BBOA particle samples were collected during the FIREX Fall 2016 lab intensive at the Missoula Fire Lab (https://www.esrl.noaa.gov/csd/projects/firex/firelab/, last access: 1 May 2019). One of the BBOA samples used in this study was from a “stack” burn and the other samples were from “room” burns. Selimovic et al. (2018) explains room and stack burns and fuels in detail. Briefly, the combustion of forest fire fuels lasted 5–20 min and during stack burns emissions were collected from a constant, diluted flow of entrained emissions by way of the stack. In room burns, the smoke from the fire was allowed to mix in the room during sample collection, and BBOA was collected during both the burn and mixing periods. Smoke was purged from the room by clean air between burns. Fuels were collected from different US regions and brought to the Missoula Fire Lab for test burns. This paper focuses on 12 fires covering gymnosperm or conifers, including ponderosa pine (Pinus ponderosa), lodgepole pine (Pinus contorta), Engelmann spruce (Picea engelmannii), Douglas fir (Pseudotsuga menziesii), juniper (Juniperus), longleaf pine (Pinus palustris), and subalpine fir (Abies lasiocarpa). Angiosperm forest fire fuels included Montana sagebrush and two types of chaparral, i.e., manzanita (Arctostaphylos) and chamise (Adenostoma fasciculatum). In some test burns, a representative “ecosystem” mix of biomass was used, including canopy, duff, litter, herbaceous, and shrub components. In other test burns, single biomass components of the ecosystem were used, such as rotten log samples. Information for each fire is provided in Table S1 in the Supplement.

Copper tubing with a PM2.5 cyclone inlet was placed in the combustion room, while the pump and filter were located in an adjacent room. The pump was operating at a flow of 16.7 L min-1 with the aid of a critical orifice, and BBOA particle samples were collected on PTFE filter substrates (FGLP04700, Millipore, 47 mm diameter, 0.2 µm pore size) during both of the combustion and smoke-mixing stages of the room burns. Loaded filters were stored at -18 ∘C until they were analyzed for BrC chromophores no more than 2 months after sampling. The room burn protocols allowed for long collection times and therefore higher aerosol mass loading, which is desirable for the analysis described below.

The molecular identity and relative abundance of BrC chromophores were determined using the HPLC–PDA–HRMS platform described by Fleming et al. (2018) and Lin et al. (2018). Segments of the filter were extracted into a mixture of organic solvents composed of 2.0 mL dichloromethane, 2.0 mL acetonitrile, and 1.0 mL of hexanes, which was shown to optimize the extraction efficiency (Lin et al., 2017). The extraction occurred overnight on a platform shaker. Extracts were filtered with polyvinylidene fluoride (PVDF) syringe filters (Millipore, Duropore, 13 mm, 0.22 µm) to remove undissolved suspended particles. Water (50 µL) and dimethyl sulfoxide (DMSO; 100 µL) were added to the extracts, which were then concentrated under a flow of N2 until the volume was reduced to roughly 150 µL, which signified that the extracting solvent evaporated and (mostly) water and DMSO remained in the solution. For photolyzed BBOA, DMSO (30 µL) was exclusively added to the extract, and evaporated to a volume of 30 µL. Visual inspection confirmed that the extracted material did not precipitate out of solution.

The HPLC utilized a reverse-phase column (Luna C18, 2×150 mm, 5 µm particles, 100 Å pore size, Phenomenex, Inc.). The injection volume was 5.0 µL for unphotolyzed or 10 µL for extractions of post-irradiated samples, with the latter providing more analyte mass since only a quarter of the filter was used in irradiation experiments. The mobile phase consisted of 0.05 % formic acid in liquid chromatography–mass spectrometry (LC–MS) grade water (A) and LC–MS grade acetonitrile (B). Gradient elution was performed with the A–B mixture at a flow rate of 200 µL min-1: 0–3 min hold at 90 % A, 3–62 min linear gradient to 10 % A, 63–75 min hold at 10 % A, 76–89 min linear gradient to 0 % A, 90–100 min hold at 0 % A, then 101–120 min hold at 90 % A. The electrospray ionization (ESI) settings of the Orbitrap HRMS were as follows: 4.0 kV spray potential, 35 units of sheath gas flow, 10 units of auxiliary gas flow, and 8 units of sweep gas flow. The solutions were analyzed in both positive and negative ion ESI-HRMS modes.

The HPLC–PDA–HRMS data were acquired and first analyzed using Xcalibur 2.4 software (Thermo Scientific). Possible exact masses were identified based on the corresponding LC retention time using the open-source software toolbox MZmine version 2.23 (http://mzmine.github.io/, last access: 28 July 2017) (Pluskal et al., 2010). Chemical formulas were assigned from exact m/z values using the Formula Calculator v1.1. More details about experimental procedures and data processing can be found elsewhere (Lin et al., 2015b, 2016, 2018).

A quarter of the filter was directly irradiated by either an ultraviolet light-emitting diode (LED, Thorlabs M300L4) or a filtered xenon arc lamp. The LED was used in experiments aimed at estimating lifetimes of individual chromophores. The LED emission spectrum was centered at 300 nm with a full width at half maximum (FWHM) of 20 nm. This wavelength was chosen because it corresponds to the most energetic UV photons available in the lower troposphere. It is common practice in photochemical experiments to use narrow band UV sources, as opposed to a broadband simulator, as it limits sample heating and evaporation (Calvert and Pitts, 1966). The LED was fixed half a centimeter away from the filter, resulting in an incident power density of 11 mW cm-2. Irradiation times for these experiments are given in Table S2. After the irradiation step, the photolyzed BBOA were extracted and analyzed using HPLC–PDA–HRMS as described in the previous section.

The irradiation time using the LED was converted into an equivalent time under sunlight by calculating the ratio of the 290–350 nm integrated spectral flux of the Sun and the 300 nm LED, given in Eq. (1). This conversion assumes that photochemistry is limited to the <350 nm range, consistent with the photochemistry of many organic molecules, which exhibit a sharp drop in the photochemical quantum yields at longer wavelengths (Turro et al., 2009). Because the radiation source does not replicate the solar spectrum, the lifetimes calculated from the formula below should be regarded as estimates. 1τatm=τLED×∫290nm350nmFLEDλdλ∫290nm350nmFsolarλ24hdλ The spectral flux density for the LED and the Sun as a function of wavelength is shown in Fig. 1. The solar flux density was estimated every hour and averaged over a 24 h period for Los Angeles, CA (34∘ N, 118∘ W), on 20 June 2017 from the quick Tropospheric Ultraviolet Visible (TUV) calculator (Madronich et al., 2002) using the following parameters: 300 DU overhead ozone column, 0.1 surface albedo (0–1), and ground elevation of 0 km with default outputs for aerosols and clouds. The procedure for calculating the spectral flux density of the LED is described in the Supplement. The maximum possible spectral flux density from the Sun was also calculated at a solar zenith angle (SZA) of 0∘ using the TUV calculator. The equation for calculating the equivalent atmospheric lifetime at an SZA of 0∘ is the same as Eq. (1), except that the 24 h averaged flux density is replaced by the peak flux density at SZA = 0. The SZA = 0∘ comparison represents the lower limit of BrC absorption lifetimes.

In a separate series of experiments, filter samples were irradiated by the filtered radiation from a xenon arc lamp to determine the characteristic lifetime for the photobleaching of the overall absorption by BrC molecules. A quarter of a PTFE filter sample was exposed to filtered light emitted from a xenon arc lamp (Newport 66902). Broadband light was reflected at a 90∘ angle using a dichroic mirror, then filtered through a 295 nm long-pass filter (Schott WG295), and finally passed through a UV bandpass filter (Schott BG1), ultimately transmitting light in the range of 290–400 nm. The incident overall power density was 196 mW cm-2. Particles were irradiated for ∼12 h to 1.8 d; the exact time varied from sample to sample depending on the offline transmission spectra. Transmission spectra were acquired directly from the PTFE filter without any material extraction using a Jasco V-670 absorption spectrometer, with a blank PTFE filter used as a reference. Four to six transmission spectra were collected at each time point as the filter was rotated, to minimize the effect of the filter orientation. The filter was then returned to the irradiation setup for further irradiation. When there was no longer any change in the transmission spectrum due to irradiation, the filter was extracted into an organic solvent mixture of 10 mL methanol, 5.0 mL acetonitrile, and 2.0 mL of hexane in a scintillation vial using a vortex mixer. While dichloromethane would be a better solvent for BBOA material, methanol was used for these experiments, since dichloromethane absorbs at longer wavelengths in the UV (up to 240 nm) and could interfere with the measurement. The solution was then evaporated down to 5 mL in order to increase the analyte concentration. For comparison, an un-irradiated quarter of the filter was prepared identically in a separate vial, and solution-phase transmission spectra of both solutions were recorded using a dual beam UV–Vis spectrometer (Shimadzu UV-2450). Sample filter-based and solution-phase spectra are shown in Fig. S2, with the y axis converted to effective base-10 absorbance, A=-log⁡(T), where T is the wavelength-dependent transmittance through the filter or the cuvette. For filter-based transmission spectra, the baseline was manually corrected by assuming the absorbance at 850 nm was zero for BrC.

In all UV irradiation experiments, the integrated absorbance from 300 to 700 nm was calculated and normalized to the starting integrated absorbance before the UV exposure. The decay constants and corresponding lifetimes were calculated as described in Fig. S1. The linear regression trend line was constrained to have a y intercept of zero. Error bars were calculated from the standard error of the slope of the linear trend line, the first-order rate constant. It should be noted that lifetimes of BrC absorption and chromophores given in this paper are lower-limit estimates since there are uncertainties due to scattering by the Teflon substrate (Presser et al., 2014).

Table 1 summarizes BrC chromophores observed in two or more fires or fuel types. The table numbers BrC chromophores by their ascending retention time on the HPLC column, i.e., with smaller, more polar compounds appearing first. Each entry includes the absorption spectrum recorded by the PDA detector, the chemical formula(s) corresponding to the detected characteristic masses at that retention time, and a potential structure based on a spectra acquired from standards or observations in previous studies. All PDA chromatograms were integrated over 300–700 nm and normalized to the maximum integrated absorbance. Chromophores in Table 1 are binned with respect to their normalized PDA absorbance as M – major (75 %–100 %); I – intermediate (25 %–75 %); or W – weak (5 %–25 %). Abundance and absorption cross sections of BrC chromophores both factor into their assigned absorbance bin, as absorbance was not mass normalized with standards. It is possible that the chromophores labeled as “M” are present in small concentrations but have a large absorption coefficient. Compounds making up less than 5 % of the normalized absorbance are not included in the table.

Lignin pyrolysis products make up one group of BrC chromophores observed. Lignin is a large, heterogeneous biopolymer that is a significant component of wood, along with cellulose and hemicellulose. Lignin monomer units vary depending on the class of the plant but generally possess phenolic moieties that are largely preserved during pyrolysis (Simoneit et al., 1993). Sinapaldehyde (8) and coniferaldehyde (9) are known lignin pyrolysis products derived from the corresponding lignin monomer units, sinapyl and coniferyl alcohol, respectively. However, they are detected in varying abundance depending on the lignin monomer units of the plant class. Sinapaldehyde and coniferaldehyde are separated by the column but elute only 0.3 min apart, as shown in Fig. 2. Sinapaldehyde is a major BrC chromophore for nearly all angiosperm or flowering fuel types, including ceanothus, chamise, and sagebrush, while coniferaldehyde is a major BrC chromophore largely among conifers or soft wood species such as subalpine fir duff, longleaf pine, juniper, and ponderosa pine litter. Coniferaldehyde has one fewer methoxy ring substituent compared to sinapaldehyde, and its PDA intensity is generally anticorrelated to that of sinapaldehyde. In other words, for fuel types with low sinapaldehyde absorbance, we observe coniferaldehyde as a major BrC chromophore and vice versa. This is consistent with the composition of lignin monomers for angiosperms and gymnosperms (Sarkanen and Ludwig, 1971; Simoneit et al., 1993).

Other BrC chromophores cannot be classified as lignin pyrolysis products but are clearly lignin-derived. Vanillic acid (1) elutes at 10.07–10.29 min as the first, shared chromophore across multiple fuel types that is notable in terms of absorption. It is observed in three fires as a weak chromophore, including subalpine fir duff, ponderosa pine rotten log, and Engelmann spruce duff. All three fires are dominated by smoldering combustion and have the lowest modified combustion efficiencies (MCEs) of all fires (Table S1). This evidence suggests that vanillic acid is a product of smoldering combustion. Further, it also has the coniferyl moiety observed for softwoods. Salicylic acid (3) is an intermediate-absorbing BrC chromophore produced during lodgepole pine burning, and weakly absorbing among other softwoods and duffs. Veratraldehyde (4) is another lignin-derived BrC chromophore, which appears in nearly all BBOA samples of this study, regardless of whether they are gymnosperm or angiosperm fuels.

There are other BrC chromophores with CxHyOz composition that can be explained as distillation products, or the volatilization of molecules originating in plants as secondary metabolites (Agati et al., 2012; Iranshahi et al., 2009). Found in plants, coumarins such as umbelliferone (5) and nodakenetin (13) have been researched because of their positive pharmacological properties (Venugopala et al., 2013). The absorption spectrum for nodakenetin has not been reported; however, the molecule has previously been detected in plant tissues (Lee et al., 2003; Wang et al., 2014) and is a major or intermediate BrC chromophore in smoke from all fuel types except chamise and ceanothus. Another type of distillation product is flavonoids, which give leaves, flowers, and fruits their color, protecting the plant from solar UV radiation, and are antioxidants, guarding the plant from reactive oxygen species (Agati et al., 2012). Flavones and flavonols have the backbone structure of 2-phenyl-1-benzopyran-4-one, and flavonols additionally require a hydroxy substituent on the only available carbon of the pyranone ring. BrC chromophores 11, 14, and 16 could have flavonoid structures based on their chemical formulas. Interestingly, tentatively assigned kaempferol (11) and diosmetin (14) are observed in only conifer species, such as lodgepole pine and longleaf pine. On the other hand, 7-hydroxy-3',4'-dimethoxyflavone (16) is only observed in angiosperm BBOA: ceanothus, chamise, and sagebrush. The former two plants appear to be related as they have the order Rosales in common, which could explain the same flavone detected in both. Coumarins and flavonoids were distillation products observed across fuel types, although the observation of specific BrC chromophores depends on the plant class, i.e., angiosperm or gymnosperm.

Nitroaromatics are a strongly absorbing class of BrC chromophores that are formed from the reaction of aromatics with NOx in plumes (Harrison et al., 2005). This class of compounds is represented in Table 1 with nitropyrogallol (2), nitrocatechol (6), hydroxynitroguaiacol (7), and methyl nitrocatechol (10). Xie et al. (2019) suggest that chromophore (12) with the chemical formula C11H13NO5 is not a nitroaromatic compound but rather a compound containing a different nitrogen-containing functional group, such as a nitrile group. We did not observe this group of chromophores for fires with low NOx levels, such as duff, as qualitatively indicated by the peak NO level (Table S1). Nitrocatechol and methyl nitrocatechol are tracers for BBOA emissions formed from the photooxidation of phenol or m-cresol, toluene, and other aromatic compounds in the presence of NOx (Iinuma et al., 2010, 2016; Lin et al., 2015a). These chromophores are most prominent in BBOA from chamise and sagebrush burns. Those two fires exhibited the highest NO mixing ratios in the entire study – 3.79 ppmv (82 % of total N emissions) and 1.62 ppmv (57 % of total N emissions) peak NO values, respectively. Nitropyrogallol (2) has an additional hydroxy group and is likely formed in the same way as nitrocatechol and methyl nitrocatechol but is more oxidized. A compound with the same formula as nitropyrogallol (2) was observed during the photooxidation of nitrocatechol in the lab (Hems and Abbatt, 2018). This is an intermediate or major BrC chromophore detected in BBOA samples from longleaf pine, manzanita, and ponderosa pine litter fires. Hydroxynitroguaiacol (7) was observed in 10 of the 12 fires and is most prominent in ponderosa pine log BBOA despite this fire having the lowest NO levels. However, it may still form through photooxidation of guaiacol in the presence of NOx (Hems and Abbatt, 2018). Nitrocatechol and methyl nitrocatechol are often used as biomass burning tracers in aged plumes (Al-Naiema and Stone, 2017; Iinuma et al., 2010; Li et al., 2016). However, in addition to these, we observed more oxidized versions of these nitroaromatic species with varying abundance depending on the BrC chromophore and test fire. This suggests that the BBOA markers nitrocatechol and methyl nitrocatechol become more functionalized on relatively short timescales (less than 2 h) due to photooxidative aging.

Polycyclic aromatic hydrocarbons (PAHs) are known to be products of incomplete combustion, and they have the potential to be long-lived BrC chromophores despite their reactivity (Keyte et al., 2013). PAHs have been observed in pristine environments, and it has been suggested that this is due to phase separation of particles and slow diffusivity of PAHs to surfaces where they react with atmospheric oxidants (Fernández et al., 2002; Keyte et al., 2013; Macdonald et al., 2000; Sofowote et al., 2011; Zhou et al., 2012, 2019). In addition to its climatic effects, PAHs are mutagenic and carcinogenic as their metabolites, diol epoxides, bind to guanidine nucleobases in DNA, effectively leading to mutations (Finlayson-Pitts and Pitts, 2000; Moorthy et al., 2015; Wood et al., 1984; Xue and Warshawsky, 2005; Zhou et al., 2017). Various PAHs (17–25, Table 1) were observed in only ceanothus, chamise, and sagebrush BBOA. PAHs in Table 1 are detected from positive ion mode ESI, and although positive mode ESI is not optimal for observing PAHs, larger PAHs are still detectable by this method (Cha et al., 2018). The same PAHs were previously observed by Lin et al. (2018) for sagebrush using atmospheric pressure photoionization (APPI) coupled with HPLC–PDA–HRMS, which is more sensitive for the detection of nonpolar aromatic compounds. In general, individual PAH chromophores are binned as “weak” in Table 1 based on their contribution to optical absorption, but, for BBOA sampled from flaming sagebrush and chamise burns, they make up a significant fraction of the overall light absorption by BrC.

Table 2 presents abundant BrC chromophores observed only in a single type of biomass fuel emissions. It should be noted that compounds making up less than 5 % of the normalized PDA absorbance (integrated from 300 to 700 nm) are not included in the tables. Due to this constraint, chromophores in Table 2 may also be present in other fires but at very low PDA absorbance values. Despite BrC chromophores in Table 2 being observed significantly for only one fuel type, they belong to the same compound classes as the BrC chromophores in Table 1. For example, a coumarin known as scopoletin (26) was observed from sagebrush BBOA. Previously we discussed that these coumarins are possible distillation products, along with flavonoids, which we also observe as a product (40) from the ceanothus fire. These distillation products (26 and 40) are among the most strongly absorbing of the BrC chromophores, characterized as intermediate or “I” in Table 2.

Gymnosperm (lodgepole pine) and angiosperm (ceanothus) BBOA particle samples were selected for the initial condensed-phase photochemistry experiments. BBOA filter samples from a lodgepole pine burn were irradiated for 6 h by an LED centered around 300 nm (which corresponds to approximately 33 h of irradiation from 24 h average solar flux density; see Eq. 1). BBOA from the ceanothus burn were irradiated by the same LED for 16 h (equivalent to 88 h of 24 h averaged atmospheric sunlight). The burning of gymnosperm (lodgepole pine) and angiosperm (ceanothus) resulted in different distributions of BrC chromophore classes. However, the same compound classes, lignin-derived and flavonoid compounds, were photo-resistant in both samples.

Most chromophores from the lodgepole pine burn sample experienced complete photobleaching during this exposure, but six of them remained observable, including coniferaldehyde (C10H10O3, 80 % decrease), salicylic acid (C7H6O3, 70 % decrease), veratraldehyde (C9H8O3, 90 % decrease), flavonoids (C15H10O6 and C16H12O6, both 70 % decrease), and nodakenetin (C14H14O4, 90 % decrease), as shown in Fig. 3. Figure 4 shows five chromophores from the ceanothus burn sample that remained observable under these conditions, including sinapaldehyde (C11H12O4, 90 % decrease), a lignin-derived chromophore (C18H16O6, 80 % decrease), and flavonoids (C16H12O5, C17H14O6, and C17H14O5, all 80 % decrease), some of which were observed exclusively in this fire. These comparatively resilient species are aromatic, which helps them be more resistant to photodegradation.

Next, we estimate the lifetime of individual BrC chromophores in BBOA. For chamise, manzanita, and lodgepole pine fires we measured the integrated PDA intensity over 300–700 nm for chromatographically separated BrC chromophores in the starting samples and for up to three irradiation time points (listed in Table S2). The limited number of samples and destructive nature of the chemical analysis only made it possible to do measurements for very few time points. Integrated PDA intensities as a function of irradiation time were fit assuming that the decay was exponential in time. LED lifetimes were then converted to equivalent lifetimes in the atmosphere, calculated from the average spectral flux density over 20 June 2017 in Los Angeles. It should be noted that due to scattering of light by the Teflon filter substrate, which effectively increases the absorption efficiency of particles trapped on the filter, lifetimes in Fig. 5 are lower limits (Presser et al., 2014). Regardless of the chromophore identities, BrC chromophores from chamise burns have shorter predicted lifetimes (0.4–0.5 d) than those from manzanita burns (0.5–0.9 d), which in turn have shorter predicted equivalent atmospheric lifetimes due to sunlight exposure than BrC from lodgepole pine burns (1.0–1.6 d), as shown in Fig. 5. These lifetimes of BrC chromophores are consistent with atmospheric observations of a rapid evolution in a California wildfire, which showed that the BrC absorbance lifetime at 370 nm was 9–15 h (Forrister et al., 2015).

The same chromophores were found to decay at different rates depending on the fuel/fire type (Fig. 5). For example, very different equivalent atmospheric lifetimes due to UV irradiation were obtained across fuel types for veratraldehyde (no. 4 in Table 1, C9H8O3), a BrC chromophore common to all three fires. One explanation is that there are multiple chromophores co-eluting at this retention time, and therefore the calculation is an average lifetime for multiple compounds. A more interesting explanation is that the surrounding matrix could affect the rate of condensed-phase photochemical transformations for individual chromophores by several possible mechanisms. First, different matrices could quench the electronic excitation in the chromophores to a different extent. Another possibility is that photodegradation of BrC chromophores could be not due to direct photolysis but rather occurring through condensed-phase photosensitized reactions (Malecha and Nizkorodov, 2017; Monge et al., 2012), in which case the rate of decomposition would depend on concentration of photosensitizers in the samples as well as viscosity of the material (Hinks et al., 2016; Kaur et al., 2019). Lastly, other absorbing species, such as black carbon, could be shielding BrC chromophores from irradiation, altering the amount of radiation absorbed by BrC chromophores. Given the different mechanisms, the potential contributions from each are difficult to distinguish in this study. The particle matrix is different for all three BBOA particle samples and could contribute to the very different equivalent atmospheric lifetimes of individual BrC chromophores observed in Fig. 5.

We also estimated the decay lifetime for the overall BrC absorption, integrated over 300–700 nm, from different fuel types. In these experiments, BBOA filters were irradiated with a filtered xenon arc lamp, which gave a spectral flux density more similar to the Sun, although more intense (Fig. 1). The advantage of taking transmission spectra directly through the filters is that it makes it possible to monitor photodegradation of BrC absorption at several irradiation times, which is not possible with the solution-phase spectrophotometry, which irreversibly destroys the filter sample by extraction. The filter transmission spectra indicated that the decay of absorbance was not actually exponential. After a certain irradiation time, the BrC absorbance no longer decreased, as observed for the samples from subalpine fir and longleaf pine burns. For example, in Fig. S2, after 21 h the recalcitrant or “baseline BrC” level has already been reached, as revealed by the next measurement at 33 h. The absorbance decreased 70 % before it reached the baseline BrC level for subalpine fir, and 60 % for longleaf pine. For estimates of the BrC absorbance lifetimes, we used only the time before reaching the final light-absorbance state. Table 3 summarizes the resulting lifetimes for BrC from four fuel types, longleaf pine, juniper, lodgepole pine, and subalpine fir. Once again, it should be noted that BrC absorption lifetimes are lower limits, due to the enhanced efficiency of absorption by particles caused by scattering of UV radiation scattering by the Teflon filter substrate (Presser et al., 2014).

Once there was no further significant change in the transmission spectrum, the filter was extracted for the solution-phase UV–Vis measurement, in order to compare the spectra obtained from the filter and in the solution. The reduction in absorbance in the solution-phase spectra was comparable to that observed in the filter transmission spectra (Fig. S2). However, there were differences in the shape of the spectra – there was no measurable absorbance above 550 nm in the extracted samples, but filter samples absorbed even at these long wavelengths (Fig. S2). It is likely that the extraction from the filter was not complete, and some of the absorbers remained on the filter after the extraction. The latter is another advantage of doing these experiments with filter samples as opposed to their solvent extracts.

BBOA from subalpine fir (litter and other components) had the shortest equivalent absorption lifetime at 10 d, and ponderosa pine (litter and canopy) had the next shortest equivalent absorption lifetime at 17 d. Different ecosystem biomass components were burned in the longleaf pine fire, such as duff, litter, and canopy, and had the next longest absorption lifetime of 25 d. The longest living BrC absorbance, at 41 d, was observed for the sample from juniper (canopy only) burn. Fuel components appear to affect BrC absorption lifetimes, as it does seem that non-canopy fuel components, such as litter and duff, lower the BrC absorption lifetimes. However, it is difficult to correlate the BrC absorption lifetimes with quantitative measures such as NO levels or MCE (Table S1). Table S1 shows that the peak NO level was lower for longleaf pine (0.67 ppmv) compared to juniper (1.72 ppmv) and ponderosa pine (1.61 ppmv), suggesting less flaming combustion may have occurred for the longleaf pine fire (although this is not reflected in the MCE trends). Regardless, the data suggest that BrC absorption can be long-lived from direct photodegradation.

In general, the lifetimes for the loss of the absorbance integrated over 300–700 nm (Table 3) are much longer than those of individual chromophores (Fig. 5). There are two likely reasons for that. First, the photochemical transformation of individual chromophores creates product(s) that may also absorb in the same wavelength range. The integrated BrC absorption (300–700 nm) may significantly decrease only after the compounds go through several successive stages of photodegradation, finally resulting in products that no longer absorb above 300 nm. The results of both UV irradiation experiments is consistent with work by Di Lorenzo et al. (2017) and Wong et al. (2017), which show that during aging, high-molecular-weight BrC chromophores are formed after lower-molecular-weight chromophores are photo-degraded. The high-molecular-weight fraction of BrC chromophores persists even at long aging times and are referred to as the recalcitrant fraction. This theory is one explanation for the short lifetimes of low-molecular-weight BrC compounds, while observing longer overall BrC absorption lifetimes. Second, Eq. (1), which we use to estimate lifetimes, does not take into account photochemical quantum yields, which tend to increase greatly at shorter wavelengths. The LED, which was used in measurements of lifetimes of individual chromophores, has a higher density of higher-energy photons compared to the Xe lamp (Fig. 1), which could accelerate the observed photodegradation rate.

The lifetimes for BrC photobleaching due to UV irradiation (10 to 41 d) are longer than what other studies have observed or approximated for other aging mechanisms. Lin et al. (2016) found that peat and ponderosa pine BBOA had similar half-lives of around 16 h based on absorption coefficients at 300 nm. However, in Lin et al. (2016), BBOA was extracted and irradiated in solution where photodegradation could occur more rapidly due to molecular diffusion (Lignell et al., 2014). Forrister et al. (2015) collected filter samples in the plumes of wildfires with different transport times during the SEAC4RS campaign and found that the BrC absorbance lifetime at 370 nm was 9–15 h. Similarly, Selimovic et al. (2019) found a significant decrease in the absorption Angstrom exponent after 10 h of daytime aging during a wildfire event in the northwestern US. Sumlin et al. (2017) aged smoldering peat BBOA in an OFR and reported a decrease of ∼40 %–50 % in the aerosol mass absorption coefficients at 375 and 405 nm after 4.5 equivalent aging days. They attributed this decrease to fragmentation of BrC chromophores due to photooxidation (oxidation by gaseous OH). Based on the comparison of these observations, photooxidation could be a more important aging mechanism affecting BrC absorption lifetimes than the UV-induced photochemical processes inside the particles.