Chemical composition of Eastern Black Sea aerosol—Preliminary results

STOTEN-15610; No of Pages 7 Science of the Total Environment xxx (2013) xxx–xxx

Contents lists available at ScienceDirect

Science of the Total Environment journal homepage: www.elsevier.com/locate/scitotenv

Chemical composition of Eastern Black Sea aerosol—Preliminary results İlker Balcılar a,⁎, Abdullah Zararsız b, Yakup Kalaycı b, Güray Doğan c, Gürdal Tuncel a a b c

Department of Environmental Engineering, Middle East Technical University, 06800 Ankara, Turkey Turkish Atomic Energy Authority, Ankara Nuclear Research and Training Center, 06983 Ankara, Turkey Department of Environmental Engineering, Akdeniz University, 07058 Antalya, Turkey

H I G H L I G H T S • • • •

We collected PM2.5 and PM2.5–10 at coastal region of the Black Sea region in Turkey. Samples analyzed for trace elements by EDXRF The source apportionment was carried out using PMF technique. PMF analysis identified four factors.

a r t i c l e

i n f o

Article history: Received 16 August 2013 Received in revised form 4 December 2013 Accepted 5 December 2013 Available online xxxx Keywords: Eastern Black Sea Trace elements Atmospheric transport Aerosol

a b s t r a c t Trace element composition of atmospheric particles collected at a high altitude site on the Eastern Black Sea coast of Turkey was investigated to understand atmospheric transport of pollutants to this semi-closed basin. Aerosol samples were collected at a timber-storage area, which is operated by the General Directorate of Forestry. The site is situated at a rural area and is approximately 50 km to the Black Sea coast and 200 km to the Georgia border of Turkey. Coarse (PM2.5–10) and fine (PM2.5) aerosol samples were collected between 2011 and 2013 using a “stacked filter unit”. Collected samples were shipped to the Middle East Technical University in Ankara, where Na, Mg, Al, Si, S, K, Ca, Ti, V, Cr, Mn, Fe, Ni, Cu, Zn, As, Ba, Pb were measured by Energy dispersive x-ray fluorescence technique (EDXRF). Comparison of measured concentrations of elements with corresponding data generated at other parts of Turkey demonstrated that concentrations of pollution derived elements are higher at Eastern Black Sea than their corresponding concentrations measured at other parts of Turkey, which is attributed to frequent transport of pollutants from north wind sector. Positive matric factorization revealed four factors including three anthropogenic and a crustal factor. Southeastern parts of Turkey, Georgia and Black Sea coast of Ukraine were identified as source regions affecting composition of particles at our site, using trajectory statistics, namely “potential source contribution function” (PSCF). © 2013 Elsevier B.V. All rights reserved.

1. Introduction Atmospheric aerosol is composed of solid particles and liquid droplets suspended in the air. In last decades, atmospheric particles attracted attention due to variety of reasons. Aerosols are emitted into atmosphere both from natural sources and anthropogenic sources. The Earth energy balance and climate are affected by aerosol, as they interact both directly and indirectly through cloud formation with Earth's radiation budget (IPCC, 2001). They also have health effects (Hoek et al., 2002; Simkhovich et al., 2008) and cause visibility degradation (Lee and Sequeira, 2002). The chemical and physical characteristics of Mediterranean and Eastern Mediterranean aerosol have been studied extensively and ⁎ Corresponding author at: Middle East Technical University, Department of Environmental Engineering, 06800 Ankara, Turkey. Tel.: + 90 3122102652. E-mail address: [email protected] (İ. Balcılar).

currently a broad database has been available for composition of eastern Mediterranean particles and implications of these compositions (e.g., Koçak et al., 2007; Koulouri et al., 2008; Koçak et al., 2012; Öztürk et al., 2012; Pey et al., 2013). Results of those studies illustrated that the Mediterranean aerosol is mainly a four component system, including crustal particles originating from arid regions at North Africa and Middle East (Koçak et al., 2012; Öztürk et al., 2012), an anthropogenic component originating from industrialized regions located at the north of the basin (Öztürk et al., 2012; Pey et al., 2013), a sea salt component and a biogenic component (Koçak et al., 2007; Koulouri et al., 2008). The anthropogenic component in the Mediterranean aerosol population was further investigated to understand types (Herut et al., 2001; Doğan et al., 2008) and locations of anthropogenic sources contributing to the aerosol composition in the Western and Eastern Mediterranean atmospheres (Koçak et al., 2004a; Güllü et al., 2005; Doğan et al., 2008). Physical and radiative characteristics of Eastern (Ichoku et al., 1999; Fotiadi et al., 2006) and Western (Mallet et al.,

0048-9697/$ – see front matter © 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.scitotenv.2013.12.023

Please cite this article as: Balcılar İ, et al, Chemical composition of Eastern Black Sea aerosol—Preliminary results, Sci Total Environ (2013), http:// dx.doi.org/10.1016/j.scitotenv.2013.12.023

2

İ. Balcılar et al. / Science of the Total Environment xxx (2013) xxx–xxx

2011; Esteve et al., 2012) Mediterranean aerosols were also investigated through modeling studies (Barnaba and Gobbi, 2004; Spyridaki et al., 2006) and lidar network was established around the basin (Balis et al., 2003). Studies to investigate chemical composition of Black Sea particles are not as abundant as the studies in the Mediterranean region. Few studies on particle composition in the Black Sea atmosphere demonstrated that the levels of particles and specific elements, particularly SO24 − are not significantly different from the levels of particles in the Eastern Mediterranean atmosphere (Hacısalihoğlu et al., 1992; Karakaş et al., 2004). However, particle composition at Eastern and Western Black Sea is not identical. Particle composition at the western parts of the Black Sea is influenced strongly from emissions in Balkan and central European countries and in that sense similar to the composition of particles sampled in the Eastern Mediterranean atmosphere (Hacısalihoğlu et al., 1992). Sources affecting Western Black Sea atmosphere originated from Central Russia (Dzubay et al., 1984; Karakaş et al., 2004) and countries surrounding the basin (Hacısalihoğlu et al., 1992; Tecer et al., 2008). Studies targeted to understand types of particles in aerosol population and to understand where those sources are located are integral parts of aerosol studies in all basins, including the Mediterranean (Koçak et al., 2007; Doğan et al., 2008; Öztürk et al., 2012). Multivariate statistics, particularly factor analysis (Hopke et al., 1976) and positive matrix factorization (Paatero and Tapper, 1994) in recent years have been widely used for that purpose. Studies towards finding locations of source regions contributing to measured concentrations of pollutants in the Eastern Mediterranean and Black Sea basins are few (Doğan et al., 2010; Tecer et al., 2012). Trajectory statistics, which combines information on concentrations of pollutants with backtrajectory information to locate source regions, is a common tool which is applied not only in the Mediterranean region (Güllü et al., 1998, 2005), but also for other locations around the world (Chen et al., 2002; Abdalmogith and Harrison, 2005). In this study multi-element data generated at a rural site on the Eastern Black Sea was investigated to understand (a) general characteristics of element concentrations, such as seasonality etc., (b) components of aerosol population at the eastern Black Sea and (c) locations

of the sources of anthropogenic component in the aerosol mass. Statistical tools at all levels are extensively used to answer these questions. Positive matrix factorization was used to identify source types and potential source contribution (PSCF) approach, which is a widely used tool in trajectory statistics, was used to find potential source regions affecting chemical composition of Eastern Black Sea particles. 2. Materials and methods 2.1. Sampling location Samples were collected at a station established in a timber-storage depot of the General Directorate of Forestry, on the Eastern Black Sea coast of Turkey (40°32′34″N 39°16′57″E). Location of the station is depicted in Fig. 1. Sampling point is approximately 50 km to the Black Sea coast and 200 km to the Georgian border of Turkey. The setting is typical rural with limited settlement around the station. There is also no significant industrial activity within a circle of 50 km radius around the station. Sampling was started in March 2011 and still continues; however, data from samples collected before January 2013 were used in this manuscript. 2.2. Sampling technique Samples were collected with a Gent Stacked Filter Unit (SFU) (Hopke et al., 1997) on polycarbonate (Nuclepore) filters. The SFU consists of a pump, two cascade filters with pore sizes of 8.0 μm and 0.4 μm and a mass flow controller that fixes the air flow rate at 16.7 L min− 1. Filters were placed in a NILU, two-stage filter holder. When air is passed through such a system at a flow rate of 16.7 L min− 1, particles with diameters N 2.5 μm are held in the top filter and particles with diameters b 2.5 μm passes through the top filter and retained at the bottom filter. A pre-impactor, which has a cut-point at 10 μm are also included to the system. When SFU is operated at a flow-rate of 16.7 L min− 1 particles with diameters N 10 μm are held at the pre-impactor and are not allowed to reach to filters. Particles with diameters between 10 μm and 2.5 μm are held at the top filter

Fig. 1. Location of the sampling station.

Please cite this article as: Balcılar İ, et al, Chemical composition of Eastern Black Sea aerosol—Preliminary results, Sci Total Environ (2013), http:// dx.doi.org/10.1016/j.scitotenv.2013.12.023

İ. Balcılar et al. / Science of the Total Environment xxx (2013) xxx–xxx

(w/pore size of 8.0 μm) and particles with diameters b2.5 μm are held on the bottom filter (w/pore size of 0.4 μm).

Table 1 Field blanks relative to sample concentrations at coarse and fine fraction. Blank average ng cm-2

2.3. Chemical analysis After sampling, filters containing samples were shipped to the Middle East Technical University where they are weighed. Samples were then analyzed for trace elements using Oxford ED-2000 energy dispersive X-Ray Fluorescence Spectrometer (EDXRF) at the Ankara Nuclear Research and Training Center. Four successive counting were performed for different group of elements. The first count was 150 second long at the tube current of 900 mA and tube voltage of 2.5 kEV. This first counting was performed for the analysis of lowZ elements, including, Na, Mg, Al, Si, S and Cl. The second counting was performed at 10 kEV, 900 mA and lasted for 100 s. K and Ca were the elements determined in this second counting. Ten samples were excited for the third time to determine Ti, V, Cr and Mn. This third excitation was performed at 15 kEV, 1000 mA and lasted for 100 s. Finally Fe, Ni, Cu, Zn, As, Ba, Pb were determined at the fourth counting, where samples were excited at 22.6 kEV, 494 mA and lasted for 100 s. A silver anode was used in all counting. Samplers were introduced to instrument using an 8-slot auto sampler. EDXRF was calibrated against NIST SRM 8785 (Air particulate matter on filter media), owing to its physical and chemical similarities to our samples. Eighteen elements measured in the study were the ones certified in this standard. 2.4. QA/QC protocol and blank levels As large numbers of samples were routinely analyzed in our laboratories, fairly extensive QA/QC protocol was an integral part of our analytical system. The protocol includes a careful recording system (sample management), repeated analysis, determination of detection and quantification limits of elements and routine analysis of field, laboratory and filter blanks. Repeated analysis includes, sending 5% of samples back to the XRF laboratory to be analyzed for the second time (without lab knowing this). Blanks include field blanks, laboratory blanks and filter blanks. Field blanks were filters sent to the station like regular filters. However, they were loaded to the sampler, pump was operated for one minute, and then they were removed sampler. From this point on, field blanks were treated like sample filters. Laboratory blanks were the filters weighted and treated like sample filters in the laboratory. These blank filters were not shipped to the station. Filter blanks were filters directly analyzed by XRF, without any manipulation in the laboratory. Generally 3 filters were spared from each box for analysis. Filter blanks were analyzed to test box-to-box variation of blank levels in Nuclepore filters. Blank levels and percent blank subtraction of elements in coarse and fine fractions are given in Table 1. The elements measured in this study can be separated into two groups based on their filter blank values. The first group consisted of Na, Mg, Al, Si, S, K, Ca, Ti, Mn, Fe. Their field blanks are smaller than 10% of their average sample concentrations. Elements that are typically associated with coarse fraction, such as Al, Na, Si etc., have higher percent blank subtraction in fine fraction and S, which is associated with fine fraction, has higher blank subtraction in coarse fraction. Remaining elements, which include V, Cr, Cu, Zn, As, Ba and Pb, have percent blank subtraction values between 16% and 40%. Such high blank subtraction is reflected in their frequency of observances. 3. Results and discussion Average concentrations of 17 elements measured in coarse (PM2.5–10) and fine (PM2.5) fractions are given in Table 2. Concentrations varied between approximately 1 ng m−3 for As and 1800 ng−3

3

Na Mg Al Si S K Ca Ti V Cr Mn Fe Ni Cu Zn As Ba Pb

86 ± 34 16 ± 9 41 ± 9 123 ± 117 246 ± 77 39 ± 16 90 ± 71 17 ± 8 5±7 25 ± 9 9±6 334 ± 98 BDL 23 ± 9 68 ± 69 0,84 ± 0,45 28 ± 11 46 ± 27

Coarse (PM2.5–10)

Fine (PM2.5)

Sample average

Sample average

Percent blank subtraction

801 1432 3871 11136 1138 1364 8162 2072 150 183 221 7518 19 76 203 3 64 50

11 1 1 1 22 3 1 1 3 14 4 4 0 30 33 33 44 92

405 690 1719 4404 5713 931 2759 197 24 145 128 4629 32 76 220 3 59 68

Percent blank subtraction 21 2 2 3 4 4 3 8 21 17 7 7 0 30 31 29 48 68

for SO24 −. All parameters had right-skewed distributions. Chi-square test demonstrated that most but not all are log-normally distributed. Concentration of Na, which is a good marker element for sea salt, was fairly low due to 50 km distance between the station and coastline. Its concentration was higher in the coarse fraction as expected. Concentrations of soil-derived elements, including Al, Si, K, Ca, Ti, V and Fe are higher in the coarse fraction as soil particles have mass median diameter of approximately 3.0 to 3.5 μm (Kuloglu and Tuncel, 2005). Concentrations of elements with mixed natural and anthropogenic sources, such as, Ti, V, Cr, Mn and Ni had generally higher concentrations in coarse fraction, which is reasonable, since part of their concentrations is due to presence of coarse soil particles. Pollution-derived elements depicted interesting variations in terms of their coarse and fine fraction concentrations. Sulfate ion had higher concentrations in the fine fraction, which was an expected behavior for a secondary specie. However, other anthropogenic elements, including Cu, Zn, As and Pb had higher concentrations in coarse fraction. This was an unexpected outcome since, association of these species with fine particles is well documented in literature (Dulac et al., 1989; Yatin et al., 2000). These pollution derived elements were also associated with fine particles in our previous studies with Mediterranean aerosol (Kuloglu and Tuncel, 2005). The reason for this unexpected behavior of anthropogenic elements is not known for the time being.

Table 2 Elemental average concentrations in the Black Sea PM (concentrations are in ng m−3). Average ± SD

Na Mg Al Si SO4 K Ca Ti V Cr Mn Fe Ni Cu Zn As Pb

Coarse (PM2.5–10)

Fine (PM2.5)

88.0 225 695.3 2063.4 274.3 170.5 940.3 915.2 43.5 57.7 48.5 2435 8.4 22.6 71.9 0.9 118.8

36.7 ± 69.9 ± 337.3 ± 630.7 ± 1809.5 ± 123.5 ± 550.9 ± 96.6 ± 4.4 ± 39.4 ± 14.5 ± 1126.5 ± 8.6 ± 14.6 ± 28.4 ± 1± 13.1 ±

± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ±

131.6 611.6 1659.7 6026.4 372.7 273.9 2073.6 5759.6 211.5 202.9 160.6 6385.7 7.2 32.2 315.4 0.4 516.4

30.7 103.2 347 1081 2509 93.4 781.8 143.9 5.2 172.3 20.6 1687.9 5.1 21.9 37.6 0.1 9

Please cite this article as: Balcılar İ, et al, Chemical composition of Eastern Black Sea aerosol—Preliminary results, Sci Total Environ (2013), http:// dx.doi.org/10.1016/j.scitotenv.2013.12.023

İ. Balcılar et al. / Science of the Total Environment xxx (2013) xxx–xxx

4

Concentrations of measured elements also depicted relatively clear seasonal variations, due seasonal variations in both meteorology and source strengths. The year is divided into two seasons as summer and winter. Division of the year into two seasons was adopted from our earlier studies at the Eastern Mediterranean atmosphere (Güllü et al., 2000; Öztürk et al., 2012). It was based on rainfall amount. In the Mediterranean region approximately 80% of annual rainfall is recorded in our winter period. Since wet scavenging accounts for a significant fraction of variations in concentrations of elements, division of the year as wet and dry periods was reasonable. Although temporal distribution of rainfall in the Black Sea atmosphere is slightly different (60% not 80% of annual rainfall is observed in “winter” period), we adopted the same division to be able to compare seasonal variations observed at the Eastern Mediterranean atmosphere with those observed in the Mediterranean region. Seasonal variations in concentrations of measured elements are given in Table 3. Crustal elements like Al, Fe, Ca had higher concentrations in summer. Similar pattern was also observed in most of the studies in the region and was attributed to easier resuspension of soil particles during dry summer season (for example, Güllü et al., 1998; Kubilay et al., 2000). Since soil is damp or ice covered in winter season its resuspension is less likely. This scenario results in higher concentrations of soil-related elements in winter. Sodium, which is a marine element in most of the coastal studies, had also higher concentrations in summer season. At coastal areas NaCl from sea salt is the dominant source of measured Na concentration. However, marine contribution to Na concentration decreases with distance from the coast, because sea salt particles are coarse and settle out quickly with distance. A simple calculation was performed by assuming Al concentration measured at our station originates entirely from soil and soil composition is similar to the composition given in Mason's (Mason and Moore, 1982) compilation demonstrated that N85% of Na at our sampling point comes from soil and not from sea salt. Although this is a crude number, it explains why Na depicts a seasonal variation, which is very similar to that observed in lithophilic elements. Anthropogenic elements As, Pb, Cu and Zn were found either in comparable concentrations in both seasons, or slightly higher concentrations in summer. Although concentrations of SO 24 − , Cu and Zn were found higher in summer, the difference between their summer and winter concentrations was not as large as the difference observed in crustal elements. Less pronounced seasonal variations in concentrations of pollution-derived elements were also observed in most of the studies performed in the Mediterranean basin (Güllü et al., 2000; Koçak et al., 2004b). Such pattern was attributed to the

long range transport of anthropogenic species to the study area and strong influence of scavenging, particularly wet scavenging, during transport, which may smooth out seasonal variations in their source strengths (Al-Momani et al., 1998; Güllü et al., 2005). and other elements measured in this study Concentrations of SO2− 4 were compared with corresponding levels reported for other locations in the Eastern Mediterranean atmosphere. Data used in the comparison are given in Table 4. This comparison depicts some interesting features of the Eastern Black Sea aerosol. Sodium concentration at our station is the lowest in all studies shown in the table. Concentrations of sea salt elements at any site depend heavily on the distance of the station to coastline. Samples in this study were collected at approximately 50 km from the Black Sea, which is the reason for small Na concentrations measured relative to other studies used in comparison. Please also note that salt content of the Black Sea is approximately half of the salt content in the Mediterranean Sea(18‰ in the Black Sea vs 38‰ in the Mediterranean Sea). These differences are probably the main reason for the observed difference in Na concentrations in the table. Concentrations of soil-related elements are comparable to the concentrations reported in other studies in the Mediterranean basin. Most of the stations in the Eastern Mediterranean (also in the Western Mediterranean) are under strong influence of dust transport from North Africa and thus expected to have higher levels of soil-related elements than sites in the Black Sea. Interestingly, this is not the case observed in Table 4. Concentrations of crustal elements like Al, Si, and Fe are among the highest in Table 4. Sources of high concentrations of lithophiles are not clear. Transport from distant sources, like Sahara, or in this case, from Asian deserts can result in the observed concentrations of crustal elements. Analysis of air mass backtrajectories, which was not performed for this work, will provide some information for the observed high levels of these elements. Although our station is far from strong anthropogenic emission sources (nearest densely populated town is approximately 50 km away, strong industrial emissions are even further away), concentrations of pollution-derived elements, such as, Cu, Zn, As, Pb, Ni, Cr are among the highest in the table, indicating that Eastern Black Sea atmosphere is under strong influence of pollution sources. Concentrations of SO24 − measured in the Eastern Mediterranean is on the average 2100 ± 2800 ng m−3. This average value is smaller than SO2− 4 concentrations reported for other sites in the table. However, concentrations depends on the year samit should be noted that, SO2− 4 ples were collected, as SO24 − concentration has been decreasing throughout Europe (Arends et al., 1997), probably in Balkans and meaTurkey as well. This may be partly responsible for the low SO2− 4 sured at our site in the year 2011.

Table 3 Seasonal average concentrations of elements (concentrations are in ng m−3).

3.1. Positive matrix factorization

Na Mg Al Si SO4 K Ca Ti V Cr Mn Fe Ni Cu Zn As Pb

Coarse (PM2.5–10) average

Fine (PM2.5) average

Summer

Winter

Summer

Winter

105.1 247.1 754.9 2521.5 311.8 181.8 1085.9 1139.5 60.7 114.5 89.8 3635.8 9.6 26.1 95.0 1.1 13.3

48.5 150.6 484.0 829.9 210.5 132.6 639.5 85.2 8.3 11.7 13.1 1005.6 6.8 18.6 23.7 0.7 435.3

41.3 76.1 371.0 711.0 1944.7 133.4 633.0 108.9 5.8 75.5 21.6 1702.7 10.2 18.9 34.8 1.0 13.7

19.0 40.7 188.9 283.7 1671.6 100.1 374.9 55.0 3.0 10.2 8.4 632.7 5.4 9.8 17.7 0.0 10.4

Aerosol components in our data set were resolved using multivariate receptor modeling, namely positive matrix factorization (PMF). PMF has been a very popular tool for source apportionment in the last ten years due to its advantages over conventional multivariate receptor modeling tools, such as factor analysis (FA) of principal component analysis (PCA) (Reff et al., 2007; Cuccia et al., 2011; Pant and Harrison, 2012). An object function minimized, with an unknown source profile, by PMF while providing a solution to uncertainties for each observation. Object function, Q, is defined as 0 12 XN m X n X xij − k¼1 gik f jk @ A Q¼ σ ij i¼1 j¼1 where σij is the uncertainty of ambient concentration of species j in sample i (Paatero and Tapper, 1993, 1994; Lee et al., 2008).

Please cite this article as: Balcılar İ, et al, Chemical composition of Eastern Black Sea aerosol—Preliminary results, Sci Total Environ (2013), http:// dx.doi.org/10.1016/j.scitotenv.2013.12.023

İ. Balcılar et al. / Science of the Total Environment xxx (2013) xxx–xxx

5

Table 4 reported for different locations in the Eastern Mediterranean and Black Sea (concentrations are in ng m−3). Concentrations of trace elements and SO2− 4 This work

Bolu Turkey

Kirklareli Turkey

Antalya Turkey

Cyprus

Ashdod (Israel)

Beirut Lebanon

Northern Greece

Crete, Greece

Erdemli Turkey

Sampling year

2011

2011

2011

2001

2003

2005

2005

2001

2001

1999

Na Mg Al Si SO4 K Ca Ti V Cr Mn Fe Ni Cu Zn As Ba Pb

125 400 1000 2700 2100 290 1500 1012 47.9 98 62 3566 17 38 100 1.9 0 132

333 285 218 470 664 825 1571 1835 7289 4194 510 349 2615 1203 83 119 8 9 37 35 47 48 1645 2370 12 12 26 33 95 92 2 1 26 54 44 23 Tuncel unpublished data

1476 371 511

1357 189 471 693 4500 353 1780 108

40172 692 776 2300 1320 630 4655 113 38.5 21.8 38 1210 14.5 14.2 165 0 0 49 Mamane et al. (2008)

2000 300

2900 440 7500

6900 290 1500

7500 300 2700

Bardouki et al. (2003)

Koçak et al. (2004b)

6624 267 968 29 3 4 7 297 3 10 1 19 Öztürk et al. (2012)

6 36 470 6 12 19 26 Bari et al. (2009)

An approach developed by Polissar et al. (2001) was used for the estimation of the uncertainties. Uncertainty estimates were defined as 10% of the measured concentration plus the detection limit. Values below the detection limit were replaced by half of the detection limit values, and their overall uncertainties were set at 5/6 of the detection limit values (Begum et al., 2005; Hammond et al., 2008; Tecer et al., 2012). The PMF 3.0 model developed by the Environmental Protection Agency (EPA, 2008) was used in this work. Positive matrix factorization was applied to fine fraction aerosols, which includes most of the anthropogenic component. Elements, which were detected in less than 60% of the samples were marked as “weak” variables. This turned out to be a weakness in PMF exercise, which is inherent to XRF technique, because important anthropogenic marker elements, such as As, Pb had to be marked as weak variables and did not contribute to fitting in the model. A four factor solution provided closest Qtheretical and Qmodeled values and generated physically explainable factors. Composition of factors (F-loading values), Fractions of elements explained by each factor (EV), which corresponds to “explained variance” in factor analysis, temporal variation of G-scores were used to assign physical sources to factors. We also calculated crustal enrichment factors of elements in each factor, which proved to be a valuable parameter to identify sources corresponding to factors. The results are represented in Fig. 2. Factor 1 explains approximately 80% of the variance for Fe and 40%–80% of the variances of anthropogenic elements such as, V, Mn, Ni, Cu and Zn. This factor is identified as particles emitted from iron and steel plants, which is expected to be an important source at the Black Sea region. Ukraine (approximately 35 × 106 t) and Turkey (approximately 30 × 106 t) are the largest steel producers in Europe. There are two large iron and steel complexes on the Black Sea cost of Turkey. One is at Ereğli and the other one at Karabük (approximately 60 km from the cost). There are nine iron and steel plants at eastern parts of Ukraine. These are the industries responsible for Factor 1. Factor 1 scores demonstrate a clear seasonality, with higher scores during winter months and lower towards summer. Since particles transported over long distances are more prone to wet scavenging with frequent rains during winter, their concentrations are expected to be low in winter and high during dry summer. Consequently, elements with higher concentrations in winter were the ones that were not removed extensively by rain. This is only possible if that element has a source that is not far away from the sampling point (Güllü et al.,

4100 7700 530 6600 170 174 44 2200 61

170 Saliba et al. (2010)

280 667 1604 2700 521 6775 89.6 5.5 17.22 33.4 1389 12.17 36.2 40.8 7.85 20.37 Samara, (2005)

1998). This scenario also supports iron and Steel plants on the Black Sea coast to explain Factor 1. Factor 2 is a typical crustal factor. It accounts for the most of the variances of lithophilic elements such as, Al and Si. Factor scores were higher at summer months, which is also typical for crustal elements, as discussed in previous sections. Formation of crustal aerosol is only possible when the soil is dry. Thus, during summer period, higher crustal dust contribution in the atmosphere is more likely as the soil is dry. Factor 3 represents emissions from metallurgy. The factor accounts for 60% of the variances in Cr, Ni, As and Pb. Factor loadings indicate high SO4 concentrations but factor 3 did not account for a significant fraction of its variance. This is a typical factor representing metallurgical industries. Black Sea region in general are surrounded large metallurgical regions. There are two large Cu smelters on the Black Sea cost of Turkey. The largest Mn smelter in the region is at Nikapol, Ukraine, which is approximately 150 km from the Black Sea coast of Ukraine. There is also a large smelter at Zestafoni, Georgia. Strong influence of metallurgical industries on composition of Black Sea aerosol was noted previously by Hacısalihoğlu et al. (1992), who reported interception of particles from metallurgical industries from Ukraine in the shipborne samples collected in the middle of the Black Sea and by Dzubay et al. (1984) who demonstrated that samples collected at Georgia in 1982 were affected from smelters in Georgia and Russia. Factor 4 was identified as secondary SO4 factor. Sulfate is the most important specie in Factor 4 profile as concentration. It explains almost variance. Although less part of the varapproximately 70% of the SO2− 4 iances of other species explained in this factor, presence of other anthropogenic elements reminiscent that they are formed from mixture of emissions from different sources during long range transport. Any distinct difference was not observed between summer and winter factor scores of this factor, which is also typical for the species that arrive to site as a result of long range transport.

4. Conclusions Fine (PM2.5) and coarse (PM2.5–10) particles were collected at the eastern Black Sea region between March 2011 and January 2013. Collected samples were analyzed for 18 major, minor and trace elements using EDXRF.

Please cite this article as: Balcılar İ, et al, Chemical composition of Eastern Black Sea aerosol—Preliminary results, Sci Total Environ (2013), http:// dx.doi.org/10.1016/j.scitotenv.2013.12.023

İ. Balcılar et al. / Science of the Total Environment xxx (2013) xxx–xxx

6

Fig. 2. Factor-(F loading), explained variances (EV), g-scores in each factor with PMF.

Although crustal and marine elements were associated with coarse particles, as expected, some of the pollution derived elements, which are expected to occur in fine particles, had comparable concentrations in coarse and fine fraction particles. The reason for this behavior is not clear. Litophilic elements had higher concentrations in summer, due to easier resuspension of soil particles during dry summer season. There was no good marine marker element. Most of the Na was from soil due to relatively long distance between the cost and the sampling point. Anthropogenic elements did not depict clear seasonal differences as observed in most of the studies in the Mediterranean atmosphere. Four factors were identified in PMF. These four main components in Eastern Black Sea aerosol included particles emitted from iron and steel plants around the basin, particles emitted from metallurgical industries, which are abundant in the region, soil particles and secondary sulfate, which represent long range transported particles to the region.

Acknowledgments This work was supported by the TÜBİTAK fund. We gratefully acknowledge the Ministry of Forestry for their corporation in operating our station in their premises.

References Abdalmogith SS, Harrison RM. The use of trajectory cluster analysis to examine the long-range transport of secondary inorganic aerosol in the UK. Atmos Environ 2005;39(35):6686–95. Al-Momani IF, Aygun S, Tuncel G. Wet deposition of major ions and trace elements in the eastern Mediterranean basin. J Geophys Res Atmos 1998;103:8287–99. Arends BG, Baard JH, Ten Brink HM. Trends in summer sulphate in Europe. Atmos Environ 1997;31(24):4063–72. Balis DS, Amiridis V, Zerefos C, Gerasopoulos E, Andreae M, Zanis P, et al. Raman lidar and sunphotometric measurements of aerosol optical properties over Thessaloniki, Greece during a biomass burning episode. Atmos Environ 2003;37(32):4529–38. Bardouki H, Liakakou H, Economou C, Sciare J, Smolík JV, Eleftheriadis DK, et al. Chemical composition of size-resolved atmospheric aerosols in the eastern Mediterranean during summer and winter. Atmos Environ 2003;37(2):195–208. Bari MA, Baumbach G, Sarachage-Ruiz L, Kleanthous S. Identification of PM10 sources in a Mediterranean island. Water Air Soil Pollut 2009;9:39–53. Barnaba F, Gobbi GP. Aerosol seasonal variability over the Mediterranean region and relative impact of maritime, continental and Saharan dust particles over the basin from MODIS data in the year 2001. Atmos Chem Phys 2004;4:2367–91. Begum BA, Hopke PK, Zhao W. Source identification of fine particles in Washington, DC, by expanded factor analysis modeling. Environ Sci Technol 2005;39(4):1129–37. Chen LWA, Doddridge BG, Dickerson BR, Chow JC, Henry RC. Origins of fine aerosol mass in the Baltimore–Washington corridor: implications from observation, factor analysis, and ensemble air parcel back trajectories. Atmos Environ 2002;36(28):4541–54. Cuccia E, Piazzalunga A, Bernardoni V, Brambilla L, Fermo P, Massabò D, et al. Carbonate measurements in PM10 near the marble quarries of Carrara (Italy) by infrared spectroscopy (FT-IR) and source apportionment by positive matrix factorization (PMF). Atmos Environ 2011;45:6481–7. Doğan G, Güllü G, Tuncel G. Sources and source regions effecting the aerosol composition of the Eastern Mediterranean. Microchem J 2008;88:142–9.

Please cite this article as: Balcılar İ, et al, Chemical composition of Eastern Black Sea aerosol—Preliminary results, Sci Total Environ (2013), http:// dx.doi.org/10.1016/j.scitotenv.2013.12.023

İ. Balcılar et al. / Science of the Total Environment xxx (2013) xxx–xxx Doğan G, Gülen G, Karakaş D, Tuncel G. Comparison of source regions affecting SO42− and NO3− concentrations at the Eastern Mediterranean and Black Sea atmospheres. Curr Anal Chem 2010;6:66–71. Dulac F, Menard PB, Ezat U, Melki S, Begametti G. Atmospheric input of trace metals to the western Mediterranean uncertainties in modeling and dry deposition from cascade impactor data. Tellus 1989;41B:362–78. Dzubay TG, Stevens RK, Haagenson PL. Composition and origins of aerosol at a forested mountain in Soviet Georgia. Environ Sci Technol 1984;18:873–8. EPA. Positive Matrix Factorization (PMF)3.0 fundamentals & user's guide; U.S. Environmental Protection Agency, Research Triangle Park, NC. available at:http://www.epa. gov/heasd/research/pmf.html, 2008. [accessed 2013]. Esteve AR, Estellés V, Utrillas MP, Martínez-Lozano JA. In-situ integrating nephelometer measurements of the scattering properties of atmospheric aerosols at an urban coastal site in western Mediterranean. Atmosp Environ 2012;47:43–50. Fotiadi A, Hatzianastassiou N, Drakakis E, Matsoukas C, Pavlakis KG, Hatzidimitriou D, et al. Aerosol physical and optical properties in the Eastern Mediterranean Basin, Crete, from Aerosol Robotic Network data. Atmos Chem Phys 2006;6(12):5399–413. Güllü G, Olmez I, Aygun S, Tuncel G. Atmospheric trace element concentrations over the Eastern Mediterranean sea: factors affecting temporal variability. J Geophys Res 1998;103:21943–54. Güllü GH, Ölmez İ, Tuncel G. Temporal variability of atmospheric trace element concentrations over the eastern Mediterranean Sea. Spectrochim Acta B 2000;55(7): 1135–50. Güllü G, Doğan G, Tuncel G. Atmospheric trace element and major ion concentrations over the eastern Mediterranean Sea: identification of anthropogenic source regions. Atmos Environ 2005;39:6376–87. Hacısalihoğlu G, Eliyakut F, Ölmez I, Balkas TI, Tuncel G. Chemical composition of particles in the Black Sea atmosphere. Atmos Environ 1992;26A:3207–18. Hammond DM, Dvonch JT, Keeler GJ, Parker EA, Kamal AS, Barres JA, et al. Sources of ambient fine particulate matter at two community sites in Detroit, Michigan. Atmos Environ 2008;42(4):720–32. Herut B, Nimmo M, Medway A, Chester R, Krom MD. Dry atmospheric inputs of trace metals at the Mediterranean coast of Israel (SE Mediterranean): sources and fluxes. Atmos Environ 2001;35(4):803–13. Hoek G, Brunekreef B, Goldbohm S, Fischer P, van den Brandt PA. Association between mortality and indicators of traffic-related air pollution in the Netherlands: a cohort study. Lancet 2002;360(9341):1203–9. Hopke PK, Gladney ES, Gordon GE, Zoller WH, Jones AG. The use of multivariate analysis to identify sources of selected elements in the Boston urban aerosol. Atmos Environ 1976;10:1015–25. Hopke PK, Xie Y, Raunema T, Biegalski S, Landsberger S, Maenhaut W, et al. Characterization of the Gent Stacked Filter Unit PM10 Sampler. Aerosol Sci Tech 1997;27:726–35. Ichoku C, Andreae MO, Andreae TW, Meixner FX, Schebeske G, Formenti P. Interrelationships between aerosol characteristics and light scattering during late winter in an Eastern Mediterranean arid environment. J Geophys Res 1999;104:24371–93. IPCC. Climate change 2001. Intergovernmental panel on climate change. London: Cambridge University Press; 2001. Karakaş D, Olmez I, Tosun S, Tuncel G. Trace and major element compositions of Black Sea aerosol. J Radioanal Nucl Chem 2004;259(1):187–92. Koçak M, Nimmo M, Kubilay N, Herut B. Spatio-temporal aerosol trace metal concentrations and sources in the Levantine Basin of the Eastern Mediterranean. Atmos Environ 2004a;38(14):2133–44. Koçak M, Kubilay N, Mihalopoulos N. Ionic composition of lower tropospheric aerosols at a Northeastern Mediterranean site: implications regarding sources and long-range transport. Atmos Environ 2004b;38(14):2067–77. Koçak M, Mihalopoulos N, Kubilay N. Chemical composition of the fine and coarse fraction of aerosols in the northeastern Mediterranean. Atmos Environ 2007;41(34):7351–68.

7

Koçak M, Theodosi C, Zarmpas P, Séguret MJM, Herut B, Kallos G, et al. Influence of mineral dust transport on the chemical composition and physical properties of the Eastern Mediterranean aerosol. Atmos Environ 2012;57:266–77. Koulouri E, Saarikoski S, Theodosi, Markaki Z, Gerasopoulos E, Kouvarakis, et al. Chemical composition and sources of fine and coarse aerosol particles in the Eastern Mediterranean. Atmos Environ 2008;42(26):6542–50. Kubilay N, Nickovi S, Moulin C, Dulac F. An illustration of the transport and deposition of mineral dust onto the eastern Mediterranean. Atmos Environ 2000;34: 1293–303. Kuloglu E, Tuncel G. Size distribution of trace elements and major ions in the eastern Mediterranean atmosphere. Water Air Soil Pollut 2005;167:221–41. Lee YL, Sequeira R. Water-soluble aerosol and visibility degradation in Hong Kong during autumn and early winter, 1998. Environ Pollut 2002;116:225–33. Lee S, Liu W, Wang Y, Russell AG, Edgerton ES. Source apportionment of PM2.5: comparing PMF and CMB results for four ambient monitoring sites in the southeastern United States. Atmos Environ 2008;42(18):4126–37. Mallet M, Gomes L, Solmon F, Sellegri K, Pont V, Roger JC, et al. Calculation of key optical properties of the main anthropogenic aerosols over the Western French coastal Mediterranean Sea. Atmos Res 2011;101:396–411. Mamane Y, Perrino C, Yossef O, Catrambone M. Source characterization of fine and coarse particles at the East Mediterranean coast. Atmos Environ 2008;42(24):6114–30. Mason B, Moore CB. Principles of geochemistry. 4th ed. Wiley; 1982. Öztürk F, Zararsız A, Dutkiewicz VA, Husain L, Hopke PK, Tuncel G. Temporal variations and sources of Eastern Mediterranean aerosols based on a 9-year observation. Atmos Environ 2012;61:463–75. Paatero P, Tapper U. Analysis of different modes of factor-analysis as least-squares fit problems. Chemometr Intell Lab 1993;18(2):183–94. Paatero P, Tapper U. Positive matrix factorization a nonnegative factor model with optimal utilization of error estimates of data values. Environmetrics 1994;5(2):111–26. Pant P, Harrison RM. Critical review of receptor modelling for particulate matter: a case study of India. Atmos Environ 2012;49:1–12. Polissar AV, Hopke PK, Poirot RL. Atmospheric aerosol over Vermont: chemical composition and sources. Environ Sci Technol 2001;35(23):4604–21. Pey J, Alastuey A, Querol X. PM10 and PM2.5 sources at an insular location in the western Mediterranean by using source apportionment techniques. Sci Total Environ 2013;456:267–77. Reff A, Eberly SI, Bhave PV. Receptor modeling of ambient particulate matter data using positive matrix factorization: review of existing methods. J Air Waste Manage 2007;57:146–54. Saliba NA, El Jam F, El Tayar G, Obeid W, Roumie M. Origin and variability of particulate matter (PM10 and PM2.5) mass concentrations over an Eastern Mediterranean city. Atmos Res 2010;97:106–14. Samara C. Chemical mass balance source apportionment of TSP in a lignite-burning area of Western Macedonia, Greece. Atmos Environ 2005;39(35):6430–43. Simkhovich BZ, Kleinman MT, Kloner RA. Air pollution and cardiovascular injury: epidemiology, toxicology, and mechanisms. Cardiology 2008;52(9):719–26. Spyridaki A, Lazaridis M, Eleftheriadis K, Smolik J, Mihalopoulos N, Aleksandropoulou V. Modelling and evaluation of size-resolved aerosol characteristics in the Eastern Mediterranean during the SUB-AERO project. Atmos Environ 2006;40(32):6261–75. Tecer LH, Süren P, Alagha O, Karaca F, Tuncel G. Effect of meteorological parameters on fine and coarse particulate matter mass concentration in a coal-mining area in Zonguldak, Turkey, Japca J Air Waste Ma 2008;58(4):543–52. Tecer LH, Tuncel G, Karaca F, Alagha O, Süren P, Zararsız A, et al. Metallic composition and source apportionment of fine and coarse particles using positive matrix factorization in the southern Black Sea atmosphere. Atmos Res 2012;118:153–69. Yatin M, Tuncel S, Aras NK, Olmez I, Aygun S, Tuncel G. Atmospheric trace elements in Ankara, Turkey: 1. factors affecting chemical composition of fine particles. Atmos Environ 2000;34(8):1305–18.

Please cite this article as: Balcılar İ, et al, Chemical composition of Eastern Black Sea aerosol—Preliminary results, Sci Total Environ (2013), http:// dx.doi.org/10.1016/j.scitotenv.2013.12.023