**Characterization and Source Identification of Elements and Water-Soluble Ions in Submicrometre Aerosols in Brno and Šlapanice (Czech Republic)**

**Pavel Mikuška 1,\* , Martin Vojt ˇešek <sup>1</sup> , Kamil Kˇr ˚umal <sup>1</sup> , Martina Mikušková-Campulov ˇ á 2 , Jaroslav Michálek <sup>3</sup> and Zbyn ˇek Veˇceˇra <sup>1</sup>**


Received: 28 May 2020; Accepted: 24 June 2020; Published: 29 June 2020

**Abstract:** Submicrometre aerosol particles (particulate matter, PM1) were collected in two Czech cities (Brno and Šlapanice) during week campaigns in winter and summer of 2009 and 2010. The aerosols were analysed for 14 elements and 12 water-soluble ions using inductively coupled plasma–mass spectrometry and ion chromatography techniques. The average PM<sup>1</sup> mass concentration was 14.4 and 20.4 µg m−<sup>3</sup> in Brno and Šlapanice, respectively. Most of the analysed elements and ions exhibit distinct seasonal variability with higher concentrations in winter in comparison to summer. The determined elements and ions together accounted for about 29% of total PM<sup>1</sup> mass, ranging between 16% and 44%. Ion species were the most abundant components in collected aerosols, accounting for 27.2% of mass of PM<sup>1</sup> aerosols, and elements accounted for 1.8% of mass of PM<sup>1</sup> aerosols. One-day backward trajectories were calculated using the Hysplit model to analyse air masses transported towards the sampling sites. The Pearson correlation coefficients between individual PM<sup>1</sup> components and PM<sup>1</sup> mass and air temperature were calculated. To identify the main aerosol sources, factor analysis was applied. Six factors were identified for each locality. The following sources of PM<sup>1</sup> particles were identified in Brno: a municipal incinerator, vehicle exhausts, secondary sulphate, a cement factory, industry and biomass burning. The identified sources in Šlapanice were as follows: a combustion source, coal combustion, a cement factory, a municipal incinerator, vehicle exhausts and industry.

**Keywords:** PM<sup>1</sup> aerosol; elements; water-soluble ions; factor analysis; source apportionment

#### **1. Introduction**

Atmospheric aerosol (particulate matter, PM) is an important airborne component with various environmental and health effects. Aerosols can deteriorate air quality, affect global climate, reduce visibility, and are involved in smog production [1]. Epidemiological studies showed an association between the concentration of aerosol particles in ambient air and adverse health effects [2–4]. The environmental and health effects of atmospheric aerosols depend on the particle size, shape and the chemical composition of particles. The atmospheric aerosol consists of a complex mixture of components including carbonaceous species (elemental and organic carbons), inorganic ions and elements in variable amounts, depending on their location and emission sources.

In the last decade, air pollution by atmospheric aerosols has also been a subject to increased interest in the Czech Republic. Numerous studies have reported the chemical composition of PM in the Czech Republic, focusing on PM<sup>10</sup> [5–10], PM2.5 particles [6,7,11–17] and also on PM<sup>1</sup> particles [6,8,10,11,18–24] that have greater toxicity compared to PM2.5 and PM10. Due to their small sizes, PM<sup>1</sup> particles can penetrate deep into the alveolar part of the lungs and cause respiratory and cardiovascular diseases [4]. Moreover, ultrafine particles (i.e., particles smaller than 100 nm) are even able to translocate from the lungs into secondary organs [25,26].

A number of individual studies have reported the aerosol chemical speciation and PM levels at specific locations in the Czech Republic, such as Prague [6,7,20], Mladá Boleslav [5,23], Brno [11,15–19,21], background site in Košetice [9,14,27] or the heavily polluted Ostrava region [8,10,12,13,22,24]. However, unlike other sites in the Czech Republic, where all components of PM (i.e., organic compounds, elements or ions) are studied, the studies dealing with aerosol composition in Brno focus almost exclusively on organic compounds [11,15,17–19], whereas particulate ions [16] and elements [21,28] have been measured in PM in Brno only marginally and information on the content of ions and elements in aerosols in this area is still incomplete.

Receptor models based on the statistical evaluation of PM chemical data acquired at receptor sites are used to identify emission sources of aerosols [29]. The chemical mass balance model assumes knowledge of the composition of the emissions for all relevant sources, however, fulfilling this requirement is often problematic [29]. More widespread are two multivariate models, principal component analysis [30–33] and positive matrix factorisation [5,8,22,24,34–36], which apportion the sources on the basis of the ambient data from the receptor site alone.

The aims of the study were to fill the gap in the missing information about the content of elements and water-soluble ions in aerosols in the Brno agglomeration; to obtain comprehensive information on the composition of PM<sup>1</sup> aerosols in Brno and Šlapanice in a combination with other studies in the same area focused on the characterization of organic compounds in PM<sup>1</sup> aerosols in Brno and Šlapanice [11,18,19]; to identify their sources. The paper presents the results of determination of elements and water-soluble ions in submicrometre aerosols (PM1, particles with aerodynamic diameter smaller than 1 µm) collected in two cities in the Czech Republic (Brno and Šlapanice) in Central Europe. The seasonal differences (winter vs. summer) were evaluated and the sources of studied aerosol components were analysed.

#### **2. Experiments**

#### *2.1. Sampling Sites*

The PM<sup>1</sup> aerosols were sampled in Brno and Šlapanice that represent a large city and a small town in the Czech Republic. Brno, the second largest city in the Czech Republic (370,000 inhabitants), is an industrial and administration centre of Moravia, the eastern part of the Czech Republic. There are various local sources of aerosols in Brno, such as traffic (cars and trams on Veveˇrí street), residential heating and large emission sources, such as heating plant, a municipal waste incinerator (on the eastern outskirts of Brno) and industry, including a foundry plant. Regional sources comprise mainly residential heating in the surrounding villages and a cement factory east of Brno. Šlapanice, a small town (6000 inhabitants), is located 3 km southeast from Brno. Overall, the sources of aerosols in Šlapanice are similar to those in Brno, but there is a difference between the composition of local and regional sources in both locations. Local sources of aerosols in Šlapanice include traffic, residential heating, small industrial factories and brickworks. Aerosols can be transported to Šlapanice from various regional sources, such as the nearby motorway between Brno and Šlapanice, Brno airport southwest of Šlapanice, a municipal waste incinerator northwest of Šlapanice, residential heating in the surrounding villages, a cement factory northeast of Šlapanice, etc. In addition, a large power plant in Hodonín that burns coal and biomass is located about 50 km southeast from Brno and Šlapanice. Moreover, a long-range transport of pollutants from distant areas or neighbouring countries to Brno and Šlapanice cannot be ignored.

#### *2.2. Aerosol Sampling*

Atmospheric aerosols in the size fraction of PM<sup>1</sup> were sampled for 24 h every day over one week in winter and one week in summer of 2009 and 2010 in Brno and Šlapanice (Figure 1) to compare the PM<sup>1</sup> composition in the large city and a nearby small town. The sampling of aerosols began every day at 9:00 a.m. Aerosol samples in Brno were collected in an urban locality on the balcony on the first floor (at the height of 8.9 m above ground level and at the distance of 15.6 m from the street Veveˇrí) of the Institute of Analytical Chemistry facing northeast toward the street Veveˇrí (49◦12'28.27´´N and 16◦35´28.00´´E). In Šlapanice the aerosols were collected in small urban locality in the garden of a family house (49◦09´55.92´´N and 16◦43´26.18´´E). Sampling locations are located relatively in the centre of both Brno and Šlapanice, and therefore, the influence of different PM sources is expected, although the sampling site in Brno is situated near the street with traffic. − −

ř ř

**Figure 1.** Location of sampling sites (Brno and Šlapanice) on a map of the Czech Republic.

− − Submicrometre aerosols were collected at each site in parallel using a high-volume (HV) and a low-volume (LV) sampler. The HV sampler (DHA-80, Digitel, 30 m<sup>3</sup> h −1 ) equipped with a PM<sup>1</sup> size selective impaction inlet (model DPM01/30/00, Digitel) collected PM<sup>1</sup> aerosols on cellulose nitrate filters (150 mm diameter, porosity 3 µm, Sartorius). A total number of 52 samples (24 samples from Brno and 28 samples from Šlapanice) were collected with HV sampler during all campaigns. The LV sampler (1 m<sup>3</sup> h −1 ), consisting of a Teflon coated aluminium cyclone inlet (cut point diameter of 1 µm, model URG-2000-30EHB), and a NILU filter unit (type 9633) collected PM<sup>1</sup> aerosols on 47mm Teflon filters (Zefluor, porosity 1 µm, PALL). To eliminate interference of gaseous pollutants, such as SO2, HNO3, NH<sup>3</sup> and others, an annular diffusion denuder [37] was placed between the cyclone and Teflon filter. A total number of 56 samples (28 samples from Brno and 28 samples from Šlapanice) were collected with the LV sampler during all campaigns.

− Meteorological parameters (i.e., temperature and relative humidity) measured by means of a commercial sensor (type T3113, Comet Systems) are given in Table 1. We also present predominant wind directions obtained from one-day backward trajectories calculated using the Hysplit v5.0.0 model. All trajectories are shown in the Supplementary Material (Figures S1–S8) separately for each location and campaign.


**Table 1.** Meteorological parameters in Brno and Šlapanice during campaigns.

#### *2.3. Processing and Filter Analysis*

Mass concentrations of collected aerosols were determined by weighing filters, using a microbalance M5P (±1 µg; Sartorius). Filters were equilibrated before weighing in an air-conditioned room under constant conditions for 48 hrs (temperature 20 ± 1 ◦C, relative humidity 50 ± 3%). Static electricity was eliminated with an ionizer prior to weighing (PRX-U, Haug). After weighing, exposed filters were cut, using ceramic scissors, into two equal pieces; each of them was weighted again.

One half of the cellulose-nitrate filters was digested in 4 mL of sub-boiling nitric acid in the UniClever microwave device (Plazmatronika). The decomposed samples were transferred quantitatively along with 4 mL of deionized water into polyethylene scintillation vials (Kartel). The extracts were analysed for the content of 14 selected elements (Al, K, Ca, Fe, Mn, Zn, Cu, Cd, Ba, As, Pb, V, Ni, Sb), employing an inductively coupled plasma–mass spectrometry (model 7500 CE, Agilent). Relative uncertainty of element analysis was in the range of 1 to 3%.

Both halves of the Teflon filters were extracted in 8 mL of deionized water under ultrasonic agitation. The extract of the first half was analysed for the content of seven anions (fluoride, chloride, nitrite, nitrate, sulphate, oxalate, phosphate), while the second half of the filter was analysed for five cations (Na+, K+, NH<sup>4</sup> <sup>+</sup>, Ca2+, Mg2+) by means of ion chromatography (ICS-2100, Dionex). Relative uncertainty of water-soluble ion analysis was in the range of 4 to 7%.

#### *2.4. Calculation of Air Trajectories*

The analysis of air mass transported towards the sampling sites was performed using the Hysplit model v5.0.0 [38,39], by calculation of one-day backward trajectories at 300, 750 and 1500 m above ground level.

#### *2.5. Factor Analysis Model*

Measured aerosol components can be grouped by their correlations. The components within a particular group are highly correlated among themselves but have relatively small correlations with aerosol components in a different group. In the factor analysis presented in this paper, it is assumed the observed correlations in a given group of aerosol components are influenced by a factor, which for given group corresponds to a source of pollution (such a pollution source can be, for example, vehicle exhausts, coal combustion and others). Factors are unmeasurable and the number of factors is much smaller than the number of aerosol components. The correlation relationship among aerosol components can be described in terms of a few underlying factors.

Suppose the observed concentrations *X*1, . . . , *X<sup>p</sup>* of elements and ions in PM particles with means (expectations) µ1, . . . , µ*p*. In our study *p* = 16. Let *F*1, . . . , *Fm*, (*m* < *p*) be considered common factors corresponding to the sources of pollution. We then consider the linear model describing how the concentrations of elements and ions *X*1, . . . , *X<sup>p</sup>* can be explained using linear combinations of these common factors *F*1, . . . , *Fm*. The model can be described by equation

$$X\_{i} - \mu\_{1} = l\_{1i} F\_{1} + \dots \; + l\_{1i} F\_{m} + \varepsilon\_{i}, i = 1, \dots, p \tag{1}$$

where *lij*, *i* = 1, . . . , *p*, *j* = 1, . . . , *m*, are the coefficients of linear combinations. They represent the so-called factor loadings indicating how strong is the statistical association between the i-th aerosol component *X<sup>i</sup>* and the j-th factor *F<sup>j</sup>* . The higher the absolute value of *lij* is, the higher is the statistical association between the aerosol component *X<sup>i</sup>* and the factor *F<sup>j</sup>* . The term ε*<sup>i</sup>* stands for random errors. Random errors are supposed to be independent with zero mean and variance *Var(*ε*<sup>i</sup> )* = ψ*<sup>i</sup>* , and common factors are assumed to have a zero mean and unit variance. The independence of common factors and random errors (*Cor(*ε*i, Fj)* = *0 for i* = *1,* . . . *, p, j* = *1,* . . . *, m*) is also considered. From the above assumptions follows Var(*X*<sup>i</sup> ) = *l*i1 <sup>2</sup> + *l*i2 <sup>2</sup> + . . . + *l*im <sup>2</sup> + ψ<sup>i</sup> = *h*<sup>i</sup> <sup>2</sup> + ψ<sup>i</sup> , where *h*<sup>i</sup> <sup>2</sup> = *l*i1 <sup>2</sup> + *l*i2 <sup>2</sup> + . . . + *l*im 2 is called the i-th communality, the proportion of the variance of the variable *X*<sup>i</sup> contributed by the *m* common factors.

Various methods [40] that allow us to estimate the parameters of model(1) exist. Here, the principal component method [41], which does not impose restrictions on distribution of analysed variables, is preferred. A crucial task of the factor analysis is the selection of appropriate number of common factors *m*. Since the higher values of *m* lead to common factors that are difficult to interpret, the aim is to choose the smallest possible value *m*, such that a sufficient proportion of variability in the data is explained. For this problem, Kaiser´s criterion and scree plot [41] are considered.

The original factor loads obtained by the principal component method can be difficult to interpret. Thus, the factor rotation is performed for better interpretation of factor loads. A common goal of rotation in this paper is to ensure that each aerosol component loads highly on a single factor and has small-to-moderate loads on the remaining single factors. The factor model with such rotated loads is called the model with a simpler structure. Usually the criterion, which is used to obtain a simpler structure, is the varimax criterion [41].

#### **3. Results**

#### *3.1. Characterization of Submicrometre Aerosols*

Mass concentrations of PM<sup>1</sup> aerosols collected at both sites during all campaigns are given in Table 2. The average measured PM<sup>1</sup> mass concentration was 19.1 µg m−<sup>3</sup> in winter and 9.65 µg m−<sup>3</sup> in summer in Brno, and 30.8 and 10.0 µg m−<sup>3</sup> in winter and in summer, respectively, in Šlapanice. PM<sup>1</sup> samples in both locations were collected at a different time; therefore, we tested if the mass concentrations of PM<sup>1</sup> aerosols in Brno and Šlapanice were similar in the same season. Agreement of mass concentration of PM<sup>1</sup> aerosols collected in Brno and Šlapanice was statistically tested by non-paired t-test for equal means. Probability of equality of mean concentrations (P) smaller than 0.05 indicates disagreement of tested mean values at both sampling sites. P value was 0.0081 for winter 2009; 0.1366 for summer 2009; 0.1339 for winter 2010 and 0.1399 for summer 2010, which indicates that the mean mass concentrations of aerosols in Brno and Šlapanice in the same season were similar in summer 2009, winter 2010 and summer 2010 but different in winter 2009. The difference in the PM<sup>1</sup> concentration in Brno and Šlapanice in winter 2009 was caused by heavy snow during days 2–5 of the campaign in Brno leading to lower concentration of PM<sup>1</sup> in Brno, while in Šlapanice (no snow during the campaign), the PM<sup>1</sup> concentration remained at a normal level for this time of a year. The concentrations of aerosols in winter were higher than in summer both in Brno and Šlapanice, which can be attributed to increased anthropogenic emissions in heating season due to household heating of local residents [11,18,19] and to lower mixing layer height due to the lower dispersive capacity of the lower atmospheric layers in winter, which favours the accumulation of pollutants and prevents air convention and the dispersion of pollutants [42,43]. The strong relationship of the PM<sup>1</sup> concentration with air temperature was confirmed by a significant negative correlation (*p* < 0.01) both

in Šlapanice (a correlation coefficient R = −0.71, Table S1) and in Brno (−0.64, Table S2). Calculated one-day backward trajectories (Figures S1–S8) indicate possible transport of not only regional air pollution from surrounding villages but also long-range transport of polluted air from more distant localities, such as heavy polluted areas in the Silesian Voivodeship in southern Poland that has been recently identified as an important PM source in winter [12]. The differences in the concentration of PM<sup>1</sup> aerosols in corresponding seasons in 2009 and 2010 (Table 2) were probably caused mainly by different meteorological and dispersal conditions in those two years.


**Table 2.** Mass concentrations of PM<sup>1</sup> aerosols during the campaigns.

The mass of determined elements and ions together accounted for about 29% of total PM<sup>1</sup> mass, ranging between 16% and 44%. The rest of the mass probably consists of organic compounds, elemental carbon and water. Previous studies at the same locations showed that organic material and elemental carbon formed on average 37.6% and 7.80% of PM<sup>1</sup> mass [11,18,19].

The concentration of PM<sup>1</sup> aerosols found in Brno and Šlapanice in 2009 and 2010 are similar to those observed in other study from this site [11] as well as in PM<sup>1</sup> from other localities in the Czech Republic [6,23], or from other sites in Europe, such as Birmingham [44], Melpitz [45], Bologna [46], Granada [34], Katowice [47] and Racibórz, whereas the concentration of PM<sup>1</sup> in Zabrze [48] was higher than in Brno and Šlapanice.

Two recent studies [21,49] from the same location in Brno do not show any decrease in air pollution with PM<sup>1</sup> aerosol. The concentration of PM<sup>1</sup> aerosols collected in summer 2014 (i.e., the mean value of 10.8 µg/m<sup>3</sup> ) was similar to the concentration found in the presented study, while the concentration found in winter 2015 (i.e., the mean value of 16.5 µg/m<sup>3</sup> ) [21] was in the middle of the concentrations found in the present study in winter 2009 and 2010. The sampling of PM<sup>1</sup> aerosols in winter 2017 was accompanied by a few days of smog event, which resulted in an increased PM concentration in Brno and the concentrations of PM<sup>1</sup> aerosols measured in this study (i.e., the mean value of 34.2 µg/m<sup>3</sup> ) [49] was thus even higher than the concentrations found in the present study.

#### *3.2. Characterization of Elements*

We analysed 14 elements in all submicrometre aerosol samples collected in Brno and Šlapanice. The average element concentrations in PM<sup>1</sup> aerosols from both sampling sites are summarized in Table 3 (campaigns of 2009) and Table 4 (campaigns of 2010). In winter 2009, the concentrations of elements in Šlapanice were higher than those in Brno, whereas in all other seasons the concentrations of elements at both localities were comparable. The sum of concentrations of the analysed elements accounted for 1.77% of PM<sup>1</sup> mass. In winter, the contribution of elements to PM<sup>1</sup> mass was 2.17% (1.70–2.76%) and in summer decreased to 1.37% (0.92–1.97%). In winter, lead and potassium were the two most abundant elements accounting for 65–84% of the total element mass in PM1, while in summer potassium prevailed (27–40% of the total element mass in PM1). The daily changes in element concentrations in 2009 and 2010 are shown in Figures 2 and 3, respectively. As, Cd, Pb and Ni are elements known to be toxic to human health [50,51]. The annual average concentration of these

elements, calculated as the average of summer and winter concentrations, do not exceed the annual limit valid in the Czech Republic [52] in both Brno and in Šlapanice in 2009 and 2010.


**Table 3.** Summary of concentrations of elements and water-soluble ions (ng m−<sup>3</sup> ) in PM<sup>1</sup> aerosols collected during the winter and summer campaigns in Brno and Šlapanice in 2009.


**Table 3.** *Cont.*

nd—not detected.

**Table 4.** Summary of concentrations of elements and water-soluble ions (ng m−<sup>3</sup> ) in PM<sup>1</sup> aerosols collected during winter and summer campaigns in Brno and Šlapanice in 2010.



**Table 4.** *Cont*.

nd—not detected.

The concentrations of elements in the PM<sup>1</sup> samples from both the Brno and Šlapanice sites are comparable with those found in Granada [34], Bologna [46] and Frankfurt [53], but lower than those in samples from other European sites, such as Katowice [47] or Tito Scalo [54]. In other Polish cities, Zabrze and Racibórz, the concentrations of several elements (Al, Mn, Fe, Cu, Zn, As) in PM<sup>1</sup> were higher than those in Brno and Šlapanice, while the concentrations of other elements were comparable [48].

A comparison of the concentrations of elements in PM<sup>1</sup> aerosols collected in Brno in 2009 and 2010 with the concentrations of elements collected at the same location in summer 2014 and winter 2015 [21] does not show a marked drop in air pollution with heavy metals in Brno location during this period. Most of the elements measured simultaneously in this and in a later study (i.e., Fe, Mn, V, Ni, Cu, Zn, Cd) show approximately similar concentrations, with the exception of lead, as the concentrations of which in winter 2015 were much lower than the concentrations found in the corresponding season in 2009 and 2010. Similarly, the concentration of lead in summer of 2014 was much lower than in summer of 2009, but practically identical with the concentrations in summer of 2010.

**Figure 2.** Daily variations in the concentrations of elements in the PM<sup>1</sup> aerosols in Brno and Šlapanice in 2009.

**Figure 3.** Daily variations in the concentrations of elements in the PM<sup>1</sup> aerosols in Brno and Šlapanice in 2010.

The enrichment factors (EFs) of elements in aerosols collected in Brno and Šlapanice were calculated to discriminate the anthropogenic and crustal (i.e., natural) origin of the studied elements. The EFs are defined by the equation

$$\text{EF} = \text{(\text{\textdegree\textdegree\textdegree\textdegree\textdegree\textdegree\textdegree\textdegree\textdegree\textdegree\textdegree\textdegree\textdegree\textdegree\textdegree\textdegree\textdegree\textdegree\textdegree\textdegree\textdegree\textdegree\textmathbf$$

where X represents the considered element and R is the reference element, while the subscripts aerosol and crust indicate concentrations in PM<sup>1</sup> and in the Earth's crust, respectively [55]. The EFs were calculated for Fe as the reference element [21]. Generally, EF < 5 indicates the crustal soil as the predominant source of the element, while the anthropogenic origin of the element may be considered at EF > 100 [56]. The enrichment factors calculated from the element concentrations determined in PM<sup>1</sup> aerosols collected in Brno and Šlapanice are shown in Table 5. Pb, Cd and Sb have the highest EFs, with Pb being the most enriched element for PM<sup>1</sup> particles both in Brno and Šlapanice followed by Cd, Sb, Zn and As with EF > 100, which indicates their anthropogenic sources. The EFs of Cu, Ba, Ni, K, Mn and V are within the range of 5–100, which suggests that these elements are of both anthropogenic origin and soil contribution. The EF of Al approaches to unity, which indicates crustal soil as a predominant source [54,57]. The EFs of most elements in the aerosols sampled in Šlapanice are larger than enrichment factors of elements in aerosols collected in Brno. The obtained value of the EFs from both localities are similar to those of the recent study in Brno [21], with the exception of V, Mn and Cd, the values of which in this study are approximately twice higher, and the EF of Pb is approximately 20 times higher.

**Table 5.** Enrichment factors calculated from the element concentrations determined in PM<sup>1</sup> aerosols collected in Brno and Šlapanice.


Most of the elements exhibit distinct seasonal variability when the concentrations of elements in winter were higher than in summer. High seasonal differences were observed especially for K, Pb, Fe, As, V, Cd, Zn, Ni, Sb and Ba, and less for Mn and Cu. Potassium is produced largely during wood combustion [58]. Other metals can originate from several different sources. Arsenic and cadmium originate mainly from coal combustion and partly from industry. Zn, Mn, Cu, Fe or Pb are produced, next to various industrial sources and traffic, mainly by biomass (wood) burning and coal combustion [59–62]. Coal combustion and biomass burning are still frequently used for energy production and for residential heating in many European countries [36,63]. Hence, it is plausible that increased concentration of these elements (with exception of Al and Ca) in heating seasons when compared to the rest of the year is largely caused by wood and coal combustion in residential heating. Recent studies [11,15,18,19] and the results from the last census in the Czech Republic in 2011 [64] indicate that the combustion of coal and wood is used for heating only in a small part of households both in Brno (0.46%) and in Šlapanice (1.36%), but the proportion of households in small villages near Šlapanice and Brno using wood or coal for heating is much higher (up to 11%). In addition, elements, such as Cd, As, Mn, Ni, Pb, V, Cu, Zn, Sb and so on, can also originate from emission of a municipal solid waste incinerator [65,66] located directly on the connecting line between the sampling site in Brno and Šlapanice, or from the incineration of waste in households (Sb, Cu, Pb, Sn, Ti, and Zn) [67]. Another possible source of elements, in particular Ca, Zn, Fe, Mn, Pb, Cd, As, Cu and Ni, is a cement plant [68–70] located north of Šlapanice and east of Brno. To verify this hypothesis, we calculated one-day backward trajectories for all the sampling campaigns in both the Brno and Šlapanice sampling sites (Figures S1–S8). Detailed analysis revealed the transport of air masses from all directions, although during individual campaigns certain wind directions prevailed (Table 1). Moreover, the backward trajectory analysis shows the possibility of a long-range transport of air masses to Brno and Šlapanice from areas as far as several hundred kilometres.

#### *3.3. Characterization of Water-Soluble Ions*

We analysed 12 water-soluble ions in total (i.e., 5 inorganic cations, 6 inorganic anions and oxalate) in all submicrometre aerosol samples collected in Brno and Šlapanice. Average water-soluble ion concentrations in PM<sup>1</sup> aerosols from both Brno and Šlapanice are summarized in Table 3 (campaigns of 2009) and Table 4 (campaigns of 2010). Ion species were important constituents of submicrometre aerosols both in Brno and Šlapanice, accounting for 27.5% and 26.8% of PM<sup>1</sup> mass in Brno and Šlapanice, respectively. Ion species on average accounted for 27.2% of mass of PM<sup>1</sup> aerosols. Ammonium, nitrate and sulphate were the three major ion species, contributing 32.3%, 36.2% and 23.4% of the total

ion concentrations, respectively. SO<sup>4</sup> <sup>2</sup>−, NO<sup>3</sup> <sup>−</sup> and NH<sup>4</sup> <sup>+</sup> are generally considered as secondary inorganic aerosol components (SIA; [71]). They derive from gas to particle conversion processes when SO<sup>2</sup> is transformed to H2SO<sup>4</sup> and nitrogen oxides to HNO3, followed by NH<sup>3</sup> neutralization to form (NH4)2SO<sup>4</sup> and NH4NO3. SO<sup>2</sup> and NO<sup>X</sup> are products of combustion processes, whereas NH<sup>3</sup> originates mainly from anthropogenic sources, such as agricultural activity, industry and traffic [36,72]. The sum of SO<sup>4</sup> <sup>2</sup>−, NO<sup>3</sup> <sup>−</sup> and NH<sup>4</sup> <sup>+</sup> accounted together for 91.9% on average (86.1–97.5%) of total ion concentration and for 25.1% of PM<sup>1</sup> mass. The concentrations of other analysed anions and cations were much lower. The daily changes in the concentrations of ions in 2009 and 2010 are shown in Figures 4 and 5, respectively.

**Figure 4.** Daily variations in the concentrations of water-soluble ions in the PM<sup>1</sup> aerosols in Brno and Šlapanice in 2009.

**Figure 5.** Daily variations in the concentrations of water-soluble ions in the PM<sup>1</sup> aerosols in Brno and Šlapanice in 2010.

The majority of the analysed ions show significant seasonal variability with winter concentrations, significantly exceeding those in the summer (Tables 3 and 4). The SIA contribution decreased from

a share of 32.4% in PM<sup>1</sup> mass in winter to 17.8% in summer. The concentration of NO<sup>3</sup> <sup>−</sup> was much higher in winter than in summer, which may be due to the low thermal stability of ammonium nitrate in the summer, favouring the conversion of particulate ammonium nitrate to gaseous nitric acid and ammonia [1,73]. In contrast, the difference between the winter and summer SO<sup>4</sup> <sup>2</sup><sup>−</sup> concentrations was much smaller compared to that of NO<sup>3</sup> <sup>−</sup>, which indicates an active photochemical production of SO<sup>4</sup> 2− in summer [74].

Increased concentrations of F−, Cl−, NO<sup>2</sup> <sup>−</sup>, SO<sup>4</sup> <sup>2</sup>−, PO<sup>4</sup> <sup>3</sup>−, NO<sup>3</sup> <sup>−</sup>, NH<sup>4</sup> <sup>+</sup> and K<sup>+</sup> in winter are mostly associated with burning of wood and coal in residential heating [11,18,19,36]. Combustion processes are the main sources of fluoride (i.e., coal), chloride (i.e., wood, coal, solid waste) and phosphate (i.e., coal, wood, traffic). K+, considered an inorganic tracer for biomass burning [14,62], formed the majority (i.e., 73.9%) of the total potassium in submicrometre aerosol. Enhanced winter concentrations of Cl−, F−, PO<sup>4</sup> <sup>3</sup><sup>−</sup> and K<sup>+</sup> were observed especially in Šlapanice, which is likely associated with local and regional combustion of wood and coal in residential heating [11,18,19]. Moreover, Cl<sup>−</sup> and F<sup>−</sup> may also be emitted from the brickworks [75,76] located directly in Šlapanice. Cl<sup>−</sup> and PO<sup>4</sup> <sup>3</sup><sup>−</sup> serving as markers of plastic waste combustion may also originate from a large municipal waste incinerator located east of Brno or from the combustion of solid waste by households [67].

The concentrations of oxalate in winter were similar to those during summer campaigns, with the exception of Šlapanice in 2010. Oxalates, considered a major water-soluble organic compound in the aerosols, were reported as both a product of primary emissions from combustion processes (traffic, biomass burning, biogenic activity) and as a secondary product of atmospheric chemistry [77,78]. Oxalate correlated both with sulphate and K<sup>+</sup> (Tables S1 and S2), which suggests that oxalate found in PM<sup>1</sup> aerosols in Brno and Šlapanice originated from both biomass burning and secondary oxidation.

The concentration of ions in PM<sup>1</sup> in Brno and Šlapanice were comparable with those found in Granada [34] and Katowice [47], while in other Polish cities, Zabrze and Racibórz, the concentrations of several ions (Cl−, K+, Na+, Ca2+, Mg2+) were higher and others (NH<sup>4</sup> <sup>+</sup>) were lower [48]. The concentrations of ions at background station in Melpitz were comparable with the exception of a higher concentration of Mg2<sup>+</sup> and a lower concentration of Ca2<sup>+</sup> in comparison with results of this study [45]. A recent study from Prague reported generally higher concentrations of the analysed Cl−, SO<sup>4</sup> <sup>2</sup>−, NO<sup>3</sup> <sup>−</sup> and NH<sup>4</sup> <sup>+</sup> in comparison with those from Brno and Šlapanice [20]. Recently, Cl−, NO<sup>2</sup> <sup>−</sup>, SO<sup>4</sup> <sup>2</sup><sup>−</sup> and NO<sup>3</sup> <sup>−</sup> have been analysed in Brno in PM2.5 particles, using the continuous aerosol sampler [16]. The concentrations of Cl<sup>−</sup> and SO<sup>4</sup> <sup>2</sup><sup>−</sup> found by the aerosol sampler were comparable with filter-based concentrations in this study, but the concentrations of NO<sup>2</sup> <sup>−</sup> and NO<sup>3</sup> <sup>−</sup> from the aerosol sampler were higher. This difference can be explained by the bimodal distribution of nitrate in PM<sup>1</sup> and PM2.5 [6] and by sampling artefacts observed during the sampling of nitrite on filters [16].

#### *3.4. Correlation Analysis*

Pearson correlation coefficients provide another way to assess the sources of the analysed components in PM<sup>1</sup> aerosols. The possible sources can be identified from the correlation matrix by analysing the value, which represents the linear coefficient of correlation between the species. The high correlation of two species suggests their identical sources. The results of correlation analysis of the studied elements and water-soluble ions and their relationship to the mass concentration of PM<sup>1</sup> and temperature of air are shown in Table S1 (Šlapanice) and Table S2 (Brno), and are discussed in detail below.

Several elements (V, Cd, As, Fe, Zn, K) and most ions (NO<sup>3</sup> <sup>−</sup>, Cl−, F−, PO<sup>4</sup> <sup>3</sup>−, K<sup>+</sup> and NH<sup>4</sup> +) both in Šlapanice and Brno (*p* < 0.01) significantly positively correlated with PM1, but negatively correlated with temperature, which indicates a strong association of the mentioned elements and ions with combustion sources, especially in heating seasons.

Many elements and ions have the same overlapping sources. For example, Zn, Cu, Mn, Pb, As, Cd were found in the emissions from several combustion processes, such as coal combustion, biomass burning, an incinerator or vehicle emissions. The effect of biomass burning on the studied elements

and ions was investigated through the correlation with K+, which is used as a marker of biomass (especially wood) burning [14,62]. K<sup>+</sup> correlated significantly (*p* < 0.01) in Brno with V, Cd, As, Sb, Cu, Ni, Mn, Fe, Zn, K, NO<sup>3</sup> <sup>−</sup>, Cl−, F−, PO<sup>4</sup> <sup>3</sup>−, NO<sup>2</sup> <sup>−</sup> and NH<sup>4</sup> <sup>+</sup>, and in Šlapanice with V, Cd, Sb, Zn, K, NO<sup>3</sup> <sup>−</sup>, F−, NH<sup>4</sup> <sup>+</sup> (*p* < 0.01) and with As and PO<sup>4</sup> <sup>3</sup><sup>−</sup> (p < 0.05), which suggests that wood burning was one of the major sources of the PM<sup>1</sup> constituents during the campaigns in Brno and Šlapanice. This finding is in good agreement with the list of elements found in the emissions from the combustion of various biomass fuels [61,62]. Other elements (i.e., Zn, Cu, Cd and As) are important markers of coal combustion [60]. Significant cross-correlations among these elements (*p* < 0.01) confirm their common origin in combustion of coal. Moreover, some elements and ions originated from the municipal waste incinerator, such as Sb, Zn, Cd, As, Mn, Ni, Cl−, F−, and PO<sup>4</sup> <sup>3</sup><sup>−</sup> correlated significantly with other elements and ions in Brno more than in Šlapanice, which suggests a stronger effect of emissions from the municipal incinerator on the composition of PM<sup>1</sup> in Brno than in Šlapanice. Ni did not correlate with any other element or ion in Šlapanice, but in Brno, Ni correlated significantly with V, Mn, Fe, K, NH<sup>4</sup> <sup>+</sup>, PO<sup>4</sup> <sup>3</sup><sup>−</sup> (*p* < 0.01), Cd, NO<sup>3</sup> <sup>−</sup> and Cl<sup>−</sup> (*p* < 0.05), which indicates the municipal incinerator as a predominant source of Ni in Brno. A similar difference in the correlations between the Brno and Šlapanice was also observed for Mn. The opposite situation was observed for Pb, which highly correlated with other PM<sup>1</sup> components more in Šlapanice (i.e., Cd, Cu, Fe, Zn, K, SO<sup>4</sup> <sup>2</sup>−, NO<sup>3</sup> <sup>−</sup>, Cl−, F <sup>−</sup>, PO<sup>4</sup> <sup>3</sup>−, NH<sup>4</sup> <sup>+</sup> (*p* < 0.01), Mn, F<sup>−</sup> (*p* < 0.05) than in Brno (Ba, NO<sup>3</sup> <sup>−</sup> (*p* < 0.01), V (*p* < 0.05)). No apparent correlation with other constituents in the PM<sup>1</sup> samples collected in the Brno and Šlapanice sites was observed for Al and Ca, suggesting that Al and Ca have different sources compared to other elements and ions in PM1, such as soil dust resuspension for Al or emissions from a cement factory for Ca and Ca2+.

#### *3.5. Source Apportionment*

The correlation analysis groups the elements and ions on the basis of high pair correlations. The method provides only general information about the possible sources. Moreover, it does not evaluate the relationship between various sources, so overlapping of sources is problematic. To enhance the accuracy of emission source identification and their relative contribution, the method of factor analysis was also applied. The factor analysis belongs to multivariate statistical methods and is commonly used to identify the sources of PM aerosols [29,31,32,79].

In order to identify the aerosol emission sources, the factors were extracted by using the principal component analysis method (PCA) [41] and further rotated by using the varimax criterion, which achieves a simpler structure of the orthogonal factor model and also better interpretable factors. The optimal number of factors was estimated based on the Kaiser criterion and scree plot [41]. All computations were performed by using the software R version 3.6.1.

The principal component analysis was performed on the concentrations of elements and ions in PM<sup>1</sup> aerosols collected in Šlapanice and Brno sites. The six extracted factor loadings from PCA analysis in Šlapanice and Brno dataset are given in Tables 6 and 7, and Figures 6 and 7, respectively. Only factor loadings above 0.1 are shown and values greater than 0.5 are in bold. The last lines in the tables (Variance and Cumul. Var., %) show the proportions of the total data variance explained by each individual factor and total explained variance, respectively.


**Table 6.** Principal component analysis results of PM<sup>1</sup> components in Šlapanice.

**Table 7.** Principal component analysis results of PM<sup>1</sup> components in Brno.


**Figure 7.** Source profiles for PM<sup>1</sup> in Brno.

#### 3.5.1. Analysis for Šlapanice

Six independent sources identified in Šlapanice dataset (F1–F6 in Table 6 and Figure 6) include a combustion source (23%), coal combustion (14%), a cement factory (13%), a municipal waste incinerator (12%), vehicle exhausts (9%) and industry (5%). The identified factors together explained 76% of the total variance.

The first factor was identified as a combustion source. It is dominated by K+, K, NH<sup>4</sup> <sup>+</sup>, V and Cd, followed by NO<sup>3</sup> <sup>−</sup>, oxalate and Zn. Other identified tracers are As, Sb, Pb, PO<sup>4</sup> <sup>3</sup><sup>−</sup> and SO<sup>4</sup> <sup>2</sup>−. These can be related to biomass (mainly wood) burning and coal combustion in residential heating [11,18,19,35] or in a power plant burning coal and biomass located 50 km southeast of Šlapanice. The factor may also include local sources, such as engine vehicle emissions (Zn, Pb, Mn) [80] or emissions from brickworks built on a small hill at the north side of Šlapanice.

The second factor was identified as coal combustion. The factor was associated with high loadings of Fe, Mn, Pb and Na<sup>+</sup> and moderate loadings of Cd, As, Cu, SO<sup>4</sup> <sup>2</sup>−, Cl−, NO<sup>3</sup> <sup>−</sup> and NH<sup>4</sup> <sup>+</sup>. Coal is used for household heating to some degree in Šlapanice and, in particular, in villages around Šlapanice. Moreover, a coal-fired power plant is located in a distance of 50 km southeast of Šlapanice.

The third factor with high loadings of Ca and Ca2<sup>+</sup> may be explained mainly by emissions from a cement factory [68–70,81]. The cement factory in Mokrá is situated in a short distance, about 6.5 km northeast of Šlapanice.

The fourth factor with high loadings of Cl−, F−, PO<sup>4</sup> <sup>3</sup><sup>−</sup> and NO<sup>3</sup> <sup>−</sup> and moderate loadings of Cd, Cu, Zn, Pb, K and Na<sup>+</sup> was identified as emissions from a municipal waste incinerator [65,66]. The large municipal waste incinerator, located at the southeast side of Brno and at a distance of 5 km northwest of Šlapanice, is used for the entire agglomeration of Brno. It is also necessary to take into account a partial contribution from waste incineration in households in Šlapanice and surrounding villages [67].

The fifth factor identified as vehicle exhausts was characterized with high loadings of Sb and Cu and moderate loadings of Cd, Zn, K, Pb and K+. Cu, Sb and Zn are used as markers of vehicle-related sources [57]. They generally originate from abrasions of tires or break wear. However, Cu, Zn and Sb present in PM<sup>1</sup> aerosols, originating probably from the exhaust emissions from diesel and petrol engines as Cu, Zn and Sb, along with other elements, are present directly in the fuel or lubricating oils [36,57,80,82].

The sixth factor with moderate loadings of Ni, Ba, Pb, SO<sup>4</sup> <sup>2</sup><sup>−</sup> and NO<sup>3</sup> <sup>−</sup>, represented emissions from industry. Several small production workshops dealing with the processing of metals are located on the outskirts of Šlapanice or nearby.

#### 3.5.2. Analysis for Brno

Six independent sources identified in the Brno dataset (F1-F6 in Table 7 and Figure 7) included a municipal waste incinerator (23%), vehicle exhausts (18%), secondary sulphate (13%), a cement factory (10%), industry (8%) and biomass burning (7%). Identified factors together explained 79% of the total variance.

The first factor was identified as the municipal waste incinerator [65,66] situated about 6 km southeast of the sampling site. The factor is associated with high loadings of Mn, Fe, Cl−, Ni, Zn, V, K, NO<sup>3</sup> <sup>−</sup> and NH<sup>4</sup> <sup>+</sup> and moderate loadings of K+, As, Sb, Cu, Cd, Pb and F−.

The second factor was identified as vehicle exhausts. It is dominated by Sb and Cu and by moderate loadings of Zn, K, Cd, F−, K+, Mn, As, Ba, Cl<sup>−</sup> and Fe [35,36,57]. The intensity of traffic, including cars and trams on the Veveˇrí street, is quite high, with frequent queues of standing cars in front of the sampling site due to a nearby intersection.

The third factor representing the secondary aerosol production was identified as a secondary sulphate with SO<sup>4</sup> <sup>2</sup><sup>−</sup> and NH<sup>4</sup> <sup>+</sup> as the two main markers. They derive from gas to particle conversion processes from SO<sup>2</sup> oxidation and NH<sup>3</sup> neutralization [30,57]. SO<sup>2</sup> originates largely from coal and

biomass combustion, while NH<sup>3</sup> results from agricultural and traffic emissions [36,72]. This factor represents regional or long-range transport of aerosols.

The fourth factor was dominated by high loadings of Ca2+, Ca and Na+, followed by moderate loadings of oxalate and SO<sup>4</sup> <sup>2</sup>−, and represented the emissions from a cement factory [68–70] situated about 14 km east of the Brno sampling site.

The fifth factor was identified as an industry source. It was characterized with high loadings of Pb and Ba and moderate loading of NO<sup>3</sup> <sup>−</sup> [57]. The factor is associated with emissions from various industrial plants around Brno.

The sixth factor with PO<sup>4</sup> <sup>3</sup>−, K+, Cl−, F−, oxalate, NH<sup>4</sup> <sup>+</sup> and Zn, as the main indicators, was related to biomass burning [30,35]. The factor relates mainly to the regional transport of emissions from biomass (mainly wood) burning within residential heating in villages near Brno.

#### 3.5.3. Comparison of Results for Šlapanice and Brno

The major sources of PM<sup>1</sup> components identified by PCA in both localities are quite similar, but they differ with their contribution. PCA showed that local pollution from the municipal waste incinerator (accounting for 23%) and vehicle exhausts (18%) prevailed in Brno, while Šlapanice was more burdened by the transport of pollution from regional sources, such as the municipal waste incinerator in Brno (12%), the cement factory in Mokrá (13%) or coal and wood combustion (14 and 23%) in the nearby villages. This hypothesis was confirmed by one-day backward trajectories (Figures S2, S4, S6 and S8) that indicate possible regional transport of air masses to Šlapanice from all directions, although the transport from east is less frequent. Villages, predominantly south or north of Šlapanice could thus be an important source of air pollution originated from biomass and coal combustion in the frame of residential heating. Moreover, there is a coal and biomass burning power plant located southeast of Šlapanice within a distance of 1 day of transport. Trajectories also confirmed the transport of air masses to Šlapanice from a northeast or northwest direction, where the cement factory and municipal incinerator are located. One-day backward trajectories calculated for Brno location (Figures S1, S3, S5 and S7) also indicated possible regional transport of air masses to Brno from all directions, which confirms findings of the factor analysis which identified the cement factory and biomass burning (i.e., within residential heating in surrounding villages or in a biomass-burning power plant located southeast of Brno) as regional sources of air pollution in Brno. Finally, we cannot neglect the contribution of long-range transport of polluted air from remote areas at distances of as far as several hundred kilometres to the pollution of both locations.

It should be borne in mind that, due to the limited sampling period at both locations, the findings concerning the source apportionment are valid only for the sampling period, while for the rest of the year, the PM<sup>1</sup> sources or their contribution may vary.

#### **4. Conclusions**

The concentrations of elements and water-soluble ions in PM<sup>1</sup> aerosols in Brno and Šlapanice, representing a large city and a small town in the Czech Republic, were compared in heating and non-heating seasons in 2009 and 2010. The average PM<sup>1</sup> mass concentration was 19.1 and 9.65 µg m−<sup>3</sup> in winter and summer in Brno, respectively, and 30.8 and 10.0 µg m−<sup>3</sup> in winter and summer in Šlapanice, respectively. In the winter season, ions formed a significant part of PM<sup>1</sup> mass of aerosols, accounting for 34.6% (26.8–41.6%), while in the summer season the contribution of ions to PM<sup>1</sup> mass decreased to 19.7% (14.8–24.9%). Ammonium, nitrate and sulphate, the three major ion species, accounted for 91.9% of a total ion concentration and 25.1% of PM<sup>1</sup> mass. The contribution of elements to PM<sup>1</sup> mass was much smaller: 2.17% (1.70-2.76%) in winter and 1.37% (0.92–1.97%) in summer. A more recent seasonal studies in Brno and Šlapanice [11,18,19] ascribed the rest of PM<sup>1</sup> mass to other components: organic material and elemental carbon that accounted for 38.8% and 7.00% of PM<sup>1</sup> mass in winter, respectively, and for 36.5% and 8.62% of PM<sup>1</sup> mass in summer, respectively.

The differences in the concentrations of PM<sup>1</sup> aerosols, elements and water-soluble ions in the corresponding seasons (especially winter) in 2009 and 2010 may have mainly been due to different meteorological and dispersal conditions in those two years. The variations in the concentrations of PM1, elements and water-soluble ions in Brno and Šlapanice during the same season result from a change in the actual emission and meteorological and dispersal situation at the relevant site.

The backward trajectory analysis confirmed that the concentrations of elements and ions in aerosols collected in Brno or Šlapanice do not depend only on local emission sources but are affected significantly also by regional transport of polluted air from various sources both nearby (e.g., surrounding villages, cement factory etc.) and by a long-range transport of polluted air from sources at larger distances from both the studied locations, such as a power plant southeast of Brno and Šlapanice, or even from more distant areas, such as a heavily polluted region in Ostrava or southern Poland situated north of Brno and Šlapanice.

The source apportionment of the PM<sup>1</sup> samples collected in Brno and Šlapanice was performed using PCA. The six major sources of PM<sup>1</sup> components identified by PCA in both localities are quite similar in composition, although differing in their fractional contribution. Coal and biomass (largely wood) combustion, a municipal waste incinerator, vehicle exhausts, a cement factory and industry were identified as major sources at both localities. Both sampling sites were burdened by both local and regional pollution. The municipal waste incinerator (23%) and vehicle exhausts (17%) identified as the two major sources of PM<sup>1</sup> in Brno indicate a predominant effect of local sources in Brno, while Šlapanice was more burdened by the transport of pollution from regional sources, such as the municipal waste incinerator in Brno (12%) or the combustion of wood and coal in the nearby villages (37%). The transport of aerosols from sources at larger distances from Šlapanice (for example a coal- and biomass-fired power plant near the border with Slovakia) or a long-range transport of PM<sup>1</sup> from neighbouring countries should also be taken into account. However, the short sampling period at both locations restricts the validity of conclusions concerning the sources of PM<sup>1</sup> aerosols at both locations only to the sampling period, while in rest of year, the PM<sup>1</sup> sources or their contribution may vary.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/2073-4433/11/7/688/s1, Figure S1: One-day back-trajectories in winter 2009 in Brno; Figure S2: One-day back-trajectories in winter 2009 in Šlapanice; Figure S3: One-day back-trajectories in summer 2009 in Brno; Figure S4: One-day back-trajectories in summer 2009 in Šlapanice; Figure S5: One-day back-trajectories in winter 2010 in Brno; Figure S6: One-day back-trajectories in winter 2010 in Šlapanice; Figure S7: One-day back-trajectories in summer 2010 in Brno; Figure S8: One-day back-trajectories in summer 2010 in Šlapanice; Table S1: Correlation analysis between PM<sup>1</sup> , temperature, elements and ions in Šlapanice; Table S2: Correlation analysis between PM<sup>1</sup> , temperature, elements and ions in Brno.

**Author Contributions:** Conceptualization, P.M.; data curation, M.V., K.K. and Z.V.; methodology, P.M., M.V. and K.K.; project administration, M.V.; software, M.M.-C. and J.M.; supervision, P.M.; writing—original draft, M.V. and ˇ Z.V.; writing—review and editing, P.M., K.K., M.M.-C. and J.M. All authors have read and agreed to the published ˇ version of the manuscript.

**Funding:** This research received no external funding.

**Acknowledgments:** This work was supported by Grant Agency of the Czech Republic under grant No. P104/19/12109S and P503/20/02203S, by the Ministry of the Environment of the Czech Republic under grant No. SP/1a3/148/08, by the Ministry of Defence under project No. PSVR II—DZRO K-110 and by the Institute of ˇ Analytical Chemistry of CAS under an Institutional research plan No. RVO 68081715. The authors gratefully acknowledge the NOAA Air Resources Laboratory (ARL) for the provision of the HYSPLIT transport and dispersion model and/or READY website (https://www.ready.noaa.gov/index.php) used in this publication.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

## *Article* **Nanoparticle Number Concentration in the Air in Relation to the Time of the Year and Time of the Day**

**Jáchym Brzezina 1,\*, Klaudia Köbölová <sup>2</sup> and Vladimír Adamec <sup>2</sup>**


Received: 6 May 2020; Accepted: 15 May 2020; Published: 19 May 2020

**Abstract:** The paper analyzes suspended particles number concentrations of 61 size fractions (184 nm to 17,165 nm) in the air at a traffic location. The average course of the individual fractions was analyzed at various intervals – daily, weekly, monthly and annually, in the period between 2017 and 2019. The data was then used to calculate the arithmetic mean for all the fractions (MS Excel, R) and then using a proprietary web application, heatmaps were constructed. The obtained results showed significant differences in both the annual and daily variation of number concentrations between the individual fractions differing in particle size. In the case of the annual variation, one can see a greater variability of smaller particles, which is most likely due to the source of the actual suspended particles. Meteorological and dispersion conditions are found as important factors for suspended particle concentrations. These can lead to significant differences from year to year. However, a comparison between 2018 and 2019 showed that even though the average absolute number concentrations can differ between years, the actual relative number concentrations, i.e., the ratios between the individual fractions remain very similar. In conclusion it can be said that the difference between the number concentration variation of the size fractions depends on both the actual pollution sources (especially in the long-term, i.e., the annual variation) and the actual size of the particles, which plays a role especially in the short-term (daily, weekly variation).

**Keywords:** PM pollution; seasonality; air quality; meteorological conditions

#### **1. Introduction**

Air pollution has recently been identified as a major issue in the field of the environment and public health [1]. Suspended particles (PM) can potentially have very undesirable effects on human health [2]. These particles are suspended in the atmosphere and can have a very complex chemical composition and have variable sizes. Sources of fine particles (aerodynamic diameter of <2.5 µm) and ultrafine particles (aerodynamic diameter of <0.1 µm) include both natural and anthropogenic sources [3]. The increase in concentrations of PM2.5 and PM0.1 has recently become a global issue due to their impact on human health, air pollution and the atmospheric and climate system [4–6]. In general, the smaller the particle, the potentially more dangerous it is for human health as it penetrates deeper into the respiratory system or even directly to the bloodstream in the case of the smallest nanoparticles.

The current legislation in the Czech Republic and European Union as a whole specifies only mass concentrations of particles PM2.5 and PM<sup>10</sup> (aerodynamic diameter <10 µm). However, in an urban environment, the ultrafine particles represent more than 90% of particles in terms of their overall count (number concentration), but their mass concentration is negligible in comparison to large particles. There is no exact regulation for air pollution in terms of the PM<sup>1</sup> fraction [7,8]. This gap in legislation is due to insufficient data available for the PM<sup>1</sup> effects on the environment and human

health, because measuring smaller particles is demanding financially and technically. Nanotoxicological studies, however, show that particles in the nano range have completely different physio-chemical properties such as lower weight, ultrahigh reactivity, a high ratio between surface area and mass etc. These unique properties can pose more serious consequences for human health compared to particles of a larger size. It is therefore very important to study the PM<sup>1</sup> and PM0.1 particles and measure and characterize their concentration and distribution [9,10].

Lots of information and studies are available for the PM<sup>10</sup> and partially also the PM2.5 particles in Europe [11–13], however, data for the PM<sup>1</sup> particles, especially regarding their chemical composition and concentrations [14–16], short-term measurements [17–19] or long term measurements [20–22] are still insufficient.

Significant changes in seasonal variability have been observed for the concentration and size distribution of ultrafine particles. Samek et al. analyzed the seasonality effect on fine and ultrafine particles from various sources. These included combustion processes (fossil fuels, biomass), secondary aerosols, and the category "other", which included traffic, industry and soil. In winter, the major sources were combustion and secondary aerosols. In the case of combustion, fine particles dominated (53% by mass for PM2.5), while ultrafine particles represented 27%. Secondary aerosols in winter were composed of especially PM<sup>1</sup> (approximately 63%). In summer, the contribution from combustion was much smaller, from 3% to 6%. The contribution of secondary aerosols in the summer was approximately 50% for both fractions. The average particle concentration in summer months for PM<sup>1</sup> was 16.4 ± 8.3 µg·m−<sup>3</sup> and for PM2,5 it was 27.2 ± 14.1 µg·m−<sup>3</sup> . In winter months the concentrations increased to 58.0 ± 18.4 µg·m−<sup>3</sup> for PM1 and 58.6 ± 29.5 µg·m−<sup>3</sup> for PM2,5 [17]. Similar correlations between seasons and PM concentrations has also been proved by other studies [23–25].

Some studies studied the effect of traffic on particle number concentration in different seasons of the year. Meteorology and traffic emissions play a significant role in urban air quality, but relationships among them are very complicated [26,27]. Dédelé et al. [28] tried to estimate the inter-seasonal differences in concentration of PM<sup>10</sup> at different site types. The highest mean concentration of PM<sup>10</sup> was determined at sites classified as urban background in the winter season (34.8 µg/m<sup>3</sup> ), while in spring and summer, the highest concentrations of PM<sup>10</sup> were determined at traffic sites, which were characterized by high traffic intensity (>10,000 vehicles per day). This is a result of the low level of emissions from domestic heating during the warm period of the year, which means that vehicle emissions contribute more to the overall concentrations. The mean PM<sup>10</sup> concentrations measured at traffic sites ranged from 20.4 µg/m<sup>3</sup> in summer to 41.8 µg/m<sup>3</sup> in the winter season. Kami ´nska [29] analyzed the relationship between pollution, traffic, and meteorological parameters. As for the PM2,5 concentrations, the meteorological conditions had the largest effect. Only in the summer, the significance of traffic intensity was comparable to that of the meteorological conditions. This can again be explained by the low level of emissions from domestic heating in that part of the year. This finding is in accordance with the analyses performed in other cities [30,31].

Results presented in this study are based on pilot measurement at the ambient air quality at a monitoring station in Ústí nad Labem, where number concentrations have been measured since mid-2017. The main goal of this analysis was to compare the variation of various size fractions in daily, weekly and annual intervals. The monitored size fractions ranged from approximately 200 nm to particles larger than 15 µm in aerodynamic diameter.

#### **2. Experiments**

The study used data from continuously measuring automated ambient air quality monitoring station in Ústí nad Labem. The city lies in the northwest of the Czech Republic and is the center of the Ústecký region. The location is in an urban, residential and commercial area. The station is classified as a traffic station as it is located 2 m from a busy road in the direction of Teplice, Prague and Dresden (D8) on the city outskirts. Local domestic heating is an important source of pollution in this location as well. A more detailed characterization of the location is provided in Table 1 and in Figure 1.

96


**Table 1.** Characterization of ambient air quality monitoring station Ústí nad Labem-Všeboˇrická.

97 ř **Figure 1.** Photograph of the ambient air quality monitoring station Ústí nad Labem-Všeboˇrická [32,33].

98 ř 99 100 101 102 103 104 105 106 107 108 109 The station is located in the city of Ústí nad Labem near the Všeboˇrická street. It is labeled as a "hot spot" station meaning it is primarily focused on air pollution from traffic. The station is equipped with the Pallas Fidas 200 analyzer (see Table 2), which works in an automated measuring mode including measurements of particle count distribution. The analyzed particles range from 180 nm to 100 µm in aerodynamic diameter and the measuring range is 0–20,000/cm<sup>3</sup> . Volume flow is 4.8 L/min (0.3 m<sup>3</sup> /h). The measurement is based on optical light-scattering. Measurement includes monitoring of PM1, PM2.5, PM<sup>10</sup> and TSP concentrations, and particle size distribution. This measurement is set to monitor over 60 different particle-size fractions. Data used in this study included the period from 15 June 2017 to 31 December 2019 with an interval of measurement of 10 min. The station is also equipped with a traffic counter. In the period of analysis, the average daily car count was 16,751. The majority of the traffic represented passenger cars (79.99%), then vans (12.15%). Large goods vehicles represented 3.87% and large trucks 3.99%.


**Table 2.** Specification of the Pallas Fidas 200 analyzer.

#### **3. Results**

The period of analysis represented the period from 15 June 2017 to 31 December 2019. Particle number concentrations per cm<sup>3</sup> were monitored in 61 fractions (from 184 nm up to larger than 17,165 µm, see Table 3) in 10-min intervals. The analysis was focused on the average course of the individual fractions for various time intervals – daily, weekly, and annual variations.


**Table 3.** Monitored size fractions.

To compare the differences between the individual fractions, the absolute number concentrations of the individual fractions have been converted to relative values, where the overall arithmetic mean for each fraction has been calculated and the value from each interval (hour, day of the week, month) has been related to this mean value.

The following heatmap (Figure 2) shows the differences in variation of the individual fractions. The X-axis represents the individual size fractions, the Y-axis represents the hours of the day. Data are shown as a relative value of each hour to the overall arithmetic mean of that particular size fraction. The visualization clearly show that fractions of smaller particles have two obvious peaks correlating with traffic peaks in the morning and in the afternoon. However, larger particles (approximately >1.5 µm) show an increase during the day, where the number concentration increases during the morning peak hours and do not go down significantly until the end of the afternoon peak hour. One can also see that the number concentration drops much more significantly during the night in the case of the larger particles. One can, therefore, say that the variability is greater in case of the larger particles. While the 213–229 nm fraction only has a difference between the minimum and the maximum ratio of the individual hours of 0.17, the larger fractions have a difference of even more than 1 (9653–10,373 nm – 1.06; 15,976–17,165 nm – 1.09; >17,165 nm – 1.18).

**Figure 2.** Heatmap showing daily variation (hourly average, Y-axis) of all the size fractions (X-axis, in nm) analyzed.

Similarly to the hourly averages, ratios between average number concentrations for weekdays (Monday-Friday) and weekends (Saturday-Sunday) have been calculated for the individual fractions (Figure 3), where the ratio corresponds to the average number concentration of that fraction on the weekend (Saturday-Sunday) divided by the average number concentration of that fraction on weekday (Monday-Friday). The graph clearly shows that the difference between weekdays and weekends is least profound in the case of the smaller fractions around 300 nm, where the ratio between average of weekday and weekend number concentration is close to 1, i.e., same values. In contrast, most significant differences were observed in case of the larger fractions, where the ratio was approximately 0.65 (smallest for 14,865–15,974 nm fraction, 0.642), i.e., the number concentration during the weekend was approximately 65% of those observed during weekdays.

146

147 148 **Figure 3.** Ratio between average number concentration on the weekdays (Monday-Friday) and the weekends (Saturday-Sunday) weekend/weekday.

149 150 151 152 Analysis of annual variability was also summarized in a heatmap (Figure 4), where the Y-axis represents the individual months and X-axis the various size fractions. The actual value represents the ratio of the particular monthly mean in relation to the overall mean value of that fraction.

**Figure 4.** Heatmap showing annual variation (monthly average, Y-axis) of all the size fractions (X-axis, in nm) analyzed.

175

183

It can be clearly seen that the number concentrations in the case of the smaller particles were high especially in the winter months. In particular, in the case of particles in the range from 320 to 800 nm. In contrast, larger particles had the average number concentrations distributed throughout the year much more evenly.

If we divide the year into two half-years – cold (October–March) and warm (April–September), we can compare the ratio between the average number concentrations for both these half-years (Figure 5), where the ratio corresponds to the average number concentration of a particular fraction during the cold half-year divided by the average number concentration of a particular fraction during the warm half-year. The graph shows the individual size fractions (X-axis) and the ratio between the cold and warm half-years (Y-axis).

176 177 178 **Figure 5.** Ratio (Y-axis) between average number concentration of all the size fractions (X-axis) between the cold half-year (October–March) and warm half-year (April–September) (cold half-year/warm half-year).

179 180 181 The trends in the ratio show that the fractions can be divided into three groups – the smallest particles (approximately 180–320 nm), which have similar number concentrations during both half-years, medium-sized particles (approximately 320 to 700 nm) where there is a gradual increase in the relative number concentrations in the winter period, with maximum ratio observed in case of the fraction 627–674 nm (4.07). Then as the particles get larger the ratio decreases. The last group of particles, with an aerodynamic larger than approximately 2 µm, has lower number concentrations in the cold half-year than in the warm half-year. In the case of the fraction >17,165 nm, the ratio is 0.63.

To see how the various years compare a comparison was made between the two complete years 2018 and 2019 Figure 6), in particular, the average number concentrations of the individual size fractions from the entire year were compared.

182 It is obvious that the two years differ in terms of the absolute values of the average number concentrations, with higher values in 2018 (most likely due to overall better meteorological and dispersion conditions in 2019, which was a very warm year). If, however, we convert the absolute values to relative ones, i.e., calculate the relative ratios between the number concentrations of each fraction and the overall mean number concentration for each year we get the relative contribution of each fraction from the overall particle count. This comparison (Figure 7) then shows that the years 2018 and 2019 were almost identical in terms of the ratios between the average number concentrations of the individual fractions.

 **Figure 6.** Comparison between average number concentration for all size fractions for 2018 and 2019.

**Figure 7.** Comparison between the contribution of each average number concentration to the overall total for 2018 and 2019.

#### **4. Discussion**

The results of the analysis proved that the annual and daily variation in number concentrations can differ a lot in relation to the particle size. It has been shown that in the case of smaller particles in the range from approximately 200 to 800 nm, there is a significant variability throughout the year, with much higher values, particularly in the winter months. This is most likely due to the variability in particle sources during the year. In cold conditions, the intensity of heating increases significantly (being almost negligible in summer months) and local domestic heating becomes a very significant source of air pollution. Even though traffic is a very important suspended particles source at this location, it is a stable source in that it is relevant both in the summer and in the winter (although meteorological and dispersion conditions [34–36], which are in general worse in the winter – lower wind speed, less precipitation, temperature inversions – can lead to a higher number of particles in the air in winter months).

Higher number concentrations in the winter months compared to the warm months were observed especially in the case of the PM<sup>1</sup> particles. Larger particles did not show such a trend. This is in accordance with other studies. For example Vecchi [17] observed an increase of PM<sup>1</sup> in the winter period by a factor of 2.5 compared to the summer, while in the case of PM2.5 the increase was only by a factor of approximately 2. A similar conclusion was also made in a study by Triantafyllou et al. [37], which showed a more significant increase of smaller particles in the winter period.

The most significant difference between the winter and summer period has been observed for particles in the range between 300 and 800 nm. Particles of this size can be a product of heating. Zhang et al. [38] analyzed emissions from coal combustion. They concluded that while primary particles generated by coal combustion have a size of approximately 10 to 30 nm, their subsequent coagulation leads to the formation of particles approximately 500 nm large, which is in accordance with the findings of this study.

Apart from heating, low temperature also affects traffic exhaust emissions. This was proved for example by a study by Weilenmanna et al. [39], which showed that a vehicle cold start has a significant negative effect especially on the emissions of CO and HC, but to a lesser extent also suspended particles.

Larger particles show higher number concentrations in the warm period of the year. This could be due to the fact that combustion generally produces smaller particles and some sources of larger particles are significant especially in the warm part of the year. This includes for example resuspension, which is more significant in the summer than in the winter when the surface is cold and soil frozen [40]. Traffic is a significant contributor to resuspension.

When looking at the differences in daily variation of the individual fractions it is obvious that there is a much more significant difference between day and night in the case of the larger particles, which show higher number concentrations during the day. The number of larger particles increases in the morning, in correlation with the morning traffic peak. Vehicles can produce these larger particles by resuspension or abrasion of brake pads, clutch, tires, and the road surface. Number concentration falls significantly in the evening. This is most likely due to the higher mass of these particles, which are therefore more likely to deposit on the ground.

In contrast, the number concentrations of the smaller particles do not differ between day and night to such an extent as the larger particles. As Figure 2 shows, two peaks can also be seen, corresponding to the morning and afternoon traffic peak, but the average number concentration of day and night do not differ as much.

The minimum number concentration is observed around noon and early afternoon hours, not during the night. This is in accordance with other studies focusing on this topic. A study by Zhu [41] showed that even though the traffic intensity at night is 75% lower than during the day, the number of submicron particles only dropped by 20%. Explanation of this could be that the wind speed at night is lower and another possible answer is that there is a weaker atmospheric dilution at night. One other factor is air temperature. Air temperature is on average lower at night and colder ambient temperatures contribute to significantly increased nuclei mode particle formation in vehicle

exhaust [42,43]. Peréz et al. [44] think that the reason for higher concentrations of smaller particles at night compared to the larger once could also be the decrease of the boundary layer height. Lower air temperature and higher air relative humidity are a favorable factor for condensation and coagulation processes between particles and precursor gases [45].

#### **5. Conclusions**

In this study, an analysis of number concentrations in 61 size fractions of suspended particles was made focusing on the difference in daily and annual variation. Data comes from ambient air quality monitoring station Ústí nad Labem-Všeboˇrická. an urban traffic station, with data available from mid-June 2017 to the end of 2019.

The results of the study showed that there are significant differences in both the annual and daily number concentration variations depending on the particle size. In the case of the annual variation, a higher variability can be seen for the smaller particles (Figure 4). This is most likely related to the sources of these particles. The most significant source of PM2.5 particles in the Czech Republic is local domestic heating. This source, however, is almost negligible in the summer months. In contrast, in the case of the daily variation there is a higher variability of the larger particles (Figure 2). This can be explained by the fact that these larger particles deposit quicker to the ground and also by meteorological conditions as explained in the paper. Given that the most significant source of suspended particles at this station is traffic, there are obvious peaks in suspended particle concentrations correlating with morning peak hour and afternoon peak hour. In the evening and at night, there is a rapid decrease in the number of large particles in the air. Smaller particles do not deposit to such an extent so their number concentration remains higher at night. Additionally, a comparison was made between number concentrations on weekdays (Monday-Friday) and weekends (Saturday-Sunday), when the traffic intensity is much lower. The comparison showed that in this case, the variation is very similar to the daily variation in that there is a much more obvious decrease of larger particles number concentrations (Figure 3).

Meteorological and dispersion conditions have a very important effect on air quality. Their variability can cause very significant differences from year to year in terms of absolute values of number concentrations. A comparison between 2018 and 2019, however, showed that the relative ratio between the number concentrations of the various size fractions remain very similar (Figure 7).

It can be concluded that the differences in variation between the various size fractions are determined by both the actual pollution sources (especially in the long-term, i.e., annual variation) and by the actual physical properties of the particles and their behavior in the atmosphere, also determined by the meteorological and dispersion conditions.

In the future, it would be ideal to research annual and daily variation of different size fractions at more station types, i.e., not just traffic stations, but also suburban and rural background stations and also repeat the analysis when a longer time-series is available.

**Author Contributions:** Conceptualization, V.A. and J.B.; methodology, J.B. and V.A.; formal analysis, K.K.; investigation, V.A., J.B. and K.K.; resources, J.B. and K.K.; data curation, J.B.; writing—original draft preparation, J.B. and K.K.; writing—review and editing, K.K.; visualization, J.B.; supervision, V.A.; project administration, K.K.; funding acquisition, J.B. All authors have read and agreed to the published version of the manuscript.

**Funding:** This paper is supported by the Czech Hydrometeorological Institute.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

*Article*
