Article pubs.acs.org/est
First Measurements of the Nitrogen Isotopic Composition of NOx from Biomass Burning Dorothy L. Fibiger†,§ and Meredith G. Hastings*,‡ †
Department of Chemistry, Brown University, Providence, Rhode Island 02912, United States Department of Earth, Environmental and Planetary Sciences and Institute at Brown for Environment and Society, Brown University, Providence, Rhode Island 02912, United States
‡
ABSTRACT: The nitrogen isotopic composition (δ15N) of NOx (NO + NO2) was measured during the fourth Fire Lab at Missoula Experiment (FLAME-4). The δ15NNOx produced by burning a variety of biomass types ranged from −7 to +12‰ (vs air N2). In the laboratory experiments, two types of emissions were sampled: “stack” fires where the emissions were measured within a few seconds of production from the fire and “chamber” fires where the emissions were held in a room for 1−2 h and sampled continuously. For both types of emissions sampled, the primary control on δ15N-NOx is the δ15N of the biomass burned (δ15N-biomass), although differences were found for δ15N-NOx between the two types of fires. For the stack emissions, δ15N-NOx = 0.41 × δ15N-biomass +1.0 (R2 = 0.83, p-value 1 μM were analyzed for isotopic composition as well. The total δ15N of the starting biomass was measured at the Marine Biological Laboratory Ecosystems Center Stable Isotope Facility. Analyses were conducted using a Europa ANCA-SL elemental analyzer−gas chromatograph preparation system interfaced with a Europa 20−20 continuous-flow gas source stable isotope ratio mass spectrometer. Analytical precision was ±0.1‰, based on replicate analyses of international reference materials.
and another when the samples were reduced. Precipitated MnO2 from all samples and blanks was later removed in the laboratory at Brown University by centrifugation and decanting. All samples and blanks were subsequently analyzed for NO3− concentration using colorimetric absorbance techniques (Smartchem 200, Westco Scientific Instruments, Inc.) at Brown University. The isotopic composition of NO3− was also determined at Brown University using the bacterial denitrifier method.31 In short, denitrifying bacteria that lack the N2O reductase enzyme convert NO3− in solution to gaseous N2O. The N2O analyte is then purified, cryofocused, separated from CO2 by gas chromatography, and analyzed on a Thermo Finnegan Delta V Plus Isotope Ratio Mass Spectrometer at m/z 44, 45, and 46. NO3− isotopic reference materials USGS34 and IAEA-N332 are processed in the same way, and the resultant N2O is used to linearly correct the 45/44 ratio to result in the final δ 15 N-NO 3 − . 33 The NO 3 − concentration and isotopic composition of the two blanks were averaged, and the δ15N-NOx is then determined for each sample by mass balance: δ15 NNOx =
δ15 Nsample[NO3−]sample − δ15 Nblank [NO3−]blank
3. RESULTS AND DISCUSSION The δ15N of NOx from all biomass fires varied between −7 and +12‰ (Figure 3, Table 1). These are the first measurements of
[NO3−]sample − [NO3−]blank (1)
The pooled standard deviation of IAEA-N3 and USGS34 were 0.4‰, while the error in δ15N-NOx across the entire method, based on repeated laboratory collections, is 1.5‰ (1σ). If realistic errors are propagated for each of the terms in eq 1, then using the minimum total [NO3−] found in this study and associated δ15N-NOx (Table 1) and a [NO3−] blank of 5 μM, we find that a [NO3−]blank/[NO3−]total ratio of 0.7 or more Table 1. δ15N-NOx and Amount of NOx Collected for Various Biomasses Measured biomass Black spruce
Giant cutgrass Organic hay
Ponderosa pine
Rice straw
Sawgrass
Wire grass Wire grass
chamber/stack
[NO3−] (μM)a
δ15N-NOx (‰)
stack chamber chamber chamber stack chamber stack chamber chamber stack stack stack chamber chamber chamber chamber chamber chamber stack stack stack stack chamber chamber
22.2 20.6 30.3 56 10.4 13.6 3.1 22.3 47.5 3.8 4.5 5.1 4.2 38.9 40.1 2.7 22.2 13.4 5.6 3.9 7.3 4.7 20.5 15.2
−2.7 −5 −5.9 −7.2 3.2 2.6 3.8 8.1 12 0.9 1.3 1.6 −1.5 −1.3 −0.1 0.5 4.4 2 2.9 1.7 3.2 1.6 −0.3 −0.9
Figure 3. Variation in δ15N-NOx from biomass burning is primarily from differences in biomasses. There are also differences in δ15N-NOx between (a) the chamber and (b) the stack fires. The error bars on both plots represent the ±1.5‰ uncertainty for the complete NOx collection and isotope determination method.12
δ15N-NOx from biomass burning and, as such, there are no other data to compare them with directly. Kundu et al.25 found the δ15N for total aerosol nitrate to be about 23‰ in Brazil. This is significantly higher than any observed δ15N-NOx measurements from this study. Turekian et al.26 conducted laboratory burns of eucalyptus and African grasses and determined that aerosol NO3− was 6.6‰ higher than the starting δ15N-biomass, and suggested that gaseous losses would have a lower δ15N than the biomass. Both of these findings indicate either that direct aerosol NO3− production from fires is heavily enriched in 15N or that there is significant fractionation in conversion of gaseous N compounds to aerosol NO3−. The observed variation here is likely caused by real variations in the
Concentration of NOx collected in solution as NO3−. Values are blank corrected and the blank was 5−7 μM. a
C
DOI: 10.1021/acs.est.6b03510 Environ. Sci. Technol. XXXX, XXX, XXX−XXX
Article
Environmental Science & Technology
0.37) and the chamber fire having a slope error of 0.15 (yintercept error, 0.78). It is currently unclear what is causing the difference between the chamber and stack fires and which, if either, is more reflective of real biomass burning derived NOx produced under environmental conditions. We can, however, eliminate some processes that would be purely a result of laboratory conditions. First, all stack fire collections occurred over the entire fire, so it should not be a difference in direct production of NOx. Additionally, we know this difference is not driven by ambient NOx in the room that is combining with the biomass-derived NOx and altering our measurements. If it were ambient NOx, then we would expect all δ15N-NOx values in the chamber fires to converge on the same value, relative to the stack fire values. Instead, we see the opposite, with the biomass with the most negative δ15N-NOx, black spruce, having a decreased δ15N-NOx in the chamber fires and the biomass with the most positive δ15N-NOx, organic hay, having a more positive δ15N-NOx in the chamber than the stack. In addition, there were several collections made that showed no detectable NO3− above the blank in the collection solution. If ambient NOx were significant, then we would have expected higher NO x concentrations derived from all collections. The difference between the chamber and stack fires also cannot be explained by NOx sticking to surfaces during the chamber fires. Teflon is generally used for NOx sampling to avoid the tendency of NOx to adhere to surfaces, but the room the emissions were held in had many surfaces of a variety of materials, including the walls, which were concrete, and the stack, which was steel. Both should provide ample surface area for NOx to adhere. We would, however, expect the isotope effect of wall losses to be consistent across all chamber fires, relative to the stack fires. In other words, wall losses should always induce a change in δ15N with the same sign. The actual results, however, have a range of isotope effects in the chamber fires relative to the stack fires (Figure 3, Table 1), which indicate that wall losses cannot explain the isotopic difference between fire types. Finally, the chamber fires resulted in generally higher NO3− concentration in solution and, therefore, a relatively smaller blank correction. This, however, is not skewing the results, as chamber fires resulting in less NO3− in solution are more similar in δ15N to other chamber fires than to stack fires of similar concentrations. For instance, the Ponderosa pine chamber burn with [NO3−] of 4.2 μM has a very similar δ15N-NOx (−1.5‰) to the other chamber burns (−1.3 and −0.1‰ and 38.9 and 40.1 μM, respectively). The δ15N-NOx is different than the three stack fires (0.9, 1.3 and 1.6‰), which have very similar concentrations (3.8, 4.5, and 5.1 μM) to that chamber fire. The same holds true for black spruce, where the stack fire has a very similar concentration to the chamber fires, but a significantly different δ15N-NOx (Table 1). The difference in δ15N-NOx between the stack and chamber fires cannot be explained by conditions incidental to the laboratory fire conditions, thus it seems likely that chemistry occurring while emissions are held in the chamber causes fractionation of the NOx. There are several loss pathways for NOx in biomass burning plumes including conversion to PAN or HNO3.34 The rates of production of both are highly variable and dependent on many plume components as well as photolysis rates. All fires were indoors, so there was no photolysis, no detectable PAN formation, and no evidence of significant oxidation chemistry.30 There was some HNO3
NOx emitted as the uncertainty of the NOx collection and isotope measurement system is only ±1.5‰ (1σ). The primary source of variability in δ15N-NOx from biomass burning is biomass type (Figure 3). Emissions from burning black spruce show the most negative δ15N values (ranging from −7 to −3‰), while emissions from burning organic hay show the most positive δ15N (+4 to +12‰) (Table 1). The range of δ15N-NOx from any biomass showed much smaller variability than across all biomass types, however, within a single biomass type, the range of δ15N still exceeded the error in the δ15N-NOx method. For example, the δ15N-NOx emitted from Ponderosa Pine ranged from −1.5 to +1.8‰. This variation is, primarily, due to the type of fire sampled, with the δ15N-NOx from stack fires ranging from +0.9‰ to +1.8‰; the δ15N-NOx from chamber fires was −1.5 to −0.1‰. The difference in δ15N-NOx between the stack and chamber fires is inconsistent in both magnitude and sign across biomass types; some showed higher δ15N-NOx in the stack fires (wire grass, ponderosa pine, giant cut grass, chamise), while others showed higher δ15N in the chamber fires (sugar cane, organic hay) (Figure 3). The variations between biomass types are significant, even when chamber and stack fires are separated. In both fire types, the variations in δ15N-NOx are primarily controlled by the δ15N of the starting biomass (Figure 4). In the stack fires, at least
Figure 4. Relationship between δ15N-NOx produced by biomass burning and the δ15N of the starting biomass. The stack burns (red x’s) and chamber burns (blue o’s) each show significant relationships, but the relationship is different between the burn types. If there are multiple measurements for a single biomass type, then a mean δ15NNOx value is used, so each point represents a distinct biomass.
83% of variation is described by a linear relationship of δ15NNOx = 0.41 × δ15N-biomass +1.0 (p-value
