Ionic Composition of Fine Particulate Matter from Urban and Regional Background Sites in Poland

Concentration and ionic composition of PM_2.5

Average daily PM_2.5 concentrations ranged from 4.54 to 77.55 μg/m³, from 7.66 to 88.79 μg/m³, and from 8.90 to 120.77 μg/m³, respectively, for the sites in Szczecin, Trzebinia, and Złoty Potok. The lowest concentrations of PM_2.5 were recorded in Szczecin (averaging 17.11 μg/m³), which can be explained by the location of a measurement station in the coastal area and the impact of clean marine air masses, which have lower aerosol loadings (Czarnecka et al., 2011).

PM_2.5 concentrations in the urban background site in Trzebinia and the regional background site in Złoty Potok were comparable. The mean values for the entire measurement period equaled 25.21 μg/m³ (Trzebinia) and 25.16 μg/m³ (Złoty Potok), and they slightly exceeded the limit value for the average annual concentration of PM_2.5 (25 μg/m³), established by the European Commission (Directive 2008/50/EC).

Clear seasonal patterns of PM_2.5 have been observed in all of the three measurement sites (Table 2), with higher levels occurring during the heating season, similar to many other locations in Europe (Putaud et al., 2010). The nonparametric U Mann–Whitney test (α = 0.05) has shown that the differences between the daily PM_2.5 concentrations in both measurement periods were statistically significant.

The cause of the observed variability lies in the increased activity of stationary emission sources (combustion of fossil fuels and biomass) in the cold season, as well as in unfavorable weather conditions during this period (shallow mixing layer, frequent temperature inversions), which hinder the diffusion and removal of contaminants (Pastuszka et al., 2010; Rogula-Kozłowska et al., 2014a, 2014b; Reizer and Juda-Rezler, 2016). Low concentrations of PM_2.5 during the nonheating season are likely associated with the occurrence of more frequent and intense precipitation (Table 1), and a deeper mixing layer (Cong et al., 2011; Pateraki et al., 2014).

Figure 3 shows the percentage of the analyzed ions in PM_2.5 along with the carbon compounds contained in PM_2.5; carbon matter concentrations in these locations are described in Błaszczak et al. (2016a, 2016b). It has been found that besides the total carbon (TC) (expressed as the sum of elemental [EC] and organic carbon [OC]), the ions have a predominant share of the total mass of PM_2.5 in each location.

OPEN IN VIEWER

FIG. 3.

Contribution of selected ions in PM2.5 mass, along with other PM2.5 constituents, for each location per heating and nonheating season (Other mass is calculated as the difference between the measured PM_2.5 mass and OC, EC, and inorganic ions.).

The largest share of other substances, namely the matter unidentified neither as the carbon compounds nor as the analyzed ions, was noted for the PM_2.5 samples from Złoty Potok (Fig. 3). Probably, in all locations, the unidentified part of the PM mass consisted of water particles associated with PM (Rogula-Kozłowska et al., 2013a), and of a number of elements not analyzed within the present work, that is, noncarbon elements associated with organic matter in aerosol (Putaud et al., 2010; Chow et al., 2015).

Total concentrations of the analyzed ions (seasonal average) in the heating season were more than two-fold higher in comparison to the nonheating season (Table 2). The differences between the average contributions of all the analyzed ions in PM_2.5 mass during both measurement periods were also recorded. The seasonal difference was more visible in the case of Trzebinia, and it was less clear for Szczecin and Złoty Potok (Fig. 3). In the heating season, the ion concentrations were lower than the concentration of total carbon (OC + EC), in contrast to the nonheating period, in which the noted ion concentrations were slightly higher (except for the Złoty Potok site). For most ions (with the exception of sulfate ions in Szczecin and Trzebinia), the seasonal differences were statistically significant (U Mann–Whitney test, α = 0.05) (Table 2).

The greatest seasonal differences have been reported in the case of Cl⁻ in Trzebinia and K⁺ in Szczecin. The mean concentrations of the earlier mentioned ions were more than nine times higher in the heating season than in the nonheating season. The Cl⁻ ions bounded with PM_2.5 may have come from a variety of sources; however, in the urban area, they are primarily of anthropogenic origin (e.g., Rogula-Kozłowska et al., 2011, 2013a, 2013b; Pant et al., 2015). Numerous scientific publications highlighted that important sources of Cl⁻ ions in PM_2.5 are combustion processes, including coal and waste combustion, as well as wood burning (e.g., Richard et al., 2011; Rogula-Kozłowska et al., 2013b; Pant et al., 2015; Shen et al., 2016).

The water-soluble potassium is used as a tracer for biomass burning (Röösli et al., 2001; Vasconcellos et al., 2007; Pant et al., 2015). Fresh wood smoke contains, apart from K⁺ ions, also Ca²⁺, Mg²⁺, Na⁺, NH₄⁺, Cl⁻, NO₃⁻, and SO₄²⁻ (Jacobson, 2002; Vasconcellos et al., 2007). Taking into account these considerations, the results of this study suggest an important role of combustion sources, including biomass burning, for the air quality in considered areas.

PM_2.5 in the selected locations differed with regard to the ionic composition; the highest concentrations of ions were recorded in Trzebinia, located in a highly industrialized and populated area, and the lowest concentrations were recorded in Złoty Potok. Despite the spatial differences in the ion concentrations, there is a clear dominance of three ions, that is, sulfate (SO₄²⁻), nitrate (NO₃⁻), and ammonium (NH₄⁺), in the mass of detected compounds in the PM_2.5 samples from all locations (Table 2 and Fig. 3). The mean concentrations of other analyzed ions were significantly lower and did not exceed 1 μg/m³ (Table 2). Attention should be paid to the relatively high concentrations of Cl⁻ in Trzebinia, especially during the heating season.

Total shares of the sulfate, nitrate, and ammonium ions, averaged over the entire measurement period, were equal to ∼78%, ∼85%, and ∼89% of all the analyzed ions in Szczecin, Trzebinia, and Złoty Potok, respectively. Among them, the highest concentrations in the air were observed for the SO₄²⁻ ions, with average shares in PM_2.5 mass amounting to 11.14% and 19.01% (Szczecin), 11.38% and 21.70% (Trzebinia), and 8.99% and 11.34% (Złoty Potok) during the heating and nonheating seasons, respectively (Table 2 and Fig. 3).

The concentration of NO₃⁻ was generally lower, and its average share ranged from 4.31% (nonheating season, Złoty Potok) to 13.47% (heating season, Szczecin). The concentration of NH₄⁺ was generally low, as compared with SO₄²⁻ and NO₃⁻ (except from the nonheating season in Trzebinia), with an average share ranging from 1.78% (nonheating season, Złoty Potok) to 7.26% (heating season, Trzebinia).

Evaluation of the SIA share in PM_2.5 mass

Due to the fact that the amount of ammonium ions in the air is the major limiting factor in the formation of SIA, as they neutralize both NO₃⁻ and SO₄²⁻ ions in the air (Erisman and Schaap, 2004), the evaluation of the SIA contents in PM should be carried out after the assessment of the ion balance between the sum of nitrate and sulfate ions, and the ammonium ions. It is worth mentioning that in the assessment of the SIA content in PM, the concentration of nonsea salt sulfates (nss-SO₄²⁻) should be taken into account, thereby eliminating the impact of the sea-salt sulfates (ss-SO₄²⁻) on the results. In the present work, the amount of nss-SO₄²⁻ was calculated according to the following formula (Sillanpää et al., 2006): \documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland, xspace}\usepackage{amsmath, amsxtra}\usepackage{upgreek}\pagestyle{empty}\DeclareMathSizes{10}{9}{7}{6}\begin{document} \begin{align*} \left[ nss - S {{ \rm{O}}_{ \rm{4}}}{^{ \rm{2 - }}} \right] = \left[ S {{ \rm{O}}_{ \rm{4}}}{^{ \rm{2 - }}} \right] {{ \rm{ - 0}}{ \rm{.246 \cdot}}} [ { \rm{N}}{{ \rm{a}} ^{ \rm{ + }}}] \end{align*} \end{document}

Average concentrations of nss-sulfates, separately for heating and nonheating periods, were presented in Table 2, for each location. During the whole measurement period, the concentrations of nss-SO₄²⁻ averaged: 2.05 μg/m³ (Szczecin), 3.83 μg/m³ (Trzebinia), and 2.25 μg/m³ (Złoty Potok). The analysis of the graph of the linear regression between the sum of the NO₃⁻ + nss-SO₄²⁻ ions and the NH₄⁺ (in μeq/m³) ^† for the entire measurement period reveals that there is a strong (high determination coefficients [R²]) linear relationship between these two parameters (Fig. 4). The value of the linear regression coefficient for all locations is close to 1, which indicates that the ammonium ions almost completely neutralize the NO₃⁻ + nss-SO₄²⁻ ions. Therefore, it allows the calculation of SIA concentration as the sum of the NH₄⁺, NO₃⁻, and nss-SO₄²⁻ concentrations (Erisman and Schaap, 2004; Squizzato et al., 2012).

OPEN IN VIEWER

FIG. 4.

Linear relationship between the NH₄⁺ ions and the sum of nss-SO₄²⁻ + NO₃⁻ for each location.

The concentrations and shares of SIA in PM_2.5 were compared with secondary organic carbon concentrations and contributions. The levels of the secondary and primary organic carbon (SOC and POC, respectively) were calculated according to the methodology proposed by Castro et al. (1999): \documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland, xspace}\usepackage{amsmath, amsxtra}\usepackage{upgreek}\pagestyle{empty}\DeclareMathSizes{10}{9}{7}{6}\begin{document} \begin{align*} { \rm{POC = EC \cdot}}{ \left( {{ \rm{OC / EC}}} \right) _{{ \rm{pri}}}} \end{align*} \end{document} \documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland, xspace}\usepackage{amsmath, amsxtra}\usepackage{upgreek}\pagestyle{empty}\DeclareMathSizes{10}{9}{7}{6}\begin{document} \begin{align*} { \rm{SOC = OC - POC}} \end{align*} \end{document}

where (OC/EC)_pri is the ratio of primary OC to EC. To estimate (OC/EC)_pri, the least-squares regression was performed for 10% of the samples with the lowest 24-h OC/EC ratio (Strader et al., 1999) (Fig. 5). The regression slope represents the ratio of the primary OC to EC (OC/EC)_pri. The results of SOC and POC determination in PM_2.5 from selected measurement sites have already been presented in Błaszczak et al. (2016a, 2016b).

OPEN IN VIEWER

FIG. 5.

Ratio of primary OC to EC (OC/EC)pri, calculated for each location.

Average concentrations of SIA in the air in the whole measurement period were equal to 5.28 μg/m³ (Szczecin), 7.22 μg/m³ (Trzebinia), and 4.30 μg/m³ (Złoty Potok). The concentrations of SIA as well as of individual secondary ions recorded during the heating period were high compared with the nonheating period (Table 2), contrary to the results recorded at different measurement sites in Europe, for example in Elche (southeastern Spain) (Galindo et al., 2013) and in Thessaloniki (Greece) (Tolis et al., 2015).

The seasonal variability of the secondary ions concentrations observed in this work was probably caused by the enhanced intensity of the local emission sources during the heating period (emission of gaseous precursors of SIA) and atmospheric conditions that were unfavorable to dispersion and favorable to the formation of secondary inorganic compounds. The obtained results also indicate the dominant role of fossil fuel combustion sources and wood burning in the affecting of SIA and PM_2.5 concentrations in the air of considered sites, similar to other areas in Poland (Rogula-Kozłowska and Klejnowski, 2013; Rogula-Kozłowska et al., 2014a, 2014b; Reizer and Juda-Rezler, 2016).

Average daily shares of SIA in the PM_2.5 mass varied within a wide range (Fig. 6): from 12.93% to 54.18% in the urban background site in Szczecin, from 10.44% to 62.73% in the urban background site in Trzebinia, and from 8.33% to 32.12% in the regional background site in Złoty Potok. The share of SIA in PM_2.5 was generally higher than the share of the SOC (Błaszczak et al., 2016a); the exception is the heating period at the urban background station in Trzebinia (the average share of SOC: 33.78%, and the share of SIA: 26.54%).

OPEN IN VIEWER

FIG. 6.

Daily average contribution of nss-SO₄²⁻, NO₃⁻, and NH₄⁺ ions and SOC in PM2.5 recorded in Szczecin (a), Trzebinia (b), and Złoty Potok (c). (SOC, secondary organic carbon; the method of calculation is presented in Błaszczak et al., 2016b).

It should be noted that both concentrations and contributions of SIA in PM_2.5 mass were visibly higher in the case of urban background stations (Szczecin, Trzebinia) than in the regional background station (Złoty Potok), in contrast to most of the literature data (Minguillón et al., 2012; Squizzato et al., 2012; Rogula-Kozłowska and Klejnowski, 2013). This could be explained by the higher concentration of emission sources in urban areas, and hence the greater availability of gaseous precursors of SIA (Alastuey et al., 2004; Seinfeld and Pandis, 2006).

Results showed that the average shares of SIA in PM_2.5 mass were similar during the heating and nonheating seasons (Fig. 6). The biggest differences in the mean share of SIA in PM_2.5 were found in Trzebinia (26.54% [heating season] against 31.47% [nonheating season]). However, particular ions included in SIA differed with regard to their shares in PM_2.5 in all averaging periods, which can be interpreted as an expected phenomenon, according to the data previously published (Rogula-Kozłowska et al., 2013a, 2014a).

The mass shares of SO₄²⁻ ions in PM_2.5 were higher in the nonheating season; this refers to PM_2.5 in each of the three locations. It is caused by higher air temperatures, an increased amount of sunlight, and high concentrations of hydroxyl radicals (HO·) in this period, which accelerate the oxidation of SO₂ and its conversion into sulfate ions (Seinfeld and Pandis, 2006; Pay et al., 2012). Conversely, clearly higher percentages of nitrate and ammonium ions in the PM_2.5 mass were recorded during the heating season. This can be associated with the occurrence of low air temperatures in winter and more stable weather conditions that favor the formation of ammonium nitrate (Squizzato et al., 2012).

Comparing the obtained results with data collected in other rural and urban background sites in Europe (Squizzato et al., 2012; Mirante et al., 2014; Moroni et al., 2015), it was found that the share of SIA in PM_2.5 in the studied areas is lower or comparable, whereas the share of SOC is relatively high. In the latter case, it stems from the fuel consumption structure typical for Poland, with the dominance of solid fossil fuels and the extensive use of biomass in the low-efficient household furnaces. Consequently, combustion favors emissions of organic carbon-rich substances (Pastuszka et al., 2010; Juda-Rezler et al., 2011; Rogula-Kozłowska et al., 2014a).

Relationship between PM_2.5, ions in PM_2.5, and meteorological parameters

PM_2.5 concentrations showed strong positive correlations with OC, EC, and SIA (Table 3). Significant positive correlations between the concentration of PM_2.5 and the concentrations of particular ions in PM_2.5, except from Ca²⁺ and Mg²⁺, were observed in Złoty Potok, where the concentration of these ions was abnormally low. The concentration of NH₄⁺ was significantly correlated with the concentrations of NO₃⁻, nss-SO₄²⁻, and K⁺. The lack of a significant correlation between the concentrations of NO₃⁻ and nss-SO₄²⁻ at the site in Trzebinia, as well as the poor correlation of these ions in the case of Szczecin, suggests that precursors of the two species were released from different emission sources (Huang et al., 2011). Besides, it would be an effect of photochemical reactions in the atmosphere and temperature dependence, especially in the case of nitrates (Alastuey et al., 2004; Squizzato et al., 2012).

Considering the two urban stations, NO₃⁻ revealed a strong correlation with PM_2.5, EC, and OC, contrary to nss-SO₄²⁻ (Table 3). A different situation was observed for the regional background station in Złoty Potok. NO₃⁻ and nss-SO₄²⁻ were relatively highly correlated, with the coefficient of 0.64, and they displayed a similar correlation coefficient with PM_2.5, EC, and OC. It may be due to the significant content of aged particles in the air, which is typical for this type of areas, where the variations in nss-SO₄²⁻ and NO₃⁻ associated with the regional pollution are more readily observed (Chan and Mozurkewich, 2007).

Results also showed that NO₃⁻ ions were strongly correlated with Cl⁻ ions, especially in the case of urban background stations (Table 3). This may indicate a common origin of these ions from the combustion processes in the open air, waste incineration, or wood combustion (Rogula-Kozłowska et al., 2014b). Cl⁻ was also strongly correlated with K⁺, suggesting their presence in the air in the form of KCl. The strong correlation between Cl⁻ and K⁺ may also imply that power generation based on coal combustion as well as biomass burning are sources of both ions (Vasconcellos et al., 2007; Rogula-Kozłowska et al., 2013b; Pant et al., 2015). In Złoty Potok, K⁺ and Cl⁻ were poorly correlated; however, a strong correlation between K⁺ and nss-SO₄²⁻ was found, which may point to biomass burning as an emission source of PM_2.5.

Interestingly, a poor correlation between Cl⁻ and Na⁺ in PM_2.5 collected in Szczecin was found; therefore, the presence of these ions is related to sources other than sea salt, such as combustion processes (Rogula-Kozłowska and Klejnowski, 2013; Pant et al., 2015; Błaszczak et al., 2016b). Finally, the negative correlation of Mg²⁺ and Ca²⁺ with other PM_2.5 constituents was probably due to the very low concentrations of these two ions, often not exceeding their detection limits.

Concentrations of PM_2.5 and the associated main components exhibited a significant, generally negative correlation with wind speed (Table 3). Both OC and EC, as well as all the analyzed ions, showed a negative correlation with air temperature; the strongest one concerns the NO₃⁻ ions (correlation coefficient r ≥ −0.80).

The analyzed PM_2.5 components showed positive correlations with relative humidity, but this relationship was not as strong as in the case of air temperature. Low temperatures and wind speeds favor the phenomenon of the inversion temperature formation, stable atmospheric conditions, and low thickness of the mixing layer, which leads to the accumulation of pollutants in the atmosphere (Deshmukh et al., 2010; Pateraki et al., 2014; Blaszczak et al., 2016b). Moreover, higher relative humidity in winter accelerates the formation of secondary compounds from the gaseous precursors (Deshmukh et al., 2010), thus worsening air pollution levels.

Correlation coefficients between PM_2.5 variables and meteorological parameters were also checked separately for heating and nonheating periods. Not surprisingly, during the heating period, meteorological parameters generally displayed the negative correlation with PM_2.5 concentrations and concentrations of its chemical constituents. During the nonheating period, the PM_2.5 variables analyzed within the study were much less correlated with meteorological parameters, and the values of the correlation coefficients were often statistically insignificant and generally lower than r = (±) 0.55.

It may be suggested that in the nonheating period the impact of other factors, that is, long-range transport sources, would have a great influence on concentrations of PM_2.5 and its constituents (Galindo et al., 2011; Mancilla et al., 2015). The impact of meteorological conditions is crucial, especially in winter when the low height of the mixing layer and frequent temperature inversions prevent propagation of pollutants in the atmosphere (Juda-Rezler et al., 2011; Rogula-Kozłowska et al., 2014a; Reizer and Juda-Rezler, 2016).

It is worth noting that the wind speed and direction are the basic parameters that allow for some estimation regarding the possible sources of emissions of particulate matter on a local scale (Fleming et al., 2012). The examination of wind-rose plots for daily samples is an approach commonly used in many scientific publications (e.g., Viana et al., 2006; Xiao et al., 2012; Majewski and Rogula-Kozłowska, 2016). However, it should be kept in mind that wind speed and direction are very sensitive parameters that may present considerable daily variability. Therefore, the assessment of possible source regions of PM_2.5 in considered areas based on the wind direction data has only a preliminary nature.

Concentration roses created for the SIA (Fig. 7) showed that during the heating season, high SIA concentrations in the air above the analyzed measurement sites were primarily related to the winds blowing from the north-eastern directions, that is NNE, NE, NEE, and in addition from the E, N, and SEE. In Szczecin, the likely source areas of the highest concentrations of SIA were the northern and eastern parts of the West-Pomeranian Province. In the case of Trzebinia, the highest concentrations of SIA would be related to the inflow of air masses from the north-eastern part of the Lesser Poland Province, as well as from Świętokrzyskie Voivodship. In Złoty Potok, the probable source regions of the highest SIA concentrations would be the Łódzkie and Świętokrzyskie Voivodships, and the northern parts of the Silesian Province.

OPEN IN VIEWER

FIG. 7.

Concentration roses of secondary inorganic aerosol for Szczecin (a), Trzebinia (b), and Złoty Potok (c).

A different situation was observed during the nonheating season. The highest concentrations of SIA in the air appeared most commonly during the inflow of air masses from the western directions, namely the NNW (Szczecin), SW and SWW (Trzebinia), and SWW, NW, and NNW (Złoty Potok).

For Trzebinia, the probable source area of the highest SIA concentrations would be the western and southern parts of the Silesian Province, as well as the southern and south-eastern parts of the Lesser Poland Province. In Złoty Potok, the likely source regions of the highest SIA concentrations would be, in general, various parts of the Silesia Province, as well as the southern parts of Łódzkie Voivodship. In the case of Szczecin, the highest concentrations of SIA would be related to the inflow of air masses from the northern parts of the city and from the West-Pomeranian Province, as well as from the Baltic Sea and from northern Germany.