A biomass burning source of C1–C4 alkyl nitrates

We report the first observations of the emission of five C1–C4 alkyl nitrates (methyl‐, ethyl‐, n‐propyl‐, i‐propyl‐, and 2‐butyl nitrate) from savanna burning. Average alkyl nitrate mixing ratios in the immediate vicinity of three bushfires in Northern Australia were 47–122 times higher than local background mixing ratios. These are the highest alkyl nitrate mixing ratios we have ever detected, with maximum mixing ratios exceeding 3 ppbv for methyl nitrate. Methyl nitrate dominated the alkyl nitrate emissions during the flaming stage of savanna burning, whereas C2–C4 alkyl nitrates were mostly emitted during the smoldering stage. To explain the formation of alkyl nitrates from biomass burning, we propose a reaction mechanism involving the combination of reactive radicals at high temperature. Bearing in mind the uncertainties associated with extrapolating small data sets to much larger scales, alkyl nitrate emissions from global savanna burning are estimated to be on the order of 8 Gg/yr.

[1] We report the first observations of the emission of five C 1 -C 4 alkyl nitrates (methyl-, ethyl-, n-propyl-, i-propyl-, and 2-butyl nitrate) from savanna burning. Average alkyl nitrate mixing ratios in the immediate vicinity of three bushfires in Northern Australia were 47-122 times higher than local background mixing ratios. These are the highest alkyl nitrate mixing ratios we have ever detected, with maximum mixing ratios exceeding 3 ppbv for methyl nitrate. Methyl nitrate dominated the alkyl nitrate emissions during the flaming stage of savanna burning, whereas C 2 -C 4 alkyl nitrates were mostly emitted during the smoldering stage. To explain the formation of alkyl nitrates from biomass burning, we propose a reaction mechanism involving the combination of reactive radicals at high temperature. Bearing in mind the uncertainties associated with extrapolating small data sets to much larger scales, alkyl nitrate emissions from global savanna burning are estimated to be on the order of 8 Gg/yr.

Introduction
[2] Alkyl nitrates (RONO 2 ) are an important component of the reactive odd nitrogen (NO y ) budget and they can serve as a reservoir for the long-range transport of nitrogen oxides (NO x = NO + NO 2 ). In the troposphere the concentration of NO x controls ozone (O 3 ) production and influences the oxidizing capacity [Talbot et al., 2000]. Whereas alkyl nitrates typically comprise less than 10% of NO y within continental air masses [Buhr et al., 1990;Shepson et al., 1993], they comprise a major component (20 -80%) of NO y in the equatorial marine boundary layer over the Pacific [Talbot et al., 2000;Blake et al., 2002].
[3] The two known pathways for alkyl nitrate production are marine emissions [Atlas et al., 1993;Blake et al., 2002;Chuck et al., 2002] and photochemical oxidation of parent hydrocarbons [Darnall et al., 1976]. Here we use air samples collected near active bushfires to investigate biomass burning as a potential source of alkyl nitrates. Biomass burning is a significant anthropogenic source of carbon dioxide (CO 2 ), carbon monoxide (CO), methane (CH 4 ), nitrous oxide (N 2 O), methyl halides, nonmethane hydrocarbons (NMHCs) and sulfur compounds [Andreae and Merlet, 2001 and references therein]. Although Friedli et al. [2001] detected enhancements of C 1 -C 5 alkyl nitrates from fires in temperate forests and sage scrub regions of the United States, the RONO 2 formation was proposed to occur via the known photochemical mechanism of alkyl radical reaction with oxygen and nitric oxide (NO). In our study, this mechanism does not explain the observed alkyl nitrate enhancements, and an alternative reaction mechanism is proposed. Alkyl nitrate emission ratios relative to CO 2 and CO are calculated, and the emissions are discussed in terms of their impact on local and regional NO y budgets.

Experimental
[4] The alkyl nitrate measurements were made at ground level in Arnhem Land, Northern Australia from Sept. 4 -9, 1999. Thirty-five whole air samples were collected in conditioned, evacuated 2-L stainless steel canisters within 3 m of three bushfires, two near Katherine (13°74 0 S, 131°73 0 E), and one further northeast in Arnhem Land (13°18 0 S, 133°58 0 E). Six additional air samples were collected in remote coastal and inland locations (Cairns, Jaibaru and Alice Springs) to determine local background mixing ratios. The samples were returned to the University of California, Irvine (UCI) and analyzed for alkyl nitrates, CO 2 , CO, CH 4 , NMHCs, halocarbons, and sulfur gases within two months of sample collection. Our group routinely performs canister integrity studies to ensure that the mixing ratios in the canisters remain stable between sample collection and analysis. During this study, an additional consideration is the freshness and high reactivity of the combustion emissions. Reactive gases such as NO and the hydroxyl radical (OH) do not survive for longer than a few minutes in our canisters, and the formation of alkyl nitrates inside the canisters is very unlikely.
[6] The alkyl nitrate measurements were calibrated using two whole air working standards that were analyzed every four samples and have been previously calibrated against a synthetic standard prepared at NCAR. Because the ECD response may not be linear at high mixing ratios, we instead injected a small sub-sample and then scaled the area under the chromatographic peak up to the proper pressure/volume. The alkyl nitrate detection limit is 0.02 pptv. The accuracy of our measurements is 10-20% for alkyl nitrates, 5 -7% for CO, and 3% for CO 2 . The measurement precision is 2% for alkyl nitrates in the 1 -10 pptv range (improving with increasing mixing ratio), 4 ppbv for CO and 2% for CO 2 .

Mixing Ratios and Emission Ratios
[7] The alkyl nitrate mixing ratios in the savanna burning samples were strongly enhanced relative to background levels (Table 1). For example the average MeONO 2 mixing ratio for the 35 bushfire samples was 105 times higher than the average local background mixing ratio. The alkyl nitrate mixing ratios from this study are also larger than mixing ratios we have measured both in cities and over productive oceanic regions (Table 1). In Karachi, Pakistan our group measured average C 1 -C 4 alkyl nitrate mixing ratios ranging from 4.7 -26.0 pptv [Barletta et al., 2002]. Over productive regions of the equatorial Pacific, we have recorded maximum C 1 -C 4 alkyl nitrate mixing ratios of 50 pptv [Blake et al., 2002]. The maximum C 1 -C 4 alkyl nitrate mixing ratios measured during the Australian fire study are the highest we have ever recorded, ranging from 130 -3300 pptv.
[9] Like CO, many trace gases are produced in reduced chemical states predominantly in the smoldering stage [Blake et al., 1996]. Consistent with this, the C 3 -C 4 alkyl nitrates showed a much better correlation with CO during the smoldering stage than with CO 2 during flaming (Table 2). Their ERs were 7 -13 times higher in the smoldering samples as compared to flaming. Interestingly, MeONO 2 showed a much better correlation with CO 2 during the flaming stage than with CO during smoldering. The magnitude of the MeONO 2 ER was similar during both fire stages. Ethyl nitrate showed intermediate characteristics in which it strongly correlated with both CO 2 and CO during the flaming and smoldering stages, respectively. The ER for EtONO 2 was five times higher during smoldering than during flaming.

Proposed Formation Mechanism
[10] The known photochemical pathway for the formation of alkyl nitrates is [Darnall et al., 1976]: where RH is a parent hydrocarbon, R is an alkyl radical, RO 2 is an alkyl peroxy radical, and RO is an alkoxy radical.  However, this mechanism does not explain the magnitude and pattern of the observed alkyl nitrate enhancements. In particular, the photochemical processing times were too short to account for the observed MeONO 2 enhancements, even with the high levels of OH that are expected in biomass burning plumes [Foberich et al., 1996], and the very high levels of CH 4 that were present (7.3 ± 2.6 ppmv during flaming; 39 ± 14 ppmv during smoldering).
[11] To explain the formation of alkyl nitrates during biomass burning, we propose a mechanism similar to equations (1) -(3b), but with RO reacting with NO 2 to form RONO 2 , rather than RO 2 with NO. Following the formation of RO 2 via equations (1) -(2), we propose that RO is quickly formed because of the abundance of radicals in the fire. The decomposition of larger RO molecules forms smaller, more stable RO radicals, which then quickly react with NO 2 to form RONO 2 (k $ 10 À11 cm 3 molec À1 s À1 ; DeMore et al., 1997): [12] By this pathway, the formation of an alkyl nitrate is not constrained by its RH + OH rate constant nor by its RO 2 + NO branching ratio (both of which are very small for MeONO 2 compared to the heavier alkyl nitrates), consistent with the relatively large amounts of MeONO 2 that were observed. The experimental data further corroborate this proposed mechanism because of the high correlation between MeONO 2 and CO 2 during the flaming stage of the fire in which temperature (T) and mixing ratios of NO x and short, stable radicals are the highest (T > $950 K during the flaming stage. By contrast, the formation of longer radicals is favoured during the cooler smoldering stage [T < $850 K; Crutzen and Goldammer, 1993], which is consistent with the better correlation of C 3 -C 4 alkyl nitrates with CO than with CO 2 .

Global Significance
[13] Despite the very high alkyl nitrate mixing ratios that were observed, the absolute magnitudes of their ERs are relatively small ( Table 2). The ERs of ethane and the biomass burning tracer methyl chloride (CH 3 Cl) relative to CO during smoldering were 3 -5 orders of magnitude larger ([4.0 ± 0.6] Â 10 À3 ; and [4.0 ± 0.9] Â 10 À4 ppmv/ ppmv, respectively). These CH 3 Cl and ethane ERs are comparable to those reported by Blake et al. [1996] for Brazilian and African savanna fires. By comparison, the alkyl nitrate ERs for the smoldering samples are 4 -11 times smaller than those reported relative to CO by Friedli et al. [2001] from North American wildfires. The reason for this difference is not clear, but could be related to different fuel composition. We note that in the case of the North American fires, a much smaller range of RONO 2 and CO mixing ratios was observed.
[14] Savanna fires are the single largest global source of biomass burning emissions [Andreae et al., 1996], and the ERs measured here can be used to estimate the annual alkyl nitrate release from global savanna burning and all types of biomass burning. Such calculations assume that the ERs from the three Australian savanna fires are representative of savanna burning and biomass burning in general, whereas ERs for individual trace gases vary with factors including CE and fuel composition (see above). Here we did not sample prefire fuel nor postfire ash, and we cannot report directly on fuel composition. However, Hurst et al. [1994] performed detailed analyses of the mass, carbon and nitrogen loads for fuel and ash in a nearby region of the Australian Northern Territory (12°S, 132°E). The reader is referred to this paper for detailed findings.
[15] Bearing the above assumptions in mind, global alkyl nitrate emissions from biomass burning were estimated using the measured ERs and the amounts of CO 2 and CO released annually from global savanna burning (6,070 Tg CO 2 yr À1 and 232 Tg CO yr À1 ) and all types of biomass burning (13,500 Tg CO 2 yr À1 and 748 Tg CO yr À1 ) [Andreae et al., 1996;Holloway et al., 2000]. The uncertainties in the CO 2 and CO estimates are at least a factor of two [Andreae et al., 1996]. Based on these values, global savanna burning emissions for the five C 1 -C 4 alkyl nitrates reported here are estimated to total roughly 6.6 Gg/yr (mainly as MeONO 2 ) during the flaming stage, and 1.2 Gg/yr (mainly as C 2 -C 4 alkyl nitrates) during the smoldering stage, for a total on the order of 8 Gg/yr (Table 3). For global biomass burning, summed C 1 -C 4 alkyl nitrate emissions on the order of 18 Gg/ yr are estimated. In both cases, MeONO 2 emissions during the flaming stage comprise a majority (roughly two-thirds) of the total C 1 -C 4 alkyl nitrate emissions.
[16] The MeONO 2 emissions from savanna and global biomass burning are estimated to total roughly 5.4 and 12.6 Gg/yr, respectively (Table 3). By comparison, for an atmosphere of about 1.7 Â 10 20 moles of air, a globally averaged MeONO 2 mixing ratio of roughly 2 pptv, and a global MeONO 2 lifetime of 0.1 yr, the global MeONO 2 source is calculated to be on the order of 300 Gg/yr. That is, MeONO 2 released from global biomass burning likely accounts for only a small portion ($ 4%) of the global MeONO 2 source.
[17] Likewise, the global release of alkyl nitrates from savanna burning (1.3 Gg N yr À1 ) is small compared to the amount of NO x that is emitted [1.1 Tg N yr À1 ; Granier et al., 1996]. However, alkyl nitrates have a relatively low reactivity compared to other components of the NO y budget, and RONO 2 to act as a reservoir for the long-range transport of NO x . In particular, MeONO 2 is the longestlived of the alkyl nitrates, and MeONO 2 emissions from biomass burning are a previously unknown source for the transport of NO x to downwind locations. In very aged biomass burning plumes, however, we have previously noted a lack of MeONO 2 enhancement [Blake et al., 1999]. This is likely a result of relatively low MeONO 2 ERs coupled with long transport times during which MeONO 2 can become depleted (thereby releasing NO x ) as a result of photochemistry, and diluted as a result of atmospheric mixing.

Conclusions
[18] Savanna biomass burning fires are major point sources of C 1 -C 4 alkyl nitrates. Average alkyl nitrate mixing ratios in the immediate vicinity of three Northern Australian bushfires are the highest we have ever detected, exceeding maximum mixing ratios we have measured both in city studies and over productive oceanic regions. Despite these high mixing ratios, the alkyl nitrate ERs are small compared to those of various NMHCs and of the biomass burning tracer CH 3 Cl. Methyl nitrate showed the highest alkyl nitrate ER, and its ERs relative to CO 2 (flaming) and CO (smoldering) were of similar magnitude. The C 2 -C 4 alkyl nitrates showed ERs were 5 -13 times higher during smoldering than during flaming. The magnitude and pattern of the alkyl nitrate enhancements cannot be explained by the traditional photochemical production mechanism, and we suggest that the alkyl nitrates were formed by RO + NO 2 reaction, rather than RO 2 + NO.
[19] Methyl nitrate emissions during the flaming stage contribute about two-thirds of the total estimated alkyl nitrate emissions during savanna and global biomass burning. However, biomass burning likely has a small impact on the global MeONO 2 budget. Similarly, global alkyl nitrate emissions from savanna fires are small in comparison to the amount of NO x that is released. Notwithstanding, the very high mixing ratios measured close to the fires (up to 3.3 ppbv of MeONO 2 ) suggest that alkyl nitrate release during biomass burning likely impacts the reactive nitrogen budget in remote regions with few local NO x sources.