Selective Collection of Particulate Ammonium for Nitrogen Isotopic

May 22, 2019 - Standard lab protocols were followed that included calibration to anion standards and blank and quality control standard measurements t...
0 downloads 0 Views 1MB Size
Download PDF
Article Cite This: Anal. Chem. 2019, 91, 7586−7594

pubs.acs.org/ac

Selective Collection of Particulate Ammonium for Nitrogen Isotopic Characterization Using a Denuder−Filter Pack Sampling Device Wendell W. Walters,*,†,§ Danielle E. Blum,‡,§ and Meredith G. Hastings†,§ Department of Earth, Environmental, and Planetary Sciences, and ‡Department of Chemistry, Brown University, 324 Brook Street, Providence, Rhode Island 02912, United States § Institute at Brown for Environment and Society, Brown University, 85 Waterman Street, Providence, Rhode Island 02912, United States Downloaded via BUFFALO STATE on July 24, 2019 at 07:07:51 (UTC). See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.

†

S Supporting Information *

ABSTRACT: Nitrogen stable isotope analysis (δ15N) of particulate ammonium (NH4+) may provide additional constraints on this critical component of fine particulate matter; however, no previous collection method has been verified for its ability to accurately and precisely characterize δ15N(NH4+). This is a critical point due to the difficulty of quantitative NH4+ collection and possible sampling artifacts. Here, we report on δ15N(NH4+) precision using an established denuder−filter pack combination with two filter configurations including (1) a nylon filter plus an acid-impregnated cellulose filter and (2) an acidimpregnated glass fiber filter for NH4+ collection in both laboratorycontrolled environments and ambient air samples. Laboratory NH4+ were generated from the nebulization of ammonium salt solutions and collected using a filter pack sampling train for off-line concentration and isotopic measurement. Quantitative collection of NH4+ was achieved using both filter configurations in both laboratory and field collections. Laboratory experiments indicate a δ15N(NH4+) precision of ±0.9‰ (1σ; n = 24) and ±0.6‰ (n = 9) for the nylon plus citric acid impregnated cellulose filter and for the citric acid impregnated glass fiber filter, respectively. Field sample reproducibility was assessed from 24 h collected side-by-side samples and indicated δ15N(NH4+) to be reproducible within 1.1‰, consistent with the laboratory findings. This work represents the first established method for speciated NH4+ collection for isotopic analysis with important implications for furthering our understanding of its atmospheric dynamics. norganic ions represent a major component of fine particulate matter (PM2.5) that can contribute up to 70% of their total masses.1 However, the inorganic contribution to PM2.5 is often difficult to predict, because it is formed via complex chemistry and interplay among gaseous precursors from a variety of sources. Important precursor emissions include sulfur dioxide (SO2), nitrogen oxides (NOx = NO + NO2), and ammonia (NH3) that can react to form particulate sulfate (SO42−), nitrate (NO3−), and ammonium (NH4+) and be transported far away from their emission sources.2 Previous regulation efforts (i.e., Clean Air Acts) have reduced anthropogenic emissions of SO2 and NOx from fossil-fuel combustion sources, but NH3 is yet to be regulated in many regions around the world from both fossil-fuel and agricultural emission sources.3 Today, NH3 and its secondary product NH4+ dominate total nitrogen that is deposited in the United States via wet and dry deposition,4 which has important implications for human health and the environment. Therefore, a fundamental understanding of NHx (NH3 + NH4+) sources and NH3−NH4+ gas-to-particle conversion mechanisms and its influences on new particle formation and coagulation is critical.5

I

© 2019 American Chemical Society

Nitrogen stable isotope analysis (δ15N) of NHx may provide additional insights into emission sources and transformation mechanisms, as they often exhibit characteristic isotopic compositions or (“fingerprints”).6 This tool may allow for distinction of emission sources as well as an understanding of the chemical and physical processes responsible for NHx removal at a process level. The partitioning of NHx between the gas, solid, and aqueous phases is driven by thermodynamic equilibrium that depends on the relative humidity, temperature, and particle chemical composition. This equilibrium is typically achieved on the order of minutes,7 suggesting that isotopic equilibrium between NH3(g) and the solid or aqueous phases of NH4+ is also likely rapidly achieved in the atmosphere (R1−R3):8,9 15

NH3(g) + 14 NH4 +(s) F 14 NH3(g) + 15 NH4 +(s)

15

NH3(g) + 14 NH4 +(aq) F 14 NH3(g) + 15 NH4 +(aq)

(R1)

(R2) Received: January 9, 2019 Accepted: May 22, 2019 Published: May 22, 2019 7586

DOI: 10.1021/acs.analchem.9b00151 Anal. Chem. 2019, 91, 7586−7594

Article

Analytical Chemistry 15

NH3(g) + 14 NH3(aq) F 14 NH3(g) + 15 NH3(aq)

tested for their suitability for isotopic characterization including identification of sampling artifacts, biases, and collection limits before their field application.19 We report on a laboratory and field-tested speciated NH4+ collection device that is suitable for ambient air collection and isotopic analysis. The device is a ChemComb speciation cartridge,20,21 which contains a PM2.5 impactor, two honeycomb denuders for the removal of acidic and basic gases, followed by a four-stage filter pack for the collection of PM2.5. Here, we have tested this method’s suitability for δ15N(NH4+) characterization via NH4+ collection under a variety of laboratory and field conditions and subsequent extraction and off-line analysis for δ15N(NH4+) using established laboratory techniques. This work is an important step in establishing robust sampling techniques for accurate and precise δ15N(NHx) characterization and will be useful in furthering our understanding of NHx dynamics in the atmosphere.

(R3)

Recent ab initio calculations predict nitrogen equilibrium constants (K) or isotopic fractionation factors (α) of 1.031 (±0.004), 1.034 (±0.004), and 1.004 (±0.003) at 25 °C for R1, R2, and R3, respectively.10 This suggests that the 15N isotope will preferentially partition into the condensed phase of NHx with an isotopic enrichment factor [ε(‰) = 1000(α − 1)] ranging from 4‰ to 34‰ relative to NH3(g) depending on the conversion mechanism, potentially scrambling original source δ15N(NH3) signatures as NH3 is converted to NH4+. Additionally, previous works have suggested that unidirectional NH3(g) neutralization reactions involving sulfuric acid (H2SO4(g)) may induce a kinetic isotope effect of approximately −28‰ based on diffusion rates of 15NH3 relative to 14 NH3.11 These isotope effects (e.g., equilibrium and kinetic) offer the possibility to use nitrogen isotope separation factors between aerosol NH4+ and NH3(g) [(Δ15δNNH4+/NH3 = δ15N(NH4+) − δ15N(NH3)] to track gas-to-particle phase conversion mechanisms during nucleation and condensation events.10 Thus, to fully utilize this isotopic tool requires a method that enables the speciated collection of NHx for separate, accurate, and precise δ15N analysis of NH3(g) and NH4+. Most previous δ15N studies involving NHx have collected a single phase of NHx for subsequent isotopic analysis. This includes collection of NH4+ on a particulate filter11 and/or collection of NH3(g) using a passive acid-coated filter.12,13 These approaches make it difficult to use δ15N as a way to evaluate an NH3 emission source without knowing the partitioning of NHx phases and its associated isotope fractionation.10 Studies that have attempted speciated NHx for δ15N analysis have been rare, and even then, questions remain as to whether quantitative NHx speciation was achieved. These studies have commonly speciated NHx utilizing a filter pack combination involving a particle filter for NH4+ collection followed by an acid-coated cellulose filter for NH3(g) collection.14,15 Problems with this approach, however, arise due to the potential for positive and negative sampling artifacts.16 Positive artifacts may result from the uptake of gases by collected particles or by the filter medium,17 while negative artifacts may result from volatilization of collected particles, which is particularly relevant for semivolatile NH4+.16 These artifacts also likely alter the δ15N values of the phase components and their interpretations. The denuder−filter combination is a well-established sampling technique that can effectively collect and speciate NHx.16,18 In this system, NH3(g) is first removed from the sampled air stream on an acid-coated glass denuder followed by collection of NH4+ on a filter pack. Utilizing such a system, we have recently reported on a laboratory- and field-tested collection method for δ15N(NH3) analysis utilizing an acidcoated honeycomb denuder;19 however, to date, there have been no NH4+ collection methods that have laboratory-verified their suitability for speciated δ15N(NH4+) analysis. This is a critical point, because to quantitatively utilize this tool requires that the isotopic measurements accurately and precisely reflect the environment in which NH4+ was collected. Atmospheric NH4+ is notoriously difficult to accurately characterize for concentration because of its volatile nature,16 which will lead to an undesirable δ15N(NH4+) fractionation effect and influence interpretation of the atmospheric dynamics involving NHx. Thus, it is critical that sampling devices are rigorously

■

EXPERIMENTAL SECTION NH4+ Collection. Collection of NH4+ was conducted using a series of filters in two different configurations. The first configuration included a nylon filter (Cole-Parmer, 0.8 μm pore, 47 mm diameter) followed by two cellulose filters (Whatman, 8 μm pore, 47 mm diameter) impregnated with 5% (w/v) citric acid (Fischer Chemical >99.5%) in ultrahighpurity water [18.2 MΩ, Milli-Q (MQ)]. This approach enables a cost-effective way for PM2.5 monitoring [i.e., see large networks such as the Interagency Monitoring of Protected Visual Environments (IMPROVE)]22,23 due to its quantitative capture of semivolatile NO3−, but NH4+ volatilization can be significant ranging between 1% and 65% based on previous field observations.16 In the first configuration, the nylon filter is used to collect PM2.5, the downstream 5% (w/v) citric acid coated cellulose filter is used to collect any volatilized NH4+, and the final 5% (w/v) citric acid filter is used as a control to check for breakthrough, potentially allowing for simultaneous monitoring of reactive species and speciated δ15N(NHx) measurements. The second filter configuration included two glass fiber filters (Whatman, 1.5 μm pore, 47 mm diameter) impregnated with 5% (w/v) citric acid. This configuration may be suitable for NHx monitoring and speciated δ15N(NHx) measurement but cannot provide the simultaneous reactive species monitoring that incorporation of the nylon filter allows for due to large potential for NO3− volatilization off the acidcoated glass fiber filter. In this setup, the first acid-coated glass fiber filter is used for the collection of NH4+, and the second filter is used to check for breakthrough. The series of filters were housed in a polytetrafluoroethylene (PTFE) four-stage filter pack contained within the ChemComb speciation cartridge (CCSC). Use of a CCSC for inorganic gas and particle speciation has been extensively described and characterized in previous works.20,21 Briefly, gases are removed on coated denuders and particles are removed using a filter pack. Due to its much smaller diffusion rate compared to gases, PM2.5 cannot migrate to the denuder walls during their residence time, allowing for speciation between gaseous and particulate components based on physical separation.17 Nylon filters were rinsed and sonicated with MQ water for 30 min three times and dried using ultrapure nitrogen gas in a vacuum desiccator prior to use. The cellulose and glass fiber filters were not washed prior to use because of disintegration issues and were directly impregnated with the 5% citric acid 7587

DOI: 10.1021/acs.analchem.9b00151 Anal. Chem. 2019, 91, 7586−7594

Article

Analytical Chemistry

tested ranging from 1 to 24 h. For the second filter configuration (acid-impregnated glass fiber filter) triplicate trials were conducted at 5, 10, and 15 min using only the nebulized NH4NO3 solution (Figure S1). Field Applicability. Utilizing the CCSC, NH4+ samples from ambient air were collected over 24 h periods in Providence, RI, U.S.A. (41.8240° N, 71.4128° W). A weather-proof enclosure for the CCSC with a built-in weather station (ProWeather Station TP3000WC) was installed on the rooftop of a building on the campus of Brown University, and air was sampled at a flow rate of 10 SLPM using a mass-flow controlled (Dakota mass flow controller 6AGC1AL55-10AB2; precision ±1‰) vacuum pump (Welch 2546B-01). Ambient NH3(g) was first denuded using two 2% citric acid coated honeycomb denuders, with the first denuder used for NH3 capture and the second used to check for breakthrough. Ambient NH4+ was collected downstream of the denuders using the two laboratory-tested filter configurations including (1) a series of nylon and two 5% (w/v) citric acid coated filters and (2) a series of two 5% (w/v) citric acid impregnated glass fiber filters. Replicate (side-by-side) samples were collected throughout the year to investigate NH4+ concentration and δ15N reproducibility under a variety of field conditions (i.e., temperature and relative humidity). The filters were extracted and processed as previously described. Concentration and δ15N Isotopic Analysis. The NH4+ concentrations ([NH4+]) in all extraction solutions were analyzed using colorimetric analysis based on the indophenol blue method27 that was automated using a discrete UV−vis analyzer (Westco SmartChem 2.0). In addition, concentration measurements were made for [NO2−], [NO3−], and [SO42−] in the nylon and glass fiber filter extraction solutions from the laboratory nebulization experiments. [NO2−] and [NO3−] were measured using a standard colorimetric technique (e.g., EPA Method 353.2) involving the cadmium reduction of NO3− to NO2− that undergoes diazotization with sulfanilamide dihydrochloride followed by detection of absorbance at 520 nm that was automated using a discrete UV−vis analyzer (Westco SmartChem 2.0). [SO42−] was measured using ion chromatography (Dionex IonPac AS22 with a 3.5 mM carbonate and 1.0 mM bicarbonate elution solution). Standard lab protocols were followed that included calibration to anion standards and blank and quality control standard measurements that were made every 5−7 samples. Reproducibility calculated from replicate measurements of quality control standards were ±2%, ±1%, ±1%, and ±1% for [NH4+], [NO2−] [NO3−], and [SO42−], respectively. Limits of detection were approximately 0.3, 0.2 0.5, and 2.0 μM for [NH4+], [NO2−], [NO3−], and [SO42−], respectively. The determination of δ15N(NH4+) is based on a coupled offline wet-chemistry technique involving hypobromite (BrO−) oxidation and acetic acid/sodium azide reduction.28 Samples are diluted to at least 10 μM NH4+ using MQ water to a volume of 10 mL in 50 mL vials. The samples are then oxidized to NO2− using BrO− in an alkaline solution, which was synthesized as previously described.28 After 30 min the reaction was stopped by 0.4 mL addition of 0.4 M sodium arsenite (NaAsO2) to remove remaining BrO−. [NO2−] oxidation yields were measured using a standard colorimetric absorption technique (e.g., U.S. EPA Method 353.2) automated using a discrete UV−vis analyzer (Westco SmartChem 2.0) to confirm quantitative conversion of NH4+ to NO2−. After conversion, analysis of δ15N was conducted

solution. A range of acid coating solutions may be used to impregnate filters for NHx capture including phosphoric acid.18 We have limited our tests to citric acid due to its low NH4+ blank compared to other acids,24 but we expect other commonly used acid solutions for NHx capture to be viable as well, provided that they are checked for “blank” NH4+ artifacts. Immediately after NH4+ collection, filters were extracted in 20 mL of MQ water, sonicated for 60 min, and filtered through 0.22 μm syringe filters to remove disintegrated filter pieces and possible microbial contaminants (particularly for the ambient air collected samples). Following this, the filters were removed, and the extraction solutions were frozen until subsequent analysis. Freezing solutions has been shown to be an adequate technique to preserve NH4+.25 Previous stability tests have indicated minimal loss of NH4+ on acidcoated filters when stored in a refrigerator for up to 2 weeks with loss of 7−9% observed for up to 5 weeks.26 Therefore, extraction of NH4+ or proper preservation initiatives (e.g., freezing samples) should be conducted soon after collection to minimize any potential for losses. Laboratory Collection Experiments. A three-jet collision nebulizer (CH Technologies) was used to generate NH4+ particles in the laboratory using two types of salt solutions that included 25 mM ammonium nitrate (NH4NO3; Acros >99%) and 12.5 mM ammonium sulfate ((NH4)2SO4; Fischer Chemical >99.7%), representing the most common types of NH4+ aerosols as well as a range of volatility. Briefly, an ultrapure nitrogen tank (PurityPlus gas) flow-controlled at 4.0 standard liters per minute (SLPM) was connected to the nebulizer. The generated NH4+ was diluted with ultrapure nitrogen that was flow-controlled at 6.0 SLPM. NHx was not detectable in the ultrapure nitrogen. The flow of the NH4+ line and the nitrogen dilution line were combined using a threeway union that was also connected to the inlet of the CCSC. Temperature and relative humidity were monitored (Elitech GS6) in the outflow of the CCSC and were typically 21.0 ± 1.0 °C and 45 ± 10%, respectively. All tubing and unions used to connect the nebulized NH4+ to the CCSC were composed of PTFE. A schematic of the experimental setup is provided in the Supporting Information (Figure S1). All experiments were conducted at the CCSC’s optimal flow rate of 10 SLPM.20 In these experiments, upstream denuders were not utilized in the sample collections. For the first filter configuration (nylon and citric acid impregnated cellulose filters), triplicate trials were conducted at 10, 20, and 30 min for both the NH4NO3 and (NH4)2SO4 solutions to test the loading capacity of the sampling system. Additionally, NH4+ desorption tests were conducted by collecting NH4+ for 10 min, and then flowing ultrapure nitrogen through the CCSC at a rate of 10 SLPM for 24 h. Due to the potential for NH4+ volatilization off the nylon filter, the limit of the collection system is likely determined by the 5% citric acid coated filter. Thus, the NH3(g) (e.g., volatilized product of NH4+) collection limit of the acid-impregnated filter was also tested in a similar manner as depicted in Figure S1 and exactly as previously described.19 For this test, a 20.1 ppmv tank of NH3 (Praxair) diluted in N2 was flowed at a rate of 0.015 LPM and further diluted with a humidified ultrapure N2 to a final flow rate of 10 SLPM that was connected to the inlet of the CCSC. Temperature and relative humidity for these experiments were typically 21.0 ± 1.0 °C and 40 ± 10%, respectively. The filter pack for these experiments only contained two acidimpregnated cellulose filters, and variable flow times were 7588

DOI: 10.1021/acs.analchem.9b00151 Anal. Chem. 2019, 91, 7586−7594

Article

Analytical Chemistry

Figure 1. Results of NH4+ collection on a nylon and backup citric acid impregnated cellulose filters from laboratory nebulization of NH4NO3 (left) and (NH4)2SO4 (right) solutions including the NH4+ collection fraction for the filter train consisting ofin ordernylon filter (orange), 5% citric acid impregnated cellulose filter 1 (green), and 5% citric acid impregnated cellulose filter 2 (purple; note that minimal NH4+ was collected on this filter (99.2% (black triangles and primary y-axis) with a consistent measured δ15N(NH4+) of −2.8 ± 0.6‰ (red triangles and secondary y-axis) (A). The red dash line in panel A corresponds to the δ15N(NH4+) of the nebulized NH4NO3 solution of −0.5 ± 0.1‰, which was slightly higher than the nebulized and collected NH4+. The relationship between the collected NH4+ and NO3− has a slope of 1.14 (±0.04) (R2 = 0.96), indicating potential loss of NO3− on the acid-impregnated glass fiber filter (B).

impregnated cellulose filter was found to be 99 ± 1% (n = 12) for all trials spanning from 0.5 to 24 h for a collection amount that ranged from 12.8 to 336.0 μg of NH3 (Figure 3). This indicates that one citric acid impregnated cellulose filter

Figure 3. Collection efficiency (CE; primary axis, gray points) and δ15N (secondary axis, red points) as a function of collected amount of NH3(g) using a 5% acid-coated cellulose filter. Quantitative collection of NH3(g) was achieved for a collection amount up to ∼350 μg, with a consistent δ15N value with a 1σ = ±1.0‰. One potential δ15N(NH3) outlier was identified (indicated by the star), and removal of this point increased the precision to 1σ = ±0.5‰. 7591

DOI: 10.1021/acs.analchem.9b00151 Anal. Chem. 2019, 91, 7586−7594

Article

Analytical Chemistry

Table 1. Summary of 24 h Ambient Air Side-by-Side Replicate Collections (A and B) Conducted in Providence, RI, U.S.A., Including Average Temperature (Temp, °C), Average Relative Humidity (RH%), [NH4+], f(NH3), δ15N(NH3), δ15N(NH4+), Δδ15N, and δ15N(NHx)a collection configuration

temp (°C)

RH (%)

[NH3(g)] μmol/m3

1A 1B

20.7

86.3

2A 2B

23.4

87.1

3A 3B

11.0

52.4

0.067 0.061 (0.005) 0.081 0.071 (0.010) 0.047 0.046 (0.001)

4A 4B

8.8

5A 5B

12.6

93.7

CD−N-CC 6A CD−CGFF 6B

11.0

52.4

CD−N-CC 7A CD−CGFF 7B

12.0

73.4

80

0.046 0.041 (0.005) 0.042 0.045 (0.003) 0.046 0.048 (0.002) 0.033 0.031 (0.002)

[NH4+] μmol/m3

f(NH3)b

CD−N-CC 0.017 0.80 0.017 0.79 (0.00) (0.01) 0.020 0.81 0.014 0.83 (0.006) (0.02) 0.017 0.73 0.016 0.74 (0.001) (0.01) CD−CGFF 0.025 0.64 0.023 0.64 (0.002) (0.00) 0.045 0.47 0.051 0.47 (0.006) (0.00) Comparison 0.016 0.74 0.014 0.77 (0.002) (0.03) 0.008 0.79 0.007 0.81 (0.001) (0.02)

δ15N(NH3) (‰)

δ15N(NH4+) (‰)

Δδ15N (‰)

δ15N(NHx) (‰)c

−14.4

19.6

−10.5

19.4

−7.0

−17.1 −17.3 (0.2)

5.2 6.3 (0.9) 8.6 7.9 (0.6) −0.1 −1.1 (1.0)

17.0 16.1 (0.9)

−12.5 −13.1 (0.6)

−24.5 −22.6 (1.8) −16.9 −17.6 (0.7)

6.9 7.7 (0.8) 10.0 10.9 (0.9)

31.4 30.4 (1.0) 26.9 28.5 (1.6)

−13.1 −11.7 (1.4) −2.7 −2.5 (0.2)

−17.1 −18.2 (1.1) −13.9 −12.8 (1.1)

−0.1 1.0 (1.1) −1.6 −2.1 (0.5)

17.0 19.2 (2.2) 12.3 10.7 (1.6)

−12.7 −13.8 (1.1) −11.3 −10.7 (0.6)

−10.8

a Two denuder filter pack configurations were tested including (1) citric acid denuder and nylon plus citric acid cellulose filters (CD−N-CC) and (2) citric acid denuder and citric acid impregnated glass fiber filter (CD−CGFF). The absolute difference between the two replicate samples is shown by the values in parentheses. bCalculated according to eq 7. cCalculated according to eq 6.

within (calculated as the absolute difference between the two collected side-by-side samples) 0.010 μmol/m3 and 1.8‰, respectively, consistent with our previous results.19 The percent error in the concentration of NH3(g) was found to be within 13.1%, which was calculated according to eq 5:

quantitatively captured NH 3(g) without any significant NH3(g) loss up to a period of 24 h in the controlled laboratory experiments, and the filter has an operative capacity (defined as the amount of target analyte that can be collected while maintaining a collection efficiency >95%)36 of at least 336.0 μg of NH3 at a flow rate at 10 SLPM. The measured δ15N(NH3) had an average value of −2.1 ± 1.0‰ (n = 12) (Figure 3). A significant δ15N(NH3) outlier was identified based on a Grubb’s test (p < 0.05). While the cause of this outlier is unclear, removal of this data point results in an average of −2.4 ± 0.5‰ (n = 11). This value is consistent with that previously measured using an acid-coated honeycomb denuder of −2.5 ± 0.8‰ from the same NH3(g) source (Praxair NH3 tank).19 Overall, quantitative NH3(g) collection as well as consistent δ15N(NH3) values suggests the robustness of NH3(g) collection on an acid-coated filter for a period up to 24 h, and this approach should be suitable for collection NH4+ volatilized off a nylon filter. Ambient Sampling. Ambient replicate (side-by-side) collections demonstrate the high precision of the tested sampling methods for a sampling period of at least 24 h (Table 1). Overall, >99% of NH3(g) was collected and retained on a single citric acid coated honeycomb denuder, as the collected [NH4+] amount on the breakthrough denuder was always within our detection limits. This is consistent with our previous findings as the collected NH3(g) never exceeded 1.2 μmol, which is well within the reported operative capacity of the citric acid honeycomb denuder of ∼23.5 μmol.19 NH3(g) concentration and δ15N(NH3) was found to be reproducible

% error =

|A − B | × 100 average(A , B)

(5)

where A and B refer to the two side-by-side collected samples. We note that, for two of the ambient replicate trials (1B and 2B; Table 1), δ15N(NH3) was not measured for both collected samples. Ambient NH4+ was collected in a downstream filter pack using both tested filter configurations. The first configuration utilized a nylon filter plus 5% citric acid coated cellulose filter. In this configuration, a large fraction of NH4+ volatilization off the nylon filter was observed that averaged 75 ± 13% (n = 8) (Table S6). This NH4+ volatilization is much higher than observed in our laboratory tests (Figure 1), which may be related to diurnal changes in temperature and relative humidity over the 24 h ambient collections as well as a much more complex ambient particle composition comprising a mixture of organic and inorganic components. We also note that our air stream was denuded prior to particle collection, which would facilitate NH4+ volatilization compared to our laboratory collections in which no denuders were utilized. Our observed NH4+ volatilization also tends to be higher than previous reported in denuded nylon filter collections with NH4+ loss up to 65%.16 However, we note that our nylon filters had a smaller 7592

DOI: 10.1021/acs.analchem.9b00151 Anal. Chem. 2019, 91, 7586−7594

Article

Analytical Chemistry pore size (0.8 μm) compared to this previous study (1.0 μm), which would have increased the pressure drop across the filter, increasing the potential for NH4+ volatilization. While significant, the volatilized NH4+ was nearly quantitatively captured using a single citric acid impregnated cellulose filter, consistent with our laboratory findings, as ambient NH4+ collection efficiency (eq 3) was found to be >97% (Table S5), as the [NH4+] measured on the breakthrough citric acid cellulose filter was found to range from 1.5 to 2.8 μM, near the blank [NH4+] of 2.4 ± 0.8 μM. The NH4+ concentration and δ15N(NH4+) were reproducible within 0.006 (n = 3) μmol/m3 (with an average percent error of 8.5 ± 8.2%; eq 5) and 0.8 ± 0.2‰ (Table 1), respectively. The isotopic separation factor between NH4+ and NH3 (Δδ15N) was found to range between 12.1‰ and 19.6‰ (Table 1). The δ15N(NHx) was found to range between −12.7‰ and −7.0‰, which was calculated according to the following (eq 6):

reproduce similar NH4+ concentrations and δ15N(NH4+) values and both are suitable for ambient NH4+ collection for off-line concentration and isotopic analysis.

■

CONCLUSION We have critically evaluated a method for NH4+ collection via two separate filter trains that includes (1) a nylon filter plus a 5% citric acid impregnated cellulose filter and (2) a citric acid impregnated glass fiber filter for δ15N(NH4+) analysis. Our results indicate significant NH4+ volatilization off the nylon filter that is quantitatively collected on the backup 5% citric acid impregnated cellulose filter in both laboratory and ambient environment collections, indicating the need to analyze the NH4+ collected on both filters. However, NH4+ was found to be quantitatively collected using a single acidimpregnated glass fiber filter in both laboratory and field tests. The laboratory results indicate a δ15N(NH4+) precision of ±0.9‰ (1σ; n = 24) and ±0.6‰ (n = 9) for the nylon plus acid-impregnated cellulose filters and the acid-impregnated glass fiber filter, respectively. This δ15N(NH4+) precision is confirmed in field replicate collections with reproducibility found to be within 1.1‰ and with [NH4+] typically within 13.3%. We note that we have limited our tests to the use of a citric acid, which we believe should represent a limiting case, as stronger acids such as phosphoric acid should perform as well and if not better for retaining NH4+. These results have important consequences for previously reported δ15N(NHx) values. Most previous studies have speciated NHx for δ15N analysis via collection of NH4+ on a particulate filter followed by collection of NH3(g) on an acidcoated filter. However, our results demonstrate that significant NH4+ volatilization off a particulate filter can occur, which has a large isotopic fractionation that was measured to range from 27.6 ± 2.5‰ to 33.1 ± 0.8‰, consistent with theoretical estimates of NH4+(s) or NH4+(aq) isotopic exchange with NH3(g). Thus, there is potential for a positive δ15N bias in previous δ15N(NH4+) measurements from aerosol filters and a negative δ15N bias in downstream acid filters utilized for speciated NH3(g) collection for δ15N(NH3) measurement. The tested NH4+ collection method (denuder plus filter train) represents the first method that has been demonstrated to properly provide NHx speciation for robust simultaneous measurement of δ15N(NH3), δ15N(NH4+), Δδ15N, and δ15N(NHx) as demonstrated by field replicate collections. Due to the simplicity and cost-effectiveness of collection of NH4+ using a single acid-impregnated glass fiber filter, this should be the recommended approach if NHx monitoring and δ15N(NHx) measurement is the primary goal. While the collection of NH4+ using both a nylon and an acid-impregnated cellulose filter may complicate the system, it may be the best approach for δ15N measurement of speciated NHx along with simultaneous inorganic gaseous/particulate concentration measurements. Utilizing a basic (e.g., sodium carbonate) and acidic (e.g., citric acid) coated denuder prior to collection of PM2.5 will selectively remove the acidic (e.g., HNO3, HCl) and basic (NH3) gases, respectively. The PM2.5 that is not removed via the denuders will be collected on the nylon filter that has been shown to quantitatively capture semivolatile particulate nitrate, and incorporation of a backup acid-coated cellulose filter will quantitatively capture the volatilized NH4+. Quantitative collection of inorganic PM2.5 and inorganic gases (denuded downstream) is desirable, as it will enable inorganic thermodynamic modeling involving the NH4+−

δ15 N(NHx) = fNH δ15 N(NH3) + (1 − fNH )δ15 N(NH4 +) 3

3

(6)

where f NH3 refers to the fraction of NHx that exists in the gaseous phase (NH3) and is calculated according to fNH = 3

[NH3] [NH3] + [NH4 +]

(7)

In only one replicate collection was δ15N(NH3) measured for both side-by-side samples for this filter configuration. Despite this small sample size, both Δδ15N and δ15N(NHx) were found to be reproducible within 0.9‰ and 0.6‰, respectively, which was consistent with the reproducibility measured for the acidimpregnated glass fiber filter configuration (Table 1). Ambient replicate collection of NH4+ was also conducted using the citric acid impregnated glass fiber filter. Overall, near quantitative collection of NH4+ was achieved using a single citric acid impregnated glass fiber filter with a collection efficiency >98% (Table S6). The NH4+ concentration and δ15N(NH4+) were found to be reproducible within 0.006 μmol/m3 (with an average percent error of 10.4 ± 2.1%) and 0.9 ± 0.1‰, respectively. Δδ15N and δ15N(NHx) were found to be reproducible within 1.6‰ and 1.4‰, respectively (Table 1). A direct comparison between the two filter configurations for the collection of ambient NH4+ was also made (Table 1). Overall, the two filter configurations agreed well with one another as average [NH4+], δ15N(NH4+), Δδ15N, and δ15N(NHx) reproducibility was found to be within 0.02 μmol/m3 (with a percent error within 13.3%, 1.1‰, 2.2‰, and 1.1‰, respectively). Interpretation of the measured δ15N(NH3), δ15N(NH4+), Δδ15N, and δ15N(NHx) is beyond the scope of this work. We note that the measured Δδ15N was found to range from 10.7‰ to 31.4‰, with the higher end values near the predicted isotopic equilibrium between NH3(g) and the condensed NH4+ phases of 31 ± 4‰ (R1) and 34 ± 4‰ (R2).10 However, lower Δδ15N values were also observed, which may reflect incomplete NH3 and NH4+ isotopic equilibrium or interactions with NH3 and aqueous interfaces (R3).19 Since the isotopic offset between gaseous and particle phase of NHx is not consistent, this suggests that δ15N(NHx) cannot be predicted based off the measurement of a single phase, indicating that simultaneous speciated NHx collection for isotopic analysis is essential. Overall, our ambient replicate collections suggest that both filter configurations tend to 7593

DOI: 10.1021/acs.analchem.9b00151 Anal. Chem. 2019, 91, 7586−7594

Article

Analytical Chemistry NO3−−SO42− system such as ISORROPIA, with new isotopic constraints (δ15N(NH3) and δ15N(NH4+)) to evaluate this thermodynamic system.

■

(11) Pan, Y.; Tian, S.; Liu, D.; Fang, Y.; Zhu, X.; Zhang, Q.; Zheng, B.; Michalski, G.; Wang, Y. Environ. Sci. Technol. 2016, 50 (15), 8049−8056. (12) Felix, J. D.; Elliott, E. M.; Gish, T. J.; McConnell, L. L.; Shaw, S. L. Rapid Commun. Mass Spectrom. 2013, 27 (20), 2239−2246. (13) Chang, Y.; Liu, X.; Deng, C.; Dore, A. J.; Zhuang, G. Atmos. Chem. Phys. 2016, 16 (18), 11635−11647. (14) Savard, M. M.; Cole, A.; Smirnoff, A.; Vet, R. Atmos. Environ. 2017, 162, 11−22. (15) Smirnoff, A.; Savard, M. M.; Vet, R.; Simard, M.-C. Rapid Commun. Mass Spectrom. 2012, 26 (23), 2791−2804. (16) Yu, X.-Y.; Lee, T.; Ayres, B.; Kreidenweis, S. M.; Malm, W.; Collett, J. L., Jr. Atmos. Environ. 2006, 40 (25), 4797−4807. (17) Ali, Z.; Thomas, C. L. P.; Alder, J. F. Analyst 1989, 114 (7), 759−769. (18) Perrino, C.; Gherardi, M. Atmos. Environ. 1999, 33 (28), 4579− 4587. (19) Walters, W. W.; Hastings, M. G. Anal. Chem. 2018, 90 (13), 8051−8057. (20) Koutrakis, P.; Sioutas, C.; Ferguson, S. T.; Wolfson, J. M.; Mulik, J. D.; Burton, R. M. Environ. Sci. Technol. 1993, 27 (12), 2497−2501. (21) Koutrakis, P.; Wolfson, J. M.; Spengler, J. D. Atmos. Environ. 1988, 22 (1), 157−162. (22) Yu, X.-Y.; Lee, T.; Ayres, B.; Kreidenweis, S. M.; Collett, J. L., Jr.; Malm, W. J. Air Waste Manage. Assoc. 2005, 55 (8), 1100−1110. (23) Chen, X.; Day, D.; Schichtel, B.; Malm, W.; Matzoll, A. K.; Mojica, J.; McDade, C. E.; Hardison, E. D.; Hardison, D. L.; Walters, S.; Van De Water, M.; Collett Jr, J. L. Atmos. Environ. 2014, 91, 118− 126. (24) Sutton, M. A.; Tang, Y. S.; Miners, B.; Fowler, D. Water, Air, Soil Pollut.: Focus 2001, 1 (5), 145−156. (25) Menchyk, N.; Park, D.; Moon, P. H.; Unruh, J. B.; Trenholm, L. E. Crop Sci. 2014, 54 (5), 2325−2327. (26) Rudling, J.; Hallberg, B.-O.; Hultengren, M.; Hultman, A. Scand. J. Work, Environ. Health 1984, 10 (3), 197−202. (27) Bolleter, W. T.; Bushman, C. J.; Tidwell, P. W. Anal. Chem. 1961, 33 (4), 592−594. (28) Zhang, L.; Altabet, M. A.; Wu, T.; Hadas, O. Anal. Chem. 2007, 79 (14), 5297−5303. (29) McIlvin, M. R.; Altabet, M. A. Anal. Chem. 2005, 77 (17), 5589−5595. (30) Sigman, D. M.; Casciotti, K. L.; Andreani, M.; Barford, C.; Galanter, M.; Böhlke, J. K. Anal. Chem. 2001, 73 (17), 4145−4153. (31) Bohlke, J. k.; Gwinn, C. j.; Coplen, T. b. Geostand. Geoanal. Res. 1993, 17 (1), 159−164. (32) Böhlke, J. K.; Coplen, T. B. Reference and Intercomparison Materials for Stable Isotopes of Light Elements. In Proceedings of the IAEA-TECDOC-825 Consultants Meeting Held in Vienna, Vienna, Austria, Dec 1−3, 1993; International Atomic Energy Agency: Vienna, Austria, 1995; Vol. 13, pp 51−67. (33) Böhlke, J. K.; Smith, R. L.; Hannon, J. E. Anal. Chem. 2007, 79 (15), 5888−5895. (34) Ge, X.; Wexler, A. S.; Clegg, S. L. Atmos. Environ. 2011, 45 (3), 524−546. (35) Fibiger, D. L.; Hastings, M. G.; Lew, A. F.; Peltier, R. E. Anal. Chem. 2014, 86 (24), 12115−12121. (36) Febo, A.; De Santis, F.; Perrino, C.; Giusto, M. Atmos. Environ. 1989, 23 (7), 1517−1530.

ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.analchem.9b00151.

■

Laboratory collection schematic, filter NH4+ blank summary, NH4NO3 nebulization and collection on nylon and acid cellulose filter summary, (NH4)2SO4 nebulization and collection of nylon and acid cellulose filter summary, NH4NO3 nebulization and collection on acid glass fiber filter, NH3(g) collection on acid-coated filter summary, and ambient replicate collection summary (PDF)

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. ORCID

Wendell W. Walters: 0000-0001-6346-9840 Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS W.W.W. acknowledges support from an Atmospheric and Geospace Sciences National Science Foundation Postdoctoral Fellow (Grant No. 1624618) during the course of this study. This research was also supported by funding to M.G.H. from the National Science Foundation (AGS 1351932). The authors also acknowledge support from an Institute at Brown for Environment and Society Internal Seed Grant (GR300123). We thank Ruby Ho and Joseph Orchardo for laboratory assistance.

■

REFERENCES

(1) Wei, L.; Duan, J.; Tan, J.; Ma, Y.; He, K.; Wang, S.; Huang, X.; Zhang, Y. Sci. China: Earth Sci. 2015, 58 (3), 345−355. (2) Pye, H. O. T.; Liao, H.; Wu, S.; Mickley, L. J.; Jacob, D. J.; Henze, D. K.; Seinfeld, J. H. J. Geophys. Res. Atmospheres 2009, 114 (D1), D01205. (3) Pan, Y.; Tian, S.; Zhao, Y.; Zhang, L.; Zhu, X.; Gao, J.; Huang, W.; Zhou, Y.; Song, Y.; Zhang, Q.; Wang, Y. Environ. Sci. Technol. 2018, 52 (7), 3926−3934. (4) Du, E. Proc. Natl. Acad. Sci. U. S. A. 2016, 113 (26), E3594− E3595. (5) Koziel, J. A.; Aneja, V. P.; Baek, B.-H. Gas-to-Particle Conversion Process between Ammonia, Acid Gases, and Fine Particles in the Atmosphere. In Animal Agriculture and the Environment: National Center for Manure and Animal Waste Management White Papers; Rice, J. M., Caldwell, D. F., Humenik, F. J., Eds.; American Society of Agricultural and Biological Engineers: St. Joseph, MI, 2006;. (6) Brenninkmeijer, C. A. M.; Janssen, C.; Kaiser, J.; Röckmann, T.; Rhee, T. S.; Assonov, S. S. Chem. Rev. 2003, 103 (12), 5125−5162. (7) Meng, Z.; Seinfeld, J. H. Atmos. Environ. 1996, 30 (16), 2889− 2900. (8) Urey, H. C. J. Chem. Soc. 1947, 7, 562−581. (9) Kirshenbaum, I.; Smith, J. S.; Crowell, T.; Graff, J.; McKee, R. J. Chem. Phys. 1947, 15 (7), 440−446. (10) Walters, W. W.; Chai, J.; Hastings, M. G. ACS Earth Space Chem. 2019, 3 (1), 79−89. 7594

DOI: 10.1021/acs.analchem.9b00151 Anal. Chem. 2019, 91, 7586−7594