Biomass burning is an important source of aerosol and
trace gases to the atmosphere, but how these emissions change chemically
during their lifetimes is not fully understood. As part of the Fire
Influence on Regional and Global Environments Experiment (FIREX 2016), we
investigated the effect of photochemical aging on biomass burning organic
aerosol (BBOA) with a focus on fuels from the western United States.
Emissions were sampled into a small (150
Biomass burning is a major source of particulate matter and trace gases to the atmosphere, and strongly affects global air quality and climate (Akagi et al., 2011; Bond et al., 2004; Liu et al., 2017). In fire-prone regions such as the western United States, the frequency and intensity of wildfires have increased over the past several decades due to fire management practices and climate change (Westerling et al., 2006), and this trend is expected to continue in the coming decades (Dennison et al., 2014; Spracklen et al., 2009). Emissions from fires have been the subject of intense study, but primary emissions alone do not determine the atmospheric impacts of biomass burning since smoke plumes can be transported thousands of kilometers and undergo dramatic chemical changes over their lifetimes in the atmosphere (Andreae et al., 1988; Cubison et al., 2011). In particular, biomass burning organic aerosol (BBOA) is subject to atmospheric aging processes that could significantly alter the climate- and health-relevant properties of biomass burning emissions (Hennigan et al., 2012; Vakkari et al., 2014). Such processes include oxidation of gas-phase compounds followed by partitioning to the particle phase, forming secondary organic aerosol (SOA); direct oxidation of molecules in the particle phase through heterogeneous reactions; and evaporation of particulate semivolatile molecules upon plume dilution, potentially followed by subsequent gas-phase oxidation. However, despite the potential importance of aging on biomass burning emissions, the effect of aging on BBOA composition and loading over multiday timescales is not well-constrained, and usually is not included in global chemical transport models (Shrivastava et al., 2017).
Field measurements provide strong evidence that the composition of BBOA
changes significantly when photochemically aged. In aircraft measurements of
biomass burning plumes, OA consistently becomes more oxidized downwind,
relative to the source of emissions
(Capes
et al., 2008; Cubison et al., 2011; Forrister et al., 2015; Jolleys et al.,
2015, 2012). Additionally, decreases in reactive tracers
from biomass burning, such as levoglucosan, are observed after aging when
compared to their contribution to fresh emissions
(Cubison
et al., 2011). Despite these consistencies, field measurements show mixed
results with regards to whether or not there is an increase in net SOA
downwind of fires. Net SOA formation is usually characterized by an OA
enhancement ratio, defined as the ratio between fresh and aged
Laboratory studies intended to constrain the effects of aging on biomass burning emissions have also had variable results. Consistent with field measurements, laboratory experiments in which emissions from open burning and wood stoves were photochemically aged found that BBOA became increasingly oxidized and tracers were depleted with increased aging time (Ahern et al., 2019; Bertrand et al., 2018; Cubison et al., 2011; Grieshop et al., 2009a; Hennigan et al., 2011; Ortega et al., 2013; Tkacik et al., 2017). Most laboratory experiments investigating the aging of biomass burning emissions find that significant amounts of SOA are formed in most, but not all, cases (Ahern et al., 2019; Bruns et al., 2016; Grieshop et al., 2009b; Hennigan et al., 2011; Ortega et al., 2013; Tiitta et al., 2016; Tkacik et al., 2017). Even under constrained laboratory experimental conditions, these studies show significant variability in SOA formation between burns of similar or even identical fuels. This variability is often attributed to differences in burning conditions (e.g., flaming and smoldering) (Hennigan et al., 2011) or the presence of unmeasured SOA precursors (Bruns et al., 2016; Grieshop et al., 2009b; Ortega et al., 2013), but predicting biomass burning SOA across fuel types and burning conditions has remained a challenge. Very recently, Ahern et al. (2019) showed that the detailed characterization of hundreds of compounds emitted from a given burn, coupled with estimated SOA yields from each, enables the prediction of SOA formation to within roughly a factor of 2. This approach establishes a clear link between the gas-phase emissions and SOA formation but relies critically on a comprehensive understanding of emission profiles, which may exhibit substantial burn-to-burn variability.
The high degree of variability in net OA observed from biomass burning studies leads to a large range of estimates of SOA from biomass burning, which span nearly 2 orders of magnitude (Shrivastava et al., 2017). The range of global estimates is thus essentially unconstrained, with some studies ranking biomass burning as an insignificant source of SOA and others ranking it as the major source of global SOA (Shrivastava et al., 2015, 2017). Understanding the evolution of biomass burning emissions is necessary to better evaluate the effects of biomass burning on air quality, human health, and climate. To this end, we describe the results from a set of laboratory aging experiments on a variety of fuels, employing an oxidation reactor coupled with real-time measurements of the composition of both the particle-phase and gas-phase emissions, to better constrain the effects of aging on biomass burning emissions.
Experiments were carried out as part of the Fire Influence on Regional and
Global Environments Experiment (FIREX 2016) at the USDA Fire Sciences
Laboratory (FSL) in Missoula, MT, USA, with the goal of better understanding the
evolution of biomass burning emissions within a controlled environment.
Experiments took place during the “stack burn” portion of FIREX, in which
fuels were burned beneath the exhaust stack (1.6
Aging experiments were conducted in a 150
Prior to sampling, the chamber was flushed with clean air from a zero-air
generator (Teledyne 701H) and humidified air (total 15 slpm, standard liters per minute) for
approximately 45
Particles and gases exiting the reactor were monitored with a suite of
analytical instruments. Particle composition measurements were made with an
aerosol mass spectrometer with a standard tungsten vaporizer (AMS, Aerodyne
Research, Inc.), which measures the mass and composition of nonrefractory
particles with diameters between 70
Particle mass and composition data from the AMS were analyzed using the
ToF-AMS analysis toolkits (Squirrel version 1.57I, Pika version 1.16I) using
the “improved-ambient” method for calculating oxygen-to-carbon (
CE and particle density were calculated by comparing AMS particle time-of-flight (PToF) and SEMS size distributions along with AMS organic mass and SEMS volume (Bahreini et al., 2005). This could only be done for the subset of data points with PToF and SEMS distributions that could be fit to lognormal functions, did not show significant particle nucleation, and had low SP-AMS rBC concentration (see below). An AMS CE correction was then applied to the entire data set by parameterizing the exponential relationship between AMS CE with the particle MFR (Fig. S4). An alternative calculation of OA mass using SEMS volume multiplied by OA density agrees with OA mass calculated using this MFR CE parameterization, and is shown in Fig. S5. Generally, POA has low organic MFR (indicating that it is relatively volatile) and particle MFR increases with oxidation, consistent with previous work (Hennigan et al., 2011). This indicates that particles are becoming less volatile and more likely to be (semi-)solid, and therefore likely to have a lower CE due to increased bounce off of the AMS vaporizer (Matthew et al., 2008; Virtanen et al., 2010). Calculated AMS collection efficiencies range from 0.35 to 0.64, with the average CE of fresh emissions equal to 0.54 and average CE of aged (i.e., end-of-oxidation) emissions equal to 0.40. We note that this CE value of 0.54 for fresh emissions is substantially lower than the value of 1 typically assumed in such experiments (Ahern et al., 2019; Hennigan et al., 2011; Heringa et al., 2011; Ortega et al., 2013; Tkacik et al., 2017).
Acetonitrile, an inert tracer species (
OH exposures in the chamber were estimated by measuring the decay of an OH
tracer, deuterated
Changes in OA mass and composition as a function of aging
time, assuming an atmospheric [OH] of
The total initial aerosol mass in the chamber varied widely from experiment
to experiment (Table S1), averaging
The chemical composition of the primary organic particulate matter varied
substantially between experiments. The initial
OH oxidation, initiated in the chamber by exposure to UV light, rapidly
changes the composition of OA, as shown in Fig. 1b–d. Gas-phase chemistry
in the mini-chamber is discussed in detail in a companion publication
(Coggon et al., 2019). Despite
differences in initial composition between experiments, OA in all
experiments undergoes a large increase in
Similar to the elemental ratios, the initial fraction of the primary organic
signal from the AMS fragment ion
End-of-experiment SOA formation vs. total NMOG concentration in the chamber prior to OH oxidation. Points are colored by the atmospheric equivalent aging time corresponding to the end of each experiment.
OA carbon mass added vs. initial NMOG carbon mass from
PTR-ToF-MS measurements at various OH exposures (0.25–4
All aging experiments show substantial SOA formation with OA mass
continuing to increase with extended aging time (Fig. 1a). Consistent with
previous studies, the correlations between OA mass enhancement ratio and
various parameters that could affect SOA production (e.g., OH exposure, POA,
monoterpenes, total NMOG concentration) are weak at best (Fig. S9). However,
the absolute amount of SOA formed does appear to correlate with some of
these. Figure 2 shows the relationship between SOA formed by the end of each
experiment (
Correlation coefficients (
The correlation between SOA formation and the initial chamber concentration
of NMOGs is reasonable since NMOGs provide the carbon that drives SOA
growth. However, the PTR-ToF-MS measures many compounds that likely do not
contribute to SOA formation (e.g., small compounds such as methanol and
acetonitrile). In addition to comparing SOA to the initial NMOG
concentration, we can examine how SOA formation correlates with the
concentration of NMOGs above some molecular weight cutoff. Figure 4 shows
the correlation coefficient for the linear fit between SOA carbon mass and
summed NMOG carbon mass at each molecular weight cutoff for 1
Previous biomass burning aging experiments with both aerosol and NMOG
measurements have not observed this relationship between SOA and total NMOGs
(Ortega et al., 2013) or individual SOA precursors
(Bruns
et al., 2016; Grieshop et al., 2009b; Ortega et al., 2013; Tkacik et al.,
2017). Most such studies identified only half or less of the NMOG signal
and/or were limited to a small number of experiments
(Grieshop et
al., 2009b; Ortega et al., 2013; Tkacik et al., 2017); this could potentially
explain why similar correlations between total volatile organic compounds (VOCs) and SOA have not been
observed before. By contrast, the use of PTR-ToF-MS in the present study
enables the measurement of
Recent work has pointed to the importance of non-traditional SOA
precursors to SOA formation for residential wood combustion of a single fuel
type (beech wood) (Bruns et al., 2016). These
precursors include semivolatile and intermediate-volatility volatile organic
compounds (S/IVOCs) such as phenols and naphthalenes
(Bruns et al., 2016). However, the majority of
gas-phase carbon observed in this study is in compounds with low carbon
number (
Estimated saturation vapor concentration distribution
From each of the relationships between SOA carbon mass and initial NMOG
carbon mass (linear fits in Fig. 3), an effective carbon yield can be
calculated. Carbon yield is defined here as the SOA carbon formed divided by the total NMOG carbon reacted at each respective OH
exposure. The amount of gas-phase carbon reacted (
SOA carbon yield from aging of biomass burning emissions.
Black points are carbon yields using our best estimate of OA carbon mass.
Yields are calculated from the ratios of SOA carbon formed to NMOG carbon reacted (estimated from the average carbon-weighted OH rate coefficient for identified compounds and accounting for chamber dilution). Error bars are
The variability in findings from previous lab and field studies on the
effect aging has on net SOA from biomass burning can be potentially
explained by the effects of dilution on the evolution of BBOA mass. Some
fraction of BBOA is semivolatile, and dilution (in chambers or ambient
smoke plumes) will cause volatile OA components to partition from the
particle phase to the gas phase
(May et al., 2013). Recent
modeling work has shown that even in plumes that show no net SOA formation,
significant condensation of secondary organic mass may occur
(Bian
et al., 2017; Hodshire et al., 2019b), but net growth can be low (or even
negligible) due to dilution-driven evaporation of OA. In ambient plumes,
dilution drives semivolatile species from the particle phase to gas phase;
although this causes a loss in OA mass, it also serves as a source of semivolatile organic compounds (SVOCs)
that can condense back onto particles after oxidation, leading to little to
no net change in OA. Related to this point, calculated net-OA values are
also sensitive to the choice of starting point (
We show that the OH-initiated aging of biomass burning emissions leads to
significant changes in BBOA composition and loading. These changes are
dependent on OH exposure and are especially large over the first few days
after emission. Significant amounts of SOA are formed from all fuels studied
here, but SOA formation is highly variable. Despite large differences in
fuel type and burning conditions, much of this variability can be explained
by differences in the initial total NMOG concentration and OH exposure.
Correlations between SOA formation and the concentration of initial measured
NMOGs in the chamber at given OH exposures are good, with
Data are available from the CSD NOAA archive at
The supplement related to this article is available online at:
Data were interpreted and the article was written by CYL and JHK. Mini-chamber construction and operation were done by CYL, DHH, and CDC. The AMS was operated and data were analyzed by CYL. PTR-ToF-MS was operated and data were analyzed by MMC, ARK, and KS. Experiments were conceived by JHK, CDC, and CW. All co-authors provided article feedback and comments.
The authors declare that they have no conflict of interest.
Christopher Y. Lim and Abigail R. Koss were supported by the NSF graduate research fellowship program. The authors would like to thank Timothy Onasch for support of the SP-AMS; Colette Heald for helpful comments; and Edward Fortner, Berk Knighton, Robert Yokelson, the entire FIREX science team, and Missoula Fire Sciences Laboratory staff for support during the project.
This research has been supported by the NOAA AC4 program (grant nos. NA16OAR4310112 and NA16OAR4310111).
This paper was edited by Ryan Sullivan and reviewed by two anonymous referees.