Enhancement of acidic gases in biomass burning impacted air masses over Canada

,


AOnospheric Formic and Acetic Acids
Over the past 10 years it has been demonstrated that carboxylic acids are ubiquitous components of the global troposphere.Keene et al. [1983] found that HCOOH and CH3COOH were the dominant acidity components of precipitation in remote regions, but these organic acids comprised only a minor portion of the free acidity in polluted regions.In general, gas phase carboxylic acids have a relatively short atmospheric lifetime, of the order of a few days due to removal by precipitation [Keene and Galloway, 1988] and dry deposition [Talbot et al., 1988b].Carboxylic acids have discernible seasonal and diurnal signals in the near-surface atmosphere, with the highest gas phase mixing ratios occurring during the summer and in the afternoon, respectively [Talbot et al., 1988b].
Known sources of vapor phase carboxylic acids include direct emission from vegetation and soils, photochemical production from natural and anthropogenic hydrocarbons, biomass-burning, and vehicular emissions [Amts and Gay, 1979;Kawamura et al., 1986;Keene and Galloway, 1986;Andreae et al., 1988b;Jacob and Wofsy, 1988;Talbot et al., 1988b].The seasonal and diurnal variations of carboxylic acid mixing ratios in the atmosphere indicate that natural sources, direct emissions and subsequent secondary production are important at midlatitudes [Talbot et al., 1988b].Biomass-burning may be a significant source of HCOOH(g) and CH3COOH(g ) to the atmosphere in some remote regions [Talbot et al., 1988b;Helas et al., 1992].
While the work of Andreae et al. [1988a] indicated that atmospheric (COO')2 may be naturally produced as a result of biomass-burning, it is still not certain if (COO')2 is directly emitted from fires or is a secondary product.Anthropogenically derived (COOH)2 may be a product of the gas or aqueous phase oxidation of glyoxal (CHOCHO) [Norton et al., 1983], a compound that has been photochemically produced from aromatic hydrocarbons in smog chamber studies [Nofima et al., 1974].Norton et al. [1983] also suspected that their HNO3 nylon filter gas-sampling system also collected small amounts of gaseous (COOH)2 or gaseous oxalyl chloride ((COC1)2), both of which become (COO')2 in aqueous solution.Gaseous (COCI)• can be produced by oxidation of perchloroethylene (C2C14) by OH [Howard, 1976].Thus far no one has been able to confirm the existence of gaseous (COOH)2 in the atmosphere.

Gas Sampling
Ambient air was brought inside the aircraft though a 4cm ID ceramic-coated low-carbon (CCLC) steel inlet at a constant flow rate of 225 L/min.This inlet provided a constant flowing airstream from which the mist chamber, a water-soluble gas sampler, subsampled for acidic gases at a rate of 30-40 L/min.The CCLC inlet was mounted perpendicular to the aircraft, extending 30 cm out from the fuselage into the free airstream.The distance to the free airstream at this location on the aircraft is about 15 cm (NASA Global Troposphere Experiment (GTE) Project Office, NASA Langley Research Center, Hampton, Virginia, unpublished results, 1983).The CCLC steel inlet was tested in our lab and found to have 100% passing efficiency for HNO 3 and SO•.We have not tested the passing efficiency of the inlet for carboxylic and oxalic acids.Given that experiments in our laboratory indicate that HNO3 is more prone to inlet wall losses than carboxylic acids, it has been assumed that it has a passing efficiency of 100% for the CCLC inlet.
The CCLC steel inlet was not found to have any detectable "memory" effect (i.e., release of HNO3 from the CCLC walls) while sampling "clean" air after 2-4 hours of exposure to 1-10 parts per billion by volume (ppbv) concentrations of HNO3.Thus after flying through concentrated plumes and then back into "clean" air, we believe that acidic gases will not be released off the inlet walls influencing cleaner samples.
Prior to each flight the mist chamber was cleaned by placing 10 ml of ultrapure deionized water in the mist chamber, pulling air though a carbonate/charcoal filter assembly to generate "clean" air, discarding the rinse water, and then repeating the process a number of times.The entire cleaning procedure took 2 hours and consisted of five short (3 min) rinses, followed by 10 long (10 min) rinses.The final two rinses were saved as blanks and treated in an identical manner to actual samples.
The carbonate/charcoal filter assembly consisted of two 47-mm carbonate-impregnated glass fiber filters and an activated charcoal filter cartridge followed by a Teflon filter.The carbonate filters remove all particles and acidic gases from the airstream, while the charcoal cartridge ensured that organic acids were completely removed from the cleaning airstream.The final Teflon filter collected any particles liberated from the charcoal cartridge.The carbonate-impregnating solution was 1.2 M Na2CO 3 in a 5% glycerol/95% deionized water solution.Filters were impregnated just prior to the field mission.This process involved soaking the filters in the carbonate/glycerol solution and then drying them at 80'C for 100 min.They were taken to the field doubly sealed in polyethylene bags.Prior to each cleaning session, the carbonate/charcoal filter assembly was loaded with fresh carbonate filters.The charcoal cartridge lasted the whole field mission.

Chemical Analysis
Mist chamber samples were analyzed immediately after each aircraft mission in a field laboratory.Our field laboratory was set up in the student biochemistry labs at Canador College, North Bay, Ontario, and moved to the Aurora Hotel in Goose Bay, Newfoundland, when the Electra switched its base of operations.
Mist chamber samples were run on two independent Dionex ion chromatographic (IC) systems (inorganic anions and carboxylic acids) modified with manual Rheodyne 7010 injection valves.In addition, peristaltic pumps were used to transport the micromembrane suppressor chemical regenerant.
The carboxylic acids were quantified on a Dionex AS4 column using a 0.4 mM Na2CO3 eluant and 25 mN H•.SO4 suppressor regenerant.Both systems have an analytical precision of 3-5 % for the species of interest here.
During ABLE 3B the mist chamber/ion chromatography (MC/IC) gas-sampling system had average detection limits of 60 parts per trillion volume (pptv) HCOOH, 90 pptv CH3COOH, 10 pptv HNO3, and 10 pptv (COOH)2 , assuming an average solution volume of 13.3 ml, an average sampled air volume of 0.5 m 3, and analytical detection limits (in/•mol/L) of 0.1 (HCOO'), 0.15 (CH3COO'), 0.02 (HNO3), and 0.02 ((COO')•).The mist chamber blanks never contained a significant level of any of the species of interest.The uncertainties assigned to the atmospheric mixing ratios reported for MC/IC samples were calculated using a propagation of errors that placed equal weighting on the following components: analytical uncertainty, air volume measurement uncertainty, collection efficiency, and inlet passing efficiency.The reported mixing ratios of HCOOH and CH3COOH have an overall uncertainty of +15% and +20% for both HNO 3 and (COOH)2.

Ancillary Measurements
The ancillary chemical and meteorological

Enhancement Factors
Historically, biomass-burning emission factors have been defined as the amount of species X that is emitted per unit of biomass combusted [Cnttzen et al., 1979;Talbot et al., 1988b].This definition has been loosely applied to measurements of species X in biomass-burning plumes [Andreae et al., 1988a;Helas et al., 1992].However, there are numerous and significant differences between measurements made directly above a fire and samples collected in an aged plume.To clarify the terminology here, we use the term enhancement factor to refer to measurements of species X in combustion plumes far away from the source, while emission factors reflect measurements of species X in combustion emissions at or very near the source.
Carbon dioxide and CO have both been used as measures of biomass combustion since the release of these species has been related to the amount of C in the fuel materials.Since CO was measured from the ABLE 3B Electra aircraft and CO2 was not, the enhancement ratios reported here were normalized relative to CO and calculated as the unitless ratio zXX//xCO.The individual plume numbers correspond to the missions in which the impacted air mass was encountered.Although there were often more than one mist chamber sample with elevated mixing ratios collected near a particular plume, many of these 10-to-20-min integrated samples were partially diluted with background air.As a result, the mist chamber sample coincident with the peak CO mixing ratio was selected as the most representative of the "plume" composition and the most suitable for use in determining a /xX//xCO enhancement factor.Plume enhancement factors were determined two different ways: (1) as the enhancement of species X (/xX) in units of its molar mixing ratio (pptv) divided by the enhancement of CO (/xCO) in pptv (see Table 1) and (2) as the slope of the linear regression relating the mixing ratio of species X (ppbv) versus the average mixing ratio of CO (ppbv) (see The species X versus CO slope enhancement factors were determined using the CO mixing ratio averaged over the start and stop times of the mist chamber sample.The regression slope enhancement factors were derived using all the mist chamber samples collected for a particular mission.Only those CO relationships with an r 2 correlation coefficient greater than 0.75 are included in Table 2.

Enhanced Plume Mixing Ratios
Enhanced mixing ratios of gaseous (COOH)2 , HCOOH, CH3COOH , and HNO3 were observed in association with biomass-burning plumes over the Canadian study regions.Examples of this relationship are illustrated using vertical composites from missions 6, 9, and 11 (Figures la,b,c), which show the coincidence of the maxima in the mixing ratios of these four acidic gases with that of CO.The combustion plumes studied here were encountered on missions 4, 6, 8, 9, 10, 11, 13, 16, and 22.The C2C14 mixing ratios for plumes 10, 16, and 22 (18-82 pptv) were significantly enhanced over the 12-14 pptv of C2C14    the foremost being insufficient data coverage.In addition, many of the plumes that are not included in this analysis were not discretely captured with our 15-min sampling resolution and as a result diluted with background air to such a degree that a meaningful enhancement factor could not be calculated.

A summary of the ancillary chemical and meteorological parameters of each of the plumes discussed in this study can be found in Table 3. Parts of this table are based on a study by
Emissions from hotter, flaming fires are more likely to be transmitted to higher altitudes initially and are therefore more likely to be transported long distances.After the primary fire, emissions from smoldering fires could also be lifted to higher altitudes by vertical convective processes.Days later, after mixing in transport, it may be difficult to distinguish if an air mass is a product of a flaming or smoldering fire.In either case, these air masses should probably be classified as biomass-burning impacted rather than biomass-burning plumes.
It is important to consider that even for the fresh plumes, a 10-min sampling resolution will not provide peak emission ratios, since the aircraft was not necessarily in the fresh plume the entire time.Dilution with background air does occur during the collection of plume samples at these temporal resolutions.In this respect, the regression slope technique of calculating enhancement factors may be a more appropriate method, assuming one still obtains a sufficient number of measurements to get a meaningful regression.In either case, for those species that are directly emitted during combustion, an enhancement factor can be viewed as a lower limit of the actual emission factor, considering the many removal mechanisms or reactions which may influence a particular species over time.It is more ambiguous as what an enhancement factor actually represents for the species that are produced in the plumes during transport, as is believed to occur in the case of HCOOH [Helas et al., 1992].Despite having varying ages and sources, there is relatively little variability in the biomass-burning enhancement factor values for the different plumes (Figure 5, Table 1).The relatively small difference between the enhancement factors for the biomass-burning plumes and the anthropogenically impacted plumes (missions 10 and 16) could lead one to conclude that during transit across the North American continent to Canada, both anthropogenic and natural processes are injecting combustion products into these air masses.Subsequently, the combustion products sampled in these plumes cannot be easily attributed to one source.
The comparison of emission characteristics related to biomass-burning versus anthropogenic/urban sources is best illustrated by these two extremes: a fresh bog fire (plume 9b) and an aged plume of industrial pollution off the eastern coast of the United States (plume 22).It is important to note that plume 22 did not impact Canadian air masses.
The anthropogenic AHNO3/ACO enhancement factor (plume 22) is approximately 6 times higher than any other HNO 3 enhancement factor (Figure 5), indicating that HNO 3 may be much more efficiently produced in urban combustion plumes.Nevertheless, biomass-burning is still a primary source of HNO 3 to the subarctic Canadian troposphere in the summertime [Talbot et al.,

this issue].
A similar comparison of (COOH)2 enhancement factors for plume 22 suggests that urban/industrial combustion may also be an important source of gaseous (COOH)2.In contrast, HCOOH appears to have higher enhancement factors in the biomass-burning plumes than in the industrial plume.This rough comparison of natural and anthropogenic sources does not show a clear trend for CH3COOH.

Comparison of Freslt and Aged ABLE 3B Plumes
Another potentially useful comparison is to see how the chemical composition of fresh differ from more aged biomass-burning emissions.A higher enhancement factor in an aged plume would indicate that a species may be produced by secondary chemical reactions during transport.A lower enhancement factor value in aged plumes suggests an influence from removal mechanisms, such as washout of soluble species or chemical decomposition.Certainly many homogeneous and heterogeneous chemical reactions (e.g., acid/base, photochemical, gas-to-particle conversion) are occurring during transport.
Unfortunately, there is not an established reliable method of determining the relative age of a biomassburning plume.However, it is possible to crudely discriminate the relative age of a few plumes using plume 9b as a fresh plume and plumes 6 and 11 (plumes encountered at higher altitudes (>3 km)) as more aged examples.Using this first-order age classification, there is no significant difference among the HNO3, HCOOH, and CH3COOH AX/ACO enhancement factors from plume 9b (fresh) and plumes 6 and 11 (more aged).An exception to this rule is plume 11, which had a much lower AHNOa/ACO and AHCOOH/ACO enhancement factor values (Figure 5 between fresh and aged biomass-burning emissions.The results of the fresh burning experiments in the laboratory portray CH3COOH as the primary carboxylic acid emitted from biomass-burning, while combustion plumes encountered over North America, South America, and Africa contain HCOOH at mixing ratios equal to or greater than those of CH3COOH.The fact that enhancement factors from fresh biomass-burning plumes (missions 4 and 6) also exhibited this enhancement of HCOOH over CH3COOH (Figure 9), implies that if HCOOH is produced via secondary chemical reactions in the plume, it occurs fairly soon after emission.Sanhueza et al. [1989] noticed that precipitation collected during periods of savanna burning in Venezuela had higher concentrations of HCOOH and CH3COOH than precipitation collected during nonburning periods.Similarly, Sanhueza [1991] noted that the biomass-burning impacted precipitation had concentrations of HCOOH about 1.5 times higher than CH3COOH.They concluded that the difference between the HCOOH/CH3COOH ratios of biomass-burning impacted precipitation and the laboratory-burning ratios of Talbot et al. [1988b] could be attributed to "atmospheric processes" rather than direct emission.Helas et al. [1992] also recognized this disparity and proposed that carboxylic acids (predominately HCOOH) are being produced in the biomass-burning plumes via the oxidation of olefins by 03.HCOOH in fire plumes when compared to ratios in "fresh" biomass-burning emissions determined from laboratory experiments.However, the similarities between the enhancement factors of the "fresh" (e.g., plume 9b) and the "aged" (e.g., plume 6) plumes encountered on ABLE 3B suggest that the HCOOH production in the plumes occurs in the first couple hours after emission.Combustion (biomass-burning and anthropogenic) appears to be the only known source of atmospheric (COO'),_.Only missions 10, 16, and 22 showed significant influence of anthropogenic pollution, as indicated by elevated C2C14 and F-12 mixing ratios.Samples from missions 16 and 22 demonstrated a strong correlation between CO and (COOH),, suggesting that urban/ industrial combustion processes may be important sources for atmospheric oxalate.Although anthropogenic combustion appears to be a principal source of HNO3 over parts of North America, biomass-burning seems to be a significant source of HNO3 in the subarctic regions of Canada.Biomass-burning appears to be a primary source of HCOOH and CH3COOH to the Canadian troposphere.
Future experiments should help determine if (COO'),_, HCOOH, and CH3COOH are produced primarily from biomass-burning or urban/industrial combustion.Stable C isotopic analysis is one tool that could be used to help separate anthropogenic and biomass-burning derived (COO')2, HCOOH, and CH3COOH.More biomassburning experiments, sampling fresh and aged emissions, might also shed some light on the processes which potentially produce HCOOH in biomass-burning plumes.In addition, HNO3 should also be measured in these experiments to determine if it is directly emitted or not from natural and anthropogenic combustion.
gases from the airstream before the droplets are trapped by a hydrophobic membrane (Teflon filter) and drip back down into the reservoir of extracting solution.The large surface-area-to-volume ratio of the mist droplets results in efficient removal of soluble gases from the airstream into solution.The extracting solution can be analyzed by a variety of colorimetric or chromatographic techniques to quantify ions of interest in solution.Ambient air samples were concentrated from 4 to 30 rain at a flow rate of 30-40 slpm, as determined by an integrating linear mass flow meter (Teledyne Hastings-Raydist).Then, the integrated sample was withdrawn (---5 ml) from the mist chamber and placed in a 15-ml high density amber polyethylene bottle.An additional 5-ml rinse of deionized water was placed in the mist chamber and ambient air was sampled for 1 additional minute.The rinse water was removed from the mist chamber, added to the sample bottle, and the entire sample was immediately placed in an ice chest.As soon as possible, after each flight, the samples were preserved with CHC13 and analyzed for NO3" CHOO'flight profiles and inferred mixing ratios (high or low) based on real-time observations of CO, 03, NOx, and NO r The vast majority of the gas samples were obtained while flying at a constant altitude.Samples collected during vertical spirals had a resolution of 1000 -2000 m.When operating the mist chamber on the Electra aircraft at altitudes above 4 km, where the ambient air temperature was below freezing, it was necessary to heat the glass exterior of the mist chamber 1 ø-2øC with resistance foil heaters to prevent ice from forming on the nozzles.

Langley
measurements utilized here were obtained and are available from the ABLE 3B data archive at the NASA GIT and TAMMS data sets were merged to the sampling times of the University of New Hampshire (UNH) mist chamber and aerosol samples by averaging all the data points that fell within the start and stop times of a given sample.The CO and odd nitrogen mixing ratios reported here are the "merged" or averaged values.The entire NASA/GTE ABLE 3B data archive is available from the NASA/GTE project office (NASA Langley Research Center, Hampton, Virginia).The hydrocarbon sampling and analysis was performed by the University of California-Irvine [Blake et al., 1992].About 20 different hydrocarbons and halocarbons were determined by gas chromatography on the 60 samples routinely collected for each flight [Blake et al., this issue].The hydrocarbon values reported for a particular UNH sample are the average of hydrocarbon samples collected between the start and the stop of the UNH sample.Since the grab sample values cannot be integrated over our sampling times, hydrocarbon values reported here for a particular combustion plume represent samples collected coincident with the peak CO mixing ratio.Five-day backward isentropic air mass trajectories were used to indicate potential air mass source regions and transport pathways [Sl•ipham et al., 1992, this issue].

Fig. 3 .
Fig. 3. Comparison of the average "plume" mixing ratios with average "plume background" mixing ratios.Number near base of bar indicates number of samples averaged.P < 0.0001 for all species.
produced from incomplete combustion, the amount of CO emitted during biomass-burning is a function of the fire conditions.Almost all of the CO is emitted during the smoldering stage of a fire, where as much as 20% of the combusted C is emitted as CO [Cmtzen et al., 1985].Hot flaming fires, on the other hand, are not O2 starved and produce much less CO.Since CO has a relatively small mixing ratio in the remote atmosphere (i.e., _• 100 ppbv), a chemical lifetime of several months, is generally passive to removal by &positional processes, and is not involved in biological respiration, CO is generally thought to be one of the more appropriate species to which biomass-burning plume enhancements are ratioed [Andreae et al., 1988].However, CO can be produced via the incomplete oxidation of hydrocarbons in biomass-burning [Wofsy et al., this issue], so •3(/,6CO enhancement factors may be a slight under estimate of the amount of species X produced per unit of carbon in the original fuel.In any case, ,6X/,6CO ratios are more suitable than ,6X/,6CO: as emission factors for the smoldering stages of a fire [Wofsy et al., 1992; Cmtzen and Andreae, 1990].