Seasonal trends of the stable nitrogen isotope ratio in particulate nitrogen compounds and their gaseous precursors in Akita, Japan

Abstract

Particulate matter (PM) can have adverse effects on human health. Moreover, because the mechanisms of PM formation and behavior in the atmosphere are notably complicated, to reduce PM concentrations effectively and meet environmental standards, source–receptor relationships must be clearly understood. Stable isotope ratios can be used to detect chemical processes and distinguish sources. In environmental science, especially in research on aerosols, stable isotope ratios have proven to constitute a powerful tool for source identification. However, there are few long-term studies of isotope fractionation during secondary aerosol formation. In this study, stable nitrogen isotope ratios (δ¹⁵N) of ammonia gas (NH₃), nitrogen dioxide gas (NO₂), nitric acid vapor (HNO₃), particulate nitrate (NO₃⁻), and ammonium (NH₄⁺) in suspended PM (SPM) were analyzed to investigate seasonal trends and isotope fractionation during aerosol formation for long term sampling in Akita, Japan. The results indicated that δ¹⁵N-NH₄⁺ in SPM and δ¹⁵N-NH₃ gas ranged from 1.3‰ to 38.5‰ (mean 16.1‰) and from −33.6‰ to −0.0‰ (−16.9‰), respectively. Furthermore, δ¹⁵N-NO₃⁻ (SPM) and δ¹⁵N-NO₂ and δ¹⁵N-HNO₃ (gaseous) ranged from −4.6‰ to 4.8‰ (mean −0.5‰), from −8.2‰ to −3.1‰ (−5.4‰), and from −7.5‰ to 2.7‰ (−5.0‰), respectively. The mean annual isotope fractionation factors for transformations from gaseous NH₃ to NH₄⁺ in SPM, from gaseous NO₂ to gaseous HNO₃, and from HNO₃ gas to NO₃^– in SPM in the atmospheric environment were +33.3‰, +0.5‰, and +4.9‰, respectively. Isotope fractionation of NH₄⁺ in SPM was much higher than that of NO₃^– in SPM. As the chemical reaction from gaseous precursors progressed, δ¹⁵N-NO₃^– in SPM became steadily heavier.

Keywords:

• stable nitrogen isotope ratio
• NH3
• NO2
• HNO3

1. Introduction

Particulate matter (PM) has adverse effects on human health (e.g., Pope et al., 1991; Brunekreef and Holgate, 2002; Hoek et al., 2002). In Japan, suspended PM (SPM) is defined as particles with a diameter of <10 μm with a cut-off of 100%. To reduce the SPM concentration in the environment, environmental air standards and effluent control have been in place since the 1970s (Japan Ministry of the Environment, 2018). Although the SPM concentration in Japan was dramatically reduced during the 1980s, it has not decreased further since the 2000s but has remained at ∼20 μg m⁻³ on average (Environment, M. o. t, 2017). Spring SPM values in Japan are generally affected dominantly by dust storms from continental Asia (e.g., Aggarwal and Kawamura, 2009), whereas winter values are more likely to be affected by sulfate, nitrite, and lead transported by continental winds (Mukai et al., 2001; Akata and Yanagisawa, 2002; Matsumoto et al., 2003). The water-soluble ion composition of aerosol particles differs among sources (Shen et al., 2009), and changes from gaseous precursors to particulate phases in SPM are promoted at low temperatures. However, owing to the complex mechanisms of SPM formation and behavior in the atmosphere and the many sources of SPM, it is difficult to define the relationship between source and receptor. Thus, to further reduce SPM concentrations, it is necessary to identify the emission sources of particulates and precursor gases and their behavior in the atmosphere.

Recent technical advances have made it possible to analyze stable isotope ratios in a small sample volume with high accuracy and precision by using techniques such as mass spectrometry together with elemental analysis and gas chromatography (e.g., Sharp (2017)). In environmental science, stable isotope ratios constitute a powerful tool for source identification. In particular, stable isotopes of light elements (e.g., carbon, nitrogen, sulfur, and oxygen) have proven useful for identifying sources and inferring atmospheric chemical reactions involving aerosols.

Following early studies of nitrogen isotopic ratios of environmental nitrogen compounds (Hoering, 1957; Moore, 1974; Moore, 1977; Freyer, 1978), many studies have examined stable nitrogen isotope ratios (δ¹⁵N) of gaseous ammonia (NH₃), gaseous nitrogen dioxide (NO₂), nitric acid vapor (HNO₃), and nitrate (NO₃⁻) and ammonium (NH₄⁺) in aerosol particles (Heaton, 1986, 1987, 1990; Freyer, 1991; Freyer et al., 1993; Russell et al., 1998; Yeatman et al., 2001a,b; Hayasaka et al., 2004; Li and Wang, 2008; Kundu et al., 2010; Pavuluri et al., 2010; Kawashima and Kurahashi, 2011). Researchers working in oceanography developed a high-precision method using a denitrifying bacterium to analyze stable nitrogen and oxygen isotopic ratios in nitric acid (Sigman et al., 2001; Casciotti et al., 2002; McIlvin and Casciotti, 2011) and a similar method to analyze stable nitrogen isotopic ratios in NH₄⁺ after oxidizing it with NO₃⁻ (Felix et al., 2013). These methods have been applied to the estimation of sources of particulate matter and gaseous nitrogen compounds and other research on environmental isotopes (Elliott et al., 2009; Felix et al., 2012; Hastings et al., 2013; Redling et al., 2013; Felix and Elliott, 2014; Felix et al., 2014; Walters et al., 2015a,b; Fibiger and Hastings, 2016; Pan et al., 2016; Walters et al., 2016; Felix et al., 2017; Yu and Elliott, 2017; Walters and Hastings, 2018). However, only a few studies have measured nitrogen isotopes in gases and particles collected simultaneously over extended periods and have examined isotopic fractionation during gas–particle reactions (Moore, 1977; Freyer, 1991). The review by Elliott et al. (2019) described that medium-, and long-term data need for future research.

In this study, stable nitrogen isotope ratios (δ¹⁵N) of gaseous NH₃, NO₂, and HNO₃ (δ¹⁵N-NH₃, δ¹⁵N-NO₂, and δ¹⁵N-HNO₃, respectively) and of NO₃⁻ and NH₄⁺ ions (δ¹⁵N-NO₃⁻ and δ¹⁵N-NH₄⁺, respectively) in SPM were measured for long term sampling, and the analysis results were used to examine isotope fractionation during gas-to-aerosol reactions and the sources and behavior of these compounds in the atmosphere.

2. Methods

2.1. SPM sampling and sampling of gases in ambient air

Ambient measurements were performed in a typical Japanese rural area at Akita Prefectural University in Yurihonjo City, Akita Prefecture, Japan (latitude 39.2°N, longitude 140.4°E). The site is ∼5 km from the coast. A rice paddy (200 m), a residential area (1 km), and a cow barn (400 m) are situated near the measurement site, but there is no large single point source of pollution such as a large factory nearby. Aerosol sampling was conducted over ∼10-day periods from April 2008 to December 2010. At least 0.24–0.42 mg of nitrogen is needed to measure δ¹⁵N-NH₄⁺ and δ¹⁵N-NO₃⁻ in SPM (i.e., 0.3–0.54 mg of N-NH₄⁺ and 1.06–1.86 mg of N-NO₃⁻) (Sakata, 2001), Typically, NH₄⁺ and NO₃⁻ concentrations in SPM exceed 0.02 and 0.07 μg m⁻³, respectively. Therefore, a high-volume air sampler (HVS-1000; Sibata Scientific Technology, Saitama, Japan) with a flow rate of ∼1000 L min⁻¹ was used to collect SPM. Glass fiber filters (GB-100R; Advantec Toyo Kaisha, Tokyo, Japan) were placed inside desiccators for a sufficient time period (at least 1 week) to remove any moisture content before they were used in the sampler. After the sampling, the sample filters were returned to the desiccator and then weighed on an electronic microbalance (LA220S-F; Sartorius Stedim Japan K.K., Tokyo, Japan; reading precision, 0.1 mg). After they were weighed, the sample filters were sealed in clean plastic bags and stored in a freezer at −31 °C.

The filter pack method was used to capture NH₃, NO₂, and HNO₃ gases on filters (AF-20Y; Advantec Toyo Kaisha) placed inside a custom-made attachment installed in the air sampler under the aerosol filters. Before use, all filters were impregnated as described below inside a glove bag filled with ultra-high-purity nitrogen gas (99.9999%) and then sealed in clean plastic bags. All impregnating reagents were selected following (Hayami and Fujita, 2004). For NH₃ gas, the filter was impregnated with 20% citric acid + 5% glycerin, and the sampling period was ∼2 days. Phosphoric acid was used for the filter pack for NH₃ gas (Hayasaka et al., 2004; Felix et al., 2013).

For NO₂ gas, the filter was coated with 10% triethanolamine + 20% potassium hydroxide + 1% glycerin, and the sampling interval was set to ∼4 days following (Moore, 1974). For HNO₃ gas, the filter was coated with 20% potassium carbonate + 1% glycerin, and the sampling interval was ∼4 days (Dasch et al., 1989; Harrison and Kitto, 1990; Noguchi et al., 2007). The alkaline filter collected HNO₃ gas, is possible that other gases such as HONO gas. However, regarding NO₂ and NO gas, the collection rate was very low (≤1%) in the alkaline filter (Noguchi et al., 2007) and it was not necessary to consider. In this study, the filter collected both HNO₃ gas and HONO gas.

Sampling of NH₃, NO₂ and HNO₃ gases was conducted from December 2009 to December 2010. Also, all blanks could not be observed. In the pretreatment method using the distillation method of this study, it is reasonable that no blank appears because the analysis method is low in sensitivity.

2.2. Sampling of gas sources: cow barn and roadside sampling

Gas sources sampled were a cow barn in Yurihonjo City, ∼400 m south of Akita Prefectural University, and a roadside site in Akita City. In the cow barn, ∼50 calves are commercially raised at any given time, and the barn is not near any other pollutant source. The roadside sampling site is in a heavy-traffic area near a crossroads, and the only pollutant source at the site is vehicle emissions.

Gaseous NH₃, NO₂, and HNO₃ (including HONO gas) were captured by passive diffusion on filters (AF-20Y; Advantec Toyo Kaisha, Ltd.) that were impregnated as described in section 2.1 and installed in a series of plastic boxes inside a custom-made wooden box. In addition to the impregnation reagents described above, the filter for NH₃ gas was also coated with 5% phosphoric acid + 5% glycerin. Sampling of the source gases was conducted over periods of ∼1 month from 14 December 2009 to 18 December 2010.

In addition, Takegawa et al. (2004) described that the average lifetimes of NO_x and NO_y are estimated to be 1.2 ± 0.4 and 1.7 ± 0.5 days, respectively. For the filter pack method in this study, sampling period for NO₂ and HNO₃ conducted 4 days. The passive sampling conducted ∼1 month. Thus, the gases seemed to be mixed sufficiently.

2.3. Sample preparation for isotope analysis

NH₄⁺ and NO₃⁻ were isolated for δ¹⁵N analysis following the method of Sakata (2001) as follows. First, NH₄⁺ and NO₃⁻ were ultrasonically extracted from the sample filters in deionized water. The extracts were diluted by adding additional deionized water and then alkalized with NaOH in a flask. Next, NH₄⁺ and NO₃⁻ were separated by a distillation procedure based on Japanese Industrial Standards Committee (1998); in this procedure, only the extracted NH₄⁺ is volatilized and absorbed by diluted H₂SO₄ (pH 2). After the distillation, the flask was cooled, and then the residue, without NH₄⁺, was mixed with deionized water and Devarda’s alloy (No. 042-00525, 50% copper, 45% aluminum, 5% zinc, Wako Pure Chemical) to reduce the NO₃⁻ to NH₄⁺. The NH₄⁺ was then separated by distillation as described above and absorbed by diluted H₂SO₄. Finally, NH₄⁺ was precipitated as the insoluble salt (C₆H₅)₄BNH₄ from each distillate by adding (C₆H₅)₄BNa (No. 346-01535, Wako Pure Chemical) to the distillate and stirring. The precipitate was separated by filtering the solution through a 0.22-μm membrane filter (GVWP04700; Nihon Millipore K.K., Tokyo, Japan), oven-dried, packed in a tin cup, and subjected to δ¹⁵N measurement by elemental analysis (EA; Flash EA, Thermo Fisher Scientific Inc., Bremen, Germany) and isotope ratio mass spectrometry (IRMS; MAT253, Thermo Fisher Scientific) within 1 day. Stable nitrogen isotope ratios, expressed as permil (‰) in δ notation, were calculated with Eq (1)(1) $${\delta }^{15}\mathrm{N}\mathrm{ }\left[\mathrm{‰}\right]=(\frac{{({}^{15}\mathrm{N}/{}^{14}\mathrm{N})}_{\mathrm{sample}}}{{({}^{15}\mathrm{N}/{}^{14}\mathrm{N})}_{\mathrm{std}}}-1)$$(1) : (1) $${\delta }^{15}\mathrm{N}\mathrm{ }\left[\mathrm{‰}\right]=(\frac{{({}^{15}\mathrm{N}/{}^{14}\mathrm{N})}_{\mathrm{sample}}}{{({}^{15}\mathrm{N}/{}^{14}\mathrm{N})}_{\mathrm{std}}}-1)$$(1) where (¹⁵N/¹⁴N)_sample and (¹⁵N/¹⁵N)_std are the ratios of ¹⁵N to ¹⁴N in the sample and the IAEA-N1 international atmospheric N₂ standard (RM8547, National Institute of Standards and Technology, +0.4 ± 0.2‰), respectively. The δ¹⁵N results were verified by measuring a l-histidine working standard (Shoko, Tokyo, Japan, δ¹⁵N = −8.42‰) and calibrated using the standard. The working standard and international standard were measured after every 6 and 30 samples, respectively. δ¹⁵N was calculated using ISODAT NT 2.0 IRMS software.

The precision and accuracy of the separation method in this study were tested following Kawashima and Kurahashi (2011) by using purchased NH₄Cl (017-02995, >99.5%; Wako Pure Chemical) and KNO₃ (160-04035, >99.5%, Wako Pure Chemical) reagents, which were individually analyzed and then measured by EA–IRMS (six times to assess reproducibility). The δ¹⁵N-NH₄⁺ and δ¹⁵N-NO₃⁻ values of the reagent standards were −5.91 ± 0.02‰ and +0.45 ± 0.09‰, respectively. The δ¹⁵N-NH₄⁺ and δ¹⁵N-NO₃⁻ values obtained for these reagents after separation by the distillation procedure described above and EA–IRMS measurement were −5.94 ± 0.04‰ (a difference of −0.03‰) and +0.30 ± 0.14‰ (−0.15‰), respectively. These results confirm that the results obtained in this study are highly accurate and precise.

2.4. Water-soluble ions in SPM

Disks were cut from the filters with a 3.5-cm diameter punch, and their water-soluble ions measured. Ions were extracted from the sample filter in deionized water by ultrasonication for 60 min. The extracted solution was filtered through a 0.45-μm membrane filter (Mini Sart RC15; Sartorius Stedim) and analyzed by ion chromatography (ICS-1000; Nippon Dionex K.K., Osaka, Japan). The detection limits were comparable to those obtained in other studies (e.g., Shen et al., 2009). The method is described in detail by Kawashima and Kurahashi (2011).

3. Results and discussion

3.1. SPM and water-soluble ions

The average SPM concentration in ambient air was 15.4 ± 6.3 μg m⁻³, which is lower than the Japan Environmental Standard (annual standard 24-h average: 100 μg m⁻³) and lower than the concentration in other regions of Japan (2008 annual average in all cities 22 μg m⁻³ (Environment, 2017)). For example, the average aerosol mass in Sapporo in 2008 was 25 μg m⁻³ (Aggarwal and Kawamura, 2009). This concentration is also much lower than the total suspended particulates (TSP) concentration of 230.5 μg m⁻³ in Shanghai, China (Wang et al., 2006) and the PM₁₀ concentration of 316.2 μg m⁻³ in Nanjing, China (Wang et al., 2003). It is also lower than average SPM concentrations in other studies.

The total concentration of water-soluble ions in SPM was slightly higher in spring (March–May) than in other seasons (Fig. 1), in agreement with previous research (e.g., Wang et al., 2006). Anion concentrations decreased in the order SO₄^2– > NO₃^– > Cl^–, and cation concentrations in the order Na⁺ > NH₄⁺ > K⁺ > Ca²⁺ > Mg²⁺. Sulfate is a major water-soluble ion in Japan (e.g., Kaneyasu et al., 1995), and concentrations of NO₃^– and Cl^– were higher in winter (December–February) than in other seasons, apparently because particulate easily in winter (Stelson et al., 1979; Stelson and Seinfeld, 1982). NH₄⁺ was higher in summer (June–August) and spring than in autumn (September–November) and winter. When mean total cations were plotted against mean total anions (not shown), the slope (i.e., the inorganic cation: anion balance) was 0.934 (R² = 0.72); this result seems reasonable.

Fig. 1. Concentrations of water-soluble ions in SPM sampled at Akita Prefectural University during spring (March–May), summer (June–August), autumn (September–November), and winter (December–February) between April 2008 and December 2010.

3.2. Nitrogen isotope ratios in ammonia gas and particulate ammonium

3.2.1. Particulate ammonium

A total of 78 SPM samples were collected at Akita Prefectural University for δ¹⁵N-NH₄⁺ measurements from April 2008 to December 2010. In these samples, δ¹⁵N-NH₄⁺ ranged from 1.3‰ to 38.5‰ (average 16.9‰) (Fig. 2a, Table 1). Although Kawashima and Kurahashi (2011) could not detect a clear δ¹⁵N-NH₄⁺ trend in the SPM samples collected from April 2008 to November 2009, over the two and a half year period studied here, a seasonal trend was detected. Specifically, δ¹⁵N-NH₄⁺ values in SPM were slightly lower in winter (11.7 ± 7.0‰) and higher in the summer (18.0 ± 7.0‰) (p < 0.01). By comparison, Moore (1977) reported that the average δ¹⁵N-NH₄⁺ value in TSP sampled in Boulder, Colorado, USA, was 5.6 ± 5.5‰, and Heaton (1987) reported a δ¹⁵N-NH₄⁺ value in dry deposition in Pretoria, South Africa, of 2.3‰ (range −3.1‰ to 6.4‰) sampled from October 1984 to March 1985. Yeatman et al. (2001b) reported δ¹⁵N-NH₄⁺ values in TSP sampled from June 1996 to May 1997 in a farming area (Weybourne) and a town (Mace Head) in England of 6 ± 6‰ (1–22‰) and −9 ± 8‰ (−20‰ to 22‰), respectively. In Gosan, Jeju Island, Korea, δ¹⁵N-NH₄⁺ values in TSP ranged from 4‰ to 32.2‰ (median 17.4 ± 4.9‰) from April 2003 to April 2004 (Kundu et al. (2010). In Niigata Prefecture, Japan, δ¹⁵N-NH₄⁺ values in TSP ranged from 12.2‰ to 39.6‰ (average 22.1 ± 8.3‰) during March–October 2002 (Hayasaka et al., 2004). In addition, δ¹⁵N-NH₄⁺ in size-segregated particles sampled in Beijing, China, ranged from −37.1‰ to 5.8‰ during January 2002 (Pan et al., 2016). Although δ¹⁵N-NH₄⁺ values determined in this study were higher than the values reported by Moore (1977), Heaton (1987), and Yeatman et al. (2001b), they were similar to those reported by Kundu et al. (2010) and Hayasaka et al. (2004). The sampling site of Hayasaka et al. (2004), in particular, was in a typical Japanese rural area near a rice paddy and the sea, similar to the sampling site in this study, and both rice paddies and the sea are NH₃ emission sources (Barrett, 1998; Das et al., 2009).

Fig. 2. Variations of (a) δ¹⁵N-NH₄⁺ values and NH₄⁺ concentrations, and (b) δ¹⁵N-NO₃^– values and NO₃^– concentrations in samples collected at Akita Prefectural University from April 2008 to December 2010.

Table 1. Seasonal concentration (± S.D.) data for NH₃ and NH₄⁺, the NH₃/NH₄⁺ ratio, δ¹⁵N-NH₃, and δ¹⁵N-NH₄⁺, and the NH₃–NH₄⁺ fractionation factor (εNH₃–NH₄⁺) for samples collected at Akita Prefectural University from December 2009 to December 2010.

	NH3 (μg/m^3)	NH4^+ (μg/m^3)	NH3/NH4^+ (mol ratio)	δ^15N-NH3 (‰)	δ^15N-NH4^+ (‰)	εNH3–NH4^+ (‰)
Spring	5.38 ± 3.08	0.50 ± 0.35	11.3	–20.8 ± 8.4	15.3 ± 5.1	+37.4 ± 5.7
Summer	5.52 ± 3.99	0.43 ± 0.36	13.6	–16.7 ± 6.5	18.0 ± 7.0	+33.8 ± 9.8
Autumn	1.44 ± 0.53	0.28 ± 0.23	5.5	–8.1 ± 4.5	18.8 ± 5.3	+28.9 ± 9.2
Winter	1.77 ± 1.51	0.37 ± 0.22	5.1	–23.9 ± 6.7	11.7 ± 7.0	+33.9 ± 7.7
All seasons	3.36 ± 3.05	0.40 ± 0.31	9.0	–16.9 ± 8.8	16.1 ± 6.6	+33.3 ± 8.2

The mean NH₄⁺ concentration in the SPM samples was 0.40 ± 0.31 μg m⁻³ (Fig. 2a, Table 1), and it tended to be slightly higher in spring and summer (0.50 μg m⁻³ and 0.43 μg m⁻³, respectively) than in autumn and winter (0.28 μg m⁻³ and 0.37 μg m⁻³, respectively). Hayasaka et al. (2004) reported a similar seasonal distribution of NH₄⁺ concentrations in TSP samples in Niigata Prefecture.

3.2.2. Ammonia gas

A total of 20 gas samples were collected at Akita Prefectural University for δ¹⁵N-NH₃ measurement from December 2009 to November 2010. In these samples, δ¹⁵N-NH₃ ranged from −33.6‰ to −0.0‰ (average −16.9‰) (Fig. 3a, Table 1). By comparison, Hayasaka et al. (2004) reported that δ¹⁵N-NH₃ ranged from −14.5‰ to −1.0‰ (average −8.3‰) collected in Niigata Prefecture, Japan, during March–October 2002. In addition, the δ¹⁵N-NH₃ of gas samples collected using a passive air sampler at nine sites in the United States ranged from −42.4‰ to +7.1‰ (annual average −15.1 ± 9.7‰; n = 86) (Felix et al. (2017)).

Fig. 3. Variations from December 2009 to December 2010 of (a) δ¹⁵N-NH₄and δ¹⁵N-NH₃ in ambient air and at the cow barn site; (b) δ¹⁵N-NO₃^–, δ¹⁵N-NO_x, and δ¹⁵N-HNO₃ (including HONO) in ambient air and δ¹⁵N-NO_x and δ¹⁵N-HNO₃ (including HONO) at the cow barn and roadside sites.

The δ¹⁵N-NH₃ values measured in this study are somewhat lower than those reported by Hayasaka et al. (2004), whereas they are in agreement with those reported by Felix et al. (2017). Hayasaka et al. (2004) collected samples only from spring to autumn, which may account for their values being higher than those in this study, because δ¹⁵N-NH₃ values of the gas samples collected at Akita Prefectural University, similar to the δ¹⁵N-NH₄⁺ values of SPM, tended to be higher in summer and lower in winter (0.05 < p < 0.10) (Table 1).

As the artifact for sampling collection, according to the recent study of Savard et al. (2017), the isotope fractionation occurs when the ammonium nitrate particles are gasified. The δ¹⁵N-HNO₃ and δ¹⁵N-NH₃ decrease and the values for δ¹⁵N-NO₃⁻ and δ¹⁵N-NH₄⁺ increase during gasified. In summer when the temperature rises, the δ¹⁵N-NH₄⁺ might become higher if gasified. Thus, the reason of the trend was possible that the particles were gasified. However, the δ¹⁵N-NH₃ also became higher in summer and lower in winter as the similar trend for δ¹⁵N-NH₄⁺, so it is not likely that the particles gasified made a tendency. In addition, NH₄⁺ in this study has a very high correlation with SO₄²⁻ (R² = 0.86), and NO₃^- (R² = 0.36) is hardly formed, and it is difficult to think that it is decomposed (Kundu et al. (2010). Thus, the artifact from aerosol to gas should not be important in this research.

The average concentration of NH₃ gas sampled at Akita Prefectural University was 3.36 ± 3.05 μg m⁻³ (range 0.69–10.97 μg m⁻³) (Table 1). By comparison, Hayasaka et al. (2004) reported slightly lower ammonia concentrations (1.65 ± 0.42 μg m⁻³; range 0.81–2.79 μg m⁻³) in Niigata Prefecture during March–October 2002. NH₃ concentrations in this study were higher in spring and summer and lower in autumn and winter, and this seasonal trend was clearer for NH₃ gas than was the case for NH₄⁺ in SPM (see Section 3.2.1). In general, the atmospheric NH₃ concentrations increase in warmer seasons and decrease in colder seasons (e.g., Okita and Kanamori, 1971; Georgii and Müller, 1974; Walker et al., 2004), so the result reported herein is very reasonable.

3.2.3. Ammonia gas at source sites

At the cow barn site in Yurihonjo City and the roadside site in Akita City, δ¹⁵N-NH₃ ranged from −22.8‰ to −14.8‰ (average −18.8‰) and from −17.0‰ to −10.5‰ (average −13.1‰), respectively (Fig. 3a). No seasonal trend was detected at either source location. By comparison, Heaton (1987) reported values of −15.2‰ and −8.9‰ for a sheep shed and a chicken shed, respectively, and Freyer (1978) measured a δ¹⁵N-NH₃ value of −8.4 ± 3.0‰ (range −12.6‰ to 4.4‰) for a cow shed, and one of −6.1 ± 1.6‰ (−7.2‰ to −4.3‰) near a coal combustion site. Felix et al. (2013) reported δ¹⁵N-NH₃ values for vehicle exhaust of –4.6‰ and –2.2‰ (n = 2), emissions from a power plant equipped with a selective catalytic reduction system of –11.3‰ and –14.6‰ (n = 2), animal waste of –33.1 ± 11.8‰ (n = 7), and volatilized fertilizer of –48.0‰ to –36.3‰ (n = 6). These results suggest that in general industrial emission sources may be slightly higher than agricultural sources. However, δ¹⁵N-NH₃ values measured at three coal-fired power plants ranged from −38.0‰ to −21.1‰ (average −25.5‰) (Savard et al., 2017). Thus, because of the small sample sizes, industrial and agricultural sources do not show a clear difference.

3.2.4. Identification of ammonia gas and particulate ammonium sources and isotope fractionation during the gas-to-aerosol transformation

As described in section 3.2.2, δ¹⁵N-NH₃ values detected in summer were slightly higher than those measured in winter. This difference can be attributed only to source differences, because isotope fractionation does not need to be considered. In addition, the NH₃/NH₄⁺ ratio of 9.0 indicated that the ambient atmosphere was enriched in NH₃ in all seasons. Because no clear differences in δ¹⁵N-NH₃ were detected among sources (see section 3.2.3), other unmeasured sources must exist.

The seasonal trend of δ¹⁵N-NH₄⁺ in aerosols may reflect three effects: first, the effect of temperature on chemical reactions and the NH₃ concentration level; second, seasonal effects on isotope fractionation; and third, the effect of different δ¹⁵N-NH₃ sources. These effects, however, are not necessarily independent; for example, temperature differs seasonally, but sources may also differ seasonally and thus would be associated with different temperatures.

The correlation between δ¹⁵N-NH₄⁺ in SPM and ambient temperature in all data (Fig. 4a) was very weak (R² [coefficient of determination] = 0.28; slope = 0.38), so temperature alone cannot account for the seasonal trend. However, the correlation was much stronger for data collected at ambient temperatures from 0 to 15 °C (R² = 0.86; Fig. 4a). This result suggests that factors such as source differences and isotope fractionation had more effect at ambient temperatures >15 °C.

Fig. 4. Scatter plots between (a) δ¹⁵N-NH₄⁺ and air temperature, and (b) δ¹⁵N-NO₃^–and air temperature at Akita Prefectural University based on data from December 2009 to December 2010.

Isotope fractionation during the chemical transformation from NH₃ gas to NH₄⁺ in particulates can be estimated from field measurements, data collected in laboratory experiments, or by theoretical calculations. In this study, the annual mean isotope fractionation (εNH₃–NH₄⁺) based on field measurements was +33.3‰, with seasonal means of +37.4‰ (spring), +33.8‰ (summer), +28.9‰ (autumn), and +33.9‰ (winter) (Table 1). By contrast, isotope fractionation during the gas-to-aerosol transformation ranged from +19.3‰ to +48.7‰ (average +30.4‰) in Niigata, Japan (Hayasaka et al., 2004), and averaged +15.6‰ in Boulder, Colorado, USA (Moore, 1977). In laboratory tests, Heaton et al. (1997) obtained an isotopic enrichment factor of +33‰ for isotope fractionation during the transformation of NH₃ gas to (NH₄)₂SO₄ particles. In addition, theoretical calculations predict an isotope fractionation factor of +39‰ at 0 °C and +35‰ at 25 °C for the transformation of NH₃ to particulate NH₄⁺ (Urey, 1947). Recently, Walters et al. (2019) was estimated that Δδ¹⁵N were 34‰ (5 °C), 31‰ (20 °C) and 29‰ (40 °C) (in calculation) under equilibrium isotope exchange using computational quantum chemistry methods. Thus, the isotope fractionation factors obtained in this study are consistent with the Niigata data (Hayasaka et al., 2004), the laboratory test results (Heaton et al., 1997), and the theoretically calculated values (Urey, 1947; Walters et al., 2019). On the other hand, the Δδ¹⁵N by Moore (1977) (15.6‰) was lower than other researches. The molar amounts of NH_3(g) against NH₄⁺_(p) by Moore (1977) (0.42) was poor than Hayasaka et al. (2004) (7.5 ± 5.8) and this study.

In general, the kinetic isotope effect initially causes light ¹⁴NH₃ to be preferentially transformed into NH₄⁺ in aerosols because molecules containing heavy ¹⁵N have stronger bonds than those that include light ¹⁴N. Over time, heavy ¹⁵N is incorporated into NH₄⁺ in aerosols if isotopic equilibrium between the gas and particles is reached. Thus, δ¹⁵N-NH₄⁺ in aerosols is initially lower than gaseous δ¹⁴N-NH₃. If the atmospheric concentration of NH₃ is high, then isotopic equilibrium is eventually achieved, and the aerosol is likely to contain some heavy ¹⁵NH₄⁺. Conversely, when the atmospheric NH₃ concentration is low, all NH₃ is consumed before equilibrium is reached; as a result, the kinetic isotope effect is dominant.

In this research, the ratio of NH₃ gas to NH₄⁺ in SPM was higher in spring (11.3) and summer (13.6) and lower in autumn (5.5), and winter (5.1) (Table 1). These results suggest that in spring and summer, the atmospheric NH₃ concentration was relatively higher and δ¹⁵N-NH₄⁺ in SPM reflected isotopic equilibrium conditions. Conversely, in winter, the atmospheric NH₃ concentration was relatively lower owing to the absence of natural sources, so δ¹⁵N-NH₄⁺ in SPM reflected non-equilibrium conditions. In addition, δ¹⁵N-NH₄⁺ values in SPM probably reflect the kinetic isotope effect as well. Together, these considerations can explain the higher δ¹⁵N-NH₄⁺ values in SPM in spring and summer and the lower values in autumn and winter.

The isotope fractionation factor becomes larger with decreasing temperature and smaller with increasing temperature (Urey, 1947; Heaton et al., 1997; Savard et al., 2017). This fact implies that the seasonal trend of δ¹⁵N-NH₄⁺ in this study (slightly lower in winter and higher in summer) cannot be explained only by simple equilibrium reactions.

3.3. Nitrogen isotope ratios of particulate nitrate and nitrogen dioxide and nitric acid gases

3.3.1. Particulate nitrate

δ¹⁵N-NO₃⁻ values of 87 SPM samples collected at Akita Prefectural University from April 2008 to December 2010 ranged from −4.6‰ to 4.8‰ (average −0.5‰) (Fig. 2b, Table 2). By comparison, δ¹⁵N-NO₃⁻ averages in PM sampled at Weybourne and Mace Head, UK, from June 1996 to May 1997 were 7 ± 6‰ and −1 ± 3‰, respectively (Yeatman et al., 2001a); Heaton (1987) reported an average value of 6.2 ± 4.3‰ in Pretoria, South Africa, for November 1984 to March 1985; Freyer (1991) reported an average value of 5.2 ± 3.1‰ in Julich, Germany, for summer 1978 to autumn 1980; and the average δ¹⁵N-NO₃⁻ value in PM sampled in Boulder, Colorado, USA, was 5.6 ± 5.9‰ (Moore (1977). Although the values obtained in this study are slightly lower overall than those obtained by these other studies, they are very similar to the Mace Head values. In this study, a seasonal trend of δ¹⁵N-NO₃⁻ in SPM was clearly detected; δ¹⁵N-NO₃⁻ was obviously lower in summer (−2.5 ± 1.4‰) and higher in winter (3.1 ± 1.1‰) (p < 0.01); this trend is opposite to that of δ¹⁵N-NH₄⁺ in PM. Yeatman et al. (2001a) described the same seasonal trend for δ¹⁵N-NO₃⁻ in Weybourne, UK (0 ± 2‰ in summer and 9 ± 5‰ in winter), and Freyer (1991) also reported lower δ¹⁵N-NO₃⁻ values in PM in summer and higher values in winter. As the artifact for sampling collection in this study, the tendency for δ¹⁵N-NO₃⁻ was reversed as the temperature. Thus, even if the particles including NO₃⁻ after sampling were gasified, it seems that there seemed to be other reasons or little reason.

Table 2. Seasonal concentration (±S.D.) data for NO₂, HNO₃ and NO₃^–, the NO₂/HNO₃ and HNO₃/NO₃^– ratios, δ¹⁵N-NO_2, δ¹⁵N-NO₃, δ¹⁵N-NO₃^–, and the fractionations factors (ε) for NO₂ to HNO₃, HNO₃ to NO₃^–, and NO₂ to NO₃^– for samples collected at Akita Prefectural University from December 2009 to December 2010. HNO₃ is including HONO.

	NO2 (μg/m^3)	HNO3 (μg/m^3)	NO3^- (μg/m^3)	NO2/HNO3 (mol ratio)	HNO3/NO3^– (mol ratio)	δ^15N-NO2 (‰)	δ^15N-HNO3 (‰)	δ^15N-NO3^– (‰)	εNO2–HNO3 (‰)	εHNO3–NO3^– (‰)	εNO2–NO3^– (‰)
Spring	1.04 ± 0.63	0.60 ± 0.34	1.21 ± 0.64	2.3	0.5	–5.9 ± 1.1	–4.6 ± 0.7	–0.4 ± 1.6	+1.5 ± 1.1	+4.2 ± 1.0	+5.7 ± 1.0
Summer	3.18 ± 2.79	1.83 ± 0.52	0.63 ± 0.33	2.4	2.9	–6.6 ± 1.3	–6.2 ± 0.8	–2.5 ± 1.5	+0.2 ± 1.6	+3.8 ± 1.1	+4.0 ± 0.9
Autumn	0.56 ± 0.17	0.51 ± 0.24	0.77 ± 0.23	1.5	0.7	–4.5 ± 1.0	–5.7 ± 1.6	–1.0 ± 2.0	–1.6 ± 2.0	+5.5 ± 2.7	+3.9 ± 2.3
Winter	0.38 ± 0.18	0.34 ± 0.11	0.85 ± 0.35	1.5	0.4	–5.1 ± 1.8	–3.5 ± 3.9	3.1 ± 1.1	+1.6 ± 3.5	+6.4 ± 4.6	+8.0 ± 1.6
All seasons	1.27 ± 1.75	0.81 ± 0.67	0.86 ± 0.46	2.1	0.9	–5.4 ± 1.5	–5.0 ± 2.3	–0.5 ± 2.5	+0.5 ± 2.4	+4.9 ± 2.7	+5.4 ± 2.2

In this study, the mean NO₃^– concentration in SPM was 0.86 ± 0.46 μg m⁻³ (Fig. 2b, Table 2), and it was higher in spring and winter than in the other seasons. Thus, NO₃^– and δ¹⁵N-NO₃⁻ in SPM showed different seasonal trends. By comparison, Wang et al. (2006) reported a similar seasonal trend for NO₃^– in TSP in Shanghai, China.

3.3.2. Nitrogen dioxide and nitric acid gases

δ¹⁵N values of NO₂ and HNO₃, the gaseous precursors of NO₃⁻ in SPM, ranged from –8.2‰ to –3.1‰ (average −5.4‰) for δ¹⁵N-NO₂ and from −7.5‰ to 2.7‰ (average −5.0‰) for δ¹⁵N-HNO₃ (Fig. 3b, Table 2). By comparison, δ¹⁵N-HNO₃ was determined to be ∼−2.5‰ in Julich, Germany, from summer 1978 to autumn 1980 (Freyer, 1978). Although both averages for δ¹⁵N-NO₂ and δ¹⁵N-HNO₃ in this study gas were slightly lower in summer (−6.4 ± 1.3‰ and −6.2 ± 0.8‰, respectively) and higher in winter (−5.1 ± 1.8‰ and −3.5 ± 3.9‰, respectively), no significant statistical differences were described between summer and winter (p > 0.1). In addition, Freyer (1991) did not report any seasonal trend of δ¹⁵N-HNO₃ in the data from Julich, Germany. In this study NO₂ and HNO₃ concentrations were 1.27 ± 1.75 μg m⁻³ (0.17–8.11 μg m⁻³) and 0.81 ± 0.67μg m⁻³ (0.22–2.65 μg m⁻³), respectively, and they were higher in summer (NO₂, 3.18 ± 2.79 μg m⁻³; HNO₃, 1,83 ± 0.67 μg m⁻³) than in winter (NO₂, 0.34 ± 0.11 μg m⁻³; HNO₃, 0.38 ± 0.18 μg m⁻³).

3.3.3. Gaseous nitrogen dioxide and nitric acid at source locations

The δ¹⁵N-NO₂ values at the cow barn and roadside sites ranged from −8.7‰ to −2.8‰ (average −6.2‰) and from −2.4‰ to 2.8‰ (average −0.48‰), respectively (Fig. 3b), whereas δ¹⁵N-HNO₃ values at the cow barn and roadside sites ranged from −5.6‰ to 5.9‰ (average −2.7‰) and from −1.0‰ to 6.1‰ (average 2.6‰), respectively (Fig. 3b). δ¹⁵N-NO₂ was not measured at the cow barn in winter, because it is difficult to analyze filters coated with triethanolamine at low temperatures (Okita et al., 1983; Dahal and Hastings, 2016). Although δ¹⁵N-HNO₃ values at the cow barn site showed a small seasonal trend, the values in winter may have been affected by vehicle emissions because no emissions from natural sources are expected in winter. Furthermore, δ¹⁵N-HNO₃, and the δ¹⁵N-NO₂ values from samples collected at the roadside in summer likely reflect agricultural sources as well as vehicle emissions. Thus, δ¹⁵N-HNO₃ and δ¹⁵N-NO₂ values obtained from the cow barn in summer and from the roadside in winter are suitable for source evaluation. Heaton (1990) reported that δ¹⁵N-NO_x of combustion products differ greatly depending on the temperature of combustion: at combustion temperatures above 2000 °C (e.g., in vehicle engines), δ¹⁵N-NO_x values range from −13‰ to −2‰, and at combustion temperatures under 1300 °C (e.g., coal combustion, home heating), they range from 6‰ to 13‰. In addition, Bowman (1975) and Hayhurst and Vince (1980) reported kinetic variations in the formation and destruction of nitrogen oxides that depended on combustion temperature. Felix et al. (2012) also reported positive δ¹⁵N-NO_x values in emissions from coal-fired power plant stacks, whereas Walters et al. (2015b) reported δ¹⁵N-NO_x values ranging from −28.1‰ to 8.5‰ in different exhaust gas samples; the tendency toward negative values in their results is attributable to the kinetic isotope effect. In addition, δ¹⁵N-NO_x values from fossil-fuel combustion sources equipped with selective catalytic reducers are relatively higher. Both NO and NO₂ are emitted from fertilizer-amended soil (Slemr and Seiler (1984). Li and Wang (2008) reported that in soil emissions due to bacterial activity, δ¹⁵N-NO ranged from −49‰ to −20‰ for fertilizer and from −49‰ to −28‰ for urea, and Yu and Elliott (2017) reported δ¹⁵N-NO values in soil emissions ranging from −59.8‰ to −23.4‰. In summary, combustion emissions are heavy δ¹⁵N-NO sources, vehicles emissions are middle-range δ¹⁵N-NO sources, and agricultural emissions are sources of light δ¹⁵N-NO.

3.3.4. Identification of nitrogen dioxide gas and particulate nitrate sources and isotope fractionation during the gas-to-aerosol transformations

The isotope fractionation factor for the transformation from NO₂ gas to HNO₃ gas (εNO₂–HNO₃) was ∼+0.5‰, and that for the transformation from HNO₃ gas to NO₃^– in SPM (εHNO₃–NO₃) was ∼+4.9‰ (Fig. 2b, Table 2). As the chemical reactions progressed, δ¹⁵N in SPM became steadily heavier. These results are consistent with theoretical calculations for these reactions (Walters and Michalski, 2015). Isotope fractionation from gas to gas was smaller than that from gas to aerosol. By contrast, isotope fractionation between HNO₃ and NO₃^– in PM was +8.9‰ (+6.7‰ to +11.9‰) (calculated by Freyer (1991), larger than the value obtained in this study, possibly because Freyer (1991) used a low-volume technique to obtain the isotope data; therefore, some HNO₃ may have been volatilized, with the result that δ¹⁵N values of the NO₃^– were higher. In addition, isotope fractionation from NO₂ to HNO₃ in the source samples was +2.4‰ for the cow barn and +2.6‰ for the roadside, larger than the isotope fractionation calculated for ambient data (+0.4‰). It is possible that at the source sites, the kinetic isotope effect was dominant because equilibrium was not reached due to a low NO₂/HNO₃ ratio.

At the Akita Prefectural University, δ¹⁵N-NO₃⁻ in SPM showed a clear seasonal trend (Fig. 3b). Freyer (1991) suggested three possible reasons for a seasonal trend in δ¹⁵N-NO₃⁻: (1) anthropogenic and natural NO_x sources may vary among the seasons; (2) the isotope exchange equilibrium between atmospheric nitrogen oxide species may be temperature dependent; and (3) there may be seasonal variations in the formation mechanisms of particulate nitrate. With regard to a possible temperature effect on δ¹⁵N-NO₃⁻ in SPM, the correlation between δ¹⁵N-NO₃⁻ in SPM and temperature was very strong (R² = 0.71, slope = −0.23) (Fig. 4b) and opposite in sign to the trend between δ¹⁵N-NH₄⁺ in SPM at temperatures under 15 °C (see Section 3.2.4). Furthermore, Freyer (1991) obtained a similar coefficient of determination for this relationship. However, this result may be due to a temperature effect on isotope fractionation or it may reflect different sources depending on the season. In general, the precursors of NO₃⁻ in aerosols (NO, NO₂, NOx, and HNO₃) are emitted during combustion (e.g., Galloway et al., 2004) and are likely to increase in winter as a result of domestic heating. Therefore, it is difficult to isolate the effect of temperature, though it is likely to be an important factor. Freyer (1991) suggested that the seasonal trend in winter of NO₃⁻ in aerosols is due to the second process, which has low temperature dependence. With regard to the source of δ¹⁵N-NO₃⁻ in SPM, the dominant source of NO₃⁻ in SPM is generally NO₂ from combustion. As summarized in Section 3.3.3, the δ¹⁵N-NO from combustion sources is high, that from vehicle emissions is intermediate, and that from natural sources is low. In this study, annual δ¹⁵N-NO₂ values primarily reflected NO_x vehicle emissions, so the baseline value could be considered to be intermediate. In summer, NO emissions due to bacterial activity would decrease the baseline because δ¹⁵N-NO from bacterial emissions is very low. In winter, the greater prevalence of emission sources such as coal combustion and domestic heating would increase the baseline, because δ¹⁵N-NO_x produced by combustion at low temperatures is very heavy. These source changes might cause there to be a slight seasonal trend in δ¹⁵N-NO_x. Then, the seasonal trend of δ¹⁵N-NO₃⁻ in SPM might be stronger than that in δ¹⁵N-NO_x because of the seasonal trend in isotope fractionation.

4. Conclusions

δ¹⁵N-NH₄⁺ and δ¹⁵N-NO₃⁻ in SPM were investigated in Akita Prefecture, Japan, from April 2008 to December 2010 to analyze seasonal trends and the sources of their precursors. The main results and conclusions from this work are as follows:

• δ¹⁵N-NH₄⁺ in SPM ranged from 1.3‰ to 38.5‰ (average 16.1‰), whereas δ¹⁵N-NH₃ gas ranged from −33.6‰ to −0.0‰ (average −16.9‰). δ¹⁵N-NH₄⁺ in SPM was slightly higher in summer and lower in winter. The ratio of NH₃ gas to NH₄⁺ in SPM was higher in summer (13.6) than in other seasons, and the NH₃ pool in the atmosphere was large; therefore, δ¹⁵N-NH₄⁺ in SPM reflected equilibrium conditions. Conversely, in winter, the ratio of NH₃ gas to NH₄⁺ in SPM was lower (5.1) owing to the absence of natural sources of NH₃, and the NH₃ pool in the atmosphere was small. Thus, δ¹⁵N-NH₄⁺ in SPM might reflect non-equilibrium conditions. In addition, the transformation of NH₃ to NH₄⁺ in SPM might partly reflect the kinetic isotope effect.

• δ¹⁵N-NO₃⁻ in SPM ranged from −4.6‰ to 4.8‰ (average −0.5‰), whereas δ¹⁵N-NO₂ gas and δ¹⁵N-HNO₃ gas ranged from –8.2‰ to –3.1‰ (average −5.4‰) and from −7.5‰ to 2.7‰ (average −5.0‰), respectively. δ¹⁵N-NO₃⁻ in SPM was obviously lower in summer and higher in winter, opposite to the trend of δ¹⁵N-NH₄⁺ in SPM. This seasonal trend is attributable to seasonal differences in NO_x sources, whether equilibrium conditions were reached, and whether there were seasonal variations in formation mechanisms. The annual δ¹⁵N-NO_x value in this study primarily reflected NO_x in vehicle emissions and can be used as a baseline. In summer, NO emitted by bacterial activity decreased the baseline because the δ¹⁵N-NO from bacteria is very low. In winter, emission sources such as coal combustion and home heating became more prevalent and increased the baseline.

• Mean annual isotope fractionation from NH₃ gas to NH₄⁺ in SPM in the ambient environment was +33.3‰. The fractionation factor from NH₃ gas to NH₄⁺ in SPM calculated in this study is consistent with previous research. Isotope fractionation from NH₃ gas to NH₄⁺ in SPM was +37.4‰ in spring, +33.8‰ in summer, +28.9‰ in autumn, and +33.9‰ in winter. Isotope fractionation from NO₂ to HNO₃ and from HNO₃ to NO₃^– in SPM was ∼+0.5‰ and +4.9‰, respectively. As the chemical reactions progressed, the δ¹⁵N became steadily higher. Isotope fractionation from gas to gas was smaller than that from gas to aerosol. These isotope fractionation factors were calculated from measurement data obtained simultaneously.

Acknowledgements

Takahiro Kurahashi and Yuya Haneishi at Akita Prefectural University contributed greatly, including conducting sampling and participating in discussions related to this research. Dr. Sakata Masahiro of the University of Shizuoka helped with the distillation of ammonia and nitrate in particulate matter. The author offers sincere thanks to all of these people.

Additional information

Funding

This study was supported by the Nippon Life Insurance Foundation, the Tostem Foundation for Construction Materials Industry Promotion, the Japan Securities Scholarship Foundation, and a Grant-in-Aid for Scientific Research on Innovative Areas, No. 23120704, a Grant-in-Aid for Scientific Research (B) No. 15H05121 from the Ministry of Education, Culture, Sports, Science and Technology of Japan, and the Environment Research and Technology Development Fund (5RF-1401) of the Environmental Restoration and Conservation Agency of Japan.