Roles of pH, ionic strength, and sulfate in the aqueous nitrate-mediated photooxidation of green leaf volatiles

Biotic and abiotic stresses can lead to terrestrial green plants releasing green leaf volatiles (GLVs), which can partition into atmospheric aqueous phases where they can undergo oxidation to form aqueous secondary organic aerosol (aqSOA). Anthropogenic emission changes have resulted in nitrate becoming an increasingly important component of atmospheric aqueous phases, which has significant implications for aqSOA formation since nitrate photolysis produces oxidants. Nevertheless, sulfate remains the main inorganic aqueous component in most regions, and thus it controls the pH and ionic strength of atmospheric aqueous phases. We report results from laboratory investigations of the effects of pH, ionic strength, and sulfate on the reaction kinetics and aqSOA formation of the aqueous nitrate-mediated photooxidation of four GLVs: cis-3-hexen-1-ol, trans-2-hexen-1-ol, trans-2-penten-1-ol, and 2-methyl-3-buten-2-ol. Our results showed that the aqueous reaction medium conditions, i.e., dilute cloud/fog vs. concentrated aqueous aerosol conditions, governed the effects that pH, ionic strength, and sulfate have on the GLV degradation rates and aqSOA mass yields. Most notably, reactions initiated by sulfate photolysis will have significant effects on the GLV degradation rates and aqSOA mass yields in aqueous aerosols but not in cloud/fog droplets. In addition to providing new insights into aqSOA formation from the aqueous reactions of GLVs in regions with substantial concentrations of nitrate in cloud, fog, and aqueous aerosols, this study highlights how nitrate and sulfate photochemistries can couple together to influence the reactions of water-soluble organic compounds and their aqSOA formation in aqueous aerosols, which have implications for our evaluations of aqueous organic aerosol lifetimes and composition.

1 Introduction

Biogenic volatile organic compounds (BVOCs) contribute more than 80 % of the global volatile organic compound (VOC) emissions (Guenther et al., 2012; Sindelarova et al., 2014). Isoprene and monoterpenes comprise more than half of the total annual BVOC emissions (Sindelarova et al., 2014). Since green leaf volatiles (GLVs) comprise a comparatively small fraction of total BVOCs, their chemical processes have received far less attention compared to isoprene and monoterpene. GLVs are C₅ to C₆ unsaturated organic compounds with aldehyde, alcohol, or ester functional groups (Sarang et al., 2021a). GLVs are emitted during the decomposition of C₁₈ polyunsaturated fatty acids in leaves when vegetation is exposed to herbivores, pathogens, or harsh weather conditions (Ameye et al., 2018; Matsui and Engelberth, 2022; Silva et al., 2021). They are also emitted by cyanobacteria and algae during bloom events (García-Plazaola et al., 2017). GLVs have the potential to contribute substantially to the local secondary organic aerosol (SOA) budget due to their increased emissions when vegetation is subjected to biotic and abiotic stresses. A previous study reported that GLV emissions from Amazon tropical forests increased significantly in the afternoon due to the plants' response to rising temperatures, whereas isoprene and monoterpene emissions decreased (Jardine et al., 2015). Global GLV emissions will potentially increase in the future due to climate change and the increasing use of new fumigation-based agricultural, horticultural, and forestry practices (Cofer et al., 2018; Matsui and Engelberth, 2022; Su et al., 2020). Thus, GLVs may play increasingly important roles in atmospheric chemistry, which necessitates increasing our knowledge of their multiphase reactions and SOA formation.

GLVs can be oxidized by ozone and hydroxyl radicals (•OH) in the gas phase to produce low-volatility products, with reported SOA mass yields ranging from 0.7 % to 20 % (Hamilton et al., 2009; Harvey et al., 2014; Jaoui et al., 2012; Mentel et al., 2013). Due to their moderately high water solubilities, GLVs can dissolve into atmospheric aqueous phases (e.g., aqueous aerosols, cloud and fog droplets), where they can be oxidized by aqueous oxidants such as •OH, sulfate anion radicals (SO₄•⁻), nitrate radicals (NO₃•), triplet organic excited states (³C^*), and singlet oxygen (¹O${}_{\mathrm{2}}^{*}$) (Richards-Henderson et al., 2014, 2015; Sarang et al., 2021a, b, 2023). Higher quantities of low-volatility products are formed from aqueous reactions compared to gas-phase reactions, with previous studies reporting aqueous SOA (aqSOA) mass yields as high as 88 %, though this will depend on the GLV and the aqueous oxidant (Richards-Henderson et al., 2014, 2015). However, these previous studies were mostly conducted under dilute aqueous conditions mimicking aqueous cloud/fog droplets. Differences in the physicochemical properties of the aqueous reaction medium will impact reaction rates in cloud/fog droplets vs. aqueous aerosols (Herrmann et al., 2015), but little is currently known about the aqueous oxidation of GLVs under more concentrated aqueous-aerosol-like conditions.

The liquid water concentrations (LWCs) of cloud and fog droplets typically fall in the range of 0.05 to 0.5 g m⁻³ (Achtert et al., 2020; Kim et al., 2022; Korolev et al., 2007), whereas the LWCs of aqueous aerosols fall in the range of 10⁻⁷ to 10⁻³ g m⁻³ (Herrmann et al., 2015). Thus, the concentrations of dissolved organic and inorganic compounds are higher in aqueous aerosols, with their dry masses close to the liquid water mass (Nguyen et al., 2016). The concentrations of inorganic ions, particularly nitrate and sulfate, primarily govern the acidities and ionic strengths of atmospheric aqueous phases. Cloud and fog droplets (pH 2 to 7) are generally less acidic than aqueous aerosols (pH 0 to 5) due to the more frequent ammonia dissolution and higher buffering capacities of cloud and fog droplets (Pye et al., 2020; Tilgner et al., 2021). The ionic strengths of atmospheric aqueous phases span a large range (10⁻⁵ to 10⁰ M), with the ionic strengths of aqueous aerosols being several orders of magnitude higher than those of cloud and fog droplets (Herrmann et al., 2015; Tilgner et al., 2021). Under the high-ionic-strength conditions in aqueous aerosols, substantial ion association occurs, which will affect the activity coefficients of organic compounds, resulting in reactions occurring under non-ideal conditions (Herrmann, 2003). Previous studies have reported that ionic strength significantly affects the aqueous reactions of some organic compounds and subsequent product formation (Herrmann, 2003; Mekic et al., 2018a, b; Mekic and Gligorovski, 2021; Zhou et al., 2019). In addition to contributing to the acidity and ionic strength of atmospheric aqueous phases, inorganic nitrate and sulfate can undergo photolysis to produce various reactive species that react with organic compounds. The tropospheric irradiation of nitrate in atmospheric aqueous phases is known to produce aqueous reactive species such as •OH, nitric oxide radicals (NO•), and nitrogen dioxide radicals (NO₂•) that can react with organic compounds (Mack and Bolton, 1999; Gen et al., 2022). Even though a recent study reported that sulfur-containing radicals (e.g., SO₄•⁻) are formed during the tropospheric irradiation of aqueous sulfate aerosols (Cope et al., 2022), the mechanisms behind their formation are still not well understood. While nitrate is increasingly important in regions with large reductions in sulfur dioxide emissions and/or with high ammonia emissions (Heald et al., 2012; Schaap et al., 2004; West et al., 1999), sulfate remains the dominant inorganic constituent of atmospheric aqueous phases in most regions (Bianco et al., 2020). At present, little is known about how inorganic nitrate and sulfate salts influence the aqueous oxidation of GLVs.

In this study, we investigated the nitrate-mediated photooxidation of four GLVs, cis-3-hexen-1-ol, trans-2-hexen-1-ol, trans-2-penten-1-ol, and 2-methyl-3-buten-2-ol (Fig. 1), under cloud-/fog-like and aqueous-aerosol-like conditions. These four GLVs, which are amongst some of the more abundant GLVs, have Henry's law constants between 60 and 120 M atm⁻¹ (Sander, 2023; Sarang et al., 2021a). Thus, they can partition efficiently into cloud/fog droplets and moderately into aqueous aerosols (Fig. S1 in the Supplement). We investigated the effects of pH, ionic strength, and sulfate have on the reaction kinetics and aqSOA mass yields under cloud-/fog-like and aqueous-aerosol-like conditions. Results from this study provide new insights into the aqueous photooxidation of GLVs in regions with substantial levels of nitrate in cloud and fog droplets and aqueous aerosols and, more generally, into how sulfate photochemistry impacts the aqueous nitrate-mediated photooxidation of other water-soluble organic compounds.

Figure 1 The four model GLVs used in this study.

2 Methods

2.1 Chemicals and solutions

All the chemicals were used as received. cis-3-Hexen-1-ol (cHxO, 98 %), trans-2-hexen-1-ol (tHxO, 96 %), 2-methyl-3-buten-2-ol (MBO, 97 %), benzoic acid (BA, 99.5 %) and p-hydroxybenzoic acid (pHBA, 99 %) were purchased from J&K Scientific. Trans-2-penten-1-ol (tPtO, ≥95 %) was purchased from Aladdin Scientific. Ammonium nitrate (NH₄NO₃, ≥95 %) and ammonium sulfate ((NH₄)₂SO₄, 99+ %) were purchased from Fisher Scientific. Sulfuric acid (H₂SO₄, 95 %) was purchased from VWR Chemicals BDH^©. Milli-Q ultrapure water (18.2 MΩ cm) was used to prepare all solutions.

Table 1 shows the concentrations of solutions used to simulate cloud-/fog-like and aqueous-aerosol-like conditions in photochemistry experiments. The same nitrate $/$ GLV molar concentration ratio (i.e., 2.5:1) was used for both cloud-/fog-like and aqueous-aerosol-like conditions. The concentrations of the GLVs and NH₄NO₃ were set to be 100 times higher for the aqueous-aerosol-like conditions than the cloud-/fog-like conditions. The nitrate concentration (i.e., 250 µM) used for cloud-/fog-like conditions is close to those measured in cloud water in Hong Kong SAR (average of 238 µM) (Li et al., 2020) and is within the global range for nitrate in continental cloud water (Bianco et al., 2020). Even though the nitrate concentration in aqueous aerosols could reach molar levels (Bzdek et al., 2020), we used 25 mM of NH₄NO₃ in experiments that simulate aerosol-like conditions to maintain the same nitrate $/$ GLV molar concentration ratio as in experiments that simulate cloud-/fog-like conditions. The pH levels of unbuffered solutions (no addition of H₂SO₄) were close to 5, and 5 was selected as the higher bound to study the pH effects on the nitrate-mediated photooxidation of GLVs. The lower bound of pH was set to 3 through the addition of H₂SO₄. (NH₄)₂SO₄ was added into the solutions to control the sulfate concentrations in and ionic strengths of the solutions. Since the acid dissociation constant (pK_a) for HSO${}_{\mathrm{4}}^{-}$ ↔ H⁺ + SO${}_{\mathrm{4}}^{\mathrm{2}-}$ is around 2.0 at 25 °C (Dickson et al., 1990), SO${}_{\mathrm{4}}^{\mathrm{2}-}$ is expected to be the dominant species even at the lower bound of pH 3. The pH levels (i.e., pH 3 vs. pH 5, Table 1) used in this study fall within the ranges for cloud and fog droplets and aqueous aerosols (Herrmann et al., 2015; Pye et al., 2020; Tilgner et al., 2021). The ionic strengths (I_total) of the solutions were calculated using the following equation:

$$\begin{array}{}\text{(1)}& I_{\mathrm{total}}={\displaystyle \frac{\mathrm{1}}{\mathrm{2}}}\times \sum _{i=\mathrm{1}}^{i=n}{c}_i{z}_i^{\mathrm{2}},\end{array}$$

where c_i and z_i are the concentration (M) and charge, respectively, of inorganic ion i for H⁺, NH${}_{\mathrm{4}}^{+}$, NO${}_{\mathrm{3}}^{-}$, and SO${}_{\mathrm{4}}^{\mathrm{2}-}$. The ionic strengths used in this study, i.e., 0.002 M vs. 0.02 M for cloud-/fog-like conditions and 0.5 M vs. 3.3 M for aqueous-aerosol-like conditions (Table 1), fall within the ranges for ionic strengths for clouds/fog droplets and continental aerosols (Herrmann et al., 2015). Only the addition of tPtO and tHxO had significant effects on the molar absorptivity of NH₄NO₃ (Fig. S2b), enhancing the peak molar absorptivity of NH₄NO₃ approximately 1.3- and 1.5-fold, respectively.

Table 1 Concentrations of GLVs, NH₄NO₃, H₂SO₄, (NH₄)₂SO₄, the pH, and I_total of solutions used to simulate cloud-/fog-like and aqueous-aerosol-like conditions in photochemistry experiments.

2.2 Photochemistry experiments

The solutions were placed into open cylindrical quartz tubes (1.2 cm inner diameter), which were placed on a rotating sample holder located in the middle of a photoreactor (Rayonet RPR-200, Southern New England UV Co.) surrounded by UVB lamps (RPR-3000Å, Southern New England UV Co.). The photon flux in the photoreactor ranged from 250 to 400 nm and peaked at 311 nm (Fig. S2a). The temperature inside the photoreactor during experiments was maintained at around 30 °C by a cooling fan located at the bottom of the photoreactor. The volume of each quartz tube was around 15 mL. The quartz tubes were filled with 10 and 1 mL of solutions during experiments simulating cloud-/fog-like and aqueous-aerosol-like conditions, respectively. Aliquots of 1 and 0.1 mL were extracted from the illuminated solutions at different reaction times for offline chemical analysis during experiments simulating cloud-/fog-like and aqueous-aerosol-like conditions, respectively. For experiments simulating aqueous-aerosol-like conditions, the extracted volume (0.1 mL) was diluted with Milli-Q ultrapure water by a factor of 10 prior to chemical analysis to ensure that the measured signals stayed within the linear detection range of the detector. The decays of the GLVs were measured using an ultrahigh-performance liquid chromatograph coupled to a photodiode array detector (UPLC-PDA, H-class, Waters). A Kinetex Polar C18 column (2.6 µm, 100×2.1 mm) equipped with a security guard and Polar C18 pre-column was used for the measurement of the four GLVs. An isocratic elution program set to 0.3 mL min⁻¹ was used. The mobile phases of water and acetonitrile were run at a ratio of 80:20 for cHxO, tHxO, and MBO and at a ratio of 85:15 for tPtO. The detection wavelengths were set to 240 nm for MBO and 210 nm for the other three GLVs. All the experiments and measurements were performed three times. All the decays of the GLVs followed apparent first-order reaction kinetics reasonably well (Figs. S3 and S4); thus they were fitted with the following equation:

$$\begin{array}{}\text{(2)}& \ln \left({\displaystyle \frac{[\mathrm{GLV}{]}_t}{[\mathrm{GLV}{]}_{\mathrm{0}}}}\right)=-k_{\mathrm{obs}}t,\end{array}$$

where k_obs is the pseudo-first-order rate obtained from the exponential fit to the photodegradation of the GLV and [GLV]_t and [GLV]₀ are the concentrations of individual GLV measured by UPLC-PDA at illumination times t and 0, respectively.

No loss in GLVs was observed in dark control experiments conducted in the absence and presence of nitrate and sulfate. During illumination in control experiments conducted in the absence of nitrate and sulfate (“light only” experiments), only MBO had significant loss under cloud-/fog-like conditions, whereas all four GLVs had significant losses under aqueous-aerosol-like conditions (Figs. S3 and S4). The four GLVs were not expected to undergo direct photolysis as they do not absorb light significantly at wavelengths larger than 280 nm (Richards-Henderson et al., 2014; Sarang et al., 2021a), as demonstrated in Figs. S5 and S6. Thus, the observed losses under illumination in control experiments conducted in the absence of nitrate and sulfate could be due to evaporation, with MBO having the largest losses due to its higher vapor pressure ($\mathrm{3.08}\times {\mathrm{10}}^{-\mathrm{2}}$ atm) compared to the other three GLVs (cHxO: $\mathrm{1.23}\times {\mathrm{10}}^{-\mathrm{3}}$ atm; tHxO: $\mathrm{1.20}\times {\mathrm{10}}^{-\mathrm{3}}$ atm; tPtO: $\mathrm{3.46}\times {\mathrm{10}}^{-\mathrm{3}}$ atm) based on estimations using EPI Suite^TM (US EPA, 2024). The k_obs values measured for the GLV decays in nitrate-mediated photooxidation experiments were subsequently corrected by subtracting the loss rates from control experiments conducted in the absence of nitrate and sulfate (“light only” experiments).

To gain insights into how concentration, pH, and ionic strength affect •OH formation during nitrate-mediated photooxidation, a separate set of experiments (i.e., GLVs were not present in the solutions) using BA (10 µM) as the •OH probe compound were conducted to estimate the steady-state concentrations of •OH ([•OH]_ss) using the same methodology as our past studies (Lyu et al., 2023; Yang et al., 2021, 2023). p-HBA, which is formed from the reaction of •OH with BA ($k_{\mathrm{BA}+\mathrm{OH}}=\mathrm{5.9}\times {\mathrm{10}}^{\mathrm{9}}$ M⁻¹ s⁻¹ (Herrmann et al., 2010)) at a yield of 0.17 (Anastasio and McGregor, 2001), was measured in these experiments using an ultrahigh-performance liquid chromatography system (1290 system, Agilent) coupled to a high-resolution quadrupole time-of-flight mass spectrometer (X500R QTOF MS/MS, Sciex) (UPLC-MS) equipped with an electrospray ionization (ESI) source that was operated in negative mode (Sect. S1 in the Supplement). Solid-phase extraction (SPE) using two different types of SPE cartridges (Oasis MAX, 60 mg, 3 mL, 60 µm, Waters; Bond PPL Elut, 200 mg, 3 mL, 125 µm, Agilent) was performed to remove inorganic salts from samples before UPLC-MS analysis. The estimated [•OH]_ss values for the cloud-/fog-like and aqueous-aerosol-like conditions are shown in Fig. S7.

2.3 Mass yields of aqSOA

Aerosol mass spectrometry was used to measure the aqSOA mass yields of the four GLVs. Aliquots of 10 and 1 mL were extracted from the illuminated solutions at one GLV lifetime (i.e., $\mathit{\tau }=\frac{\mathrm{1}}{k_{\mathrm{obs}}}$, when 37 % of the initial concentration of the GLV remained) in experiments simulating cloud-/fog-like and aqueous-aerosol-like conditions, respectively (Tables S2 and S3). The time points equivalent to one GLV lifetime were determined from the aforementioned kinetic experiments. For experiments simulating aqueous-aerosol-like conditions, the extracted volume (1 mL) was diluted with Milli-Q ultrapure water by a factor of 300 prior to aerosol mass spectrometry analysis to ensure that the measured signals stayed within the linear detection range of detector. Each sample solution was injected at 10 mL h⁻¹ by a syringe pump (model 100, KD Scientific) into an aerosol generation system (model 9200, Brechtel Manufacturing Incorporated), which aerosolized the solution. The atomizer system used nitrogen gas (99.999 % purity) as the carrier gas at a flow rate of 4.5 L min⁻¹. The aerosols generated were passed through an inline dryer before entering a time-of-flight aerosol chemical speciation monitor (ACSM, Aerodyne Research Inc.). All the experiments and measurements were performed three times.

The aqSOA mass yield (Y_SOA) was calculated using the following equation (Jiang et al., 2023; Ma et al., 2021):

$$\begin{array}{}\text{(3)}& \begin{array}{rl}& Y_{\mathrm{SOA}}={\displaystyle \frac{\mathrm{organic}\mathrm{mass}\mathrm{increased}}{\mathrm{GLV}\mathrm{mass}\mathrm{decreased}}}={\displaystyle \frac{[\mathrm{Org}{]}_{\mathit{\tau }}-[\mathrm{Org}{]}_{\mathrm{0}}}{[\mathrm{GLV}{]}_{\mathrm{0}}-[\mathrm{GLV}{]}_{\mathit{\tau }}}}\\ & ={\displaystyle \frac{[\mathrm{Org}{]}_{\mathrm{ACSM},\mathit{\tau }}\times \frac{{\left[{\mathrm{SO}}_{\mathrm{4}}^{\mathrm{2}-}\right]}_{\mathit{\tau }}}{{\left[{\mathrm{SO}}_{\mathrm{4}}^{\mathrm{2}-}\right]}_{\mathrm{ACSM},\mathit{\tau }}}-[\mathrm{Org}{]}_{\mathrm{ACSM},\mathrm{0}}\times \frac{{\left[{\mathrm{SO}}_{\mathrm{4}}^{\mathrm{2}-}\right]}_{\mathrm{0}}}{{\left[{\mathrm{SO}}_{\mathrm{4}}^{\mathrm{2}-}\right]}_{\mathrm{ACSM},\mathrm{0}}}}{(\mathrm{1}-\mathrm{0.37})\times [\mathrm{GLV}{]}_{\mathrm{0}}}}\end{array},\end{array}$$

where [Org], [GLV], and [SO${}_{\mathrm{4}}^{\mathrm{2}-}$] are the concentrations of organics, the GLV of interest, and sulfate, respectively. The terms with ACSM as a subscript indicate the mass concentrations (in mg L⁻¹) of aerosols measured by the ACSM, which is different from the molar concentrations (mol L⁻¹) of solutions which do not have the ACSM subscripts. The subscripts τ and 0 indicate the sample solutions obtained at one GLV lifetime (Tables S2 and S3) and before illumination, respectively. The concentrations of sulfate in the solutions before and after illumination were assumed to be the same (i.e., [SO${}_{\mathrm{4}}^{\mathrm{2}-}$]_τ = [SO${}_{\mathrm{4}}^{\mathrm{2}-}$]₀). The concentration of sulfate in each sample solution (Table 1) was used as the internal standard to scale the concentrations of organics measured by ACSM to those of the solutions. This is because sulfate is non-refractory and is expected to be collected and quantified by the ACSM, which had a capture vaporizer with a collection efficiency of 1 (Daellenbach et al., 2016; Xu et al., 2018; Joo et al., 2021). Additionally, the sulfate and organic composition in the atomized aerosols are expected to be internally mixed together (Ma et al., 2021).

3 Results and discussion

3.1 Cloud-/fog-like conditions

3.1.1 Reaction kinetics

The concentrations of the four GLVs decreased upon irradiation in the presence of nitrate and sulfate. In contrast to MBO, no decays were observed for cHxO, tHxO, and tPtO when only sulfate was present (but not nitrate) in the solutions. In the absence of nitrate, the MBO decay rates obtained in the absence and presence of sulfate were similar. The MBO decay in the absence of nitrate could be due to MBO evaporation since it has a higher vapor pressure than the other three GLVs. Overall, these results indicated that sulfate has an insignificant effect on the kinetics of the four GLVs under cloud-/fog-like conditions.

Figure 2 shows the k_obs values for the four GLVs upon irradiation in the presence of nitrate at different pH levels (i.e., 3 vs. 5) and ionic strengths (i.e., 0.002 M vs. 0.02 M). Separate experiments performed in the absence of GLVs and using BA as the •OH probe compound showed that the estimated [•OH]_ss values decreased with pH under these cloud-/fog-like conditions (Fig. S7a), consistent with results reported by Lyu et al. (2023). The k_obs values for the four GLVs were on the orders of 10⁻⁵ to 10⁻⁴ s⁻¹ for the four GLVs. The decays in the four GLVs upon irradiation in the presence of nitrate were due to the reactions of the GLVs with reactive species produced from nitrate photolysis such as •OH, NO•, and NO₂• (Table S1 in the Supplement). Even though approximately equal quantities of •OH and NO₂• are produced during nitrate photolysis (Chen et al., 2019; Zhang et al., 2021), the typical reactivities of NO₂• are 2 to 5 orders of magnitude lower than those of •OH (Ford et al., 2002; Chen et al., 2019; Zhang et al., 2021). Other reactive species produced during nitrate photolysis (e.g., hydroperoxide radicals (HO₂•) and superoxide ions (O${}_{\mathrm{2}}^{-}\mathrm{•}$)) are also expected to have lower reactivities compared to •OH (Bielski et al., 1985; Mack and Bolton, 1999). Reaction with •OH was also shown to be the main contributor to the reaction kinetics of other non-photolyzable organic compounds (e.g., formic acid, glycolic acid) during aqueous nitrate-mediated photooxidation (Lyu et al., 2023). While it is possible that sulfur-containing radicals and other reactive species were formed from the photolysis of (NH₄)₂SO₄ (Table S4), their effects on k_obs are small due to their low concentrations under diluted cloud-/fog-like conditions (Cope et al., 2022). Additionally, the k_obs values measured under illumination in control experiments conducted in the presence of sulfate only were not statistically different (p>0.05) from the k_obs values measured under illumination in control experiments conducted in the absence of nitrate and sulfate (“light only” experiments). Thus, the decays of the GLVs were likely governed mostly by their reactions with •OH, though minor contributions from their reactions with reactive species other than •OH cannot be discounted.

Figure 2 The k_obs values for the four GLVs during nitrate-mediated photooxidation under cloud-/fog-like conditions (Table 1). The error bars represent 1 standard deviation originating from triplicate experiments and measurements. Statistical analyses (Student's t test) on the differences in the k_obs values are presented in Tables S6 to S9.

Under the same ionic-strength conditions, the four GLVs had higher k_obs at pH 3 than at pH 5, though the pH-dependent trends for cHxO at I_total=0.002 M and 0.02 M were statistically insignificant (p>0.05) (Table S5). The four GLVs do not have acidic H atoms; thus they do not undergo acid dissociation to form different relative abundances of deprotonated and neutral forms with different reactivities at different pH levels. Additionally, Richards-Henderson et al. (2014) showed that the •OH rate constants for many GLVs do not depend on pH. Hence, the pH-dependent k_obs trends seen in Fig. 2 were due to the pH-dependent formation of •OH (Fig. S7) and other reactive species from nitrate photolysis. HNO₂, whose production from nitrate photolysis is favored over NO${}_{\mathrm{2}}^{-}$ production at pH ≤ 3.5 (Marussi and Vione, 2021), has a higher quantum yield for •OH formation than NO${}_{\mathrm{2}}^{-}$ in the near-UV region (Arakaki et al., 1999; Marussi and Vione, 2021). Thus, the formation rates and concentrations of •OH produced at pH 3 are higher than at pH 5, which would explain the higher k_obs at pH 3. There were no statistically significant differences in the k_obs values for I_total of 0.002 M vs. 0.02 M under the same pH conditions for the four GLVs (p>0.05). This indicated that ionic strength (and sulfate) has an insignificant effect on the reaction kinetics of the four GLVs under cloud-/fog-like conditions. Overall, only pH impacted the reaction kinetics of the four GLVs significantly under cloud-/fog-like conditions.

3.1.2 Mass yields of aqSOA

Figure 3 shows the Y_SOA values for the four GLVs measured at one GLV lifetime during irradiation in the presence of nitrate at different pH levels (i.e., 3 vs. 5) and ionic strength (i.e., 0.002 M vs. 0.02 M). The measured Y_SOA values (0 % to 53 %) are in line with the range of Y_SOA values (10 % to 88 %) measured by Richards-Henderson et al. (2014) for five GLVs (including cHxO and MBO) at one GLV lifetime in their reactions with •OH generated from H₂O₂ photolysis under cloud-/fog-like conditions.

Figure 3 Y_SOA values for the four GLVs at one GLV lifetime during nitrate-mediated photooxidation under cloud-/fog-like conditions. The error bars represent 1 standard deviation originating from triplicate experiments and measurements and include errors propagated from the standard deviations of the sulfate concentrations measured by ACSM. The ACSM-measured organic signals for cHxO at pH 5 and I_total=0.02 M and for MBO at pH 3 and 5 at I_total=0.02 M and at pH 5 at I_total=0.002 M were very low, resulting in close to zero organic concentrations and Y_SOA values. Therefore, there was essentially no formation of low-volatility products from these three experiments, and the Y_SOA values were essentially zero. Statistical analyses (Student's t test) on the differences in the Y_SOA values are presented in Tables S10 to S13.

Given the high reactivity of •OH, reactions of the GLVs with •OH are expected to produce products that contribute substantially to aqSOA formation. However, we cannot discount the possibility that other reactive species produced from nitrate/sulfate photolysis (Tables S1 and S4) also contributed to aqSOA formation. The reaction of •OH with the GLV is expected to occur either by •OH addition to the C=C bonds to form hydroxy alkyl radicals or by H abstraction from the C–H or O–H bonds to form alkyl radicals (Figs. S8 and S9). The H-bond dissociation energies at the CH₂, CH₃, and OH groups are around 393, 419, and 436 kJ mol⁻¹, respectively (Benson, 1976). H abstraction is expected to occur preferentially at C–H sites on the carbon α to the OH group (Cooper et al., 2009; Sarang et al., 2023). The hydroxy alkyl radicals and alkyl radicals subsequently react with O₂ to form peroxy radicals (RO₂•), which then react with other RO₂• to form either higher-molecular-weight carbonyls and alcohols or alkoxy radicals (RO•). RO• can undergo fragmentation reactions to form lower-molecular-weight compounds. Even though they are not shown in our proposed reaction mechanisms in Figs. S8 and S9, bimolecular combination reactions involving RO₂• and RO• (e.g., RO₂• + RO₂•) that form oligomers could have also contributed to aqSOA formation.

Attempts to identify prominent low-volatility products (and their formation pathways) that contributed to aqSOA using liquid chromatography–mass spectrometry were unsuccessful due to the presence of large quantities of inorganic salts in the samples, which negatively impacted the ionization efficiencies of the products. Nevertheless, low-volatility products from both the •OH addition and the H-abstraction channels likely contributed to aqSOA formation for the four GLVs. Sarang et al. (2023) previously detected products formed from both channels in their study of the aqueous •OH oxidation of various GLVs. In the case of cHxO, the dominant products from the H-abstraction channel were reported to be at least 15 kcal mol⁻¹ more stable than the products from the •OH-addition channel (Sarang et al., 2023). Subsequent density functional theory calculations indicated that both channels were important contributors to product formation due to the barrierless pathway in the •OH-addition channel and the formation of thermodynamically stable allylic alkyl radicals in the H-abstraction channel (Sarang et al., 2023). Allylic alkyl radicals from the H-abstraction channel similarly play important roles in product formation in the •OH oxidation of large unsaturated organic compounds (Nah et al., 2014).

In contrast to cHxO, tHxO, and tPtO, the Y_SOA values for MBO were not statistically different (p>0.05) from 0 %. The substantial differences in the Y_SOA values for MBO vs. cHxO, tHxO, and tPtO could be attributed to the molecular structure of MBO. MBO contains a terminal C=C bond that is adjacent to its OH group, whereas cHxO, tHxO, and tPtO contain non-terminal C=C bonds that are non-adjacent to their OH groups (Fig. 1). Due to the molecular structure of MBO, the formation of RO• with oxygen radical centers adjacent to at least one OH functional group is enhanced for both the •OH-addition and H-abstraction channels (Fig. S9). Their close proximity to oxygenated functional groups increases the susceptibility of these RO• to fragmentation (Atkinson, 1997), which forms lower-molecular-weight compounds that may volatilize into the gas phase. Thus, the enhanced formation of RO• that preferentially fragments into higher-volatility products during the reaction of MBO with •OH under cloud-/fog-like conditions would explain its low Y_SOA values. In contrast, RO• formation (and, thus, fragmentation) is not enhanced in the •OH reactions of cHxO, tHxO, and tPtO due to the formation of primary and secondary RO₂• formed from the •OH-addition and H-abstraction channels.

Under the same ionic-strength conditions, the Y_SOA values for the four GLVs were generally higher at pH 3 than at pH 5, though these pH-dependent trends were not statistically significant (p>0.05) in some instances (Tables S10 to S13). It is possible that the enhanced aqSOA formation at lower pH was due to acid-catalyzed reactions. The aqueous reaction of •OH with the four GLVs likely forms various higher- and lower-molecular-weight carbonyls (Figs. S8 and S9) (Sarang et al., 2021a, 2023). Some of these carbonyls could have undergone acid-catalyzed reactions (e.g., hydration, polymerization, aldol condensation) to form low-volatility products (Ervens et al., 2011; Maben and Ziemann, 2023; Presberg et al., 2024; Cooke et al., 2024). Additionally, it is possible that NO• enhanced the formation of low-volatility products that contributed to aqSOA at lower pH, possibly through the formation of low-volatility organonitrates via the RO₂• + NO• → RONO₂ pathway (Atkinson and Arey, 2003). This is because NO• formation from nitrate photolysis would be enhanced at pH 3 (Table S1). HNO₂ is favored over its conjugated base NO${}_{\mathrm{2}}^{-}$ at pH < 3.3 (Marussi and Vione, 2021). The production of NO• from the photolysis of HNO₂ (R15 in Table S1) is nearly 1 order of magnitude faster than its formation from NO${}_{\mathrm{2}}^{-}$ photolysis (R10 in Table S1).

Interestingly, with the exception of cHxO at pH 3, the Y_SOA values for the four GLVs were higher at I_total=0.002 M than at I_total=0.02 M under the same pH conditions, though these ionic-strength-dependent trends were not statistically significant (p>0.05) in some instances (Tables S10 to S13). This is in contrast to the insignificant effect that ionic strength had on k_obs (Fig. 2). (NH₄)₂SO₄ was used to control the ionic strengths of the solutions (Table 1). Thus, our results indicated that even though ionic strength and/or sulfate concentration had insignificant effects on the reaction kinetics under cloud-/fog-like conditions, they could significantly affect the formation of low-volatility products. The lower Y_SOA values measured at higher I_total and sulfate concentrations under the same pH conditions implied that fragmentation reactions that form volatile lower-molecular-weight products were enhanced at higher I_total and/or sulfate concentrations. Additionally, the higher-I_total conditions could have enhanced the partitioning of products into the gas phase due to the salting-out effect (Peng and Wan, 1998).

3.2 Aqueous-aerosol-like conditions

3.2.1 Reaction kinetics

Figure 4 shows the k_obs values for the four GLVs upon irradiation in the presence of nitrate at different pH levels (i.e., 3 vs. 5) and ionic strengths (i.e., 0.5 M vs. 3.3 M). The concentrations of the GLVs and NH₄NO₃ used in this set of experiments to simulate aqueous-aerosol-like conditions were both 100 times higher than those used to simulate cloud-/fog-like conditions while maintaining the same nitrate $/$ GLV molar concentration ratio of 2.5:1 (Table 1). Separate experiments performed in the absence of GLVs and using BA as the •OH probe compound showed that the estimated [•OH]_ss values decreased with pH under these cloud-/fog-like conditions (Fig. S7b), consistent with results for the cloud-/fog-like conditions (Fig. S7a). Unsurprisingly, the [•OH]_ss values for aqueous-aerosol-like conditions were higher than those for the cloud-/fog-like conditions due to the higher concentrations of NH₄NO₃ used to simulate aqueous-aerosol-like conditions. The k_obs values for the four GLVs were on the orders of 10⁻⁶ to 10⁻³ s⁻¹.

Figure 4 The k_obs values for the four GLVs during nitrate-mediated photooxidation under aqueous-aerosol-like conditions (Table 1). The error bars represent 1 standard deviation originating from triplicate experiments and measurements. Statistical analyses (Student's t test) on the differences in the k_obs values are presented in Tables S14 to S17.

The k_obs values measured for cHxO, tHxO, and tPtO under aqueous-aerosol-like conditions were factors of 1.1 to 6.1 lower than those measured for cloud-/fog-like conditions (Fig. 2). The lower k_obs values measured for these three GLVs under aqueous-aerosol-like conditions could be due, in part, to the exponential decrease in the nitrate photolysis rate with increasing nitrate concentration (Ye et al., 2017). Consequently, the production of reactive species does not increase linearly with the nitrate concentration. Additionally, sulfur-containing radicals and reactive species produced from sulfate photolysis are expected to contribute significantly to the degradation of the GLVs under aqueous-aerosol-like conditions due to the high concentration of (NH₄)₂SO₄ in the solutions (Cope et al., 2022). Work by Sarang et al. (2021b) suggests that the rate constants for the reactions of GLVs with SO₄•⁻ are about 1 order of magnitude lower than those of their reactions with •OH (Table S5). Nonetheless, the exponential decrease in the nitrate photolysis rate with increasing nitrate concentration does not completely explain the other noticeable differences in the k_obs results for aqueous-aerosol-like vs. cloud-/fog-like conditions. Firstly, the k_obs values for MBO under aqueous-aerosol-like conditions were factors of 1.2 to 5.1 higher than those under cloud-/fog-like conditions. Reasons for MBO's higher k_obs values under aqueous-aerosol-like conditions are currently unknown. Secondly, in contrast to the insignificant effect that the ionic strength and/or sulfate concentration had on the reaction kinetics under the same pH conditions for cloud-/fog-like conditions, the k_obs values for the four GLVs were significantly higher at I_total=3.3 M than at I_total=0.5 M (p<0.05) under the same pH conditions for aqueous-aerosol-like conditions.

The higher k_obs values at higher I_total under the same pH conditions for aqueous-aerosol-like conditions could be due to higher •OH reactivities at higher ionic strength (Herrmann, 2003; Mekic and Gligorovski, 2021; Weller et al., 2010). Additionally, since (NH₄)₂SO₄ was used to control the ionic strengths of the solutions, sulfate photolysis likely contributed to the ionic-strength-dependent/sulfate-dependent k_obs trends for the aqueous-aerosol-like conditions (Table S4). Cope et al. (2022) showed that sulfur-containing radicals (e.g., SO₄•⁻) were formed in (NH₄)₂SO₄-containing concentrated solutions and aqueous aerosols when they were irradiated with UVB light or simulated sunlight. Even though the mechanism for the formation of sulfur-containing radicals from aqueous (NH₄)₂SO₄ photolysis remains unknown, the authors showed that the SO₄•⁻ formed could easily react with various organic compounds in aqueous aerosols. The (NH₄)₂SO₄ concentrations used in this study (0.16 and 1.09 M) were substantially lower than the (NH₄)₂SO₄ concentration used by Cope et al. (2022) (3.7 M). Nevertheless, sulfate photolysis occurred under this study's aqueous-aerosol-like conditions since the concentrations of the four GLVs decreased upon irradiation when only sulfate was present in the solutions (Fig. S10), likely driven primarily by their reactions with SO₄•⁻.

Our results clearly show that sulfur-containing radicals produced from sulfate photolysis can participate in aqueous reactions with GLVs in aqueous aerosols but not in cloud and fog droplets. This could be due to the low hydration numbers in aqueous aerosols that would reduce the energy needed to produce sulfur-containing radicals from sulfate photolysis (Cope et al., 2022; Xu et al., 1998). As explained by Cope et al. (2022), a fully solvated SO${}_{\mathrm{4}}^{\mathrm{2}-}$ anion in dilute solutions has about 16 water molecules in its first solvation shell (Plumridge et al., 2000). The energy needed to detach an electron from SO${}_{\mathrm{4}}^{\mathrm{2}-}$(H₂O)_n to produce SO₄•⁻(H₂O)_n decreases with the number of water molecules in its hydration shell, and electron detachment potentially occurs spontaneously at 3 water molecules (Pathak, 2014). Thus, the effective potential barrier for the electron detachment of SO${}_{\mathrm{4}}^{\mathrm{2}-}$(H₂O)₄ under concentrated conditions akin to aqueous aerosols is likely substantially lower compared to that under dilute conditions akin to those of cloud and fog droplets (Yang et al., 2002). Consequently, the likelihood of sulfur-containing radical production in concentrated aqueous aerosols would be substantially higher than that in diluted cloud and fog droplets.

Comparisons of the k_obs values obtained in the presence of sulfate and nitrate (Fig. 4) vs. only sulfate (Fig. S10) indicated that sulfate photolysis had a complex non-additive effect on the GLVs' reaction kinetics. Only approximately half of the k_obs values measured in the presence of sulfate and nitrate were significantly higher (p<0.05) than those measured in the presence of only sulfate under the same pH conditions. The non-additive effect that sulfate photolysis had on the GLVs' reaction kinetics could be due to its mechanism coupling with the nitrate photolysis mechanism. For instance, SO₄•⁻ could react with the NO${}_{\mathrm{3}}^{-}$ anion to form NO₃• and the SO${}_{\mathrm{4}}^{\mathrm{2}-}$ anion (de Semainville et al., 2007). However, since the mechanism for the formation of sulfur-containing radicals from sulfate photolysis remains unknown, we were unable to assess the extent to which the sulfate photolysis mechanism coupled with the nitrate photolysis mechanism under our experimental conditions. Furthermore, contributions of the •OH, SO₄•⁻, and NO₃• reactions to the measured k_obs would require knowledge of both the reaction rate constants and the concentrations of •OH, SO₄•⁻, and NO₃•. While the SO₄•⁻ and NO₃• concentrations in our study are not known, work by Richards-Henderson et al. (2014) and Sarang et al. (2021b) suggests that the rate constants for the reactions of GLVs with •OH, SO₄•⁻, and NO₃• are on the orders of 10⁹ M⁻¹ s⁻¹, 10⁸ to 10⁹ M⁻¹ s⁻¹, and 10⁷ to 10⁸ M⁻¹ s⁻¹, respectively (Table S5). However, the aforementioned reaction rate constants were measured using dilute solutions with low ionic strengths (Sarang et al., 2021b), and it is unclear whether they could be extrapolated to aqueous aerosols which have high ionic strengths (Herrmann et al., 2015).

With the exception of tPtO at I_total=0.5 M, the k_obs values were higher at pH 5 than at pH 3 under the same ionic-strength conditions, though these pH-dependent trends were not statistically significant (p>0.05) in some instances (Tables S14 to S17). The increase in k_obs with pH could potentially be due to the formation of •OH, SO₄•⁻, and NO₃• from the coupled sulfate and nitrate photolysis mechanisms being pH-dependent, though this would require future studies to elucidate the mechanisms. The k_obs values measured in the presence of only sulfate did not have an obvious pH dependence (Fig. S10), which could be due to the four GLVs being pH-insensitive organic compounds. Cope et al. (2022) previously reported that pH had substantial effects on the reactions of SO₄•⁻ with pH-sensitive organic compounds but not on the reactions of SO₄•⁻ with pH-insensitive organic compounds.

3.2.2 Mass yields of aqSOA

Figure 5 shows the Y_SOA values for the four GLVs measured at one GLV lifetime during irradiation in the presence of nitrate at different pH levels (i.e., 3 vs. 5) and ionic strengths (i.e., 0.5 M vs. 3.3 M). The Y_SOA values measured under aqueous-aerosol-like conditions were substantially higher than those measured under cloud-/fog-like conditions (Fig. 3). The enhanced aqSOA formation under aqueous-aerosol-like conditions could be attributed to the higher concentrations of GLVs, which were 100 times higher than those used under cloud-/fog-like conditions. Consequently, the reaction of higher concentrations of GLVs enhanced RO₂• and RO• combination reactions that led to oligomer formation.

Figure 5 Y_SOA values for the four GLVs at one GLV lifetime during nitrate-mediated photooxidation under aqueous-aerosol-like conditions (Table 1). The error bars represent 1 standard deviation originating from triplicate experiments and measurements and include errors propagated from the standard deviations of the sulfate concentrations measured by ACSM. The ACSM-measured organic signals for tPtO at pH 5 at I_total=3.3 M were very low, resulting in close to zero organic concentrations and Y_SOA values. Therefore, there was essentially no formation of low-volatility products from this experiment, and the Y_SOA value was essentially zero. Statistical analyses (Student's t test) on the differences in the Y_SOA values are presented in Tables S18 to S21.

Similarly to the Y_SOA measured under cloud-/fog-like conditions (Fig. 3), the Y_SOA values for the four GLVs generally decreased with increasing pH under the same ionic-strength conditions and with increasing ionic strength and sulfate concentration under the same pH conditions, though these trends were not statistically significant (p>0.05) in some instances (Tables S18 to S21). The enhanced aqSOA formation at lower pH could be due to the formation of low-volatility products from acid-catalyzed reactions (e.g., hydration, polymerization, aldol condensation) (Ervens et al., 2011; Maben and Ziemann, 2023; Presberg et al., 2024; Cooke et al., 2024) and/or the enhanced formation of low-volatility organonitrates via the RO₂• + NO• → RONO₂ pathway (Atkinson and Arey, 2003). Reduced aqSOA formation at higher ionic strength and sulfate concentration was likely due to the enhancement of fragmentation pathways in the reactions of GLVs with sulfur-containing radicals formed from sulfate photolysis. For instance, SO₄•⁻ addition to C=C bonds to form higher-molecular-weight organosulfates is a minor channel compared to fragmentation pathways that form lower-molecular-weight products induced from electron transfer and other reactions by SO₄•⁻ (Ren et al., 2021). The higher concentrations of SO₄•⁻ formed from the photolysis of high concentrations of sulfate (≥1085 M) likely enhanced fragmentation pathways that led to the formation of lower-molecular-weight products. Additionally, the higher-I_total conditions could have enhanced the partitioning of products into the gas phase due to the salting-out effect (Peng and Wan, 1998).

Most noticeably, Y_SOA values as high as 59 % were measured for MBO at I_total=0.5 M under aqueous-aerosol-like conditions, in contrast to the substantially lower Y_SOA values measured under cloud-/fog-like conditions (≤9 %). While this could be due to the enhancement of RO₂• and RO• combination reactions induced by the higher concentrations of MBO in aqueous-aerosol-like conditions, the formation of low-volatility organosulfates induced by reactions involving sulfate could have contributed to the higher Y_SOA values as well. The concentrations of sulfate used to control the ionic strength in aqueous-aerosol-like conditions were up to 1861 times higher than those used in cloud-/fog-like conditions (Table 1). Thus, the reaction of MBO with sulfur-containing radicals formed from sulfate photolysis likely played a significant role in aqSOA formation under aqueous-aerosol-like conditions. Organosulfates (e.g., 2-hydroxy-2-methyl-4-sulfate-3-butanone) were previously identified as products from the reaction of MBO with SO₄•⁻ in the aqueous phase (Ren et al., 2021). Additionally, organosulfates could have been formed by acid-catalyzed reactions between sulfate and a MBO-derived epoxide (e.g., (3,3-dimethyloxiran-2-yl)methanol) formed from the •OH reaction of MBO (Zhang et al., 2012). Nevertheless, increasing the sulfate concentration 7-fold to achieve I_total=3.3 M led to substantial reductions in Y_SOA (9 % and 2 %). This indicated that fragmentation pathways would eventually be enhanced in the reactions of GLVs with sulfur-containing radicals formed from the photolysis of high concentrations of sulfate.

4 Conclusions and implications

We investigated the nitrate-mediated photooxidation of four GLVs in dilute cloud-/fog-like and concentrated aqueous-aerosol-like conditions, focusing on the effects that pH, ionic strength, and sulfate on the reaction kinetics and aqSOA mass yields. Our results showed that the aqueous reaction medium conditions governed the effects that pH, ionic strength, and sulfate had on the reaction kinetics and aqSOA mass yields. Under dilute cloud-/fog-like conditions, the four GLVs had higher k_obs at lower pH, which could be attributed to the pH-dependent formation of •OH and other reactive species from nitrate photolysis. Ionic strength and sulfate had insignificant effects on k_obs. In contrast, under concentrated aqueous-aerosol-like conditions, the four GLVs had higher k_obs at higher pH, as well as higher k_obs values at higher ionic strength and sulfate concentration. Many of these differences could be attributed to sulfur-containing radicals produced from sulfate photolysis participating in the reactions of GLVs under aqueous-aerosol-like conditions but not in cloud-/fog-like conditions. Under cloud-/fog-like conditions where the sulfate concentrations were low, k_obs was governed by the reactions of GLVs with •OH and other reactive species from nitrate photolysis. In contrast, the high sulfate concentrations in the aqueous-aerosol-like conditions enhanced the formation of sulfur-containing radicals from sulfate photolysis, which participated in the reactions of GLVs. Higher Y_SOA values were measured under aqueous-aerosol-like conditions, likely due to enhanced oligomer formation from RO₂• and RO• combination reactions caused by the higher concentrations of GLVs reacted. Despite the different effects that pH, ionic strength, and sulfate had on the reaction kinetics in cloud-/fog-like vs. aqueous-aerosol-like conditions, similar Y_SOA trends were observed for these two reaction conditions. Higher Y_SOA was measured at lower pH, which could be due to the enhanced formation of low-volatility products from acid-catalyzed reactions and/or RO₂• + NO• → RONO₂ reactions. Lower Y_SOA was measured at higher ionic strength and sulfate concentration, which could be attributed to the enhancement of fragmentation pathways in the reactions of GLVs with sulfur-containing radicals formed from sulfate photolysis.

Overall, the results provide new insights into the aqueous photooxidation of GLVs in areas with substantial levels of nitrate in cloud and fog droplets and aqueous aerosols. These insights are expected to be useful in modeling studies of the atmospheric fates of GLVs and their contributions to the SOA budget. These insights build upon those provided by previous studies that were conducted under dilute cloud-/fog-like conditions and in the absence of inorganic salts (Richards-Henderson et al., 2014, 2015; Sarang et al., 2021b, 2023). Results from this study highlight the influences that nitrate and sulfate, the two main inorganic constituents in cloud and fog droplets and aqueous aerosols in most regions, can have on the aqueous photooxidation of GLVs. Additionally, the magnitudes of their influences depend on the aqueous reaction medium (i.e., dilute cloud and fog droplets vs. concentrated aqueous aerosols) in which the reactions occur.

Our study also highlights many questions about the sulfate photolysis mechanism that need to be addressed in future studies. These include the mechanism for the formation of sulfur-containing radicals from aqueous (NH₄)₂SO₄ photolysis and how the sulfate photolysis mechanism can couple with the nitrate photolysis mechanism to affect the formation of reactive species including •OH. With the exception of pH 5 under aqueous-aerosol-like conditions, Fig. S7 shows that the [•OH]_ss generally decreased with increasing sulfate concentration, though the magnitude of the decrease depended on the pH. Additional studies are needed to elucidate how the presence of sulfate will affect the formation of reactive species under different conditions (e.g., pH, ionic strength, aqueous reaction medium). While not investigated in this study due to our inability to completely remove inorganic salts prior to UPLC-MS analysis, we hypothesize that [•OH]_ss will likely similarly decrease with increasing sulfate concentration under concentrated aqueous-aerosol-like conditions. Additionally, even though this study focuses on the aqueous photooxidation of GLVs, it is likely that nitrate, sulfate, and the aqueous reaction medium will influence the aqueous photooxidation of other water-soluble organic compounds as well. More importantly, the manner in which nitrate and sulfate influence the reaction kinetics and aqSOA formation will depend on not only the aqueous reaction medium in which the reactions occur, but also whether the water-soluble organic compound is pH-sensitive or pH-insensitive (Cope et al., 2022; Lyu et al., 2023; Yang et al., 2023). The aforementioned factors need to be considered in future studies on the photooxidation of water-soluble organic compounds in different atmospheric aqueous phases.

There are several caveats that should be noted. First, we were unable to completely distinguish between the effects of sulfate and ionic strength on the aqueous photooxidation of GLVs since H₂SO₄ and (NH₄)₂SO₄ were used to control both the pH and ionic strength of the solutions. Second, the effects of only two ionic-strength conditions, 0.5 and 3.3 M, were investigated in experiments simulating aqueous aerosols. However, ionic strengths in atmospheric aqueous aerosols span a large range and can reach 45 M (Herrmann et al., 2015; Volkamer et al., 2007). Future studies could consider using a chemically inert inorganic salt (e.g., sodium perchlorate (Mekic et al., 2018a)) to control the ionic strength of solutions and investigate reactions in aqueous aerosols with very high ionic strengths. Third, many of our conclusions regarding key reaction pathways were drawn based on Y_SOA measurements performed using an ACSM due to our inability to completely remove inorganic salts from experimental samples before UPLC-MS analysis. Future studies should consider using alternative analytical methods that are not adversely impacted by inorganic salts (e.g., gas chromatography–mass spectrometry (Sarang et al., 2023) and nuclear magnetic resonance (Ren et al., 2021)) to identify prominent products and reaction pathways, though the detection of oligomers is still expected to be analytically challenging.