Guaiacol (99 %), sodium sulfite (Na2SO3, >99 %), 2,2,6,6-tetramethyl-1-piperinedinyloxy (TEMPO, 98 %), ethanol (99 %), and tert-butanol (99 %) were purchased from Macklin. Zero air is made up of 21 % O2 and 79 % N2. All water used in the experiments was ultrapure Milli-Q water (18.2 MΩcm-1).

Bulk aqueous experiment. All experiments were performed in a 25 mL airtight Pyrex tube equipped with a magnetic stir bar and a gas inlet tube for feeding high-purity zero air or nitrogen (∼0.4 Lmin-1) under 370 nm light or Xenon lamp (300w) irradiation. Based on preliminary experiments, we identified the critical role of UVA irradiation in driving the reaction; therefore, a high-power UVA lamp was employed to ensure sufficient photon flux and to obtain reliable kinetic data (see Supplement, Figs. S1–S4). A 20 mL reaction solution containing guaiacol, Na2SO3, and other reactants was prepared. The pH of the reaction solution was adjusted using H2SO4 and NaOH and measured with a pH meter REDOX potentiometer Conductivity meter (AZ-86555) that was calibrated with commercial pH standards. In experiments requiring the measurement of total inorganic sulfur, the pH is adjusted with either phosphoric acid or phosphate. Aliquots (3 mL) were sampled every 20 min for 1 h, with 0.30 mL MeOH added immediately to quench the reaction. Each experiment was repeated at least twice.

HPLC analysis. The concentrations of the guaiacol and phenol were determined using an HPLC (Thermo Scientific™ UltiMate™ 3000) equipped with a diode array detector (DAD) and an Agilent 5 TC-C18 column (150×4.60mm, 5 µm). The column temperature was maintained at 25 °C, and the flow rate was set to 1 mLmin-1. Detection was performed at 274 nm. The mobile phase consisted of 60/40 (v/v) acetonitrile/water acidified with 0.1 % trifluoroacetic acid (TFA).

Direct infusion HRMS. Reaction solutions were filtered through a membrane and then directly introduced into an Agilent 6546 quadrupole time-of-flight mass spectrometer (QTOF-MS, Santa Clara, CA) with electrospray ionization (ESI) source in negative mode. The MS parameters were as follows: nebulizer, 25 psi; gas flow, 10 Lmin-1; sheath gas temperature, 330 °C; capillary voltage, 3500 V; sheath gas flow, 12 Lmin-1. MS data were collected in an m/z range of 90–500. Agilent MassHunter Qualitative Analysis software (version 10.0) was used for data analysis.

UV-vis spectroscopy. An ultraviolet-visible Spectrophotometer (Youke, T2602, Shanghai, China) was used to monitor guaiacol absorbance during reaction with Na2SO3 and to record sample spectra from 200 to 500 nm. The reaction solution was directly loaded without dilution or modification. Spectra of the guaiacol – Na2SO3 reaction were collected every 20 min for 1 h.

IC measurements. Sulfite (SO32-) and sulfate (SO42-) concentrations were analyzed using a Metrohm 883 Basic IC system quipped with a Metrosep A supply 5-250/4.0 analytical column and a conductivity detector (Liu et al., 2025). Prior to analysis, 2 % isopropanol and 1.0 mM NaOH was added into the reaction solution. The eluent used was 3.2 mM Na2CO3/1.0 mM NaHCO3, with a flow rate of 0.8 mLmin-1. For total inorganic sulfur analysis, samples were pre-oxidized to SO42- using hydrogen peroxide (H2O2) before IC detection.

FIDI-MS experiment. The working principle of FIDI-MS (field-induced droplet ionization mass spectrometry) is described as follows (Gong et al., 2022). Droplets approximately 2 mm in diameter (∼4 µL volume) were suspended from the tip of a stainless-steel capillary, which was positioned equidistantly between two parallel plate electrodes separated by 6.3 mm apart. The droplets were formed by injecting the analyte solution through the capillary using a syringe pump. The chemical composition and solute concentrations of the suspended droplets were identical to those of the corresponding bulk stock solutions used for droplet generation. The parallel plates were mounted on a translation stage to align the front electrode's aperture with the atmospheric pressure inlet of a Thermo-Fischer LTQ-XL mass spectrometer (Waltham, MA), which was operated under laboratory ambient air conditions at a relative humidity of approximately 50 %. Once the droplets were formed, a 60 s equilibration period was allowed to enable compound diffusion and achieve equilibrium coverage at the air-water interface.

Sampling of the suspended droplets was accomplished by applying a high-voltage pulse (3–5 kV, 100 ms duration, variable polarity) to the rear electrode and capillary, with half the voltage simultaneously applied to the rear plate, thereby establishing a uniform electric field. This field induced a dipole in the suspended droplets, causing it to elongate and form a double Taylor cone at both ends, which ejected oppositely charged submicron-sized droplets. These negatively charged droplets passed through the aperture of the front plate and entered the mass spectrometer for gas-phase ion detection. Due to the significant disturbance caused by the ionization droplet interface (IDI) sampling, a new droplet was generated for each measurement. In this study, a negative voltage polarity was applied to the rear plate and capillary to facilitate detection of deprotonated guaiacol ions ([GUA]⁻).

HR-ToF-AMS experiment. During photochemical experiments, reaction solutions were aerosolized with a constant output atomizer (TSI Inc.) using N2 as the carrier gas. The resulting aerosols were dried with a diffusion dryer and then introduced into a high-resolution time-of-flight aerosol mass spectrometer (HR-ToF-AMS; Aerodyne Research, Inc.) for chemical characterization. Drying allowed evaporation of volatile and semi-volatile species; therefore, the AMS primarily measured the mass concentration and bulk composition of the remaining low-volatility products. The operating principles of AMS have been described previously (Decarlo et al., 2006; Canagaratna et al., 2007). Briefly, the AMS analyzes non-refractory aerosols that vaporize at ∼600 °C under high vacuum via 70 eV electron impact ionization. In this study, the AMS was operated in “V” ion optical mode (mass resolution ∼3000) to acquire mass spectra up to m/z 422. The AMS data were processed with the standard AMS toolkits SQUIRREL (v1.67) and PIKA (v1.27), available at http://cires.colorado.edu/jimenez-group/ToFAMSResources/ToFSoftware/ (last access: 23 April 2026).

DFT and TDDFT calculations. Geometry optimizations and frequency calculation for all molecular structures (reactants, products, and transition states) were performed using the M06-2X (Zhao and Truhlar, 2007) functional with the ma-TZVP basis set (Zheng et al., 2010), employing the SMD solvation model (Marenich et al., 2009) to simulate aqueous-phase effects in water, as implemented in the Gaussian 16 software package (Frisch et al., 2016). Optimized structures were verified by frequency computations to confirm local minima (zero imaginary frequencies) or transition structures (single imaginary frequency). Intrinsic reaction coordinate (IRC) calculations were performed to ensure that the first-order saddle points found were true transition states (TS) connecting the reactants and the products. Single-point energy calculations and solvation effects were evaluated at the CCSD(T)/aug-cc-pVTZ (Guo et al., 2018; Noga and Bartlett, 1987) level using the SMD solvation model, with geometries optimized at M06-2X/ma-TZVP and zero-point energy (ZPE) correction applied. The calculations were carried out using the ORCA 5.0.3 program package (Neese, 2025). Multiwfn 3.8 (Lu and Chen, 2012) and Shermo 2.4 (Lu and Chen, 2021) were used for further data analysis.

Classical MD calculations. Classical molecular dynamics (MD) calculations were performed using GROMACS 4.5.5 (Hess et al., 2008). In a cubic box with periodic boundary conditions, the system consisted of 1000 SPC/E water molecules and one GUA molecule using the OPLS-AA force field. Electrostatics were treated with the particle-mesh Ewald (PME) method; van der Waals interactions were truncated at 10 Å. A leap-frog integrator was used with a 2 fs timestep, and the trajectories were recorded every 10 steps.

Umbrella Sampling: To determine the average volume for each system, 10 ns simulations were conducted in the NVT ensemble, where the temperature was set to 300 K using the V-rescale method. The potentials of mean force (PMF) were calculated using the Weighted Histogram Analysis Method (WHAM) calculations, which were performed in one additional 10 ns simulation with initial configurations from the preceding simulations. The GUA moved in the z-dimension around their frozen positions under a harmonic restoring force. The force constant was set at 1×103 (kJmol-1nm-1), and configurations were recorded every 0.5 ps. Visualization and trajectory analysis were implemented using VMD (Humphrey et al., 1996).

Box model conditions. Based on the empirically determined apparent photooxidation rate constants of S(IV) under UVA irradiation, the apparent photon efficiency (APE) was calculated (see Supplement). Assuming that the APE remains constant, the apparent rate constants under UVA irradiation corresponding to the AM0 standard solar spectrum (Seinfeld and Pandis, 2016) were then derived (derivation details are provided in the Supplement). Sulfate production rates at 271 K were calculated for different aqueous-phase reaction pathways with O3, H2O2, TMIs, and NO2, following Cheng (Cheng et al., 2016), excluding ionic strength effects.

The Henry's law constants at 271 K for SO2, O3, H2O2, and NO2 are 3.521 Matm-1, 0.025 Matm-1, 1.147×106 Matm-1, and 2.319×10-2 Matm-1, respectively. Equilibrium constants for SO2⋅H2O are KS1=0.025 M and KS2=1.09×10-7 M (Cheng et al., 2016).

Scenario Conditions. “Cloud droplets” scenario: [SO2(g)]=5 ppb, [NO2(g)]=1 ppb, [H2O2(g)]=1 ppb, [O3(g)]=50 ppb, [Fe(III)]=0.3 µM, [Mn(II)]=0.03 µM, liquid water content (LWC)=0.1 gm-3.

“Beijing haze” scenario: [SO2(g)]=40 ppb, [NO2(g)]=66 ppb, [H2O2(g)]=0.01 ppb, [O3(g)]=1 ppb, LWC=300 µgm-3. The concentrations of Fe(III) and Mn(II) were assumed to vary with pH (Cheng et al., 2016).

The sulfate formation rate was calculated using the following equation. [S(IV)](M)=[SO2]g(ppb)×10-9×1+Ks1[H+]+Ks1×Ks2[H+]2×Hso2P[SO42-](µgm-3h-1)=[S(IV)]0×(1-e-kobs(s-1)×3600s)×Vwater×96(gmol-1)×106(µgg-1)

To investigate the photodegradation of GUA in sodium sulfite (Na2SO3) solutions under UVA irradiation, we first examined the photooxidation behavior of Na2SO3 in the UVA region. Na2SO3 solutions with controlled initial pH were prepared and continuously bubbled with zero air (Fig. S5). At pH 4.0 and 0.5 mM Na2SO3, sulfite loss in the dark was slow, with an observed rate constant of 2.27×10-5 s-1 (Fig. S6) (Brandt and van Eldik, 1995). Under UVA irradiation (370 nm), the sulfite loss rate increased nearly tenfold to 2.02×10-4 s-1, with sulfate (SO42-) as the primary product (Fig. S7). Increasing Na2SO3 concentration to 2.0 mM had only a moderate effect, with rate constants averaging (2.41 ± 0.79) × 10-4 s-1 (Figs. S8 and S9). In contrast, pH significantly influenced photooxidation kinetics (Fig. S10): the apparent sulfite decay rate increased by nearly 14-fold from pH 4.0 to 7.0, reflecting shifts in dominant S(IV) species (HSO3- vs. SO32-) with different photochemical reactivities. These findings demonstrate that UVA light substantially enhances S(IV) oxidation in metal-free systems and that the reaction is strongly pH-dependent.

Guaiacol (GUA), a methoxyphenol emitted primarily from biomass burning (4.7 Tgyr-1 globally) (Liu et al., 2022; Li et al., 2023a), was used as a molecular probe to trace reactive intermediates formed during UVA-driven S(IV) oxidation. Given its Henry's law constant (McFall et al., 2020), up to 40 % of atmospheric GUA can partition into the aqueous phase (Fig. S11), making it a relevant proxy for aqueous-phase organic transformations.

At pH 4.0, GUA (0.1 mM) was added to 2.0 mM Na2SO3 solution under continuous zero-air bubbling. GUA remained stable in the dark, with minor losses attributed to evaporation. Under UVA irradiation (370 nm), however, it degraded rapidly following pseudo-first-order kinetics (k≈0.023 min-1; Fig. S12a) that is approximately 14 times faster than direct photolysis (Fig. S12b), highlighting the critical role of S(IV)-derived reactive intermediates. Suppressing O2 via N2 purging significantly reduced GUA degradation (Fig. 1a), confirming the importance of O2-dependent photochemistry induced by UVA.

We further investigated how reagent concentrations influence degradation kinetics. At high Na2SO3 : GUA molar ratios (≥20), GUA degradation followed pseudo-first-order kinetics, with rates increasing linearly with Na2SO3 concentration (Fig. 1b). At lower ratios, deviations from first-order behavior were observed (Fig. S13), suggesting a shift in the limiting reagent or changes in radical propagation dynamics.

Figure S14 presents the kinetics of UVA-irradiated solutions containing 0.1 mM GUA and 0.5 mM Na2SO3. The apparent oxidation rate constant for SO32- was 6.34×10-4 s-1 (Fig. S15), about three times higher than that without GUA (2.02×10-4 s-1) (Fig. S7b), indicating that GUA significantly promoted S(IV) oxidation. The concurrent decrease in total inorganic sulfur closely tracked GUA degradation, suggesting that GUA reacted with photochemically generated intermediates to form S-containing organic species, such as organosulfates (OSs).

High-resolution mass spectrometry (HRMS; m/Δm=5×104) was used to identify reaction products. Negative-mode ESI analysis of a solution containing 0.1 mM GUA and 2.0 mM Na2SO3 at pH 4.0 (Fig. 2a) revealed unreacted GUA (C7H7O2-, m/z=123.0446) along with multiple sulfate ester derivatives: C6H5O5S- (m/z=188.9858), C7H7O5S- (m/z=203.0014), C7H7O6S- (m/z=218.9963), and C7H7O7S- (m/z=234.9912). These signals indicate OSs formation from GUA reacting with SO4⚫- radicals photochemically generated from SO32- and O2 under UVA.

To verify SO4⚫- involvement, we introduced 2,2,6,6-tetramethyl-1-piperinedinyloxy (TEMPO; C9H18NO) as a radical scavenger (Bai et al., 2016). In the Na2SO3+TEMPO system without UVA (Fig. 2b) or after 30 min in the dark (Fig. 2c), only the TEMPO⁺ – SO32- adduct (C9H18NO4S-, m/z=236.0968) was observed. However, under 370 nm irradiation, new peaks appeared at m/z=252.0903 and 220.1019, corresponding to the TEMPO-SO4⚫- adduct (C9H18NO5S-) and its O2-loss fragment (C9H18NO3S-, Fig. 2d), respectively, confirming SO4⚫- generation.

Finally, adding TEMPO to the GUA+Na2SO3+UVA system (Fig. 2e) eliminated all OS peaks, leaving only signals for the TEMPO-SO4⚫- adduct and its fragment. This demonstrates that TEMPO scavenged SO4⚫- and suppressed GUA-derived OS formation, confirming SO4⚫- as the key intermediate driving the observed OSs production.

SO4⚫- can also oxidize water or OH- to form hydroxyl radicals (⚫OH) (Wojnárovits and Takács, 2019), which effectively oxidize GUA in aqueous phase (Yu et al., 2014). To assess the relative contributions of ⚫OH versus SO4⚫-, we used ethanol (EtOH) and tert-butyl alcohol (tBuOH) as radical scavengers: EtOH reacts rapidly with both ⚫OH (1.2×109 M-1s-1) and SO4⚫- (1.6×107 M-1s-1), while tBuOH reacts primarily with ⚫OH (3.8×108 M-1s-1) and only weakly with SO4⚫- (4×105 M-1s-1) (Liang and Su, 2009). At pH 4.0, adding 0.5 M EtOH significantly suppressed GUA photodegradation, while tBuOH had little effect, supporting SO4⚫- as the dominant oxidant (Fig. S16).

The kinetics of the reaction between GUA and SO4⚫- were assessed using a relative rate method with phenol as the reference compound (Fig. S17) (Tran et al., 2022; Liang and Su, 2009). After correcting for direct photodegradation, the results indicate that GUA reacts very rapidly with SO4⚫-, with an effective rate approaching the diffusion-controlled regime under the experimental conditions. It is important to note that these values represent condition-dependent, relative estimates rather than absolute intrinsic rate constants. Given the uncertainties inherent in the relative rate approach (e.g., reference rate selection, radical distribution, and irradiation heterogeneity), the derived rate should be interpreted as an upper-limit estimate of reactivity. The observed fast kinetics are nevertheless consistent with prior quantum chemical calculations (Li et al., 2023b), supporting the high reactivity of GUA toward sulfate radicals..

In aqueous Na2SO3, the primary S(IV) species are SO2⋅H2O, HSO3-, and SO32-, with HSO3- dominant under the experimental pH range (Fig. S18). At pH 4.0, Na2SO3 showed nearly no UV-vis absorption above 250 nm (Fig. 3a). Adding 0.1 mM GUA introduced a strong 274 nm peak from π–π∗ transitions in GUA's aromatic ring. Although initial UVA absorption was minimal, it increased markedly during irradiation, indicating the formation of new light-absorbing products.

Since 370 nm UVA light (∼3.35 eV) lacks sufficient energy to directly excite either HSO3- or the HSO3--GUA complex, the formation of SO4⚫- likely involved photoactivation of intermediate complexes such as [HSO3-+O2] or [SO32-+O2]. Time-dependent density functional theory (TDDFT) calculations support this, showing that [SO32-+O2] can absorb UVA light (Fig. S19) and subsequently form reactive radicals. This is consistent with previous findings that halide-O2 complexes can be photoexcited by UVA to yield radicals like X⚫ and HO2⚫ (Cao et al., 2024a, b).

Density functional theory (DFT) calculations (Fig. 3b) reveal that electron transfer from the triplet state (T1) of [SO32-+O2] to form SO3⚫- and O2⚫- is endergonic (∼13 kcalmol-1) and unfavorable without light. TDDFT results indicate that UVA can excite T1 to higher-energy triplet states (T2), enabling this electron transfer. The resulting SO3⚫- is oxidized by O2 to SO5⚫-, which decomposes to SO4⚫-, while O2⚫- further oxidizes S(IV) species.

The photodegradation of GUA in aqueous Na2SO3 solution under UVA irradiation proceeds through three major mechanisms, summarized in Table 1:

(1) Formation of sulfur-centered radicals: Under UVA irradiation, the [SO32-+O2] complex is photoexcited from the triplet state (T₁) to a higher triplet (T₂), enabling electron transfer to produce SO3⚫- and O2⚫-. Although HSO3- is the predominant S(IV) species at pH 4.0, it exists in rapid dynamic equilibrium with SO32-, which is generated via fast dissociation of HSO3- (k=6.75×103 s-1), ensuring a sufficient concentration of SO32- for complex formation. SO3⚫- is rapidly oxidized by molecular O2 to form SO5⚫- at a high rate (k=1.5×109 M-1s-1).

(2) Oxidation of sulfites to sulfate: SO5⚫- reacts with HSO3- or SO32- to produce SO4⚫-, SO3⚫- and SO52-. SO3⚫- re-enters the cycle by reacting with O2 to regenerate SO5⚫-. SO52- protonates to HSO5-, which continues oxidizing S(IV) species to SO42-. O2⚫- also oxidizes S(IV), but more slowly than SO5⚫- or SO4⚫- (Table S1). Importantly, SO5⚫- reacts approximately 100 times faster with SO32- than with HSO3-, leading to a significant acceleration of sulfite photooxidation at pH>4.0 where SO32- dominates (Fig. S7).

(3) Formation of organosulfates: SO4⚫- reacts with GUA extremely rapidly, much faster than with S(IV) species. This rapid reaction leads to substantial formation of low-volatility organics compounds, including OSs and GUA dimers and derivatives, with a SOA yield of ∼80 % (Fig. S20). GUA also increases the overall rate of sulfite oxidation by nearly threefold, probably via additional reactive radicals generated during its reaction with SO4⚫-. Proposed mechanisms for the GUA-SO4⚫- reaction are shown in Figs. S21 and S22.

In atmospheric environments, cloud and fog droplets typically range from a few to tens of micrometers in diameter. Within these microdroplets, surface-active solutes often concentrate at the air-water interface, where reactions are accelerated due to surface enrichment and reduced activation energies (Ruiz-Lopez et al., 2020). Classical molecular dynamics (MD) simulations revealed that GUA is energetically favored at the interface, with an interfacial free energy 2.8 kcalmol-1 lower than in bulk water (Figs. 4a and S23). SO4⚫- also shows an interfacial preference, although much smaller (0.17 kcalmol-1 difference) (Xie et al., 2024), suggesting that both species are enriched at the interface.

Microdroplets also facilitate gas exchange, boosting [SO32-+O2] complex formation and SO4⚫- production under UVA. Thus, GUA photodegradation is expected to be far greater in microdroplets than in bulk water – potentially by several orders of magnitude.

To test this, we used field-induced droplet ionization mass spectrometry (FIDI-MS) (Huang et al., 2018; Gong et al., 2022; Zhang et al., 2023) to monitor UVA-induced photodegradation of 0.1 mM GUA in microdroplets, with and without 3.0 mM Na2SO3 (see Methods). Figure 4b shows averaged FIDI-MS signals from five droplets, fitted to pseudo-first-order kinetics (Figs. S24 and S25). GUA degraded nearly 200 times faster at the interface than in bulk (kbulk=2.6×10-5 s-1 vs. kinterface=4.8×10-3 s-1). With Na2SO3, the rate similarly increased ∼60-fold, indicating interfacial SO4⚫- concentrations of ∼10-12 M, about two orders magnitude higher than in bulk.

Overall, these findings demonstrate that phenolic compounds like GUA are enriched and highly reactive at air-water interfaces, where UVA-driven SO4⚫- formation greatly accelerates photodegradation and OS production.

The experimental system used in this study represents a simplified aqueous-phase environment designed to isolate key photochemical processes. Consequently, the derived kinetic parameters should be interpreted as condition-dependent estimates rather than direct quantitative representations of atmospheric reaction rates.

When gas-phase SO2 dissolves into cloud and fog droplets, it hydrates to form S(IV) species such as SO32-. In the presence of O2 and UVA irradiation, SO32- can be oxidized to SO42- through radical pathways. Based on our experimental measurements, the apparent photon efficiency (APE) of S(IV) oxidation under UVA irradiation was estimated. Using this experimentally derived APE as a constant parameter, we simulated sulfate formation induced by UVA under the AM0 standard solar spectrum, representing an upper-limit estimate of the sulfate production efficiency via this pathway. This efficiency was then compared with sulfate formation driven by conventional atmospheric oxidants, including NO2, O3, and transition metal ions (TMIs) (Cheng et al., 2016) (Fig. 5, see Methods). Under “Cloud droplets” conditions (Seinfeld and Pandis, 2016; Herrmann et al., 2015) (Fig. 5a), sulfate formation induced by UVA in the bulk solution was comparable in magnitude to that driven by NO2. Under “Beijing haze” conditions (Cheng et al., 2016), where the photonic flux in the UVA range is reduced to 34 %, UVA-induced sulfate formation remained comparable to the O3 oxidation pathways (Fig. 5b).

At the same time, it should be noted that the UVA-driven pathway identified here is not intended to replace or dominate established sulfate formation mechanisms, such as transition metal ion (TMI)-catalyzed oxidation. Rather, it represents a complementary pathway that may contribute to sulfate formation under specific conditions, particularly in aqueous environments where phenolic compounds and UVA irradiation coexist. Therefore, this pathway should be viewed as a complementary and condition-dependent source of sulfate, rather than a dominant process in complex atmospheric systems.

In summary, our results reveal a metal-free pathway for SO2 oxidation to sulfate in atmospheric aqueous phases under UVA irradiation. Unlike traditional mechanisms that rely on metal catalysts or high-energy UVB/UVC lights, we show that the [SO32-+O2] complex can initiate sulfate radical production under UVA – wavelengths far more prevalent in the solar spectrum.

In the presence of guaiacol – a common phenolic compound from biomass burning, these sulfate radicals drive rapid GUA oxidation, producing low-volatility organic compounds, including organosulfates. Moreover, microdroplet experiments show that GUA photodegradation is dramatically accelerated in small droplets under UVA light due to intensified interfacial chemistry. The high surface-area-to-volume ratio of microdroplets promotes efficient generation of reactive oxidants, particularly sulfate radicals, which accelerate both S(IV) oxidation and organics transformations. Together, these findings uncover a sunlight-accessible, metal-free pathway for sulfate and SOA formation, especially relevant to slightly acidic, sunlit, and water-rich atmospheric environments.