A total of 4 burning experiments were conducted using 4 different types of burning materials, including beech, straw, cow dung, and plastic, as shown in Table S1 in the Supplement. Please note that each burning experiment had no replicate experiment. This could lead to uncertainties in the results since the biomass burning experiments involve inherent variability. The burning materials are described in more detail by previous publications (Zhang et al., 2023; Wang et al., 2025; Li et al., 2024). Briefly, the straw and beech were sourced from a local forestry company in Würenlingen, Switzerland. The cow dung cakes (made of cow dung and straw) were sourced from Goyla dairy, Delhi, India and polyethylene plastic materials were bought in Delhi, India. We categorized four burning types for this experiment: beech, cow dung, straw, and plastic. We selected these four solid fuels and conducted emissions tests to simulate certain types of burning in the atmosphere. The beech, straw, and cow dung are representative of residential combustion fuels, which is consistent with the materials used in previous articles (Zhang et al., 2023; Li et al., 2024; Wang et al., 2025). The polyethylene plastic materials are representative of plastic burning in India and China (Haque et al., 2019).

The stainless-steel cylinder is 62 cm in diameter and 35 cm in height. This cylinder has been used to investigate wood beech burning emissions in the previous studies (Sardena et al., 2026; Bogler et al., 2025). The burning cow dung and plastic materials on top of stainless-steel cylinder were presented as waste burning in India (Zhang et al., 2023; Wang et al., 2025). In addition, the cow dung and plastic burning on the top of stainless-steel cylinder was more efficiency than in the stainless-steel cylinder. The experimental setup is shown in Fig. S1 in the Supplement. The fuels were ignited with kindling, and the emissions were pulled into a hood. After the kindling burned away (∼ 3 to 10 min after ignition), the emissions were introduced into a holding tank through stainless steel sampling lines and passing through an ejection dilutor (DI-1000, Dekati Ltd.) with a dilution ratio of ∼ 10. The holding tank is a stainless-steel container (1 m³) used to store emissions. The aim of holding tank is to average the emissions at different combustion efficiency in order to fully characterize the emission in ambient (Zhang et al., 2023). The holding tank was flushed overnight with clean air before each experiment, ensuring the background particle concentrations were less than 10 particles cm⁻³. To avoid cross-contamination from residual SVOCs in the holding tank, it was flushed overnight with clean air before each experiment, ensuring the background particle concentrations were less than 10 particles cm⁻³. This way residual SVOCs did not significantly affect subsequent measurements (Li et al., 2024). The emissions were injected into the holding tank for 10 to 30 min, depending on the emission source (Zhang et al., 2023). When the particle mass concentration was between 1 to 5 mg m⁻³, we stopped injecting particles into the holding tank. The particles in the holding tank were collected by quartz filters and Teflon filters and samples can be considered as primary organic aerosol. Blank quartz and Teflon filters were collected during the campaign and analyzed in the same way as the other samples. Please note that quartz filters and Teflon filters have different adsorption behaviour of gases which could lead to different filter loading and hence differences between EEM and mass spectrometer analysis. The partitioning of the VOC into the aerosol particles loaded on the filters may have a larger impact than the adsorption on the filter surface and this effect is expected to be similar for both filter types. However, the sampling time for fresh particles was 5 min and the sampling flow for fresh and aged particles was 8 L min⁻¹, which will help mitigate uptake of gases onto the filters. The overall adsorption of SVOC on the filter was small compared to the typical aerosol particle mass loading.

The holding tank was connected to an oxidative flow reactor (OFR) to generate secondary organic aerosol from the volatile organic compounds and NO_x present in the holding tank. Please note that the NO_x was generated during the burning. There were no instruments to measure the NO_x concentration. However, according to previous studies, the concentrations of NO_x were around a few hundred ppm (Winter et al., 1999; Courtemanche and Levendis, 1998). Furthermore, the volatile organic compounds were generated from the combustion (Li et al., 2024). After the filter collection of the primary organic aerosol (POA), an aged OA filter was collected including both the POA and secondary organic aerosol (SOA) generated in the OFR. The OFR was maintained at 293 K and 50 % relative humidity, and a constant O₃ concentration of 4 ppm was present. The OFR has 254 nm lights, which photolyzes O₃ to produce OH radicals (Li et al., 2024). The OH exposure was set to 5.4 × 10¹¹ molec. cm⁻³ s during the experiment (Li et al., 2024). The holding chamber provided a constant source of VOCs and POA to generate SOA (Li et al., 2024; Wang et al., 2025). The output of the OFR was directed to either a quartz or Teflon filter. Similar to previous studies (Li et al., 2024), ∼ 50 µgm-3 of POA entered the OFR and ∼ 50–100 µgm-3 of SOA was formed in the OFR. Unfortunately, there was not a concurrent measurement of aerosol mass concentration during filter collection since the entire output of the OFR was directed to the filter samples. Therefore, it is not possible to discuss the results below in quantitative terms. Since the samples contain primary emissions and secondary aerosol, we named these filters as the aged organic aerosol e.g. Fig. 1. It is helpful to classify our samples. Before starting the experiments, the background filters were collected. The filter samples are shown in Table S1. For this paper, we will mainly discuss the results from excitation emission spectroscopy and FIGAERO-CIMS measurements.

Teflon filters collected at different phases of experiments were analyzed with a filter inlet for gases and aerosols coupled to a high-resolution time-of-flight chemical ionization mass spectrometer (FIGAERO-HR-ToF-CIMS, Aerodyne Research Inc. hereafter CIMS) employing iodide (I⁻) for chemical ionization (Jiang et al., 2022; Lopez-Hilfiker et al., 2014) which provide information of individual organic compounds. In brief, particles on the Teflon filter were desorbed by a flow of ultra-high-purity nitrogen (99.9999 %) heated from room temperature to 200 °C over the course of 35 min (Lopez-Hilfiker et al., 2014; Huang et al., 2019). The resulting mass spectral signal evolutions as a function of desorption temperature are termed thermograms (Lopez-Hilfiker et al., 2014). Integration of thermograms of individual compounds yielded their signal in counts per second, which were converted to mass concentrations assuming an average maximum sensitivity of 22 counts s⁻¹ ppt⁻¹ (cps ppt⁻¹, ppt: parts per trillion) (Lopez-Hilfiker et al., 2014). Please note that the sensitivity of CIMS for different organic compounds can vary by a few orders of magnitude.

During the measurements, the mass resolution of FIGAERO-CIMS was relatively stable with about 4000 m/Δ. The interference from isomers with different vapor pressures or thermal fragmentation of larger oligomeric molecules can lead to more complex, multimodal and broader thermograms (Lopez-Hilfiker et al., 2014). The signal integration can include the different isomers or thermal fragmentation of larger oligomers. Therefore, the isomers or thermal decomposition can lead to increase errors of estimating the organic mass concentrations (Jiang et al., 2024). In this study, organic molecules of Teflon filters were identified by FIGAERO-CIMS. Please note that the iodide CIMS has sensitivities varying over several orders magnitude for different compounds, for example, different oxidation states (Lopez-Hilfiker et al., 2016). Keeping this in mind, it can still be meaningful to do a relative comparison of the large number of highly oxidized compounds assuming the same sensitivity. The raw data were analysed using the toolkit Tofware (v3.1.2, Tofwerk, Thun, Switzerland, and Aerodyne, Billerica) with Igor Pro software (v7.08, Wavemetrics, Portland, OR). Particle-phase backgrounds were assessed by putting an additional clean Teflon filter upstream of the particle phase sampling port during the deposition (Huang et al., 2019; Lee et al., 2018). In this study, we observed typically about 2800 mass peaks from particles corresponding to different oxygenated organic compounds using FIGAERO-CIMS.

Methanol-soluble organic carbon (MSOC) was extracted from the quartz filters with 5 mL methanol (for analysis purity, Merck) via ultrasonication of filter punches for 30 min. The methanol extraction had high extraction efficiency for organic compounds on the filters. The methanol extraction efficiency for organic carbon was around 85 % (Cheng et al., 2016). Obtained extracts were filtered through a 0.45 µm polytetrafluoroethylene membrane into a glass bottle to remove the insoluble material. Please note that the MSOC contains methanol-soluble but also water-soluble compounds since also water-soluble compounds are partially soluble in methanol. We did this direct methanol extraction in order to dissolve a maximum number of compounds facilitating a good comparison with the mass spectrometric analysis in which there is no differentiation between different solubilities. However, this has to be kept in mind when comparing our results with studies that separated the water and methanol soluble fractions. Absorption and excitation–emission spectra of these extracts were measured by an Aqualog fluorometer (HORIBA Scientific, USA). Please note that since the organic mass loadings were not clear, we did not calculate the mass absorption coefficients. The Aqualog measurements were described by previous studies (Jiang et al., 2022). In the brief, we used an excitation wavelength range from 239–500 nm and an emission wavelength range from 247–700 nm. The wavelength increments of the scans for excitation and emissions were 3 and 2.33 nm, respectively. The excitation emission spectrum of each sample is shown in Fig. S4. The resulting excitation–emission spectra were analyzed with the PARAFAC model to identify potential chromophoric components in MSOC.

The details of the data analysis procedure are given by Pucher et al. (2019) and Murphy et al. (2013). We used the StaRdom package for RStudio version 2026.04.0 of PARAFAC model (Pucher et al., 2019), which was downloaded from https://cran.r-project.org/web//packages/staRdom/vignettes/PARAFAC_analysis_of_EEM.html (last access: 19 February 2026). In brief, light absorption measurements were used to correct the excitation-emission matrices (EEM) for inner-filter effects. The highest absorbance was not greater than 2 (mostly below 0.5 at 237 nm), which is appropriate for inner-filter corrections of the EEMs. Afterwards, all EEMs were normalized to the Raman peak area of water so that their unit is in Raman units (RU) whose excitation wavelength was 350 nm. Additionally, an interpolation method was used to remove the signals of the first-order Rayleigh and Raman scattering as well as the second-order Rayleigh scattering in the EEMs. Using all 8 EEMs obtained for combustion emission particles in the PARAFAC analysis, three different components were adopted by comparisons of the residual errors and by visual inspection for the three- to seven component PARAFAC model (Fig. 2). They successfully passed the split-half validation with the split style of S4C6T3 for the 8 samples (Fig. S2). The corresponding model parameters and detailed information of the Split-half validation are shown in Table S2 and following contents.

The PARAFAC component intensities were normalized to the sum of components fluorescence intensities for given samples. The mass concentration of each molecule was normalized by the total mass concentration of 2871 organic molecules detected by FIGAERO-CIMS. The Spearman correlations were derived between each molecule and PARAFAC data across 8 samples (straw fresh OA, straw aged OA, plastic fresh OA, plastic aged OA, beech fresh OA, beech aged OA, cow dung fresh OA, and cow dung aged OA). The molecules correlated to PARAFAC component intensities with Spearman ≥ 0.643 (one tailed test, see the detailed information in Sect. S2) were assigned to each PARAFAC component (Tables S5, S6, and S7) (Tang et al., 2024; Stubbins et al., 2014). The molecular formulas were categorized into four groups based on their Modified Aromaticity Index (AI_mod) and H/C ratio, e.g, condensed aromatic compounds (AI_mod ≥ 0.67); aromatic compounds (0.5 < AI_mod < 0.67); highly unsaturated and phenolic compounds (H/C < 1.5, AI_mod ≤ 0.5); and aliphatic compounds (1.5 ≤ H/C ≤ 2.5).

Filter samples collected for the eight different cases were dissolved by methanol and measured their absorption and excitation-emission spectra. A PARAFAC model was used to investigate the major chromophores in methanol-soluble organic carbon extracted from the biomass-burning aerosol particles (Chen et al., 2020). With this approach, three different characteristic chromophore components were identified and named C1, C2, and C3, hereafter (Fig. 2). The peaks of excitation/emission (Ex/Em) values for C1 were at 239, 300 nm for excitation and at 372 nm for emission. Correlation analysis of PARAFAC components and Aerosol Mass Spectrometer data resulted that a similar water-soluble chromophore component can be considered as less-oxygenated humic-like substances (Chen et al., 2016). The peak of excitation/emission (Ex/Em) values for C2 was at 245 nm for excitation and at 372 nm for emissions. The maximum emission wavelength of C2 ranges above 400 nm. Similar components as C2 were found in water-soluble organic carbon and considered highly oxygenated humic-like substances (Chen et al., 2016; Yan and Kim, 2017). The maximum emission wavelength of C3 was less than 350 nm. It was located at the phenolic-like region which contains phenol and naphthalene compounds (Jiang et al., 2022), since the peaks of shorter excitation wavelength (< 250 nm) and shorter emission wavelength (<350 nm) were associated with aromatic proteins like tyrosine (Cory and McKnight, 2005). Jiang et al. (2022) found that phenol- and naphthalene-like components had a good correlation (r = 0.7) with phenol, which most likely originates from biomass burning and fossil fuel combustion. Compared with the significant similarity of the characteristic spectra identified using the PARAFAC model with literature data, C1 was considered as a less-oxygenated HULIS (LO-HULIS), C2 as a highly oxygenated HULIS (HO-HULIS), C3 as a phenolic-like substance (PLS).

Figure 3 shows that there were different relative contributions of chromophore components from burning the four different fuels. The LO-HULIS has a relatively high contribution of (45 ± 10 %) in biomass-burning fresh OA. The LO-HULIS was also identified as chromophore in brown carbon from biomass burning (straw, pinewood, and corn straw) with a contribution of 15 %–25 % (Fan et al., 2020). Tang et al. (2020) found that a similar chromophore was enriched in biomass burning aerosol with a fraction of 21 ± 6.9 %. The relative contribution of LO-HULIS was with 56 ± 7 % higher for beech wood than for the other three fuels with all nearly 40 %. However, compared to LO-HULIS, the HO-HULIS in four different fuels shows only lower contributions to fluorescence with 14 ± 8 %. This is in contrast to findings by Jiang et al. (2022) who found that the HO-HULIS components dominated (96 ± 6) % in summer and had much less but still substantial contributions of (31 ± 8) % in winter, both for urban aerosol. However, this finding somewhat agrees with the HO-HULIS fraction of about 10 % of brown carbon POA from pinewood burning (Fan et al., 2020). In contrast to LO-HULIS, the relative contribution of the PLS in four different fuels was higher with 40 ± 12 % which has an important contribution to the chromophore. This is consistent with a previous study where a similar component dominated the fluorescence in biomass-burning aerosol with about 35 %–45 % (Fan et al., 2020).

Variations of the relative chromophore contributions for the oxidized primary emissions of the four fuels are displayed in Fig. 3 as well. After aging with ozone and OH radicals during photooxidation, the relative contribution of PLS significantly decreased, from 54 % to 8 % for plastic, from 33 % to 5 % for straw, from 52 % to 9 % for beech wood, and from 43 % to 20 % for cow dung. These findings suggest that the phenolic-like chromophore is easily susceptible to photooxidation. The HO-HULIS exhibits a significant increase with photooxidation from 5 % to 46 % for plastic, from 23 % to 34 % for straw, from 12 % to 31 % for beech wood, and from 8 % to 19 % for cow dung. This indicates that excitation-emission associated with HO-HULIS increased substantially with the formation of secondary organic aerosol (SOA) and SOA is the dominant contributor to HO-HULIS. In comparison, the LO-HULIS also moderately increased, from 41 % to 47 % for plastic, from 44 % to 61 % for straw, from 36 % to 60 % for beech wood, and from 49 % to 61 % for cow dung. Taken together, the relative contributions of humic-like fluorophores (LO-HULIS and HO-HULIS) accounted for 77 %–91 % of the aged combustion BrC (aged OA), which is much higher than the range of 44 %–55 % observed for the primary samples (fresh OA). After photooxidation, the light absorption and Absorption Ångström Exponent were also significantly deceased (Fig. S5). Consistently, after photooxidation, smaller m/z organic molecules were formed from SOA formation (Fig. 1). The major difference after SOA formation is the shift in the emission spectra from lower wavelengths (333–370 nm) to higher wavlengths (462 nm).

Correspondingly, the O/C ratio increased with an enhancement of 0.1–0.4 (Fig. S3). Compared with previous studies, the O₃ oxidation of oxyaromatics could produce polyhydroxylated aromatics (Fan et al., 2020). Fan et al. (2020) also found that a component (protein-like or phenol-like organic matter, PLOM) similar as PLS was predominantly decomposed by ozone aging, while the relative fraction of highly oxygenated humic-like components significantly increased. Though, ozonolysis of biomass burning POA generated from wood combustion is relatively minor on the time scale of oxdative flow reactor, where the O : C ratio increased by ∼ 10 % (Bogler et al., 2025).

The chemical composition of organic aerosol could be assigned to the three different chromophores factors (LO-HULIS, HO-HULIS, and PLS) identified by the PARAFAC analysis based on their individual correlations. The detailed correlation and assignments are shown in Sect. S2 and Table S5, S6, and S7 in the Supplement. The properties of molecules related to the three chromophore components are given in Table 1. As shown in Table S5, 49 molecules were associated with LO-HULIS chromophore and accounted for 6.5 ± 4.2 % of total organic molecules detected by FIGAERO-CIMS. The number of LO-HULIS assocated molucules was lower than HO-HULIS and PLS. Compared with HO-HULIS and PLS, LO-HULIS molecules showed a relatively low molecular weight (156.6 Da), an low O/C ratio of 0.6, higher unsaturation degrees (DBE and AI_mod), higher level of condensed aromatic compounds (23.4 %), and higher fractions of highly unsaturated and phenolic compounds (36.0 %) (Fig. 4a, Table 1). This indicates that LO-HULIS chromophores were highly unsaturated and have high fractions of highly unsaturated and phenolic compounds. Consistently, Jiang et al. (2022) found that a similar chromophore group as HO-HULIS had a high DBE (5.8 ± 0.04), a high AI_mod (0.23 ± 0.02), and a relatively low average O/C ratio of 0.8 ± 0.01 in field measurements.

The 143 molecules associated with HO-HULIS chromophore accounted for 5.9 ± 4.4 % of total organic molecular signals detected by FIGAERO-CIMS. The number of molucules assocated with HO-HULIS was higher than LO-HULIS and PLS. Compared with LO-HULIS and PLS, HO-HULIS molecules showed a middle level of molecular weight (171.9 Da), higher O/C ratio (0.8), higher OSc (0.3), higher fraction of nitrogen containning molecules, and lower fractions of highly unsaturated and phenolic compounds (21.3 %) (Fig. 4b, Table 1). Therefore, it indicates that the HO-HULIS chromophore was more oxidized and contained high fraction of nitrogen containing molecules. Compared with previous studies, a samilar chromphore as HO-HULIS shown that the average O/C ratio, molecular weight, DBE, and AI_mod were 0.9 ± 0.01, 170 ± 1 Da, 3.6 ± 0.03, and 0.16 ± 0.01, respectively (Jiang et al., 2022). Tang et al. (2024) found that a similar chomophore as HO-HULIS had more than 60 % of nigrogen containing molecules. In addition, the HO-HULIS chromophore was associated with more compounds that possessed higher aromaticity (AI_mod > 0.5) and higher oxidation state (OSC > 0) (Tang et al., 2024).

The 108 molecules associated with PLS chromophore accounted for 12.8 ± 7.6 % of total organic molecules detected by FIGAERO-CIMS. Compared with LO-HULIS and HO-HULIS, these molecules show a higher molecular mass (206.7 Da), lower O/C ratio (0.6), higher H/C ratio (1.5), lower unsaturation degrees (DBE and AI_mod), lower OSc (-0.4), low fraction of nitrogen containing molecules (2 %), and higher fraction of aliphatic compounds (Fig. 4c, Table 1). In addition, the associated molecules of PLS chromophores had some monomers or dimmers, e.g., C₁₆H₁₄O₂, C₁₈H₃₂O₂, C₁₈H₃₄O₂, C₁₈H₃₅O₂N₃. Therefore, it indicates that the PLS chromophore was low oxidation state, contained lower fraction of nitrogen containing molecules and higher fraction of aliphatic compounds. Compared with previous stuides, the PLS chromophore was not exclusively associated with nitrogen-contianing compounds (Tang et al., 2024; Stubbins et al., 2014). Tang et al. (2024) found that PLS chromophore in a field measurment was consisted of lowconjugation, nitrogen-depleted, and oxygen-depleted species.

After photooxidation, the relative contents of the PLS chromophore decreased. The photooxidation of PLS and volatile organic compounds in presence of NO_x could lead to form the HO-HULIS and LO-HULIS (Fig. 3). Please note that the NO_x was generated during the burning. According to previous studies, the concentrations of NO_x were around a few hundred ppm (Winter et al., 1999; Courtemanche and Levendis, 1998). The conceptual illustration is shown in Fig. 5. The PLS chromophore contained molecules with lower-conjugation, lower nitrogen content, and lower oxygen content. During the photooxidation, the PLS molecules were oxidized by O₃ and OH radicals in presence of NO_x. The chemcial bonds in large molecules (C_9.9H_14.8N_0.05O_4.5) were broken. In addition, the photooxidation of VOCs in presence of NO_x can formed the small molecules with higher fractions of nitrogen-containing molecules and higher oxidation state (Li et al., 2024). Consequencely, they led to form LO-HULIS (C_7.2H_7.7N_0.1O_3.8) and HO-HULIS (C_7.3H_8.6N_0.2O_4.6) with higher O/C ratio and lower H/C ratio. Consistently, the O₃ oxidation of oxy-aromatics can cause the cleavage of the aromatic bond (C=C) to generate polyfunctional low-molecular-weight carboxylic acids (C=O) and also formation of polyhydroxylated aromatics (phenol-OH) (Fan et al., 2020). However, evolutions of chromophore were different at ambinent environments compared with primary aerosol at oxidative flow reactor experiments. For example, the photochemical reaction and oxidation reaction drive degradation of less-oxygenated humic-like substances and led to forming highly-oxygenated humic-like substances (Chen et al., 2021a, b). This study has not yet achieved quantitative characterization of each individual process (the degradation of large molecules and oxidation of VOCs) contributing to HULIS chormophore formation. Future studies need to integrate isotopic techniques or mass spectromters to address this limitation and accurately quantify the contribution of each process.