Collection of NO and NO2 for Isotopic Analysis of NOx Emissions

There have been several measurements made of the nitrogen isotopic composition of gaseous NOx (NOx = NO + NO2) from various emission sources, utilizin...
5 downloads 0 Views 447KB Size
Download PDF
Article pubs.acs.org/ac

Collection of NO and NO2 for Isotopic Analysis of NOx Emissions Dorothy L. Fibiger,*,†,¶ Meredith G. Hastings,*,‡ Audrey F. Lew,‡ and Richard E. Peltier§ †

Brown University, Department of Chemistry, 324 Brook Street, Providence, Rhode Island 02912, United States Brown University, Department of Earth, Environmental and Planetary Sciences, 324 Brook Street, Providence, Rhode Island 02912, United States § University of Massachusetts, Environmental Health Sciences, 686 North Pleasant Street, Amherst, Massachusetts 01003, United States ‡

ABSTRACT: There have been several measurements made of the nitrogen isotopic composition of gaseous NOx (NOx = NO + NO2) from various emission sources, utilizing a wide variety of methods to collect the NOx in solution as nitrate or nitrite. However, previous collection techniques have not been verified for complete or efficient capture of NOx such that the isotopic composition of NOx remains unaltered during collection. Here, we present a method of collecting NOx (NO + NO2) in solution as nitrate to evaluate the nitrogen isotopic composition of the NOx (δ15N-NOx). Using a 0.25 M KMnO4 and 0.5 M NaOH solution, quantitative NOx collection was achieved under a variety of conditions in laboratory and field settings, allowing for isotopic analysis without correcting for fractionations. The uncertainty across the entire analytic procedure is ±1.5‰ (1σ). With this method, a more robust inventory of NOx source isotopic composition is possible, which has implications for studies of air quality and acid deposition. −0.5‰ and 1.4‰. Li and Wang5 created NOx sourced from fertilized soils in the laboratory and found δ15N ranging from −49 to −28 (n = 20) (Figure 1), with the range representing the evolution of NOx over time from microbial processes (δ15N values became progressively higher with subsequent measurements); however, no direct measures of natural soil emissions have been conducted. Moore6 reported δ15N of NO2 from “natural soil emanations,” but this was based only on the similarity of values obtained from a passive diffusion collector placed in a greenhouse (with no indication of fertilizer application in the greenhouse) and those obtained from ambient outdoor NOx in Boulder, CO. Additionally, there are no reported values of δ15N-NOx from biomass burning, which in many locations, should be the primary natural NOx source.7 More data exists for coal combustion and vehicle emissions; however, both positive and negative values for δ15N-NOx have been reported (see below). Recently, Felix and Elliott12 used Ogawa passive collectors to measure δ15N-NO2. These proprietary collectors use a small disc impregnated with triethanolamine to collect NO2 as NO2−. While they report that these values help constrain the δ15N of NOx sources, they offer no evidence that these collection techniques are appropriate for isotopic applications. The passive diffusion samplers utilized have been shown to be robust for concentration measurements, but that is accomplished, primarily, through rigorous laboratory calibration. It is

N

itrogen oxides (NOx = NO + NO2) are of great interest due to their influence on the main oxidants in the atmosphere (i.e., ozone and hydroxyl radical), as well as their subsequent transformation and deposition as nitric acid. Emissions of NOx are primarily as NO, and NO is rapidly oxidized to NO2 in the ambient atmosphere. Today, more than half of the global budget of NOx is sourced from fossil fuel combustion (i.e., coal, oil, gasoline), while lightning, biomass burning, and nitrification and denitrification by microbes in soils combined makes up most of the rest of the budget. A number of measurements of δ 1 5 N-NO x (δ 1 5 N = [(15N/14N)sample/(15N/14N)air−N2 − 1] × 1000‰, where air N2 is a reference) have been made for assorted NOx sources under a wide variety of conditions (Figure 1). While most δ15N-NOx measurements have used some combination of a basic and oxidizing solution to collect NOx in solution as nitrate (NO3−) or nitrite (NO2−), none have been laboratory verified to collect all of the NOx they are exposed to. This is a critical weakness in the attempts to use δ15N-NOx to trace NOx sources. If only a portion of the NOx is collected, isotopic fractionation can occur, altering the δ15N-NOx value measured such that it does not reflect the source value. It is vitally important, before attempting to apply these source values quantitatively, that we are confident about their accuracy. It has been suggested that natural sources of NOx have a more positive δ15N and anthropogenic sources have a negative δ15N,1−3 yet direct measures of δ15N-NOx do not support this. The only measurement of δ15N-NOx from a purely natural source is the value for lightning reported by Hoering,4 which was simulated with sparks in a laboratory and had values of © 2014 American Chemical Society

Received: August 8, 2014 Accepted: November 21, 2014 Published: November 21, 2014 12115

dx.doi.org/10.1021/ac502968e | Anal. Chem. 2014, 86, 12115−12121

Analytical Chemistry

Article

Figure 1. δ15N-NOx values by source of emissions. Prior measurements used a wide variety of NOx collection methods (see legend) and show great variation in δ15N (symbols). It is also noted in the legend if the method collects NO2 only. The laboratory study of soil emissions (purple open triangles) represents an evolution over time of the signal due to microbial processing of nitrate and ammonium added to the soils. Studies using Ogawa brand passive collectors collect NO2 only, and the filters are exposed for at least 1 week.

could be due to the difference in engine designs in the years between the studies (e.g., in 1990 South Africa, cars were not equipped with catalytic converters), the form of NOx collected (i.e., Heaton8 collected total NOx, while Ammann et al.9 collected NO and NO2 separately), differences in what is actually being measured (i.e., direct tailpipe versus roadside collections), differences in fuel N content, or fractionations associated with the collection methods. In addition to collection efficiency affecting the isotopes measured, there is the possibility of interference from other nitrogen containing compounds collected into solution. Ammonia is highly soluble and so is easily collected from the gas phase into solution and, because most of these solutions utilize an oxidizing environment to convert the NOx, can then also be converted to nitrate or nitrite. Felix et al.10 tested several NOx collection methods when making power plant measurements and found a range of δ15N-NOx of +9.0‰ to +25.6‰. The large range was accounted for by differences in fractionation with different NOx scrubber technology used in various power plants. The array of NOx collection methods, however, was only tested in a power plant utilizing over fire air scrubbing technology. Most other plants tested use selective catalytic reduction (SCR) or selective noncatalytic reduction (SNCR), both of which use an injection of ammonia into the stream to reduce NOx emissions (i.e., 4NO + 4 NH3 + O2 → 4N2 + 6H2O). Ammonia is known to interfere with certain NOx collection methods,16 and the H2SO4 and H2O2 collection utilized by Felix et al.10 in the rest of the power plants is very similar to the Environmental Protection Agency method for ammonia collection (EPA Conditional Test Method 027). Therefore, it is unknown if the difference in δ15N-NOx measured in the SCR/SNCR (13.6‰ to 25.6‰) and nonammonia NOx reduction (9.0‰ to 12.6‰) power plants (Figure 1) is driven by fractionation in the NOx removal step or by possible ammonia interference in the measurements. For all previous studies, efficiency of NOx collection has not been reported. If less than 100% of the NOx (or NO or NO2) is collected, fractionations can occur, leading to divergent measurements. For example, for a simple unidirectional reaction, kinetics dictate that 14N will react faster than 15N

known that the efficiency of NO2 uptake on the passive diffusion samplers can vary significantly under differing atmospheric conditions, such as temperature and humidity,14 while any fractionations that can occur at differing uptake rates are not quantified. While the range of δ15N-NO2 reported may be accurate, without rigorous testing, the reported values cannot be used as a quantitative measure and cannot be compared to other δ15N values in the literature. Prior to Felix and Elliott,12 the data in the literature represented primarily “active” collections of NOx, typically using a combination of basic and oxidizing solutions to collect NOx in solution as NO3− or NO2−. Most have utilized NaOH or KOH as the base, while the oxidants have included H2O2 and Guiacol.5,8,9 Additionally, collections have been made in solutions of triethanolamine (TEA) in water, which collects only NO2.15 Some of these prior measurements purport to collect total NOx;8,10 others collect NO and NO2 separately,9 while still others collect only NO2.12,15 These prior collections, however, show wide variations in the δ15N-NOx from the same source (Figure 1). Whether these variations show true variations within emissions from a single source or are derived from the measurement difference is impossible to determine with the lack of methodological testing. For example, δ15N-NOx from vehicle emissions has been reported in the range of −13‰ to +10‰. Whether this range reflects real differences in NO x production or fractionations with measurement cannot be determined with current information. When measuring δ15N-NOx from vehicles, Heaton8 used a NaOH/H2O2 solution, trapping direct tailpipe emissions inside a container and leaving them in contact with the solution for several days. The resulting δ15N-NOx ranged from −13‰ to −2‰. Ammann et al.,9 in contrast, placed denuder tubes coated with NaOH/Guiacol by the side of a highway and collected NO and NO2 separately for 3 days at a time. The δ15N-NO2 measured here ranged from +1.8‰ to +10.2‰ and δ15N-NO ranged from −4.7‰ to +9.8‰ (n = 9). Because NO and NO2 concentrations are not reported for each measurement, total δ15N-NOx cannot be calculated but falls somewhere in the −4.7‰ to +10.2‰ range. The difference in δ15N-NOx between the Heaton8 and Ammann et al.9 studies 12116

dx.doi.org/10.1021/ac502968e | Anal. Chem. 2014, 86, 12115−12121

Analytical Chemistry

Article

with N2 (99.998%) at a known flow through a mass flow controller (MKS Type 1179A) with a full scale of 5000 sccm. The NOx mixture was passed through the solution in a 250 or 500 mL gas-washing bottle with a fritted cylinder to scrub the NOx from solution. With a known flow rate and NOx concentration, the amount of NOx passed through could be calculated and the amount of NO3− collected in solution could be compared. The collection method was tested under a variety of conditions, including differing NaOH concentrations, frit coarseness, collection time, solution amount, and air flow rate. The method was effective at collecting 100% (±5%, 1σ) of the NOx and producing consistent isotope results under a wide variety of conditions (see below). Isotopic Analysis. After collecting the NOx in solution as NO3−, the KMnO4 must be removed from solution. If it is not removed quickly, any ammonia (NH3) collected can be oxidized to NO2− or NO3− and interfere with isotopic measurement.16 Interference of the isotopic measurement by NH3 is simple to avoid by fast removal of KMnO4 from solution. Margeson et al.16 found that conversion of NH3 to NO3− was both linear and slow; with collection of NOx in a 200 ppm NH3 atmosphere, the [NO3−] increased by 2.8% after 7 days. Under this scenario, after 36 h (the longest time before solution reduction in our experiments), there would be an increase of 0.6%. In the atmosphere, NH3 concentrations much lower than 200 ppm are more typical, so the percent of NO3− increase should be even smaller. Nonetheless, we removed KMnO4 as soon as possible after collection, and it is important to account for NH3 if solutions are stored with KMnO4 present for several days or weeks. We found the conversion of NO and NO2 to NO3− to be very fast, as all tank NOx collection solutions were reduced within 15 min of stopping collection and showed complete collection of the NOx. The KMnO4 is removed through reduction with 30% H2O2 to MnO2, which precipitates out of solution as a fine, dark precipitate. The H2O2 is added until the solution left behind is colorless. Excess H2O2 addition is not a concern, as it is removed catalytically by MnO2.16 After the KMnO4 is reduced, the solution is stable at room temperature in an HDPE bottle for several months. The MnO2 was removed from solution by centrifugation and decanting. The solution was divided into 3 60 mL centrifuge tubes and spun for 15 min at 4000 rpm. After decanting, the solution was neutralized using 12.1 N HCl (Fisher brand, reagent grade), based on the original solution volume and NaOH concentration. After reduction and neutralization, the solution can be analyzed for NO3− concentration. This was quantified utilizing standard colorimetric absorbance techniques (Smartchem 200, Westco Scientific Instruments, Inc.). The reproducibility for the NO3− concentration is ±0.3 μM (1σ) for repeated measurements of a sample. The NO3− concentration was then used to calculate injection volumes for isotopic analysis of 10 nmol of N. Isotopes were analyzed using the bacterial denitrifier method,18 and data was corrected following the scheme in Kaiser et al.19 In short, denitrifying bacteria that lack the N2O reductase enzyme convert NO3− to gaseous N2O. Using He as a carrier gas, the N2O is then purified, cryo-focused, and finally analyzed on a Thermo Finnegan Delta V Plus Isotope Ratio Mass Spectrometer at m/z 44, 45, and 46. Internationally recognized NO3− reference standards USGS34 and IAEA-N320 are run alongside samples and used to correct the resulting 45/ 44 ratio to result in the final δ15N.

such that the collection solution will reflect an overall lower δ15N-NOx than the true δ15N value in the environment. In addition, real source variations can lead to differences among δ15N-NOx from the same source type. Use of and the type of NOx removal technology employed when a measurement was made can have a large influence on δ15N-NOx.10 Additionally, whether emissions are measured directly from the source or after mixing with ambient air has also varied between studies and may influence the δ15N-NOx. The ambient conditions under which collections are made, such as temperature, humidity, and length of collection may contribute to variations observed. Finally, some measurements have collected NO and NO2 separately, while others have collected total NOx. In an effort to address the clear, fundamental need for an analytically verified, consistent technique for measuring the δ15N of NOx, we report here on laboratory and field testing of a method that efficiently collects NO and NO2 in solution as NO3− and can be used for isotopic analysis of the NOx without correction for fractionation. This is an important step toward being able to build a comprehensive inventory of δ15N-NOx from NOx sources, with implication for studies of air quality and acid deposition.17 By deploying a laboratory tested, reliable method for collecting NOx, we can accurately evaluate the δ15N-NOx from various sources and begin to apply this information quantitatively. Ultimately, having separate isotopic measures of different reactive nitrogen species (e.g., NO, NO2, total NOx, HNO3, etc.) will also be possible.

■

EXPERIMENTAL SECTION NOx (as NO, NO2, or NO + NO2) was collected in solution as NO3− and analyzed, offline, for δ15N. To collect the NOx, NO and/or NO2 was bubbled through a gas-washing bottle containing a 0.25 M KMnO4 and 0.5 M NaOH solution. The solution was then reduced and neutralized, and the resulting NO3− was analyzed for δ15N. The NOx was collected quantitatively under a wide range of conditions, including a variety of flow rates, temperatures, humidity, and gas-phase NOx concentrations, allowing for direct determination of δ15NNOx. Solution Preparation. To prepare 500 mL of the solution, 125 mL of 1 N KMnO4 (Fisher Brand, Fisher Scientific) is diluted with approximately 300 mL of 18.2 MΩ water. Concentrated NaOH, approximately 10 M, is prepared from reagent grade NaOH (Fisher Brand, Fisher Scientific). Twentyfive mL of the 10 M NaOH is then added to the KMnO4 solution, and the entire solution is diluted to 500 mL with additional 18.2 MΩ water. Note that it is important to dilute the KMnO4 significantly before adding the NaOH, as NaOH can cause concentrated KMnO4 to precipitate. The 10 M NaOH can be prepared ahead of time and stored in a HDPE amber bottle. The 0.25 M KMnO4 and 0.5 M NaOH solution should be prepared daily and must be stored in amber glass, as the NO3− blank (see below) was higher and more variable when stored in plastic. The complete solution should not be stored for significant lengths of time because NOx from the ambient air can be incorporated as NO3−. For the experiments conducted here, the solution was typically used (i.e., exposed to NOx) and reduced within 48 h. Experimental Setup. To test the collection methods in the lab, analytical tanks of 50 ppmv NO or NO2 in N2 were used at a known flow rate through an MKS flow controller with a full scale of 50 sccm. Both NO and NO2 were tested to ensure complete collection of total NOx. The NOx gas was diluted 12117

dx.doi.org/10.1021/ac502968e | Anal. Chem. 2014, 86, 12115−12121

Analytical Chemistry

Article

Blank Collection. Background NO3− was always found in the KMnO4, and it is critical to correct for the interference of any background NO3− in the starting solution. After preparing the KMnO4 and NaOH solution, a 50 mL aliquot of the solution was collected and treated the same way as the sample solution. While no detectable NO3− was found in any of the NaOH used, using the total solution as a blank allows for correction of any trace NO3− contribution from the NaOH or the water used. If a solution was made to use over the course of several hours, one aliquot was collected at the beginning and another at the end of the experiment. For these aliquots, reproducibility was excellent: concentration values typically agreed within 0.3 μM and isotopic values differed by less than 0.2‰, similar to the precision of the individual techniques. After analysis of the sample solution and the blank, the final sample isotopes can be calculated by mass balance: δ15 Nsample =

For the collections, the setup was identical to that in the laboratory except for air being drawn with a pump instead of pushed by the tank and the addition of air filtration. The air from the smog chamber was run first through a Millipore Fluoropore membrane filter (1.0 μm, 90 mm) to remove particulates and then through a Pall Life Sciences Nylasorb filter (1.0 μm, 90 mm) to remove nitric acid. Five sequential 0.5 h samples were collected from the chamber. For all samples, KMnO4 was reduced shortly after the final collection, less than 24 h after the collection solution was made. Field Deployment of Method. The method was also deployed on the roof of a building (17.1 m above ground) located between two heavily traveled highways in Providence, RI (222 Richmond St.) located less than 1 km from Interstate 95 and Interstate 195. Several separate measurements were made during the fall of 2013, at times of heavy traffic (7−9:30 and 15:30−18:30). During some of the field measurements, two identical setups were run to test the reproducibility of the method. Future field collections will include direct NOx concentration measurements to determine the collection efficiency in the field. Teflon tubing was attached to a railing at the side of the building and run 1−1.5 m to the sampling table. The air was first passed through a Millipore Fluoropore membrane filter (1.0 μm, 90 mm) to remove particulates. No precautions were taken in these experiments to limit HNO3 collection (i.e., via a Nylasorb filter), so it is possible that the final δ15N values reflect a combination of ambient NOx and HNO3. Other than the lack of HNO3 filter, the setup of this system was identical to that for the smog chamber measurements. All collections were done at a flow rate of approximately 4.5 L/min. Between collection times, the solution was stored in the gas-washing bottle in an indoor location. The sample KMnO4 was reduced shortly after the final collection time, which was typically within 48 h from the solution being made and always less than 48 h from start of the collection time. A blank was sampled from each solution made to correct the final NO3− concentration and isotopic values for its contribution.

δ15 Ntotal[NO3−]total − δ15 Nblank [NO3−]blank [NO3−]total − [NO3−]blank (1)

Blank Reduction. Several methods were tested to attempt to reduce the NO3− blank associated with KMnO4. Ion exchange resins that should remove NO3− actually resulted in an increased [NO3−] in solution, probably due to the ammonia substrate used in these resins. Recrystallization of KMnO4 showed no difference in [NO3−]. Several different brands and preparations of KMnO4 were tested, and the lowest background NO3− found was around 6 μM in solutions prepared with 1 N KMnO4 (Fisher Brand, Fisher Scientific). The background could be reduced to 5 μM using Fisher brand alumina (Adsorption, 80−200 mesh) as an ion exchange medium. In a batch method, 150 g of the alumina was mixed with 500 mL of 18.2 MΩ water and shaken vigorously. The water was removed through vacuum filtration with a Buchner funnel and filter paper. After rinsing, the alumina was added to 1 L of 1 N KMnO4 and shaken vigorously. The alumina was separated from the KMnO4, once again, through vacuum filtration. The alumina processing can impact the remaining NO3− isotopically, though the fractionations seem to be small, so it is critical to collect a blank with each batch and maximize the signal to blank ratio. Laboratory Measurement of Real Emissions. In addition to the NOx tank tests, the analytical method was used in both a laboratory and a field setting to measure source generated NOx emissions. In the laboratory, the method was used to measure NOx from diesel engine emissions in a diesel smog chamber at the University of Massachusetts at Amherst. Diesel exhaust was generated by a single cylinder Yanmar diesel engine (9.1hp, 406 cm3 displacement) coupled to a Pramac generator set (Pramac P6000s, Siena, Italy) which was located outside of the laboratory. The engine was run for 6 min at idle to warm the engine and used commercially available ultra low sulfur diesel fuel. After warm-up, a continuous 50% load was applied to the engine, and whole diesel exhaust was then withdrawn through a stainless steel pipe from the exhaust plenum at a location prior to the muffler. Exhaust was shunted to a 13m3 Teflon-lined smog chamber in the laboratory for 2 min 10 s, and then, the filtered and VOC-denuded room air was added until the correct dilution factor was achieved. In this case, whole diesel exhaust was diluted 100-fold.

■

RESULTS AND DISCUSSION Alternate Methods Tested. In addition to the KMnO4/ NaOH collection method we developed, we tested several Table 1. Variations in Fraction of NOx Collected and δ15NNOx with Varying [NOx] and Total Flow Rate N2 flow rate (L/min)

[NOx] (ppm)

fraction of NOx collecteda

δ15N (‰)

1 2 3 5 5 10

1.07 0.53 0.35 0.22 0.22 0.11

0.97 1.08 1.04 0.97 1.02 0.41

−60.3 −58.7 −60.7 −63.1 −61.2 n.a.b

a Moles of NOx passed through solution/mol NO3− collected in solution. bDue to the low fraction collected, the sample was not analyzed for isotopic composition.

other NOx collection techniques that have been used for isotopic analysis in the past. A solution of TEA in water has been used for NOx collection but has never been verified in the laboratory as an efficient collection method for isotopic analysis. Under a very select set of circumstances, we were 12118

dx.doi.org/10.1021/ac502968e | Anal. Chem. 2014, 86, 12115−12121

Analytical Chemistry

Article

Table 2. Reproducibility of δ15N-NOx for Sampling Across Different NOx Tanks NOx tank a

A A A A A A A Bb B B B B

[NOx] (ppm)

fraction of NOx collected

δ15N (‰)

blank/total Nc

0.032 0.032 0.032 0.032 0.032 0.022 0.022 0.22 0.22 0.22 0.22 0.22

0.91 0.96 1.04 0.98 1.06 0.99 1.06 0.97 1.02 0.96 1.04 0.93

−68.5 −72.1 −69.0 −71.6 −69.3 −70.4 −69.1 −63.1 −61.2 −60.4 −63.1 −64.1

0.18 0.17 0.16 0.16 0.15 0.23 0.22 0.13 0.14 0.15 0.14 0.15

TEA, which is important for any NOx collection method as sample NOx concentrations cannot be precisely predicted prior to sampling. In addition, many sources of NOx emissions have highly variable NOx concentrations.16 It should also be noted that the TEA collection approach can only scrub NO2 from air. In order to collect NOx with TEA, an oxidizer must be set up before the TEA solution to convert any NO to NO2 for collection. We also tested collection in NaOH/H2O2 solution. This solution has been used for NOx collection in the past;8 however, there was a persistent, very high nitrate background that we were unable to eliminate. We tested several different NaOH and H2O2 concentrations and were unable to reduce the background below 25 μM, making field application of this method of minimal use. Final Method. The KMnO4/NaOH solution proved the most effective at reliably collecting 100% of the NOx and demonstrating consistent isotopic results across a range of conditions. Margeson et al.16 demonstrated that a 0.25 M KMnO4 and 0.5 M NaOH solution can consistently collect 100% of NOx passed through when using this solution for NOx concentration analysis in power plant stacks. Margeson et al.21 later adjusted the method to use a 0.25 M KMnO4 and 1.0 M NaOH solution to collect additional acidic gases in solution, but when tested for isotopic analysis of NOx, this solution had a higher NO3− background and showed greater variability in isotopic composition. Using the 0.25 M KMnO4 and 0.5 M NaOH solution, 100% of NOx was collected under a variety of laboratory conditions. Total flow rate, NOx concentration, solution volume, and frit types were tested. While there are limits to the method, it proved robust under a wide range of conditions. Solution volume and frit type were unimportant in how efficiently NOx was collected in solution. As long as the solution was sufficient to cover the frit, which was placed less than 0.5 cm from the bottom of the gas-washing bottle, the fraction of NOx collected did not change. In addition, whether the frit was fine or coarse made no difference in fraction collected. The coarse frit may provide some benefit, as a MnO2 precipitate is formed when NOx is converted to NO3− and, over time, this may cause finer frits to become clogged. The bottle form (tall or short) was also irrelevant to collection, and custom gas-washing bottles of different widths tested as well were equally efficient. In most bottles tested, 100 mL of the 0.25 KMnO4 and 0.5 NaOH solution was sufficient to cover the frit and resulted in collecting all the NOx passed through. Minimizing the solution volume, while maintaining collection efficiency, allows for minimization of the background NO3− relative to the sample size, which is ideal.

Tank A was NO in N2. Mean δ15N-NO for tank A was −70.0‰, the standard deviation was 1.4‰. bTank B was NO2 in N2. Mean δ15NNO2 for tank B was −62.4‰, the standard deviation was 1.5‰. (Note that the tanks were supplied by different vendors and the δ15N difference may be due to differences in NOx concentration techniques or the difference in production of NO or NO2.) cThe ratio of moles of NO3− in the blank to the total moles of NO3− in the final solution. a

Table 3. Consecutive Collections of NOx from a Smog Chamber for δ15N-NOx Analysis collection

[NO3−] (μM)

δ15N (‰)

average flow (L/min)

blank/total N

1 2 3 4 5

108.1 43.2 55.8 53.8 56.6

−16.7 −17.5 −18.1 −19.2 −18.6

4.0 1.9 2.4 2.3 2.2

0.06 0.19 0.15 0.15 0.14

able to achieve 100% efficiency of NOx collection with TEA; however, this was observed only under specific methodological conditions that have not been presented in previous literature. For example, the age of the TEA solution effects the percentage of NO2 collected. A 20% TEA solution left in a sealed container at room temperature for 24 h resulted in a decrease in the fraction of NO2 collected by 10%. Concentrated TEA is more stable but still showed degradation over several weeks. Further, TEA concentration in solution also effected collection efficiency. Maximum efficiency was achieved with a TEA concentration of 10−20%. These low concentration solutions are advantageous because they are less viscous than higher concentrations; however, the lower concentrations show a larger drop in efficiency with solution age. NO2 concentration in air had limited effect on the efficiency of collection with

Table 4. NOx Collections Conducted on a Rooftop in Providence, RI to Analyze δ15N-NOx in an Urban Environment

a

dates of collection

collection time (hours)

[NO3−] (μM)

δ15N (‰)

temperature (°C)a

7/25/2013 to 7/26/2013 10/8/2013 to 10/9/2013 10/8/2013 to 10/9/2013 11/6/2013 to 11/7/2013 11/6/2013 to 11/7/2013 11/20/2013 to 11/21/2013 11/20/2013 to 11/21/2013

8.17 6.75 6.75 2.5 5.25 8.9 8.9

17.46 14.43 16.78 30.86 37.05 44.29 29.66

−5.5 −0.6 −1.3 −7.7 −6.7 −7.1 −6.7

19.2 15.8 15.8 17.1 17.1 3.28 3.28

humiditya wind direction (°E of N)a blank/total N 82.5 57.5 57.5 73.0 76.0 45.0 45.0

105 133 133 151 151 169 169

0.41 0.3 0.26 0.20 0.17 0.14 0.21

Temperature, humidity, and wind direction were measured at the sampling site with a Davis Instruments 6250 Vantage Vue weather station. 12119

dx.doi.org/10.1021/ac502968e | Anal. Chem. 2014, 86, 12115−12121

Analytical Chemistry

Article

The first approach to reducing NO3− background was to test different forms and brands of KMnO4 and find the one with the lowest starting NO3−. Across a wide array of crystal and solution KMnO4, the lowest NO3− background was in Fisher brand 1 N KMnO4. Many bottles of the Fisher brand 1 N KMnO4 have been tested, and all resulted in a final solution with [NO3−] from 5 to 7 μM. This range was true across multiple bottles from the same lot number, as well as between lot numbers. The starting [NO3−] in solution can be reduced farther by treatment with alumina. The Al2O3 (Fisher Brand, Fisher Scientific) was used in a batch method. 150 g of alumina was rinsed with H2O and then added to 1 L of 1 M KMnO4. After vigorous shaking, the alumina was filtered off. The 1 M KMnO4 remaining had a NO3− background reduced to 4 μM, creating a 1 μM reduction in the final solution. Note also that it is critical to only store and work with the KMnO4 solution in glass containers, as storage in plastic resulted in a higher and more variable background [NO3−]. Laboratory Emissions Measurements. The method was deployed to measure the δ15N of emissions in a smog chamber at University of Massachusetts at Amherst. Five sequential samples were taken from the chamber at 100-fold dillution, with each collection being 0.5 h in duration. NO x concentrations exceeded the measurement limit of the chemiluminescent analyzer on the smog chamber (1 ppm) but were targeted to be 1900 ppb based on the dilution. The five samples have a mean value of −18.0‰ and a standard deviation of 0.97‰ (Table 3). Field Emissions Measurements. The method was deployed on the rooftop of a building located between two highways in Providence, RI for several periods during the summer and fall of 2013. During several of the collection periods, identical setups were deployed to test the reproducibility of the method. The primary NOx emissions measured should be from vehicles; however, the primary intent of the collections was to test the collection method under varying field conditions (Table 4). Duplicate measurements made with identical setups show agreement better than the 1.5‰ error found, in the laboratory, across the entire collection method. All the urban measurements show somewhat negative δ15N values, consistent with the measurements of δ15N in the diesel smog chamber. The negative δ15N is also consistent with automobile measurements made by Heaton.8 The difference between the smog chamber δ15N-NOx (Table 3) values and the ambient urban δ15N-NOx (Table 4) could be driven by many factors, including the mix of diesel and gasoline vehicles on the road, other NOx sources in urban areas, or photochemical processing of NOx in the outdoor environment. The variations between different dates of the rooftop measurements show no correlation with collection time, [NO3−] collected, temperature, humidity, or wind direction. Whether the differences are caused by variations in the composition of traffic, the presence of HNO3, or photolytic processing will be the subject of future study. Both the smog chamber and rooftop measurements demonstrate the versatility and ease of use of this method. In both a controlled laboratory setting with high NO x concentrations and in the field with variable NOx concentrations, δ15N-NOx was very consistent and duplicate field measurements also showed high repeatability.

When testing gas tanks with extremely low humidity, appreciable evaporation was observed after an hour of collection at 5 L/min. The percent of NOx collected, however, was not affected as long as the solution level remained above the frit, such that the bubbling action was preserved. In the field, where humidity was much higher than the tanks, but highly variable, there was no observable solution evaporation. Flow Rate. A wide variety of air flow rates, from 2 mL/min to 10 L/min, were tested in the laboratory and very few limitations were found up to 5 L/min. N2 flow rates were varied from 1 to 10 L/min, while the NO flow was held constant. This served to vary the NO concentration simultaneously with total flow rate with remarkably consistent results. Any flow rate up to 5 L/min showed 100% collection of the NO and consistent isotopic results. At flow rates greater than 5 L/min, the collection efficiency dropped significantly (Table 1). This may be due to the rate at which NO dissolved in solution. The fraction of the NOx collected remains consistent around one at flow rates up to 5 L/min but still contains some variations. These variations can be accounted for with several sources of error in the NOx collection procedure. The flow controller and the concentration measurement can both induce error into the measurement, but the primary source of error is blank correction of the [NO3−]. Small variations in [NO3−] in the blank or sample can result in variable final [NO3−] and, therefore, fraction of NOx collected. Reproducibility. In two separate sets of experiments run with one NO2 and one NO tank, multiple collections were done with a total flow rate of 5 L/min. Results for each tank show very consistent fractions of the NOx collected and consistent δ15N, despite somewhat different δ15N between the two tanks (Table 2). The set of collections with tank A have a standard deviation of 1.4‰, while the set with tank B have a standard deviation of 1.5‰. Thus, we conclude the uncertainty across the entire method from NOx collection to isotope analysis and data correction is ±1.5‰. While the pooled standard deviation of the USGS34 standard analyzed with all collected NOx samples is 0.4‰ (n = 28), significantly lower than the 1.5‰ across the entire method, the 1.5‰ is somewhat smaller than the error propagated across the blank correction equation, above. At typical ambient collection values, δ 15 N total of −15‰, [NO3−]total of 15 μM, δ15Nblank of 2‰, and [NO3−]blank of 5 μM, the propagated error is 1.7‰. At δ15N values further from 0‰, such as the tank values, the error increases (10‰ at δ15N of −60‰.) Since our actual measurement variation is lower, it is implied that some of the error is systematic, rather than random. This is, perhaps, not surprising since the individual component errors are based on a standard deviation across analytical sets, while the blank and samples are typically analyzed within the same set. Comparison of the error value with prior NOx collections is not possible as no other previously documented methods have quantified an error across the entire analytical method. Blank Reduction. The largest obstacle to using the 0.25 M KMnO4 and 0.5 M NaOH method successfully is a background of NO3− in the KMnO4. Many means of eliminating this background were tested, and while none were successful in completely eliminating the NO3−, we were able to keep the background concentration consistent around 5 μM. Furthermore, despite varying the relative concentration of the blank to that of the sample, consistent isotopic results were achieved (Tables 2−4). 12120

dx.doi.org/10.1021/ac502968e | Anal. Chem. 2014, 86, 12115−12121

Analytical Chemistry

■

Article

(16) Margeson, J. H.; Knoll, J. E.; Midgett, M. R.; Oldaker, G. B.; Loder, K. R.; Grohse, P. M.; Gutknecht, W. F. Anal. Chem. 1984, 56, 2607−2610. (17) Elliott, E. M.; Kendall, C.; Wankel, S. D.; Burns, D. A.; Boyer, E. W.; Harlin, K.; Bain, D. J.; Butler, T. J. Environ. Sci. Technol. 2007, 41, 7661−7667. (18) Sigman, D. M.; Casciotti, K. L.; Andreani, M.; Barford, C.; Galanter, M.; Bohlke, J. K. Anal. Chem. 2001, 73, 4145−4153. (19) Kaiser, J.; Hastings, M. G.; Houlton, B. Z.; Roeckmann, T.; Sigman, D. M. Anal. Chem. 2007, 79, 599−607. (20) Böhlke, J. K.; Mroczkowski, S. J.; Coplen, T. B. Rapid Commun. Mass Spectrom. 2003, 17, 1835−1846. (21) Margeson, J. H.; Knoll, J. E.; Midgett, M. R.; Oldaker, G. B.; Reynolds, W. E. Anal. Chem. 1985, 57, 1586−1590.

CONCLUSIONS Using a 0.25 M KMnO4 and 0.5 M NaOH solution, it is possible to collect NOx quantitatively. This allows for isotopic analysis of the NOx without correction for potentially fractionating processes. This procedure can be used to measure δ15N-NOx from various sources with an error of ±1.5‰. The method proved robust under a wide variety of flow rate, temperature, and humidity conditions, both in the laboratory and in the field. Consistent, negative δ15N was found for NOx generated from a diesel engine in a laboratory setting. Additionally, consistent δ15N was measured with replicate collections of NOx in an urban environment.

■

AUTHOR INFORMATION

Corresponding Authors

*E-mail: dorothy_fi[email protected]. *E-mail: [email protected]. Present Address ¶

D.L.F.: National Oceanic and Atmospheric Administration, Chemical Sciences Division, Earth System Research Laboratory, 325 Broadway Boulder, CO, 80302.

Author Contributions

The manuscript was written through contributions of all authors. All authors have given approval to the final version of the manuscript. Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS This research was partially supported by the National Science Foundation (Award No. AGS-1351932 to M.G.H.), the American Association of University Women (D.L.F.), and Brown University Research and Teaching Awards (A.F.L.). We thank Cate Levey, Connor Hilton, and Ruby Ho for laboratory assistance.

■

REFERENCES

(1) Freyer, H. D. Tellus, Ser. B 1991, 43, 30−44. (2) Morin, S.; Savarino, J.; Frey, M. M.; Domine, F.; Jacobi, H. W.; Kaleschke, L.; Martins, J. M. F. J. Geophys. Res.: Atmos. 2009, 114, D05303. (3) Geng, L.; Alexander, B.; Cole-Dai, J.; Steig, E. J.; Savarino, J.; Sofen, E. D.; Schauer, A. J. Proc. Natl. Acad. Sci. U. S. A. 2014, 111, 5808−5812. (4) Hoering, T. Geochim. Cosmochim. Acta 1957, 12, 97−102. (5) Li, D. J.; Wang, X. M. Atmos. Environ. 2008, 42, 4747−4754. (6) Moore, H. Atmos. Environ. 1977, 11, 1239−1243. (7) Jaegle, L.; Steinberger, L.; Martin, R. V.; Chance, K. Faraday Discuss. 2005, 130, 407−423. (8) Heaton, T. H. E. Tellus 1990, 42, 304−307. (9) Ammann, M.; Siegwolf, R.; Pichlmayer, F.; Suter, M.; Saurer, M.; Brunold, C. Oecologia 1999, 118, 124−131. (10) Felix, J. D.; Elliott, E. M.; Shaw, S. L. Environ. Sci. Technol. 2012, 46, 3528−3535. (11) Snape, C. E.; Sun, C. G.; Fallick, A. E.; Irons, R.; Haskell, J. Prepr. Symp. - Am. Chem. Soc., Div. Fuel Chem. 2003, 48, 3−5. (12) Felix, J. D.; Elliott, E. M. Atmos. Environ. 2014, 92, 359−366. (13) Felix, J. D.; Elliott, E. M. Geophys. Res. Lett. 2013, 40, 1642− 1646. (14) Hagenbjork-Gustafsson, A.; Tornevi, A.; Forsberg, B.; Eriksson, K. J. Environ. Monit. 2010, 12, 1315−1324. (15) Freyer, H. D.; Kley, D.; Volzthomas, A.; Kobel, K. J. Geophys. Res.: Atmos. 1993, 98, 14791−14796. 12121

dx.doi.org/10.1021/ac502968e | Anal. Chem. 2014, 86, 12115−12121