STABLE CARBON AND NITROGEN ISOTOPE RATIO IN PM₁ AND SIZE SEGREGATED AEROSOL PARTICLES OVER THE BALTIC SEA

1. Introduction

The coastal environment is a special environment where the mixing of marine boundary and continental boundary layers occurs [1, 2] and it is influenced by both marine and continental sources. Carbonaceous material accounts for a significant fraction of both marine and continental aerosols  [3–5]. The  coastal marine atmosphere adjacent to large urban and industrial centres can be strongly impacted by pollution emissions, resulting in high loading of pollutants in the ambient air. According to several authors, the  carbonaceous aerosol concentration in European coastal areas, including the  coastal region of the  Baltic Sea, exceeds typical clean marine concentrations of 0.14–0.6 µg/m³ and nearly reaches the values of urban particle concentrations of 5–10 µg/m³ [6]. A  substantial influence of continental emissions has been widely observed, not only in the coastal areas but also in the  open seas  [7, 8]. In addition to emissions originated from the  oceans, continental outflow is also an important source of carbonaceous aerosols in marine environment [3, 9, 10]. Chesselet et al. (1981) [11] concluded that more than 80% of atmospheric particulate organic carbon was of continental origin over remote marine areas. Fu et al. (2011) [9] reported that fossil fuel combustion and biomass burning are the major sources of organic aerosols over the Western Pacific. Marine natural emissions of carbonaceous aerosol (biological sources) were found to be the  most important contributor only over some remote oceanic areas [3, 4]. Anthropogenic carbonaceous aerosols mainly originate from fuel combustion, industrial processes and biomass/biofuel burning. Due to a complex mixture of the marine and continental air masses source, the apportionment of the  carbonaceous fraction is especially difficult in the coastal environment.

The stable carbon isotope ratio (δ¹³C) of carbonaceous aerosol was used in a number of studies to identify and apportion main pollution sources [12–18]. The stable carbon isotope ratio was also used to quantify aerosol particles of marine origin [19, 20].

Nitrogen isotope analysis is also a useful tool to discriminate the origin of nitrogen containing aerosol particles, but less used than δ¹³C, because δ¹⁵N depends on the sources of PM and can also be modified by chemical and physical processes in the atmosphere [21–25]. Particulate matter derived from biomass burning (C3 plants) had δ¹⁵N values ranging from 2 to 19.5‰  [26]. Widory (2007) [21] determined δ¹⁵N values for the main heating sources such as natural gas 7.7±5.9‰, coal combustion –5.6‰ and fuel oil –7.5±8.3‰. Traffic related PM had δ¹⁵N values about 4.6±0.8‰ [21]. Total nitrogen (TN) in aerosol particles is the sum of ammonium (NH⁺₄), nitrate (NO^–₃) and organic nitrogen (ON). These forms of nitrogen have different sources and transformation pathways in the atmosphere, resulting in different δ¹⁵N values of TN  [27–31]. Previous authors demonstrated that using a combination of stable carbon and stable nitrogen isotopic compositions (in NH⁺₄, NO^–₃ and ON) facilitates the study of the contribution of anthropogenic (biomass burning and fossil fuel) and biogenic (marine) sources and chemical aging to ambient aerosol [32, 33].

The major goal of this study was to gain an insight into the origin of carbon and nitrogen in submicron and size-segregated aerosol particles in the costal environment and characterize the advection of continental pollution to the marine environment using TC and TN isotopic analysis.

2. Materials and methods

Aerosol samples were collected over the Baltic Sea during a cruise from 9 to 17 November 2012. PM₁ samples were collected at a flow rate of 30 L/min using a  low volume sampler (Leckel). Five PM₁ samples were collected during the cruise on quartz fiber filters (Millipore). In addition, size segregated aerosol samples were collected with a Micro-Orifice Uniform Deposit Impactor (MOUDI) model 110 with a flow rate of 30 L/min, and 50% aerodynamic cutoff diameters of 11 stages of 18, 9.9, 6.2, 2.5, 1.8, 1.0, 0.56, 0.32, 0.18, 0.10 and 0.056 µm. Aluminum foil was used as an impaction surface. The  quartz filters were pre-combusted at 600°C for 24  h to remove organic contaminants. After sampling all aerosol samples were stored in a refrigerator.

Figure  1 shows the  cruise tracks which are indicated by sample ID from F1 to F5. The  meteorological conditions, sampling periods and air mass back trajectory directions are presented in Table 1.

All aerosol samples were analysed for TC and TN contents and their isotopic compositions using an elemental analyzer (EA) FlashEA 1112 connected to a stable isotope ratio mass spectrometer (IRMS) ThermoFinnigan Delta Plus Advantage by the method described in  [34, 35]. Caffeine (IAEA-600) was used as a secondary reference material to determine TC, TN, and their isotopic compositions (δ¹³C and δ¹⁵N).

Table 1. Data about the samples (PM₁ and MOUDI) and weather conditions during the cruises in the Baltic Sea on 9–17 November 2012. See Fig.  1, coloured online.

Meteorological parameters recorded at a  meteorological synoptic station at shipboard during the  cruise include ambient temperature, relative humidity, wind speed and wind direction. The average ambient temperature was 7°C ranging from 3 to 10°C. The minimum air temperature was reported on 14–15 November 2012, while the maximum one on 9–12 November (see Table  1). The daily average relative humidity ranged from 64 to 88% (average 83%). The local wind direction varied between 120 to 280° and the average wind speed was 6.5 m/s.

Backward trajectories were calculated using the hybrid single particle Lagrangian (HYSPLIT) model from the  National Oceanic and Atmospheric Administration (NOAA) [36]. The trajectories were calculated every 6 h with a total 72 h duration at 500  m AGL (Above Ground Level). The analysis of air mass back trajectories indicates that during the sampling period S, SW and W air masses were prevailing (Fig. 2).

3. Results

3.1. Concentrations of total carbon and total nitrogen in PM₁

The variations of TC and TN mass concentration measured aboard RV Oceania during the  cruise on the Baltic Sea are shown in Fig. 3. TC concentrations ranged from 1.95 to 5.91  µg/m³ (average 4.27±1.56  µg/m³). TN concentrations varied from 1.3 to 2.8 µg/m³ (average 2.4±0.8 µg/m³). As shown in Fig. 3, the TN concentration followed the TC concentration which could point to the  same aerosol sources or source regions. The TN and TC concentrations depended on the ship location and on the air mass origin during the sampling campaign. The highest concentrations of TC (average 5.3±0.6 µg/m³) and TN (average 2.8±0.8  µg/m³) were measured when southern and southwestern air masses from Poland, Germany and the Czech Republic and the southern wind from Poland were prevailing (F1, F4, F5). These findings suggest that the majority of TC and TN in PM₁ were derived from regional anthropogenic sources during these time periods.

The lowest TC and TN concentrations (F2, F3) were observed in the  western outflow when air masses generally came from the North Atlantic, but passed over the industrial regions of England. The  observed TN and TC concentrations were higher than those reported for ‘clean marine’ aerosol (TC about 0.1–0.5 µg/m³, TN about 0.02–0.03  µg/m³) at Mace Head in the  North Atlantic [19, 37, 38]. Previous studies showed that carbonaceous aerosol particles carried to the coastal site of the Baltic Sea (Preila) from the northwestern direction via England were affected by anthropogenic sources (continental and marine transport) rather than biogenic sources (marine) [39, 40]. Ceburnis  et  al. (2011)  [19] revealed that in clean marine air masses (which had no continental pathways) 8–20% of carbonaceous aerosol could be attributed to fossil fuel sources (e.g. shipping). The  area of our study, namely, the  Baltic Sea, is one of the busiest seas in the world (>2,000 ships per day) and the amount of maritime traffic is predicted to keep growing (http://www.helcom.fi). The modelling results have shown that the relative contribution of ship emissions to the annual mean NO₂ is more than 40% over the Baltic Sea and 22– 28% for the entire Baltic Sea region. The average contributions of ships to the levels of PM_2.5 and EC (elemental carbon) are in a range of 4.3–6.5% and 5–7%, respectively [41].

The sample with the lowest total carbon concentration F2 (TC = 1.95±0.08 µg/m³, 1.34±0.05 µg/m³) was obtained when the  ship was located about 30–40  km from the  northern part of the  Polish coast and the  westerly wind direction was prevailing. This sample was most likely influenced by long-range transport (anthropogenic) and marine (biogenic and shipping) sources, rather than regional coastal sources. However, we assume that due to a low sea biological activity in the late fall period (9–17 November 2012) the contributions from marine biogenic sources to samples F1–F5 were insignificant. During the  relatively clean western air mass period (F3) the concentrations of TC (3.55±0.33 µg/m³) and TN (2.34±0.29 µg/m³) notably increased compared to those of sample F2. This happens when the ship spends more time nearby the  coastline of Poland. Thus, sample F3 was influenced by both long-range transport and regional coastal land-based pollution sources. Relatively high concentrations of TC and TN together with the  back trajectory analysis suggest that PM₁ samples during this study were affected by anthropogenic regional sources and long-range transport.

3.2. Carbon and nitrogen isotopic composition of PM₁

The analysis of δ¹³C-TC and δ¹⁵N-TN can give an indication of the origin of carbonaceous and nitrogen species in PM₁. δ¹³C-TC of PM₁ varied from –27.5 to –26.2‰ with an average of –26.8±0.5‰. The mean δ¹³C observed over the Baltic Sea is comparable to that determined in several urban and anthropogenically influenced locations in Europe [16, 17, 42, 43]. The  minimum δ¹³C value (–27.5±0.2‰) was observed on 11–12 November (F2) when the lowest TC concentration and the western outflow was prevailing (Fig. 4). Higher δ¹³C (–26.6±0.4‰) values were registered in the PM₁ samples with higher TC concentrations (F1, F3, F4) when the ship was located near the coastline of Poland and the highest δ¹³C value was observed in F5 near the harbour of Klaipėda. It seems that continental sources responsible for higher carbon concentrations are associated with higher δ¹³C values. The variation (from –27.0 to –26.2‰) in these samples indicated consistent regional coastal sources.

A wide variation of the total nitrogen isotopic ratios of PM₁ (from –0.2 to 10‰) was observed during the  study period (Fig.  4). The  mean of δ¹⁵N in PM₁ shows larger variations than those in δ¹³C. These variations of δ¹⁵N-TN may be due to the changes in the contribution of regional pollution sources, chemical aging of nitrogen species and abundances of NH⁺₄ and NO^–₃ in TN 33]. A scatter plot of δ¹⁵N against TN in Fig. 4 shows a tendency towards depletion in ¹⁵N at higher TN concentrations, that was probably affected by a  significant anthropogenic input from fossil fuel combustion which are generally depleted in ¹⁵N (δ¹⁵N < 0) [21] compared to other sources.

Figure 5 shows that δ¹³C is anti-correlated with δ¹⁵N values during the campaign. The highest δ¹³C (–26.4‰) and lowest δ¹⁵N (–0.2 and 0.8‰) values were registered in samples F4 and F5 when air masses arrived mainly from the  southwest direction (Poland). In this region, aerosol particles enriched in ¹³C are expected due to high contributions of particles from coal combustion [13]. According to the  official Polish Government Energy Policy Strategy, 89% of heat in Poland is generated from coal (https://www.worldenergy.org; https://euracoal.eu, 2018). Górka et al. (2014) [43] found that during the heating season in Poland coal-derived aerosol particles (δ¹³C  =  –24.5±0.5‰) comprise about 40% of the elemental carbon fraction. Previous studies revealed that traffic-derived aerosol particles in Eastern European countries (Lithuania, Poland) typically have the  mean δ¹³C signature of –28±0.5‰ [13, 18, 44]. Relying on the carbon isotopic composition, we can conclude that possible sources of PM₁ over the  coastal areas were combined emissions of solid fuel (coal) burning and vehicle exhaust. Furthermore, ¹³C enriched PM₁ samples in air masses from Poland are associated with higher contributions from coal combustion. The δ¹³C value for F2 is consistent with liquid fossil fuel combustion as the main source.

The lowest δ¹⁵N values (from –0.2 to 0.8‰) together with the  elevated TN concentration were observed on 14–17 November (samples F4, F5). Lower δ¹⁵N values indicate a  significant impact from fuel oil and coal combustion [21]. However, coal combustion yields particles low in nitrogen content (0.33%). The nitrogen concentration in samples F4 and F5 was high and this suggests strong contributions of secondary nitrate. Vodička et al. (2019) [33] showed that a decrease in δ¹⁵N of TN was associated with a higher content of NO^–₃, and elevated δ¹⁵N values were caused by higher levels of NH⁺₄ or OrgN. NO^–₃ originates mainly from fossil fuel combustion and biomass burning  [45], while NH⁺₄ originates from agricultural activities, biological emissions and, to a minor extent, from anthropogenic combustion sources (fossil fuel and biomass burning) [46–50]. Heaton  et  al. (1991)  [51] reported that δ¹⁵N of NO_x derived from diesel engines ranged from –13 to –2‰, while those from coal-fired power stations ranged from 6 to 13‰. Changes in the δ¹⁵N values during conversion of NO_x to nitrate are minor, therefore δ¹⁵N values of particulate nitrate imply NO_x sources [21, 52]. It has been reported by other authors that the transformation of NO_x to atmospherics nitrates elevates δ¹⁵N values (by about 10–15‰) from the  initial δ¹⁵N values of the  NO_x sources  [53]. We assume that NO^–₃ was more abundant in samples F4 and F5 when the ship was close to a nearby harbour (Gdansk) and the conversion of NO_x to NO^–₃ was more likely to be responsible for the observed δ¹⁵N of TN rather than the conversion of NH₃ ↔ NH⁺₄. Bearing in mind the nitrogen oxides as a source of nitrogen in the aerosol particles we assume that δ¹⁵N values (from –0.2 to 0.8‰) of samples F4 and F5 are similar to those of NO_x generated from vehicular exhaust or ship emissions. We can conclude that the dominant sources of TN in samples F4, F5 most probably reflect the NO_x input generated by liquid fuel combustion (on road traffic, shipping).

The highest δ¹⁵N values (up to 10‰) were observed on 9–12 November (F1, F2) and were accompanied by lower TN concentrations. High δ¹⁵N values imply that secondary formation of NH⁺₄ played a  larger role in nitrogen contents of PM₁ in these samples. Previous studies based on chemical analysis and the  δ¹⁵N(NH⁺₄) approach showed that in Poland fossil fuel burning is the main source of NH⁺₄ in precipitation during the heating season [54]. Precipitation samples during November are characterized by a low NH⁺₄ concentration and δ¹⁵N (NH⁺₄) values in a range of 5–10‰. However, it is also possible that δ¹⁵N values have become more enriched during long-range transport to the site.

As shown in Fig.  5, samples F1, F2 are characterized by lower δ¹³C (–27.0‰ and 27.5‰, respectively) and the  highest δ¹⁵N (9.8±0.1‰ and 8.3±0.2‰, respectively) values. The  δ¹³C values for samples F1, F2 imply that they were much less affected by coal, but more by diesel/gasoline combustion (shipping and traffic).

Combining stable carbon and nitrogen isotopic data with the TN and TC concentrations we were able to discriminate likely aerosol sources. Relatively high δ¹³C and low δ¹⁵N values indicate fossil fuel (coal and diesel) as a dominant source for the aerosol particles consistent with air mass back trajectories crossing Poland, while relatively low δ¹³C and high δ¹⁵N values indicate that the main source of TC and TN could be shipping activities and aged aerosol particles, respectively.

3.3. δ¹³C and δ¹⁵N in size segregated aerosol particles

The total carbon and total nitrogen size distributions obtained from two MOUDI sampling periods (M1 9–13 November and M2 13–17 November) are presented in Fig. 6. During the M1 period (9–13 November) air masses were mainly transported from the western direction via the North Atlantic and England and only partly passed over southwestern Europe’s industrial regions (Poland, Germany, the Czech Republic). This period is characterized as a relatively clean period. During the M2 period air masses were mainly passing over Southern Europe (polluted period).

As shown in Fig.  6, the  size distributions of TC for relatively clean and polluted periods were different. During the  cleaner period, TC exhibited bimodal mass-size distribution with the  first peak clearly identified in the accumulation mode between 0.56 and 1 µm and other slight and less-marked peak centered at approximately 1.8–2.5 µm. During the polluted period, the shape of the  TC size distribution had a  unimodal peak in a size range of 1–1.8 µm.

Figure  7 gives an overview of a  stable carbon and nitrogen isotopic composition in size segregated aerosol particles. The  stable carbon isotopic composition varied from –26.3±0.1‰ to –28.2±0.1‰ during the  M1 period and from –25.2±0.2‰ to –27.9±0.1‰ during the  M2 period. The  first period M1 was characterized by a lower TC concentration and a lower variability of δ¹³C values (1.9‰). During the polluted period M2 the highest TC concentration and a larger variability of δ¹³C values (2.7‰) were observed. The δ¹³C values had a similar distribution within the particle size during the relatively clean M1 and polluted M2 periods: the  δ¹³C values increased with an increase of the size of fine particles. Most ¹³C depleted aerosol particles were observed in the size range from 0.056 to 0.18 µm. The mean δ¹³C value in PM_0.056–0.18 was –28.05±0.12‰, equivalent to δ¹³C values for vehicular exhaust (–28.0±0.9‰) [13, 18]. These ultra-fine particles have typically short lifetimes and therefore are characteristic of local sources. We can conclude that local coastal sources are, therefore, mainly liquid fuel combustion. The observed δ¹³C values of PM_0.18–2.5 (–26.4±0.1‰) are a possible mixture of coal and diesel/gasoline combustion particles. Particles in this size range have longer lifetimes and can be transported from further distances, which explains the stronger influence of coal combustion. Moreover, the carbonaceous species during the M2 period were only slightly enriched in ¹³C (0.5±0.1‰) compared to M1, suggesting similar sources for TC during both the periods. Nevertheless, higher δ¹³C values during the polluted period M2 point to a larger proportion of particles from coal combustion.

δ¹⁵N shows a strong enrichment of ¹⁵N during the  M1 period (δ¹⁵N about 11.5‰) in comparison with the  M2 period (δ¹⁵N about 2.9‰). As mentioned above (Subsection 3.2), in relatively clean air masses (M1) chemical aging of nitrogen species including the  gas-to-particle exchange (NH₃ ↔ NH⁺₄) reaction during long-range transport can cause an enrichment of ¹⁵N. Possibly, higher δ¹⁵N values during the M1 period could also be attributed to the impact from biomass burning (δ¹⁵N  =  10–20‰). Fossil fuel combustion (traffic or ship emissions) was presumably the dominant source of aerosol nitrogen during the polluted period (δ¹⁵N about 2.9‰). During this period biomass burning sources are less likely, because δ¹⁵N-TN were depleted in ¹⁵N.

Conclusions

Stable carbon and stable nitrogen isotope measurements were performed on PM₁ and size resolved aerosol samples collected during the cruise in the Baltic Sea from 9 to 17 November 2012. Total carbon (TC) and total nitrogen (TN) concentrations varied from 1.95 to 5.91 µg/m³ and from 1.3 to 2.8 µg/m³, respectively. Temporal variation of TC and TN concentrations during the  campaign depended on the ship location and air mass origin. The back trajectory analysis suggests that relatively high TC and TN loadings are associated with continental air masses (S, SW) and indicates that PM₁ samples are influenced by anthropogenic pollution. The total carbon isotopic composition of PM₁ varied from –27.5 to –26.2‰ with an average of –26.8±0.5‰. The  minimum δ¹³C value (–27.5±0.2‰) was observed in PM₁ samples with the lowest TC concentration when the western air masses were prevailing and the  air mass did not have a  recent contact with land. Higher δ¹³C (–26.6±0.4‰) values were associated with the highest TC levels when the ship was located nearby the  coastline of Poland and the  air mass came from the south.

Based on the δ¹³C analysis of TC we conclude that combined emissions of solid fuel (coal) and diesel/gasoline combustion (vehicular exhaust/ shipping) were the  most possible sources of PM₁ during continental outflow.

A wide variation of the total nitrogen isotopic ratios of PM₁ (from –0.2 to 10‰) was observed during the study period. Lower δ¹⁵N values indicated a significant influence from fossil fuel combustion, including ship emission and traffic. The high δ¹⁵N implies that the secondary formation of NH⁺₄ played a certain role in nitrogen contents of PM₁.

The δ¹³C values of size segregated aerosol particles (PM_0.056–2.5) increased with increase in the  size of aerosol particles. Most ¹³C depleted (δ¹³C = –28.05±0.12‰) aerosol particles were observed in the size range from 0.056 to 0.18 µm, suggesting predominant diesel/gasoline combustion sources for the ultra-fine particles of local, coastal origin. The  carbonaceous species during the  polluted period were slightly ¹³C enriched (0.5±0.1‰) compared to those during the relatively clean period (M1), suggesting similar sources for TC during both periods. The higher δ¹³C values obtained during the polluted period (M2) implied a slightly larger proportion from coal combustion consistent with the air mass transport mainly from the south. On the other hand, δ¹⁵N shows a strong enrichment of ¹⁵N during the M1 period (δ¹⁵N about 11.5‰) in comparison with that during the M2 period (δ¹⁵N about 2.9‰). The  highest δ¹⁵N values imply that TN during a relatively clean period was significantly influenced by chemical aging of nitrogen species. The lower δ¹⁵N values (δ¹⁵N about 2.9‰) obtained during the polluted period indicate that fossil fuel sources were dominant.

In summary, the study of stable carbon and stable nitrogen isotopes provides useful information on different sources and production pathways of TC and TN in fine aerosol particles.

STABILIŲJŲ ANGLIES IR AZOTO IZOTOPŲ SANTYKIO TYRIMAI ĮVAIRAUS DYDŽIO AEROZOLIO DALELĖSE (KD₁, KD0,_056–2,5)

Santrauka

Pateikiami stabiliųjų anglies (δ¹³C) ir azoto (δ¹⁵N) izotopų santykio verčių matavimai įvairaus dydžio (KD₁, KD_0,056–2,5) aerozolio dalelėse. Dalelės rinktos kruizo Baltijos jūroje metu 2012 m. lapkričio 9–17 dienomis.

Nustatyta, kad pietvakarinėse oro masėse KD₁ dalelėms būdingos didelės δ¹³C vertės (–26,4 ‰), kurių tikėtinas anglies turinčių aerozolio dalelių šaltinis yra akmens anglies ir dyzelino  /  benzino deginimas, o mažos δ¹⁵N (–0,2, 0,8  ‰) vertės parodo skysto kuro deginimo šaltinius (laivų ir autotransporto išmetalai). Bandiniuose, kuriuos renkant vyravo vakarinės oro masės, buvo priešingai – itin mažos δ¹³C (–27,5 ‰) ir didelės δ¹⁵N vertės (iki 10 ‰). Tokios δ¹⁵N vertės labiausiai siejamos su azotinių medžiagų fotocheminiais degradacijos procesais tolimosios pernašos metu, o mažos δ¹³C vertės rodo dyzelino / benzino deginimo šaltinį, kuris šiuo atveju labiausiai siejamas su laivų emisijomis.

Skirtingo dydžio aerozolio dalelių δ¹³C analizė parodė, kad mažiausios δ¹³C vertės buvo registruotos 0,056–0,18 µm dydžio dalelėse, o 0,18–2,5 µm dalelių dydžio intervale δ¹³C vertės nuosekliai didėjo. Toks kitimas atspindi aerozolio dalelių formavimosi mechanizmus ir šaltinius.