Emissions from the Open Laboratory Combustion of Cheatgrass ( Bromus tectorum )

: Cheatgrass ( Bromus tectorum ) is a highly invasive species in the Great Basin of the Western USA that is increasing the frequency and intensity of wildland ﬁres. Though cheatgrass plays a signiﬁcant role in the ﬁre ecology of the Great Basin, very little is known about its combustion emissions. The fresh smoke from 16 open laboratory burns of cheatgrass was analyzed using real-time measurements and ﬁlter analysis. We presented measured intensive optical properties of the emitted smoke, including absorption Ångström exponent (AAE), scattering Ångström exponent (SAE), single scattering albedo (SSA), and other combustion properties, such as modiﬁed combustion e ﬃ ciency (MCE) and fuel-based emission factors (EFs). In addition, we gave a detailed chemical analysis of polar organic species in cheatgrass combustion emissions. We presented EFs that showed a large variation between fuels and demonstrated that analysis of combustion emissions for speciﬁc fuels was important for studying and modeling the chemistry of biomass-burning emissions.


Introduction
Cheatgrass (Bromus tectorum) is an invasive grass that has extensive ecological impacts [1] and is rapidly spreading through the sagebrush ecosystem of the Great Basin [1] that covers most of Nevada, much of Utah, and smaller portions of Oregon, California, Idaho, and Wyoming, USA. This invasion results in new uncertainties for wildland fire emissions in the Great Basin. Specifically, the emission factors (EFs) and the properties of aerosols emitted from sagebrush combustion have been characterized [2], but emissions from cheatgrass combustion have not. Furthermore, cheatgrass is increasing the frequency and magnitude of rangeland fires ten-fold [3] by creating higher fuel loads and higher vegetative coverage in areas where shrubs and desert grasses were once separated by open areas of soil. Regions, where spatial cheatgrass coverage reaches more than 15%, are more than twice as likely to experience a wildfire compared to regions where the cheatgrass coverage is minimal [1]. The increased fine fuel accumulation also increases the likelihood of the destruction of native plants during wildland fire events [4]. Cheatgrass is rapidly invading much of the Great Basin, causing additional carbon emissions [5] through increased fire frequency in rangelands [6].
Our study of cheatgrass combustion emissions was conducted with cheatgrass harvested from a site near Reno, Nevada, USA, that had most recently burned during the Hawken fire of June 2016. This fire burned through native vegetation, destroying most of the sagebrush and other chaparral plants, which allowed for increased cheatgrass growth in this area. On average, the Reno area (NV, has been given by Tian et al. [37]. The facility consists of a sealable chamber (1.8 m × 1.8 m × 2.0 m) with aluminum walls that taper at the roof, funneling smoke into an exhaust pipe. Combustion exhaust is sampled from this pipe through copper tubes and is delivered to instrumentation for real-time and filter sampling. The chamber flow rate is controlled by a fan in the exhaust pipe and by a gate controlling the air inlet, located underneath the chamber.
For our combustion experiments, the chamber fan was turned off, and the gate was fully closed, allowing combustion emissions to accumulate within the chamber, similar to the methods used by McMeeking et al. [2]. Cheatgrass was piled on top of a continuously weighed ceramic disk in the chamber to monitor fuel mass loss during the burning period. The copper pipe connecting the chamber to the real-time sampling instruments was disconnected, and the fuel was ignited with a butane lighter. The chamber was allowed to fill with smoke for five minutes, resulting in a uniform aerosol distribution in the chamber. At the end of this five-minute time period, the chamber was reconnected to the sampling line. The real-time instruments characterized smoke from the chamber until smoke concentrations dropped to near-ambient levels.
The PASS-3 utilizes the photoacoustic method to measure aerosol light absorption coefficients (b abs Mm −1 ) and integrated nephelometry to measure light scattering coefficients (b scat Mm −1 ) at the wavelengths of 405 nm, 532 nm, and 781 nm simultaneously [11,38,39]. Particle mass concentrations and size distributions were measured with a scanning mobility particle sizer (SMPS), which consisted of a TSI 3081 differential mobility analyzer (DMA) and TSI 3775-condensation particle counter (CPC). These real-time instruments sampled an input flow of 1 L per minute each, reporting results every second, except for the SMPS, which was operated at a 10:1 sheath to sample flow ratio, with a 0.3 L per minute sample flow rate. This SMPS setup measured particles with the size in the range of 8-580 nm. The CO 2 and CO monitors were checked for accuracy prior to each burn, and calibration was performed, if needed.

Fuel Characteristics
Cheatgrass is a plant native to Eurasia that has aggressively invaded North America with an ability to produce over 10,000 plants per square mile [40]. Cheatgrass grows quickly in spring before native species develop and dies by early summer, thereby contributing large amounts of dry, fine fuels that are easily ignitable. The plant is recognizable by the drooping seed heads, as seen in Figure 1, and fine hairs covering the plant. The young plants are green, turning to purple, during further maturation, their color fades into a pale golden brown as the plant dies. The typical height of the plant is between 7.5 and 75 cm, and its seeds are covered with bristle-like hairs that allow for dispersion by wind and animal fur. Fuels and ashes from each cheatgrass burn were analyzed for total carbon, nitrogen, hydrogen, and oxygen mass content at DRI's Environmental Analysis Facility (EAF) using a Thermo Scientific™ EA 1112 Series CHNS-O Analyzer (Waltham, MA, USA) for total carbon, nitrogen, hydrogen, and oxygen ratios [41,42].

Modified Combustion Efficiency
The modified combustion efficiency (MCE) approximates the fraction of emitted carbon that is in the form of CO 2 and depends on the fuel and its combustion phase (e.g., flaming vs. smoldering). MCE is calculated as the ratio of molar CO 2 and CO excess concentrations as where ∆CO 2 and ∆CO are the excess molar concentrations of CO 2 and CO, respectively, with the background concentrations subtracted. Here, the subtracted background concentrations were the mean of the concentrations measured in the chamber over the 2 min prior to the burn's ignition. The MCE is a qualitative indicator of the amount of flaming and smoldering combustion present since, in a typical burn, there is a mixture of both combustion phases. In flaming combustion, the reaction oxidizes C, N, H, and S and turns them into common greenhouse gases like CO 2 , H 2 O, NO x , and SO 2 [43] with an MCE of about 0.99 for pure flaming combustion [44]. The MCE of smoldering combustion ranges from 0.65 to 0.85 [44]. MCE is also related to the amounts of OC and BC emitted; flaming combustion emits more BC, while smoldering combustion emits more OC and its light-absorbing component BrC.

Modified Combustion Efficiency
The modified combustion efficiency (MCE) approximates the fraction of emitted carbon that is in the form of CO2 and depends on the fuel and its combustion phase (e.g., flaming vs. smoldering). MCE is calculated as the ratio of molar CO2 and CO excess concentrations as where ΔCO2 and ΔCO are the excess molar concentrations of CO2 and CO, respectively, with the background concentrations subtracted. Here, the subtracted background concentrations were the mean of the concentrations measured in the chamber over the 2 min prior to the burn's ignition. The MCE is a qualitative indicator of the amount of flaming and smoldering combustion present since, in a typical burn, there is a mixture of both combustion phases. In flaming combustion, the reaction oxidizes C, N, H, and S and turns them into common greenhouse gases like CO2, H2O, NOx, and SO2 [43] with an MCE of about 0.99 for pure flaming combustion [44]. The MCE of smoldering combustion ranges from 0.65 to 0.85 [44]. MCE is also related to the amounts of OC and BC emitted; flaming combustion emits more BC, while smoldering combustion emits more OC and its light-absorbing component BrC.

Fuel-Based Emission Factors
The fuel-based emission factor (EFn) for a species, n, is the ratio of the mass of species n emitted (M n ) to dry fuel mass burned (M fuel ) as

Fuel-Based Emission Factors
The fuel-based emission factor (EF n ) for a species, n, is the ratio of the mass of species n emitted (M n ) to dry fuel mass burned (M fuel ) as where the ratio of masses emitted (i.e., M n /M C ) and M fuel is determined as the ratio of total carbon mass emitted (M C ) and the carbon mass fraction of the dry fuel (CMF fuel ) as where the total carbon mass emitted (M C ) is well approximated by the sum of carbon mass in carbon dioxide (M C CO 2 ) and carbon monoxide (M C CO ) as Atmosphere 2020, 11, 406 5 of 15 In this equation, we neglected the ash carbon content, which is typically less than 5% of the fuel mass [7]. EFs for particulate matter (EF PM ) were calculated from the SMPS particles per unit volume of air using a density value of 1.2 g/cc to determine the mass concentration. CO 2 and CO mass concentrations were measured with the LiCor and Thermo analyzers. Background concentrations were subtracted, and the excess of CO 2 and CO mass concentrations were used to calculate the EFs.

Single Scattering Albedo
Aerosol absorption (b abs ) and scattering (b sca ) coefficients were measured with the PASS-3 and used to calculate the single scattering albedo (SSA). SSA is an intensive aerosol property that equals the ratio of scattering coefficient to total extinction coefficient as , and λ is the wavelength of light [19,20]. The term b ext , the extinction coefficient, is the sum of the scattering and absorption coefficients. The aerosol SSA is the most important intensive particle parameter, controlling aerosol direct radiative forcing [19,20], and it influences radiative fluxes and heating in the atmosphere. Mostly absorbing aerosols (SSA ≈ 0) have a heating effect, while mostly scattering ones (SSA ≈ 1) have a cooling effect [45,46].

Absorption Ångström Exponent
The absorption Ångström exponent (AAE) is an intensive aerosol optical property, describing the wavelength dependence of the aerosol absorption coefficient. AAE has also been used quantitatively to separate brown carbon (BrC) absorption from black carbon (BC) absorption. BC is highly absorbing in the visible and near-visible spectrum, while BrC is highly absorbing in the near-UV and blue-spectrum. AAE is the negative slope of the absorption coefficient as a function of wavelength on a log-log plot and can be written as where b abs is the wavelength-dependent absorption coefficient [18]. AAE is calculated from two b abs coefficients measured by the PASS-3 at two different wavelengths. AAE values much larger than one are typical for aerosols emitted by smoldering biomass burning and are caused by BrC absorption. AAE values of about 1 typically indicate absorption from BC [46].

Scattering Ångström Exponent
The scattering Ångström exponent (SAE) characterizes the wavelength dependence of aerosol light scattering. SAE depends mostly on the dominant size mode of a mixture of aerosols, and, combined with the AAE, it can give insight into the nature of the aerosol [47]. The value of the SAE, which is the negative slope of the scattering coefficient as a function of wavelength on a log/log plot, is where b sca is the wavelength-dependent scattering coefficient.

Analysis of Organic Species
Sampling and analysis of organic emissions collected on XAD (gas-phase organic compounds) resin (Sigma-Aldrich, St. Louis, MO, USA) and filter samplers (particle-phase organic compounds) are described in detail by Sengupta et al. [48] and Samburova et al. [36]. Briefly, combustion emissions were collected on 100-mm diameter Teflon-impregnated glass fiber (TIGF) filters (Fiber Film T60A20, Pall Life Sciences, Ann Arbor, MI, USA), followed by a XAD-4 cartridge at a flow rate of 0.11 m 3 /min. Immediately after sampling, the samples were stored at -20 • C. Prior to extraction, collected filters and XAD cartridges were spiked with deuterated internal standards of polar organic compounds [48]. The samples were extracted with an accelerated solvent extractor (ASE) system (DIONEX, ASE-300, Salt Lake City, UT, USA) at the following conditions: temperature: 80 • C, solvents: dichloromethane (150 mL) followed by acetone (150 mL), extraction cell pressure: 10.3 MPa, extraction time: 15 min. After the extraction, the samples were pre-concentrated to 1 mL using a rotary evaporator at 35 • C and a gentle vacuum. The extracts were derivatized with N,O-Bis(trimethylsilyl)trifluoroacetamide (BSTFA) agent, as described elsewhere [49], and analyzed with gas chromatography-mass spectrometry (GC-MS) method [48] using Varian CP-3800 GC equipped with a CP-8400 autosampler and interfaced to a Varian 4000 Ion Trap Mass Spectrometer (Varian, Inc. Walnut Creek, CA, USA). Injections were performed into a 30 m length, 5% phenylmethylsilicone fused silica capillary column (DB-5MS, Agilent Technologies, Palo Alto, CA, USA) with an integrated deactivated guard column (length: 10 m).

Modified Combustion Efficiency
Once ignited, cheatgrass burned rapidly, and the flaming combustion phase dominated carbon emission. Cheatgrass MCE ranged from 0.93 to 0.98, confirming that cheatgrass combustion was dominated by flaming combustion, as is typical for dry, fine fuels. The MCEs reported for cheatgrass in Table 1 were the average emission ratios for the first 5 min of 12 burns. Cheatgrass ignited quickly, and initially, flaming-dominated combustion caused a rapid increase in CO 2 concentrations, followed by an increase in CO concentrations, indicating a transition to more smoldering combustion. The MCE averaged over 12 burns was calculated to be 0.96 ± 0.03, with ±0.03 being the standard deviation for the 12 burns. This MCE measured for cheatgrass was in reasonable agreement with previously published MCE values for a prescribed grassland burn [50] for Dambo grass and Montana grass open laboratory combustion [7] and for savanna and grasslands fires [51,52], as shown in Table 1.

Single Scattering Albedo (SSA)
The SSA, shown in Figure 2, was the average ratio of the scattering coefficient to the extinction coefficient for the first 5 min of 12 burns. At the 405 nm wavelength, the SSA of cheatgrass combustion aerosols was the lowest (0.81-0.85); here, a significant part of the extinction (i.e., 15%-19%) was due to absorption. The SSA at 405 nm had the least variation between the burns; it was indicative of the absorption of BrC, which was highly absorbing in the UV. The SSA at the 532 nm and 781 nm wavelengths indicated some absorption from BC, which is highly absorbing in the visible spectrum [53]. Lewis et al. [11] reported the optical properties of rice straw, a fuel similar to cheatgrass. The SSA for the 405 nm wavelength for cheatgrass was reported here to be 0.83 ± (0.02), showing similar absorption as the rice straw with an SSA of 0.88 with an error of 5% reported by Lewis et al. [11].  [50], open laboratory combustion of Dambo and Montana grass [7], and savanna and grasslands fires [51,52].

Single Scattering Albedo (SSA)
The SSA, shown in Figure 2, was the average ratio of the scattering coefficient to the extinction coefficient for the first 5 min of 12 burns. At the 405 nm wavelength, the SSA of cheatgrass combustion aerosols was the lowest (0.81-0.85); here, a significant part of the extinction (i.e., 15%-19%) was due to absorption. The SSA at 405 nm had the least variation between the burns; it was indicative of the absorption of BrC, which was highly absorbing in the UV. The SSA at the 532 nm and 781 nm wavelengths indicated some absorption from BC, which is highly absorbing in the visible spectrum [53]. Lewis et al. [11] reported the optical properties of rice straw, a fuel similar to cheatgrass. The SSA for the 405 nm wavelength for cheatgrass was reported here to be 0.83 ± (0.02), showing similar absorption as the rice straw with an SSA of 0.88 with an error of 5% reported by Lewis et al. [11].

Absorption and Scattering Ångströ m Exponents
The AAE is approximately one (~1) for BC but can be larger than two for a mix of BC and BrC [18]. The SAE can be as small as zero for very large particles in the geometric optics regime and as large as four for very small particles in the Rayleigh regime [54]. Here, Å ngström exponents were calculated for pairs of 405/532 nm, 405/781 nm, and 532/781 nm wavelengths. The differences between 405/532 and 405/781 AAE were driven by BrC that absorbed mostly in the UV and blue and by BC absorption that was less dependent on wavelength (AAEBC ~ 1). SAE gives information on particle sizes, indicating coarse-mode particles for small values (i.e., <1.5) and fine-mode particles for larger SAE values (i.e., >1.5) [55]. The SAE values observed for cheatgrass (i.e., 0.5-3.5,

Absorption and Scattering Ångström Exponents
The AAE is approximately one (~1) for BC but can be larger than two for a mix of BC and BrC [18]. The SAE can be as small as zero for very large particles in the geometric optics regime and as large as four for very small particles in the Rayleigh regime [54]. Here, Ångström exponents were calculated for pairs of 405/532 nm, 405/781 nm, and 532/781 nm wavelengths. The differences between 405/532 and 405/781 AAE were driven by BrC that absorbed mostly in the UV and blue and by BC absorption that was less dependent on wavelength (AAE BC~1 ). SAE gives information on particle sizes, indicating coarse-mode particles for small values (i.e., <1.5) and fine-mode particles for larger SAE values (i.e., >1.5) [55]. The SAE values observed for cheatgrass (i.e., 0.5-3.5, Table 2) indicated a mixture of fine, medium, and coarse mode particles. The AAE values (405/781 nm) observed for cheatgrass (i.e., 2.04-6.91, Table 2) indicated moderate to strong dominance of BrC absorption.

Comparison of Intensive Optical Aerosol Properties for Grass Combustion Emissions
While there has been little work on SSA, AAE, and SAE of aerosols emitted from the open combustion of grasses, Lewis et al. [11] reported SSA, AAE, and SAE values for aerosols emitted from the combustion of rice straw ( Table 2). The AAE 405/781 nm range observed by us for cheatgrass emissions (2.0-6.9) for the wavelength pair of 405/870 included the AAE 405/870 nm value of 2.8 that was reported by Lewis et al. [11] for rice straw emissions. Their SAE (405/532 nm) values (1.7-2.2) measured for rice straw combustion emissions were also well within the wider range (0.5-3.5) observed by us for cheatgrass combustion emissions. SSA measured by them at 405 nm for rice straw combustion emissions was 0.88, while our result for 405 nm SSA of cheatgrass combustion emissions was quite similar (i.e., SSA = 0.083 ± 0.02). Figure 3 is a plot of the SAE 405/532 nm values found in this study along with rice straw SAE 405/532 nm derived from Figure 4 of Lewis et al. [11], showing that the rice straw SAE was generally higher, more variable, and without a monotonic decrease when compared to our measurements for cheatgrass, likely due to combustion conditions and fuel properties. At the beginning of our cheatgrass burn (see Figure 3), the SAE was quite high (i.e., 2.26), indicating the predominant emission of fine mode particles. As the burn evolved, the SAE of cheatgrass decreased, indicating an increase of particle sizes with a shift to a mixture of fine and medium mode particles.
Atmosphere 2020, 11, x; doi: FOR PEER REVIEW www.mdpi.com/journal/atmosphere for rice straw emissions. Their SAE (405/532 nm) values (1.7-2.2) measured for rice straw combustion emissions were also well within the wider range (0.5-3.5) observed by us for cheatgrass combustion emissions. SSA measured by them at 405 nm for rice straw combustion emissions was 0.88, while our result for 405 nm SSA of cheatgrass combustion emissions was quite similar (i.e., SSA = 0.083 ± 0.02). Figure 3 is a plot of the SAE 405/532 nm values found in this study along with rice straw SAE 405/532 nm derived from Figure 4 of Lewis et al. [11], showing that the rice straw SAE was generally higher, more variable, and without a monotonic decrease when compared to our measurements for cheatgrass, likely due to combustion conditions and fuel properties. At the beginning of our cheatgrass burn (see Figure 3), the SAE was quite high (i.e., 2.26), indicating the predominant emission of fine mode particles. As the burn evolved, the SAE of cheatgrass decreased, indicating an increase of particle sizes with a shift to a mixture of fine and medium mode particles.

Cheatgrass Emission Factors and Comparison to Previously Reported EFs for Grasses
Our averaged cheatgrass EFs are listed in Table 3, along with values that had previously been reported for similar types of fuel. This table includes previously reported EFs [50] for prescribed grassland burns and results from Chen et al. [7], who performed open-combustion in the laboratory similar to the methods used here. Results listed in Table 3 showed that our cheatgrass CO 2 EFs were close (within 1%) to the average of the other CO 2 EFs included and within 10% of the extremes of the other CO 2 EFs included. As expected, CO emission factors reported had a much larger range (i.e., 20 to 69 g/kg) than those for CO 2 with our cheatgrass CO EFs near (within~6%) the average of the other CO EFs included. The larger range was likely due to different prevalence of flaming and smoldering combustion for the different fuels and burns. Our PM EFs for cheatgrass were significantly higher (i.e., 18 ± 4 g/kg) than the other PM EFs included that ranged from 2.2 ± 1.1 to 10 ± 1, indicating a cheatgrass PM EF about 2.5 times larger than the average of the other three PM EFs included in Table 3. Table 3. EFs (emission factors) for cheatgrass are listed with previously reported data from a prescribed grassland fire [51,52], open combustion of Dambo and Montana grass [7], and savanna and grassland fire [51,52]. Our reported cheatgrass EFs of CO and CO 2 showed good agreement (within 1% and~6%) with reported EFs of other fuels, while the cheatgrass particulate matter (PM) EFs were~2.5 times larger than the average of the other three EFs.
The EFs of 82 individual polar organic compounds in gas and particle-particle phases of cheatgrass combustion emissions are listed in Table S1. The most abundant compound among all the analyzed organic species in the particle-phase emissions was levoglucosan, with an EF of 95.0 ± 24.9 µg/g ( Figure 4a). For savanna grass combustion [56], it was approximately five times higher (500 µg/g). Suberic and palmitic acids were the second and third largest contributors to the organic mass of the cheatgrass particle-phase emissions, with EFs of 22.9 ± 5.7 and 5.0 ± 1.2 µg/g, respectively. As expected, the top three contributors to the particle-phase organic emissions (Figure 4a) were different from those for the gas phase emissions (Figure 4b). The EFs for gas-phase guaiacol (17.6 ± 4.4 µg/g), syringol (14.8 ± 3.7 µg/g), and 4-methyl guaiacol (5.0 ± 1.2 µg/g) were highest among all the volatile and semi-volatile organic species collected on the XAD cartridges and analyzed by us (Table S1). Yatavelli et al. [57] and Sengupta et al. [48] also reported guaiacol and 4-methyl guaiacol as major contributors to the gas-phase organic emissions. Atmosphere 2020, 11, x FOR PEER REVIEW 10 of 17 Atmosphere 2020, 11, x; doi: FOR PEER REVIEW www.mdpi.com/journal/atmosphere   Table 4. EFs for gas-and particle-phase organic compounds in combustion emissions from cheatgrass, ponderosa pine needles, and savanna grass (units: µg g −1 of consumed fuel); BDL-below detection limit; the standard deviations represent analytical uncertainty.

Conclusions
Ecosystems in the Great Basin of the Western USA are greatly affected by the invasion of cheatgrass. While cheatgrass has comparatively low available fuel mass for wildland fires compared to the fuel available to burn in conifer or boreal forests, wildland fires in the Great Basin will continue to become more frequent and intense, while cheatgrass continues to invade the Great Basin.
The open laboratory combustion of cheatgrass was observed to be mostly complete combustion with high MCE. Optical properties reported here (i.e., SAE, AAE, and SSA) for cheatgrass combustion emissions were similar to those previously reported for rice straw combustion emissions. Optical absorption by particulate emissions was dominated by BrC with only little absorption by BC. Comparison between our fuel-based emission factors (i.e., EF CO2 , EF CO , and EF PM ) and other previously reported EFs for grassland and savannah fires showed fuel-based mass emission factors for cheatgrass fires, compared with those for other grasses and grasslands, were comparable for CO 2 (within 10%), with larger discrepancies (up to~50%) for CO and even larger ones (up to a factor of 10) for PM EFs. While our cheatgrass CO 2 and CO emission factors were near the average of those for comparable fuels included in this comparison, cheatgrass PM emission factors were significantly (~2.5 times) higher.
The EFs of 82 gas-and particle-phase organic compounds in the combustion emissions of cheatgrass were analyzed and compared with those of ponderosa pine needles, eucalyptus, and the savanna grass emissions. Levoglucosan was the largest contributor of all the analyzed organic species, which agreed with results for the analyses of combustion emissions from other wildland fuels. Compared with the EFs for ponderosa pine needles and eucalyptus, the total (sum of gas and particle-phase emissions) combustion EFs for cheatgrass were significantly lower (~23 times) than those for ponderosa pine needle, while they were approximately seven times higher than those for eucalyptus. This large variation of the total EFs for the organic species between fuels demonstrated that analysis of combustion emissions for specific fuels is very important, especially for studying and modeling the chemistry of biomass-burning emissions.