The gas-phase oxidation reactions of n-aldehydes with OH radicals were conducted in a flow reactor setup in the laboratory as shown in Fig. 1. All the experiments were conducted at room temperature and 1 atm pressure of air. The precursor aldehydes were introduced into the reactor from their individual gas cylinders. The oxidant OH radicals were produced in situ by the reaction of tetramethylethylene (TME, C6H12) with ozone. We used an ozone generator that photolyzes zero-air by a mercury lamp (UVP, Analytik Jena) to provide ozone to the reactor while TME was supplied from a gas cylinder. The zero-air was produced by feeding compressed clean air to a zero-air generator (AADCO-737-15) which was also used as bath gas in the reactor maintaining a total sample flow of 8–10 slpm. The initial concentrations of reactant precursors were determined by controlling the individual gas flows using calibrated mass flow controllers (Alicat Scientific). An ozone analyzer (2B Technologies model 205) was used to measure the ozone concentrations. Further details regarding reactant concentration measurements, as well as specifications for the chemicals and gas cylinders, are provided in Sects. S1 and S2 in the Supplement.

We studied the reactions over a range of reaction times (short: 1–3 s, and long: 11–13 s). The experimental conditions are presented in Table 1. Long reaction time experiments were conducted using a borosilicate flow reactor (length: 100 cm, and i.d.: 4.7 cm) while a quartz flow reactor (length: 100 cm, and i.d.: 2.2 cm) was used for the short reaction time experiments. We utilized the full volume of the reactor to achieve a long reaction time. However, the short reaction times were achieved by controlling the distance between the mass spectrometer orifice and the position where the precursor aldehyde meets the oxidant OH inside the reactor. This was done by using a movable injector that brings the aldehyde of interest at variable positions inside the reactor. The shortest possible reaction time of an individual aldehyde was chosen by the detection of any HOMs from its oxidation initiated by OH radicals. In short reaction time experiments, the highest concentration of VOC (6.4 ppmv) was used for pentanal while the concentrations of other aldehydes were up to 1 ppmv. The VOC concentrations of 0.2–2.5 ppmv were used in long reaction time experiments. The other reactants including 43–97 ppbv of TME, and 77–295 ppbv of ozone were maintained nearly constant with respect to individual VOC in all experiment types (see Table 1). Among all the studied n-aldehyde systems, the lowest level of aldehyde and O3 concentrations were maintained for heptanal oxidation experiment. This is because higher concentrations led to irrelevant products likely originating from the ozonolysis of heptanal stabilizers (see Fig. S2). To observe the effect of NOx on the OH induced oxidation of n-aldehydes, we conducted the experiments with variable NO concentrations (2–1000 ppbv). Additionally, one more set of experiments, hydrogen to deuterium (H/D) exchange, were conducted by the addition of D2O flow to the n-aldehyde OH oxidation reaction. These experiments give an estimate of the number of functional groups with labile H atoms (e.g., OH, OOH, and C(O)OOH) in the oxidation products. A near complete H/D conversion was confirmed by monitoring the reagent ion signals of HNO3NO3- and (HNO3)2NO3- fully converting to DNO3NO3- and (DNO3)2NO3-, respectively (see Fig. S14). The time series of reactive species, VOC, OH, RO2, O3, and NO under different reaction conditions are shown in Figs. S16 and S17.

The oxidation products were detected using a nitrate ion time-of-flight chemical ionization mass spectrometer (NO3--ToF-CIMS) as their NO3- adducts. A zero-air sheath flow of 20 slpm was provided to the chemical ionization inlet. The NO3- ions were produced from gas-phase HNO3 flow under soft X-ray exposure while being carried by N2 to the inlet. The mass spectrometric data processing, including averaging, baseline removal, mass axis calibration, and peak integration were done using the tofTools v6.03 package for MATLAB.

We estimated the concentrations of reactive species in the flow reactor during different n-aldehyde oxidation experiments using a kinetic simulator Kinetiscope (version 1.1.1136.x64) (Hinsberg and Houle, 2022). These include the average concentrations of OH radicals and initial RO2 radicals (i.e., CnH2n-1O3, the first peroxy radicals formed in the oxidation process) produced in the experiments. In the simulations, a single-reactor model with constant volume, pressure, and temperature was employed. The temperature was set to 298.15 K. The simulation setting parameters such as total number of particles (1 × 10⁹) and random number seed (12947) were kept constant for consistency, while the maximum simulation time was matched to that of individual experiments. Details of all the simulations are provided in Sect. S13.

In this section, we discuss how early HOMs formed and how they evolved with the progress of reaction time in different n-aldehyde OH oxidation experiments. The hexanal OH oxidation spectra included in Figs. 2 and 3 are reproduced from our previous study (Barua et al., 2023) for comparison with the other aldehydes. Figure 2 shows the results from short reaction time (Δt = 1–3 s) experiments. We observed the formation of O6 and O7 HOMs within 1.0, 1.1, and 2.3 s reaction times in the oxidation experiments of octanal, hexanal, and pentanal, respectively. It is essential to mention that with the increase of number of carbon atoms in the studied aldehydes, the required precursor concentrations for first HOM observation decreased (from 6.4 ppm pentanal, 1 ppm hexanal to 0.72 ppm octanal; corresponding reacted concentrations 1.15, 0.34, and 0.30 ppb, respectively). This shows a clear effect of carbon chain length on the reactivity of linear aldehydes towards HOM formation. Because the oxidation process of n-aldehydes (CnH2nO) with OH is initiated by the abstraction of a H atom (aldehydic or non-aldehydic), the first formed acyl (or alkyl) peroxy radical CnH2n-1O3 contains an odd number of oxygen atoms. If autoxidation outcompetes any other bimolecular reactions (e.g., RO2 + RO2, RO2 + HO2, etc.), the product spectrum will be mostly dominated by odd number of oxygen containing products. In all studied n-aldehyde systems, the intensity of O6 HOM is higher than that of O7 HOM (Figs. 2 and 3). The formation of CnH2n-1O6 peroxy radical indicates that the process certainly involves a bimolecular reaction step. In hexanal OH oxidation, Barua et al. (2023) computationally showed that the formation of C6H11O5 peroxy radical via autoxidation is very fast while the subsequent isomerization reaction leading to C6H11O7 is slower. The same is likely to hold true for other aldehydes and it is expected that the CnH2n-1O5 peroxy radical undergoes a bimolecular reaction converting it to CnH2n-1O4 alkoxy radical, followed by a H-shift, and subsequent O2 addition reactions producing the dominant CnH2n-1O6 HOM (see Figs. S7–S11). As the reaction time increased, we observed the formation of monomeric HOM up to O7 and accretion products up to O10 composition within 2.9 s in hexanal oxidation (see Fig. 2d) with the consumption of ∼ 8.88 × 10⁻⁴ of its initial concentration. In the case of octanal, monomeric O8 HOM and accretion products up to O10 formed within 2.1 s reaction time (see Fig. 2e) with a comparable reacted fraction (∼ 8.71 × 10⁻⁴) of its initial concentration.

In long reaction time (11–13 s) experiments, we observed higher intensities of product signals (see Fig. 3) in comparison to their intensities in the short reaction time experiments – as expected. For pentanal, Fig. 3a shows that HOM accretion products up to O10 formed within 12.8 s reaction time which were not seen in the short reaction time experiment (2.3 s). A lower precursor concentration, 2.5 ppm of pentanal in long reaction time experiment compared to earlier 6.4 ppm in short reaction time, was sufficient to produce the observed HOMs in this case. A close observation of C6–C8 n-aldehyde oxidation spectra (Fig. 3b–d) reveals that HOM accretion products up to O11 formed within 11–13 s reaction time under the experimental conditions. In all n-aldehyde experiments, we also observed the accretion products resulting from different combinations of aldehyde-derived peroxy radicals CnH2n-1O6-8 and TME-derived peroxy radical C3H5O3 (see Fig. S13 for details) which are marked with dark red arrows. Figure 3d implies that the highest oxygenation (C8H15O9) in the monomeric HOM products is associated with octanal, whereas Fig. 3c shows that the most oxygenated products produced from heptanal are C7H12-14O8. In the case of pentanal and hexanal, monomeric HOMs are limited to seven oxygen atoms (see Fig. 3a and b). All in all, we notice a near identical distribution of oxidation products in the experiments with all C5–C8 n-aldehydes. However, the tendency of oxidation gets faster and advances to higher oxygenated products when the carbon chain length increases.

The bar plot (see Fig. 4) compares the yields of major oxidation products from different n-aldehydes. It clearly shows that the yields of higher oxygenated products increase as we move from pentanal to octanal. In the accretion product regime, the yields of O9–O10 products in octanal are lower than that of hexanal which is also reflected in their dimer to monomer ratios with octanal being 8.8 × 10⁻² and hexanal being 2.5 × 10⁻¹. The lower ratio for octanal compared to hexanal is observed despite both precursors producing comparable quantities of initial RO2 radicals (9.12 × 10⁹ and 9.37 × 10⁹ molec.cm-3 from octanal and hexanal, respectively; see Table S1 in the Supplement). This can lie in the variation of RO2 + RO2 reaction rate coefficients forming the accretion products (RO2 + RO2 → ROOR + O2) which is highly dependent on specific RO2 structures (Shallcross et al., 2005; Berndt et al., 2018).

It has been widely acknowledged that the formation of HOM is suppressed in high NOx conditions (Wildt et al., 2014; Praske et al., 2018; McFiggans et al., 2019; Pullinen et al., 2020), thus reducing the SOA yields. In this process, the reduction in SOA yield is largely attributed to the suppression of highly condensable HOM accretion products (RO2 + RO2 → ROOR) (Kirkby et al., 2016; Pullinen et al., 2020). However, other studies have shown that NO can also enhance HOM formation by producing reactive RO radicals (RO2 + NO → RO + NO2) that can propagate autoxidation (Rissanen, 2018; Yan et al., 2020; Wang et al., 2021; Shen et al., 2022; Nie et al., 2023; Barua et al., 2025; Kang et al., 2025). In our different n-aldehyde oxidation experiments in the presence of variable concentrations of NO, we observed the general tendency of dropping of the accretion product signals (C2nH4n-2Oz) and their corresponding yields as expected (see Figs. 5 and 6). Figure 5 represents the mass spectra recorded in the experiments without (in blue) and with the presence of 100 ppb initial NO (in red). At this condition, most of the monomeric products are quenched at different extents alongside the formation of organonitrates while the accretion products are quenched nearly completely. Interestingly, we observed some enhancement in the intensities of O4 products in the case of hexanal and octanal (Fig. 5b and c) under 100 ppb NO. A closer look at Fig. 6a reveals that the yields of closed-shell products C5H8O4-6 somewhat increased under around 1 ppb average NO condition which then started to decrease under higher NO of around 30 ppb and above in pentanal oxidation. In both pentanal and hexanal oxidations, the dominant O6 peroxy radicals (C5H9O6 and C6H11O6, respectively, blue markers) gained higher yields under around 30 ppb of NO (Fig. 6a and b) compared to without NO condition.

On the other hand, in octanal oxidation, the yield of dominant O6 peroxy radical (C8H15O6) increased even under 70 ppb of NO (see Fig. 6c). These observations indicate that the suppressing effect of NO on the yields of HOMs in n-aldehyde oxidation is perturbed with the increase of carbon chain length of the precursor aldehyde. Also, in all the studied n-aldehydes, although highly oxygenated products are suppressed under higher NO concentrations, we see some enhancement in the early oxygenated closed shell products (O4–O5, and even O6 product in pentanal) under relatively lower NO concentrations. It should be noted that the formation of organonitrates with chemical composition CnH2n-1OzNO strongly supports our assignment of reactive peroxy radical intermediates CnH2n-1Oz. Additional insights into the abundance of HOMs under varying NO concentrations are discussed in Sect. S12.

With the addition of D2O in the OH initiated oxidation experiments of n-aldehydes, we observed a shift in individual product signals in the mass spectra equivalent to the number of exchangeable H atoms in the product structures (see Fig. 7). This provides additional insight into the product identities in terms of the total number of OH, OOH, and (or) C(O)OOH groups present in their molecular structures. Figure 7 shows that the oxidation products with same number of oxygen atoms in all the studied n-aldehydes undergo identical mass shifts (H/D exchange) in the presence of D2O. This observation indicates that the autoxidation mechanism derived by Barua et al. (2023) for hexanal OH oxidation (see Fig. S5) is directly applicable to the other linear aldehydes studied here and thus produces similar product structures. The original mechanism is extended to HOMs up to nine oxygen atoms and presented in Fig. S6. The likely formation process of HOM accretion products (C2nH4n-2O9-11) is shown in Fig. S12. The proposed structures of O5 closed-shell products, O6–O8 monomeric HOMs as well as O9-11 HOM accretion products (see Figs. S6–S12) agree with the H/D exchange experiments in terms of their 2–4 units of mass shifts in the respective n-aldehyde mass spectra (see Fig. 7).

Ambient concentration of individual longer chain aldehyde (≥ C5) can vary from sub-ppb to several ppb depending on time and location (Williams et al., 1996; Duan et al., 2008; Li et al., 2018; Ma et al., 2019). A total concentration of C6–C10 n-aldehydes in Monti Cimini Forest in Italy was measured to be 8.8 ppb (Ciccioli et al., 1993). In indoor air, the concentration can be significantly higher, even around 50 ppb (Birmili et al., 2022). Atmospheric lifetime of n-aldehydes due to their reactivity with OH radicals is generally less than 10 h (Albaladejo et al., 2002; Aguirre et al., 2025). Because of their significant photochemical ozone formation potential (Jenkin et al., 2017; Aguirre et al., 2025), they are good candidates for generating photochemical smog in NOx rich polluted urban atmosphere. On the other hand, previous studies have shown that atmospheric oxidation products of longer chain n-aldehydes are direct contributors to the formation of SOA (Chacon-Madrid et al., 2010; Fan et al., 2024). Here, we demonstrated that the studied C5–C8 n-aldehydes can rapidly form HOM via autoxidation initiated by OH radicals, and the length of carbon chain controls the efficiency of the process. Therefore, with the increase of carbon chain length in n-aldehydes, the fast formation of HOMs is expected to take part in the early stages of gas-to-particle formation and growth contributing to atmospheric SOA. Our experiments in the presence of NO showed a general decreasing trend of HOM accretion products with increasing NO but also formed the corresponding highly oxygenated organic nitrates (HOM-ONs). In our study with the n-aldehydes under 1–70 ppb NO conditions, some yield enhancement with oxidation products up to 6 O atoms was also seen. Pullinen et al. (2020) showed that both HOMs and HOM-ONs originated from the same peroxy radicals with more than six O atoms condensed on particles by about 50 % and those with more than eight O atoms condensed by about 100 % to form SOA in monoterpene photooxidation. Moreover, other reports show that HOM-ONs originated from different VOCs can contribute to low volatility products (Barua et al., 2025) and thereby particle growth and aerosol mass loading (Fry et al., 2014; Lee et al., 2016; Huang et al., 2019).