Photolytic aging of organic aerosol from pyrolyzed urban materials

Emissions from large-scale fires significantly contribute to the atmospheric burden of primary organic aerosol (OA). The frequency of fires occurring at the wildland-urban interface (WUI) is increasing, with biomass and...


Scheme S1.
Experimental flowchart (a) for pyrolysis experiments conducted with all ten urban materials (separately).Organic aerosol generated through pyrolysis is labeled as "OA".All OA samples were generated in a tube furnace at 600 °C under N2 (b).First, experiments were conducted by collecting on Teflon filters (top arrow).An evaluation of solvent effects was performed to determine the best solvent for extraction, as each material yielded OA with unknown composition.Each filter was cut into four equal pieces, and each was extracted in a different solvent: water, methanol, octanol, and orgmix (acetonitrile/dichloromethane/hexane 2:2:1 by volume).UV-vis absorption spectra were utilized to determine which solvent was best suited per material.Photolysis experiments (downward arrow) were conducted on separately collected OA.Similar to the solvent investigations, each filter was cut into quarters and irradiated with various times (short-term aging and long-term aging).Subsequent extraction and UV-Vis absorption measurements were utilized to determine how irradiated fractions differed from the non-irradiated fraction.A schematic of the laboratory pyrolysis setup is shown in panel b.
Table S1.Summary of urban materials pyrolyzed in this study.All pyrolysis runs were conducted at 600 °C unless noted otherwise.All values are reported as the average amongst three replicates with uncertainty values reported to one standard deviation.Emission factors (from Teflon filter collection experiments) are reported in the units of grams of collected OA per kilogram of urban material utilized in pyrolysis.Approximately 200 mg of urban material was pyrolyzed in each experiment.Carpet pyrolyzed at 800 and 1000 ° C were only tested once (no uncertainty values reported).Best extraction solvent was determined based on greatest overall absorbance and expression of spectral features (Figure S3).Each quarter of the filter was aged separately for 0.5 h (red), 1 h (blue), or 2 h (grey) and is compared with the unaged fraction (black).The integrated MAC values (normalized to the unaged value) are shown as a subplot in each panel, with the UV integrated over 280-400 nm and the visible integrated over 400-700 nm.Boxcar averaging of the MAC data (over 5 nm range) was utilized to improve signal-to-noise ratio.Each quarter of the filter was aged separately for 0.5 h (red), 1 h (blue), or 2 h (grey) and is compared with the unaged fraction (black).The integrated MAC values (normalized to the unaged value) are shown as a subplot in each panel, with the UV integrated over 280-400 nm and the visible integrated over 400-700 nm.Boxcar averaging of the MAC data (over 5 nm range) was utilized to improve signal-to-noise ratio.Figure S13e shows the absorption spectrum for thin PVC wire OA (purple trace, Figure S13e) as originally presented in Figure 1 of the manuscript.This spectrum exhibits distinct absorption bands at 253 nm, 290 nm, and 300 nm.Phthalic acid standard was dissolved in orgmix and an absorption measurement was taken using a UV Vis spectrometer (black trace, Figure S13e).Thin PVC wire OA absorption spectrum from the PDA detector of the HRMS (dashed line, Figure S13e) and phthalic acid standard solution exhibited the same spectral features at 230 nm, 275 nm, and 283 nm.As these peak wavelengths do not match that of the thin PVC wire OA (purple trace), we conclude that phthalic acid is not the compound responsible for the distinctive absorption in thin PVC wire OA, however, it is responsible for most of the overall sample's absorption in the PDA.Follow up studies on urban OA chemical composition will look further into identifying the compound responsible for absorption bands at 253 nm, 290 nm, and 300 nm.The integrated MAC values (normalized to the unaged value) for the UV integrated over 280-400 nm are shown in the subplot, with the UV integrated over 280-400 nm.The drywall OA aged samples bleached similar amount, within 5% of each other.Consistent with previous laboratory drywall OA experiments, photolytic exposure darkens this urban material's pyrolysis products.This is represented in the pictures of the unaged OA vs. aged OA filter and the "U" cutout which darkened during the sunlight exposure.

Figure S1 .
Figure S1.Sample photographs of filters containing organic aerosol (OA) generated through the pyrolysis of ten urban materials at 600 °C under N2.The OA differed in appearance amongst urban materials.

Figure S2 .
Figure S2.Sample photographs of filters containing OA generated through the pyrolysis of carpet at 600, 800, and 1000 °C under N2.Pyrolysis temperatures greater than 600 °C yielded OA that was poorly soluble due to presence of black carbon (BC).

Figure S3 .
Figure S3.UV-Vis absorption spectra recorded after the pyrolysis of urban materials (a-j) and extraction in orgmix (grey), methanol (red), water (blue), and n-octanol (green).The MAC values assume complete extraction of OA from the filter.The lowest MAC observed for water indicates that it does not extract the OA fully and is the least effective solvent tested for all materials.Plywood extracts were diluted by a factor of ten for absorbance measurements.

Figure S4 .
Figure S4.The spectral photon flux density (F(λ)) over the actinic range of the Xenon arc lamp used in laboratory photolysis experiments (dashed line) compared to the averaged solar spectrum in Los Angeles, California (Latitude/Longitude: 34°/118°) for November 9 th between the hours of 11 AM to 1 PM (solid line).Panel (a) shows F(λ) as a function of wavelength.Panel (b) shows a product of spectral photon flux density and the mass absorption coefficient average across all 10 unaged urban OA (F(λ) × MACaverage), which better illustrates the wavelength at which photolysis is taking place.The "Quick TUV" calculator [Madronich, S. ACOM: Quick TUV https://www.acom.ucar.edu/Models/TUV/Interactive_TUV/(accessed November 03, 2022)] was used to estimate the spectral flux densities.The parameters used in the Quick TUV calculator were:

Figure S5 .
Figure S5.Panel (a) shows the wavelength-dependent MAC values recorded after pyrolysis of urban materials, photolytic aging for 2h, and extraction in the best-suited organic solvent (TableS1 and Figure S3).MAC values were calculated assuming 100% extraction efficiency from the Teflon filter.Panels (b) and (c) show log-log plots of the 2h aged OA MAC and unaged OA MAC, respectively, as a function of wavelength.The effective AAE, shown in Figure 1b of the manuscript, were obtained from the linear fits of these spectra in the 280-480 nm range.The linear trendlines are not shown in panels (b) and (c) for simplicity.Boxcar averaging of the MAC data (over 5 nm range) was utilized to improve signal-to-noise ratio.

Figure S6 .
Figure S6.Wavelength-dependent mass absorption coefficient (MAC) values recorded after photolysis on Teflon filters and the subsequent extractions of fiberboard OA (a and b) and carpet OA(c and d).Each quarter of the filter was aged separately for 0.5 h (red), 1 h (blue), or 2 h (grey) and is compared with the unaged fraction (black).The integrated MAC values (normalized to the unaged value) are shown as a subplot in each panel, with the UV integrated over 280-400 nm and the visible integrated over 400-700 nm.Boxcar averaging of the MAC data (over 5 nm range) was utilized to improve signal-to-noise ratio.

Figure S7 .
Figure S7.Wavelength-dependent mass absorption coefficient (MAC) values recorded after photolysis on Teflon filters and the subsequent extractions of thick PVC wire OA (a and b), ceiling tile OA (c and d), and plywood OA (e and f).Each quarter of the filter was aged separately for 0.5 h (red), 1 h (blue), or 2 h (grey) and is compared with the unaged fraction (black).The integrated MAC values (normalized to the unaged value) are shown as a subplot in each panel, with the UV integrated over 280-400 nm and the visible integrated over 400-700 nm.Boxcar averaging of the MAC data (over 5 nm range) was utilized to improve signal-to-noise ratio.

Figure S8 .
Figure S8.Pictures of Teflon filters after photolysis and subsequent extraction of lumber OA (a), fabric OA (b), and vinyl tile OA (c).Orgmix was used in the extraction of all OA.Only these 3 urban materials left behind residual OA on Teflon filters after extraction.

Figure S10 .
Figure S10.Wavelength-dependent mass absorption coefficient (MAC) values recorded after photolysis and subsequent extraction of fabric OA in orgmix.Teflon experimental results are shown in panels a and b.Fused silica window experimental results are shown in panels c and d.Each quarter of the filter was aged separately for 0.5 h (red), 1 h (blue), or 2 h (grey) and is compared with the unaged fraction (black).The integrated MAC values (normalized to the unaged value) are shown as a subplot in each panel, with the UV integrated over 280-400 nm and the visible integrated over 400-700 nm.Boxcar averaging of the MAC data (over 5 nm range) was utilized to improve signal-to-noise ratio.

Figure S11 .
Figure S11.Wavelength-dependent mass absorption coefficient (MAC) values recorded after photolysis and subsequent extraction of vinyl tile OA in orgmix.Teflon experimental results are shown in panels a and b.Fused silica window experimental results are shown in panels c and d.Each quarter of the filter was aged separately for 0.5 h (red), 1 h (blue), or 2 h (grey) and is compared with the unaged fraction (black).The integrated MAC values (normalized to the unaged value) are shown as a subplot in each panel, with the UV integrated over 280-400 nm and the visible integrated over 400-700 nm.Boxcar averaging of the MAC data (over 5 nm range) was utilized to improve signal-to-noise ratio.

Figure S12 .
Figure S12.Ratios of the 2 h aged to unaged integrated, UV range MAC values.The reported ratios were calculated by integrating the MAC values in the UV range (280-400 nm) then referencing the 2 h aged integrated MAC value to that of the unaged, integrated MAC value.The dashed line (0.9 to 1.1) indicates the threshold for the change.

Figure S13 .
Figure S13.UHPLC-PDA-HRMS data for thin PVC wire OA and phthalic acid standard: a) Thin PVC Wire OA PDA chromatogram (red) and TIC chromatogram (black); b) mass spectrum of thin PVC wire OA at TIC retention time 7.39 min; c) phthalic acid standard PDA chromatogram (red) and TIC chromatogram (black); d) mass spectrum of phthalic acid standard at TIC retention time 7.33 min; e) absorption spectra of thin PVC wire OA (purple, from Figure 1), thin PVC wire OA detected by PDA (dashed), and phthalic acid (black).A 0.07 min correction was made to TIC chromatograms to account for the delay time between PDA and Orbitrap detectors.All PDA traces were offset by x10 4 to account for differences in intensity between PDA and Orbitrap detection.Note in panel (c): retention time is slightly earlier than panel (a) due to phthalic acid standard dissolved in water only and not acetonitrile.

Figure S14 .
Figure S14.Reconstructed mass spectra of individual organic components identified in the timeintegrated TPD-DART-HRMS experiments with the OA collected from pyrolysis of carpet (a), vinyl tile (b) and fiberboard (c) materials.Unaged OA is presented in blue, whereas OA aged 2 h under UV radiation are presented in red.To facilitate visual comparison, MS peaks detected in aged OA samples are plotted as negative signal, and intensities of all peaks are scaled to the cubic root of the originally recorded values.

Figure S15 .
Figure S15.Reconstructed mass spectra of the unique OA components.Peaks unique to unaged OA + peaks that decreased by 10 % after aging are presented in blue, whereas peaks unique to aged OA + peaks that increased by 10% after aging are presented in red.These components were identified in the time-integrated TPD-DART-HRMS experiments with the OA collected from pyrolysis of carpet (a), vinyl tile (b) and fiberboard (c) materials.To facilitate visual comparison, MS peaks detected in aged OA samples (red) are plotted as negative signal, and intensities of all peaks are scaled to the cubic root of the originally recorded values.

Figure S16 .
Figure S16.Long-term photolysis experiments.Wavelength-dependent mass absorption coefficient (MAC) values recorded after photolysis on Teflon filters and the subsequent extractions of carpet OA (a and b), vinyl flooring OA (c and d), and fiberboard OA (e and f).Each quarter of the filter was aged separately for 1 h (red), 6 h (blue), or 18 h (grey) and is compared with the unaged fraction (black).The integrated MAC values (normalized to the unaged value) are shown as a subplot in each panel, with the UV integrated over 280-400 nm and the visible integrated over 400-700 nm.Boxcar averaging (over 5 nm range) was utilized.

Figure S17 .
Figure S17.Wavelength-dependent mass absorption coefficient (MAC) values recorded for unaged drywall OA sample (black, dotted), drywall OA sample after photolysis using the Xenon arc lamp (black), and drywall OA sample after photolysis on the rooftop in direct sunlight (red).The integrated MAC values (normalized to the unaged value) for the UV integrated over 280-400 nm are shown in the subplot, with the UV integrated over 280-400 nm.The drywall OA aged samples bleached similar amount, within 5% of each other.Consistent with previous laboratory drywall OA experiments, photolytic exposure darkens this urban material's pyrolysis products.This is represented in the pictures of the unaged OA vs. aged OA filter and the "U" cutout which darkened during the sunlight exposure.