Light Absorption of Secondary Organic Aerosol: Composition and Contribution of Nitroaromatic Compounds

Abstract

Secondary organic aerosol (SOA) can affect the atmospheric radiation balance through absorbing light at shorter visible and UV wavelengths. However, the composition and optical properties of light-absorbing SOA is poorly understood. In this work, SOA filter samples were collected during individual chamber experiments conducted with three biogenic and eight aromatic volatile organic compound (VOC) precursors in the presence of NO_X and H₂O₂. Compared with the SOA generated using the aromatic precursors, biogenic SOA generally exhibits negligible light absorption above 350 nm; the aromatic SOA generated in the presence of NO_X shows stronger light absorption than that generated with H₂O₂. Fifteen nitroaromatic compound (NAC) chemical formulas were identified and quantified in SOA samples. Their contributions to the light absorption of sample extracts were also estimated. On average, the m-cresol/NO_X SOA sample has the highest mass contribution from NACs (10.4 ± 6.74%, w/w), followed by naphthalene/NO_X (6.41 ± 2.08%) and benzene/NO_X (5.81 ± 3.82%) SOA. The average contributions of NACs to total light absorption were at least two times greater than their average mass contributions at 365 and 400 nm, revealing the potential use of chromophoric NACs as brown carbon (BrC) tracers in source apportionment and air quality modeling studies.

Introduction

The light absorption of atmospheric aerosols significantly impacts the Earth’s radiation balance.(1, 2) Black carbon (BC) and mineral dusts have been recognized as efficient light absorbing agents for decades,(3–6) while global climate models commonly treat organic matter (OM) as a purely light scattering component.(7, 8) Many studies show that OM emitted during open biomass burning can strongly absorb light at shorter visible and UV wavelengths (λs) with strong λ dependency.(9–12) This light-absorbing OM is often referred to as “brown carbon” (BrC). Measurement-based estimates of radiative forcing due to OM have suggested that BrC accounts for ∼20% of the solar absorption of carbonaceous aerosols globally.(13, 14) Feng et al.(15) applied chemical transport and radiative transfer models to estimate the enhanced absorption of solar radiation due to BrC and found that the contribution of BrC to aerosol forcing could account for more than a quarter of the estimated radiative effects of BC globally. However, the available modeling studies on aerosol absorption with the inclusion of BrC only parametrized primary emissions of BrC from biomass or biofuel combustion,(13) and the light absorption of secondary organic aerosol (SOA) is rarely considered or assumed to be less important.(16)

Multiple studies have demonstrated the formation of secondary BrC through laboratory experiments.(9, 17–25) Laboratory-based chamber experiments suggest that a variety of factors impact the light absorption of SOA, including precursor composition (volatile organic compounds, VOC),(17) oxidants (e.g., hydroxyl radical, OH; nitrogen oxides NO_X),(20) aging,(9) and seed acidity.(25) These studies conclude that the SOA derived from the photo-oxidation of anthropogenic aromatic VOCs (e.g., toluene) has stronger light absorption than typical biogenic (e.g., α-pinene) SOA reaction products. Additionally, BrC formation is favored in reactions in the presence of gaseous ammonia and amines, high NO_X, and highly acidic seed particles. Some nitroaromatic compounds (NACs, e.g., nitrocatechol) were identified in laboratory-generated SOA and accounted for part of the total light absorption at near-UV (300–400 nm) irradiance.(26) However, the molecular structure and composition of NACs from different reaction conditions (VOC/oxidant) and their contributions to light absorption remain uncertain. Most previous studies investigated the light absorption of SOA from only a few biogenic and/or anthropogenic precursors with NO_X and/or OH.(17, 19, 20, 26) The potential of different chemical precursors to produce secondary BrC under identical conditions (e.g., oxidant, seed type) is poorly understood.

In the present study, SOA was generated by irradiating three biogenic (α-pinene, isoprene, and β-caryophyllene) and eight aromatic (1,3,5-trimethylbenzene, 1,2,4-trimethylbenzene, m-xylene, toluene, ethylbenzene, benzene, naphthalene, and m-cresol) VOCs individually in the presence of NO_X and hydrogen peroxide (H₂O₂), respectively, in a smog chamber. The SOA was collected on filters, extracted in methanol, and analyzed with a UV/vis spectrometer and a high performance liquid chromatograph/diode array detector (HPLC/DAD)-quadrupole (Q)-time-of-flight mass spectrometer (ToFMS). First, we compared the bulk mass absorption efficiency (MAE, m² g^–1) – a measure of BrC – of SOA from different precursors reacted under NO_X or H₂O₂. Next, the identification and quantification of NACs were conducted, and the mass contributions of NACs to SOA were calculated. Finally, the contribution of identified NACs to the light absorption of SOA was estimated. These study results provide a comprehensive picture about the light-absorbing characteristics of typical biogenic and aromatic SOA at bulk chemical and molecular levels, benefiting chemical source apportionment models and climate models using optical properties.

Methods

Smog Chamber Experiments

All SOA experiments were conducted at the U.S. Environmental Protection Agency, RTP, NC location within a fixed volume 14.5 m³ indoor chamber. Details of the equipment and experimental methods for the indoor chamber system were described elsewhere.(27–29) A summary of experimental conditions for the photo-oxidation of all VOC precursors is provided in Table S1 of the Supporting Information. Each VOC precursor was photo-oxidized under both NO_X and H₂O₂ conditions. For reactions with NO_X, nitric oxide (NO) was continuously injected from high pressure cylinders into the smog chamber through a mixing manifold. In the H₂O₂ experiments, 50 wt % H₂O₂ solution was vaporized and flushed into the chamber using via a heated glass bulb and photolyzed in the chamber to generate OH radicals in the absence of NO_X. Ammonium sulfate [(NH₄)₂SO₄] seed aerosol at a concentration of 0.1–1 μg m^–3 was introduced for all experiments to promote SOA formation. All experiments were conducted in dynamic flow mode with a ∼4 h residence time.

During experiments, the VOC precursor concentration in the chamber was determined using a cryogenic trap followed by GC-FID analysis (model 5890, Hewlett-Packard, Palo Alto, CA). NO and NO_X were measured using an oxides of nitrogen analyzer (TECO model 42C, Franklin, MA). Ozone was measured using an ozone monitor (Bendix model 8002, Lewisburg, WV). The radiation intensity was continuously monitored with an integrating radiometer (Eppley Laboratory, Inc., Newport, RI). The relative humidity (RH) and temperature were measured with an Omega digital thermohydrometer (model RH411, Omega Engineering, Inc., Stamford, CT). Once steady-state conditions were obtained (24 h), particle samples were collected at 16.7 L min^–1 using 47 mm Teflon-impregnated glass fiber filters (Pall Gelman Laboratory, Ann Arbor, MI). A 60 cm XAD4-coated annular denuder (URG, Inc., Chapel Hill, NC) was applied upstream of the filter to remove gas-phase organics in the air stream. Mass concentrations of SOA (μg m^–3) in the chamber were determined from gravimetric measurements of the filters. Air volumes and SOA concentrations for the filter samples analyzed in this work are listed in Table S1.

UV/Vis Analysis

Details of the extraction and instrument setup are given in the Supporting Information. Briefly, a 1.5 cm² punch of each filter sample was extracted ultrasonically with methanol (5 mL) for 15 min, followed by filtration and analysis with a UV/vis spectrometer. To convert the light absorption due to chromophores measured in solution to ambient aerosol chromophore concentrations, the light absorption values of all the filter extracts are converted to light absorption coefficients (Abs_λ, Mm^–1) by(30)

$${\mathrm{A}\mathrm{b}\mathrm{s}}_{\lambda }=\mathrm{}\left(A_{\lambda }-A_{700}\right)\times \frac{V_I}{V_a\times L}\mathrm{l}\mathrm{n}\left(10\right)$$ (1)

where A₇₀₀ is referenced to account for systematic baseline drift,(30) absorption was not observed at λ ≥ 700 nm for any sample, V_l (m³) is the volume of methanol (5 mL) used for extraction, V_a (m³) is the volume of the sampled air, and L (m) is the optical path length of the quartz cuvette (1 cm) in the UV/vis spectrometer. For ease of analysis, this work focused on Abs_λ at λ = 300–550 nm, where most of the BrC absorption was observed.(31)

To identify which reaction (VOC/oxidant) generated the strongest light-absorbing SOA, the Abs_λ value was normalized by the SOA concentration for each sample using eq 2:

$${\mathrm{M}\mathrm{A}\mathrm{E}}_{\lambda }=\mathrm{}\frac{{\mathrm{A}\mathrm{b}\mathrm{s}}_{\lambda }}{\mathrm{O}\mathrm{M}}$$ (2)

in which MAE_λ is the bulk mass absorption efficiency (m² g^–1), and OM (μg m^–3) is the mass concentration of SOA in each sample. It was assumed that the SOA is 100% extracted by methanol. The solution absorption Ångström exponent (Å_a) is a measure of spectral dependence of light absorption and is determined from the linear regression fit of log₁₀(Abs_λ) vs log₁₀(λ) over the λ range of 300–550 nm. Finally, the imaginary part of the complex refractive index (k), a parameter used to evaluate the light-absorbing characteristics of particulate matter, was calculated for each SOA sample based on the spectroscopic data. See the Supporting Information for calculation method details.

Sample Extraction and HPLC/DAD-Q-ToFMS Analysis

Internal standard (25 μL of 10 ng μL^–1 nitrophenol-d₄) was spiked onto each filter sample prior to extraction. Approximately 3–5 mL of methanol was used to ultrasonically extract each sample twice (15 min each time). Extract was filtered into a pear-shaped glass flask and rotary evaporated to ∼0.1 mL. Concentrated extracts were then transferred into a solvent-rinsed 2 mL amber vial. The final extract sample volume was ∼500 μL prior to HPLC/DAD-Q-ToFMS analysis. All glassware used during the extraction procedure was prebaked at 550 °C overnight. Sample and extract storage temperatures were ≤ −20 °C.

The identification and quantification of target NACs were conducted with an HPLC (1200 series, Agilent Technologies, Santa Clara, CA) coupled to a UV/vis DAD (G1315C, Agilent Technologies, Santa Clara, CA) followed by a Q-ToFMS (Agilent 6520). The Q-ToFMS used a multimode source operating in electrospray ionization (ESI) and negative ion (−) modes. Details of the HPLC/DAD-Q-ToFMS instrument conditions and methods are provided in the Supporting Information. Briefly, samples were analyzed in full scan mode (40–1000 Da). The combination of HPLC, online light absorbance, and accurate MS detection (2 ppm) permits direct identification of chemical compound formulas responsible for absorbing photons in the near UV and visible range. Selected samples were re-examined using MS-MS mode under identical chromatographic conditions. [M–H]⁻ ions were targeted for collision-induced dissociation (CID) using predetermined accurate mass and retention times. MS/MS data offered structural m/z data and were used to compare standard compounds to those identified in SOA samples. The standard, surrogate assignments, proposed structures, and 15 sample-identified chemical formulas are listed in Table S2. Acceptance criterion for compound identification and quantification was set at ±10 ppm mass accuracy. The extracted ion chromatograms (EICs) and Q-ToF MS/MS spectra for standard compounds are shown in Figure S1, Supporting Information. The EICs, MS/MS spectra, and UV/vis chromatograms (DAD signal at 365 nm) for identified compounds in selected SOA samples are provided in Figures S2–S10. Because the sample extract was analyzed by UV/vis detector before entering the MS, the retention times of the identified compounds in UV/vis chromatogram was ∼0.15 min earlier than the EIC.

Quantification was performed using the internal standard method. Multipoint (n = 9) calibration curves were generated using the relative MS response following standard dilution (∼0.01–2 ng μL^–1). The identified NACs (including isomers) represented by each formula were quantified individually using standard or surrogate compounds with similar structure or [M–H]⁻ ion (Table S2) and then added together for the calculation of mass contribution to the total SOA in each sample. Standard recoveries were determined after spiking blank filters with known target compound concentrations and following the extraction and analysis procedures detailed above. Average recoveries of standard compounds ranged from 75.1 to 116% (Table S3). Table S3 also provides the method detection limit (MDL) of each standard compound. Each MDL was calculated as 3σ following multiple injections (n = 7) at the lowest calibration concentration. All reported data were above MDL. Blank filters were extracted and analyzed in the same manner as SOA samples, and no contamination was observed for identified compounds.

Estimation of NACs Contribution to Abs_λ

The MAE values of individual compound standards in methanol were used to estimate NAC absorption at a given UV/vis λ, similar to the method applied in Zhang et al.(32) Standards were diluted in series (5–11 points) over the concentration ranges in the sample extracts. Concentrations of identified compounds in sample extracts (C_l, ng μL^–1) for UV/vis analysis were calculated as

$${\mathrm{C}}_{\mathrm{I}}=\mathrm{}\frac{C_a\times V_a}{V_I}$$ (3)

where C_a (ng m^–3) is the air concentration of the NAC based on HPLC-ToFMS quantification, V_a (m³) is the sample volume corresponding to the filter aliquot (1.5 cm²) used for extraction, and V_l (μL) is the solvent volume (5 mL). The UV/vis spectra of standard compounds are shown in Figure S11. Details of MAE calculations for standard compounds and Abs_λ attribution for identified NACs are given in the Supporting Information. The MAE values at 365 nm, 400 nm, 450 nm, 500 nm, and 550 nm and the corresponding linear ranges of standard compounds are provided in Table S4.

Results and Discussion

Light-Absorbing Characteristics of SOA Extracts

Average MAE spectra for biogenic and aromatic SOA generated under NO_X and H₂O₂ conditions are shown in Figure 1. The summary of average bulk MAE and k at five representative wavelengths over the λ range of 300–550 nm (λ = 365, 400, 450, 500, 550 nm) and Å_a is given in Table S5. In Figure 1a-c, biogenic SOA generated under both NO_X and H₂O₂ conditions shows minimal absorption from 300 to 550 nm, while the MAE spectra for aromatic SOA (Figure 1d-k) reflect typical atmospheric BrC with high UV absorption. Among the eight aromatic VOCs, SOA generated from toluene, ethylbenzene, benzene, naphthalene and m-cresol showed significantly (p < 0.05) higher average MAE₃₆₅ under NO_X conditions as compared to H₂O₂ conditions (Table S5). In the presence of NO_X, the SOA of m-cresol had the highest average MAE at 365 (2.47 ± 1.05 m² g^–1) and 400 nm (1.19 ± 0.37 m² g^–1); naphthalene had the highest average MAE at 450 (0.43 ± 0.034 m² g^–1), 500 (0.25 ± 0.015 m² g^–1), and 550 nm (0.12 ± 0.011 m² g^–1), followed by benzene SOA (Table S5). Light absorption was also observed in the range of 300–550 nm for SOA generated under H₂O₂ conditions with m-xylene, toluene, ethylbenzene, benzene, naphthalene, and m-cresol. Naphthalene had the strongest light-absorbing SOA in the presence of H₂O₂ at each of the five representative wavelengths, followed by benzene, ethylbenzene, and m-cresol. The imaginary part of complex refractive index (k) was linearly derived from MAE, so the difference in average k values across different VOCs or between NO_X and H₂O₂ conditions is reflected by the average MAE values. The average k values were summarized in Table S5. For the five aromatic VOCs with relatively strong light-absorbing SOA under NO_X conditions (Figure 1g-k), the average Å_a values were lower for NO_X conditions (range 4.92 ± 0.23 to 6.89 ± 0.63) than H₂O₂ conditions (6.46 ± 0.25 to 7.75 ± 0.24, Table S5). The Å_a values calculated for those SOA extracts (e.g., isoprene) with large percentages of A_λ – A₇₀₀ (λ = 300–550 nm, >20%) measurements with close to zero or negative values and those with very high standard deviations are subject to large uncertainty because the SOA had weak or no absorption (Figure 1a-d) and thus may not contain BrC. The above results suggest that the photo-oxidation of biogenic VOCs may not generate light-absorbing SOA on neutral seed [(NH₄)₂SO₄]; the aromatic SOA contains BrC, for which the light-absorption is sensitive to differences in molecular structures of SOA precursors and the availability of NO_X.

Figure 1.

Bulk MAE spectra for SOA formed under NO_X and H₂O₂ conditions, from (a) α-pinene, (b) isoprene, (c) β-caryophyllene, (d) 1,3,5-trimethylbenzene, (e) 1,2,4-trimethylbenzene, (f) m-xylene, (g) toluene, (h) ethylbenzene, (i) benzene, (j) naphthalene, and (k) m-cresol. For sample number N > 2, gray and cyan areas represent ±1 standard deviation; and for sample number N = 2, cyan and gray areas represent ± half the difference between the two measurements.

Liu et al.(20) have investigated the light absorption of SOA generated under NO_X and NO_X-free conditions for two aromatic (toluene and trimethylbenzene) and two biogenic VOCs (α-pinene and isoprene). The average MAE₃₆₅ for toluene/NO_X SOA (0.55 ± 0.17 m² g^–1) in this study is comparable to that observed by Liu et al.(20) (∼0.8 m² g^–1). However, the MAE₃₆₅ for the NO_X-free condition is 1 order of magnitude lower in Liu et al.(20) (∼0.01 m² g^–1) than here (0.17 ± 0.051 m² g^–1). For naphthalene SOA in the presence of NO_X, the MAE₄₀₀ (0.78 ± 0.074 m² g^–1) obtained here is comparable to the MAE₄₀₅ obtained by Metcalf et al.(33) (0.81 m² g^–1) but much higher than the MAE₄₀₅ values in Updyke et al.(24) (0.10 m² g^–1). The k₃₆₅ values for toluene (with NO_X ∼ 0.025, with H₂O₂ ∼ 0.008) and m-xylene (with NO_X ∼ 0.008, with H₂O₂ ∼ 0.002) SOA in Liu et al.(19) are obtained similarly and are comparable to the corresponding average k₃₆₅ values in the present study (Table S5). The low absorbance for biogenic SOA materials observed presently is consistent with previous results.(17, 20, 34) The differences in the light absorption of aromatic SOA across studies may be ascribed to the different reaction conditions, including VOC/NO_X ratio, irradiation power, reaction time in the chamber, etc., which warrants further study.

Identification and Quantification of NACs and Their Contributions to Bulk Abs_λ

Since NACs have been detected as light-absorbing components in the atmosphere(32, 35) and can be generated from photochemical reactions,(26) the NAC composition and its contribution to bulk absorption of SOA samples are investigated. First, 15 NAC chemical formulas in the SOA samples were determined (Table S2) from the HPLC/DAD-Q-ToF MS analysis. Multiple NAC compound structures were examined using MS/MS data. The relative mass contribution (%) to total SOA of each NAC formula categorized by precursor VOC and oxidant type (NO_X or H₂O₂) is given in Table S6, and the average relative mass contributions (%) of each NAC in SOA samples can be visualized in Figure 2. The m-cresol/NO_X SOA has the highest average mass contribution from total NACs (10.4 ± 6.74%), followed by naphthalene/NO_X (6.41 ± 2.08%) and benzene/NO_X (5.81 ± 3.82%) SOA. The contribution of each identified NAC to the bulk Abs_λ of SOA samples was estimated from the mass concentration (ng m^–3) and MAE (m² g^–1) of individual standards. Average contributions of identified NACs to the light absorption (Abs_λ) of SOA samples are listed in Tables S7–S11 for λ’s of 365 nm, 400 nm, 450 nm, 500 nm, and 550 nm. For simplicity, the average contributions of identified NACs to Abs_λ of SOA samples from different experiments at 365 nm, 400 nm, and 450 nm are visualized in Figures 3, S13, and S14, respectively. In Figure 3, the m-cresol/NO_X SOA had the highest Abs₃₆₅ contribution of all identified NACs (average ± standard deviation, 50.5 ± 15.8%), followed by benzene/NO_X (28.0 ± 8.86%) and naphthalene/NO_X (20.3 ± 8.01%) SOA; for these three SOAs, roughly 36.5 ± 15.8–47.4 ± 20.0% of Abs₄₀₀ could be attributed to identified NACs (Figure S13). The biogenic SOA samples show weak light absorption (Figure 1) and comprise virtually no NACs under NO_X and H₂O₂ conditions (Table S6 and Figure 2). The lack of aromaticity in the biogenic precursors and secondary reaction products presumably explains this observation.(36) Clearly, light-absorbing conjugated double bonds are retained in aromatic SOA as NACs; thus, NAC compositions and contributions to Abs_λ in the aromatic SOA samples are the focus henceforth.

Figure 2.

Average mass contributions (%) of nitroaromatic compounds to total SOA generated in different chamber experiments.

Figure 3.

Average contributions (%) of nitroaromatic compounds absorption to the light absorption coefficients of SOA extracts at 365 nm (Abs₃₆₅) for different chamber experiments.

NAC Composition in Aromatic SOAs

For the m-cresol/NO_X experiment, C₇H₇NO₄ products dominate identified NACs (Figure 2), accounting for 8.34 ± 5.87% of the total SOA (Table S6), consistent with the observations (9.9%) of Iinuma et al.(37) Three structural isomers are observed for C₇H₇NO₄ with the strongest EIC signal intensity (∼10⁶) and light absorption (DAD signal at 365 nm; Figure S2a-c). These three compounds show similar MS/MS spectra (Figure S2A-C). In the MS/MS spectra, the m/z 137 ion explains the loss of H and NO ([M–H]⁻, 168), and the m/z 109 ion explains continuous CO loss possibly due to aromatic ring opening. Iinuma et al.(37) identified three C₇H₇NO₄ isomers using authentic standards in field and m-cresol/NO_X photo-oxidation samples. Their separation and analytical methodology is similar to our own, so the elution order of the three C₇H₇NO₄ isomers peaks in the EIC is also presumably similar. According to Iinuma et al.,(37) the C₇H₇NO₄ signals at 8.46, 8.55, and 9.22 min in Figure S2b are 4-methyl-5-nitrocatechol, 3-methyl-5-nitrocatechol, and 3-methyl-6-nitrocatechol, respectively. In our study, three C₇H₇NO₅ isomers were also identified, and their MS/MS spectra showed abundant m/z 125 and 153 ions (Figure S3a, A-C). Comparing to the MS/MS spectra of 4-nitrocatechol standard and C₇H₇NO₄ (Figures S1C, S2A-C), the C₇H₇NO₅ compounds should have a nitrocatechol skeleton with either one methyl and one hydroxyl group or one hydroxymethyl group. The three C₇H₇NO₅ isomers also have discernible light absorption at 365 nm (Figure S3b). Due to the weak EIC signal intensity for the C₈H₉NO₅ compound eluting at 4.14 min (∼10³), the MS/MS spectra are only available for C₈H₉NO₅ compounds at 8.51 and 9.27 min (Figure S4). After losing one CH₃ and one NO₂, the fragmentation patterns of C₈H₉NO₅ compounds are similar as C₇H₇NO₄ with abundant m/z ions of 137 and 109. Thus, the C₈H₉NO₅ compounds may have a methyl nitrocatechol skeleton with either one methyl and one hydroxyl group or one hydroxymethyl group. None of the 15 identified NACs is observed in m-cresol/H₂O₂ samples.

Nitronaphthol and hydroxyl nitronaphthol were identified in naphthalene SOA under both NO_X and H₂O₂ conditions (Figure 2) from their predicted formula and by comparing their MS/MS spectra (Figure S5A-C) to the reference compound (2-nitro-1-naphthol, Figure S1H). These two compounds accounted for 5.44 ± 1.73% and 1.63 ± 1.83% of the naphthalene SOA under NO_X and H₂O₂ conditions, respectively. However, their light absorption signals at 365 nm could hardly be resolved from background noise in the UV/vis chromatogram (Figure S5c). This might be ascribed to their low concentrations in the sample extract, as reflected by their weak EIC signal intensity (Figure S5a,b). Among all the chamber experiments with H₂O₂ as the oxidant, only naphthalene SOA contained a significant amount (mass contribution >1%) of NACs. Updyke et al.(24) observed strong light-absorbing SOA at visible wavelengths from naphthalene/H₂O₂ experiments. They also demonstrated that the exposure of SOA to gaseous NH₃ (RH ∼ 100%, 72 h) can increase SOA absorption. In the current work, the strong light absorption and high concentrations of NACs observed for naphthalene/H₂O₂ samples might be explained as follows: the trace amount of residue NO_X (<10 ppb) or nitrogen containing substance in the chamber tends to react with naphthalene in a short-term; the interactions between gas phase NH₃ or dissolved NH₄⁺ from seed aerosols and naphthalene/H₂O₂ SOA lead to NAC generation. However, the latter might be less likely and warrants further study, since the photochemical reactions of naphthalene in the current work take place under dry conditions (RH < 20%) and the residence time of SOA is short (4 h), and trivial NACs were observed in other VOC/H₂O₂ reactions.

In the benzene/NO_X SOA sample, the C₆H₅NO₄ and C₆H₅NO₅ dominated the composition of NACs (Figure 2), contributing 4.65 ± 3.78% and 1.08 ± 0.39% to the total SOA mass on average (Table S6). Referring to the EIC and MS/MS spectrum of 4-nitrocatechol (Figure S1c, C), the C₆H₅NO₄ compound could be identified as 4-nitrocatechol; the C₆H₅NO₅ compound should contain a nitrocatechol skeleton with an extra hydroxyl group on the benzene ring. These two compounds also had strong light absorption from the UV/vis chromatogram (Figure S6c). In Figure S7a, the C₆H₅NO₃ peak at 8.27 min was identified as 4-nitrophenol using standard reagent (Figure S1a, A), and the peak at 8.54 min might be a structural isomer of 4-nitrophenol. The C₆H₄N₂O₆ compound contains two nitro groups based on its fragmentation pattern (Figure S7B). Among the eight benzene/NO_X samples analyzed in this work, four were collected from the chamber experiments with added methyl nitrite (CH₃ONO), and the other four were from the experiment without methyl nitrite. As shown in Figure S12, the presence of methyl nitrite did not change the composition of NACs in SOA products, as compared with the experiment with only benzene, but it might lead to a decrease in SOA yield (Table S1) and the mass contribution of C₆H₅NO₄ to total SOA.

In Figure S8a, three C₈H₉NO₄ isomers are observed in the ethylbenzene/NO_X sample. In their MS/MS spectra, the compound eluting at 8.27 min had the dominant ions of m/z 138 and 108 (Figure S8A); the two C₈H₉NO₄ compounds eluting at 9.59 and 10.22 min had the same dominant ions (m/z 109 and 137; Figure S8B, C) as methyl nitrocatechols (Figure S2A-C) and were postulated as ethyl nitrocatechols. In contrast to the latter two C₈H₉NO₄ isomers, the C₈H₉NO₄ compound eluting at 8.27 min might experience ring opening (loss of CO) before the loss of NO during the fragmentation. Three carboxylic acid-like compounds (C₈H₇NO₄, C₇H₅NO₅, and C₉H₉NO₄) were observed for the ethylbenzene/NO_X sample (Figure S9). These compounds have low EIC signal intensity and are more fragile than nitrophenolic compounds with the CID fragmentation, so their MS/MS spectra were not available in the current work. For the ethylbenzene/NO_X experiment, the C₈H₉NO₄ (0.32 ± 0.26%), C₈H₇NO₄ (0.27 ± 0.21%), C₆H₅NO₄ (0.23 ± 0.10%), and C₇H₅NO₅ (0.21 ± 0.095%) compounds had significant and comparable mass contribution to total SOA products (Table S6).

In Figure S10a, the C₇H₇NO₃ compound eluting at 9.74 min is identified, using standard reagent (Figure S1b, B), as 2-methyl-4-nitrophenol, which dominates the C₇H₇NO₃ isomers from the toluene/NO_X sample. This is consistent with the observation of toluene/NO_X products from Sato et al.(38) The C₇H₇NO₄ isomers from toluene/NO_X SOA eluting during 8.43–9.20 min in Figure S10b are the same as those in Figure S2b and could be identified as methyl nitrocatechols. The MS/MS spectrum for the C₇H₇NO₄ compound eluting at 6.18 min had the same dominant ions (m/z 108 and 138; Figure S10B) as the C₈H₉NO₄ compound from ethylbenze/NO_X SOA eluting at 8.27 min (Figure S8A), indicating a common molecular skeleton. The retention times and fragmentation patterns of the C₇H₆N₂O₆ compounds (Figure S10c, C) are different from the reference compound (4,6-dinitromethylresorcinol; Figure S1j, J), but the functional groups composition on the benzene ring may be similar, including two nitro, two hydroxyl, and one methyl groups. Among all the NACs identified for toluene/NO_X SOA, C₇H₇NO₄ had the highest average mass contribution (0.27 ± 0.17%, Table S6), followed by C₇H₆N₂O₆ (0.19 ± 0.093%) and C₆H₅NO₄ (0.088 ± 0.036%). The identified NACs in the SOA products generated with the other aromatic VOCs (m-xylene, 1,2,4-trimethylbenzene, and 1,3,5-trimethylbenzene) in this work had low mass contributions (∼0.2%, Table S6).

Several previous studies have investigated the identification of NACs in aerosols from chamber reactions(26, 37–39) and biomass burning emissions.(40, 41) These studies demonstrated that the NACs could be generated from atmospheric processing of aromatic VOCs and raised the possibility of using the identified NACs as source markers for BrC. However, both biomass burning and fossil fuel combustion could emit a complex mixture of gas-phase aromatic VOCs (e.g., benzene, toluene),(42–47) serving as precursors for NACs. It is impractical to link these NACs to specific pollution sources (e.g., biomass burning), although NACs can be linked with specific aromatic precursors. Moreover, the BrC chromophores (NACs) identified in this work may be subject to aging and photobleaching in the atmosphere. Lin et al.(41) and Forrister et al.(48) examined the light absorption decay of OM from biomass burning and obtained a half-life of ∼15 h. The photobleaching of secondary BrC from reactions of limonene/O₃ SOA + NH₃ and glyoxal or methyl glyoxal + ammonium sulfate can be quite fast (a few minutes to hours).(49, 50) While the light absorption decay for some NACs (e.g., 4-nitrophenol and 4-nitrocatechol) and naphthalene/NO_X SOA in water solutions under UV photolysis (without the addition of H₂O₂) is much slower, with a half-life comparable to biomass burning OM.(49, 50) Due to the longer half-lives of NACs, several NACs identified in this work (e.g., nitrocatechol, methylnitrocatechol) have been observed in ambient aerosols and cloudwater with the highest concentrations of 1–3 ng m^–3 for individual compounds.(32, 35)

The chamber experiments performed here are limited (e.g., light intensity and spectra, temperature and humidity variation, gas and particle composition) in representing the atmosphere, and the identified BrC chromophores are vulnerable to photobleaching. However, the emissions of aromatic VOCs and NO_X are prominent in urban areas, and the NACs identified in this work could be generated continuously during the daytime. Moreover, the lifetimes of these NACs are long enough for accumulation and detection in the atmosphere. Thus, the identified NACs in this work can serve as source tracers to represent secondary BrC chromophores. As shown in Figure 2, several aromatic precursors have characteristic NAC compositions, and some NACs are uniquely generated from specific aromatic VOCs (e.g., C₇H₆N₂O₆ in toluene/NO_X SOA). Therefore, in future ambient analysis or source apportionment studies, the identified NACs with high mass contributions (e.g, C₁₀H₇NO₃ for naphthalene) to total SOA in this study could be used as source markers for secondary BrC derived from specific aromatic VOCs in urban areas, when and where biomass burning is less prominent; the monitoring data of aromatic VOCs can serve as supporting evidence, benefiting the identification of the dominant VOC precursor. Those NACs might also be used to estimate the SOA contributions of aromatic VOCs to ambient organic aerosol with the organic tracer-based method by Kleindienst et al.(51) However, the associated uncertainties due to the variability in mass contribution of specific NACs in different SOA samples and the influence from biomass burning and photobleaching need further study.

Contributions of NACs to Abs_λ

In general, the NACs in aromatic SOA samples had higher contributions to Abs₃₆₅ than their mass contributions (Figures 2 and 3), suggesting that most of the identified NACs are strong BrC chromophores. Similar to their mass compositions, the C₇H₇NO₄ and C₇H₇NO₅ compounds had the largest contributions to Abs₃₆₅ (40.5 ± 15.2% and 8.25 ± 1.79%) in m-cresol/NO_X SOA; C₁₀H₇NO₃ and C₁₀H₇NO₄ were the characteristic compounds in Abs₃₆₅ contribution (NO_X 13.2 ± 5.62% and 4.52 ± 1.56%; H₂O₂ 8.09 ± 8.84% and 1.49 ± 1.61%) for naphthalene SOA; the C₆H₅NO₄ and C₆H₅NO₅ compounds lead the contributions to Abs₃₆₅ (17.5 ± 9.28% and 10.1 ± 4.22%) in benzene/NO_X SOA; the C₇H₇NO₄ and C₇H₆N₂O₆ compounds are the two major Abs₃₆₅ contributors (6.68 ± 3.09% and 2.43 ± 1.08%) in toluene/NO_X SOA. In this work, 4-nitrocatechol, 2-methyl-4-nitroresorcinol, and 2-nitrophloroglucinol were used as the reference compounds for C₆H₅NO₄, C₇H₇NO₄, C₆H₅NO₅, and C₇H₇NO₅ in estimating their Abs₃₆₅ contributions. All three compounds have peak absorption at around 350 nm (Figure S11), explaining the shoulders observed in the UV/vis spectra for benzene/NO_X and m-cresol/NO_X samples in Figure 1i,k. In ethylbenzene/NO_X samples, the C₈H₇NO₄ and C₇H₅NO₅ compounds had comparable mass contributions to C₆H₅NO₄ and C₈H₉NO₄ (Figure 2), but the contributions to Abs₃₆₅ were dominated by the latter two compounds (Figure 3). This is because the Abs₃₆₅ contribution from C₈H₇NO₄ and C₇H₅NO₅ was estimated using two carboxylic acid compounds (3,5-dimethyl-4-nitrobenzoic acid and 2-methyl-5-nitrobenzoic acid) with very low absorption at wavelengths above 350 nm (Figure S11).

At 400 nm, the light absorption of 2-nitro-1-naphthol and 2-nitrophloroglucinol, two surrogate compounds used to estimate the light absorption of C₁₀H₇NO₃, C₁₀H₇NO₄, C₆H₅NO₅, and C₇H₇NO₅, respectively, almost reaches the maximum in their UV/vis spectra (Figure S11). While the light absorption of the reference compounds (4-nitrocatechol and 2-methyl-4-nitroresorcinol) for C₆H₅NO₄ and C₇H₇NO₄ is much lower at 400 nm than at 365 nm (Figure S11). As a result, the NACs generated from naphthalene/NO_X experiments had the highest average contribution (47.4 ± 20.0%) to Abs₄₀₀, followed by m-cresol/NO_X (39.3 ± 2.94%) and benzene/NO_X (36.5 ± 9.56%) samples (Figure S13). Since all the reference compounds have their maximum absorption below 450 nm, the Abs₄₅₀ contributed by NACs (<20%) in different experiments was lower than their contributions to Abs₄₀₀ by more than 50%. At 500 and 550 nm, very few identified compounds (e.g., C₆H₅NO₄, C₆H₄N₂O₆) had contributions to the light absorption of sample extracts, and the sum of their contributions was no more than 3% of the total SOA (Tables S10 and S11). Zhang et al.(32) and Teich et al.(52) measured the light absorption from NACs in ambient particles, which accounted for no more than 10% (0.02–9.86%) of the aqueous extract light absorption at 365–370 nm, much lower than the total NACs contribution to Abs₃₆₅ of benzene/NO_X, naphthalene/NO_X, and m-cresol/NO_X SOA in this work. This might be due to the fact that the ambient SOA is derived from the photochemical reaction of a complex mixture of VOCs, and most of them (e.g., biogenic VOC, toluene, and trimethylbenzene) have weak potential in generating NACs. Additionally, atmospheric processes (e.g., aging and photobleaching) can remove BrC chromophores in the atmosphere.

To understand if the difference in bulk MAE_λ observed in Figure 1 between NO_X and H₂O₂ conditions is caused by the difference in NACs production, the average contributions of NACs to MAE at 365, 400, and 450 nm for the different experiments were calculated using eq 1. The difference in bulk MAE and MAE contributions from NACs between aromatic/NO_X and aromatic/H₂O₂ samples at 365, 400, and 450 nm is provided in Table S12. The difference in MAE contributions from NACs explained the highest fraction (59.0%) of the bulk MAE difference between m-cresol/NO_X and m-cresol/H₂O₂ samples at 365 nm. However, for other reactions, no more than 50% of the bulk MAE differences between NO_X and H₂O₂ conditions could be explained by the difference in MAE contributions from NACs, especially at long wavelengths (∼450 nm). For 1,3,5-trimethylbenzene, 1,2,4-trimethylbenzene, and m-xylene, the explainable fractions could be even lower (<10%). However, these three aromatic VOCs show no significant differences (Table S5) in bulk MAE₃₆₅ between NO_X and H₂O₂ conditions. These results suggested that besides the identified NACs in this work, other light-absorbing products could be generated in reactions with NO_X and contributing to the differences in bulk MAE between NO_X and H₂O₂ experiments. Lin et al.(26, 41) investigated the molecular characterization of BrC in biomass burning aerosols and toluene/NO_X SOA. They identified a list of nitrogen-containing compounds with carbon number up to 48; Di Lorenzo et al.(53) found that most BrC absorption was due to high molecular weight (HMW > 500) molecules in both fresh and aged biomass burning aerosols. As such, the unexplained differences in bulk MAE between NO_X and H₂O₂ experiments might be ascribed to unidentified nitrogen-containing compounds and HMW components with extremely low volatility.

Atmospheric Implications

The stronger light absorption of aromatic SOA compared to biogenic SOA reveals the potential significance of aromatic VOCs as important precursors for secondary BrC; the comparisons of bulk MAE spectra derived from SOA extracts of NO_X versus H₂O₂ conditions imply that the secondary BrC should have more impact on aerosol absorption in urban atmospheres with abundant aromatic VOCs and NO_X. Fifteen NAC chemical formulas were derived from aromatic SOA samples. The average total amount of characteristic compounds in each sample can explain up to 10% of SOA mass, which is greater than other anthropogenic tracers (e.g., phthalic acid, 2,3-dihydroxy-4-oxopentanoic acid) in aromatic SOA reported by Kleindienst et al.(51, 54) Some of the identified NACs are predominantly associated with specific aromatic VOCs (e.g., C₆H₅NO₅ for benzene, C₁₀H₇NO₃ for naphthalene, and C₇H₇NO₅ for m-cresol), suggesting potential application of these compounds in ambient BrC or PM source apportionment studies for the estimation of aromatic SOA contribution. However, the uncertainties associated with mass contribution, as induced by the application of surrogate compounds and the difference between reactions (residence time) in chamber and that in the atmosphere, need to be addressed in future work; the influences from biomass burning and atmospheric processing also need to be evaluated. Up to 50% of bulk Abs_λ and MAE_λ could be attributed to the characteristic NACs at 365 and 400 nm for m-cresol/NO_X, benzene/NO_X, and naphthalene/NO_X SOA, which had the strongest light absorption among all types of SOA, indicating that NACs could be important BrC chromophores in the atmosphere. Little light absorption of SOA extracts could be explained by identified NACs at wavelengths longer than 450 nm, and further work is needed to elucidate the structures and quantity of HMW BrC components in aromatic SOA with strong chromophores in the visible region.

The views expressed in this article are those of the authors and do not necessarily represent the views or policies of the U.S. Environmental Protection Agency.