Article pubs.acs.org/est
Air Quality Impact and Physicochemical Aging of Biomass Burning Aerosols during the 2007 San Diego Wildfires Melanie D. Zauscher,†,∥ Ying Wang,‡,⊥ Meagan J. K. Moore,‡,# Cassandra J. Gaston,§,▽ and Kimberly A. Prather*,‡,§ †
Department of Mechanical and Aerospace Engineering, University of California, San Diego, California Department of Chemistry and Biochemistry, University of California, San Diego, California § Scripps Institution of Oceanography, University of California, San Diego, California ‡
S Supporting Information *
ABSTRACT: Intense wildfires burning >360 000 acres in San Diego during October, 2007 provided a unique opportunity to study the impact of wildfires on local air quality and biomass burning aerosol (BBA) aging. The size-resolved mixing state of individual particles was measured in real-time with an aerosol time-of-flight mass spectrometer (ATOFMS) for 10 days after the fires commenced. Particle concentrations were high county-wide due to the wildfires; 84% of 120−400 nm particles by number were identified as BBA, with particles 5 times higher submicrometer aerosol mass concentrations than regional pollution transport events in San Diego. 3.2. Particle Types. Figure 1d shows the temporal variation of particle types between 120 and 400 nm. After 10/30/2007, the relative amounts OC and soot increased. The ocean-faring ship particle type became more prevalent after 10/31/2007. Although all BBA mass spectra have similar characteristics, including 39K+ and carbonaceous markers,25 five different BBA particle types were observed in this study (Figure S6) based on the relative strength of the carbonaceous markers, 62NO3−, and 97 HSO4 −. Combined together, the BBA particle types represented ∼84% of the total number of particles detected by the ATOFMS, although nearly 100% of all particles were BBAs until 10/25/2007. It should be noted that BBA particles were sampled after transport to UCSD, and thus the particles had undergone various degrees of aging. The known marker of fresh BBAs,10,12,57 113K2Cl+, peaked on 10/22/2007 when particles with the smallest SMPS CMD (∼100 nm) were detected, confirming the presence of fresh BBAs (Figure 1b). However, when the SMPS CMD increased to ∼170 nm, also on 10/22/ 2007, the peak area of 113K2Cl+ decreased while that of 213 K3SO4+ increased, and thus these chemical markers serve as effective indicators of the relative physical age of BBAs. These changes in BBA size and composition with age will affect particle hygroscopicity and the overall cloud nucleating properties of BBAs and hence play an important role in regional climate.56 In this study, 44−65% of BBA mass spectra had peaks at 71 C3H3O2−, 73C2HO3−/C3H5O2−, 75C2H3O3−, and 87C3H3O3− likely due to organic acid anions, which typically produce negative molecular ions, such as oxalate (89HC2O4−).36 Although Silva et al.25 concluded that 45HCO2−, 59CH3COO−, and 71C3H3O2− were indicative of levoglucosan, other ATOFMS studies with organic standards, meat burning, and plant detritus showed that these ions were not unique to levoglucosan.29,57 As explained in section 2.2, 45HCO2− was not used in this study due to interference. Because levoglucosan is primarily emitted from biomass burning, it should be found across all BBA sizes. However, because the size distribution of BBA with 71C3H3O2− was shifted toward particles >225 nm (Figure S7), 71C3H3O2− must be mostly secondary in this study. Similar results were found for 59CH3COO−.These results suggest that 59CH3COO− and 71C3H3O2− observed in BBA during this study were not representative of levoglucosan. In order to try to identify the 71C3H3O2−, 73C2HO3−/C3H5O2−, 75 C2H3O3−, and 87C3H3O3− ions, organic standards were analyzed with ATOFMS. Figure S8 shows representative mass spectra of acrylate, glyoxalate, and other standards, in addition to levoglucosan, highlighting that multiple compounds typically present in biomass burning emissions can produce these ions. Thus, it is not possible to identify the source of these ions using solely the ATOFMS. However, the prevalence of these ions on larger particles suggests that these peaks resulted from the condensation of secondary organic acids. More information on how these standards were analyzed and a discussion of these results are given in the Supporting Information.
22/2007 23:00. The CMD increased again from 10/22/2007 23:00 to 10/23/2007 9:00, at which time the CMD remained stable at ∼140 nm. The CMD decreased to ∼100 nm on the morning of 10/25/2007 and then increased to ∼180 nm coinciding with the increase of RH on the afternoon of 10/25/ 2007. After 10/27/2007, the effects of the wildfires were less evident on the particle size distributions, and periodically an ultrafine mode, indicative of local traffic emissions, was observed. The observed particle CMD of 100−180 nm at the beginning of this study are comparable to the range of previously measured BBAs between 100 and 230 nm depending on particle age.3,48,49 The broad CMD range observed indicates BBAs were processed to different degrees during transport from the wildfires to the UCSD sampling site or alternatively that fire conditions changed.50 Detailed analysis linking chemical markers of fresh and aged BBAs with the shifts in size distribution are described in section 3.3. Figure 1c shows the PM2.5 measured at Escondido and downtown San Diego along with the estimated mass concentrations at UCSD. Due to the proximity of the Witch and Poomacha Fires (Figure S1), the PM2.5 concentrations were higher at Escondido than at UCSD and downtown. However, the national ambient air quality air standard (NAAQS) 24-h average PM2.5 exposure limit of 35 μg/m3 was exceeded at all sites, with hourly maximum values as high as 397, 177, and 110 μg/m3 at Escondido, UCSD, and downtown, respectively. During 10/22/2007 12:00 to 10/23/2007 16:00, when the wildfires were most intense and air mass trajectories passed through the Witch Fire before arriving at UCSD, the PM2.5 concentrations were highest with median (interquartile range) values of 72.0 (43.0−186.0), 47.6 (35.1−61.7), and 42.0 (29.5−48.3) μg/m3 at Escondido, UCSD, and downtown, respectively. In contrast, the annual 2007 average ± standard deviation PM2.5 concentrations were 15.0 ± 17.5 and 10.9 ± 9.1 μg/m3 at the Escondido and downtown air quality stations, respectively. Therefore, the 2007 wildfires emitted high fine aerosol mass concentrations dangerous to the public health of communities downwind. Estimated PM1.0 and PM0.1−0.4 from particle size concentrations at UCSD were also the highest throughout the period of intense wildfires (10/22/2007 12:00 to 10/23/2007 16:00) with median (interquartile range) values of 44.5 (30.9−56.3) and 37.7 (26.6−48.5) μg/m3 and maximum values of 168 and 148 μg/m3, respectively. The majority of particle mass was due to particles 225 nm (Figure 3a), it can be concluded that 62NO3− was present in all stages of aging, while 97HSO4− was only found in aged BBAs. This conclusion is supported by a previous aircraft study of BBAs over Wyoming, which also determined that aged BBAs contained more 97HSO4− compared with 62 NO3−.9 In contrast, a previous Mexico City study observed more 62NO3− than 97HSO4− in aged BBAs.33 The difference in the abundance of HSO4− versus 62NO3− in aged BBAs is likely dependent on the concentrations of different precursor gas phase species and meteorological conditions. For example, high humidity and increased cloud cover were likely more abundant in the air mass during the aircraft BBA study in order for extensive cloud processing of BBAs to occur, whereas in the Mexico City study there were probably higher concentrations of NOx(g) and solar radiation, leading to increased photochemistry. Overall, the nitrate factor and sulfate factor followed the temporal variations of 62NO3− and 97HSO4−, respectively (Figure 2b). BBA 62NO3− exhibited a diurnal pattern at the beginning of the fires, reaching a maximum in the midmorning and decreasing at noon when the temperature was the highest. Thus, the likely mechanism involved in this diurnal pattern is partly explained by the semivolatile nature of the nitrate, which makes the ambient temperature an important driver for its partitioning between the gas and particulate phases. The NOx temporal distribution (Figure S9) was characterized with a maximum during the midmorning and a smaller one at night. Thus, in addition to temperature, photochemistry must be playing a role. Despite the thick haze over San Diego County during the wildfires, solar radiation still had a strong diurnal profile overall (Figure S3), and thus photochemistry probably initiated the rapid conversion of NOx(g), emitted directly from the wildfires, to HNO3(g), which would then partition to the particulate phase as nitrate.59 The addition of nitrate to BBA is representative of BBA aging due to interaction with wildfire coemissions in the presence of the appropriate meteorological conditions. When the RH increased on 10/25/2007 (Figure S3), the nitrate factor no longer tracked 62NO3−, which spiked in parallel with 125HNO3NO3− and 18NH4+ during this period (Figure 3b). Instead, PMF analysis separated this 62NO3− spike into the ammonium nitrate factor. The high humidity probably led to the formation of particulate phase NH4NO3 from gaseous NH3 and HNO 3 ,60,61 both of which are emitted from biomass burning.62−66 Albeit the 18NH4+ was associated with 97HSO4− during the rest of the study, indicating the presence of ammonium sulfate ((NH4)2SO4), when the ammonium nitrate factor spiked there was a prevalence of ammonium nitrate (NH4NO3; Figure S10), as previously observed in BBAs.59,67 It is worth noting that sulfate was found in smaller particle sizes than those with ammonium (Figure 3a) and that ammonium was often mixed with nitrate in the larger sized particles. Although previous BBA studies have also found correlations between ammonium and oxalate,68 in the current study these species were independent of each other. Though present at the beginning of the fires, the RPA of 97 HSO4− remained low until the afternoon of 10/23/2007 and then peaked mostly in the afternoons between 10/23/2007 and 10/26/2007 (Figure 3b). Although in-cloud aqueous processing of SO2(g) is the main formation pathway of sulfate,69,70 the RH remained low (90% during this change, 97HSO4− did not increase in parallel to the RH likely because of low SO 2 atmospheric concentrations. Additionally, because shipping is a more significant source of SO2(g) than biomass burning,71 the 97 HSO4− detected on BBA after 10/25/2007 likely originated from SO2(g) emitted from the ships around the ports of San Diego, Los Angeles, and Long Beach, facilitating the formation of sulfate through cloud processing, perhaps over the ocean, during this time. The 10/27/2007 97HSO4− spike coincided with the presence of a widespread fog event, indicating it was produced through aqueous processing during this time. Therefore, the addition of sulfate to BBA at the end of this study is representative of BBA aging upon interaction with regional pollutants depending on meteorology. BBAs in areas affected by different regional pollutants and meteorological conditions will exhibit different aging processes. 3.3.2. Organic Carbon Species. As illustrated in Figure 2, PMF analysis separated organic carbon markers into four distinct factors described below with species given in order of prevalence. The organic carbon factor was mainly composed of 43 C2H3O+ and 27C2H3+, which are organic carbon fragments resulting from the ionization of larger organic carbon species. The organic carbon factor tracked the temporal profile of the sulfate factor while the RH remained low on the afternoons of 10/23/2007 and 10/24/2007 (Figure 2b), suggesting that the organic species represented by the 43C2H3O+ and 27C2H3+ fragments were formed through a similar primary mechanism as sulfate during this time. The organic nitrogen factor, comprised of 26CN−, 42CNO−, 59 C2H3O2−, and 46NO2−, was more prevalent at the beginning of the wildfires. The 26CN− and 42CNO− markers, which can be the resulting fragment of many nitrogen containing organic nitrogen species, have been observed in biogenic particles such as plant detritus and bacteria.29,57,72 Heterocyclic nitrogen compounds, including some naturally produced by plants and living organisms, and water-soluble organic nitrogen species have been previously measured in BBAs.73−75 Previous work has also observed significant gas phase hydrogen cyanide (HCN) and isocyanic acid (HCNO) emitted in fresh biomass burning;76−79 however, based on the vapor pressure of these species they will not partition to the particulate phase. Thus, it is likely that the 26CN− and 42CNO− markers are resulting fragments of organic nitrogen biogenic material involved in biomass pyrolysis. Furthermore, rapid formation of gas phase acetic acid, seen as 59C2H3O2−, and other organic acids have been observed in relatively young biomass burning plumes.15,79−81 Thus, it is likely that the species found in the organic nitrogen factor were either emitted directly in the particulate phase or they were rapidly produced from gas phase precursors emitted during flaming conditions and partitioned quickly to the particulate phase as the smoke plume cooled during transport. The amines factor consisted mainly of 86(C2H5)2NCH2+ with minor contributions from 18NH4+ and 26CN−, which were both likely also amine fragments during this time. The ammonium nitrate factor spike, which coincided with the first increase in RH on the night of 10/25/2007 (Figure S3), was 12 h before 7639
dx.doi.org/10.1021/es4004137 | Environ. Sci. Technol. 2013, 47, 7633−7643
Environmental Science & Technology
■
the amines factor grew in (Figure 2b). BBAs with 18NH4+ were also larger than those with 86(C2H5)2NCH2+ (Figure 3a), likely because water drives nitrate onto the particulate phase as NH4NO3. The BBAs with 86(C2H5)2NCH2+ also grew in size over time but did not grow as large as the BBAs with 18NH4+ (Figure 3a). It is also interesting that the amines and organic nitrogen factors were split into two, but based on their temporal and size profiles this makes sense. While the organic nitrogen factor was prevalent at the beginning of the fires, the amines factor spiked after the increase in RH on 10/25/2007. In controlled laboratory burns, gas-phase amines have been observed to be emitted during the smoldering phase.1,62 Amines remained in the gas phase while the RH was low, as shown by previous work,38,82 and partitioned to the particulate phase once the RH increased to >90%. The organic acids factor, which had contributions from 89 HC2O4−, 87C3H3O3−, 75C2H3O3−, 73C2HO3−/73C3H5O2−, and 71 C3H3O2−, peaked between 10/22/2007 12:00 and 10/22/ 2007 19:00, 10/23/2007 9:00 and 10/25/2007 14:00, 10/26/ 2007 10:00 and 10/27/2007 22:00, and 10/28/2007 5:00 and 10/30/2007 11:30. These organic acids tracked each other temporally and by size (Figure S7), indicating similar formation mechanisms, which may have varied during the sampling period due to changing meteorological conditions and differing concentrations of gas-phase precursors. The organic acids at the beginning of the wildfires were likely primary or a result of fast secondary production, as has been previously observed.15,80,81 However, organic acids spiked between 10/28/ 2007 5:00 and 10/30/2007 11:30, coinciding with the period when the air mass back trajectories came from offshore, and thus the BBAs sampled during this time had significantly more time to age. Therefore, the organic acids from this period could have formed through slow gas-to-particle conversion of VOCs emitted during biomass burning, which further oxidized into the organic acids.83,84 Although it was not possible to identify each organic species present in BBAs, the four different organic carbon factors identified in this study indicate that ATOFMS data can still be used to learn about different processes that occur in biomass burning. New methods for online sampling of BBA would be useful, such as the aerosol chip electrophoresis (ACE) system85 and the two-dimensional gas chromatography thermal desorption (2D-TAG) instrument,86 in order to complement the capabilities of ATOFMS measurements. In addition, it would be greatly beneficial to sample ultrafine (
