydroxylamines formation, which subsequently dehydrate to produce amides and imines. These findings provide a mechanistic framework for understanding amine-driven aerosol aging processes that affects atmospheric chemistry, air quality, and climate systems.

1 Introduction

Atmospheric aerosols undergo complex chemical transformations that significantly affect human health, environmental quality, and climate systems (Shen et al., 2023; George and Abbatt, 2010). The heterogeneous evolution of organic aerosols, initiated by gaseous amines, drives the formation and growth of nitrogen-containing secondary organic aerosols (SOAs), which are critical components of atmospheric pollution (Na et al., 2007; De Haan et al., 2011; Tian et al., 2024). These transformations are governed by composition-dependent amination mechanisms, with distinct pathways for different organic aerosols, such as carboxylic acids (RCOOHs), aldehydes (RCHOs), and secondary ozonides (SOZs). Quantitative analysis of multiple amination reactions of these particles provides fundamental insights into the chemical evolution process of atmospheric SOAs.

The decomposition of SOZs upon amine exposure is initiated by a nucleophilic attack on the carbon atom of SOZs (Jørgensen and Gross, 2009). However, subsequent reaction pathways remain controversial. Na et al. (2006) demonstrated that the nucleophilic attack of NH₃ on a 3,5-diphenyl-1,2,4-trioxolane (denoted as C₁₄ SOZ), derived from the gas-phase ozonolysis of styrene, induces ring-opening reaction to form a C₁₄ amino hydroperoxide. This crucial intermediate subsequently decompose to yield H₂O₂, benzaldehyde, and phenylmethanimine (C₇H₇N). Consistently, Almatarneh et al. (2020) and Jørgensen and Gross (2009) identified C₂ amino hydroperoxide intermediates from the reactions of NH₃ with a C₂ SOZ, derived from the ozonolysis of ethene. In contrast, Zahardis et al. (2008) observed that the attack of octadecylamine (ODA, C₁₈H₃₉N) on a C₁₈ SOZ, produced from the ozonolysis of oleic acid, generates a C₃₆ hydroxyl peroxyamine intermediate, ultimately forming H₂O, nonanal, and C₂₇ amide. More recently, Qiu et al. (2024) reported that the attack of ethylamine on a C₁₅ SOZ, derived from the ozonolysis of β-caryophyllene, directly open the ring and generates H₂O and a C₁₇ amine.

To our knowledge, these key intermediates (amino hydroperoxide and hydroxyl peroxyamine) have not been experimentally measured in prior studies (Na et al., 2006; Jørgensen and Gross, 2009; Almatarneh et al., 2020; Zahardis et al., 2008), creating uncertainty about their mechanistic roles in controlling the evolution of SOZ upon amine exposure. The measured stable amination products (amides, imines, and amines) additionally exhibit inconsistency across studies (Na et al., 2006; Jørgensen and Gross, 2009; Almatarneh et al., 2020; Zahardis et al., 2008). Furthermore, previous experimental investigations primarily focused on qualitative products identification, lacking both quantitative reaction rates of SOZ and amine, as well as kinetics analyses of product formation. These factors limit the evaluation of the amination chemistry in the atmosphere. To mimic the heterogeneous reactions of SOZs and amine, we generated SOZ particles via the heterogeneous ozonolysis of alkene, their dominant natural formation pathway (Qiu et al., 2024, 2022). Specifically, the ozonolysis of squalene (Sqe) was chosen as model system to generate SOZ particles, building on the demonstration of high SOZ yields (maximum ∼ 21 % total yield) from Sqe ozonolysis (Heine et al., 2017). Meanwhile, this Sqe ozonolysis system produces some carbonyl byproducts (e.g., aldehydes and carboxylic acids) enabling simultaneous quantification of carbonyl aerosols upon amine exposure.

It is widely established that reactions between carboxylic acids and amines typically proceed via acid-base neutralization to form ammonium salts (Na et al., 2007; Liu et al., 2012; Smith et al., 2010; Gao et al., 2018). However, Ditto et al. (2022) demonstrated experimentally that the heterogeneous reaction of oleic acid (C₁₈H₃₄O₂) and NH₃ yields an oleamide (C₁₈H₃₅ON) and H₂O. This observation aligns with the calculated pathways by Charville et al. (2011), who suggested that the reaction between carboxylic acid and amine generates a dihydroxyamine intermediate that subsequently dehydrates to form an amide. Moreover, significant uncertainties persist regarding the heterogeneous reaction rates of such carboxylic acid and amine reactions (Fairhurst et al., 2017a, b; Liu et al., 2012). Fairhurst et al. (2017a) reported the heterogeneous reaction rates (uptake coefficient, γ) ranging from 10⁻¹ (malonic acid) to 10⁻⁵ (adipic acid) upon ethylamine (C₂H₇N) exposure. They further revealed that the uptake of amines onto low-molecular-weight diacids (C₃–C₈) is structure-dependent, with higher γ values observed for odd-carbon diacids than even-carbon ones. Additionally, γ decreases with increasing carbon chain length of diacids. However, these trends remain unestablished for long-chain acids. Furthermore, to our knowledge, the heterogeneous uptake coefficients for aldehyde particles upon amine exposure have not been experimentally measured.

Our objective is to investigate the heterogeneous reactions of particulate SOZs, carboxylic acids, and aldehydes upon exposure to gaseous ethylamine (selected as model amine), using a tandem flowtube reactor. As a representative atmospheric amine (Lee and Wexler, 2013), ethylamine was selected for its remarkable heterogeneous reactivity and simple structure. The atmospheric pressure photoionization high-resolution mass spectrometer (APPI-HRMS) is used to identify reaction products and measure reaction kinetics as a function of ethylamine exposure. Additionally, the heterogeneous uptake coefficients for SOZs, aldehydes, and acids are quantified. To interpret the experimental data, the multiphase reaction mechanisms governing the decomposition of SOZs, carboxylic acids, and aldehydes, as well as the formation of featured amination products are revealed.

2 Experimental methods

An integrated experimental system employing a tandem flowtube reactor coupled with APPI-HRMS was developed to examine the heterogeneous reactions between target organic aerosols upon ethylamine exposure, as illustrated in Fig. 1a. The apparatus contains three key components: (i) in-situ generation of organic particles in the first flowtube reactor from the ozonolysis of Sqe aerosols, (ii) controlled multiphase reactions between organic particles and ethylamine in the secondary flowtube reactor, and (iii) online monitoring of chemical compositions of organic particles as a function of ethylamine exposure. To isolate the contribution of heterogeneous reactions in the secondary flowtube reactor (Fig. 1a), the controlled experiments were conducted using a single flowtube configuration, as illustrated in Fig. 1b.

Figure 1 Schematic diagrams of (a) the tandem flowtube system, and (b) the controlled one flowtube experiment.

2.1 Generation of organic particles in the first flowtube reactor

Polydisperse Sqe aerosols were generated via homogeneous nucleation by passing N₂ (300 mL min⁻¹) through a Pyrex tube filled with liquid Sqe (Sigma-Aldrich, 99 % purity), located in a tube furnace setting at 145 °C (Fig. 1). Upon exiting the Pyrex tube, the Sqe vapor cooled and homogeneously nucleated to form aerosols, which were subsequently passed through an annular activated charcoal denuder to remove any residual gas-phase organics produced in the oven. The average Sqe particle distribution was log-normal with a mass concentration of 6120 µg m⁻³ and an average diameter of 209 nm (Fig. S1 in the Supplement).

The Sqe aerosol flow was then introduced into the first quartz flowtube reactor (130 cm long and 2.6 cm inner diameter) where they reacted with O₃ to generate organic particles, mainly composed of SOZs, aldehydes, and carboxylic acids (Heine et al., 2017). O₃ was generated by passing 20 mL min⁻¹ O₂ through a corona discharge generator (com-ad-02, Anseros, China), with its concentration monitored using an O₃ analyzer (GM-6000-OEM, Anseros, China). Dilute dry N₂ (580 mL min⁻¹) was also introduced into the first flowtube reactor to achieve a total flow rate of 900 mL min⁻¹, corresponding to an average residence time of 46 s. The O₃ concentration in the first flowtube reactor was varied from 0 to 0.678 ppm.

The chemical compositions of organic aerosols generated in the first flowtube reactor were analyzed using APPI-HRMS (Liu et al., 2024). As illustrated in Figs. S2a and S3a, the major components were identified as SOZs, carboxylic acids, and aldehydes, consistent with the product distributions measured using vacuum ultraviolet aerosol mass spectrometer (VUV-AMS) (Heine et al., 2017; Arata et al., 2019). Figure S4 displays the mass signals of representative compounds, including C₂₇ aldehyde (C₂₇H₄₄O), C₂₂ acid (C₂₂H₃₆O₂), and C₃₅ SOZ (C₃₅H₅₈O₃), as a function of O₃ concentration in the first flowtube reactor. It is demonstrated that, at fixed O₃ concentration (e.g., 0.57 ppm), the mass signals of these compounds are stable over 30 min-operation periods with little signal fluctuation. The size distribution of organic particles was monitored using a Scanning Mobility Particle Sizer (SMPS, TSI 3080L DMA and 3776 CPC), revealing an average diameter of 201 nm (Fig. S1).

2.2 Reactions of organic particles with ethylamine in the secondary flowtube reactor

Upon exiting the first flowtube reactor and subsequent O₃ denuder, the organic particles were introduced into a secondary quartz flowtube reactor (130 cm long and 2.6 cm inner diameter) to investigate their heterogeneous reactions with ethylamine (Fig. 1a). This coupled reactor configuration, combining organic aerosols generated in the first flowtube reactor with amine exposure in the secondary flowtube, is designated as the tandem 2FT experimental system. Ethylamine was supplied from a standard gas cylinder (1000 ppm ethylamine balanced with nitrogen; Wetry, Shanghai, China). A mixture of organic particle flow, ethylamine, and diluent N₂ was introduced into the secondary flowtube reactor, maintaining a total flow rate of 1100 mL min⁻¹ (corresponding to an average residence time of 37 s). The ethylamine concentration in the secondary flowtube reactor was varied from 0 to 43.45 ppm. All experiments were conducted at atmospheric pressure and room temperature. The experiments have been conducted under dry condition, corresponding to a relative humidity of approximately 3 % (Heine et al., 2017).

2.3 Controlled experiments in one flowtube reactor

As widely demonstrated (Bos et al., 2006; Fredenhagen and Kuhnol, 2014), the reaction kinetics measured using the APPI technique could be influenced by potential photochemical side reactions (Fig. S5). To eliminate contributions from interactions between ethylamine and organic aerosols in the APPI region, control experiments were designed (Fig. 1b). In these control experiments, organic aerosols generated in the first flowtube reactor were introduced directly into the APPI region, bypassing the secondary flowtube reactor. By subtracting the reaction kinetics of the 1FT controlled experiments from those obtained in the tandem 2FT experiments, the net contribution of heterogeneous reactions occurring in the secondary flowtube reactor was quantitatively determined (Fig. S6).

2.4 Real-time detection system and data analysis for heterogeneous reactions

A portion of the particle stream was sampled by the SMPS to measure particle size distribution and concentration. The remaining flow (800 mL min⁻¹) was directed into the ionization region of the APPI-HRMS (Orbitrap Fusion, Thermo Scientific) for real-time chemical characterization (Fig. S5). Additional details on the application of APPI-HRMS for quantifying heterogeneous reactions of particles are available in our previous work (Liu et al., 2024).

By monitoring the mass signals of organic particles (denoted as [Particle]) as a function of ethylamine exposure, defined as the concentration of ethylamine ([ethylamine]) × residence time (t), the heterogeneous decay rate (k_particle) was determined through fitting the decay profiles to an exponential function (Eq. 1) (Smith et al., 2009; Liu et al., 2024). The effective uptake coefficient (γ_eff), representing the probability of reactive particle decay upon ethylamine collisions, was then calculated using Eq. (2) (Liu et al., 2024; Smith et al., 2009). In Eq. (2), D, ρ₀, N_A, $\stackrel{\mathrm{‾}}{c}$, and M correspond to particle diameter, density, Avogadro's number, mean speed of ethylamine, and molar mass of reactant molecules, respectively. In this work, the heterogeneous reaction rates measured in the 2FT flowtube reactor experiments and single flowtube reactor (1FT) experiments were designated as γ_eff, 2FT and γ_eff, 1FT, respectively. The net contribution of heterogeneous reactions in the secondary flowtube reactor was quantified by subtracting γ_eff, 1FT from γ_eff, 2FT, yielding the differential uptake coefficient (Δγ_eff) as defined in Eq. (3). Figure S6 presents the decay kinetics and corresponding Δγ_eff values for representative compounds: C₂₇ aldehyde, C₂₇ acid, and C₂₀ SOZ. For instance, γ_eff, 2FT and γ_eff, 1FT values for C₂₀ SOZ were determined to be 8.0 × 10⁻⁵ and 6.2 × 10⁻⁵, respectively, resulting in Δγ_eff = 1.8 × 10⁻⁵.

$$\begin{array}{}\text{(1)}& {\displaystyle \frac{\left[\mathrm{Particle}\right]}{[\mathrm{Particle}{]}_{\mathrm{0}}}}=\exp \left(-k_{\mathrm{particle}}\times \left[\mathrm{Ethylamine}\right]\times t\right)\text{(2)}& {\displaystyle }{\mathit{\gamma }}_{\mathrm{eff}}={\displaystyle \frac{\mathrm{4}\times k_{\mathrm{particle}}\times D\times {\mathit{\rho }}_{\mathrm{0}}\times N_{\mathrm{A}}}{\mathrm{6}\times \stackrel{\mathrm{‾}}{c}\times M}}\text{(3)}& {\displaystyle }\mathrm{\Delta }{\mathit{\gamma }}_{\mathrm{eff}}={\mathit{\gamma }}_{\mathrm{eff},\mathrm{2}\mathrm{FT}}-{\mathit{\gamma }}_{\mathrm{eff},\mathrm{1}\mathrm{FT}}\end{array}$$

3 Results and discussion

3.1 Heterogeneous reactions rates of organic aerosols upon ethylamine exposure

The O₃ addition reactions to the C=C bonds of Sqe in the first flowtube reactor generate primary ozonides (POZs) (Heine et al., 2017), as illustrated in Fig. S8. These POZs subsequently decompose to form three ketones (C₃H₆O, C₈H₁₄O, and C₁₃H₂₂O) with molecular weights (MWs) of 58, 126, and 194, three aldehydes (C₁₇H₂₈O, C₂₂H₃₆O, and C₂₇H₄₄O) with MWs of 248, 316, and 384, and six Criegee intermediates (CIs) (Heine et al., 2017). Unimolecular isomerization reactions of C₁₇, C₂₂, and C₂₇ CIs produce carboxylic acids (C₁₇H₂₈O₂, C₂₂H₃₆O₂, and C₂₇H₄₄O₂) with MWs of 264, 332, and 400 (Arata et al., 2019; Zahardis et al., 2005). Bimolecular reactions between CIs and aldehydes (or ketones) generate a series of SOZs (Fig. S9), including C₆, C₁₁, C₁₆, C₂₀, C₂₁, C₂₅, C₂₆, C₃₀, C₃₄, C₃₅, C₃₉, C₄₀, C₄₄, C₄₉, and C₅₄ SOZs (Heine et al., 2017). Considering the relative higher partitioning of smaller species (e.g., ketones and smaller SOZs) into the gas phase, the present work mostly focused on larger SOZs, aldehydes, and carboxylic acids.

Figure 2 (a–b) Decay of SOZs, aldehydes, and carboxylic acids as a function of ethylamine (EA) exposure in tandem flowtube experiments. (c–d) Differential effective uptake coefficients (Δγ_eff) for C₂₀, C₂₁, C₂₅, C₂₆, C₃₀, C₃₄, C₃₅, C₃₉, C₄₀, C₄₄, C₄₉, and C₅₄ SOZs; C₁₇, C₂₂, and C₂₇ carboxylic acids, and C₁₇, C₂₂, and C₂₇ aldehydes. Uptake coefficients for carboxylic acids from Refs (Liu et al., 2012; Fairhurst et al., 2017a) are included.

Figure 2a illustrates the relative abundance of C₂₀ to C₅₄ SOZs as a function of ethylamine exposure. These SOZs exhibit distinct heterogeneous decay rates (k_particle), with the C₂₁ SOZ exhibiting the highest rate. The differential effective uptake coefficients (Δγ_eff), quantifying the contribution of SOZ reactions with ethylamine in the secondary flowtube reactor, were then calculated using Eqs. (2) and (3) (Fig. 2c and Table S2). The Δγ_eff values of SOZs generally show consistent tendencies with the decay kinetics and exhibit a zigzag pattern that decreases with increasing carbon chain length of SOZs.

Differences in the initial concentrations of SOZs may contribute to their distinct heterogeneous reactivities (Δγ_eff) shown in Fig. 2c. As demonstrated by Heine et al. (2017), in multi-component particles, the heterogenous reactivity of each component depends on its initial concentration (Jacobs et al., 2016; Zeng et al., 2020; Arata et al., 2019). Heine et al. (2017) reported that the abundance of SOZs formed during the ozonolysis of squalene varies, following the order: C₃₀ > C₂₅, C₃₅ > C₄₄ > C₂₀, C₂₁, C₃₉, C₄₀, C₄₉ > C₂₆, C₃₄, C₅₄ SOZs, as illustrated in Fig. S7a. Consequently, in this work, the initial concentrations of SOZs formed from the ozonolysis of Sqe in the flowtube reactor differ. To quantify the influence of the initial SOZ concentrations on their heterogeneous reaction rates, the Δγ_eff values for SOZs were normalized for their corresponding initial concentrations. As illustrated in Fig. S7b, the normalized Δγ_eff values exhibit smaller differences compared to the original Δγ_eff values. This observation supports the hypothesis that differences in initial SOZ concentrations affect their decay rate upon ethylamine exposure. Additionally, SOZs with long-chain substituents (e.g., C₅₄ SOZ) exhibit lower reactivity (Ponec et al., 1997). This reduced reactivity may be attributed to the steric hindrance effects (Hon et al., 1995), which restrict the conformational flexibility of SOZ molecules during attack by ethylamine.

Figure 2b illustrates the decay kinetics of representative aldehydes and carboxylic acids. These aldehydes and carboxylic acids decay faster than SOZs. Consistently, their differential effective uptake coefficients (Δγ_eff from 10⁻⁵ to 10⁻⁴) are larger than those of SOZs (10⁻⁵ to 10⁻⁶), as illustrated in Fig. 2d. This difference could be explained by the higher acidity of carboxylic acids, which enhances the heterogeneous reactions between these acids upon ethylamine exposure. As reported by Liu et al. (2012) the heterogeneous uptake coefficients of dimethylamine (C₂H₇N, an isomer of ethylamine), with citric acid (a triacid, C₆H₈O₇, γ ∼ 10⁻³) is significantly larger than with humic acid (a diacid, C₉H₉NO₆, γ ∼ 10⁻⁶). They attributed this difference to the stronger acidity of citric acid relative to humic acid.

Figure 2d illustrates that the measured Δγ_eff values for carboxylic acids follow the trend: C₁₇H₂₈O₂ > C₂₂H₃₆O₂ > C₂₇H₄₄O₂, indicating a negative dependence of heterogeneous reactivity on carbon chain length. A similar trend is observed for aldehydes, i.e., C₁₇H₂₈O > C₂₂H₃₆O > C₂₇H₄₄O. To our knowledge, no prior experimental measurements exist for the heterogeneous reactivity of aldehyde particles with ethylamine, whereas reactivity trends for carboxylic acids of varying carbon chain length have been investigated (Fairhurst et al., 2017a). For example, Fairhurst et al. (2017a) measured the heterogeneous reactivities of ethylamine by solid dicarboxylic acids with varying carbon numbers: malonic acid (C₃H₄O₄), succinic acid (C₄H₆O₄), glutaric acid (C₅H₈O₄), adipic acid (C₆H₁₀O₄), and pimelic acid (C₇H₁₂O₄). Their measured uptake coefficients are approximately 10⁻¹ for C₃ diacids, 7 × 10⁻⁵ for C₄ diacid, and 1 × 10⁻⁵ for the C₆ diacid. They attributed this trend to differences in the crystalline surface structures of these solid acids. Thus, these reported uptake coefficients for carboxylic acids reacting with amine span a wide range (10⁻¹ to 10⁻⁶), suggesting that more comprehensive data are needed to elucidate the reactivity trends for both carboxylic acids and aldehydes across varying carbon chain length.

3.2 Products distribution during the heterogeneous reactions of organic particles

Figure 3a illustrates the product distribution during the ozonolysis of Sqe aerosols. Consistent with prior observations (Liu et al., 2024), maximum product yields were achieved at approximately 60 % Sqe conversion (corresponding to 0.678 ppm O₃), with representative products (SOZs, aldehydes, and acids) shown in Figs. S2 and S3. Figures 3b and S3 show representative products from heterogeneous reactions between SOZs with ethylamine (C₂H₇N), with mass spectral analysis revealing three product classes. First, protonated molecular ions (denoted as [M + H]⁺) appear at $m/z$ 314, 368, 382, 436, 450, 504, 558, 572, 626, 640, 694, and 762, corresponding to adducts from reactions between SOZs and ethylamine, i.e., SOZs + C₂H₇N. For instance, the $m/z$ 504 peak corresponds to C₃₂H₅₇O₃N (MW 503) from the reaction of C₃₀ SOZ (C₃₀H₅₀O₃) with C₂H₇N. Detailed reaction mechanisms are discussed in Sect. 3.3. Second, the deprotonated ions ([M − H]⁻) at $m/z$ 294, 348, 362, 416, 430, and 484 derive from subsequent dehydration products from SOZs + C₂H₇N reactions, i.e., SOZs + C₂H₇N − H₂O. For instance, the $m/z$ 484 peak matches the H₂O loss product of C₃₂H₅₇O₃N, i.e., C₃₂H₅₇O₃N − H₂O. Third, the deprotonated ions ([M − H]⁻) at $m/z$ 278, 332, 346, 400, 414, 468, and 536 correspond to the H₂O₂ elimination products from SOZs and C₂H₇N reactions, i.e., SOZs + C₂H₇N − H₂O₂. For instance, the $m/z$ 468 peak represents the H₂O₂ elimination product of C₃₂H₅₇O₃N, i.e., C₃₂H₅₇O₃N − H₂O₂. Notably, dehydration and H₂O₂ elimination products were only observed for smaller SOZs (C_n < 35), suggesting that substituent effects and chain length of SOZs could influence the amination reaction pathways (see Sect. 3.3 for mechanistic details).

Figure 3 Mass spectra of (a) representative SOZs (the signals within $m/z$ range of 500–800 are amplified) formed during the ozonolysis of Sqe in the first flowtube reactor, and (b) their amination products (adducts of SOZ + EA are in red and dehydration products are in blue) upon exposure to ethylamine (EA) in the secondary flowtube reactor.

Figures S2b and S3b illustrate the products formed from heterogeneous reactions of C₁₇, C₂₂, and C₂₇ aldehydes (RCHOs), as well as C₁₇, C₂₂, and C₂₇ carboxylic acids (RCOOHs) upon ethylamine exposure in the secondary flowtube reactor. The protonated molecular ions ([M + H]⁺) at $m/z$ 294, 310, 362, 378, 430, and 446 correspond to products from reactions of these aldehydes (or carboxylic acids) with ethylamine (C₂H₇N), i.e., RCHOs (or RCOOHs) + C₂H₇N (Bain et al., 2016; De Haan et al., 2011). For instance, the peak at $m/z$ 294 is assigned to C₁₉H₃₅ON (MW 293) formed from the reaction of C₁₇ aldehyde (C₁₇H₂₈O) with C₂H₇N. These RCHOs (or RCOOHs) + C₂H₇N adducts subsequently undergo H₂O elimination reactions (Shen et al., 2023; Bain et al., 2016; De Haan et al., 2009; Tuguldurova et al., 2024), producing characteristic peaks at $m/z$ 276, 344, and 412 (from aldehyde reactions); as well as $m/z$ 292, 360, and 428 (from acid reactions). For instance, the [M + H]⁺ peak at $m/z$ 276 corresponds C₁₉H₃₃N (MW 275), resulting from H₂O elimination from C₁₉H₃₅ON, i.e., C₁₉H₃₅ON − H₂O.

3.3 Amination mechanisms of secondary ozonides, carboxylic acids, and aldehydes

A reaction mechanism was developed to elucidate the heterogeneous reactions of organic aerosols, including SOZs, carboxylic acids (C₁₇H₂₈O₂, C₂₂H₃₆O₂, and C₂₇H₄₄O₂), and aldehydes (C₁₇H₂₈O, C₂₂H₃₆O, and C₂₇H₄₄O).

Figure 4 Amination mechanisms of secondary ozonides (SOZs), carboxylic acids (RCOOHs), and aldehydes (RCHOs) upon ethylamine (EA) exposure, with representative chemical structures and compositions of reactants and products.

For SOZs, the electronegativity of neighboring oxygen atoms induces a net positive charge on the α-carbon atoms (Fig. 4), facilitating nucleophilic attack by an amine (Jørgensen and Gross, 2009). This could lead to the formation of either hydroxyl peroxyamines (R1) or amino hydroperoxides (R2), depending on the nucleophilic attack sites (Zahardis et al., 2008; Jørgensen and Gross, 2009; Almatarneh et al., 2020). However, these competing pathways and their proposed products remain controversial (Fig. S11). Zahardis et al. (2008) proposed that SOZ reaction with octadecylamine generates a hydroxyl peroxyamine intermediate, which subsequently decomposes to nonanal and a C₂₇ amide via H₂O elimination (Fig. S11a). Conversely, Almatarneh et al. (2020), Jørgensen and Gross (2009), and Na et al. (2006) suggested that SOZ reactions with ammonia form amino hydroperoxide intermediates (Fig. S11b to e). Notably, neither hydroxyl peroxyamine nor amino hydroperoxide intermediates have been experimentally detected in these studies (Zahardis et al., 2008; Almatarneh et al., 2020; Jørgensen and Gross, 2009; Na et al., 2006). Contrasting both pathways, Qiu et al. (2024) demonstrated that ethylamine addition to a cyclic SOZ directly yields a linear amination product through simultaneously H₂O elimination (Fig. S11a), i.e., bypassing formation of either proposed intermediates (Fig. S11f).

Figure 5 (a) Structures of two representative isomers (C₃₂H₅₇O₃N, MW 503), i.e., hydroxyl peroxyamine (I) and amino hydroperoxide (II). (b–d) MS² fragmentation of their protonated ions at HCD energies: (b) 0 %, (c) 10 %, and (d) 20 %.

In this work, mass peaks corresponding to SOZs + ethylamine (C₂H₇N) adducts were observed (Fig. 3b), providing direct experimental evidence for nucleophilic attack by ethylamine on SOZs. The MS² spectra provide complementary evidence for their structural characterization. Figure 5 illustrates representative MS² fragmentation patterns of the [C₃₂H₅₈O₃N]⁺ ion (denoted as [M + H]⁺), corresponding to the C₃₂H₅₇O₃N product formed from C₃₀ SOZ (C₃₀H₅₀O₃) + ethylamine (C₂H₇N) reaction. Two representative isomers, a hydroxyl peroxyamine (I) and amino hydroperoxide (II), were selected for analysis (Fig. 5a). Both isomers undergo C₂H₅NH elimination, yielding [M − C₂H₆N]⁻ ions ($m/z$ 459). The hydroxyl peroxyamine isomer (I) subsequently loses C₃H₇O and C₁₁H₁₈ to form $m/z$ 294, while the amino hydroperoxide isomer (II) eliminates HO₂ yielding $m/z$ 426. Additional fragmentation peaks ($m/z$ 60, 86, 149, 383, and 441) may derive from these isomers. It is also noted that C₃₀ SOZ isomers formation during the ozonolysis of Sqe (Figs. S9 and S10b) suggests additional C₃₂H₅₇O₃N isomers likely contribute to these fragmentation patterns (Fig. 5). Figure 6a illustrates the kinetics of representative C₃₀ hydroxyl peroxyamines (or amino hydroperoxides) as a function of ethylamine exposure.

Figure 6 Experimental signal as a function of ethylamine (EA) exposure for: (a) hydroxyl peroxyamine (or amino hydroperoxide), (b) peroxyamine, (c) amino ether, (d) dihydroxylamine and hydroxylamine, (e) amide, and (f) imine.

Consumption reactions for the amino hydroperoxide and hydroxyl peroxyamine intermediates were previously proposed (Zahardis et al., 2008; Na et al., 2006; Almatarneh et al., 2020; Jørgensen and Gross, 2009). Amino hydroperoxide decomposition has been demonstrated to generate smaller imines and aldehydes via H₂O₂ elimination (Fig. 4). For instance, Almatarneh et al. (2020) demonstrated that a C₂ amino hydroperoxide decomposition yields methylenimine, formaldehyde, and H₂O₂ (Fig. S11). Consistently, products with compositions of C₅H₁₁N, C₁₀H₁₉N, C₁₅H₂₇N, C₁₉H₃₃N, C₂₄H₄₁N, and C₂₉H₄₉N are assigned to be imines (Fig. S13) that could be attributed to these reactions. In contrast, for hydroxyl peroxyamine, Zahardis et al. (2008) reported that hydroxyl peroxyamine decomposition generates a C₂₇ amide, C₉ aldehyde with H₂O elimination. Observed products with compositions of C₁₉H₃₃ON, C₂₄H₄₁ON, and C₂₉H₄₉ON are proposed to be amides (Fig. S12) that could be formed from these pathways. Moreover, hydroxyl peroxyamines lacking α-H atoms form hydroxyimines (C₅H₁₁ON, C₁₀H₁₉ON, and C₁₅H₂₇ON), rather than amides. For instance, a C₃₆ hydroxyl peroxyamine with available α-H atom (Fig. S12a) decomposes to a C₁₉ amide, C₁₇ aldehyde and H₂O, whereas a C₂₈ hydroxyl peroxyamine without α-H atom (Fig. S12b) yields C₁₅ hydroxyimine, C₁₃ ketone, and H₂O.

These hydroxyimines, amides, and imines (Fig. 4) exhibit significantly lower carbon numbers than their precursor hydroxy peroxyamines (R3) or amino hydroperoxides (R4). Consequently, neither R3 nor R4 pathways explain observed products retaining same carbon numbers with amino hydroperoxides or hydroxy peroxyamines. This work therefore proposes two new pathways: R5 (hydroxyl peroxyamine → H₂O + peroxyamine), and R6 (amino hydroperoxide → H₂O₂ + amino ether). With C₃₂ intermediates for example, these pathways reasonably explain observed C₃₂H₅₅O₂N (peroxyamine) and C₃₂H₅₅ON (amino ether) products formed from C₃₂H₅₇O₃N via H₂O or H₂O₂ elimination, respectively (Fig. 6b and c). The observed trend of their experimental signal, which initially increased and then decreased as a function of ethylamine exposure, providing evidence for mediating the conversion of hydroxy peroxyamine (or amino hydroperoxide) to hydroxylamine, amide, and imine. Given the absence of these pathways in prior work (Almatarneh et al., 2020; Na et al., 2006; Zahardis et al., 2008; Qiu et al., 2024; Jørgensen and Gross, 2009), these R5 and R6 require further investigations. Figure S15 summarizes a simplified mechanism involving four key intermediates, including hydroxyl peroxyamines, peroxyamines, amino hydroperoxides, and amino ethers.

Figure 4 illustrates reaction mechanisms for ethylamine reactions with C₁₇, C₂₂, and C₂₇ carboxylic acids, as well as C₁₇, C₂₂, and C₂₇ aldehydes, forming dihydroxylamines (R7) and hydroxylamines (R8), respectively (De Haan et al., 2011; Ditto et al., 2022; Bain et al., 2016; Shashikala et al., 2023; Sarkar et al., 2019). For instance, the nucleophilic attack reaction of ammonia at carbonyl (C=O) site of acetaldehyde generates a hydroxylamine as demonstrated by Sarkar et al. (2019). Here, dihydroxylamines (C₁₉H₃₅O₂N, C₂₄H₄₃O₂N, C₂₉H₅₁O₂N) and hydroxylamines (C₁₉H₃₅ON, C₂₄H₄₃ON, and C₂₉H₅₁ON) have been measured by APPI-HRMS (Fig. S2b) (Sarkar et al., 2019), as well as their kinetics as a function of ethylamine exposure (e.g., C₁₉ dihydroxylamine and hydroxylamine in Fig. 6d). Subsequent H₂O elimination reactions of dihydroxylamines and hydroxylamines yield amides (R9, C₁₉H₃₃ON, C₂₄H₄₁ON, and C₂₉H₄₉ON) and imines (R10, e.g., C₁₉H₃₃N, C₂₄H₄₁N, and C₂₉H₄₉N) (Montgomery and Day, 1965), respectively, as shown in Fig. S2b. For example, C₁₇ aldehyde (C₁₇H₂₈O, Fig. S14a) reacts with ethylamine to generate C₁₉ hydroxylamine intermediate (C₁₉H₃₅ON, MW 293), which dehydrates to C₁₉ imine (C₁₉H₃₃N, MW 275). Additionally, these amides and imines can be also produced from hydroxyl peroxyamines (R3) or amino hydroperoxides (R4), as shown in Fig. 6e and f.

4 Conclusions

Atmospheric amines critically modulate the evolution of aerosols and particulate pollution. Here, we investigate heterogeneous reactions of ethylamine with SOZs, carboxylic acids, and aldehydes aerosols using a tandem flowtube reactor combined with online APPI-HRMS. Heterogeneous reactivities (Δγ_eff) for SOZs decrease 17.5-fold with increasing carbon chain length, from Δγ_eff = 1.0 × 10⁻⁴ for C₂₁ SOZ to 5.7 × 10⁻⁶ for C₄₉ SOZ, with nonmonotonic behavior suggesting substitution effects. Crucially, reactions of ethylamine with carboxylic acids and aldehydes exhibit Δγ_eff values of 10⁻⁴ to 10⁻⁵, exceeding SOZs reactivity at equivalent carbon numbers, with reactivity similarly declining with chain length. This reactivity dependence implies atmospheric lifetimes ($\mathit{\tau }\propto {\mathit{\gamma }}^{-\mathrm{1}}$) spanning two orders of magnitude across these organic aerosols, thereby controlling their differential impacts on air quality, health, and climate. Moreover, hydroxyl peroxyamines, amino hydroperoxides, peroxyamines, and amino ethers, were measured as characteristic intermediates linking SOZ consumption to stable nitrogenous products (hydroxyimines, amides, and imines). Dihydroxylamines and hydroxylamines from reactions of carboxylic acids and aldehydes were characterized as crucial intermediates. Further reaction mechanism analysis reveals that nucleophilic addition of ethylamine to SOZs initiates the formation of hydroxyl peroxyamines and amino hydroperoxides. Beyond established cleavage pathways yielding form hydroxyimines, amides, and imines, this work demonstrates new elimination pathways: hydroxyl peroxyamines (amino hydroperoxides) → H₂O (or H₂O₂) + peroxyamines (or amino ethers). These mechanistic insights elucidate the transformation of organic aerosols to nitrogen-containing secondary organic aerosols (SOAs), providing fundamental parameters for more accurate modeling of atmospheric processes.