Atmospheric aerosols significantly affect visibility, air quality and climate1. Exposure to atmospheric aerosols has been associated with increased cardiopulmonary mortality and morbidity2. Atmospheric organic aerosols (OAs) may originate from natural and anthropogenic sources and account for 20–50% of aerosol loading globally3,4. The composition of OAs may be complex, comprising of thousands of compounds from simple hydrocarbons to highly oxidized compounds. Directly emitted hydrocarbon components are referred to as primary organic aerosols (POAs) and OA formed via the atmospheric oxidation of volatile and semi-volatile organic compounds (VOCs) are referred to as secondary organic aerosol (SOA).

The formation of SOAs has recently received much attention as they have been shown to be a major component of atmospheric OAs5. For example, 50% of fine particle mass is composed of OAs in the urban locations within the Northern Hemisphere. The SOA fraction ranged from 65 to 95% between urban and remote regions6. The precursors of SOAs include VOCs emitted from both biogenic (e.g., terrestrial vegetation, grassland, peatlands and forest) and anthropogenic sources (e.g., biomass burning, coal combustion, transportation, solvent utilization and industry). Under atmospheric conditions, VOCs may undergo atmospheric oxidation with ozone (O3), nitrate (NO3) and hydroxyl radicals (OH) to form less volatile products which may undergo further reactions and/or partition into the condensed phase, leading to complex chemical composition profiles within the aerosol.

Atmospheric models indicate that annual SOA fluxes are largely dominated by biogenic SOA (BSOA, 88 TgC/yr), followed by biomass burning SOA (BBSOA, 17 TgC/yr), and anthropogenic SOA (ASOA, 10 TgC/yr)7. This highlights the dominance of biogenic VOCs over other SOA precursors, where isoprene alone contributes ~50% to the global VOCs flux8. ASOA precursors are mostly composed of alkanes (40%), followed by aromatics (20%) and alkenes (10%) with the rest being oxygenated and unidentified compounds9. In urban environments, light aromatic hydrocarbons contribute up to 30% of VOCs10 and 3–25 TgC/yr to global SOA production, up to 2–3 fold higher than previous model estimates11. Ambient measurements have also shown the importance of ASOA in urban environments as they were found to be contributing significantly to the OA mass and can be substantially higher in concentration than BSOA12,13.

Atmospheric models use parameterizations of SOA yield curves obtained from laboratory experiments to predict SOA mass concentrations. However, SOA yield data are not available for all ASOA precursors due to a limited number of laboratory studies, with some ASOA precursors and processes (e.g., cloud processing or other aerosol-phase reactions) that contribute to SOA not being considered14. OAs need to be well characterised within atmospheric models to develop effective air quality management plans, that mitigate adverse health and climate impacts of atmospheric aerosol loading and exposure. There are several potential causes for the widespread negative bias in model predictions of OAs. This review focuses on SOA formation from ASOA precursors, which historically has not been investigated extensively.

The objectives of this work are to summarize the outcomes of existing knowledge on ASOA precursors and their oxidation products from atmospheric simulation chamber studies. We also compare SOA yields from different precursors and the factors controlling the SOA mass yield. Since this is a very large subject, the coverage does not extend to natural precursors such as isoprene and terpenes, or to reaction processes in the aqueous phase. The latter have been reviewed by Blando and Turpin15, Carlton et al.16, Ervens et al.17 and Mahilang et al.18 Finally, gaps in knowledge between SOA contribution from lab, field and modelling studies are discussed.

Laboratory experiments

SOA-forming potentials of precursors are generally estimated through atmospheric simulation chamber experiments to derive a yield. A brief description of their design is presented hereafter.

Smog chamber/ potential aerosol mass (PAM) designs

Smog (environmental) chamber designs

Predominantly, chambers are several cubic metres in volume and made from Teflon film and are often equipped with a standard set of instruments, including an integrated temperature and humidity probe and a differential pressure indicator19,20,21,22. In addition, chambers are often equipped with sets of narrow-band blacklights and/or UV broadband lamps for irradiation20,21. The lamps usually have a standard 300 − 400 nm spectrum, centred at 350 nm. Different light intensities can be used depending on the experimental requirements (e.g., high or low nitrogen oxides (NOx))23.

In most high-NOx experiments, OH radicals are produced by the photolysis of nitrous acid (HONO) using UV lamps24,25,26,27. Even though NOx is produced from HONO photolysis, additional NO can be added to the chamber to reach the desired VOC/ NOx ratios.

For low-NOx experiments, H2O2 is used as the OH precursor25,28. An atmospherically relevant OH concentration (ca.106 molecule cm−3) is produced by photolysing the H2O2 in the chamber29,30,31. OH radicals can also be generated by photolysis of isopropyl/methyl nitrite. (Reactions: 1–3) when added into the chamber32,33.

$$\left( {{{{\mathrm{CH}}}}_{{{\mathrm{3}}}}} \right)_2{{{\mathrm{CHONO/CH}}}}_{{{\mathrm{3}}}}{{{\mathrm{NO}}}} + h\nu \to \left( {{{{\mathrm{CH}}}}_{{{\mathrm{3}}}}} \right)_2{{{\mathrm{CHO/CH}}}}_3 + {{{\mathrm{NO}}}}$$
$$\left( {{{{\mathrm{CH}}}}_{{{\mathrm{3}}}}} \right)_{{{\mathrm{2}}}}{{{\mathrm{CHO/CH}}}}_3 + {{{\mathrm{O}}}}_2 \to {{{\mathrm{CH}}}}_{{{\mathrm{3}}}}{{{\mathrm{COCH}}}}_3/{{{\mathrm{CH}}}}_2 + {{{\mathrm{HO}}}}_2$$
$${{{\mathrm{HO}}}}_2 + {{{\mathrm{NO}}}} \to {{{\mathrm{OH + NO}}}}_2$$

Oxidation flow reactor (OFR) designs

Oxidation Flow Reactors (OFRs) produce highly oxidizing conditions to accelerate processes which would usually take a number of days in the atmosphere to a few minutes in the laboratory/field. In addition to being used to generate SOA particles, the OFRs are also used to simulate atmospheric processing of soot and other model POAs. OFR185 and OFR254 (i.e., 185 and 254 nm photolysis lamps) are the most basic operation modes and have been used most widely to investigate the accelerated atmospheric chemistry34. In OFR254 mode, filtered low-pressure mercury lamps provide 254 nm light to photolyze externally generated O3 in the reactor to form OH (Reactions: 4–5). However, in OFR185 mode the low-pressure mercury lamps are not filtered and emit 185 nm light that can photolyze H2O producing OH, and O2 into O(3P), meaning externally generated O3 is not required. In addition, O(3P) recombines with O2 to form O3, which can also produce OH with the help of 254 nm UV lamps, as in OFR254. Hence, OH radical (8 × 108 − 2 × 1010 molecule cm−3) production processes for both, OFR254 and OFR185, is given by Reactions: 4–5 and 6–9, respectively34,35,36,37. OFRs are operated under continuous flow conditions, where environmental chambers are typically run in batches. The OH/HO2 and OH/O3 ratios in the OFRs are similar to those found in the troposphere, while daytime concentrations of OH, HO2 and O3 are 100 to 10,000 times larger.

For OFR254

$${{{\mathrm{O}}}}_3 + hv\left( {\lambda = 254{{{\mathrm{nm}}}}} \right) \to {{{\mathrm{O}}}}_2 + {{{\mathrm{O}}}}\left({{}^1{{{\mathrm{D}}}}} \right)$$
$${{{\mathrm{O}}}}\left( {{}^1{{{\mathrm{D}}}}} \right) + {{{\mathrm{H}}}}_2{{{\mathrm{O}}}} \to 2{{{\mathrm{OH}}}}$$

For OFR185

$${{{\mathrm{H}}}}_{{{\mathrm{2}}}}{{{\mathrm{O + O}}}}_2 + hv\left( {\lambda = 185{{{\mathrm{nm}}}}} \right) \to \; \to {{{\mathrm{OH + HO}}}}_2$$
$${{{\mathrm{3O}}}}_2 + hv\left( {\lambda = 185{{{\mathrm{nm}}}}} \right) \to \to 2{{{\mathrm{O}}}}_3$$
$${{{\mathrm{O}}}}_3 + hv\left( {\lambda = 254{{{\mathrm{nm}}}}} \right) \to {{{\mathrm{O}}}}_2 + {{{\mathrm{O}}}}\left( {{}^1{{{\mathrm{D}}}}} \right)$$
$${{{\mathrm{O}}}}\left( {{}^{{{\mathrm{1}}}}{{{\mathrm{D}}}}} \right){{{\mathrm{ + H}}}}_{{{\mathrm{2}}}}{{{\mathrm{O}}}} \to {{{\mathrm{2OH}}}}$$

*Two consecutive arrows represent that some basic steps are ignored.

The Potential Aerosol Mass (PAM) reactor is the most popular OFR to study oxidation chemistry, and only one is available commercially (from Aerodyne Research, Inc., Billerica, MA, USA). The reactor is cylindrical (internal diameter ~ 20 cm and volume ~ 10 L) with a residence time of usually 100–200 s34. The role of OFRs in atmospheric chemistry research has already been reviewed by Peng and Jimenez34.

Analytical methods for SOA composition (tracers)

To identify and quantify the chemical composition of SOA, a wide range of online (i.e. CIMS (chemical ionisation mass spectrometry), HR-ToF-AMS (high-resolution time of flight aerosol mass spectrometry))38,39 and off-line (i.e. GC/MS, 2D-GC/ToF-MS) techniques have been used in recent decades. In terms of off-line measurements, SOA products are collected on filters (Quartz or PTFE filters) and then go directly for analysis using thermal desorption-GCxGC-ToF/MS techniques40, or the collected filters can be extracted, derivatised and then analysed by GC/MS or 2D-GC-ToF/MS techniques41.

Soa formation pathways and reaction products

This section reviews the reaction products observed from the oxidation of AVOCs. Details on the reaction products from alkane and aromatic AVOC oxidation are given in Tables 1 and 2. Most studies have involved single precursor VOC, but many have varied the concentrations of NOx, which can fundamentally influence reaction pathways, and have included seed inorganic or organic aerosol, both of which influence SOA yields. The mechanisms and effects of such additional factors are explored in later sections. Overall, this section begins with providing a short description of alkanes and aromatics, and their relevance in the atmosphere, followed by a generalised reaction scheme based on the oxidation from OH, chlorine (Cl) and nitrate (NO3) radical. Later, the section discusses major reaction products obtained from the oxidation of individual precursors under different experimental conditions based on the results from published studies. In addition, a brief introduction to organosulfates formation from the oxidation of alkanes and aromatics is also included.

Table 1 Aliphatic precursors and reaction products.
Table 2 Aromatic precursors and reaction products.

Alkanes and alkyl groups

Alkanes are a main class of AVOCs (40–50% of total VOCs) in urban areas32,42. They are predominantly emitted from combustion sources, such as traffic emissions, petroleum and fossil fuel combustion31. Ensberg et al.43 reported that a major fraction of diesel and gasoline fuels contains cyclic, branched and linear alkanes, while long-chain alkanes constitute a substantial fraction of the unresolved complex mixture from fossil fuels44. Their ambient lifetimes are predominantly governed by reaction with OH and range from 0.5 days for n-hexadecane (C16) to 1.4 days for n-octane (C8)45,46. This is sufficiently long for their transport from urban to rural areas, indicating their importance within the atmosphere. Therefore, following the development sequence, the generalised alkane oxidation reaction scheme (SOA formation pathways) and the oxidation products are discussed below.

Formation pathways

In the atmosphere, OH is the dominant daytime sink of alkanes forming alkyl peroxy radicals (RO2). RO2 subsequently reacts with NO to form alkyl nitrate or alkoxy radicals (RO), which could react with O2, decompose or isomerize23,33. The reaction mechanism of OH with linear, branched, or cyclic alkanes in the presence of NOx is shown in Fig. 1. For linear and branched alkanes, decomposition leads to fragmentation and forms carbonyls and RO2 after addition of O2. For cyclic alkanes, the formation of a single ring-opened carbonyl RO2 is observed. All RO2 formed at different stages (alkyl peroxy, carbonyl alkyl peroxy, and 1,4-hydroxy alkyl peroxy radicals) can react further with NO/O2 or isomerise or react again with OH radicals to form multigeneration products.

Fig. 1: OH radical-initiated reactions of alkanes in the presence of NOx.
figure 1

R, R1, and R2 represent alkyl groups (based on the reaction scheme provided by Lim and Ziemann33).

Major reaction products

A wide range of SOA products have been identified, which not only strongly depend on molecular size and structure of each precursor (linear/ branched/cyclic and functional groups) and degree of oxidation, but also on other chamber conditions during the ageing process (i.e., NOx concentration and relative humidity (RH))47,48. Alcohols, carboxylic acids, ketones and carbonyls are commonly the first oxidation products of branched and linear alkanes22,48,49. Carbonyls are predominantly formed from reactions of OH and C5-C8 n-alkanes with NO49. Reisen et al.49 found that formation of 1,4-hydroxycarbonyls accounts for 54, 57, 51, and 53% of the reaction products from C5-C8 n-alkanes, respectively. The formation of relatively more complex structures occurs as second phase oxidation products come from the aerosol chemistry of carbonyl groups. For example, 1,4-substituted hydroxycarbonyl compounds may undergo cyclization and result in the formation of cyclic hemiacetals that can dehydrate to form highly reactive substituted dihydrofurans or aldehydes that undergo imine formation with amines.

OH-initiated reactions of n-dodecane (C12)/2-methylundecane/hexylcyclohexane and cyclododecane with low-NOx result in the formation of the RO2 radical that rapidly produces a hydroperoxide upon reaction with HO2 as a first-generation product, and subsequently several high generation products, such as hydroxycarbonyls, hydroperoxides, hydroperoxycarbonyls, hydroxy hydroperoxides, peracids, peroxyhemiacetals (PHA), and dihydroperoxide31,50,51,52. It has also been reported that acidic species are expected to form from the photolysis of hydroperoxy groups which include carboxylic acids, peracids, hydroxy carboxylic acids, and hydroxy peracids. It was also reported that the dominant species in n-dodecane SOA formed under low-NO conditions are functionalized hemiacetals and peroxyhemiacetals, which account for 98% of the estimated SOA mass31. They are formed from aerosol chemistry traditionally considered to be acid-catalysed. In addition, low-NOx SOA formed from cyclododecane is mainly composed of ethers and peroxyhemiacetals and contains minor contributions of hydroxycarbonyls, hydroperoxides, and alcohols. Similarly, hexylcyclohexane SOA is composed of hydroperoxides, peroxyhemiacetals, hemiacetals, and hydroxycarbonyls under low-NOx conditions.

Under high NOx, more SOA products are identified, such as nitrate and ester groups44. The formation of hydroxy-nitrates, dinitrates, and alkyl nitrates are observed by oxidation of alkanes (C6-C15) with OH under high concentration of NOx21,22,33. The oxidation products formed from the reactions of OH with linear, branched, and cyclic alkanes are similar. The first-generation SOA products rapidly isomerize in the aerosol phase to form cyclic hemiacetals, which go on to dehydrate to form volatile dihydrofurans. Any volatile products react with OH forming multigeneration products, which are acyclic or monocyclic for all alkanes, and can also be bicyclic from cyclic alkanes. In addition, 1,4-hydroxycarbonyls are an important first generation product in SOA formation from alkanes, as reported by Reisen et al.49.

Similarly, the formation of alkoxyhydroperoxy aldehyde from reactions of cyclic alkenes and O3 was reported53. These compounds can isomerize via intramolecular reaction between the hydroperoxyl and carbonyl groups to form a cyclic peroxyhemiacetal. Other products such as ketones and lactones are identified from the reaction of cholestane, squalane and octacosane, with OH48,54. The octacosanone and octacosanol positional isomers, which are the most prominent first-generation products, were observed during octacosane oxidation. In addition, cholestanones, cholestanals, and cholestanols, are the dominant first-generation functionalization products observed from cholestane oxidation accounting for <70% of the mass of total speciated compounds. Second-generation oxidation products (i.e., fragmentation) and higher-generation products (i.e., functionalization) are much less abundant. For the cyclic compounds, tertiary-carbon alkoxy radicals may have a role in the distribution of functionalization (via isomerization) and the fragmentation products (via decomposition)48,54.

In the photooxidation of dodecane under high-NOx, RO2 + NO gas-phase chemistry forms RO species, which decompose to aldehydes or isomerize to alcohols31. Alcohols, hemiacetals, nitrate-bearing hemiacetals, and acid anhydrides are the major SOA products with minor components of dihydrofurans, furans and ketones. Whereas cyclododecane SOA produced under high-NOx conditions is primarily composed of ethers, esters, hydroxycarbonyls, alcohols, and nitrates. Similarly, hexylcyclohexane SOA produced under similar conditions contains an abundance of oligomeric compounds: hydroxycarbonyl dimers, hemiacetals, and esters31. For dodecane oxidation, a distinct feature is the formation of an ester or ether group formed from the dihydrofuran oxidation by either OH or O355. The formation of tetrahydrofuran and carbonyl ester solely occur via the OH oxidation channel. However, ozonolysis produces an energy-rich primary ozonide, which rapidly decomposes into Criegee intermediates. Similarly, Li et al.56 also observed the formation of hydroxycarbonyls, hemiacetals, and acid anhydrides, under high-NOx conditions from the oxidation of long-chain n-alkanes (C12, C15 and C17). Recent chamber experiments that examined SOA composition from n-alkane (C10-C17) precursors in the presence of NOx also reported the formation of cyclic hemiacetal, hemiacetal nitrate and hemiacetal dinitrates to be first, second and later generation products57.

Overall, the hydroxycarbonyl, the most dominant first-generation product from the alkane oxidation (via isomerization of RO), can undergo heterogeneous cyclisation and dehydration to form a substituted dihydrofuran21,22,33,49,50,51,52,55, which is likely to be highly reactive with OH, O3 and NO3 due to the presence of a double bond. In addition, tertiary-carbon RO are thought to be key species in the formation of the functionalization and fragmentation products for the cyclic compounds.

Aromatic compounds

Aromatic compounds are generally associated with anthropogenic activities, such as combustion emissions and industrial solvents10. Their contribution can reach up to 40% of the total mass of anthropogenic VOC emissions within cities58. In addition, the total aromatic emissions (~15.8 Tg y−1) account for 15% of the annual anthropogenic non-methane hydrocarbon (NMHC) global budget59,60. Aromatic VOCs can yield a wide variety of secondary oxygenated and nitrated volatile, semi-volatile and non-volatile organic products. Considering the complexity involved in the aromatic VOCs oxidation in the atmosphere, we will discuss the reaction scheme first. This is followed by the discussions about the major oxidation products for different aromatic precursors (mono & poly aromatics).

Formation pathways

Atmospheric oxidation of aromatic hydrocarbons is dominated by reaction with OH, resulting in the formation of a variety of oxidation products via ring-retaining and -opening chemical pathways58. Briefly, the OH initiated oxidation of aromatics (≥90%) proceeds via OH-addition to the ring yielding an aromatic-OH-adduct (Ar-OH-adduct), also known as the hydroxycylclohexadienyl radical (and its methylated derivates)61,62,63. An example of the available reaction pathways has been given for the reaction of 1,3,5-trimethylbenzene (1,3,5-TMB) with the OH radical (Fig. 2). Dealkylation and the H-atom abstraction from substituent alkyl groups are minor reaction pathways62,64,65. The fate of the Ar-OH-adduct is dominated by reactions with O2 via reversible/ irreversible pathways. The products of the Ar-OH-adduct + O2 reactions are phenols (C), epoxides (E), and a bicyclic peroxy radical (G). Formation of oxepins (H) has been observed but not in the case of OH + benzene66. The bicyclic peroxy radical (biRO2) is a major product from aromatic oxidation with formation yields of >50% for alkylbenzenes67. The reaction of biRO2 is further assumed to yield bicyclic hydroperoxides (K), bicyclic diols (L), and bicyclic ketones (M). The ring opening could potentially lead to the formation of α-dicarbonyls (e.g., glyoxal, methylglyoxal, and dimethyl-glyoxal), unsaturated γ-dicarbonyls (e.g., butenedial, 4-oxo-2-pentenal) and furanones or a combination of these products.

Fig. 2: OH radical-initiated reaction of 1,3,5-trimethylbenzene (1,3,5-TMB).
figure 2

AS The intermediate products (based on reaction scheme provided by Nehr240).

Major reaction products

Outdoor smog chamber photooxidations of toluene, m-xylene, p-xylene, ethylbenzene, m-ethyltoluene, p-ethyltoluene, and 1,2,4-TMB have been conducted in the presence of sunlight and hydrocarbon−NOx mixtures68. This study reported that unsaturated anhydrides (furandiones) are the predominant components of aerosol from all the aromatics, while a significant amount of saturated anhydrides was also observed, which could have resulted from the hydrogenation of the furandiones within the aerosol phase. In particular, the more dominant species identified in toluene SOA include furandiones, phenols and benzoic acid where ~40% are linked to ring-retaining products and ~60% are most probably arising from ring-fragmentation pathways. In the case of m-xylene and ethylbenzene, ring-fragmentation products comprise a significant portion of the SOA. The dominant organic products include furandiones and m-toluic acid for m-xylene with carbonyls, furandiones and phenols for ethylbenzene. In the case of p-xylene, SOA was composed of similar products as from m-xylene including dicarbonyls and substituted phenols. The ring-retaining and ring-fragmentation pathways were equally as important in 1,2,4-TMB SOA and mainly included benzoic acids, furandiones and carbonyls. Similar products were also observed in m- and p-ethyltoluene SOA68. In addition, ring cleavage products from the gas-phase reactions of the OH radical with p-xylene and 1,2,3- and 1,2,4-TMB under varying concentrations of NO2 included 3-hexene-2,5-dione (32%, 31% percentage yields for p-xylene and 1,2,4-TMB oxidation), 2,3-butanedione (52%) from 1,2,3-TMB oxidation69. The formation of other ring cleavage products, such as unsaturated dicarbonyls and epoxydicarbonyls, were also observed. Similarly, Jang and Kamens70 also studied SOA from the photooxidation of toluene. The gas-phase reaction led to substituted aromatics, and non-aromatic ring—retaining and—opening products. The major SOA products included: nitrophenols, quinones, aromatic carbonyls, oxoacids, and unsaturated dicarbonyls. Results also showed that aldehyde products (glyoxal, methylglyoxal, hydroxyacetaldehyde, and some ring-opening products) could further react through heterogeneous processes and contribute to SOA formation. The formation of substituted aromatics and ring cleavage products from the photooxidation of toluene under different NOx conditions were also reported71. The major chemical constituents of the aerosol are hemiacetal, peroxy hemiacetal oligomers and low molecular weight dicarboxylic acids. In addition, peroxy hemiacetal oligomers and dicarboxylic acids were products formed via heterogeneous reactions of second-generation products.

Glyoxal is a major primary product from reactions of benzene, toluene, or xylene (BTX) with OH, indicating that ring-cleavage pathways (bicycloalkyl radical) play a major role in OH-initiated oxidation of monocyclic aromatic VOCs72. The formation of primary glyoxal is considered as a marker for the bicycloalkyl radical pathway for benzene oxidation giving a lower limit for the branching ratio for the alkyl-substituted aromatics. In addition, the formation of first-generation products such as α-dicarbonyls (glyoxal and methylglyoxal) and unsaturated 1,4-dicarbonyls via ring scission pathways are also observed62,68,70. Similar products were also observed by Birdsall and Elrod62 and Arey et al.73 from OH initiated reactions of toluene, xylenes and TMBs. Other products such as dimethylphenols, diunsaturated dicarbonyls and epoxy carbonyls were observed from the photooxidation of o- and m-xylene in a fast flow reactor at two limiting NO concentrations74. Similarly, Song et al.75 also reported the formation of unsaturated diketones and unsaturated aldehydes following the ring-opening pathways, observing dicarbonyls from xylene isomers oxidation.

Oxidation of 1,3,5-TMB was studied under varying NOx conditions to investigate its potential role in SOA formation76. The main reaction products are 3,5-dimethylbenzaldehyde, nitrogenated compounds, furanone and dicarbonyls including low molecular weight organic acids. Methylglyoxal, 3,5-dimethyl-5(2H)-2-furanone, 3-methyl-furan-2,5-dione (methyl maleic anhydride) and the dicarbonyl 2-methyl-4-oxo-2-pentenal, are the first oxidation products formed via the RO and the bicyclic nitrate under high-NOx. It was proposed that methylglyoxal, the O2-bridged ketone and the m/z 113 isobaric furanone and dicarbonyl ring-opening products could act as markers to ‘represent’ SOA formation.

The NO-dependency of the products formed from the OH oxidation of aromatic compounds was also studied62. For all aromatic species, the bicyclic peroxy radical and bicyclic peroxide products were uniquely observed. As shown in Fig. 3, the bicyclic species are major products under NO-free conditions with yields ranging from 10 to 30%. It also shows that NO plays a role in the partitioning of the observed products between bicyclic and ring scission products (such as butenedial and methylbutendial for toluene specifically). The presence of NO leads to a higher proportion of ring scission products for each aromatic. The dienedial and epoxide products are observed as minor species in all systems, but not in the case of benzene. For the alkyl carbonyl products, this pathway is much more important for ethylbenzene than for the other alkylated aromatics, which is most likely due to the increased hydrogen abstraction pathway reactivity of the methylene group in ethylbenzene. Similarly, Sato et al.77 reported the composition of SOA to be abundant with carboxylic acids or hydroxycarbonyls from benzene and 1,3,5-TMB oxidation. The carboxylic acids from benzene oxidation contain an aldehyde or carboxyl group while the carboxylic acids from 1,3,5-TMB oxidation consist of a ketone and carboxyl group at each end.

Fig. 3: Summary of products formed from the OH oxidation of aromatic compounds.
figure 3

Yields are also reported for NO-free and NO-present conditions for each compound (taken from Birdsall and Elrod62).

Products from toluene photooxidation under low to modest NO conditions contained 2-methyl-4-nitrophenol, a second-generation product, and methyl nitrocatechol, a third-generation product78. In the same study, Pereira et al.78 also investigated SOA composition formed during the photooxidation of 4-methylcatechol, which did not yield nitrophenols in the aerosol phase; only one nitrophenol (4-methyl-5-nitrocatechol) was observed.

SOA components from the photooxidation of monocyclic aromatics were also investigated by Li et al.79 and observed the formation of both carbonyls (ketone or aldehyde) and acids (carbonyl acid and hydroxycarbonyl) as seen previously62,68,70,72,76. Results also reported that oligomerisation is expected to be an important pathway for SOA formation from monocyclic aromatics. In addition, benzene and toluene were considered as the most important monocyclic aromatic precursors to SOA formation due to their high SOA yields. In addition, SOA products formed during o-, m-, and p-xylene photooxidation were compared by Zhang et al.80. This study found that the alkyl substitute position in xylene significantly affects SOA formation, and the effects of NO on the products formed during SOA growth were different for the three xylene isomers. Dicarbonyls, TBM (C3-trione + 2,3-butanedione + 3-methyl-2-oxiranecarbaldehyde), and highly oxidized species (HOS) were determined to be the predominant SOA components arising from xylene photooxidation80. High NO levels were noted to inhibit the formation of C3-trione and 2,3-butanedione in the SOA from m- and o-xylene, whereas the formation of 3-methyl-2-oxiranecarbaldehyde during p-xylene photooxidation was significantly promoted.

Overall, the major photooxidation products from aromatics (mainly BTX) are glyoxal, methylglyoxal, unsaturated dicarbonyls and furanones via ring-opening pathways, and aromatic aldehydes/acids and phenols via ring-retaining pathways. The hydroxycylclohexadienyl radical, an aromatic-OH adduct, is the main primary product of the reaction of the aromatic ring with OH.

Another important class of compounds is polycyclic aromatic hydrocarbons (PAHs). Those with <4 rings mainly exist in the gas phase and can undergo gas-phase oxidation processes to produce oxygenated and nitro compounds that contribute to SOA formation81,82,83,84. PAHs can be a potential contributor to the SOA budget, especially in urban areas81,82. The OH-initiated photooxidation of naphthalene was examined85, showing ring-opening products (e.g., 2-formylcinnamaldehyde, phthaldialdehyde, and phthalic anhydride) are the primary gas-phase products under high-NOx conditions via the RO2 + NO pathway or a bicyclic peroxy mechanism. The low-NOx chemistry includes the formation of ring-retaining compounds (e.g., naphthol, naphthoquinone, and epoxyquinone) via the RO2 + HO2 pathway85. Results also suggested that acidic compounds and organic peroxides account for a majority of identified high- and low-NOx SOA. Organic peroxides are found to contribute ~28 and 26% of the total high-NOx SOA and low-NOx SOA mass, respectively. In addition, the study also proposed the use of 4-nitro-1-naphthol as a valid ambient organic tracer for naphthalene high-NOx SOA.

Simulation chamber experiments studying gas-phase oxidation of acenaphthylene with OH showed the major products were naphthalene-1,8-dicarbaldehyde, 1,8-naphthalic anhydride, and a 10-carbon ring opened dialdehyde83. Acenaphthenequinone, a compound known to promote the formation of reactive oxygen species at the cellular level, was also identified as a product of oxidation with OH and NO3. Ring-retaining and -opening products were also identified from the oxidation of acenaphthene, although nitroacenaphthene, and 1,8-naphthalic anhydride were more abundant in the aerosol phase84. In particular, hydrogen abstraction from the cyclopenta-fused ring is reported to be an important reaction pathway for the formation of 1-acenaphthenone and naphthalene-1,8-dicarbaldehyde84. Studies of photochemical aging of SOA produced by the gas-phase oxidation of naphthalene by OH and acenaphthylene by O3 reported naphthaldehyde, oxaacenaphthylene-2-one, and hydroxyl-naphthaldehyde as major SOA products86.

Role of chlorine (Cl) and nitrate (NO3) radical initiated oxidation

The above discussion illustrates that OH is central to oxidative processing of anthropogenic VOCs in the troposphere. However, the significance of Cl reactions in the Arctic lower troposphere, marine boundary layer and coastal regions has become apparent87,88,89. The rate constants of reactions of VOCs with Cl are a factor of ~102 higher than those with OH90, implying that even when tropospheric Cl concentrations are relatively low, reactions of VOCs with Cl can be competitive with or even exceed those of OH with VOCs. Similarly, volatile phenolic derivatives are most likely to be emitted from biomass burning and are also produced from the photooxidation of aromatic VOCs. These phenol derivatives can undergo rapid reactions with NO3 at night and represent a significant source of SOA and brown carbon (BrC) in the atmosphere. Therefore, rapid atmospheric removal is expected upon reactions with OH during the day and NO3 during the night, suggesting a key role of the NO3 radical in VOCs oxidation. Details on Cl and NO3 initiated oxidation of VOCs including their reaction pathway and the major reaction products are discussed here.

Cl radical-initiated oxidation

In the atmosphere, Cl atom-initiated oxidation is possible either via Cl addition to the aromatic ring or H atom abstraction from the ring91. Oxidation products from the reaction of Cl atoms with aromatic hydrocarbons include benzaldehyde and benzyl alcohol (95% of total products) from toluene and 1-naphthaldehyde and 1-naphthyl alcohol (48% of total products) from 1-methylnaphthalene (1-MN)92. The benzyl peroxy radicals undergo self-reaction to form benzaldehyde and benzyl alcohol in the molecular channel and benzaldehyde via the radical channel.

These products confirm that the hydrogen abstraction reaction is the dominant pathway for oxidation of alkylbenzenes and alkylnaphthalenes, unlike the OH radical reactions, where OH radical addition to the aromatic ring dominates.

Experiments investigating Cl atom-initiated oxidation of naphthalene, acenaphthylene, and acenaphthene show phthalic anhydride and chloronaphthalene as the major products, indicating that both reaction pathways, H atom abstraction and Cl addition are important91, contrary to the findings reported by Wang et al.92. As an example, the proposed mechanism for the Cl atom initiated oxidation of naphthalene is given in Fig. 4. The Cl-naphthalene reaction could result in the formation of an RO2, chlorinated radical, chlorinated oxy radical and acyl radical, where the acyl radical can react with O2 and then undergo cyclisation to form phthalic anhydride. Acenaphthenone is the main product observed from the reaction of Cl with acenaphthene, while 1,8-naphthalic anhydride, acenaphthenone, acenaphthenequinone, and chloroacenaphthenone were identified as products of acenaphthylene oxidation. The identified oxidation products such as chloronaphthalene and chloroacenaphthenone can be used as potential markers for the Cl atom-initiated oxidation of PAHs.

Fig. 4: Formation pathways for the Cl atom-initiated oxidation of naphthalene.
figure 4

AE represent the products observed in both gas and particulate phase (based on the reaction scheme provided by Riva et al.91).

SOA formation from Cl-initiated oxidation of toluene under different conditions has also been investigated93. The difference between high and low NOx experiments specific to chlorine-initiated chemistry is the formation of ClNO2 in the high NOx environment, which can result in decreased concentrations of Cl at the beginning of the experiment and also a sustained source of Cl throughout the rest of the experiment. In contrast, in low NOx experiments most Cl forms HCl, which is relatively stable to photolysis and does not result in significant recycling of Cl. The investigation also showed that bulk organic composition was different for SOA from Cl-dominated reactions compared to OH-dominated reactions, including highly oxidised SOA formed through Cl-initiated reactions. The identified reaction products included benzenetriol, benzoquinone, benzaldehyde, phenols, and benzoic acid.

The formation of organic chlorine upon reaction with alkanes occurs through the heterogeneous production of dihydrofuran via 1,4-hydroxycarbonyl uptake, acid-catalysed isomerization, and dehydration reactions (Fig. 5)22,33. Wang and Hildebrandt Ruiz94 reported the gas-phase alkane—Cl oxidation products formed under high-NOx conditions were dominated by organonitrates. Organochloride formation is found to be minimal under humid conditions, where dihydrofuran formation is inhibited95. The dominant gas-phase organic chlorine compounds were: C2H3ClO2 and C2H2Cl2O2. In another study, the Cl-initiated photo-oxidation reactions of n-alkanes: C10 ~ C14, were also investigated and the formation of carbonyls without any chlorinated products was reported96, contrary to the results observed by Wang and Hildebrandt Ruiz94 Decanal and decyl alcohol were the dominant gas phase products of the reaction for n-C10 alkane with Cl atom via H-atom abstraction, including two types of reaction with site-internal H-atoms (methylene H atom) and terminal H-atoms (methyl H atom). The oxygenated products continued to react with Cl atoms and then decomposed to form short-chain oxygenated products. In addition, some oxygenated compounds retained the carbon chain after abstracting the methyl H atom and were formed by subsequent reactions with O2. Similarly, C11 ~ 14 n-alkanes also react with Cl atoms by an H-atom abstraction channel; however, they react further with Cl to decompose into small molecule products without retaining the main chain compounds. The reaction of C11~14 n-alkanes with Cl atoms and O2 led to the initial formation of RO, and further RO arose from three reaction pathways: reaction with O2 to form carbonyls and continued reaction with the Cl atoms; an intramolecular hydrogen transfer reaction to form alkyl radicals; and decomposition reactions. These reaction pathways could finally form stable compounds with multifunctional aldehydes and ketones as products.

Fig. 5: Formation pathway for the reaction of alkanes with Cl atoms to form chlorinated organics.
figure 5

R1 and R2 represent alkyl groups (based on the reaction scheme provided by Wang and Hildebrandt Ruiz94).

NO3 radical-initiated oxidation

The most common gas-phase reaction scheme for the oxidation of phenolic VOCs is the reaction with NO3, which leads to the addition of at least one nitro group (−NO2) to the phenolic backbone. The main products from the NO3 oxidation of phenolic VOCs include the nitro- and dinitro-phenols as well as a variety of their dimer products, which likely undergo gas-particle partitioning (GPP)97,98. An example of the reaction pathways is given for the oxidation of phenol with the NO3 radical in Fig. 6.

Fig. 6: General mechanism for the NO3 oxidation of phenol.
figure 6

Green and brown colour represent the products containing single nitro group, blue and red colour represent the products containing two nitro group and orange colour represents the product containing three nitro group (based on reaction scheme provided by Mayorga et al.97).

The reaction pathway outlined in Fig. 6 has the capability to form a wide variety of dimers upon NO3 ∙ oxidation with up to three nitrogen atoms per dimer molecule. Many of the dimers in the systems are formed in the gas phase via phenoxy radical pathways, followed by GPP. This is supported as phenol and catechol are present in the dimers; they are unlikely to be present in the particle phase in substantial amounts to participate in condensed phase dimerisation. Similar gas-phase phenoxy radical chemistry has also been proposed previously99,100.

The reaction of phenol with NO3 was found to yield 2-nitrophenol as the main nitration product and was independent of the concentration of NO298. In addition, in the presence of O3 the formation of 4-nitrophenol and p-benzoquinone was observed. The products formed from the NO3 radical-initiated oxidation of o-, m-, and p-cresol with high yield were alkylnitrophenols, substituted quinones and HNO3101. The formation of 4-nitrocatechol from the reaction of NO3/OH with catechol in the presence of NOx comprises a major component of SOA formed102. In addition, the formation of 5-nitro-1,2,3-benzenetriol was also observed from the OH reaction. It was postulated that the formation of 4-nitrocatechol is the result of phenolic hydrogen abstraction by OH or NO3 to form a β-hydroxyphenoxy/o-semiquinone radical, which then goes on to react with NO2 to form the final product102.

The reaction of resorcinol (1,3-benzenediol) with NO3 resulted in a lower SOA yield than the OH oxidation103. This can be explained by differences in the reaction mechanisms as OH radicals primarily add to the aromatic ring to form a variety of low-volatility products, while reactions with NO3 radicals occurred solely by abstraction of phenolic hydrogen to produce much more volatile nitroresorcinol and hydroxybenzoquinone products. The major oxidation products identified were benzenetriol, nitrobenzenetriol, and hydroxymuconic semialdehyde in the aerosol phase and hydroxybenzoquinone and nitroresorcinol in the gas phase.

The NO3 oxidation of five phenolic derivatives (including phenol, catechol, 3-methylcatechol, 4-methylcatechol, and guaiacol) resulted in several nitro-containing products and nitrophenol compounds, including the nitrophenol type of products with additional hydroxyl functional groups, non-aromatic/ring-opening nitro-products, nitrated diphenyl ether dimers and fragmentation products with carbon-containing substituents were investigated97. This study indicated that NO3 oxidation of phenolic VOCs may contribute an important portion of SOA and can lead to a significant portion of light-absorbing BrC in the atmosphere.

Additionally, NO3 initiated oxidation reactions of PAHs are also particularly important as they can lead to the formation of nitro-PAHs, some of which are known to be mutagenic and carcinogenic104. The oxidation products formed from the reaction of NO3 with acenaphthylene were predominantly composed of oxygenated compounds, including the formation of hydroxylated and nitro-PAHs, which appears to be very minor83. The major gas- and aerosol-phase products from the oxidation of acenaphthene were 1-acenaphthenone and nitroacenaphthene, respectively84.

Organosulfates—a major SOA constituent

This section is designed to provide a brief introduction to Organosulfates (OS) including their possible formation pathways. In addition, we also focus on the OS compounds identified from the oxidation of anthropogenic precursors such as alkanes and aromatics. During the past decade, OS has gained constant attention with studies reporting a high concentration of OS species in the ambient environment. However, estimates of OS concentrations and their role in atmospheric physicochemical processes are still associated with large uncertainties, suggesting the need for comprehensive research involving field, laboratory, and modelling approaches. The latest review performed by Brüggemann et al.105 presents extensive information on the formation pathways of OS and their atmospheric relevance, transformations, reactivity, and fate in the atmosphere, and measurement techniques. In addition, the review also highlights a global overview on the OS measurements from different precursors. Therefore, in this section, we will briefly discuss OS and their relevance in the atmosphere avoiding any detailed information which can be easily accessed from the existing review105.

OS are another important class of compounds and generally considered as potential markers for SOA formation under acidic conditions by aerosol-phase reactions with sulfuric acid106, formed by oxidation of sulphur dioxide (SO2). SO2 is mainly of anthropogenic origin, which suggests the formation of OS can be significant in highly polluted regions around the globe. OS formation possibly occurs through a wider variety of formation mechanisms in the atmosphere, either via heterogeneous or via bulk aerosol-phase reactions, than initially expected. This includes epoxide pathways107,108, sulfate esterification reactions109, nucleophilic substitution109,110, sulfoxy radical reactions111,112, and heterogeneous reactions of SO2113. The formation of OS via epoxide intermediates is kinetically possible under tropospheric conditions while, OS formation via a direct sulfate esterification pathway may occur rarely. The sulfate esterification needs to be conducted in the absence of water, which is unlikely in the real atmosphere. Despite this, lab studies have reported the formation of OS by direct sulfate esterification109, but expected to play a less important role in the atmosphere. The OS formation may also occur via nucleophilic substitution of alcohols with sulfuric acid109,110. As mentioned, an OS formation pathway is also linked with sulfoxy radical reactions in the aqueous phase via the addition of sulfoxy radicals to double bond molecules114 or the reaction between a sulfate radical and an alkyl radical111,112. The second pathway (radical − radical reactions) is expected to be minor in the presence of dissolved oxygen. In addition, the formation of OS has been also reported through heterogeneous reactions where gas-phase SO2 reacts with unsaturated hydrocarbons in the absence of gas-phase oxidant like OH or O3113.

Several studies have already indicated that OS could account for a substantial fraction (up to 30 %) of the organic mass measured in ambient PM2.5115. Although most of the evidence for OS identification comes from field measurements115,116,117, only a few OS precursors have been characterised through laboratory studies. In most of the laboratory studies, OS has been identified in the presence of acidified sulfate seed aerosol from OH, NO3 or O3 oxidation of biogenic VOCs (BVOCs), including isoprene106,118, 2-methyl-3-buten-2-ol (MBO)119,120, unsaturated aldehydes121,122,123, monoterpenes109,115 and sesquiterpenes111,124,125,126. Only a few laboratory studies have reported OS formation from anthropogenic precursors127,128. However, Ma et al.129observed that aromatic OS could account for two-thirds of the OS mass identified in Shanghai. In addition, the formation of aliphatic OS was also reported in the ambient samples from other urban locations116,117, suggesting that gas-phase oxidation of long-chain or cyclic alkanes could be an important source of OS116.

The formation of OS from the photooxidation of PAHs (naphthalene and 2-MN) was investigated in the presence of non-acidified and acidified sulfate seed aerosol127. Most of the OS compounds identified from the smog chamber experiments were also reported in the samples collected at Lahore, Pakistan, and Pasadena, USA, indicating that PAH photooxidation in the presence of sulfate aerosols can be a source of anthropogenic OS compounds. Results also suggest that some of the identified OS such as C10H9O6S (m/z 257.0139) from naphthalene photooxidation, and C6H4NO6S (m/z 217.9751), C9H11O5S (231.0333), and C11H13O7S (289.0330) from 2-MN photooxidation can be used as a new PAH SOA tracer127. Similarly, the same outdoor smog chamber was used to study the OS formation from alkanes (dodecane, cyclodecane and decalin) photooxidation128. Most of the OS identified could be explained by reactive uptake of epoxides onto sulfate aerosol and/or heterogeneous reactions of hydroperoxides. However, OS formation via acid catalysed reactive uptake of epoxides has been observed only for cyclic alkanes. In addition, identified OS such as C7H15O4S(m/z 195), C10H17O5S(m/z 249), C9H19O6S−(m/z 255), C10H19O6S(m/z 267), C10H17O7S(m/z 281), C10H19O8S(m/z 299), C12H19O7S(m/z 307), C10H15O9S(m/z 311), can be used as tracers for SOA production from the oxidation of alkanes in urban areas. Besides, laboratory evidence for OS formation from anthropogenic precursors is scarce, and thus, more studies are needed to assess the importance and implications of OS for atmospheric composition.

SOA yields

In this section, we summarize SOA yields from the oxidation of alkanes and aromatics based on relevant laboratory studies conducted to date. The data presented here include SOA yields from different precursors under varying atmospheric conditions/parameters. The influence of atmospheric parameters on the SOA yield is discussed later (section 4.2) but also mentioned here when appropriate.

SOA yields: oxidation of alkanes and aromatics from lab studies

Conventionally, SOA formation potential (SOA yield) is defined as the mass ratio of formed SOA and the consumed precursor VOCs (Eq. 10) for a given set of experimental conditions17.

$$Y\left( {SOA} \right) = \frac{{m(SOA)}}{{{\Delta}\left[ {VOC} \right]}}$$

In addition, SOA yield can also be represented as a function of the organic aerosol mass concentration using a semiempirical absorptive GPP model130,131.

$$Y\left( {SOA} \right) = {\Delta}M_o\sum _{i} \frac{{\alpha _{iK_{om,i}}}}{{1 + {\Delta}M_o}}$$

where ∆Mo (µg m−3) is the total OA mass concentration, and αi is the mass-based stoichiometric coefficient for product i that is formed, and Kom,i (m3 µg−1) is an equilibrium absorptive partitioning coefficient. Generally, two surrogate species (i = 2) have been proven sufficient to parametrize yields.

Details on the measured SOA yields from alkanes and aromatics precursors are given in Tables 3 and 4. Yield estimates are usually based upon quite long reaction times to consume a large proportion of the precursor, but at extended reaction times SOA mass can both increase as more heavily oxidised products are formed and decrease due to formation of volatile species in fragmentation reactions of primary oxidation products.

Table 3 SOA yields from aliphatic precursors.
Table 4 SOA yields from aromatic precursors.

Alkane mass yields

SOA yields for n-alkanes ranged from 0.005 for C8 to 0.53 for C15, with a sharp increase from 0.08 for C11 to 0.50 for C13, suggesting this could be linked to an increase in the contribution of first-generation products (hydroxycarbonyls)21. Lim and Ziemann33 also showed a decrease in the SOA yields formed from the oxidation of C15, C12, and C10 from 0.63 to 0.35 to 0.15 due to an increase in the amount of volatile reaction products with decreasing carbon number, which subsequently reduces their tendency to form SOA. For branched alkanes, the yields follow the order: C12 > 2-methylundecane > 2, 3-dimethyldecane from 0.35 to 0.21 to 0.11. SOA yields were also reported to increase with increasing carbon number (lower volatility) for long-chain n-alkanes (C10, C12, C15, and C17), reaching a yield of 0.51 for C17 at an organic mass concentration (COA) of 15.4 µg m−3 under high NOx concentration26. However, Lim and Ziemann33 measured much higher SOA yields than those observed by Presto et al.26 as their goal was to identify SOA products, thus, the experiments were conducted at much higher COA (> ~1000 μg m−3).

Tkacik et al.27 extended the work done by Lim and Ziemann33 and performed smog chamber experiments with different alkanes, including cyclic, branched, and linear compounds to determine SOA yields with OH radical-initiated reactions under high-NOx conditions at an atmospheric COA. SOA yields from cyclic alkanes were comparable to those from linear alkanes with 3–4 carbons larger in size. For alkanes with equivalent carbon numbers, branched alkanes had the lowest SOA yields, ranging between 0.05 and 0.08 at a COA of 15 μg m3. The SOA yields from a subset of n-alkane precursors (C10, C15, tricyclo[,6]decane (C10H16)) were also investigated and reported maximum SOA yields were 0.39, 0.69, and 1.1132.

Loza et al.23 found that the SOA yields were highest from cyclododecane under both NOx conditions. SOA yields ranged from 0.033 (C12, low-NOx) to 1.6 (cyclododecane, high-NOx). Under high-NOx conditions, SOA yields followed the order 2-methylundecane (0.11–0.38) < C12 (0.23–0.62) ~ hexylcyclohexane (0.34–0.61) < cyclododecane (0.8–1.6). Under low-NOx conditions, SOA yields increased from 2-methylundecane (0.14–0.31) ~ C12 (0.03–0.28) < hexylcyclohexane (0.30–0.65) < cyclododecane (0.22–0.86), suggesting a positive influence of cyclization present in the alkane structure. The yields of SOA formed from the reaction of OH with a series of acyclic (C10), monocyclic (cyclodecane (C10H20)), polycyclic (decalin (C10H18), pinane (C10H18), tricyclo[,6]decane (C10H16), and adamantane (C10H16)) alkanes, were also examined to better understand the role of multiple cyclic moieties on SOA formation24. Results showed that the SOA yield for C10, cyclodecane, decalin, pinane, tricyclo[,6]decane, adamantane ranged from 0.096–0.254, 0.964–1.37, 1.13–1.38, 0.38–0.396, 0.838–1.81 and 0.99–1.14, respectively. The SOA yields were also investigated from OH oxidation of selected alkanes (C7, C10, C12 and decalin) under low-NOx condition in an OFR by Li et al.35 . The SOA yield for C7, C10, C12, cyclodecane, and decalin ranged from 0.044–0.099, 0.295–0.534, 0.663–0.960, 1.639–2.121, and 1.532–2.298, respectively in the presence or absence of seed aerosols.

Another environmental chamber experiment was carried out to investigate the Cl-initiated oxidation of n-alkanes (C8–12) under high NOx by Wang and Hildebrandt Ruiz94. They showed that the observed SOA yields were higher for C8 (0.16) and C12 (1.65) than those resulted from OH-initiated oxidation (C8 = 0.04; C12 = 0.35). Under humid conditions, the bulk SOA production was found to be suppressed, though the SOA yields observed for Cl-initiated oxidation of C8 (0.24), C10 (0.50), and for C12 (1.10) were still much higher than those observed for OH-initiated oxidation.

Aromatic compounds mass yield

Izumi and Fukuyama133 studied the photochemical aerosol formation from mixtures of aromatic hydrocarbons and NOx by exposing the gas mixtures to simulated solar radiation in a simulation chamber with the aim of determining the aerosol yields. It was found that yields were greater for toluene, ethylbenzene, and o- and m-ethyltoluene (0.03, 0.031, 0.033 and 0.037, respectively) but lesser for para-substituted toluene derivatives (<0.015). The strong dependency of SOA yields on COA were observed by Odum et al.130 who showed that the yield for m-xylene and 1, 2, 4-TMB ranged from 0.19–10% and 2.6–9.67%, respectively. Their results also suggest the yield is higher at low temperatures for a given COA. The SOA yields for 1, 2, 4-TMB derive from experiments conducted at temperatures from 22 to 26 °C. The yield values are like those for both the lower and higher temperature m-xylene at COA below 60 µg m−3. At COA above 60 µg m−3, the yields for 1, 2, 4-TMB are higher than those for the higher temperature m-xylene data and lower than those for the lower temperature m-xylene data. Thus, for most atmospherically relevant COA values (i.e., ≤60 µg m−3), 1, 2, 4-TMB has a similar yield to m-xylene.

In addition, an extensive series of sunlight radiated smog chamber experiments were performed with 17 individual aromatic species to determine their SOA yields131. The studied VOCs were divided into two distinct classes, namely low- (i.e., ≥ two methyl substituents) and high-yield (i.e., ≤ one methyl or ethyl substituents) aromatics. The SOA yields for low- and high-yield aromatics ranged from 0.019–0.082 and 0.030–0.124, respectively. Similarly, the irradiation experiments conducted in the presence of NOx reported the largest SOA yield of 1.59 for toluene, followed by 1.09 for p-xylene and 0.41 for 1,3,5-TMB134. However, Edney et al.19 observed a lower yield for toluene. The SOA yields were also investigated from the photooxidation of m-xylene, toluene, and benzene under two limiting NOx conditions25. Under low-NOx conditions, the SOA yields for m-xylene, toluene, and benzene are broadly constant (0.36, 0.30, and 0.37, respectively), indicating that the SOA formed is effectively nonvolatile under the range of organic mass concentration (>10 µg m−3) studied, whereas under high-NOx conditions aerosol growth is immediate. Losses of semivolatiles by chemical reaction or to walls lead to lower SOA formation in chambers than in the atmosphere. Considering the slower oxidation rate observed in the photooxidation experiments performed by Odum et al.130 and Odum et al.131, it is likely that their SOA yield parameters underestimate SOA formation from aromatic hydrocarbons in the atmosphere. Similarly, the SOA yield from xylene oxidation during m-xylene/NOx experiments showed that at lower NOx considerably more organic aerosol mass was generated than for those with higher NOx135. The observed SOA yield from xylene oxidation ranged from 0.032–0.141. In a similar experiments, Song et al.75 investigated SOA formation from the photooxidation of xylene isomers (m-, p-, and o-xylenes), and found that the yield from xylenes ranged from 0.011 to 0.234, strongly depending upon the level of NOx. A similar SOA yield from the photooxidation of p-xylene in the presence of NOx was also reported by Healy et al.136. In addition, Lu et al.137 investigated another xylene isomer and reported yields with seed (0.05–0.12) and no seed conditions (0.05–0.08), indicating higher yield in the presence of (NH4)2SO4 aerosols.

The enhanced SOA yields from 1,3,5-TMB oxidation under low NOx reported by Wyche et al.76 indicate a role for oxygen-bridged species and organic acids during aerosol growth. The observed SOA yield ranged from 0.0621–0.0637 and 0.0029–0.0747 under low- and high-NOx, respectively. Sato et al.77 showed the SOA yield from benzene to be higher, however, the yield measured (0.025) in one of the experiments was slightly lower than the SOA yield measured by Odum et al.131 (0.036). SOA formation from the photooxidation of 12 C8 and C9 aromatic hydrocarbons under low-NOx was investigated138. Ortho isomers (0.044–0.237) have the highest SOA yield while para isomers (0.024–0.122) have the lowest SOA yield. The lower SOA yields for para isomers are consistent with previous observation by Izumi and Fukuyama133 and Li et al.79. Also, SOA yields of benzene under comparable low-NOx were higher than those in Sato et al.77. A similar trend was also observed for SOA yields of toluene139.

SOA yields were also estimated from PAHs oxidation. SOA formation from photooxidation of naphthalene, 1-MN, 2-MN, and 1,2-dimethylnaphthalene (1,2-DMN) were examined by Chan et al.140. This study found SOA yields were 0.19–0.30 for naphthalene, 0.19–0.39 for 1-MN, 0.26–0.45 for 2-MN, and 0.31 for 1,2-DMN under high-NOx and aerosol mass loadings (10–40 μg m−3). Under low-NOx, yields were 0.73, 0.68, and 0.58, for naphthalene, 1-MN, and 2-MN, respectively. More ring-retaining products were observed in the gas phase under low-NOx, resulting in low volatility SOA and constant yields. Their results also showed the SOA yields for naphthalene and alkyl naphthalenes are on the order of 0.25–0.45 under high-NOx, which are about three times those of light aromatics. Shakya and Griffin82 determined the aerosol yields for PAHs to be in the range of 0.02–0.22 with 0.20–0.30 uncertainty and in the following order: naphthalene (0.13), 1-MN (0.08), acenaphthylene (0.07), 2-MN (0.06), and acenaphthene (0.06), respectively. High yields for naphthalene may be attributed to larger concentrations of OH and initial ratios of hydrocarbons and NO (HC/NO) compared with the PAH experiments. In addition, the yields reported in their study are within the range of 0.25–0.40 for high NOx observed by Chan et al.140. In another study, the yields (Cl oxidation) were determined to be 0.91, 0.85, and 0.98 for naphthalene, acenaphthylene, and acenaphthene, respectively91, which is ~3 × higher than from OH oxidation140.

The yields of SOA formed from reactions of catechol with OH and NO3 radicals were 1.34 ± 0.2 and 1.50 ± 0.2102. These values are exceptionally high, considering that those measured for the photooxidation of simple aromatics ranged from about 0.01 to 0.78 shown above. The SOA yield reported for the OH radical-initiated reaction is much higher than the value of 0.53 reported previously by Borrás and Tortajada-Genaro141. In another study, Finewax et al.103 investigated the SOA formation from resorcinol with OH and NO3, and SOA yields of 0.86 and 0.09, respectively were reported.

Factors affecting SOA yields

Aerosol yields are dependent on several parameters, such as the concentration of NOx in the system, the oxidation rate of the precursor, RH, temperature (T) and the type of seed aerosols used. Such factors can influence SOA yields from both anthropogenic and biogenic precursors, although the magnitude of the effect is dependent upon the properties of the individual reaction system. Lower volatile and reactive species are particularly affected by chamber walls losses, which can significantly impact SOA formation experiments as they involve highly reactive oxidants and the production of lower volatility products.

Effects of experimental conditions (i.e., temperature, RH, UV intensity)

RH controls the liquid water content of the aerosols and thus the chemical or physical processes that involve water (i.e., reactant, product, or solvent) are affected14. However, the number of studies that have systematically investigated the impacts of RH on SOA formation are limited.

Cocker et al.142 reported the effect of RH on SOA mass yield for the m-xylene and 1,3,5-TMB photooxidation in the presence and absence of inorganic (ammonium sulphate, (NH4)2SO4) seed aerosol at <2% RH and 50% RH and concluded the impact of RH was not substantial within this range. The effect of RH on SOA formation from the photooxidation of p-xylene was also studied at atmospheric pressure and ambient temperature by Healy et al.136. They showed an increase in RH results in higher SOA yields and HONO formation, which leads to increased OH concentration. The OH concentration showed a gradual increase with RH and correlated with aerosol yields and RH. In addition, study of SOA formation from toluene and xylenes at different RH also found that increasing liquid water led to increasing SOA formation143,144.

The formation of SOA from benzene and ethylbenzene in the presence of NOx at different RH was also investigated145. Their results suggested that aqueous-phase reactions and the hydration from glyoxal could be enhanced under high RH conditions, leading to an increase in the formation of SOA from both benzene and ethylbenzene. Hinks et al.29 studied the effect of RH on the chemical composition of SOA formed from low-NOx toluene oxidation and showed a significant loss in the amount of oligomers present in the SOA generated at 75% RH compared to that under dry conditions. In addition, their results also confirmed that low-NOx toluene SOA is more volatile when formed under high-RH conditions29. RH-induced oligomer suppression was also reported for Cl-initiated oxidation of n-alkanes (C8–12) under high NOx94. SOA yields decreased (0.37 for decane and 0.20 for dodecane) under humid condition (35–67%), similar to the reduction reported in SOA yields for dodecane-OH SOA formation under humid conditions by Fahnestock et al.31. Similar results were also observed by Lamkaddam et al.146 from high-NOx dodecane SOA formation. Their study reported a reduction in SOA yield by a factor of 2 as conditions went from dry (RH < 1%) to more humid conditions (RH ≥ 5%). The role of water in SOA formation from the irradiation of a toluene-NO2 system was also investigated by Jia and Xu147 who reported that the yield of SOA from toluene almost doubled as RH was increased from 5 to 85 %, mainly due to the contribution of O–H-containing products, such as poly alcohols formed from aqueous reactions.

Temperature is another important factor affecting SOA formation due to its effect on both the vapour pressure of the SVOCs and the rate constants of the oxidation processes. Takekawa et al.20 performed photooxidation experiments to study the temperature dependence of SOA formation from three aromatic hydrocarbons (i.e., toluene, m-xylene and 1,2,4-TMB) and one alkane (i.e., n-undecane). Their results showed that a higher SOA yield (~2×) was obtained at 283 K than that at 303 K. In another study the dependence of aerosol mass yield on temperature was investigated in the toluene SOA system30, which also showed higher SOA yields at lower temperatures (284 K > 305 K) under both low- and high-NOx, consistent with the previous observation20. These results all suggest that many of the oxidation products are semi-volatile and hence partition more extensively into the condensed phase at lower temperatures. An investigation of the temperature dependence of SOA yields for n-dodecane (C12H26) in the presence of NOx with and without seed aerosol between 283.0–304.5 K (Fig. 7) showed no strong dependence32. This was attributed to a change of rate constants leading to different SOA composition, or non-volatile oxidation products.

Fig. 7: The n-dodecane temperature-dependent SOA yields as a function of the organic aerosol mass.
figure 7

The colours represent the different temperature at which SOA yields were calculated (taken from Lamkaddam et al.32).

UV intensity has also shown a positive effect on SOA yield26,30. SOA formed under high- and low-UV intensity showed similar composition as reported by Presto et al.26. High SOA yield was expected corresponding to more oxygenated products from increased radical cycling and multigenerational chemistry, but AMS data did not confirm this. In addition, Hildebrandt et al.30 showed that the toluene SOA formation is sensitive to UV intensity.

Overall, most of the studies showed a positive effect of RH on SOA yields from aromatics through processes which can enhance the SOA formation by aqueous reactions143,144,145,147,148. Unlike RH, temperature has shown a negative effect or little influence on SOA formation due to enhanced partitioning into the condensed phase at lower temperature. In addition, UV intensity also exhibited positive effect on SOA yields.

Effects of molecular structure

The effect of molecular structure/alkyl group position on the measured SOA yield has been investigated33. SOA yields from OH radical-initiated reactions of alkanes in the presence of NOx increase with carbon number due to enhanced formation of non-volatile reaction products. The effect of carbon number of alkanes on SOA composition/yield was also investigated through the oxidation of a range of n-alkanes (C10-C17)57. The relative contribution of first-generation cyclic hemiacetal gradually increased with n-alkane size, but the relative contribution of second generation and higher nitrate-containing species (i.e., hemiacetal nitrate and hemiacetal dinitrate) decreased with an increasing alkane size (carbon number). Also, larger decreases for C11 were observed compared to C17 with increasing RH, indicating that the SOA yield also depends on the size of the parent alkane.

Branched alkanes have lower yields than linear alkanes because of fragmentation and the formation of more volatile products. Cyclic alkanes have higher yields than linear alkanes due to the formation of less volatile ring opened products with an additional aldehyde group. The lower volatility of these products, and their tendency to form oligomers, appears to enhance SOA yields. Presto et al.26 observed the positive influence of carbon number on SOA yields from n-alkanes (C10, C12, C15, and C17) oxidation. SOA yield of branched alkanes also depends on the position of the methyl group on the carbon backbone as suggested by Tkacik et al.27. It is possible that a RO adjacent to a branch point predominantly undergoes decomposition leading to fragmentation, and forms products containing fewer carbon atoms that are likely to be too volatile to partition to the particle-phase.

The presence of cyclization in the parent alkane structure has been shown to increase measured SOA yields, whereas the presence of branch points decreases SOA yields, which was more pronounced under high-NOx than low-NOx23. Higher SOA yields from cyclic and polycyclic precursors have a negligible dependence on the number of rings24.

For aromatic precursors, alkyl substitution and the o-, m- and p-position affects the SOA yield131,133. Odum et al.131 reported that the single-substituted aromatics might generate a higher ratio of ring-retaining to ring-cleavage products than the multiple-substituted aromatics. Ring-retaining products may have lower vapour pressures than the smaller ring-cleavage products, resulting in higher yields for single-substituted aromatics. In addition, Li et al.138 investigated the SOA yield from aromatic oxidation systems and concluded that ortho isomers have the highest SOA yield for similar aerosol concentrations, while para isomers have the lowest SOA yield, showing that SOA yields of aromatic hydrocarbons depends more on the substituent location than substituent length.

Overall, SOA yields from of the oxidation of linear, branched and cyclic alkanes followed the order cyclic > linear > branched33. In addition, the position of the methyl group in the branched alkanes and presence of cyclization in the parent alkanes were also found to affect SOA mass yield. SOA yields for aromatic oxidation systems follow the order: ortho > one substituent > meta > three substituent > para, thus showing a significant influence of the alkyl substitute number, substituent location, carbon chain length and branching structure on aromatic hydrocarbon oxidation.

Effects of aerosol seeds

Particle seed characteristics also play an important role in SOA formation. Kleindienst et al.134 (toluene, p-xylene, and 1,3,5-TMB/NOx), Edney et al.19 (toluene/ NOx) and Cocker et al.142 found that (NH4)2SO4 background seed had no significant effect on SOA formation in aromatic photooxidation systems.

However, Kroll et al.149 and Lu et al.137 observed that SOA yield was enhanced in the presence of (NH4)2SO4 seeds with RH at 4–7 and 56%. The enhanced SOA yield from the photooxidation of aromatic hydrocarbons (e.g., toluene and n-xylene) might be attributed to particle phase reactions, which form products with high-molecular weight and low volatility149. Lu et al.137 investigated the effect of inorganic seeds on SOA formation in irradiated m-xylene/ NOx photooxidation systems and proposed that high concentrations of the seed aerosol and thin organic layers are the major factors in particle-phase heterogeneous reactions. The observation that the presence of (NH4)2SO4 seed particles impact SOA yields, where CaSO4 and Ca(NO3)2 particles have no effect, is indicative that the acidic ammonium ion maybe playing a role in the heterogeneous reactions. In another study25, several m-xylene and toluene photooxidation experiments under different NOx conditions showed that seed acidity has no effect upon SOA yields (Fig. 8).

Fig. 8: SOA growth curves for toluene photooxidation in the presence of neutral seed versus acidic seed.
figure 8

Solid and hollow circles represent nonacidic and acidic seed condition (taken from Ng et al.25).

Kamens et al.143 showed that experiments with higher initial seed concentrations (and particle phase water) generated more toluene-derived SOA than the lower seed experiments. In addition, Zhou et al.144 reported that increasing initial seed concentrations resulted in higher SOA yield in the o-, p-xylene, and toluene SOA system. Liu et al.150 also investigated the SOA formation from toluene in the absence of NOx on initially wet or dry (NH4)2SO4 seed particles. An increase in OH exposure from 0.47 × 1011 to 5.28 × 1011 molecules cm−3 s resulted in a decrease in the ratio of SOA mass yield on wet (NH4)2SO4 seeds to that on dry (NH4)2SO4 seeds from 1.31 to 1.01. Results showed a relatively higher SOA yield and higher oxidation state of SOA formed with initially wet (NH4)2SO4 seeds including SOA composition being dominated by earlier generation products containing carbonyl groups at low OH exposure and by later-generation products containing acidic groups at high OH exposures. In another study, experiments were conducted to assess the effect of seed particles on SOA formation from 2-methoxyphenol (guaiacol) by Liu et al.151. With (NH4)2SO4 and NaCl seed particles, SOA yield was enhanced by about 23 and 30%, respectively, which further increased to about 30–53% in the presence of SO2, suggesting that SO2 and seed particles make a synergistic contribution to SOA formation.

These findings suggest that the presence of seed aerosols may or may not affect the SOA yield, depending upon several factors such as RH, initial seed concentration, acidity etc.

Effects of presence of gaseous pollutants (low/high NOx conditions)

Another important parameter is the level of NOx; many studies have been conducted to define the influence of NOx on the SOA yield25,80,135,140,152,153,154,155. The presence of high NOx was found to suppress O3 formation and aerosol formation occurred mainly through reactions with the OH radical in toluene/NOx photooxidation152. Further, the low-NOx experiments promoted O3 and NO3 formation, they can further react with toluene oxidation products, subsequently increasing aerosol formation141. In another study, Klotz et al.153 found that the phenol yield decreased at high NOx concentrations (>100 ppb) while ring-opening products are favoured during benzene oxidation, suggesting significant differences in aerosol formation between low- and high-NOx experiments.

Ng et al.25 showed that the SOA yields from the oxidation of aromatics under low-NOx is substantially higher than those under high-NOx. This is likely due to competition between RO2 + NO (high-NOx) and RO2 + HO2 (low-NOx) reactions, suggesting the importance of RO2 chemistry in SOA formation from BTX. Under high-NOx, the RO radical is produced from the reaction of RO2 + NO which undergoes fragmentation to form higher volatility species, whereas under low-NOx, the carbon chain is preserved, resulting in higher yields when compared with those observed for high-NOx. Similar results were also noticed for the toluene and o-, m- and p-xylene isomers oxidation systems80,154. Chan et al.140 also reported the formation of a higher fraction of ring-retaining products under low-NOx conditions from the oxidation of PAHs. The SOA was composed of semi-volatile products under high-NOx and nonvolatile under low-NOx. In a recent study, Yang et al.155 showed that the SOA yield increases notably with larger NOx concentrations under low-NOx conditions during TMB photooxidation, while an opposite trend is observed in high-NOx experiments. The increase in SOA yield in the low-NOx regime was attributed to the increase of NOx-induced OH concentrations; the formation of low-volatility species might be suppressed, thereby leading to a lower SOA yield in high-NOx conditions.

For linear alkanes with >6 carbon atoms, isomerization of RO is favoured over fragmentation reactions, or reaction with O2, subsequently resulting in the formation of lower volatility compounds than fragmentation products22. Loza et al.23 observed similar SOA yields under low- and high-NOx conditions for the larger alkanes (C12), 2-methylundecane, hexylcyclohexane, cyclododecane, suggesting that yields of SOA from alkane oxidation do not display a systematic NOx dependence. Recently, the roles of NOx were also investigated by generating individual RO via alkyl nitrite photolysis and measuring their product distributions156. Results show that lower-NO/NO2 ratios promote the formation of secondary OH, peroxyacyl nitrates, and highly oxidized products that result in more SOA production.

The presence of SO2 has been shown to promote SOA formation in cyclohexane, 2-methothoxyphenol, and TMB oxidation system151,155,157. As discussed before (section 3.4), high SO2 concentrations may lead to the formation of OS115,116,117.

Overall, studies so far have shown that high levels of NOx may be important factors that lead to lower SOA yields compared to those obtained under low-NOx conditions for aromatic oxidation systems. However, NOx dependence for alkane oxidation systems is still not clear. Thus, more studies are needed to investigate the effect of gaseous pollutants in the oxidation system that could deepen the understanding of SOA formation in relatively complex polluted environments.

Effects of OA concentrations, VOC mixtures and TiO2

The dependence of SOA on OA mass concentration was studied by varying the amounts of VOCs such as m-xylene and p-xylene158. As VOC concentrations increased, the measured OA yield also increased. The OA mass formed in the PAM chamber was directly proportional to the VOC amount during the direct photooxidation of VOCs without seed particles or POA. This is likely due to an increase in the partitioning into the condensed phase which could enhance OA mass concentration130,159.

SOA yields can also be profoundly affected by oxidation of VOC mixtures. McFiggans et al.160 demonstrated that highly reactive compounds (e.g., isoprene) can scavenge OH radical, inhibiting their reaction with other VOCs, and the resulting RO2 can also scavenge highly oxygenated products. These processes can reduce the yield of low-volatility products that may suppress both particle number and mass of SOA.

The effect of photocatalytic materials on SOA formation is still not clear. TiO2 suppressed SOA formation from the photooxidation of m-xylene161. Yields of SOA were 0.003–0.04, whereas negligible SOA was formed when TiO2 was added. This is likely to be due to the decreased concentrations of reactive carbonyl compounds because of TiO2 addition. The suppression effect was also influenced by the presence of seed particles and NOx. Reaction mechanisms for the photocatalysis of aromatics should be considered in future under varying experimental conditions.

Overall, results suggest a positive influence of OA mass concentration on SOA formation, while the presence of photocatalytic material and VOC mixtures shows a negative effect.

Effects of chamber walls and OH exposure

SOA yields are highly sensitive to experimental conditions, as well as corrections of the data for chamber wall losses30,162,163,164,165. Investigations of gas-wall partitioning on SOA yield, that included the oxidation of C8–C16 n-alkanes, 1-alkenes, C8–C13 2-alcohols and 2-ketones in Teflon chambers indicate that losses of OC to the walls can be large for SVOCs and could result in significant underestimation of SOA yields162. The possibility of further oxidation or particle-phase oligomer formation is reduced by the loss of gas-phase OC to the walls, thereby directly reducing the SOA yield. Experiments studying the toluene-NOx oxidation system demonstrated the underestimation of SOA formation by factors up to 4 due to losses of SOA-forming vapours to chamber walls163. In addition, gas-phase wall losses for both types of chambers (continuous flow and batch) under similar conditions showed that the continuous flow experiments mitigate some effects of gas-phase wall losses with long (>2 days) experiment run times165. These studies show that gas-phase wall losses still affect the absolute SOA yield for both types of chamber and warrant more investigation.

Another important parameter observed during the experiments is OH exposure. Studies have shown a decrease in SOA yield for aromatics/alkanes oxidation systems subsequent to an increase in OH exposure. This is possibly due to further oxidation of gas-phase species and breaking of carbon–carbon bonds or heterogeneous OH oxidation reactions24,132,166.

Following the above discussion, the observational evidence suggests a significant effect of gas phase wall loss on SOA mass yields, showing a suppression in SOA formation due to loss of SOA -forming vapour. In addition, experimental variables such as OH exposure may have a negative effect on SOA yield.

Perspective on factors affecting SOA yields

As discussed above many parameters affect aerosol yield. Generally, factors like RH, UV intensity, and initial OA mass concentration seem to have a positive influence on SOA formation while temperature, NOx, OH exposure, the presence of photocatalytic material and VOC mixture shows a negative effect. SOA yield was also found to be highly sensitive to the design of chamber walls, molecular structure of the individual precursor and the presence of seed aerosols. While the role of these parameters is relatively well understood for alkanes and the common mono/poly aromatics-OH oxidation system (using single precursor approach), most of the current challenges and uncertainties, by far, concern their role in real atmospheric conditions where SOA precursors are highly complex mixtures with different volatilities including the presence of other oxidants (e.g., Cl and NO3). It is thus urgent to assess the effect of these parameters under the real atmospheric conditions by photo-oxidising dilutions from whole major emission sources, such as from vehicular exhaust or biomass burning (and not solely a specific compound from these sources).

SOA estimation from different anthropogenic sources

In the previous section, we have discussed alkane and aromatic compound oxidation systems, including their reaction pathways, SOA forming potential (yield), and major reaction products. We briefly present here SOA formation from specific anthropogenic sources such as vehicular exhaust (diesel/gasoline), residential wood burning, cooking etc. Diesel exhaust is an important source of OA in the urban environment, primarily composed of light aromatics, PAHs and long-chain n-alkanes167,168,169. The gas-phase emissions from diesel exhaust are dominated by light aromatics, accounting for 27% of the SOA formed. The SOA contribution from PAHs can reach up to 54% from the oxidation of diesel emissions, representing a potentially large source of urban SOA140. Intermediate volatility organic compounds (IVOCs) are found to be a major component of diesel fuel and mainly exist as vapour in the atmosphere170. Results also suggest some of the first generation products of IVOC (diesel fuel) oxidation are vapours that can undergo multiple generations of oxidation. IVOCs also contribute similarly to the total SOA budget as PAHs, but higher (~4) than single-ring aromatics27,171. The findings also show twice as much SOA formation from diesel emissions after the inclusion of IVOCs as an SOA precursor. Therefore, IVOCs are likely an important source of SOA mass in the atmosphere. In addition, photooxidation of IVOCs also illustrates that the IVOCs are more likely to form oxygenated OA under high NOx conditions, contrary to the oxidation of traditional anthropogenic and biogenic SOA precursors for which reduced aerosol formation is observed at high NOx172. For gasoline exhaust, the emissions of PAHs compared to light aromatics are low, hence, their contribution to SOA formation is almost insignificant173. In another study, the ageing of gasoline exhaust was performed using a mobile environmental reaction chamber174. Results illustrated that a modern gasoline car can lead to the formation of significantly higher SOA than the emitted primary aerosol mass. Du et al.175 reported higher SOA production from gasoline exhaust, showing a greater contribution of IVOCs and SVOCs to the SOA formation. Similarly, another experimental study with gasoline exhaust also showed the formation of higher amounts of IVOCs and SVOC, which promoted the GPP and SOA formation176. It has also been shown that with the increase of aromatic content in gasoline exhaust from 23 to 50%, an enhancement was reported in the SOA yield by a factor of 4.0–6.7176. Residential wood combustion can contribute considerably to the atmospheric OA/SOA burden, particularly in regions with cooler climates, and contribute up to ~80% of the total observed SOA177. Cooking may also be a potential source of SOA because of the abundant emissions of non-methane organic gases178. It has also been shown that the alkenals, which are not traditional SOA precursors accounted for 5 − 34% of the observed SOA179. In addition, SOA formation from gas-phase emissions of heated oils were also investigated178, showing the major SOA precursors were linked to the content of monounsaturated fat and omega-6 fatty acids present in cooking oils. In another study, SOA formation from heated oils also suggested that the major precursor was aldehyde, which accounted for a majority of the oxidation products from cooking emissions180. Additionally, there is a growing concern with SOA formation from non-combustion anthropogenic sources (e.g., coatings, adhesives, cleaning agents, printing inks, pesticides, and personal care products), which may contribute a substantial amount of VOCs to the atmosphere and subsequently results in SOA production181. However, many of these chemicals are currently not included in the emission inventories and require better quantification.

Factors affecting SOA properties

The composition and properties (i.e., optical properties, oxidation state, and phase state) of SOA are affected by RH, temperature and SO2, which has significant implications for regional haze formation and global radiative forcing. Therefore, we discuss in further detail these SOA properties and their implications for air quality and climate.

Optical properties

The optical properties of SOA and their effects on the global radiative balance remains poorly understood. Refractive index (RI) is one of the most important parameters when describing aerosol optical properties as it accounts for other optical factors, such as scattering and extinction efficiencies. RIs measured of seeded aerosols are a little lower than those under non-seeded conditions in low-NOx environments, but are higher for high-NOx environments56. This was the case most noticeably for dodecane derived SOAs when comparing SOA derived from other long-chain alkanes. Measured RI values generally ranged from 1.33 to 1.57 (i.e., 1.33–1.54 for C12, 1.40–1.57 for C15 and 1.41–1.53 for C17-derived SOAs). Long-chain alkanes, a major contributor to IVOC, may contribute largely to the optical properties of SOAs. Multiphase processes can also affect the RI and light-extinction of m-xylene-derived SOA. They can significantly enhance the RI due to oligomerisation processes that results in an increase of the light-scattering efficiency and direct radiative forcing by ~20–90%182.

The optical properties of m-xylene-derived SOA are affected by GPP. As organic mass increases and the particles grow, the proportions of SVOCs and IVOCs in the aerosol phase change resulting in a decrease of SOA RI by 0.09 ± 0.02 without seeds and 0.15 ± 0.02 with seeds183. Pre-existing seeds could promote condensation and could inhibit the formation of some high-molecular-weight products, like highly oxygenated multifunctional compounds (HOMs), leading to a further decrease of the RI. In addition, the optical properties of SOA particles are related to the total organic mass and molecular composition. Similarly, the optical properties of n-dodecane SOA showed that low temperatures could enhance the real part of the RI at wavelengths of 532 and 375 nm due to the formation of large amounts of oligomers184.

Experiments investigating the effect of different experimental conditions (i.e., dry, dry with SO2, wet and wet with SO2) on the composition and optical properties of toluene-derived SOA showed that an increase in the RH enhances both light absorption and the scattering property of the SOA, possibly due to the formation of oligomers through multiphase reactions185. The addition of SO2 resulted in a decrease in the real part of the RI of toluene-derived SOA, probably due to the partitioning of low-oxidation products. The extinction coefficients of toluene-derived SOA with SO2 increased by ~30% under wet conditions compared to dry conditions.

Carbon oxidation state

The average carbon oxidation state (OSC) is widely used to describe the degree of oxidation of atmospheric organic species186. Briefly, the OSC is defined as the charge a carbon atom would acquire if it loses all electrons with more electronegative atoms but gains all electrons with less electronegative atoms. Liu et al.151 reported that SO2 was found to favour an increase of the OSC of SOA formed from methoxyphenol oxidation, indicating that a functionalisation reaction is dominant over oligomerisation and leads to the formation of highly oxidised products with relatively low volatility, which consequently result in an increase in OSC and SOA yields. The formation and ageing of SOA from the photooxidation of toluene and other small aromatics were also studied187. Higher exposure to OH resulted in a different OA composition with high OSC and SOA mass yield and reduced SOA volatility. The OSC generally increased during photooxidation ranging from 0.16–0.29 for final oxidation products187. The cloud condensation nuclei (CCN) potential of aerosols is a function of their hygroscopicity188. Hildebrandt Ruiz et al.187 reported no clear correlation between hygroscopicity and O:C or between hygroscopicity and OSC in the toluene photooxidation system, contrary to the conventional view that suggests oxidative ageing of aerosol has a positive effect on hygroscopicity189. This is possible either due to the presence of surface-active compounds or if the OA is mostly composed of compounds with similar OSC but different molecular size (e.g., oligomers), as the size of molecules affects their solubility and hence their hygroscopicity. It has also been reported that OSC is a better indicator of aerosol oxidation than O:C as the latter may have some influence from non-oxidative processes (i.e., hydration and dehydration) while OSC increases continually with oxidation186,187. In another study, experiments investigating the relationship between the oxidation level and hygroscopic properties of SOA particles confirmed that the hygroscopic growth varied linearly with O:C, but the CCN activity mostly followed a nonlinear trend190.

Physical phase state

The physical phase state (e.g., solid, semisolid or liquid) of SOA particles can significantly affect a number of atmospheric processes, such as gaseous uptake, particle-phase rate of reaction, GPP equilibrium and ageing191,192,193. For example, the diffusion of molecules within the bulk of a viscous particle is comparatively slow and may become the rate-limiting step. The water uptake of highly viscous SOA particles may not occur at low temperatures, which has a significant influence on particle size and scattering properties193,194. The RH-dependent viscosities at room temperature of SOA particles produced from toluene photooxidation indicate that the toluene-derived SOA particles are in the liquid phase at RH > 60%195. Saukko et al.192 investigated the phase state of laboratory-generated SOA produced in a flow tube reactor by photooxidation of anthropogenic precursors, naphthalene and n-heptadecane. Their measurements of the bounced fraction (BF) suggests that SOA particles are in an amorphous solid or semisolid state at RH < 60%. SOA particles generated from naphthalene showed a decrease in the BF from 0.65 ± 0.1 at RH < 50% to 0.4 ± 0.1 at RH = 64 %, suggesting a decrease in viscosity possibly due to an RH-induced phase change from solid to liquid-like particles. Unlike the naphthalene SOA system, n-heptadecane SOA has a low BF indicating liquid-like, less viscous particles at low oxidation levels181. In addition, the bounce behaviour of the SOA systems did not always show a correlation with the particle oxidation state192.

Koop et al.193 have suggested that the oligomerisation and polymerisation processes should be given greater consideration as they will affect the physical state of organic aerosols beyond the decrease in vapour pressure. Oligomerisation may result in a strong increase in viscosity and the formation of semisolid state aerosols, which has implications for heterogeneous oxidation reactions and the extended atmospheric lifetime of SOA particles193. Several other studies have also shown that laboratory-produced SOA particles have an amorphous solid state191,196,197,198. However, recent studies are predominantly based on biogenic precursors suggesting that more studies using anthropogenic precursors are required to better evaluate the change in SOA properties and their impact on the urban environment.

Flow reactors vs smog chambers: effect on precursor oxidation and yields

Flow reactors (e.g., PAM) and smog chambers have both been used extensively to study SOA formation for a variety of atmospherically relevant conditions and emission sources. However, large-scale smog chambers age pollutant mixtures over the course of hours to days, whereas flow reactors rapidly age the mixtures in seconds to minutes. They do this by using high oxidant concentrations to achieve higher degrees of oxygenation, which may affect the atmospheric relevance of the generated products. In addition, the ability of flow reactors to simulate slow atmospheric processes, like condensed phase reactions, is also of concern. The chemical composition and yields of the oxidation products formed from the photooxidation of α-pinene and of wood-combustion emissions were generally consistent in both PAM and smog chamber199. However, the yields during the PAM experiments for the α-pinene system were slightly lower than those in the smog chamber due to increasing wall losses of gas-phase species as PAM has a higher surface area to volume ratio. Similarly, an intercomparison study of SOA generated from gas-phase precursors in a PAM and several environmental chambers showed a similar SOA composition across all systems, whether low OH concentrations are used over long exposure times (chamber) or high OH concentrations over short exposure times (PAM)200. The use of acidic seed particles in the flow reactor increased the yield of SOA by a factor of 3–5, where seed particles are routinely used in the chambers. The study also highlights that both, flow reactors and chambers, are capable of characterising SOA ageing mechanisms representative of ambient conditions. Despite concerns, flow reactors are likely to remain within the study of atmospheric ageing as they produce a high, consistent and reproducible output, are mobile and age aerosols over short timescales, all very convenient attributes for laboratory studies. More intercomparison studies with varying atmospheric conditions for both types of reactors should be considered when evaluating their use within aerosol aging experiments, as commercially available PAM chambers have gained rapid popularity due to their ease of use within field and laboratory experiments.

Gaps in SOA contribution from lab, field and modelling studies

This section focuses on areas of uncertainty in the understanding of SOA formation from anthropogenic precursors/sources.

Laboratory studies

Most of our knowledge about SOA formation from anthropogenic VOCs (i.e., alkanes, aromatics, alkenes) derives from laboratory studies. These studies are designed to mimic atmospheric conditions using large simulation chambers (also known as smog chambers) or aerosol flow reactors. In general, the potentials of compounds to form SOA are measured using experiments with a single precursor approach130,201. However, in reality, SOA precursors occur within packets of air as highly complex mixtures of VOCs and SVOCs with different volatilities202. As a result, using a single-species approach for studying SOA formation is not atmospherically representative, especially in the urban environment. Some studies have been performed to avoid issues with the single-species approach by photo-oxidising whole vehicular exhaust or biomass burning emissions170,177,203,204,205,206,207,208. Tiitta et al.205 and Bruns et al.177 investigated the SOA formation from fuel‐wood‐burning emissions and wiregrass fires and found pyrogenic VOCs (e.g., benzene, naphthalene, and toluene) and furans were significant precursors for SOA production. This is in contrast to the findings of Ahern et al.203, who investigated SOA formation in smoke from the open burning of various biomass fuels representative of canopy and grassland fires and observed significant OA mass enhancement from fuels representing canopy fires containing significant amounts of volatile precursors, such as monoterpenes. Their experiments were conducted using freshly harvested branches of coniferous canopy fuels, a species known as monoterpene emitters209,210. Similarly, the OH-initiated ageing of biomass burning emissions was also investigated204, showing the formed SOA is found to be highly variable and the observed variability may be related to the initial total non-methane organic vapour concentration and OH exposure. In addition, experiments investigating the ageing of emissions from flaming and smouldering-dominated wood fires showed single-ring and polycyclic aromatics are major precursors in flaming emissions, while furans are important SOA precursors in smouldering fires206. Results also suggest that the oxidation products from lignin pyrolysis are likely to dominate SOA formation, which does not depend on the combustion or ageing conditions.

Like biomass burning emissions, photooxidation of low volatility organics found in motor vehicle emissions have been investigated170,207,208. Experiments with light-duty gasoline vehicle exhaust suggest that the catalyst warm-up is an important factor in controlling SOA precursor emissions207. In addition, the majority (<50%) of measured SOA from diesel emissions can be explained by traditional SOA precursors while the rest (~30 %) may originate from the non-methane organic gas emissions208. These observations provide a comprehensive overview on SOA formation considering all VOCs present, which is not possible using the single precursor approach. More studies are needed to get detailed insight on SOA formation mechanisms from key VOC precursors to improve the development of next-generation SOA models and emissions control strategies.

Additionally, laboratory studies have identified linear, branched, and cyclic alkanes with carbon numbers ranging from C7 to C25 as possible SOA precursors while monoaromatics (i.e., benzene, toluene, xylene, TMB) are the most common aromatic precursors as discussed in the previous sections. Cyclic compounds are abundant in diesel exhaust, comprising of ~30% monocyclic and 50% polycyclic alkanes24,211 Industrial and combustion emissions are also likely to be other sources of polycyclic species. However, SOA formation from polycyclic alkanes has not been explored much, including the effect of multiple, fused rings on SOA formation. PAHs, an important precursor for SOA formation in an urban environment, have not been investigated extensively, except in a few cases91,92,212.

Field studies

There has been an improvement in the quantity and quality of ambient measurements that are helpful in evaluating both models and emission inventories. The offline measurement of SOA tracers in the field relies on suitable marker compounds being identified from laboratory experiments, however, the lack of identified ASOA tracers limits the understanding of ASOA abundance and variability.

A number of advanced instruments such as PILS-WSOC (Particle in Liquid Sampler—Water Soluble Organic Carbon)213 AMS/ACSM (Aerosol chemical speciation monitor)39,214, TAG-AMS (thermal desorption aerosol gas chromatograph-AMS)215, EESI-TOF-MS (extractive electrospray ionization-time of flight-MS)216,217,218,219, PTR-MS (proton-transfer reaction-MS) or CIMS (chemical ionization MS)220, are available to quantify OM at a high time resolution (i.e., minutes or seconds). These techniques have been used in many field studies and are able to separate SOA from POA. The online techniques often reduce the likely errors that are introduced by filter-based measurements as the latter are performed over several hours to days, making it challenging to differentiate atmospheric processes. However, the deployment of these instruments for long periods is labour intensive and expensive. In addition, the identification of SOA from different sources may not be always possible as the mass fragments from the ionisation of molecules in mass spectrometry do not often represent a signature for a particular precursor or source. However, this is not always the case as instruments like CIMS or EESI-TOF-MS with good mass-resolution use softer ionization methods. Therefore, at present, the combination of available methods during field campaigns is highly recommended for a complete overview.

Modelling studies

There have been several improvements made in the parameterizations used for atmospheric modelling since the last decade, such as revisions in SOA yields of different precursors based on chamber studies under varying conditions and the use of a new framework (volatility basis set ‘VBS’ approach) to handle the complexities of SOA formation, reaction, and volatility in order to understand the thermodynamics of mixed inorganic-organic systems4. SOA formation has been represented using two approaches in the atmospheric models; empirical models (based on laboratory data)221,222,223, and explicit/semi-explicit models (e.g., Master Chemical Mechanism)224,225. The heterogeneous reactions involving neutral to highly acidic aerosols are not very well accounted for in the existing SOA models. As a result, the current models mostly underestimate SOA mass in comparison with what is typically observed7,201. This could be also due to some missing SOA precursors, mostly anthropogenic, such as IVOCs and PAHs27. A modelling study, which attempted to parametrize SOA formation from alkanes and PAHs, showed 20–30% of total SOA produced from these precursors226. SVOCs or IVOCs are also not included in current emissions inventories and could be an important SOA precursor. In addition, the oxidation of existing POA or oxidation of semi-volatile SOA in the gas phase may enhance SOA formation from anthropogenic hydrocarbons. Bruns et al.177 characterized primary and aged emissions from residential wood burning and showed that traditional SOA precursors account for only ~3–27% of the observed SOA, but ~84–116% of the SOA can be explained by including non-traditional precursors in models.

Overall, this discussion suggests that atmospheric models still require experimental data to parameterise SOA formation from different precursors, followed by evaluation using field data.

Integrated studies-laboratory/modelling/field

Despite rigorous research, several issues are emerging which can only be addressed via an integrated approach combining laboratory, field and modelling studies. For example, Xu et al.227 combined all three approaches to investigate sources of OA and better constrain the OA budget. Their lab-in-the-field experiments reported that formed SOA from biogenic VOCs (α-pinene and β-caryophyllene) can be used as a proxy to apportion low oxygenated OA in the Southeastern US. In addition, their model also reproduces the magnitude and diurnal variability of low oxygenated OA measured at multiple sites in the Southeastern US. Their results also highlight that an integrated approach can facilitate the study of SOA formation under realistic atmospheric conditions, and improve atmospheric models which can be used on a large spatial scale to assess the sources of OA/SOA. Lamkaddam et al.228 designed a study to investigate the aqueous-phase processing of isoprene oxidation products in cloud droplets using a similar approach. They integrate their experimental findings into a global model and reported that clouds significantly increase the amount of SOA. Similarly, Bruns et al.177. showed an improvement in the total SOA budget by including non-traditional precursors identified using laboratory experiments. Their findings suggest that identifying the main precursors responsible for SOA formation from residential wood burning enables improved model parameterizations for SOA contribution. However, such integrated studies are few and require more attention, especially those focusing on AVOCs.

Closing thoughts

Significant advances have been made in the understanding of the formation and properties of ASOAs, but substantial gaps remain that present challenges for accurate characterization of the role of ASOAs in environmental degradation and climate. A better understanding of SOA chemical and physical properties is required to improve the modelling of global OA budgets. Thus, a major goal is to investigate/improve simulation of missing ASOA in atmospheric models through laboratory and field studies.

As outlined, in most of the studies reactions with OH are the dominant initial transformation pathway for AVOCs in the troposphere, although the importance of Cl reactions in the Arctic regions, marine and coastal areas have become apparent due to the rapid rate of reaction with SOA precursors. Additionally, reactions with NO3 at night may represent a significant source of SOA and BrC formation in the atmosphere. Reaction pathways linked to Cl/NO3 initiated oxidation of different AVOCs have not received as much attention and probably should be considered as a newly emerging focus of future research.

Another important line of research is the role of autoxidation reactions. The rapid autoxidation of RO2 formed during VOC oxidation results in HOMs that efficiently form SOA. However, most previous work is based on biogenic precursors (e.g., terpenes and sesquiterpenes)229,230,231. Therefore, although some knowledge has been developed for biogenic precursors, the following key questions remain:

What is the role of autoxidation in SOA formation from AVOCs?

How will it affect the SOA formation in urban environments where AVOC precursors dominate?

As the oxidation of aromatic hydrocarbons can rapidly form HOMs with extremely low volatility, it makes aromatic hydrocarbons a potential contributor to nucleation events observed in the urban environment232,233. There is a possibility that the study of these phenomena in highly pollutant environments may aid in discovering some new pathways for SOA formation chemistry.

Photochemical ageing of aerosols results in a change to aerosol composition and physical properties through various chemical and physical processes234. For example, UV-induced photodegradation can reduce the average size, mass and volatility distribution of SOA compounds resulting from photoinduced fragmentation and serve as a source of oxygenated VOCs235,236. This will cause SOA to scatter sunlight less efficiently, which in turn affects the Earth’s climate. Alternatively, certain photochemical processes occurring on the surfaces of aerosol particles can also increase the average size and complexity of particulate organics237. Condensed-phase photochemical processes may produce a gaseous species comparable to its primary sources and could therefore measurably affect the budgets of both particulate and gaseous atmospheric organic compounds on a global scale. These processes are not fully characterised or accurately included in the global models. For example, Hodzic et al.238 showed that the inclusion of condensed-phase photolysis in a global model can potentially result in ~50% of particle mass loss after 10 days of ageing with an assumption that condensed-phase photolysis of organics occurs at the same rate as in the gas-phase. Another recent combined experimental and modelling study also predicted about a 50% reduction in biogenic SOA due to SOA photodegradation239. This suggests the study of photochemical processes in the highly polluted environment through field campaigns, or of photochemical processes linked to AVOCs oxidation in the laboratory chambers can potentially aid in closing the SOA budget.

Finally, we conclude that our knowledge on SOA formation from AVOCs precursors is far from being complete. More integrated studies (laboratory/modelling/field) are needed to improve atmospheric models to capture the magnitude and variability of SOA to assess the impact on aerosol climate effects. Laboratory studies should include mixtures of precursors to simulate atmospheric conditions as closely as possible.